exp run: Support hydra basic sweeper.

Requires `--queue` to be used alongside `--set-param`.

Doesn't strictly require `hydra.config.enabled`.

Add `get_hydra_sweeps` to `dvc/utils/hydra.py`.  Uses hydra `BasicSweeper` to generate the different sweeps.

`glob` syntax for sweeps is not supported in this P.R.

Closes #8083

---

Using setup from https://hydra.cc/docs/tutorials/basic/your_first_app/composition/ :

```console
$ dvc exp run -S "db=mysql,postgresql" -S "schema=warehouse,school" --queue
Queueing with '{'params.yaml': ['db=mysql', 'schema=warehouse']}'.
Queued experiment '5ab98b8' for future execution.
Queueing with '{'params.yaml': ['db=mysql', 'schema=school']}'.
Queued experiment '57c2fb6' for future execution.
Queueing with '{'params.yaml': ['db=postgresql', 'schema=warehouse']}'.
Queued experiment 'b9d6391' for future execution.
Queueing with '{'params.yaml': ['db=postgresql', 'schema=school']}'.
Queued experiment '145cd55' for future execution.
```

.git_archival.txt

@@ -1 +1,4 @@
+node: $Format:%H$
+node-date: $Format:%cI$
+describe-name: $Format:%(describe:tags=true)$
 ref-names: $Format:%D$

.github/CODEOWNERS

@@ -1 +0,0 @@
-*       @iterative/dvc

.github/FUNDING.yml

@@ -1,3 +0,0 @@
-# These are supported funding model platforms
-
-patreon: DVCorg

.github/workflows/benchmarks.yaml

@@ -11,10 +11,10 @@ jobs:
         os: [ubuntu-18.04]
     steps:
       - name: Cancel Previous Runs
-        uses: styfle/cancel-workflow-action@0.9.1
+        uses: styfle/cancel-workflow-action@0.10.0
         with:
           access_token: ${{ github.token }}
-      - uses: actions/setup-python@v3
+      - uses: actions/setup-python@v4
         with:
             python-version: 3.8
       - uses: iterative/dvc-bench@main

.github/workflows/packages.yaml

@@ -6,152 +6,92 @@ on:
   release:
     types: [released, prereleased]
   workflow_dispatch:
+
 jobs:
-  deb:
-    runs-on: ubuntu-18.04
+  binary:
+    strategy:
+      matrix:
+        include:
+         - {pkg: "deb",  os: "ubuntu-18.04", asset: "dvc_${{ github.event.release.tag_name }}_amd64.deb"}
+         - {pkg: "rpm", os: "ubuntu-18.04", asset: "dvc-${{ github.event.release.tag_name }}-1.x86_64.rpm"}
+         - {pkg: "osxpkg", os: "macos-10.15", asset: "dvc-${{ github.event.release.tag_name }}.pkg"}
+         - {pkg: "exe",  os: "windows-2019", asset: "dvc-${{ github.event.release.tag_name }}.exe"}
+
+    name: ${{ matrix.pkg }}
+    runs-on: ${{ matrix.os }}
     steps:
     - uses: actions/checkout@v3
       with:
           fetch-depth: 0
+
     - name: Set up Python 3.9
-      uses: actions/setup-python@v3
+      uses: actions/setup-python@v4
       with:
         python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
+        cache: pip
+        cache-dependency-path: |
+          setup.cfg
+          scripts/build-requirements.txt
+
+    - name: Set up Ruby 2.6
+      uses: ruby/setup-ruby@v1
+      if: matrix.pkg != 'exe'
       with:
-        ruby-version: 2.4
-    - name: Install deps
-      run: |
-        pip install .[all]
-        pip install -r scripts/build-requirements.txt
-        gem install --no-document fpm
-    - name: Build deb
-      run: python scripts/build.py deb
-    - name: Publish deb
-      if: github.event_name == 'release'
-      uses: actions/[email protected]
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      with:
-        upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/fpm/dvc_${{ github.event.release.tag_name }}_amd64.deb
-        asset_name: dvc_${{ github.event.release.tag_name }}_amd64.deb
-        asset_content_type: binary/octet-stream
-  rpm:
-    runs-on: ubuntu-18.04
-    steps:
-    - uses: actions/checkout@v3
-      with:
-        fetch-depth: 0
-    - name: Set up Python 3.9
-      uses: actions/setup-python@v3
-      with:
-        python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
-      with:
-        ruby-version: 2.4
-    - name: Install deps
-      run: |
-        pip install .[all]
-        pip install -r scripts/build-requirements.txt
-        gem install --no-document fpm
-    - name: Build rpm
-      run: python scripts/build.py rpm
-    - name: Publish rpm
-      if: github.event_name == 'release'
-      uses: actions/[email protected]
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      with:
-        upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/fpm/dvc-${{ github.event.release.tag_name }}-1.x86_64.rpm
-        asset_name: dvc-${{ github.event.release.tag_name }}-1.x86_64.rpm
-        asset_content_type: binary/octet-stream
-  osxpkg:
-    runs-on: macos-10.15
-    steps:
-    - uses: actions/checkout@v3
-      with:
-        fetch-depth: 0
-    - name: Set up Python 3.9
-      uses: actions/setup-python@v3
-      with:
-        python-version: 3.9
-    - name: Set up Ruby 2.4
-      uses: actions/[email protected]
-      with:
-        ruby-version: 2.4
-    - name: Install deps
-      run: |
-        # available lxml (webdav dependency) wheels are built with an old
-        # SDK, which breaks osxpkg notarization. Building from scratch
-        # instead.
-        pip install .[all] --no-binary lxml
-        pip install -r scripts/build-requirements.txt
-        gem install --no-document fpm
-    - name: Build
-      run: python scripts/build.py osxpkg
-    - name: Publish
-      if: github.event_name == 'release'
-      uses: actions/[email protected]
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      with:
-        upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/fpm/dvc-${{ github.event.release.tag_name }}.pkg
-        asset_name: dvc-${{ github.event.release.tag_name }}.pkg
-        asset_content_type: binary/octet-stream
-  exe:
-    runs-on: windows-2019
-    steps:
-    - uses: actions/checkout@v3
-      with:
-        fetch-depth: 0
-    - name: Set up Python 3.9
-      uses: actions/setup-python@v3
-      with:
-        python-version: 3.9
+        ruby-version: '2.6'
+
+    - name: Install fpm
+      if: matrix.pkg != 'exe'
+      run: gem install --no-document fpm
+
     - name: Install deps
       run: |
+        pip install -U wheel pip setuptools
         pip install .[all]
         pip install -r scripts/build-requirements.txt
+
     - name: Pull images
-      run: |
-        dvc pull
-    - name: Build
-      run: python scripts\build.py exe
-      shell: cmd
-    - name: Publish
+      run: dvc pull
+
+    - name: Build ${{ matrix.pkg }}
+      run: python scripts/build.py ${{ matrix.pkg }}
+
+    - name: Publish ${{ matrix.pkg }}
       if: github.event_name == 'release'
+      uses: actions/[email protected]
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      uses: actions/[email protected]
       with:
         upload_url: ${{ github.event.release.upload_url }}
-        asset_path: scripts/innosetup/dvc-${{ github.event.release.tag_name }}.exe
-        asset_name: dvc-${{ github.event.release.tag_name }}.exe
+        asset_path: scripts/${{ matrix.pkg == 'exe' && 'innosetup' || 'fpm' }}/${{ matrix.asset }}
+        asset_name: ${{ matrix.asset }}
         asset_content_type: binary/octet-stream
+
   pip:
-    runs-on: ubuntu-18.04
+    runs-on: ubuntu-20.04
     steps:
     - uses: actions/checkout@v3
       with:
         fetch-depth: 0
+
     - name: Set up Python 3.8
-      uses: actions/setup-python@v3
+      uses: actions/setup-python@v4
       with:
         python-version: 3.8
+        cache: pip
+        cache-dependency-path: setup.cfg
+
     - name: Install
-      run: pip install ".[all,tests]"
+      run: |
+        pip install -U wheel pip build setuptools
+        pip install ".[all,tests]"
+
     - name: Build packages
       run: |
         pip install -U pip
         ./scripts/build_package.sh
+
     - name: Publish packages to PyPI
       if: github.event_name == 'release'
-      uses: pypa/gh-action-pypi-publish@master
+      uses: pypa/gh-action-pypi-publish@release/v1
       with:
-        user: __token__
         password: ${{ secrets.PYPI_TOKEN }}

.github/workflows/tests.yaml

@@ -17,14 +17,14 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - name: Cancel Previous Runs
-      uses: styfle/cancel-workflow-action@0.9.1
+      uses: styfle/cancel-workflow-action@0.10.0
       with:
         access_token: ${{ github.token }}
     - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python 3.8
-      uses: actions/setup-python@v3
+      uses: actions/setup-python@v4
       with:
         python-version: 3.8
     - name: Install requirements
@@ -43,12 +43,15 @@ jobs:
       matrix:
         os: [ubuntu-20.04, windows-latest, macos-latest]
         pyv: ["3.8", "3.9", "3.10"]
+        include:
+        - {os: ubuntu-latest, pyv: "3.11-dev"}
+
     steps:
     - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python
-      uses: actions/setup-python@v3
+      uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.pyv }}
     - name: get pip cache dir
@@ -68,6 +71,8 @@ jobs:
       run: |
         pip install --upgrade pip setuptools wheel
         pip install -e ".[dev]"
+      env:
+        AIOHTTP_NO_EXTENSIONS: ${{ matrix.pyv == '3.11-dev' && '1' }}
     - name: run tests
       timeout-minutes: 40
       run: >-

.pre-commit-config.yaml

@@ -6,9 +6,9 @@ repos:
       - id: black
         language_version: python3
     repo: https://github.com/ambv/black
-    rev: 22.3.0
+    rev: 22.6.0
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.2.0
+    rev: v4.3.0
     hooks:
       - id: check-toml
       - id: check-yaml

README.rst

@@ -3,106 +3,114 @@
 `Website <https://dvc.org>`_
 • `Docs <https://dvc.org/doc>`_
 • `Blog <http://blog.dataversioncontrol.com>`_
-• `Twitter <https://twitter.com/DVCorg>`_
-• `Chat (Community & Support) <https://dvc.org/chat>`_
 • `Tutorial <https://dvc.org/doc/get-started>`_
-• `Mailing List <https://sweedom.us10.list-manage.com/subscribe/post?u=a08bf93caae4063c4e6a351f6&id=24c0ecc49a>`_
+• `Related Technologies <https://dvc.org/doc/user-guide/related-technologies>`_
+• `How DVC works`_
+• `VS Code Extension`_
+• `Installation`_
+• `Contributing`_
+• `Community and Support`_
 
-|CI| |Maintainability| |Coverage| |Donate| |DOI|
+|CI| |Python Version| |Coverage| |VS Code| |DOI|
 
 |PyPI| |Packages| |Brew| |Conda| |Choco| |Snap|
 
 |
 
-**Data Version Control** or **DVC** helps you develop reproducible machine
-learning projects:
+**Data Version Control** or **DVC** is a command line tool and `VS Code Extension`_ to help you develop reproducible machine learning projects:
 
-#. Version your data and models.
+#. **Version** your data and models.
+   Store them in your cloud storage but keep their version info in your Git repo.
 
-#. Connect them to your code and parameters to form a pipeline.
+#. **Iterate** fast with lightweight pipelines.
+   When you make changes, only run the steps impacted by those changes.
 
-#. Modify data, code, or parameters and run your modified pipeline.
+#. **Track** experiments in your local Git repo (no servers needed).
 
-#. Iterate and track each experiment.
+#. **Compare** any data, code, parameters, model, or performance plots.
 
-#. Share, compare, and reconstruct experiments.
-
-**Why DVC?**
-
-#. **Store** data and models (big) in your cloud storage, but keep their version
-   info (small) in your Git repo.
-
-#. **Iterate** fast with straightforward YAML pipelines. When you make changes,
-   only run the steps impacted by those changes.
-
-#. **Track** experiments in your local Git repo (no servers needed) and choose
-   which to keep and share.
-
-#. **Compare** any data, code, parameters, metrics, or plots and analyze however
-   you want.
-
-#. **Reconstruct** anyone's experiment with a single command.
-
-.. contents:: **Contents**
-  :backlinks: none
-
-How DVC works
-=============
-
-We encourage you to read our `Get Started <https://dvc.org/doc/get-started>`_ guide to better understand what DVC
-is and how it can fit your scenarios.
-
-The easiest (but not perfect!) *analogy* to describe it: DVC is Git (or Git-LFS to be precise) & Makefiles
-made right and tailored specifically for ML and Data Science scenarios.
-
-#. ``Git/Git-LFS`` part - DVC helps store and share data artifacts and models, connecting them with a Git repository.
-#. ``Makefile``\ s part - DVC describes how one data or model artifact was built from other data and code.
-
-DVC usually runs along with Git. Git is used as usual to store and version code (including DVC meta-files). DVC helps
-to store data and model files seamlessly out of Git, while preserving almost the same user experience as if they
-were stored in Git itself. To store and share the data cache, DVC supports multiple remotes - any cloud (S3, Azure,
-Google Cloud, etc) or any on-premise network storage (via SSH, for example).
-
-|Flowchart|
-
-The DVC pipelines (computational graph) feature connects code and data together. It is possible to explicitly
-specify all steps required to produce a model: input dependencies including data, commands to run,
-and output information to be saved. See the quick start section below or
-the `Get Started <https://dvc.org/doc/get-started>`_ tutorial to learn more.
+#. **Share** experiments and automatically reproduce anyone's experiment.
 
 Quick start
 ===========
 
-Please read `Get Started <https://dvc.org/doc/get-started>`_ guide for a full version. Common workflow commands include:
+    Please read our `Command Reference <https://dvc.org/doc/command-reference>`_ for a complete list.
 
-+-----------------------------------+----------------------------------------------------------------------------+
-| Step                              | Command                                                                    |
-+===================================+============================================================================+
-| Track data                        | | ``$ git add train.py``                                                   |
-|                                   | | ``$ dvc add images.zip``                                                 |
-+-----------------------------------+----------------------------------------------------------------------------+
-| Connect code and data by commands | | ``$ dvc run -n prepare -d images.zip -o images/ unzip -q images.zip``    |
-|                                   | | ``$ dvc run -n train -d images/ -d train.py -o model.p python train.py`` |
-+-----------------------------------+----------------------------------------------------------------------------+
-| Make changes and reproduce        | | ``$ vi train.py``                                                        |
-|                                   | | ``$ dvc repro model.p.dvc``                                              |
-+-----------------------------------+----------------------------------------------------------------------------+
-| Share code                        | | ``$ git add .``                                                          |
-|                                   | | ``$ git commit -m 'The baseline model'``                                 |
-|                                   | | ``$ git push``                                                           |
-+-----------------------------------+----------------------------------------------------------------------------+
-| Share data and ML models          | | ``$ dvc remote add myremote -d s3://mybucket/image_cnn``                 |
-|                                   | | ``$ dvc push``                                                           |
-+-----------------------------------+----------------------------------------------------------------------------+
+A common CLI workflow includes:
+
+
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+| Task                              | Terminal                                                                                           |
++===================================+====================================================================================================+
+| Track data                        | | ``$ git add train.py params.yaml``                                                               |
+|                                   | | ``$ dvc add images/``                                                                            |
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+| Connect code and data             | | ``$ dvc stage add -n featurize -d images/ -o features/ python featurize.py``                     |
+|                                   | | ``$ dvc stage add -n train -d features/ -d train.py -o model.p -M metrics.json python train.py`` |
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+| Make changes and experiment       | | ``$ dvc exp run -n exp-baseline``                                                                |
+|                                   | | ``$ vi train.py``                                                                                |
+|                                   | | ``$ dvc exp run -n exp-code-change``                                                             |
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+| Compare and select experiments    | | ``$ dvc exp show``                                                                               |
+|                                   | | ``$ dvc exp apply exp-baseline``                                                                 |
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+| Share code                        | | ``$ git add .``                                                                                  |
+|                                   | | ``$ git commit -m 'The baseline model'``                                                         |
+|                                   | | ``$ git push``                                                                                   |
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+| Share data and ML models          | | ``$ dvc remote add myremote -d s3://mybucket/image_cnn``                                         |
+|                                   | | ``$ dvc push``                                                                                   |
++-----------------------------------+----------------------------------------------------------------------------------------------------+
+
+How DVC works
+=============
+
+    We encourage you to read our `Get Started
+    <https://dvc.org/doc/get-started>`_ docs to better understand what DVC
+    does and how it can fit your scenarios.
+
+The closest *analogies* to describe the main DVC features are these:
+
+#. **Git for data**: Store and share data artifacts (like Git-LFS but without a server) and models, connecting them with a Git repository. Data management meets GitOps!
+#. **Makefiles** for ML: Describes how data or model artifacts are built from other data and code in a standard format. Now you can version your data pipelines with Git.
+#. Local **experiment tracking**: Turn your machine into an ML experiment management platform, and collaborate with others using existing Git hosting (Github, Gitlab, etc.).
+
+Git is employed as usual to store and version code (including DVC meta-files as placeholders for data).
+DVC `stores data and model files <https://dvc.org/doc/start/data-management>`_ seamlessly in a cache outside of Git, while preserving almost the same user experience as if they were in the repo.
+To share and back up the *data cache*, DVC supports multiple remote storage platforms - any cloud (S3, Azure, Google Cloud, etc.) or on-premise network storage (via SSH, for example).
+
+|Flowchart|
+
+`DVC pipelines <https://dvc.org/doc/start/data-management/pipelines>`_ (computational graphs) connect code and data together.
+They specify all steps required to produce a model: input dependencies including code, data, commands to run; and output information to be saved.
+
+Last but not least, `DVC Experiment Versioning <https://dvc.org/doc/start/experiments>`_ lets you prepare and run a large number of experiments.
+Their results can be filtered and compared based on hyperparameters and metrics, and visualized with multiple plots.
+
+.. _`VS Code Extension`:
+
+Visual Studio Code Extension
+============================
+
+|VS Code|
+
+To use DVC as a GUI right from your VS Code IDE, install the `DVC Extension <https://marketplace.visualstudio.com/items?itemName=Iterative.dvc>`_ from the Marketplace.
+It currently features experiment tracking and data management, and more features (data pipeline support, etc.) are coming soon!
+
+|VS Code Extension Overview|
+
+    Note: You'll have to install core DVC on your system separately (as detailed
+    below). The Extension will guide you if needed.
 
 Installation
 ============
 
-There are four options to install DVC: ``pip``, Homebrew, Conda (Anaconda) or an OS-specific package.
+There are several ways to install DVC: in VS Code; using ``snap``, ``choco``, ``brew``, ``conda``, ``pip``; or with an OS-specific package.
 Full instructions are `available here <https://dvc.org/doc/get-started/install>`_.
 
-Snap (Snapcraft/Linux)
-----------------------
+Snapcraft (Linux)
+-----------------
 
 |Snap|
 
@@ -111,11 +119,10 @@ Snap (Snapcraft/Linux)
    snap install dvc --classic
 
 This corresponds to the latest tagged release.
-Add ``--beta`` for the latest tagged release candidate,
-or ``--edge`` for the latest ``main`` version.
+Add ``--beta`` for the latest tagged release candidate, or ``--edge`` for the latest ``main`` version.
 
-Choco (Chocolatey/Windows)
---------------------------
+Chocolatey (Windows)
+--------------------
 
 |Choco|
 
@@ -123,8 +130,8 @@ Choco (Chocolatey/Windows)
 
    choco install dvc
 
-Brew (Homebrew/Mac OS)
-----------------------
+Brew (mac OS)
+-------------
 
 |Brew|
 
@@ -132,8 +139,8 @@ Brew (Homebrew/Mac OS)
 
    brew install dvc
 
-Conda (Anaconda)
-----------------
+Anaconda (Any platform)
+-----------------------
 
 |Conda|
 
@@ -142,11 +149,10 @@ Conda (Anaconda)
    conda install -c conda-forge mamba # installs much faster than conda
    mamba install -c conda-forge dvc
 
-Depending on the remote storage type you plan to use to keep and share your data, you might need to
-install optional dependencies: `dvc-s3`, `dvc-azure`, `dvc-gdrive`, `dvc-gs`, `dvc-oss`, `dvc-ssh`.
+Depending on the remote storage type you plan to use to keep and share your data, you might need to install optional dependencies: `dvc-s3`, `dvc-azure`, `dvc-gdrive`, `dvc-gs`, `dvc-oss`, `dvc-ssh`.
 
-pip (PyPI)
-----------
+PyPI (Python)
+-------------
 
 |PyPI|
 
@@ -154,10 +160,8 @@ pip (PyPI)
 
    pip install dvc
 
-Depending on the remote storage type you plan to use to keep and share your data, you might need to specify
-one of the optional dependencies: ``s3``, ``gs``, ``azure``, ``oss``, ``ssh``. Or ``all`` to include them all.
-The command should look like this: ``pip install dvc[s3]`` (in this case AWS S3 dependencies such as ``boto3``
-will be installed automatically).
+Depending on the remote storage type you plan to use to keep and share your data, you might need to specify one of the optional dependencies: ``s3``, ``gs``, ``azure``, ``oss``, ``ssh``. Or ``all`` to include them all.
+The command should look like this: ``pip install 'dvc[s3]'`` (in this case AWS S3 dependencies such as ``boto3`` will be installed automatically).
 
 To install the development version, run:
 
@@ -165,13 +169,13 @@ To install the development version, run:
 
    pip install git+git://github.com/iterative/dvc
 
-Package
--------
+Package (Platform-specific)
+---------------------------
 
 |Packages|
 
-Self-contained packages for Linux, Windows, and Mac are available. The latest version of the packages
-can be found on the GitHub `releases page <https://github.com/iterative/dvc/releases>`_.
+Self-contained packages for Linux, Windows, and Mac are available.
+The latest version of the packages can be found on the GitHub `releases page <https://github.com/iterative/dvc/releases>`_.
 
 Ubuntu / Debian (deb)
 ^^^^^^^^^^^^^^^^^^^^^
@@ -191,64 +195,32 @@ Fedora / CentOS (rpm)
    sudo yum update
    sudo yum install dvc
 
-Comparison to related technologies
-==================================
-
-#. Data Engineering tools such as `AirFlow <https://airflow.apache.org/>`_,
-   `Luigi <https://github.com/spotify/luigi>`_, and others - in DVC data,
-   model and ML pipelines represent a single ML project focused on data
-   scientists' experience.  Data engineering tools orchestrate multiple data
-   projects and focus on efficient execution. A DVC project can be used from
-   existing data pipelines as a single execution step.
-
-#. `Git-annex <https://git-annex.branchable.com/>`_ - DVC uses the idea of storing the content of large files (which should
-   not be in a Git repository) in a local key-value store, and uses file hardlinks/symlinks instead of
-   copying/duplicating files.
-
-#. `Git-LFS <https://git-lfs.github.com/>`_ - DVC is compatible with many
-   remote storage services (S3, Google Cloud, Azure, SSH, etc). DVC also
-   uses reflinks or hardlinks to avoid copy operations on checkouts; thus
-   handling large data files much more efficiently.
-
-#. Makefile (and analogues including ad-hoc scripts) - DVC tracks
-   dependencies (in a directed acyclic graph).
-
-#. `Workflow Management Systems <https://en.wikipedia.org/wiki/Workflow_management_system>`_ - DVC is a workflow
-   management system designed specifically to manage machine learning experiments. DVC is built on top of Git.
-
-#. `DAGsHub <https://dagshub.com/>`_ - online service to host DVC
-   projects.  It provides a useful UI around DVC repositories and integrates
-   other tools.
-
-#. `DVC Studio <https://studio.iterative.ai/>`_ - official online
-   platform for DVC projects.  It can be used to manage data and models, run
-   and track experiments, and visualize and share results.  Also, it
-   integrates with `CML (CI/CD for ML) <https://cml.dev/>`__ for training
-   models in the cloud or Kubernetes.
-
-
 Contributing
 ============
 
-|Maintainability| |Donate|
+|Maintainability|
 
-Contributions are welcome! Please see our `Contributing Guide <https://dvc.org/doc/user-guide/contributing/core>`_ for more
-details. Thanks to all our contributors!
+Contributions are welcome!
+Please see our `Contributing Guide <https://dvc.org/doc/user-guide/contributing/core>`_ for more details.
+Thanks to all our contributors!
 
 |Contribs|
 
-Mailing List
-============
+Community and Support
+=====================
 
-Want to stay up to date? Want to help improve DVC by participating in our occasional polls? Subscribe to our `mailing list <https://sweedom.us10.list-manage.com/subscribe/post?u=a08bf93caae4063c4e6a351f6&id=24c0ecc49a>`_. No spam, really low traffic.
+* `Twitter <https://twitter.com/DVCorg>`_
+* `Forum <https://discuss.dvc.org/>`_
+* `Discord Chat <https://dvc.org/chat>`_
+* `Email <mailto:[email protected]>`_
+* `Mailing List <https://sweedom.us10.list-manage.com/subscribe/post?u=a08bf93caae4063c4e6a351f6&id=24c0ecc49a>`_
 
 Copyright
 =========
 
 This project is distributed under the Apache license version 2.0 (see the LICENSE file in the project root).
 
-By submitting a pull request to this project, you agree to license your contribution under the Apache license version
-2.0 to this project.
+By submitting a pull request to this project, you agree to license your contribution under the Apache license version 2.0 to this project.
 
 Citation
 ========
@@ -265,6 +237,9 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://dvc.org
    :alt: DVC logo
 
+.. |VS Code Extension Overview| image:: https://raw.githubusercontent.com/iterative/vscode-dvc/main/extension/docs/overview.gif
+   :alt: DVC Extension for VS Code
+
 .. |CI| image:: https://github.com/iterative/dvc/workflows/Tests/badge.svg?branch=main
    :target: https://github.com/iterative/dvc/actions
    :alt: GHA Tests
@@ -273,14 +248,14 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://codeclimate.com/github/iterative/dvc
    :alt: Code Climate
 
+.. |Python Version| image:: https://img.shields.io/pypi/pyversions/dvc
+   :target: https://pypi.org/project/dvc
+   :alt: Python Version
+
 .. |Coverage| image:: https://codecov.io/gh/iterative/dvc/branch/main/graph/badge.svg
    :target: https://codecov.io/gh/iterative/dvc
    :alt: Codecov
 
-.. |Donate| image:: https://img.shields.io/badge/patreon-donate-green.svg?logo=patreon
-   :target: https://www.patreon.com/DVCorg/overview
-   :alt: Donate
-
 .. |Snap| image:: https://img.shields.io/badge/snap-install-82BEA0.svg?logo=snapcraft
    :target: https://snapcraft.io/dvc
    :alt: Snapcraft
@@ -316,3 +291,7 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
 .. |Contribs| image:: https://contrib.rocks/image?repo=iterative/dvc
    :target: https://github.com/iterative/dvc/graphs/contributors
    :alt: Contributors
+
+.. |VS Code| image:: https://vsmarketplacebadge.apphb.com/version/Iterative.dvc.svg
+   :target: https://marketplace.visualstudio.com/items?itemName=Iterative.dvc
+   :alt: VS Code Extension

dvc/analytics.py

@@ -50,7 +50,7 @@ def is_enabled():
             Config(validate=False).get("core", {}).get("analytics", "true")
         )
 
-    logger.debug("Analytics is {}abled.".format("en" if enabled else "dis"))
+    logger.debug("Analytics is %sabled.", "en" if enabled else "dis")
 
     return enabled
 
@@ -187,4 +187,4 @@ def _find_or_create_user_id():
             return user_id
 
     except LockError:
-        logger.debug(f"Failed to acquire '{lockfile}'")
+        logger.debug("Failed to acquire '%s'", lockfile)

dvc/annotations.py

@@ -0,0 +1,35 @@
+from dataclasses import asdict, dataclass, field, fields
+from typing import Any, ClassVar, Dict, List, Optional
+
+from funcy import compact
+
+
+@dataclass
+class Annotation:
+    PARAM_DESC: ClassVar[str] = "desc"
+    PARAM_TYPE: ClassVar[str] = "type"
+    PARAM_LABELS: ClassVar[str] = "labels"
+    PARAM_META: ClassVar[str] = "meta"
+
+    desc: Optional[str] = None
+    type: Optional[str] = None
+    labels: List[str] = field(default_factory=list)
+    meta: Dict[str, Any] = field(default_factory=dict)
+
+    def update(self, **kwargs) -> "Annotation":
+        for attr, value in kwargs.items():
+            if value and hasattr(self, attr):
+                setattr(self, attr, value)
+        return self
+
+    def to_dict(self) -> Dict[str, str]:
+        return compact(asdict(self))
+
+
+ANNOTATION_FIELDS = [field.name for field in fields(Annotation)]
+ANNOTATION_SCHEMA = {
+    Annotation.PARAM_DESC: str,
+    Annotation.PARAM_TYPE: str,
+    Annotation.PARAM_LABELS: [str],
+    Annotation.PARAM_META: object,
+}

dvc/api.py

@@ -1,124 +0,0 @@
-import os
-from contextlib import _GeneratorContextManager as GCM
-
-from funcy import reraise
-
-from dvc.exceptions import OutputNotFoundError, PathMissingError
-from dvc.repo import Repo
-
-
-def get_url(path, repo=None, rev=None, remote=None):
-    """
-    Returns the URL to the storage location of a data file or directory tracked
-    in a DVC repo. For Git repos, HEAD is used unless a rev argument is
-    supplied. The default remote is tried unless a remote argument is supplied.
-
-    Raises OutputNotFoundError if the file is not tracked by DVC.
-
-    NOTE: This function does not check for the actual existence of the file or
-    directory in the remote storage.
-    """
-    with Repo.open(repo, rev=rev, subrepos=True, uninitialized=True) as _repo:
-        fs_path = _repo.dvcfs.from_os_path(path)
-        with reraise(FileNotFoundError, PathMissingError(path, repo)):
-            info = _repo.dvcfs.info(fs_path)
-
-        dvc_info = info.get("dvc_info")
-        if not dvc_info:
-            raise OutputNotFoundError(path, repo)
-
-        dvc_repo = info["repo"]
-        md5 = dvc_info["md5"]
-
-        return dvc_repo.cloud.get_url_for(remote, checksum=md5)
-
-
-def open(  # noqa, pylint: disable=redefined-builtin
-    path, repo=None, rev=None, remote=None, mode="r", encoding=None
-):
-    """
-    Open file in the supplied path tracked in a repo (both DVC projects and
-    plain Git repos are supported). For Git repos, HEAD is used unless a rev
-    argument is supplied. The default remote is tried unless a remote argument
-    is supplied. It may only be used as a context manager:
-
-        with dvc.api.open(
-                'path/to/file',
-                repo='https://example.com/url/to/repo'
-                ) as fd:
-            # ... Handle file object fd
-    """
-    args = (path,)
-    kwargs = {
-        "repo": repo,
-        "remote": remote,
-        "rev": rev,
-        "mode": mode,
-        "encoding": encoding,
-    }
-    return _OpenContextManager(_open, args, kwargs)
-
-
-class _OpenContextManager(GCM):
-    def __init__(
-        self, func, args, kwds
-    ):  # pylint: disable=super-init-not-called
-        self.gen = func(*args, **kwds)
-        self.func, self.args, self.kwds = func, args, kwds
-
-    def __getattr__(self, name):
-        raise AttributeError(
-            "dvc.api.open() should be used in a with statement."
-        )
-
-
-def _open(path, repo=None, rev=None, remote=None, mode="r", encoding=None):
-    with Repo.open(repo, rev=rev, subrepos=True, uninitialized=True) as _repo:
-        with _repo.open_by_relpath(
-            path, remote=remote, mode=mode, encoding=encoding
-        ) as fd:
-            yield fd
-
-
-def read(path, repo=None, rev=None, remote=None, mode="r", encoding=None):
-    """
-    Returns the contents of a tracked file (by DVC or Git). For Git repos, HEAD
-    is used unless a rev argument is supplied. The default remote is tried
-    unless a remote argument is supplied.
-    """
-    with open(
-        path, repo=repo, rev=rev, remote=remote, mode=mode, encoding=encoding
-    ) as fd:
-        return fd.read()
-
-
-def make_checkpoint():
-    """
-    Signal DVC to create a checkpoint experiment.
-
-    If the current process is being run from DVC, this function will block
-    until DVC has finished creating the checkpoint. Otherwise, this function
-    will return immediately.
-    """
-    import builtins
-    from time import sleep
-
-    from dvc.env import DVC_CHECKPOINT, DVC_ROOT
-    from dvc.stage.monitor import CheckpointTask
-
-    if os.getenv(DVC_CHECKPOINT) is None:
-        return
-
-    root_dir = os.getenv(DVC_ROOT, Repo.find_root())
-    signal_file = os.path.join(
-        root_dir, Repo.DVC_DIR, "tmp", CheckpointTask.SIGNAL_FILE
-    )
-
-    with builtins.open(signal_file, "w", encoding="utf-8") as fobj:
-        # NOTE: force flushing/writing empty file to disk, otherwise when
-        # run in certain contexts (pytest) file may not actually be written
-        fobj.write("")
-        fobj.flush()
-        os.fsync(fobj.fileno())
-    while os.path.exists(signal_file):
-        sleep(0.1)

dvc/api/__init__.py

@@ -0,0 +1,9 @@
+from .data import (  # noqa, pylint: disable=redefined-builtin
+    get_url,
+    open,
+    read,
+)
+from .experiments import make_checkpoint
+from .params import params_show
+
+__all__ = ["get_url", "make_checkpoint", "open", "params_show", "read"]

dvc/api/data.py

@@ -0,0 +1,213 @@
+from contextlib import _GeneratorContextManager as GCM
+from typing import Optional
+
+from funcy import reraise
+
+from dvc.exceptions import OutputNotFoundError, PathMissingError
+from dvc.repo import Repo
+
+
+def get_url(path, repo=None, rev=None, remote=None):
+    """
+    Returns the URL to the storage location of a data file or directory tracked
+    in a DVC repo. For Git repos, HEAD is used unless a rev argument is
+    supplied. The default remote is tried unless a remote argument is supplied.
+
+    Raises OutputNotFoundError if the file is not tracked by DVC.
+
+    NOTE: This function does not check for the actual existence of the file or
+    directory in the remote storage.
+    """
+    with Repo.open(repo, rev=rev, subrepos=True, uninitialized=True) as _repo:
+        fs_path = _repo.dvcfs.from_os_path(path)
+        with reraise(FileNotFoundError, PathMissingError(path, repo)):
+            info = _repo.dvcfs.info(fs_path)
+
+        dvc_info = info.get("dvc_info")
+        if not dvc_info:
+            raise OutputNotFoundError(path, repo)
+
+        dvc_repo = info["repo"]
+        md5 = dvc_info["md5"]
+
+        return dvc_repo.cloud.get_url_for(remote, checksum=md5)
+
+
+class _OpenContextManager(GCM):
+    def __init__(
+        self, func, args, kwds
+    ):  # pylint: disable=super-init-not-called
+        self.gen = func(*args, **kwds)
+        self.func, self.args, self.kwds = func, args, kwds
+
+    def __getattr__(self, name):
+        raise AttributeError(
+            "dvc.api.open() should be used in a with statement."
+        )
+
+
+def open(  # noqa, pylint: disable=redefined-builtin
+    path: str,
+    repo: Optional[str] = None,
+    rev: Optional[str] = None,
+    remote: Optional[str] = None,
+    mode: str = "r",
+    encoding: Optional[str] = None,
+):
+    """
+    Opens a file tracked in a DVC project.
+
+    This function may only be used as a context manager (using the `with`
+    keyword, as shown in the examples).
+
+    This function makes a direct connection to the remote storage, so the file
+    contents can be streamed. Your code can process the data buffer as it's
+    streamed, which optimizes memory usage.
+
+    Note:
+        Use dvc.api.read() to load the complete file contents
+        in a single function call, no context manager involved.
+        Neither function utilizes disc space.
+
+    Args:
+        path (str): location and file name of the target to open,
+        relative to the root of `repo`.
+        repo (str, optional): location of the DVC project or Git Repo.
+            Defaults to the current DVC project (found by walking up from the
+            current working directory tree).
+            It can be a URL or a file system path.
+            Both HTTP and SSH protocols are supported for online Git repos
+            (e.g. [user@]server:project.git).
+        rev (str, optional): Any `Git revision`_ such as a branch or tag name,
+            a commit hash or a dvc experiment name.
+            Defaults to HEAD.
+            If `repo` is not a Git repo, this option is ignored.
+        remote (str, optional): Name of the `DVC remote`_ used to form the
+            returned URL string.
+            Defaults to the `default remote`_ of `repo`.
+            For local projects, the cache is tried before the default remote.
+        mode (str, optional): Specifies the mode in which the file is opened.
+            Defaults to "r" (read).
+            Mirrors the namesake parameter in builtin `open()`_.
+            Only reading `mode` is supported.
+        encoding(str, optional): `Codec`_ used to decode the file contents.
+            Defaults to None.
+            This should only be used in text mode.
+            Mirrors the namesake parameter in builtin `open()`_.
+
+    Returns:
+        _OpenContextManager: A context manager that generatse a corresponding
+            `file object`_.
+            The exact type of file object depends on the mode used.
+            For more details, please refer to Python's `open()`_ built-in,
+            which is used under the hood.
+
+    Raises:
+        AttributeError: If this method is not used as a context manager.
+        ValueError: If non-read `mode` is used.
+
+    Examples:
+
+        - Use data or models from a DVC repository.
+
+        Any file tracked in a DVC project (and stored remotely) can be
+        processed directly in your Python code with this API.
+        For example, an XML file tracked in a public DVC repo on GitHub can be
+        processed like this:
+
+        >>> from xml.sax import parse
+        >>> import dvc.api
+        >>> from mymodule import mySAXHandler
+
+        >>> with dvc.api.open(
+        ...     'get-started/data.xml',
+        ...     repo='https://github.com/iterative/dataset-registry'
+        ... ) as fd:
+        ...     parse(fd, mySAXHandler)
+
+        We use a SAX XML parser here because dvc.api.open() is able to stream
+        the data from remote storage.
+        The mySAXHandler object should handle the event-driven parsing of the
+        document in this case.
+        This increases the performance of the code (minimizing memory usage),
+        and is typically faster than loading the whole data into memory.
+
+        - Accessing private repos
+
+        This is just a matter of using the right repo argument, for example an
+        SSH URL (requires that the credentials are configured locally):
+
+        >>> import dvc.api
+
+        >>> with dvc.api.open(
+        ...     'features.dat',
+        ...     repo='[email protected]:path/to/repo.git'
+        ... ) as fd:
+        ...     # ... Process 'features'
+        ...     pass
+
+        - Use different versions of data
+
+        Any git revision (see `rev`) can be accessed programmatically.
+        For example, if your DVC repo has tagged releases of a CSV dataset:
+
+        >>> import csv
+        >>> import dvc.api
+        >>> with dvc.api.open(
+        ...     'clean.csv',
+        ...     rev='v1.1.0'
+        ... ) as fd:
+        ...     reader = csv.reader(fd)
+        ...     # ... Process 'clean' data from version 1.1.0
+
+    .. _Git revision:
+        https://git-scm.com/docs/revisions
+
+    .. _DVC remote:
+        https://dvc.org/doc/command-reference/remote
+
+    .. _default remote:
+        https://dvc.org/doc/command-reference/remote/default
+
+    .. _open():
+        https://docs.python.org/3/library/functions.html#open
+
+    .. _Codec:
+        https://docs.python.org/3/library/codecs.html#standard-encodings
+
+    .. _file object:
+        https://docs.python.org/3/glossary.html#term-file-object
+
+    """
+    if "r" not in mode:
+        raise ValueError("Only reading `mode` is supported.")
+
+    args = (path,)
+    kwargs = {
+        "repo": repo,
+        "remote": remote,
+        "rev": rev,
+        "mode": mode,
+        "encoding": encoding,
+    }
+    return _OpenContextManager(_open, args, kwargs)
+
+
+def _open(path, repo=None, rev=None, remote=None, mode="r", encoding=None):
+    with Repo.open(repo, rev=rev, subrepos=True, uninitialized=True) as _repo:
+        with _repo.open_by_relpath(
+            path, remote=remote, mode=mode, encoding=encoding
+        ) as fd:
+            yield fd
+
+
+def read(path, repo=None, rev=None, remote=None, mode="r", encoding=None):
+    """
+    Returns the contents of a tracked file (by DVC or Git). For Git repos, HEAD
+    is used unless a rev argument is supplied. The default remote is tried
+    unless a remote argument is supplied.
+    """
+    with open(
+        path, repo=repo, rev=rev, remote=remote, mode=mode, encoding=encoding
+    ) as fd:
+        return fd.read()

dvc/api/experiments.py

@@ -0,0 +1,33 @@
+import builtins
+import os
+from time import sleep
+
+from dvc.env import DVC_CHECKPOINT, DVC_ROOT
+from dvc.repo import Repo
+from dvc.stage.monitor import CheckpointTask
+
+
+def make_checkpoint():
+    """
+    Signal DVC to create a checkpoint experiment.
+
+    If the current process is being run from DVC, this function will block
+    until DVC has finished creating the checkpoint. Otherwise, this function
+    will return immediately.
+    """
+    if os.getenv(DVC_CHECKPOINT) is None:
+        return
+
+    root_dir = os.getenv(DVC_ROOT, Repo.find_root())
+    signal_file = os.path.join(
+        root_dir, Repo.DVC_DIR, "tmp", CheckpointTask.SIGNAL_FILE
+    )
+
+    with builtins.open(signal_file, "w", encoding="utf-8") as fobj:
+        # NOTE: force flushing/writing empty file to disk, otherwise when
+        # run in certain contexts (pytest) file may not actually be written
+        fobj.write("")
+        fobj.flush()
+        os.fsync(fobj.fileno())
+    while os.path.exists(signal_file):
+        sleep(0.1)

dvc/api/params.py

@@ -0,0 +1,278 @@
+from collections import Counter
+from typing import Dict, Iterable, Optional, Union
+
+from funcy import first
+
+from dvc.exceptions import DvcException
+from dvc.repo import Repo
+
+
+def params_show(
+    *targets: str,
+    repo: Optional[str] = None,
+    stages: Optional[Union[str, Iterable[str]]] = None,
+    rev: Optional[str] = None,
+    deps: bool = False,
+) -> Dict:
+    """Get parameters tracked in `repo`.
+
+    Without arguments, this function will retrieve all params from all tracked
+    parameter files, for the current working tree.
+
+    See the options below to restrict the parameters retrieved.
+
+    Args:
+        *targets (str, optional): Names of the parameter files to retrieve
+        params from. For example, "params.py, myparams.toml".
+        If no `targets` are provided, all parameter files tracked in `dvc.yaml`
+        will be used.
+        Note that targets don't necessarily have to be defined in `dvc.yaml`.
+        repo (str, optional): location of the DVC repository.
+            Defaults to the current project (found by walking up from the
+            current working directory tree).
+            It can be a URL or a file system path.
+            Both HTTP and SSH protocols are supported for online Git repos
+            (e.g. [user@]server:project.git).
+        stages (Union[str, Iterable[str]], optional): Name or names of the
+            stages to retrieve parameters from.
+            Defaults to `None`.
+            If `None`, all parameters from all stages will be retrieved.
+            If this method is called from a different location to the one where
+            the `dvc.yaml` is found, the relative path to the `dvc.yaml` must
+            be provided as a prefix with the syntax `{relpath}:{stage}`.
+            For example: `subdir/dvc.yaml:stage-0` or `../dvc.yaml:stage-1`.
+        rev (str, optional): Name of the `Git revision`_ to retrieve parameters
+            from.
+            Defaults to `None`.
+            An example of git revision can be a branch or tag name, a commit
+            hash or a dvc experiment name.
+            If `repo` is not a Git repo, this option is ignored.
+            If `None`, the current working tree will be used.
+        deps (bool, optional): Whether to retrieve only parameters that are
+            stage dependencies or not.
+            Defaults to `False`.
+
+    Returns:
+        Dict: See Examples below.
+
+    Raises:
+        DvcException: If no params are found in `repo`.
+
+    Examples:
+
+        - No arguments.
+
+        Working on https://github.com/iterative/example-get-started
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show()
+        >>> print(json.dumps(params, indent=4))
+        {
+            "prepare": {
+                "split": 0.2,
+                "seed": 20170428
+            },
+            "featurize": {
+                "max_features": 200,
+                "ngrams": 2
+            },
+            "train": {
+                "seed": 20170428,
+                "n_est": 50,
+                "min_split": 0.01
+            }
+        }
+
+        ---
+
+        - Filtering with `stages`.
+
+        Working on https://github.com/iterative/example-get-started
+
+        `stages` can a single string:
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show(stages="prepare")
+        >>> print(json.dumps(params, indent=4))
+        {
+            "prepare": {
+                "split": 0.2,
+                "seed": 20170428
+            }
+        }
+
+        Or an iterable of strings:
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show(stages=["prepare", "train"])
+        >>> print(json.dumps(params, indent=4))
+        {
+            "prepare": {
+                "split": 0.2,
+                "seed": 20170428
+            },
+            "train": {
+                "seed": 20170428,
+                "n_est": 50,
+                "min_split": 0.01
+            }
+        }
+
+        ---
+
+        - Using `rev`.
+
+        Working on https://github.com/iterative/example-get-started
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show(rev="tune-hyperparams")
+        >>> print(json.dumps(params, indent=4))
+        {
+            "prepare": {
+                "split": 0.2,
+                "seed": 20170428
+            },
+            "featurize": {
+                "max_features": 200,
+                "ngrams": 2
+            },
+            "train": {
+                "seed": 20170428,
+                "n_est": 100,
+                "min_split": 8
+            }
+        }
+
+        ---
+
+        - Using `targets`.
+
+        Working on `multi-params-files` folder of
+        https://github.com/iterative/pipeline-conifguration
+
+        You can pass a single target:
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show("params.yaml")
+        >>> print(json.dumps(params, indent=4))
+        {
+            "run_mode": "prod",
+            "configs": {
+                "dev": "configs/params_dev.yaml",
+                "test": "configs/params_test.yaml",
+                "prod": "configs/params_prod.yaml"
+            },
+            "evaluate": {
+                "dataset": "micro",
+                "size": 5000,
+                "metrics": ["f1", "roc-auc"],
+                "metrics_file": "reports/metrics.json",
+                "plots_cm": "reports/plot_confusion_matrix.png"
+            }
+        }
+
+
+        Or multiple targets:
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show(
+        ...     "configs/params_dev.yaml", "configs/params_prod.yaml")
+        >>> print(json.dumps(params, indent=4))
+        {
+            "configs/params_prod.yaml:run_mode": "prod",
+            "configs/params_prod.yaml:config_file": "configs/params_prod.yaml",
+            "configs/params_prod.yaml:data_load": {
+                "dataset": "large",
+                "sampling": {
+                "enable": true,
+                "size": 50000
+                }
+            },
+            "configs/params_prod.yaml:train": {
+                "epochs": 1000
+            },
+            "configs/params_dev.yaml:run_mode": "dev",
+            "configs/params_dev.yaml:config_file": "configs/params_dev.yaml",
+            "configs/params_dev.yaml:data_load": {
+                "dataset": "development",
+                "sampling": {
+                "enable": true,
+                "size": 1000
+                }
+            },
+            "configs/params_dev.yaml:train": {
+                "epochs": 10
+            }
+        }
+
+        ---
+
+        - Git URL as `repo`.
+
+        >>> import json
+        >>> import dvc.api
+        >>> params = dvc.api.params_show(
+        ...     repo="https://github.com/iterative/demo-fashion-mnist")
+        {
+            "train": {
+                "batch_size": 128,
+                "hidden_units": 64,
+                "dropout": 0.4,
+                "num_epochs": 10,
+                "lr": 0.001,
+                "conv_activation": "relu"
+            }
+        }
+
+
+    .. _Git revision:
+        https://git-scm.com/docs/revisions
+
+    """
+    if isinstance(stages, str):
+        stages = [stages]
+
+    def _onerror_raise(result: Dict, exception: Exception, *args, **kwargs):
+        raise exception
+
+    def _postprocess(params):
+        processed = {}
+        for rev, rev_data in params.items():
+            processed[rev] = {}
+
+            counts = Counter()
+            for file_data in rev_data["data"].values():
+                for k in file_data["data"]:
+                    counts[k] += 1
+
+            for file_name, file_data in rev_data["data"].items():
+                to_merge = {
+                    (k if counts[k] == 1 else f"{file_name}:{k}"): v
+                    for k, v in file_data["data"].items()
+                }
+                processed[rev] = {**processed[rev], **to_merge}
+
+        if "workspace" in processed:
+            del processed["workspace"]
+
+        if not processed:
+            raise DvcException("No params found")
+
+        return processed[first(processed)]
+
+    with Repo.open(repo) as _repo:
+        params = _repo.params.show(
+            revs=rev if rev is None else [rev],
+            targets=targets,
+            deps=deps,
+            onerror=_onerror_raise,
+            stages=stages,
+        )
+
+    return _postprocess(params)

dvc/cli/__init__.py

@@ -107,7 +107,7 @@ def _log_exceptions(exc: Exception) -> Optional[int]:
         )
         return 251
 
-    from dvc_objects.cache import DiskError
+    from dvc_data.hashfile.cache import DiskError
 
     if isinstance(exc, DiskError):
         from dvc.utils import relpath
@@ -121,7 +121,7 @@ def _log_exceptions(exc: Exception) -> Optional[int]:
         )
         return None
 
-    from dvc_data.stage import IgnoreInCollectedDirError
+    from dvc_data.hashfile.build import IgnoreInCollectedDirError
 
     if isinstance(exc, IgnoreInCollectedDirError):
         logger.exception("")
@@ -189,6 +189,15 @@ def main(argv=None):  # noqa: C901
     except KeyboardInterrupt:
         logger.exception("interrupted by the user")
         ret = 252
+    except BrokenPipeError:
+        import os
+
+        # Python flushes standard streams on exit; redirect remaining output
+        # to devnull to avoid another BrokenPipeError at shutdown
+        # See: https://docs.python.org/3/library/signal.html#note-on-sigpipe
+        devnull = os.open(os.devnull, os.O_WRONLY)
+        os.dup2(devnull, sys.stdout.fileno())
+        ret = 141  # 128 + 13 (SIGPIPE)
     except NotDvcRepoError:
         logger.exception("")
         ret = 253

dvc/cli/actions.py

@@ -0,0 +1,20 @@
+from argparse import _AppendAction
+
+
+class CommaSeparatedArgs(_AppendAction):  # pylint: disable=protected-access
+    def __call__(self, parser, namespace, values, option_string=None):
+        from funcy import ldistinct
+
+        items = getattr(namespace, self.dest) or []
+        items.extend(map(str.strip, values.split(",")))
+        setattr(namespace, self.dest, ldistinct(items))
+
+
+class KeyValueArgs(_AppendAction):
+    def __call__(self, parser, namespace, values, option_string=None):
+        items = getattr(namespace, self.dest) or {}
+        for value in filter(bool, values):
+            key, _, value = value.partition("=")
+            items[key.strip()] = value
+
+        setattr(namespace, self.dest, items)

dvc/cli/command.py

@@ -13,7 +13,7 @@ class CmdBase(ABC):
 
         os.chdir(args.cd)
 
-        self.repo = Repo(uninitialized=self.UNINITIALIZED)
+        self.repo: "Repo" = Repo(uninitialized=self.UNINITIALIZED)
         self.config = self.repo.config
         self.args = args
 

dvc/cli/parser.py

@@ -14,6 +14,7 @@ from dvc.commands import (
     config,
     daemon,
     dag,
+    data,
     data_sync,
     destroy,
     diff,
@@ -33,6 +34,7 @@ from dvc.commands import (
     move,
     params,
     plots,
+    queue,
     remote,
     remove,
     repro,
@@ -50,6 +52,7 @@ logger = logging.getLogger(__name__)
 
 COMMANDS = [
     init,
+    queue,
     get,
     get_url,
     destroy,
@@ -86,6 +89,7 @@ COMMANDS = [
     experiments,
     check_ignore,
     machine,
+    data,
 ]
 
 

dvc/commands/add.py

@@ -2,12 +2,43 @@ import argparse
 import logging
 
 from dvc.cli import completion
+from dvc.cli.actions import KeyValueArgs
 from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 
 
+def _add_annotating_args(parser: argparse.ArgumentParser) -> None:
+    parser.add_argument(
+        "--desc",
+        type=str,
+        metavar="<text>",
+        help="User description of the data.",
+    )
+    parser.add_argument(
+        "--meta",
+        metavar="key=value",
+        nargs=1,
+        action=KeyValueArgs,
+        help="Custom metadata to add to the data.",
+    )
+    parser.add_argument(
+        "--label",
+        dest="labels",
+        type=str,
+        action="append",
+        metavar="<str>",
+        help="Label for the data.",
+    )
+    parser.add_argument(
+        "--type",
+        type=str,
+        metavar="<str>",
+        help="Type of the data.",
+    )
+
+
 class CmdAdd(CmdBase):
     def run(self):
         from dvc.exceptions import (
@@ -28,6 +59,9 @@ class CmdAdd(CmdBase):
                 glob=self.args.glob,
                 desc=self.args.desc,
                 out=self.args.out,
+                type=self.args.type,
+                labels=self.args.labels,
+                meta=self.args.meta,
                 remote=self.args.remote,
                 to_remote=self.args.to_remote,
                 jobs=self.args.jobs,
@@ -109,15 +143,8 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<number>",
     )
-    parser.add_argument(
-        "--desc",
-        type=str,
-        metavar="<text>",
-        help=(
-            "User description of the data (optional). "
-            "This doesn't affect any DVC operations."
-        ),
-    )
+
+    _add_annotating_args(parser)
     parser.add_argument(
         "targets", nargs="+", help="Input files/directories to add."
     ).complete = completion.FILE

dvc/commands/dag.py

@@ -1,11 +1,15 @@
 import argparse
+from typing import TYPE_CHECKING
 
 from dvc.cli.command import CmdBase
 from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
+if TYPE_CHECKING:
+    from networkx import DiGraph
 
-def _show_ascii(G):
+
+def _show_ascii(G: "DiGraph"):
     from dvc.dagascii import draw
     from dvc.repo.graph import get_pipelines
 
@@ -18,12 +22,24 @@ def _show_ascii(G):
     return "\n".join(ret)
 
 
-def _show_dot(G):
+def _quote_label(node):
+    label = str(node)
+    # Node names should not contain ":" unless they are quoted with "".
+    # See: https://github.com/pydot/pydot/issues/258.
+    if label[0] != '"' and label[-1] != '"':
+        return f'"{label}"'
+    return label
+
+
+def _show_dot(G: "DiGraph"):
     import io
 
+    import networkx as nx
     from networkx.drawing.nx_pydot import write_dot
 
     dot_file = io.StringIO()
+
+    nx.relabel_nodes(G, _quote_label, copy=False)
     write_dot(G.reverse(), dot_file)
     return dot_file.getvalue()
 
@@ -70,9 +86,9 @@ def _collect_targets(repo, target, outs):
             targets.extend([str(out) for out in stage.outs])
             continue
 
-        for out in outs_trie.itervalues(
+        for out in outs_trie.itervalues(  # noqa: B301
             prefix=repo.fs.path.parts(path)
-        ):  # noqa: B301
+        ):
             targets.extend(str(out))
 
     return targets

dvc/commands/data.py

@@ -0,0 +1,293 @@
+import argparse
+import logging
+from operator import itemgetter
+from typing import TYPE_CHECKING, Any, Dict, Iterable, Set
+
+from funcy import chunks, compact, log_durations
+
+from dvc.cli import completion
+from dvc.cli.actions import CommaSeparatedArgs
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.ui import ui
+from dvc.utils import colorize
+
+if TYPE_CHECKING:
+    from dvc.repo.data import Status as DataStatus
+
+
+logger = logging.getLogger(__name__)
+
+
+class CmdDataStatus(CmdBase):
+    COLORS = {
+        "not_in_cache": "red",
+        "committed": "green",
+        "uncommitted": "yellow",
+        "untracked": "cyan",
+    }
+    LABELS = {
+        "not_in_cache": "Not in cache",
+        "committed": "DVC committed changes",
+        "uncommitted": "DVC uncommitted changes",
+        "untracked": "Untracked files",
+        "unchanged": "DVC unchanged files",
+    }
+    HINTS = {
+        "not_in_cache": ('use "dvc fetch <file>..." to download files',),
+        "committed": (
+            "git commit the corresponding dvc files to update the repo",
+        ),
+        "uncommitted": (
+            'use "dvc commit <file>..." to track changes',
+            'use "dvc checkout <file>..." to discard changes',
+        ),
+        "untracked": (
+            'use "git add <file> ..." or '
+            '"dvc add <file>..." to commit to git or to dvc',
+        ),
+        "git_dirty": (
+            "there are {}changes not tracked by dvc, "
+            'use "git status" to see',
+        ),
+    }
+
+    @staticmethod
+    def _process_status(status: "DataStatus"):
+        """Flatten stage status, and filter empty stage status contents."""
+        for stage, stage_status in status.items():
+            items = stage_status
+            if isinstance(stage_status, dict):
+                items = {
+                    file: state
+                    for state, files in stage_status.items()
+                    for file in files
+                }
+            if not items:
+                continue
+            yield stage, items
+
+    @classmethod
+    def _show_status(cls, status: "DataStatus") -> int:
+        git_info = status.pop("git")  # type: ignore[misc]
+        result = dict(cls._process_status(status))
+        if not result:
+            no_changes = "No changes"
+            if git_info.get("is_empty", False):
+                no_changes += " in an empty git repo"
+            ui.write(f"{no_changes}.")
+
+        for idx, (stage, stage_status) in enumerate(result.items()):
+            if idx:
+                ui.write()
+
+            label = cls.LABELS.get(stage, stage.capitalize() + " files")
+            header = f"{label}:"
+            color = cls.COLORS.get(stage, None)
+
+            ui.write(header)
+            if hints := cls.HINTS.get(stage):
+                for hint in hints:
+                    ui.write(f"  ({hint})")
+
+            if isinstance(stage_status, dict):
+                items = [
+                    ": ".join([state, file])
+                    for file, state in stage_status.items()
+                ]
+            else:
+                items = stage_status
+
+            tabs = "\t".expandtabs(8)
+            for chunk in chunks(1000, items):
+                out = "\n".join(tabs + item for item in chunk)
+                ui.write(colorize(out, color))
+
+        if (hints := cls.HINTS.get("git_dirty")) and git_info.get("is_dirty"):
+            for hint in hints:
+                message = hint.format("other " if result else "")
+                ui.write(f"[blue]({message})[/]", styled=True)
+        return 0
+
+    def run(self) -> int:
+        with log_durations(logger.trace, "in data_status"):  # type: ignore
+            status = self.repo.data_status(
+                granular=self.args.granular,
+                untracked_files=self.args.untracked_files,
+            )
+
+        if not self.args.unchanged:
+            status.pop("unchanged")  # type: ignore[misc]
+        if self.args.untracked_files == "no":
+            status.pop("untracked")
+        if self.args.json:
+            status.pop("git")  # type: ignore[misc]
+            ui.write_json(compact(status))
+            return 0
+        return self._show_status(status)
+
+
+class CmdDataLs(CmdBase):
+    @staticmethod
+    def _show_table(
+        d: Iterable[Dict[str, Any]],
+        filter_types: Set[str],
+        filter_labels: Set[str],
+        markdown: bool = False,
+    ) -> None:
+        from rich.style import Style
+
+        from dvc.compare import TabularData
+
+        td = TabularData(
+            columns=["Path", "Type", "Labels", "Description"], fill_value="-"
+        )
+        for entry in sorted(d, key=itemgetter("path")):
+            typ = entry.get("type", "")
+            desc = entry.get("desc", "-")
+            labels = entry.get("labels", [])
+
+            if filter_types and typ not in filter_types:
+                continue
+            if filter_labels and filter_labels.isdisjoint(labels):
+                continue
+
+            rich_label = ui.rich_text()
+            for index, label in enumerate(labels):
+                if index:
+                    rich_label.append(",")
+                style = Style(bold=label in filter_labels, color="green")
+                rich_label.append(label, style=style)
+
+            if markdown and desc:
+                desc = desc.partition("\n")[0]
+
+            path = ui.rich_text(entry["path"], style="cyan")
+            type_style = Style(bold=typ in filter_types, color="yellow")
+            typ = ui.rich_text(entry.get("type", ""), style=type_style)
+            td.append([path, typ or "-", rich_label or "-", desc])
+
+        td.render(markdown=markdown, rich_table=True)
+
+    def run(self):
+        from dvc.repo.data import ls
+
+        filter_labels = set(self.args.labels)
+        filter_types = set(self.args.type)
+        d = ls(
+            self.repo, targets=self.args.targets, recursive=self.args.recursive
+        )
+        self._show_table(
+            d,
+            filter_labels=filter_labels,
+            filter_types=filter_types,
+            markdown=self.args.markdown,
+        )
+        return 0
+
+
+def add_parser(subparsers, parent_parser):
+    data_parser = subparsers.add_parser(
+        "data",
+        parents=[parent_parser],
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    data_subparsers = data_parser.add_subparsers(
+        dest="cmd",
+        help="Use `dvc data CMD --help` to display command-specific help.",
+    )
+    fix_subparsers(data_subparsers)
+
+    DATA_STATUS_HELP = (
+        "Show changes between the last git commit, "
+        "the dvcfiles and the workspace."
+    )
+    data_status_parser = data_subparsers.add_parser(
+        "status",
+        parents=[parent_parser],
+        description=append_doc_link(DATA_STATUS_HELP, "data/status"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        help=DATA_STATUS_HELP,
+    )
+    data_status_parser.add_argument(
+        "--json",
+        action="store_true",
+        default=False,
+        help="Show output in JSON format.",
+    )
+    data_status_parser.add_argument(
+        "--show-json",
+        action="store_true",
+        default=False,
+        dest="json",
+        help=argparse.SUPPRESS,
+    )
+    data_status_parser.add_argument(
+        "--granular",
+        action="store_true",
+        default=False,
+        help="Show granular file-level info for DVC-tracked directories.",
+    )
+    data_status_parser.add_argument(
+        "--unchanged",
+        action="store_true",
+        default=False,
+        help="Show unmodified DVC-tracked files.",
+    )
+    data_status_parser.add_argument(
+        "--untracked-files",
+        choices=["no", "all"],
+        default="no",
+        const="all",
+        nargs="?",
+        help="Show untracked files.",
+    )
+    data_status_parser.set_defaults(func=CmdDataStatus)
+
+    DATA_LS_HELP = "List data tracked by DVC with its metadata."
+    data_ls_parser = data_subparsers.add_parser(
+        "ls",
+        aliases=["list"],
+        parents=[parent_parser],
+        description=append_doc_link(DATA_LS_HELP, "data/ls"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        help=DATA_LS_HELP,
+    )
+    data_ls_parser.add_argument(
+        "--md",
+        "--show-md",
+        dest="markdown",
+        action="store_true",
+        default=False,
+        help="Show tabulated output in the Markdown format (GFM).",
+    )
+    data_ls_parser.add_argument(
+        "--type",
+        action=CommaSeparatedArgs,
+        default=[],
+        help="Comma-separated list of type to filter.",
+    )
+    data_ls_parser.add_argument(
+        "--labels",
+        action=CommaSeparatedArgs,
+        default=[],
+        help="Comma-separated list of labels to filter.",
+    )
+    data_ls_parser.add_argument(
+        "-R",
+        "--recursive",
+        action="store_true",
+        default=False,
+        help="Recursively list from the specified directories.",
+    )
+    data_ls_parser.add_argument(
+        "targets",
+        default=None,
+        nargs="*",
+        help=(
+            "Limit command scope to these tracked files/directories, "
+            ".dvc files, or stage names."
+        ),
+    ).complete = completion.DVCFILES_AND_STAGE
+
+    data_ls_parser.set_defaults(func=CmdDataLs)

dvc/commands/data_sync.py

@@ -43,7 +43,8 @@ class CmdDataPull(CmdDataBase):
             )
             self.log_summary(stats)
         except (CheckoutError, DvcException) as exc:
-            self.log_summary(getattr(exc, "stats", {}))
+            if stats := getattr(exc, "stats", {}):
+                self.log_summary(stats)
             logger.exception("failed to pull data from the cloud")
             return 1
 

dvc/commands/experiments/__init__.py

@@ -15,6 +15,7 @@ from dvc.commands.experiments import (
     ls,
     pull,
     push,
+    queue_worker,
     remove,
     run,
     show,
@@ -30,6 +31,7 @@ SUB_COMMANDS = [
     ls,
     pull,
     push,
+    queue_worker,
     remove,
     run,
     show,

dvc/commands/experiments/branch.py

@@ -29,6 +29,6 @@ def add_parser(experiments_subparsers, parent_parser):
         "experiment", help="Experiment to be promoted."
     )
     experiments_branch_parser.add_argument(
-        "branch", help="Git branch name to use."
+        "branch", help="Name for the new Git branch"
     )
     experiments_branch_parser.set_defaults(func=CmdExperimentsBranch)

dvc/commands/experiments/diff.py

@@ -27,7 +27,7 @@ class CmdExperimentsDiff(CmdBase):
 
         if self.args.json:
             ui.write_json(diff)
-        else:
+        elif diff:
             from dvc.compare import show_diff
 
             precision = self.args.precision or DEFAULT_PRECISION

dvc/commands/experiments/gc.py

@@ -12,8 +12,8 @@ logger = logging.getLogger(__name__)
 def _raise_error_if_all_disabled(**kwargs):
     if not any(kwargs.values()):
         raise InvalidArgumentError(
-            "Either of `-w|--workspace`, `-a|--all-branches`, `-T|--all-tags` "
-            "or `--all-commits` needs to be set."
+            "Either of `-w|--workspace`, `-a|--all-branches`, `-T|--all-tags`,"
+            " `--all-commits` or `--date` needs to be set."
         )
 
 
@@ -23,6 +23,7 @@ class CmdExperimentsGC(CmdBase):
             all_branches=self.args.all_branches,
             all_tags=self.args.all_tags,
             all_commits=self.args.all_commits,
+            commit_date=self.args.commit_date,
             workspace=self.args.workspace,
         )
 
@@ -31,13 +32,21 @@ class CmdExperimentsGC(CmdBase):
         msg += "the workspace"
         if self.args.all_commits:
             msg += " and all git commits"
-        elif self.args.all_branches and self.args.all_tags:
-            msg += " and all git branches and tags"
-        elif self.args.all_branches:
-            msg += " and all git branches"
-        elif self.args.all_tags:
-            msg += " and all git tags"
+        else:
+            if self.args.all_branches and self.args.all_tags:
+                msg += " and all git branches and tags"
+            elif self.args.all_branches:
+                msg += " and all git branches"
+            elif self.args.all_tags:
+                msg += " and all git tags"
+
+            if self.args.commit_date:
+                msg += (
+                    " and all git commits before date "
+                    f"{self.args.commit_date}"
+                )
         msg += " of the current repo."
+
         if self.args.queued:
             msg += " Run queued experiments will be preserved."
         else:
@@ -54,6 +63,7 @@ class CmdExperimentsGC(CmdBase):
             all_tags=self.args.all_tags,
             all_commits=self.args.all_commits,
             workspace=self.args.workspace,
+            commit_date=self.args.commit_date,
             queued=self.args.queued,
         )
 
@@ -110,6 +120,17 @@ def add_parser(experiments_subparsers, parent_parser):
         default=False,
         help="Keep experiments derived from all Git commits.",
     )
+    experiments_gc_parser.add_argument(
+        "--date",
+        type=str,
+        dest="commit_date",
+        metavar="<YYYY-MM-DD>",
+        default=None,
+        help=(
+            "Keep experiments from the commits after (inclusive) a certain "
+            "date. Date must match the extended ISO 8601 format (YYYY-MM-DD)."
+        ),
+    )
     experiments_gc_parser.add_argument(
         "--queued",
         action="store_true",

dvc/commands/experiments/queue_worker.py

@@ -0,0 +1,25 @@
+import logging
+
+from dvc.cli.command import CmdBase
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueWorker(CmdBase):
+    """Run the exp queue worker."""
+
+    def run(self):
+        self.repo.experiments.celery_queue.worker.start(self.args.name)
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    QUEUE_WORKER_HELP = "Run the exp queue worker."
+    parser = experiments_subparsers.add_parser(
+        "queue-worker",
+        parents=[parent_parser],
+        description=QUEUE_WORKER_HELP,
+        add_help=False,
+    )
+    parser.add_argument("name", help="Celery worker name.")
+    parser.set_defaults(func=CmdQueueWorker)

dvc/commands/experiments/remove.py

@@ -10,23 +10,22 @@ logger = logging.getLogger(__name__)
 
 
 class CmdExperimentsRemove(CmdBase):
-    def raise_error_if_all_disabled(self):
+    def check_arguments(self):
         if not any(
             [
-                self.args.experiment,
                 self.args.all_commits,
                 self.args.rev,
                 self.args.queue,
             ]
-        ):
+        ) ^ bool(self.args.experiment):
             raise InvalidArgumentError(
                 "Either provide an `experiment` argument, or use the "
-                "`--rev` or `--all-commits` flag."
+                "`--rev` or `--all-commits` or `--queue` flag."
             )
 
     def run(self):
 
-        self.raise_error_if_all_disabled()
+        self.check_arguments()
 
         removed_list = self.repo.experiments.remove(
             exp_names=self.args.experiment,

dvc/commands/experiments/show.py

@@ -138,7 +138,7 @@ def _collect_rows(
         row_dict["Experiment"] = exp.get("name", "")
         row_dict["rev"] = name_rev
         row_dict["typ"] = typ
-        row_dict["Created"] = _format_time(
+        row_dict["Created"] = format_time(
             exp.get("timestamp"), fill_value, iso
         )
         row_dict["parent"] = parent
@@ -218,7 +218,7 @@ def _sort_exp(experiments, sort_path, sort_name, typ, reverse):
     return ret
 
 
-def _format_time(datetime_obj, fill_value=FILL_VALUE, iso=False):
+def format_time(datetime_obj, fill_value=FILL_VALUE, iso=False):
     if datetime_obj is None:
         return fill_value
 

dvc/commands/freeze.py

@@ -16,7 +16,7 @@ class CmdFreezeBase(CmdBase):
             try:
                 func(target)
             except DvcException:
-                logger.exception(f"failed to {name} '{target}'")
+                logger.exception("failed to %s '%s'", name, target)
                 ret = 1
         return ret
 

dvc/commands/gc.py

@@ -11,29 +11,46 @@ logger = logging.getLogger(__name__)
 
 class CmdGC(CmdBase):
     def run(self):
-        from dvc.repo.gc import _raise_error_if_all_disabled
+        from dvc.repo.gc import _validate_args
 
-        _raise_error_if_all_disabled(
+        _validate_args(
             all_branches=self.args.all_branches,
             all_tags=self.args.all_tags,
             all_commits=self.args.all_commits,
+            all_experiments=self.args.all_experiments,
+            commit_date=self.args.commit_date,
             workspace=self.args.workspace,
+            rev=self.args.rev,
+            num=self.args.num,
         )
 
+        if self.args.rev:
+            self.args.num = self.args.num or 1
+
         msg = "This will remove all cache except items used in "
 
         msg += "the workspace"
         if self.args.all_commits:
             msg += " and all git commits"
-        elif self.args.all_branches and self.args.all_tags:
-            msg += " and all git branches and tags"
-        elif self.args.all_branches:
-            msg += " and all git branches"
-        elif self.args.all_tags:
-            msg += " and all git tags"
-        elif self.args.all_experiments:
-            msg += " and all experiments"
+        else:
+            if self.args.all_branches and self.args.all_tags:
+                msg += " and all git branches and tags"
+            elif self.args.all_branches:
+                msg += " and all git branches"
+            elif self.args.all_tags:
+                msg += " and all git tags"
+            if self.args.commit_date:
+                msg += (
+                    " and all git commits before date "
+                    f"{self.args.commit_date}"
+                )
+            if self.args.rev:
+                msg += (
+                    f" and last {self.args.num} commits from {self.args.rev}"
+                )
 
+        if self.args.all_experiments:
+            msg += " and all experiments"
         if self.args.repos:
             msg += " of the current and the following repos:"
 
@@ -53,12 +70,15 @@ class CmdGC(CmdBase):
             all_tags=self.args.all_tags,
             all_commits=self.args.all_commits,
             all_experiments=self.args.all_experiments,
+            commit_date=self.args.commit_date,
             cloud=self.args.cloud,
             remote=self.args.remote,
             force=self.args.force,
             jobs=self.args.jobs,
             repos=self.args.repos,
             workspace=self.args.workspace,
+            rev=self.args.rev,
+            num=self.args.num,
         )
         return 0
 
@@ -83,6 +103,26 @@ def add_parser(subparsers, parent_parser):
         default=False,
         help="Keep data files used in the current workspace.",
     )
+    gc_parser.add_argument(
+        "--rev",
+        type=str,
+        default=None,
+        help="Keep data files used in the specified <commit>.",
+        metavar="<commit>",
+    )
+    gc_parser.add_argument(
+        "-n",
+        "--num",
+        type=int,
+        dest="num",
+        metavar="<num>",
+        help=(
+            "Keep data files used in the last `num` commits "
+            "starting from the `--rev` <commit>. "
+            "Only used if `--rev` is also provided. "
+            "Defaults to `1`."
+        ),
+    )
     gc_parser.add_argument(
         "-a",
         "--all-branches",
@@ -104,6 +144,18 @@ def add_parser(subparsers, parent_parser):
         default=False,
         help="Keep data files for all Git commits.",
     )
+    gc_parser.add_argument(
+        "--date",
+        type=str,
+        dest="commit_date",
+        metavar="<YYYY-MM-DD>",
+        default=None,
+        help=(
+            "Keep cached data referenced in the commits after ( inclusive )"
+            " a certain time. Date must match the extended ISO 8601 format "
+            "(YYYY-MM-DD)."
+        ),
+    )
     gc_parser.add_argument(
         "--all-experiments",
         action="store_true",

dvc/commands/get_url.py

@@ -18,7 +18,7 @@ class CmdGetUrl(CmdBaseNoRepo):
             Repo.get_url(self.args.url, out=self.args.out, jobs=self.args.jobs)
             return 0
         except DvcException:
-            logger.exception(f"failed to get '{self.args.url}'")
+            logger.exception("failed to get '%s'", self.args.url)
             return 1
 
 

dvc/commands/imp.py

@@ -19,7 +19,11 @@ class CmdImport(CmdBase):
                 fname=self.args.file,
                 rev=self.args.rev,
                 no_exec=self.args.no_exec,
+                no_download=self.args.no_download,
                 desc=self.args.desc,
+                type=self.args.type,
+                labels=self.args.labels,
+                meta=self.args.meta,
                 jobs=self.args.jobs,
             )
         except DvcException:
@@ -33,6 +37,8 @@ class CmdImport(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
+    from .add import _add_annotating_args
+
     IMPORT_HELP = (
         "Download file or directory tracked by DVC or by Git "
         "into the workspace, and track it."
@@ -69,20 +75,19 @@ def add_parser(subparsers, parent_parser):
         help="Specify name of the .dvc file this command will generate.",
         metavar="<filename>",
     )
-    import_parser.add_argument(
+    no_download_exec_group = import_parser.add_mutually_exclusive_group()
+    no_download_exec_group.add_argument(
         "--no-exec",
         action="store_true",
         default=False,
-        help="Only create .dvc file without actually downloading it.",
+        help="Only create .dvc file without actually importing target data.",
     )
-    import_parser.add_argument(
-        "--desc",
-        type=str,
-        metavar="<text>",
-        help=(
-            "User description of the data (optional). "
-            "This doesn't affect any DVC operations."
-        ),
+    no_download_exec_group.add_argument(
+        "--no-download",
+        action="store_true",
+        default=False,
+        help="Create .dvc file including target data hash value(s)"
+        " but do not actually download the file(s).",
     )
     import_parser.add_argument(
         "-j",
@@ -94,4 +99,6 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<number>",
     )
+
+    _add_annotating_args(import_parser)
     import_parser.set_defaults(func=CmdImport)

dvc/commands/imp_url.py

@@ -17,9 +17,13 @@ class CmdImportUrl(CmdBase):
                 out=self.args.out,
                 fname=self.args.file,
                 no_exec=self.args.no_exec,
+                no_download=self.args.no_download,
                 remote=self.args.remote,
                 to_remote=self.args.to_remote,
                 desc=self.args.desc,
+                type=self.args.type,
+                labels=self.args.labels,
+                meta=self.args.meta,
                 jobs=self.args.jobs,
             )
         except DvcException:
@@ -34,6 +38,8 @@ class CmdImportUrl(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
+    from .add import _add_annotating_args
+
     IMPORT_HELP = (
         "Download or copy file from URL and take it under DVC control."
     )
@@ -66,11 +72,19 @@ def add_parser(subparsers, parent_parser):
         help="Specify name of the .dvc file this command will generate.",
         metavar="<filename>",
     ).complete = completion.DIR
-    import_parser.add_argument(
+    no_download_exec_group = import_parser.add_mutually_exclusive_group()
+    no_download_exec_group.add_argument(
         "--no-exec",
         action="store_true",
         default=False,
-        help="Only create .dvc file without actually downloading it.",
+        help="Only create .dvc file without actually importing target data.",
+    )
+    no_download_exec_group.add_argument(
+        "--no-download",
+        action="store_true",
+        default=False,
+        help="Create .dvc file including target data hash value(s)"
+        " but do not actually download the file(s).",
     )
     import_parser.add_argument(
         "--to-remote",
@@ -94,13 +108,5 @@ def add_parser(subparsers, parent_parser):
         ),
         metavar="<number>",
     )
-    import_parser.add_argument(
-        "--desc",
-        type=str,
-        metavar="<text>",
-        help=(
-            "User description of the data (optional). "
-            "This doesn't affect any DVC operations."
-        ),
-    )
+    _add_annotating_args(import_parser)
     import_parser.set_defaults(func=CmdImportUrl)

dvc/commands/ls/__init__.py

@@ -42,7 +42,7 @@ class CmdList(CmdBaseNoRepo):
                 ui.write("\n".join(entries))
             return 0
         except DvcException:
-            logger.exception(f"failed to list '{self.args.url}'")
+            logger.exception("failed to list '%s'", self.args.url)
             return 1
 
 

dvc/commands/plots.py

@@ -24,6 +24,60 @@ def _show_json(renderers, split=False):
     ui.write_json(result)
 
 
+def _adjust_vega_renderers(renderers):
+    from dvc.render import VERSION_FIELD
+    from dvc_render import VegaRenderer
+
+    for r in renderers:
+        if isinstance(r, VegaRenderer):
+            if _data_versions_count(r) > 1:
+                summary = _summarize_version_infos(r)
+                for dp in r.datapoints:
+                    vi = dp.pop(VERSION_FIELD, {})
+                    keys = list(vi.keys())
+                    for key in keys:
+                        if not (len(summary.get(key, set())) > 1):
+                            vi.pop(key)
+                    if vi:
+                        dp["rev"] = "::".join(vi.values())
+            else:
+                for dp in r.datapoints:
+                    dp.pop(VERSION_FIELD, {})
+
+
+def _summarize_version_infos(renderer):
+    from collections import defaultdict
+
+    from dvc.render import VERSION_FIELD
+
+    result = defaultdict(set)
+
+    for dp in renderer.datapoints:
+        for key, value in dp.get(VERSION_FIELD, {}).items():
+            result[key].add(value)
+    return dict(result)
+
+
+def _data_versions_count(renderer):
+    from itertools import product
+
+    summary = _summarize_version_infos(renderer)
+    x = product(summary.get("filename", {None}), summary.get("field", {None}))
+    return len(set(x))
+
+
+def _filter_unhandled_renderers(renderers):
+    # filtering out renderers currently unhandled by vscode extension
+    from dvc_render import VegaRenderer
+
+    def _is_json_viable(r):
+        return not (
+            isinstance(r, VegaRenderer) and _data_versions_count(r) > 1
+        )
+
+    return list(filter(_is_json_viable, renderers))
+
+
 class CmdPlots(CmdBase):
     def _func(self, *args, **kwargs):
         raise NotImplementedError
@@ -35,10 +89,28 @@ class CmdPlots(CmdBase):
         props = {p: getattr(self.args, p) for p in PLOT_PROPS}
         return {k: v for k, v in props.items() if v is not None}
 
+    def _config_files(self):
+        config_files = None
+        if self.args.from_config:
+            config_files = {self.args.from_config}
+        return config_files
+
+    def _html_template_path(self):
+        html_template_path = self.args.html_template
+        if not html_template_path:
+            html_template_path = self.repo.config.get("plots", {}).get(
+                "html_template", None
+            )
+            if html_template_path and not os.path.isabs(html_template_path):
+                html_template_path = os.path.join(
+                    self.repo.dvc_dir, html_template_path
+                )
+        return html_template_path
+
     def run(self):
         from pathlib import Path
 
-        from dvc.render.match import match_renderers
+        from dvc.render.match import match_defs_renderers
         from dvc_render import render_html
 
         if self.args.show_vega:
@@ -58,9 +130,10 @@ class CmdPlots(CmdBase):
                 return 1
 
         try:
-
             plots_data = self._func(
-                targets=self.args.targets, props=self._props()
+                targets=self.args.targets,
+                props=self._props(),
+                config_files=self._config_files(),
             )
 
             if not plots_data:
@@ -69,56 +142,51 @@ class CmdPlots(CmdBase):
                     "visualization file will not be created."
                 )
 
-            out: str = self.args.out or "dvc_plots"
+            out: str = self.args.out or self.repo.config.get("plots", {}).get(
+                "out_dir", "dvc_plots"
+            )
 
             renderers_out = (
                 out if self.args.json else os.path.join(out, "static")
             )
-            renderers = match_renderers(
-                plots_data=plots_data,
+
+            renderers = match_defs_renderers(
+                data=plots_data,
                 out=renderers_out,
                 templates_dir=self.repo.plots.templates_dir,
             )
+            if self.args.json:
+                renderers = _filter_unhandled_renderers(renderers)
+                _show_json(renderers, self.args.split)
+                return 0
+
+            _adjust_vega_renderers(renderers)
 
             if self.args.show_vega:
                 renderer = first(filter(lambda r: r.TYPE == "vega", renderers))
                 if renderer:
                     ui.write_json(json.loads(renderer.get_filled_template()))
                 return 0
-            if self.args.json:
-                _show_json(renderers, self.args.split)
-                return 0
-
-            html_template_path = self.args.html_template
-            if not html_template_path:
-                html_template_path = self.repo.config.get("plots", {}).get(
-                    "html_template", None
-                )
-                if html_template_path and not os.path.isabs(
-                    html_template_path
-                ):
-                    html_template_path = os.path.join(
-                        self.repo.dvc_dir, html_template_path
-                    )
 
             output_file: Path = (Path.cwd() / out).resolve() / "index.html"
 
-            render_html(
-                renderers=renderers,
-                output_file=output_file,
-                template_path=html_template_path,
-            )
+            if renderers:
+                render_html(
+                    renderers=renderers,
+                    output_file=output_file,
+                    template_path=self._html_template_path(),
+                )
 
-            ui.write(output_file.as_uri())
-            auto_open = self.repo.config["plots"].get("auto_open", False)
-            if self.args.open or auto_open:
-                if not auto_open:
-                    ui.write(
-                        "To enable auto opening, you can run:\n"
-                        "\n"
-                        "\tdvc config plots.auto_open true"
-                    )
-                return ui.open_browser(output_file)
+                ui.write(output_file.as_uri())
+                auto_open = self.repo.config["plots"].get("auto_open", False)
+                if self.args.open or auto_open:
+                    if not auto_open:
+                        ui.write(
+                            "To enable auto opening, you can run:\n"
+                            "\n"
+                            "\tdvc config plots.auto_open true"
+                        )
+                    return ui.open_browser(output_file)
 
             return 0
 
@@ -155,29 +223,22 @@ class CmdPlotsModify(CmdPlots):
 
 
 class CmdPlotsTemplates(CmdBase):
-    TEMPLATES_CHOICES = [
-        "simple",
-        "linear",
-        "confusion",
-        "confusion_normalized",
-        "scatter",
-        "smooth",
-    ]
-
     def run(self):
-        from dvc_render.vega_templates import dump_templates
+        from dvc.exceptions import InvalidArgumentError
+        from dvc_render.vega_templates import TEMPLATES
 
         try:
-            out = (
-                os.path.join(os.getcwd(), self.args.out)
-                if self.args.out
-                else self.repo.plots.templates_dir
-            )
+            target = self.args.template
+            if target:
+                for template in TEMPLATES:
+                    if target == template.DEFAULT_NAME:
+                        ui.write_json(template.DEFAULT_CONTENT)
+                        return 0
+                raise InvalidArgumentError(f"Unexpected template: {target}.")
 
-            targets = [self.args.target] if self.args.target else None
-            dump_templates(output=out, targets=targets)
-            templates_path = os.path.relpath(out, os.getcwd())
-            ui.write(f"Templates have been written into '{templates_path}'.")
+            else:
+                for template in TEMPLATES:
+                    ui.write(template.DEFAULT_NAME)
 
             return 0
         except DvcException:
@@ -186,10 +247,7 @@ class CmdPlotsTemplates(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
-    PLOTS_HELP = (
-        "Commands to visualize and compare plot metrics in structured files "
-        "(JSON, YAML, CSV, TSV)."
-    )
+    PLOTS_HELP = "Commands to visualize and compare plot data."
 
     plots_parser = subparsers.add_parser(
         "plots",
@@ -205,7 +263,10 @@ def add_parser(subparsers, parent_parser):
 
     fix_subparsers(plots_subparsers)
 
-    SHOW_HELP = "Generate plots from metrics files."
+    SHOW_HELP = (
+        "Generate plots from target files or from `plots`"
+        " definitions in `dvc.yaml`."
+    )
     plots_show_parser = plots_subparsers.add_parser(
         "show",
         parents=[parent_parser],
@@ -216,9 +277,10 @@ def add_parser(subparsers, parent_parser):
     plots_show_parser.add_argument(
         "targets",
         nargs="*",
-        help="Files to visualize (supports any file, "
-        "even when not found as `plots` in `dvc.yaml`). "
-        "Shows all plots by default.",
+        help=(
+            "Plots files or plot IDs from `dvc.yaml` to visualize. "
+            "Shows all plots by default."
+        ),
     ).complete = completion.FILE
     _add_props_arguments(plots_show_parser)
     _add_output_argument(plots_show_parser)
@@ -226,8 +288,8 @@ def add_parser(subparsers, parent_parser):
     plots_show_parser.set_defaults(func=CmdPlotsShow)
 
     PLOTS_DIFF_HELP = (
-        "Show multiple versions of plot metrics "
-        "by plotting them in a single image."
+        "Show multiple versions of a plot by overlaying them "
+        "in a single image."
     )
     plots_diff_parser = plots_subparsers.add_parser(
         "diff",
@@ -240,8 +302,8 @@ def add_parser(subparsers, parent_parser):
         "--targets",
         nargs="*",
         help=(
-            "Specific plots file(s) to visualize "
-            "(even if not found as `plots` in `dvc.yaml`). "
+            "Specific plots to visualize. "
+            "Accepts any file path or plot name from `dvc.yaml` file. "
             "Shows all tracked plots by default."
         ),
         metavar="<paths>",
@@ -263,7 +325,7 @@ def add_parser(subparsers, parent_parser):
 
     PLOTS_MODIFY_HELP = (
         "Modify display properties of data-series plots "
-        "(has no effect on image-type plots)."
+        "defined in stages (has no effect on image plots)."
     )
     plots_modify_parser = plots_subparsers.add_parser(
         "modify",
@@ -273,7 +335,8 @@ def add_parser(subparsers, parent_parser):
         formatter_class=argparse.RawDescriptionHelpFormatter,
     )
     plots_modify_parser.add_argument(
-        "target", help="Metric file to set properties to"
+        "target",
+        help="Plots file to set properties for (defined at the stage level).",
     ).complete = completion.FILE
     _add_props_arguments(plots_modify_parser)
     plots_modify_parser.add_argument(
@@ -285,8 +348,7 @@ def add_parser(subparsers, parent_parser):
     plots_modify_parser.set_defaults(func=CmdPlotsModify)
 
     TEMPLATES_HELP = (
-        "Write built-in plots templates to a directory "
-        "(.dvc/plots by default)."
+        "List built-in plots templates or show JSON specification for one."
     )
     plots_templates_parser = plots_subparsers.add_parser(
         "templates",
@@ -296,13 +358,14 @@ def add_parser(subparsers, parent_parser):
         formatter_class=argparse.RawDescriptionHelpFormatter,
     )
     plots_templates_parser.add_argument(
-        "target",
+        "template",
         default=None,
         nargs="?",
-        choices=CmdPlotsTemplates.TEMPLATES_CHOICES,
-        help="Template to write. Writes all templates by default.",
+        help=(
+            "Template for which to show JSON specification. "
+            "List all template names by default."
+        ),
     )
-    _add_output_argument(plots_templates_parser, typ="templates")
     plots_templates_parser.set_defaults(func=CmdPlotsTemplates)
 
 
@@ -383,3 +446,9 @@ def _add_ui_arguments(parser):
         help="Custom HTML template for VEGA visualization.",
         metavar="<path>",
     )
+    parser.add_argument(
+        "--from-config",
+        default=None,
+        metavar="<path>",
+        help=argparse.SUPPRESS,
+    )

dvc/commands/queue/__init__.py

@@ -0,0 +1,34 @@
+import argparse
+
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands.queue import kill, logs, remove, start, status, stop
+
+SUB_COMMANDS = [
+    start,
+    stop,
+    status,
+    logs,
+    remove,
+    kill,
+]
+
+
+def add_parser(subparsers, parent_parser):
+    QUEUE_HELP = "Commands to manage experiments queue."
+
+    queue_parser = subparsers.add_parser(
+        "queue",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_HELP, "queue"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        help=QUEUE_HELP,
+    )
+
+    queue_subparsers = queue_parser.add_subparsers(
+        dest="cmd",
+        help="Use `dvc queue CMD --help` to display " "command-specific help.",
+    )
+
+    fix_subparsers(queue_subparsers)
+    for cmd in SUB_COMMANDS:
+        cmd.add_parser(queue_subparsers, parent_parser)

dvc/commands/queue/kill.py

@@ -0,0 +1,34 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueKill(CmdBase):
+    """Kill exp task in queue."""
+
+    def run(self):
+        self.repo.experiments.celery_queue.kill(revs=self.args.task)
+
+        return 0
+
+
+def add_parser(queue_subparsers, parent_parser):
+    QUEUE_KILL_HELP = "Kill actively running experiment queue tasks."
+    queue_kill_parser = queue_subparsers.add_parser(
+        "kill",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_KILL_HELP, "queue/kill"),
+        help=QUEUE_KILL_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    queue_kill_parser.add_argument(
+        "task",
+        nargs="*",
+        help="Tasks in queue to kill.",
+        metavar="<task>",
+    )
+    queue_kill_parser.set_defaults(func=CmdQueueKill)

dvc/commands/queue/logs.py

@@ -0,0 +1,56 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueLogs(CmdBase):
+    """Show output logs for a queued experiment."""
+
+    def run(self):
+        self.repo.experiments.celery_queue.logs(
+            rev=self.args.task,
+            encoding=self.args.encoding,
+            follow=self.args.follow,
+        )
+
+        return 0
+
+
+def add_parser(queue_subparsers, parent_parser):
+    QUEUE_LOGS_HELP = (
+        "Show output logs for running and completed experiment queue tasks."
+    )
+    queue_logs_parser = queue_subparsers.add_parser(
+        "logs",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_LOGS_HELP, "queue/logs"),
+        help=QUEUE_LOGS_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    queue_logs_parser.add_argument(
+        "-e",
+        "--encoding",
+        help=(
+            "Text encoding for log output. Defaults to system locale encoding."
+        ),
+        metavar="<encoding>",
+    )
+    queue_logs_parser.add_argument(
+        "-f",
+        "--follow",
+        help=(
+            "Attach to task and follow additional live output. Only "
+            "applicable if the task is still running."
+        ),
+        action="store_true",
+    )
+    queue_logs_parser.add_argument(
+        "task",
+        help="Task to show.",
+        metavar="<task>",
+    )
+    queue_logs_parser.set_defaults(func=CmdQueueLogs)

dvc/commands/queue/remove.py

@@ -0,0 +1,95 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.exceptions import InvalidArgumentError
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueRemove(CmdBase):
+    """Remove exp in queue."""
+
+    def check_arguments(self):
+        clear_flag = any(
+            [
+                self.args.all,
+                self.args.queued,
+                self.args.failed,
+                self.args.success,
+            ]
+        )
+        if not (clear_flag ^ bool(self.args.task)):
+            raise InvalidArgumentError(
+                "Either provide an `tasks` argument, or use the "
+                "`--all`, `--queued`, `--failed`, `--success` flag."
+            )
+
+    def run(self):
+
+        self.check_arguments()
+
+        if self.args.all:
+            self.args.queued = True
+            self.args.failed = True
+            self.args.success = True
+
+        if self.args.queued or self.args.failed or self.args.success:
+            removed_list = self.repo.experiments.celery_queue.clear(
+                success=self.args.success,
+                queued=self.args.queued,
+                failed=self.args.failed,
+            )
+        else:
+            removed_list = self.repo.experiments.celery_queue.remove(
+                revs=self.args.task,
+            )
+
+        if removed_list:
+            removed = ", ".join(removed_list)
+            ui.write(f"Removed tasks in queue: {removed}")
+        else:
+            ui.write(f"No tasks found named {self.args.task}")
+
+        return 0
+
+
+def add_parser(queue_subparsers, parent_parser):
+
+    QUEUE_REMOVE_HELP = "Remove queued and completed tasks from the queue."
+    queue_remove_parser = queue_subparsers.add_parser(
+        "remove",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_REMOVE_HELP, "queue/remove"),
+        help=QUEUE_REMOVE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    queue_remove_parser.add_argument(
+        "--all",
+        action="store_true",
+        help="Remove all queued and completed tasks from the queue.",
+    )
+    queue_remove_parser.add_argument(
+        "--queued",
+        action="store_true",
+        help="Remove all queued tasks from the queue.",
+    )
+    queue_remove_parser.add_argument(
+        "--success",
+        action="store_true",
+        help="Remove all successful tasks from the queue.",
+    )
+    queue_remove_parser.add_argument(
+        "--failed",
+        action="store_true",
+        help="Remove all failed tasks from the queue.",
+    )
+    queue_remove_parser.add_argument(
+        "task",
+        nargs="*",
+        help="Tasks to remove.",
+        metavar="<task>",
+    )
+    queue_remove_parser.set_defaults(func=CmdQueueRemove)

dvc/commands/queue/start.py

@@ -0,0 +1,48 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueStart(CmdBase):
+    """Start exp queue workers."""
+
+    def run(self):
+        started = self.repo.experiments.celery_queue.start_workers(
+            self.args.jobs
+        )
+
+        suffix = "s" if started > 1 else ""
+        ui.write(
+            f"Started '{started}' new experiments task queue worker{suffix}."
+        )
+
+        return 0
+
+
+def add_parser(queue_subparsers, parent_parser):
+
+    QUEUE_START_HELP = "Start the experiments task queue worker."
+    queue_start_parser = queue_subparsers.add_parser(
+        "start",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_START_HELP, "queue/start"),
+        help=QUEUE_START_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    queue_start_parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        default=1,
+        help=(
+            "Maximum number of concurrent queue workers to start. "
+            "Defaults to 1."
+        ),
+        metavar="<number>",
+    )
+    queue_start_parser.set_defaults(func=CmdQueueStart)

dvc/commands/queue/status.py

@@ -0,0 +1,62 @@
+import argparse
+import logging
+from typing import List, Mapping, Optional
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.compare import TabularData
+from dvc.ui import ui
+
+from ..experiments.show import format_time
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueStatus(CmdBase):
+    """Show queue task and worker status."""
+
+    def run(self):
+        result: List[
+            Mapping[str, Optional[str]]
+        ] = self.repo.experiments.celery_queue.status()
+        if result:
+            all_headers = ["Task", "Name", "Created", "Status"]
+            td = TabularData(all_headers)
+            for exp in result:
+                created = format_time(exp.get("timestamp"))
+                td.append(
+                    [
+                        exp["rev"][:7],
+                        exp.get("name", ""),
+                        created,
+                        exp["status"],
+                    ]
+                )
+            td.render()
+        else:
+            ui.write("No experiment tasks in the queue.")
+        ui.write()
+
+        worker_status = self.repo.experiments.celery_queue.worker_status()
+        active_count = len(
+            [name for name, task in worker_status.items() if task]
+        )
+        idle_count = len(worker_status) - active_count
+
+        ui.write(f"Worker status: {active_count} active, {idle_count} idle")
+
+        return 0
+
+
+def add_parser(queue_subparsers, parent_parser):
+    QUEUE_STATUS_HELP = (
+        "Show the status of experiments queue tasks and workers."
+    )
+    queue_status_parser = queue_subparsers.add_parser(
+        "status",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_STATUS_HELP, "queue/status"),
+        help=QUEUE_STATUS_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    queue_status_parser.set_defaults(func=CmdQueueStatus)

dvc/commands/queue/stop.py

@@ -0,0 +1,43 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdQueueStop(CmdBase):
+    """Stop exp queue workers."""
+
+    def run(self):
+        self.repo.experiments.celery_queue.shutdown(kill=self.args.kill)
+
+        if self.args.kill:
+            ui.write(
+                "All running tasks in the queue have been killed."
+                "Queue workers are stopping."
+            )
+        else:
+            ui.write("Queue workers will stop after running tasks finish.")
+
+        return 0
+
+
+def add_parser(queue_subparsers, parent_parser):
+
+    QUEUE_STOP_HELP = "Stop all experiments task queue workers."
+    queue_stop_parser = queue_subparsers.add_parser(
+        "stop",
+        parents=[parent_parser],
+        description=append_doc_link(QUEUE_STOP_HELP, "queue/stop"),
+        help=QUEUE_STOP_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    queue_stop_parser.add_argument(
+        "--kill",
+        action="store_true",
+        help="Kill all running tasks before stopping the queue workers.",
+    )
+    queue_stop_parser.set_defaults(func=CmdQueueStop)

dvc/commands/repro.py

@@ -16,9 +16,7 @@ class CmdRepro(CmdBase):
         if len(stages) == 0:
             ui.write(CmdDataStatus.UP_TO_DATE_MSG)
         else:
-            ui.write(
-                "Use `dvc push` to send your updates to " "remote storage."
-            )
+            ui.write("Use `dvc push` to send your updates to remote storage.")
 
         if self.args.metrics:
             from dvc.compare import show_metrics

dvc/commands/stage.py

@@ -76,7 +76,7 @@ class CmdStageList(CmdBase):
         # removing duplicates while maintaining order
         collected = chain.from_iterable(
             self.repo.stage.collect(
-                target=target, recursive=self.args.recursive, accept_group=True
+                target=target, recursive=self.args.recursive
             )
             for target in self.args.targets
         )

dvc/commands/update.py

@@ -18,6 +18,7 @@ class CmdUpdate(CmdBase):
                 rev=self.args.rev,
                 recursive=self.args.recursive,
                 to_remote=self.args.to_remote,
+                no_download=self.args.no_download,
                 remote=self.args.remote,
                 jobs=self.args.jobs,
             )
@@ -28,7 +29,10 @@ class CmdUpdate(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
-    UPDATE_HELP = "Update data artifacts imported from other DVC repositories."
+    UPDATE_HELP = (
+        "Update data artifact imported (via dvc import or dvc import-url) "
+        "from an external DVC repository or URL."
+    )
     update_parser = subparsers.add_parser(
         "update",
         parents=[parent_parser],
@@ -52,6 +56,13 @@ def add_parser(subparsers, parent_parser):
         default=False,
         help="Update all stages in the specified directory.",
     )
+    update_parser.add_argument(
+        "--no-download",
+        action="store_true",
+        default=False,
+        help="Update .dvc file hash value(s)"
+        " but do not download the file(s).",
+    )
     update_parser.add_argument(
         "--to-remote",
         action="store_true",

dvc/config.py

@@ -187,7 +187,7 @@ class Config(dict):
         filename = self.files[level]
         fs = self._get_fs(level)
 
-        logger.debug(f"Writing '{filename}'.")
+        logger.debug("Writing '%s'.", filename)
 
         fs.makedirs(os.path.dirname(filename))
 

dvc/config_schema.py

@@ -78,7 +78,9 @@ REMOTE_COMMON = {
     "url": str,
     "checksum_jobs": All(Coerce(int), Range(1)),
     "jobs": All(Coerce(int), Range(1)),
+    Optional("worktree"): Bool,
     Optional("no_traverse"): Bool,  # obsoleted
+    Optional("version_aware"): Bool,
 }
 LOCAL_COMMON = {
     "type": supported_cache_type,
@@ -152,6 +154,8 @@ SCHEMA = {
                     "ssl_verify": Any(Bool, str),
                     "sse": str,
                     "sse_kms_key_id": str,
+                    "sse_customer_algorithm": str,
+                    "sse_customer_key": str,
                     "acl": str,
                     "grant_read": str,
                     "grant_read_acp": str,
@@ -166,6 +170,7 @@ SCHEMA = {
                 "gs": {
                     "projectname": str,
                     "credentialpath": str,
+                    "endpointurl": str,
                     Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
@@ -217,6 +222,7 @@ SCHEMA = {
                     **REMOTE_COMMON,
                 },
                 "gdrive": {
+                    "profile": str,
                     "gdrive_use_service_account": Bool,
                     "gdrive_client_id": str,
                     "gdrive_client_secret": str,
@@ -224,6 +230,7 @@ SCHEMA = {
                     "gdrive_service_account_user_email": str,
                     "gdrive_service_account_json_file_path": str,
                     Optional("gdrive_trash_only", default=False): Bool,
+                    Optional("gdrive_acknowledge_abuse", default=False): Bool,
                     Optional("verify", default=True): Bool,
                     **REMOTE_COMMON,
                 },
@@ -269,6 +276,7 @@ SCHEMA = {
     "plots": {
         "html_template": str,
         Optional("auto_open", default=False): Bool,
+        "out_dir": str,
     },
     "exp": {
         "code": str,
@@ -279,4 +287,13 @@ SCHEMA = {
         "plots": str,
         "live": str,
     },
+    "parsing": {
+        "bool": All(Lower, Choices("store_true", "boolean_optional")),
+        "list": All(Lower, Choices("nargs", "append")),
+    },
+    "hydra": {
+        Optional("enabled", default=False): Bool,
+        "config_dir": str,
+        "config_name": str,
+    },
 }

dvc/daemon.py

@@ -6,6 +6,7 @@ import os
 import platform
 import sys
 from subprocess import Popen
+from typing import List
 
 from dvc.env import DVC_DAEMON
 from dvc.utils import fix_env, is_binary
@@ -81,7 +82,7 @@ def _spawn_posix(cmd, env):
 
 
 def _spawn(cmd, env):
-    logger.debug(f"Trying to spawn '{cmd}'")
+    logger.debug("Trying to spawn '%s'", cmd)
 
     if os.name == "nt":
         _spawn_windows(cmd, env)
@@ -90,7 +91,7 @@ def _spawn(cmd, env):
     else:
         raise NotImplementedError
 
-    logger.debug(f"Spawned '{cmd}'")
+    logger.debug("Spawned '%s'", cmd)
 
 
 def daemon(args):
@@ -99,12 +100,14 @@ def daemon(args):
     Args:
         args (list): list of arguments to append to `dvc daemon` command.
     """
+    daemonize(["daemon", "-q"] + args)
+
+
+def daemonize(cmd: List[str]):
     if os.environ.get(DVC_DAEMON):
         logger.debug("skipping launching a new daemon.")
         return
 
-    cmd = ["daemon", "-q"] + args
-
     env = fix_env()
     if not is_binary():
         file_path = os.path.abspath(inspect.stack()[0][1])

dvc/data_cloud.py

@@ -3,16 +3,37 @@
 import logging
 from typing import TYPE_CHECKING, Iterable, Optional
 
-from dvc_data.db import get_index
+from funcy import cached_property
+
+from dvc_data.hashfile.db import get_index
 
 if TYPE_CHECKING:
-    from dvc_data.status import CompareStatusResult
-    from dvc_objects.db import ObjectDB
-    from dvc_objects.hash_info import HashInfo
+    from dvc_data.hashfile.db import HashFileDB
+    from dvc_data.hashfile.hash_info import HashInfo
+    from dvc_data.hashfile.status import CompareStatusResult
 
 logger = logging.getLogger(__name__)
 
 
+class Remote:
+    def __init__(self, path, fs, **config):
+        self.path = path
+        self.fs = fs
+
+        self.worktree = config.pop("worktree", False)
+        self.config = config
+
+    @cached_property
+    def odb(self):
+        from dvc_data.hashfile.db import get_odb
+
+        path = self.path
+        if self.worktree:
+            path = self.fs.path.join(path, ".dvc", "cache")
+
+        return get_odb(self.fs, path, **self.config)
+
+
 class DataCloud:
     """Class that manages dvc remotes.
 
@@ -27,18 +48,23 @@ class DataCloud:
     def __init__(self, repo):
         self.repo = repo
 
-    def get_remote_odb(
+    def get_remote(
         self,
         name: Optional[str] = None,
         command: str = "<command>",
-    ) -> "ObjectDB":
+    ) -> "Remote":
         from dvc.config import NoRemoteError
 
         if not name:
             name = self.repo.config["core"].get("remote")
 
         if name:
-            return self._init_odb(name)
+            from dvc.fs import get_cloud_fs
+
+            cls, config, fs_path = get_cloud_fs(self.repo, name=name)
+            fs = cls(**config)
+            config["tmp_dir"] = self.repo.index_db_dir
+            return Remote(fs_path, fs, **config)
 
         if bool(self.repo.config["remote"]):
             error_msg = (
@@ -55,13 +81,13 @@ class DataCloud:
 
         raise NoRemoteError(error_msg)
 
-    def _init_odb(self, name):
-        from dvc.fs import get_cloud_fs
-        from dvc_data.db import get_odb
-
-        cls, config, fs_path = get_cloud_fs(self.repo, name=name)
-        config["tmp_dir"] = self.repo.index_db_dir
-        return get_odb(cls(**config), fs_path, **config)
+    def get_remote_odb(
+        self,
+        name: Optional[str] = None,
+        command: str = "<command>",
+    ) -> "HashFileDB":
+        remote = self.get_remote(name=name, command=command)
+        return remote.odb
 
     def _log_missing(self, status: "CompareStatusResult"):
         if status.missing:
@@ -76,13 +102,13 @@ class DataCloud:
 
     def transfer(
         self,
-        src_odb: "ObjectDB",
-        dest_odb: "ObjectDB",
+        src_odb: "HashFileDB",
+        dest_odb: "HashFileDB",
         objs: Iterable["HashInfo"],
         **kwargs,
     ):
         from dvc.exceptions import FileTransferError
-        from dvc_data.transfer import TransferError, transfer
+        from dvc_data.hashfile.transfer import TransferError, transfer
 
         try:
             return transfer(src_odb, dest_odb, objs, **kwargs)
@@ -94,7 +120,7 @@ class DataCloud:
         objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        odb: Optional["ObjectDB"] = None,
+        odb: Optional["HashFileDB"] = None,
     ):
         """Push data items in a cloud-agnostic way.
 
@@ -121,7 +147,7 @@ class DataCloud:
         objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        odb: Optional["ObjectDB"] = None,
+        odb: Optional["HashFileDB"] = None,
     ):
         """Pull data items in a cloud-agnostic way.
 
@@ -149,7 +175,7 @@ class DataCloud:
         objs: Iterable["HashInfo"],
         jobs: Optional[int] = None,
         remote: Optional[str] = None,
-        odb: Optional["ObjectDB"] = None,
+        odb: Optional["HashFileDB"] = None,
         log_missing: bool = True,
     ):
         """Check status of data items in a cloud-agnostic way.
@@ -164,7 +190,7 @@ class DataCloud:
             log_missing: log warning messages if file doesn't exist
                 neither in cache, neither in cloud.
         """
-        from dvc_data.status import compare_status
+        from dvc_data.hashfile.status import compare_status
 
         if not odb:
             odb = self.get_remote_odb(remote, "status")
@@ -180,5 +206,5 @@ class DataCloud:
 
     def get_url_for(self, remote, checksum):
         odb = self.get_remote_odb(remote)
-        path = odb.hash_to_path(checksum)
+        path = odb.oid_to_path(checksum)
         return odb.fs.unstrip_protocol(path)

dvc/dependency/param.py

@@ -8,8 +8,8 @@ import dpath.util
 from voluptuous import Any
 
 from dvc.exceptions import DvcException
-from dvc.utils.serialize import LOADERS, ParseError
-from dvc_objects.hash_info import HashInfo
+from dvc.utils.serialize import ParseError, load_path
+from dvc_data.hashfile.hash_info import HashInfo
 
 from .base import Dependency
 
@@ -39,12 +39,13 @@ class ParamsDependency(Dependency):
 
     def __init__(self, stage, path, params=None, repo=None):
         self.params = list(params) if params else []
-        info = (
-            {self.PARAM_PARAMS: params} if isinstance(params, dict) else None
-        )
+        hash_info = HashInfo()
+        if isinstance(params, dict):
+            hash_info = HashInfo(self.PARAM_PARAMS, params)
         repo = repo or stage.repo
         path = path or os.path.join(repo.root_dir, self.DEFAULT_PARAMS_FILE)
-        super().__init__(stage, path, info=info, repo=repo)
+        super().__init__(stage, path, repo=repo)
+        self.hash_info = hash_info
 
     def dumpd(self):
         ret = super().dumpd()
@@ -139,12 +140,9 @@ class ParamsDependency(Dependency):
             )
 
     def read_file(self):
-        _, ext = os.path.splitext(self.fs_path)
-        loader = LOADERS[ext]
-
         self.validate_filepath()
         try:
-            return loader(self.fs_path, fs=self.repo.fs)
+            return load_path(self.fs_path, self.repo.fs)
         except ParseError as exc:
             raise BadParamFileError(
                 f"Unable to read parameters from '{self}'"

dvc/dependency/repo.py

@@ -10,9 +10,9 @@ from dvc.prompt import confirm
 from .base import Dependency
 
 if TYPE_CHECKING:
+    from dvc_data.hashfile.file import HashFile
+    from dvc_data.hashfile.hash_info import HashInfo
     from dvc_objects.db import ObjectDB
-    from dvc_objects.file import HashFile
-    from dvc_objects.hash_info import HashInfo
 
 
 class RepoDependency(Dependency):
@@ -31,7 +31,7 @@ class RepoDependency(Dependency):
 
     def __init__(self, def_repo, stage, *args, **kwargs):
         self.def_repo = def_repo
-        self._staged_objs: Dict[str, "HashFile"] = {}
+        self._objs: Dict[str, "HashFile"] = {}
         super().__init__(stage, *args, **kwargs)
 
     def _parse_path(self, fs, fs_path):
@@ -63,7 +63,7 @@ class RepoDependency(Dependency):
         return {self.PARAM_PATH: self.def_path, self.PARAM_REPO: self.def_repo}
 
     def download(self, to, jobs=None):
-        from dvc_data.checkout import checkout
+        from dvc_data.hashfile.checkout import checkout
 
         for odb, objs in self.get_used_objs().items():
             self.repo.cloud.pull(objs, jobs=jobs, odb=odb)
@@ -103,14 +103,12 @@ class RepoDependency(Dependency):
         from dvc.config import NoRemoteError
         from dvc.exceptions import NoOutputOrStageError, PathMissingError
         from dvc.utils import as_posix
-        from dvc_data.stage import stage
-        from dvc_data.tree import Tree, TreeError
+        from dvc_data.hashfile.build import build
+        from dvc_data.hashfile.tree import Tree, TreeError
 
         local_odb = self.repo.odb.local
         locked = kwargs.pop("locked", True)
-        with self._make_repo(
-            locked=locked, cache_dir=local_odb.cache_dir
-        ) as repo:
+        with self._make_repo(locked=locked, cache_dir=local_odb.path) as repo:
             used_obj_ids = defaultdict(set)
             rev = repo.get_rev()
             if locked and self.def_repo.get(self.PARAM_REV_LOCK) is None:
@@ -133,7 +131,7 @@ class RepoDependency(Dependency):
                     pass
 
             try:
-                staging, _, staged_obj = stage(
+                object_store, _, obj = build(
                     local_odb,
                     as_posix(self.def_path),
                     repo.dvcfs,
@@ -143,31 +141,31 @@ class RepoDependency(Dependency):
                 raise PathMissingError(
                     self.def_path, self.def_repo[self.PARAM_URL]
                 ) from exc
-            staging = copy(staging)
-            staging.read_only = True
+            object_store = copy(object_store)
+            object_store.read_only = True
 
-            self._staged_objs[rev] = staged_obj
-            used_obj_ids[staging].add(staged_obj.hash_info)
-            if isinstance(staged_obj, Tree):
-                used_obj_ids[staging].update(oid for _, _, oid in staged_obj)
-            return used_obj_ids, staged_obj
+            self._objs[rev] = obj
+            used_obj_ids[object_store].add(obj.hash_info)
+            if isinstance(obj, Tree):
+                used_obj_ids[object_store].update(oid for _, _, oid in obj)
+            return used_obj_ids, obj
 
     def _check_circular_import(self, odb, obj_ids):
         from dvc.exceptions import CircularImportError
         from dvc.fs.dvc import DvcFileSystem
-        from dvc_data.db.reference import ReferenceObjectDB
-        from dvc_data.tree import Tree
+        from dvc_data.hashfile.db.reference import ReferenceHashFileDB
+        from dvc_data.hashfile.tree import Tree
 
-        if not isinstance(odb, ReferenceObjectDB):
+        if not isinstance(odb, ReferenceHashFileDB):
             return
 
         def iter_objs():
             for hash_info in obj_ids:
                 if hash_info.isdir:
                     tree = Tree.load(odb, hash_info)
-                    yield from (odb.get(oid) for _, _, oid in tree)
+                    yield from (odb.get(hi.value) for _, _, hi in tree)
                 else:
-                    yield odb.get(hash_info)
+                    yield odb.get(hash_info.value)
 
         checked_urls = set()
         for obj in iter_objs():
@@ -190,8 +188,8 @@ class RepoDependency(Dependency):
     def get_obj(self, filter_info=None, **kwargs):
         locked = kwargs.get("locked", True)
         rev = self._get_rev(locked=locked)
-        if rev in self._staged_objs:
-            return self._staged_objs[rev]
+        if rev in self._objs:
+            return self._objs[rev]
         _, obj = self._get_used_and_obj(
             obj_only=True, filter_info=filter_info, **kwargs
         )

dvc/dvcfile.py

@@ -197,7 +197,7 @@ class SingleStageFile(FileMixin):
         assert not isinstance(stage, PipelineStage)
         if self.verify:
             check_dvcfile_path(self.repo, self.path)
-        logger.debug(f"Saving information to '{relpath(self.path)}'.")
+        logger.debug("Saving information to '%s'.", relpath(self.path))
         dump_yaml(self.path, serialize.to_single_stage_file(stage))
         self.repo.scm_context.track_file(self.relpath)
 
@@ -243,9 +243,11 @@ class PipelineFile(FileMixin):
         self._lockfile.dump(stage)
 
     @staticmethod
-    def _check_if_parametrized(stage):
+    def _check_if_parametrized(stage, action: str = "dump") -> None:
         if stage.raw_data.parametrized:
-            raise ParametrizedDumpError(f"cannot dump a parametrized {stage}")
+            raise ParametrizedDumpError(
+                f"cannot {action} a parametrized {stage}"
+            )
 
     def _dump_pipeline_file(self, stage):
         self._check_if_parametrized(stage)
@@ -291,6 +293,7 @@ class PipelineFile(FileMixin):
         self._lockfile.remove()
 
     def remove_stage(self, stage):
+        self._check_if_parametrized(stage, "remove")
         self._lockfile.remove_stage(stage)
         if not self.exists():
             return

dvc/external_repo.py

@@ -145,7 +145,7 @@ def _get_remote_config(url):
             name = "auto-generated-upstream"
             return {
                 "core": {"remote": name},
-                "remote": {name: {"url": repo.odb.local.cache_dir}},
+                "remote": {name: {"url": repo.odb.local.path}},
             }
 
         # Use original remote to make sure that we are using correct url,

dvc/fs/__init__.py

@@ -1,26 +1,18 @@
 from urllib.parse import urlparse
 
+from dvc_http import HTTPFileSystem, HTTPSFileSystem  # noqa: F401
+
 # pylint: disable=unused-import
 from dvc_objects.fs import utils  # noqa: F401
 from dvc_objects.fs import (  # noqa: F401
-    FS_MAP,
-    AzureFileSystem,
-    GDriveFileSystem,
-    GSFileSystem,
-    HDFSFileSystem,
-    HTTPFileSystem,
-    HTTPSFileSystem,
     LocalFileSystem,
     MemoryFileSystem,
-    OSSFileSystem,
-    S3FileSystem,
     Schemes,
-    SSHFileSystem,
-    WebDAVFileSystem,
-    WebDAVSFileSystem,
-    WebHDFSFileSystem,
     generic,
     get_fs_cls,
+    known_implementations,
+    localfs,
+    registry,
     system,
 )
 from dvc_objects.fs.base import AnyFSPath, FileSystem  # noqa: F401
@@ -29,18 +21,26 @@ from dvc_objects.fs.errors import (  # noqa: F401
     ConfigError,
     RemoteMissingDepsError,
 )
-from dvc_objects.fs.implementations.azure import AzureAuthError  # noqa: F401
-from dvc_objects.fs.implementations.gdrive import GDriveAuthError  # noqa: F401
-from dvc_objects.fs.implementations.local import localfs  # noqa: F401
-from dvc_objects.fs.implementations.ssh import (  # noqa: F401
-    DEFAULT_PORT as DEFAULT_SSH_PORT,
-)
 from dvc_objects.fs.path import Path  # noqa: F401
 
 from .data import DataFileSystem  # noqa: F401
 from .dvc import DvcFileSystem  # noqa: F401
 from .git import GitFileSystem  # noqa: F401
 
+known_implementations.update(
+    {
+        "dvc": {
+            "class": "dvc.fs.dvc.DvcFileSystem",
+            "err": "dvc is supported, but requires 'dvc' to be installed",
+        },
+        "git": {
+            "class": "dvc.fs.git.GitFileSystem",
+            "err": "git is supported, but requires 'dvc' to be installed",
+        },
+    }
+)
+
+
 # pylint: enable=unused-import
 
 
@@ -104,11 +104,6 @@ def get_cloud_fs(repo, **kwargs):
     except Invalid as exc:
         raise RepoConfigError(str(exc)) from None
 
-    if "jobs" not in remote_conf:
-        jobs = core_config.get("jobs")
-        if jobs:
-            remote_conf["jobs"] = jobs
-
     if "checksum_jobs" not in remote_conf:
         checksum_jobs = core_config.get("checksum_jobs")
         if checksum_jobs:
@@ -116,11 +111,8 @@ def get_cloud_fs(repo, **kwargs):
 
     cls = get_fs_cls(remote_conf)
 
-    if cls == GDriveFileSystem and repo:
-        remote_conf["gdrive_credentials_tmp_dir"] = repo.tmp_dir
-
     url = remote_conf.pop("url")
-    if issubclass(cls, WebDAVFileSystem):
+    if cls.protocol in ["webdav", "webdavs"]:
         # For WebDAVFileSystem, provided url is the base path itself, so it
         # should be treated as being a root path.
         fs_path = cls.root_marker

dvc/fs/data.py

@@ -1,183 +1,16 @@
 import logging
+import os
 import threading
-import typing
 
-from fsspec import AbstractFileSystem
 from funcy import cached_property, wrap_prop
 
+from dvc.utils import as_posix
+from dvc_data.fs import DataFileSystem as _DataFileSystem
 from dvc_objects.fs.base import FileSystem
-from dvc_objects.fs.callbacks import DEFAULT_CALLBACK
-
-if typing.TYPE_CHECKING:
-    from dvc.types import AnyPath
 
 logger = logging.getLogger(__name__)
 
 
-class _DataFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
-    """DVC repo fs.
-
-    Args:
-        repo: DVC repo.
-    """
-
-    root_marker = "/"
-
-    def __init__(self, workspace=None, **kwargs):
-        super().__init__(**kwargs)
-        self.repo = kwargs["repo"]
-        self.workspace = workspace or "repo"
-
-    @cached_property
-    def path(self):
-        from . import Path
-
-        def _getcwd():
-            return self.root_marker
-
-        return Path(self.sep, getcwd=_getcwd)
-
-    @property
-    def config(self):
-        raise NotImplementedError
-
-    def _get_key(self, path):
-        if self.workspace != "repo":
-            from . import get_cloud_fs
-
-            cls, kwargs, fs_path = get_cloud_fs(None, url=path)
-            fs = cls(**kwargs)
-            return (self.workspace, *fs.path.parts(fs_path))
-
-        path = self.path.abspath(path)
-        if path == self.root_marker:
-            return (self.workspace,)
-
-        key = self.path.relparts(path, self.root_marker)
-        if key == (".") or key == (""):
-            key = ()
-
-        return (self.workspace, *key)
-
-    def _get_fs_path(self, path: "AnyPath", remote=None):
-        from dvc.config import NoRemoteError
-
-        info = self.info(path)
-        if info["type"] == "directory":
-            raise IsADirectoryError
-
-        value = info.get("md5")
-        if not value:
-            raise FileNotFoundError
-
-        cache_path = self.repo.odb.local.hash_to_path(value)
-
-        if self.repo.odb.local.fs.exists(cache_path):
-            return self.repo.odb.local.fs, cache_path
-
-        try:
-            remote_odb = self.repo.cloud.get_remote_odb(remote)
-        except NoRemoteError as exc:
-            raise FileNotFoundError from exc
-        remote_fs_path = remote_odb.hash_to_path(value)
-        return remote_odb.fs, remote_fs_path
-
-    def open(  # type: ignore
-        self, path: str, mode="r", encoding=None, **kwargs
-    ):  # pylint: disable=arguments-renamed, arguments-differ
-        fs, fspath = self._get_fs_path(path, **kwargs)
-        return fs.open(fspath, mode=mode, encoding=encoding)
-
-    def ls(self, path, detail=True, **kwargs):
-        info = self.info(path)
-        if info["type"] != "directory":
-            return [info] if detail else [path]
-
-        root_key = self._get_key(path)
-        try:
-            entries = [
-                self.sep.join((path, name)) if path else name
-                for name in self.repo.index.tree.ls(prefix=root_key)
-            ]
-        except KeyError as exc:
-            raise FileNotFoundError from exc
-
-        if not detail:
-            return entries
-
-        return [self.info(epath) for epath in entries]
-
-    def isdvc(self, path, recursive=False, strict=True):
-        try:
-            info = self.info(path)
-        except FileNotFoundError:
-            return False
-
-        recurse = recursive or not strict
-        return bool(info.get("outs") if recurse else info.get("isout"))
-
-    def info(self, path, **kwargs):
-        from dvc_objects.meta import Meta
-
-        key = self._get_key(path)
-
-        try:
-            outs = list(self.repo.index.tree.iteritems(key))
-        except KeyError as exc:
-            raise FileNotFoundError from exc
-
-        ret = {
-            "type": "file",
-            "size": 0,
-            "isexec": False,
-            "isdvc": False,
-            "outs": outs,
-            "name": path,
-        }
-
-        if len(outs) > 1 and outs[0][0] != key:
-            shortest = self.repo.index.tree.shortest_prefix(key)
-            if shortest:
-                assert shortest[1][1].isdir
-                if len(shortest[0]) <= len(key):
-                    ret["isdvc"] = True
-
-            ret["type"] = "directory"
-            return ret
-
-        item_key, (meta, hash_info) = outs[0]
-
-        meta = meta or Meta()
-
-        if key != item_key:
-            assert item_key[: len(key)] == key
-            ret["type"] = "directory"
-            return ret
-
-        ret["size"] = meta.size
-        ret["isexec"] = meta.isexec
-        ret[hash_info.name] = hash_info.value
-        ret["isdvc"] = True
-        ret["isout"] = True
-        ret["meta"] = meta
-        if hash_info and hash_info.isdir:
-            ret["type"] = "directory"
-        return ret
-
-    def get_file(  # pylint: disable=arguments-differ
-        self, rpath, lpath, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        fs, path = self._get_fs_path(rpath)
-        fs.get_file(path, lpath, callback=callback, **kwargs)
-
-    def checksum(self, path):
-        info = self.info(path)
-        md5 = info.get("md5")
-        if md5:
-            return md5
-        raise NotImplementedError
-
-
 class DataFileSystem(FileSystem):
     protocol = "local"
 
@@ -194,6 +27,8 @@ class DataFileSystem(FileSystem):
     def isdvc(self, path, **kwargs):
         return self.fs.isdvc(path, **kwargs)
 
-    @property
-    def repo(self):
-        return self.fs.repo
+    def from_os_path(self, path):
+        if os.path.isabs(path):
+            path = os.path.splitdrive(path)[1]
+
+        return as_posix(path)

dvc/fs/dvc.py

@@ -4,8 +4,7 @@ import os
 import posixpath
 import threading
 from contextlib import suppress
-from itertools import takewhile
-from typing import TYPE_CHECKING, Callable, Optional, Tuple, Type, Union
+from typing import TYPE_CHECKING, Any, Callable, Optional, Tuple, Type, Union
 
 from fsspec.spec import AbstractFileSystem
 from funcy import cached_property, wrap_prop, wrap_with
@@ -22,29 +21,19 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 RepoFactory = Union[Callable[[str], "Repo"], Type["Repo"]]
+Key = Tuple[str, ...]
 
 
 def as_posix(path: str) -> str:
     return path.replace(ntpath.sep, posixpath.sep)
 
 
-def _wrap_walk(dvc_fs, *args, **kwargs):
-    for root, dnames, fnames in dvc_fs.walk(*args, **kwargs):
-        yield dvc_fs.path.join(dvc_fs.repo.root_dir, root), dnames, fnames
+# NOT the same as dvc.dvcfile.is_dvc_file()!
+def _is_dvc_file(fname):
+    from dvc.dvcfile import is_valid_filename
+    from dvc.ignore import DvcIgnore
 
-
-def _ls(fs, path):
-    dnames = []
-    fnames = []
-
-    for entry in fs.ls(path, detail=True):
-        name = fs.path.name(entry["name"])
-        if entry["type"] == "directory":
-            dnames.append(name)
-        else:
-            fnames.append(name)
-
-    return dnames, fnames
+    return is_valid_filename(fname) or fname == DvcIgnore.DVCIGNORE_FILE
 
 
 def _merge_info(repo, fs_info, dvc_info):
@@ -70,6 +59,10 @@ def _merge_info(repo, fs_info, dvc_info):
     return ret
 
 
+def _get_dvc_path(dvc_fs, subkey):
+    return dvc_fs.path.join(*subkey) if subkey else ""
+
+
 class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
     """DVC + git-tracked files fs.
 
@@ -82,28 +75,25 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
 
     root_marker = "/"
 
-    PARAM_REPO_URL = "repo_url"
-    PARAM_REPO_ROOT = "repo_root"
-    PARAM_REV = "rev"
-    PARAM_CACHE_DIR = "cache_dir"
-    PARAM_CACHE_TYPES = "cache_types"
-    PARAM_SUBREPOS = "subrepos"
-
     def __init__(
         self,
+        url: Optional[str] = None,
+        rev: Optional[str] = None,
         repo: Optional["Repo"] = None,
-        subrepos=False,
-        repo_factory: RepoFactory = None,
-        **kwargs,
-    ):
-        super().__init__()
-
+        subrepos: bool = False,
+        repo_factory: Optional[RepoFactory] = None,
+        **repo_kwargs: Any,
+    ) -> None:
         from pygtrie import Trie
 
+        super().__init__()
         if repo is None:
-            repo, repo_factory = self._repo_from_fs_config(
-                subrepos=subrepos, **kwargs
+            repo = self._make_repo(
+                url=url, rev=rev, subrepos=subrepos, **repo_kwargs
             )
+            assert repo
+            # pylint: disable=protected-access
+            repo_factory = repo._fs_conf["repo_factory"]
 
         if not repo_factory:
             from dvc.repo import Repo
@@ -135,82 +125,37 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         """Keep a datafs instance of each repo."""
 
         if hasattr(repo, "dvc_dir"):
-            self._datafss[key] = DataFileSystem(repo=repo)
+            self._datafss[key] = DataFileSystem(index=repo.index.data["repo"])
 
-    def _get_key(self, path):
+    def _get_key(self, path) -> Key:
         parts = self.repo.fs.path.relparts(path, self.repo.root_dir)
         if parts == (".",):
             parts = ()
         return parts
 
+    def _get_key_from_relative(self, path) -> Key:
+        parts = self.path.relparts(path, self.root_marker)
+        if parts and parts[0] == os.curdir:
+            parts = parts[1:]
+        return parts
+
+    def _from_key(self, parts: Key) -> str:
+        return self.repo.fs.path.join(self.repo.root_dir, *parts)
+
     @property
     def repo_url(self):
         if self.repo is None:
             return None
         return self.repo.url
 
-    @property
-    def config(self):
-        return {
-            self.PARAM_REPO_URL: self.repo_url,
-            self.PARAM_REPO_ROOT: self.repo.root_dir,
-            self.PARAM_REV: getattr(self.repo.fs, "rev", None),
-            self.PARAM_CACHE_DIR: os.path.abspath(
-                self.repo.odb.local.cache_dir
-            ),
-            self.PARAM_CACHE_TYPES: self.repo.odb.local.cache_types,
-            self.PARAM_SUBREPOS: self._traverse_subrepos,
-        }
-
     @classmethod
-    def _repo_from_fs_config(
-        cls, **config
-    ) -> Tuple["Repo", Optional["RepoFactory"]]:
-        from dvc.external_repo import erepo_factory, external_repo
+    def _make_repo(cls, **kwargs) -> "Repo":
         from dvc.repo import Repo
 
-        url = config.get(cls.PARAM_REPO_URL)
-        root = config.get(cls.PARAM_REPO_ROOT)
-        assert url or root
+        with Repo.open(uninitialized=True, **kwargs) as repo:
+            return repo
 
-        def _open(*args, **kwargs):
-            # NOTE: if original repo was an erepo (and has a URL),
-            # we cannot use Repo.open() since it will skip erepo
-            # cache/remote setup for local URLs
-            if url is None:
-                return Repo.open(*args, **kwargs)
-            return external_repo(*args, **kwargs)
-
-        cache_dir = config.get(cls.PARAM_CACHE_DIR)
-        cache_config = (
-            {}
-            if not cache_dir
-            else {
-                "cache": {
-                    "dir": cache_dir,
-                    "type": config.get(cls.PARAM_CACHE_TYPES),
-                }
-            }
-        )
-        repo_kwargs: dict = {
-            "rev": config.get(cls.PARAM_REV),
-            "subrepos": config.get(cls.PARAM_SUBREPOS, False),
-            "uninitialized": True,
-        }
-        factory: Optional["RepoFactory"] = None
-        if url is None:
-            repo_kwargs["config"] = cache_config
-        else:
-            repo_kwargs["cache_dir"] = cache_dir
-            factory = erepo_factory(url, root, cache_config)
-
-        with _open(
-            url if url else root,
-            **repo_kwargs,
-        ) as repo:
-            return repo, factory
-
-    def _get_repo(self, path: str) -> "Repo":
+    def _get_repo(self, key: Key) -> "Repo":
         """Returns repo that the path falls in, using prefix.
 
         If the path is already tracked/collected, it just returns the repo.
@@ -218,33 +163,21 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         Otherwise, it collects the repos that might be in the path's parents
         and then returns the appropriate one.
         """
-        if not self.repo.fs.path.isin_or_eq(path, self.repo.root_dir):
-            # outside of repo
-            return self.repo
-
-        key = self._get_key(path)
         repo = self._subrepos_trie.get(key)
         if repo:
             return repo
 
         prefix_key, repo = self._subrepos_trie.longest_prefix(key)
-        prefix = self.repo.fs.path.join(
-            self.repo.root_dir,
-            *prefix_key,  # pylint: disable=not-an-iterable
-        )
-
-        parents = (parent for parent in self.repo.fs.path.parents(path))
-        dirs = [path] + list(takewhile(lambda p: p != prefix, parents))
-        dirs.reverse()
-        self._update(dirs, starting_repo=repo)
+        dir_keys = (key[:i] for i in range(len(prefix_key) + 1, len(key) + 1))
+        self._update(dir_keys, starting_repo=repo)
         return self._subrepos_trie.get(key) or self.repo
 
     @wrap_with(threading.Lock())
-    def _update(self, dirs, starting_repo):
+    def _update(self, dir_keys, starting_repo):
         """Checks for subrepo in directories and updates them."""
         repo = starting_repo
-        for d in dirs:
-            key = self._get_key(d)
+        for key in dir_keys:
+            d = self._from_key(key)
             if self._is_dvc_repo(d):
                 repo = self.repo_factory(
                     d,
@@ -252,7 +185,9 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
                     scm=self.repo.scm,
                     repo_factory=self.repo_factory,
                 )
-                self._datafss[key] = DataFileSystem(repo=repo)
+                self._datafss[key] = DataFileSystem(
+                    index=repo.index.data["repo"]
+                )
             self._subrepos_trie[key] = repo
 
     def _is_dvc_repo(self, dir_path):
@@ -265,41 +200,23 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         repo_path = self.repo.fs.path.join(dir_path, Repo.DVC_DIR)
         return self.repo.fs.isdir(repo_path)
 
-    def _get_fs_pair(
-        self, path
-    ) -> Tuple[
-        Optional[FileSystem],
-        Optional[str],
-        Optional[DataFileSystem],
-        Optional[str],
-    ]:
+    def _get_subrepo_info(
+        self, key: Key
+    ) -> Tuple["Repo", Optional[DataFileSystem], Key]:
         """
-        Returns a pair of fss based on repo the path falls in, using prefix.
+        Returns information about the subrepo the key is part of.
         """
-        parts = self.path.relparts(path, self.root_marker)
-        if parts and parts[0] == os.curdir:
-            parts = parts[1:]
-
-        fs_path = self.repo.fs.path.join(self.repo.root_dir, *parts)
-        repo = self._get_repo(fs_path)
-        fs = repo.fs
-
-        repo_parts = fs.path.relparts(repo.root_dir, self.repo.root_dir)
-        if repo_parts[0] == os.curdir:
-            repo_parts = repo_parts[1:]
-
-        dvc_parts = parts[len(repo_parts) :]
-        if dvc_parts and dvc_parts[0] == os.curdir:
-            dvc_parts = dvc_parts[1:]
-
-        key = self._get_key(repo.root_dir)
-        dvc_fs = self._datafss.get(key)
-        if dvc_fs:
-            dvc_path = dvc_fs.path.join(*dvc_parts) if dvc_parts else ""
+        repo = self._get_repo(key)
+        repo_key: Key
+        if repo is self.repo:
+            repo_key = ()
+            subkey = key
         else:
-            dvc_path = None
+            repo_key = self._get_key(repo.root_dir)
+            subkey = key[len(repo_key) :]
 
-        return fs, fs_path, dvc_fs, dvc_path
+        dvc_fs = self._datafss.get(repo_key)
+        return repo, dvc_fs, subkey
 
     def open(
         self, path, mode="r", encoding="utf-8", **kwargs
@@ -307,37 +224,43 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         if "b" in mode:
             encoding = None
 
-        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
+        key = self._get_key_from_relative(path)
+        fs_path = self._from_key(key)
         try:
-            return fs.open(fs_path, mode=mode, encoding=encoding)
+            return self.repo.fs.open(fs_path, mode=mode, encoding=encoding)
         except FileNotFoundError:
+            _, dvc_fs, subkey = self._get_subrepo_info(key)
             if not dvc_fs:
                 raise
 
+        dvc_path = _get_dvc_path(dvc_fs, subkey)
         return dvc_fs.open(dvc_path, mode=mode, encoding=encoding, **kwargs)
 
     def isdvc(self, path, **kwargs):
-        _, _, dvc_fs, dvc_path = self._get_fs_pair(path)
+        key = self._get_key_from_relative(path)
+        _, dvc_fs, subkey = self._get_subrepo_info(key)
+        dvc_path = _get_dvc_path(dvc_fs, subkey)
         return dvc_fs is not None and dvc_fs.isdvc(dvc_path, **kwargs)
 
     def ls(  # pylint: disable=arguments-differ
         self, path, detail=True, dvc_only=False, **kwargs
     ):
-        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
-
-        repo = dvc_fs.repo if dvc_fs else self.repo
-        dvcignore = repo.dvcignore
-        ignore_subrepos = kwargs.get("ignore_subrepos", True)
+        key = self._get_key_from_relative(path)
+        repo, dvc_fs, subkey = self._get_subrepo_info(key)
 
         names = set()
         if dvc_fs:
             with suppress(FileNotFoundError):
+                dvc_path = _get_dvc_path(dvc_fs, subkey)
                 for entry in dvc_fs.ls(dvc_path, detail=False):
                     names.add(dvc_fs.path.name(entry))
 
-        if not dvc_only and fs:
+        ignore_subrepos = kwargs.get("ignore_subrepos", True)
+        if not dvc_only:
+            fs = self.repo.fs
+            fs_path = self._from_key(key)
             try:
-                for entry in dvcignore.ls(
+                for entry in repo.dvcignore.ls(
                     fs, fs_path, detail=False, ignore_subrepos=ignore_subrepos
                 ):
                     names.add(fs.path.name(entry))
@@ -345,26 +268,21 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
                 pass
 
         dvcfiles = kwargs.get("dvcfiles", False)
-
-        def _func(fname):
-            from dvc.dvcfile import is_valid_filename
-            from dvc.ignore import DvcIgnore
-
-            if dvcfiles:
-                return True
-
-            return not (
-                is_valid_filename(fname) or fname == DvcIgnore.DVCIGNORE_FILE
-            )
-
-        names = filter(_func, names)
+        if not dvcfiles:
+            names = (name for name in names if not _is_dvc_file(name))
 
         infos = []
         paths = []
         for name in names:
             entry_path = self.path.join(path, name)
+            entry_key = key + (name,)
             try:
-                info = self.info(entry_path, ignore_subrepos=ignore_subrepos)
+                info = self._info(
+                    entry_key,
+                    entry_path,
+                    ignore_subrepos=ignore_subrepos,
+                    check_ignored=False,
+                )
             except FileNotFoundError:
                 continue
             infos.append(info)
@@ -378,46 +296,52 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
     def get_file(  # pylint: disable=arguments-differ
         self, rpath, lpath, callback=DEFAULT_CALLBACK, **kwargs
     ):
-        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(rpath)
-
-        if fs:
-            try:
-                fs.get_file(fs_path, lpath, callback=callback, **kwargs)
-                return
-            except FileNotFoundError:
-                if not dvc_fs:
-                    raise
+        key = self._get_key_from_relative(rpath)
+        fs_path = self._from_key(key)
+        fs = self.repo.fs
+        try:
+            fs.get_file(fs_path, lpath, callback=callback, **kwargs)
+            return
+        except FileNotFoundError:
+            _, dvc_fs, subkey = self._get_subrepo_info(key)
+            if not dvc_fs:
+                raise
+        dvc_path = _get_dvc_path(dvc_fs, subkey)
         dvc_fs.get_file(dvc_path, lpath, callback=callback, **kwargs)
 
     def info(self, path, **kwargs):
-        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
-
-        repo = dvc_fs.repo if dvc_fs else self.repo
-        dvcignore = repo.dvcignore
+        key = self._get_key_from_relative(path)
         ignore_subrepos = kwargs.get("ignore_subrepos", True)
+        return self._info(key, path, ignore_subrepos=ignore_subrepos)
+
+    def _info(self, key, path, ignore_subrepos=True, check_ignored=True):
+        repo, dvc_fs, subkey = self._get_subrepo_info(key)
 
         dvc_info = None
         if dvc_fs:
             try:
-                dvc_info = dvc_fs.info(dvc_path)
+                dvc_info = dvc_fs.fs.index.info(subkey)
+                dvc_path = _get_dvc_path(dvc_fs, subkey)
+                dvc_info["name"] = dvc_path
             except FileNotFoundError:
                 pass
 
         fs_info = None
-        if fs:
-            try:
-                fs_info = fs.info(fs_path)
-                if dvcignore.is_ignored(
-                    fs, fs_path, ignore_subrepos=ignore_subrepos
-                ):
-                    fs_info = None
-            except (FileNotFoundError, NotADirectoryError):
-                if not dvc_info:
-                    raise
+        fs = self.repo.fs
+        fs_path = self._from_key(key)
+        try:
+            fs_info = fs.info(fs_path)
+            if check_ignored and repo.dvcignore.is_ignored(
+                fs, fs_path, ignore_subrepos=ignore_subrepos
+            ):
+                fs_info = None
+        except (FileNotFoundError, NotADirectoryError):
+            if not dvc_info:
+                raise
 
         # NOTE: if some parent in fs_path turns out to be a file, it means
         # that the whole repofs branch doesn't exist.
-        if fs and not fs_info and dvc_info:
+        if dvc_info and not fs_info:
             for parent in fs.path.parents(fs_path):
                 try:
                     if fs.info(parent)["type"] != "directory":
@@ -433,14 +357,6 @@ class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         info["name"] = path
         return info
 
-    def checksum(self, path):
-        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
-
-        try:
-            return fs.checksum(fs_path)
-        except FileNotFoundError:
-            return dvc_fs.checksum(dvc_path)
-
 
 class DvcFileSystem(FileSystem):
     protocol = "local"
@@ -469,10 +385,6 @@ class DvcFileSystem(FileSystem):
     def repo_url(self):
         return self.fs.repo_url
 
-    @property
-    def config(self):
-        return self.fs.config
-
     def from_os_path(self, path):
         if os.path.isabs(path):
             path = os.path.relpath(path, self.repo.root_dir)

dvc/fs/git.py

@@ -15,6 +15,7 @@ class GitFileSystem(FileSystem):  # pylint:disable=abstract-method
     """Proxies the repo file access methods to Git objects"""
 
     protocol = "local"
+    PARAM_CHECKSUM = "md5"
 
     def __init__(
         self,

dvc/info.py

@@ -8,12 +8,20 @@ import psutil
 
 from dvc import __version__
 from dvc.exceptions import NotDvcRepoError
-from dvc.fs import FS_MAP, generic, get_fs_cls, get_fs_config
+from dvc.fs import Schemes, generic, get_fs_cls, get_fs_config, registry
 from dvc.repo import Repo
 from dvc.scm import SCMError
 from dvc.utils import error_link
 from dvc.utils.pkg import PKG
 
+SUBPROJECTS = (
+    "dvc_data",
+    "dvc_objects",
+    "dvc_render",
+    "dvc_task",
+    "dvclive",
+    "scmrepo",
+)
 package = "" if PKG is None else f"({PKG})"
 
 
@@ -23,6 +31,7 @@ def get_dvc_info():
         "---------------------------------",
         f"Platform: Python {platform.python_version()} on "
         f"{platform.platform()}",
+        f"Subprojects:{_get_subprojects()}",
         f"Supports:{_get_supported_remotes()}",
     ]
 
@@ -32,9 +41,9 @@ def get_dvc_info():
             # can't auto-create it, as it might cause issues if the user
             # later decides to enable shared cache mode with
             # `dvc config cache.shared group`.
-            if os.path.exists(repo.odb.local.cache_dir):
+            if os.path.exists(repo.odb.local.path):
                 info.append(f"Cache types: {_get_linktype_support_info(repo)}")
-                fs_type = get_fs_type(repo.odb.local.cache_dir)
+                fs_type = get_fs_type(repo.odb.local.path)
                 info.append(f"Cache directory: {fs_type}")
             else:
                 info.append("Cache types: " + error_link("no-dvc-cache"))
@@ -58,7 +67,7 @@ def _get_caches(cache):
     caches = (
         cache_type
         for cache_type, cache_instance in cache.by_scheme()
-        if cache_instance
+        if cache_instance and cache_type != "repo"
     )
 
     # Caches will be always non-empty including the local cache
@@ -80,7 +89,7 @@ def _get_linktype_support_info(repo):
     links = generic.test_links(
         ["reflink", "hardlink", "symlink"],
         odb.fs,
-        odb.fs_path,
+        odb.path,
         repo.fs,
         repo.root_dir,
     )
@@ -88,9 +97,29 @@ def _get_linktype_support_info(repo):
     return ", ".join(links)
 
 
+def _get_subprojects():
+    subprojects = []
+    for subproject in SUBPROJECTS:
+        try:
+            version = importlib_metadata.version(subproject)
+            subprojects.append(f"{subproject} = {version}")
+        except ImportError:
+            pass
+
+    return "\n\t" + "\n\t".join(subprojects)
+
+
 def _get_supported_remotes():
     supported_remotes = []
-    for scheme, fs_cls in FS_MAP.items():
+    for scheme in registry:
+        if scheme in [Schemes.LOCAL, Schemes.MEMORY, "dvc", "git"]:
+            continue
+
+        try:
+            fs_cls = registry[scheme]
+        except ImportError:
+            continue
+
         if not fs_cls.get_missing_deps():
             dependencies = []
             for requirement in fs_cls.REQUIRES:

dvc/lock.py

@@ -70,6 +70,8 @@ class LockNoop(LockBase):
         self._lock = True
 
     def unlock(self):
+        if not self.is_locked:
+            raise DvcException("Unlock called on an unlocked lock")
         self._lock = False
 
     @property
@@ -93,6 +95,7 @@ class Lock(LockBase):
         super().__init__(lockfile)
         self._friendly = friendly
         self._lock = None
+        self._lock_failed = False
 
     @property
     def files(self):
@@ -100,6 +103,7 @@ class Lock(LockBase):
 
     def _do_lock(self):
         try:
+            self._lock_failed = False
             with Tqdm(
                 bar_format="{desc}",
                 disable=not self._friendly,
@@ -111,6 +115,7 @@ class Lock(LockBase):
             ):
                 self._lock = zc.lockfile.LockFile(self._lockfile)
         except zc.lockfile.LockError:
+            self._lock_failed = True
             raise LockError(FAILED_TO_LOCK_MESSAGE)
 
     def lock(self):
@@ -120,6 +125,12 @@ class Lock(LockBase):
         lock_retry()
 
     def unlock(self):
+        if self._lock_failed:
+            assert self._lock is None
+            return
+
+        if not self.is_locked:
+            raise DvcException("Unlock called on an unlocked lock")
         self._lock.close()
         self._lock = None
 

dvc/logger.py

@@ -163,7 +163,7 @@ class LoggerHandler(logging.StreamHandler):
                 msg, file=self.stream, end=getattr(self, "terminator", "\n")
             )
             self.flush()
-        except RecursionError:
+        except (BrokenPipeError, RecursionError):
             raise
         except Exception:  # noqa, pylint: disable=broad-except
             self.handleError(record)

dvc/machine/backend/base.py

@@ -3,7 +3,8 @@ from contextlib import contextmanager
 from typing import TYPE_CHECKING, Iterator, Optional
 
 if TYPE_CHECKING:
-    from dvc.fs import SSHFileSystem
+    from dvc_ssh import SSHFileSystem
+
     from dvc.types import StrPath
 
 

dvc/machine/backend/terraform.py

@@ -3,8 +3,9 @@ from contextlib import contextmanager
 from functools import partial, partialmethod
 from typing import TYPE_CHECKING, Iterator, Optional
 
+from dvc_ssh import DEFAULT_PORT, SSHFileSystem
+
 from dvc.exceptions import DvcException
-from dvc.fs import DEFAULT_SSH_PORT, SSHFileSystem
 from dvc.utils.fs import makedirs
 
 from .base import BaseMachineBackend
@@ -68,7 +69,7 @@ class TerraformBackend(BaseMachineBackend):
             resource = tpi.default_resource(name)
         return {
             "host": resource["instance_ip"],
-            "port": DEFAULT_SSH_PORT,
+            "port": DEFAULT_PORT,
             "username": "ubuntu",
             "fs_factory": partial(_sshfs, dict(resource)),
         }

dvc/odbmgr.py

@@ -1,8 +1,8 @@
-from dvc.fs import Schemes
-from dvc_data.db import get_odb
+from dvc.fs import GitFileSystem, Schemes
+from dvc_data.hashfile.db import get_odb
 
 
-def _get_odb(repo, settings):
+def _get_odb(repo, settings, fs=None):
     from dvc.fs import get_cloud_fs
 
     if not settings:
@@ -10,7 +10,8 @@ def _get_odb(repo, settings):
 
     cls, config, fs_path = get_cloud_fs(repo, **settings)
     config["tmp_dir"] = repo.tmp_dir
-    return get_odb(cls(**config), fs_path, state=repo.state, **config)
+    fs = fs or cls(**config)
+    return get_odb(fs, fs_path, state=repo.state, **config)
 
 
 class ODBManager:
@@ -24,7 +25,7 @@ class ODBManager:
     ]
 
     def __init__(self, repo):
-        self.repo = repo
+        self._repo = repo
         self.config = config = repo.config["cache"]
         self._odb = {}
 
@@ -42,13 +43,19 @@ class ODBManager:
                 if opt in config:
                     settings[str(opt)] = config.get(opt)
 
-        self._odb[Schemes.LOCAL] = _get_odb(repo, settings)
+        kwargs = {}
+        if not isinstance(repo.fs, GitFileSystem):
+            kwargs["fs"] = repo.fs
+
+        odb = _get_odb(repo, settings, **kwargs)
+        self._odb["repo"] = odb
+        self._odb[Schemes.LOCAL] = odb
 
     def _init_odb(self, schemes):
         for scheme in schemes:
             remote = self.config.get(scheme)
             settings = {"name": remote} if remote else None
-            self._odb[scheme] = _get_odb(self.repo, settings)
+            self._odb[scheme] = _get_odb(self._repo, settings)
 
     def __getattr__(self, name):
         if name not in self._odb and name in self.CLOUD_SCHEMES:

dvc/output.py

@@ -17,25 +17,19 @@ from dvc.exceptions import (
     MergeError,
     RemoteCacheRequiredError,
 )
-from dvc_data import check as ocheck
-from dvc_data import load as oload
-from dvc_data.checkout import checkout
-from dvc_data.stage import stage as ostage
-from dvc_data.transfer import transfer as otransfer
-from dvc_data.tree import Tree
+from dvc_data.hashfile import Tree
+from dvc_data.hashfile import check as ocheck
+from dvc_data.hashfile import load as oload
+from dvc_data.hashfile.build import build
+from dvc_data.hashfile.checkout import checkout
+from dvc_data.hashfile.hash_info import HashInfo
+from dvc_data.hashfile.istextfile import istextfile
+from dvc_data.hashfile.meta import Meta
+from dvc_data.hashfile.transfer import transfer as otransfer
 from dvc_objects.errors import ObjectFormatError
-from dvc_objects.hash_info import HashInfo
-from dvc_objects.istextfile import istextfile
-from dvc_objects.meta import Meta
 
-from .fs import (
-    HDFSFileSystem,
-    LocalFileSystem,
-    RemoteMissingDepsError,
-    S3FileSystem,
-    Schemes,
-    get_cloud_fs,
-)
+from .annotations import ANNOTATION_FIELDS, ANNOTATION_SCHEMA, Annotation
+from .fs import LocalFileSystem, RemoteMissingDepsError, Schemes, get_cloud_fs
 from .utils import relpath
 from .utils.fs import path_isin
 
@@ -65,10 +59,12 @@ CASE_SENSITIVE_CHECKSUM_SCHEMA = Any(
 #
 # so when a few types of outputs share the same name, we only need
 # specify it once.
+HDFS_PARAM_CHECKSUM = "checksum"
+S3_PARAM_CHECKSUM = "etag"
 CHECKSUMS_SCHEMA = {
     LocalFileSystem.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
-    HDFSFileSystem.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
-    S3FileSystem.PARAM_CHECKSUM: CASE_SENSITIVE_CHECKSUM_SCHEMA,
+    HDFS_PARAM_CHECKSUM: CHECKSUM_SCHEMA,
+    S3_PARAM_CHECKSUM: CASE_SENSITIVE_CHECKSUM_SCHEMA,
 }
 
 
@@ -85,9 +81,9 @@ def loadd_from(stage, d_list):
         plot = d.pop(Output.PARAM_PLOT, False)
         persist = d.pop(Output.PARAM_PERSIST, False)
         checkpoint = d.pop(Output.PARAM_CHECKPOINT, False)
-        desc = d.pop(Output.PARAM_DESC, False)
         live = d.pop(Output.PARAM_LIVE, False)
         remote = d.pop(Output.PARAM_REMOTE, None)
+        annot = {field: d.pop(field, None) for field in ANNOTATION_FIELDS}
         ret.append(
             _get(
                 stage,
@@ -98,9 +94,9 @@ def loadd_from(stage, d_list):
                 plot=plot,
                 persist=persist,
                 checkpoint=checkpoint,
-                desc=desc,
                 live=live,
                 remote=remote,
+                **annot,
             )
         )
     return ret
@@ -192,6 +188,7 @@ def load_from_pipeline(stage, data, typ="outs"):
                 Output.PARAM_PERSIST,
                 Output.PARAM_CHECKPOINT,
                 Output.PARAM_REMOTE,
+                *ANNOTATION_FIELDS,
             ],
         )
 
@@ -257,7 +254,6 @@ class Output:
     PARAM_PLOT_TITLE = "title"
     PARAM_PLOT_HEADER = "header"
     PARAM_PERSIST = "persist"
-    PARAM_DESC = "desc"
     PARAM_LIVE = "live"
     PARAM_LIVE_SUMMARY = "summary"
     PARAM_LIVE_HTML = "html"
@@ -289,11 +285,25 @@ class Output:
         checkpoint=False,
         live=False,
         desc=None,
+        type=None,  # pylint: disable=redefined-builtin
+        labels=None,
+        meta=None,
         remote=None,
         repo=None,
     ):
+        self.annot = Annotation(
+            desc=desc, type=type, labels=labels or [], meta=meta or {}
+        )
         self.repo = stage.repo if not repo and stage else repo
-        fs_cls, fs_config, fs_path = get_cloud_fs(self.repo, url=path)
+        meta = Meta.from_dict(info or {})
+        # NOTE: when version_aware is not passed into get_cloud_fs, it will be
+        # set based on whether or not path is versioned
+        fs_kwargs = {"version_aware": True} if meta.version_id else {}
+        fs_cls, fs_config, fs_path = get_cloud_fs(
+            self.repo,
+            url=path,
+            **fs_kwargs,
+        )
         self.fs = fs_cls(**fs_config)
 
         if (
@@ -325,38 +335,52 @@ class Output:
         # By resolved path, which contains actual location,
         # should be absolute and don't contain remote:// refs.
         self.stage = stage
-        self.meta = Meta.from_dict(info)
-        self.hash_info = HashInfo.from_dict(info)
+        self.meta = meta
         self.use_cache = False if self.IS_DEPENDENCY else cache
         self.metric = False if self.IS_DEPENDENCY else metric
         self.plot = False if self.IS_DEPENDENCY else plot
         self.persist = persist
         self.checkpoint = checkpoint
         self.live = live
-        self.desc = desc
 
         self.fs_path = self._parse_path(self.fs, fs_path)
         self.obj = None
 
         self.remote = remote
 
-    def _parse_path(self, fs, fs_path):
-        if fs.protocol != "local":
-            return fs_path
+        if self.fs.version_aware:
+            _, version_id = self.fs.path.coalesce_version(
+                self.def_path, self.meta.version_id
+            )
+            self.meta.version_id = version_id
 
+        if self.is_in_repo:
+            self.hash_name = "md5"
+        else:
+            self.hash_name = self.fs.PARAM_CHECKSUM
+
+        self.hash_info = HashInfo(
+            name=self.hash_name,
+            value=getattr(self.meta, self.hash_name, None),
+        )
+
+    def _parse_path(self, fs, fs_path):
         parsed = urlparse(self.def_path)
-        if parsed.scheme != "remote":
+        if (
+            parsed.scheme != "remote"
+            and self.stage
+            and self.stage.repo.fs == fs
+            and not fs.path.isabs(fs_path)
+        ):
             # NOTE: we can path either from command line or .dvc file,
             # so we should expect both posix and windows style paths.
             # paths accepts both, i.e. / works everywhere, \ only on win.
             #
             # FIXME: if we have Windows path containing / or posix one with \
             # then we have #2059 bug and can't really handle that.
-            if self.stage and not os.path.isabs(fs_path):
-                fs_path = fs.path.join(self.stage.wdir, fs_path)
+            fs_path = fs.path.join(self.stage.wdir, fs_path)
 
-        abs_p = fs.path.abspath(fs.path.normpath(fs_path))
-        return abs_p
+        return fs.path.abspath(fs.path.normpath(fs_path))
 
     def __repr__(self):
         return "{class_name}: '{def_path}'".format(
@@ -380,15 +404,17 @@ class Output:
 
         return self.fs.path.relpath(self.fs_path, self.repo.root_dir)
 
+    def clear(self):
+        self.hash_info = HashInfo.from_dict({})
+        self.meta = Meta.from_dict({})
+        self.obj = None
+
     @property
     def protocol(self):
         return self.fs.protocol
 
     @property
     def is_in_repo(self):
-        if self.fs.protocol != "local":
-            return False
-
         if urlparse(self.def_path).scheme == "remote":
             return False
 
@@ -409,7 +435,8 @@ class Output:
 
     @property
     def odb(self):
-        odb = getattr(self.repo.odb, self.protocol)
+        odb_name = "repo" if self.is_in_repo else self.protocol
+        odb = getattr(self.repo.odb, odb_name)
         if self.use_cache and odb is None:
             raise RemoteCacheRequiredError(self.fs.protocol, self.fs_path)
         return odb
@@ -417,25 +444,31 @@ class Output:
     @property
     def cache_path(self):
         return self.odb.fs.unstrip_protocol(
-            self.odb.hash_to_path(self.hash_info.value)
+            self.odb.oid_to_path(self.hash_info.value)
         )
 
     def get_hash(self):
+        _, hash_info = self._get_hash_meta()
+        return hash_info
+
+    def _get_hash_meta(self):
         if self.use_cache:
             odb = self.odb
-            name = self.odb.fs.PARAM_CHECKSUM
         else:
             odb = self.repo.odb.local
-            name = self.fs.PARAM_CHECKSUM
-        _, _, obj = ostage(
+        _, meta, obj = build(
             odb,
             self.fs_path,
             self.fs,
-            name,
+            self.hash_name,
             ignore=self.dvcignore,
             dry_run=not self.use_cache,
         )
-        return obj.hash_info
+        return meta, obj.hash_info
+
+    def get_meta(self) -> Meta:
+        meta, _ = self._get_hash_meta()
+        return meta
 
     @property
     def is_dir_checksum(self):
@@ -471,6 +504,11 @@ class Output:
         except (FileNotFoundError, ObjectFormatError):
             return True
 
+    def changed_meta(self) -> bool:
+        if self.fs.version_aware and self.meta.version_id:
+            return self.meta.version_id == self.get_meta().version_id
+        return False
+
     def workspace_status(self):
         if not self.exists:
             return {str(self): "deleted"}
@@ -541,19 +579,19 @@ class Output:
             raise self.IsNotFileOrDirError(self)
 
         if self.is_empty:
-            logger.warning(f"'{self}' is empty.")
+            logger.warning("'%s' is empty.", self)
 
         self.ignore()
 
-        if self.metric or self.plot:
+        if self.metric:
             self.verify_metric()
 
         if not self.use_cache:
-            _, self.meta, obj = ostage(
+            _, self.meta, obj = build(
                 self.repo.odb.local,
                 self.fs_path,
                 self.fs,
-                self.fs.PARAM_CHECKSUM,
+                self.hash_name,
                 ignore=self.dvcignore,
                 dry_run=True,
             )
@@ -566,11 +604,11 @@ class Output:
 
         assert not self.IS_DEPENDENCY
 
-        _, self.meta, self.obj = ostage(
+        _, self.meta, self.obj = build(
             self.odb,
             self.fs_path,
             self.fs,
-            self.odb.fs.PARAM_CHECKSUM,
+            self.hash_name,
             ignore=self.dvcignore,
         )
         self.hash_info = self.obj.hash_info
@@ -580,9 +618,10 @@ class Output:
             self.odb.set_exec(self.fs_path)
 
     def _checkout(self, *args, **kwargs):
-        from dvc_data.checkout import CheckoutError as _CheckoutError
-        from dvc_data.checkout import LinkError, PromptError
+        from dvc_data.hashfile.checkout import CheckoutError as _CheckoutError
+        from dvc_data.hashfile.checkout import LinkError, PromptError
 
+        kwargs.setdefault("ignore", self.dvcignore)
         try:
             return checkout(*args, **kwargs)
         except PromptError as exc:
@@ -607,11 +646,11 @@ class Output:
             if granular:
                 obj = self._commit_granular_dir(filter_info)
             else:
-                staging, _, obj = ostage(
+                staging, _, obj = build(
                     self.odb,
                     filter_info or self.fs_path,
                     self.fs,
-                    self.odb.fs.PARAM_CHECKSUM,
+                    self.hash_name,
                     ignore=self.dvcignore,
                 )
                 otransfer(
@@ -627,7 +666,6 @@ class Output:
                 obj,
                 self.odb,
                 relink=True,
-                ignore=self.dvcignore,
                 state=self.repo.state,
                 prompt=prompt.confirm,
             )
@@ -637,15 +675,15 @@ class Output:
         prefix = self.fs.path.parts(
             self.fs.path.relpath(filter_info, self.fs_path)
         )
-        staging, _, save_obj = ostage(
+        staging, _, save_obj = build(
             self.odb,
             self.fs_path,
             self.fs,
-            self.odb.fs.PARAM_CHECKSUM,
+            self.hash_name,
             ignore=self.dvcignore,
         )
         save_obj = save_obj.filter(prefix)
-        checkout_obj = save_obj.get(self.odb, prefix)
+        checkout_obj = save_obj.get_obj(self.odb, prefix)
         otransfer(
             staging,
             self.odb,
@@ -670,9 +708,7 @@ class Output:
         if self.IS_DEPENDENCY:
             return ret
 
-        if self.desc:
-            ret[self.PARAM_DESC] = self.desc
-
+        ret.update(self.annot.to_dict())
         if not self.use_cache:
             ret[self.PARAM_CACHE] = self.use_cache
 
@@ -708,22 +744,21 @@ class Output:
             raise DvcException(
                 f"verify metric is not supported for {self.protocol}"
             )
-
-        if not self.metric or self.plot:
+        if not self.metric:
             return
 
         if not os.path.exists(self.fs_path):
             return
 
-        name = "metrics" if self.metric else "plot"
         if os.path.isdir(self.fs_path):
             msg = "directory '%s' cannot be used as %s."
-            logger.debug(msg, str(self), name)
+            logger.debug(msg, str(self), "metrics")
             return
 
         if not istextfile(self.fs_path, self.fs):
-            msg = "binary file '{}' cannot be used as {}."
-            raise DvcException(msg.format(self.fs_path, name))
+            raise DvcException(
+                f"binary file '{self.fs_path}' cannot be used as metrics."
+            )
 
     def download(self, to, jobs=None):
         from dvc.fs.callbacks import Callback
@@ -740,7 +775,7 @@ class Output:
         elif self.hash_info:
             try:
                 obj = oload(self.odb, self.hash_info)
-            except FileNotFoundError:
+            except (FileNotFoundError, ObjectFormatError):
                 return None
         else:
             return None
@@ -748,7 +783,7 @@ class Output:
         fs_path = self.fs.path
         if filter_info and filter_info != self.fs_path:
             prefix = fs_path.relparts(filter_info, self.fs_path)
-            obj = obj.get(self.odb, prefix)
+            obj = obj.get_obj(self.odb, prefix)
 
         return obj
 
@@ -843,13 +878,12 @@ class Output:
 
         upload = not (update and from_fs.isdir(from_info))
         jobs = jobs or min((from_fs.jobs, odb.fs.jobs))
-        staging, self.meta, obj = ostage(
+        staging, self.meta, obj = build(
             odb,
             from_info,
             from_fs,
             "md5",
             upload=upload,
-            jobs=jobs,
             no_progress_bar=no_progress_bar,
         )
         otransfer(
@@ -885,7 +919,7 @@ class Output:
         if not self.is_dir_checksum:
             raise DvcException("cannot get dir cache for file checksum")
 
-        obj = self.odb.get(self.hash_info)
+        obj = self.odb.get(self.hash_info.value)
         try:
             ocheck(self.odb, obj)
         except FileNotFoundError:
@@ -913,10 +947,10 @@ class Output:
         except RemoteMissingDepsError:  # pylint: disable=try-except-raise
             raise
         except DvcException:
-            logger.debug(f"failed to pull cache for '{self}'")
+            logger.debug("failed to pull cache for '%s'", self)
 
         try:
-            ocheck(self.odb, self.odb.get(self.hash_info))
+            ocheck(self.odb, self.odb.get(self.hash_info.value))
         except FileNotFoundError:
             msg = (
                 "Missing cache for directory '{}'. "
@@ -973,7 +1007,7 @@ class Output:
         else:
             obj = self.get_obj(filter_info=kwargs.get("filter_info"))
             if not obj:
-                obj = self.odb.get(self.hash_info)
+                obj = self.odb.get(self.hash_info.value)
 
         if not obj:
             return {}
@@ -1024,7 +1058,7 @@ class Output:
         other = out.dumpd()
 
         ignored = [
-            self.fs.PARAM_CHECKSUM,
+            self.hash_name,
             Meta.PARAM_SIZE,
             Meta.PARAM_NFILES,
         ]
@@ -1044,8 +1078,8 @@ class Output:
             )
 
     def merge(self, ancestor, other):
-        from dvc_data.tree import MergeError as TreeMergeError
-        from dvc_data.tree import du, merge
+        from dvc_data.hashfile.tree import MergeError as TreeMergeError
+        from dvc_data.hashfile.tree import du, merge
 
         assert other
 
@@ -1065,7 +1099,7 @@ class Output:
         except TreeMergeError as exc:
             raise MergeError(str(exc)) from exc
 
-        self.odb.add(merged.fs_path, merged.fs, merged.hash_info)
+        self.odb.add(merged.path, merged.fs, merged.oid)
 
         self.hash_info = merged.hash_info
         self.meta = Meta(
@@ -1087,24 +1121,29 @@ class Output:
 
     @property
     def is_plot(self) -> bool:
-        return bool(self.plot)
+        return bool(self.plot) or bool(self.live)
 
 
+META_SCHEMA = {
+    Meta.PARAM_SIZE: int,
+    Meta.PARAM_NFILES: int,
+    Meta.PARAM_ISEXEC: bool,
+    Meta.PARAM_VERSION_ID: str,
+}
+
 ARTIFACT_SCHEMA = {
     **CHECKSUMS_SCHEMA,
+    **META_SCHEMA,
     Required(Output.PARAM_PATH): str,
     Output.PARAM_PLOT: bool,
     Output.PARAM_PERSIST: bool,
     Output.PARAM_CHECKPOINT: bool,
-    Meta.PARAM_SIZE: int,
-    Meta.PARAM_NFILES: int,
-    Meta.PARAM_ISEXEC: bool,
 }
 
 SCHEMA = {
     **ARTIFACT_SCHEMA,
+    **ANNOTATION_SCHEMA,
     Output.PARAM_CACHE: bool,
     Output.PARAM_METRIC: Output.METRIC_SCHEMA,
-    Output.PARAM_DESC: str,
     Output.PARAM_REMOTE: str,
 }

dvc/parsing/__init__.py

@@ -303,7 +303,7 @@ class EntryDefinition:
     ) -> DictStr:
         try:
             return context.resolve(
-                value, skip_interpolation_checks=skip_checks
+                value, skip_interpolation_checks=skip_checks, key=key
             )
         except (ParseError, KeyNotInContext) as exc:
             format_and_raise(

dvc/parsing/context.py

@@ -1,5 +1,4 @@
 import logging
-import os
 from abc import ABC, abstractmethod
 from collections import defaultdict
 from collections.abc import Mapping, MutableMapping, MutableSequence, Sequence
@@ -18,6 +17,7 @@ from dvc.parsing.interpolate import (
     normalize_key,
     recurse,
     str_interpolate,
+    validate_value,
 )
 
 logger = logging.getLogger(__name__)
@@ -159,7 +159,7 @@ class Value(Node):
 PRIMITIVES = (int, float, str, bytes, bool)
 
 
-class Container(Node, ABC):
+class Container(Node, ABC):  # noqa: B024
     meta: Meta
     data: Union[list, dict]
     _key_transform = staticmethod(identity)
@@ -356,17 +356,14 @@ class Context(CtxDict):
     def load_from(
         cls, fs, path: str, select_keys: List[str] = None
     ) -> "Context":
-        from dvc.utils.serialize import LOADERS
+        from dvc.utils.serialize import load_path
 
         if not fs.exists(path):
             raise ParamsLoadError(f"'{path}' does not exist")
         if fs.isdir(path):
             raise ParamsLoadError(f"'{path}' is a directory")
 
-        _, ext = os.path.splitext(path)
-        loader = LOADERS[ext]
-
-        data = loader(path, fs=fs)
+        data = load_path(path, fs)
         if not isinstance(data, Mapping):
             typ = type(data).__name__
             raise ParamsLoadError(
@@ -506,7 +503,7 @@ class Context(CtxDict):
                 self.data.pop(key, None)
 
     def resolve(
-        self, src, unwrap=True, skip_interpolation_checks=False
+        self, src, unwrap=True, skip_interpolation_checks=False, key=None
     ) -> Any:
         """Recursively resolves interpolation and returns resolved data.
 
@@ -522,10 +519,10 @@ class Context(CtxDict):
         {'lst': [1, 2, 3]}
         """
         func = recurse(self.resolve_str)
-        return func(src, unwrap, skip_interpolation_checks)
+        return func(src, unwrap, skip_interpolation_checks, key)
 
     def resolve_str(
-        self, src: str, unwrap=True, skip_interpolation_checks=False
+        self, src: str, unwrap=True, skip_interpolation_checks=False, key=None
     ) -> str:
         """Resolves interpolated string to it's original value,
         or in case of multiple interpolations, a combined string.
@@ -543,10 +540,12 @@ class Context(CtxDict):
             expr = get_expression(
                 matches[0], skip_checks=skip_interpolation_checks
             )
-            return self.select(expr, unwrap=unwrap)
+            value = self.select(expr, unwrap=unwrap)
+            validate_value(value, key)
+            return value
         # but not "${num} days"
         return str_interpolate(
-            src, matches, self, skip_checks=skip_interpolation_checks
+            src, matches, self, skip_checks=skip_interpolation_checks, key=key
         )
 
 

dvc/parsing/interpolate.py

@@ -1,11 +1,12 @@
 import re
 import typing
-from collections.abc import Mapping
+from collections.abc import Iterable, Mapping
 from functools import singledispatch
 
 from funcy import memoize, rpartial
 
 from dvc.exceptions import DvcException
+from dvc.utils.flatten import flatten
 
 if typing.TYPE_CHECKING:
     from typing import List, Match
@@ -80,6 +81,45 @@ def _(obj: bool):
     return "true" if obj else "false"
 
 
+@to_str.register(dict)
+def _(obj: dict):
+    from dvc.config import Config
+
+    config = Config().get("parsing", {})
+
+    result = ""
+    for k, v in flatten(obj).items():
+
+        if isinstance(v, bool):
+            if v:
+                result += f"--{k} "
+            else:
+                if config.get("bool", "store_true") == "boolean_optional":
+                    result += f"--no-{k} "
+
+        elif isinstance(v, str):
+            result += f"--{k} '{v}' "
+
+        elif isinstance(v, Iterable):
+            for n, i in enumerate(v):
+                if isinstance(i, str):
+                    i = f"'{i}'"
+                elif isinstance(i, Iterable):
+                    raise ParseError(
+                        f"Cannot interpolate nested iterable in '{k}'"
+                    )
+
+                if config.get("list", "nargs") == "append":
+                    result += f"--{k} {i} "
+                else:
+                    result += f"{i} " if n > 0 else f"--{k} {i} "
+
+        else:
+            result += f"--{k} {v} "
+
+    return result.rstrip()
+
+
 def _format_exc_msg(exc: "ParseException"):
     from pyparsing import ParseException
 
@@ -148,23 +188,33 @@ def get_expression(match: "Match", skip_checks: bool = False):
     return inner if skip_checks else parse_expr(inner)
 
 
+def validate_value(value, key):
+    from .context import PRIMITIVES
+
+    not_primitive = value is not None and not isinstance(value, PRIMITIVES)
+    not_foreach = key is not None and "foreach" not in key
+    if not_primitive and not_foreach:
+        if isinstance(value, dict):
+            if key == "cmd":
+                return True
+        raise ParseError(
+            f"Cannot interpolate data of type '{type(value).__name__}'"
+        )
+
+
 def str_interpolate(
     template: str,
     matches: "List[Match]",
     context: "Context",
     skip_checks: bool = False,
+    key=None,
 ):
-    from .context import PRIMITIVES
-
     index, buf = 0, ""
     for match in matches:
         start, end = match.span(0)
         expr = get_expression(match, skip_checks=skip_checks)
         value = context.select(expr, unwrap=True)
-        if value is not None and not isinstance(value, PRIMITIVES):
-            raise ParseError(
-                f"Cannot interpolate data of type '{type(value).__name__}'"
-            )
+        validate_value(value, key)
         buf += template[index:start] + to_str(value)
         index = end
     buf += template[index:]

dvc/prompt.py

@@ -52,5 +52,5 @@ def password(statement):
     Returns:
         str: password entered by the user.
     """
-    logger.info(f"{statement}: ")
+    logger.info("%s: ", statement)
     return getpass("")

dvc/render/__init__.py

@@ -1,6 +1,7 @@
 INDEX_FIELD = "step"
 REVISION_FIELD = "rev"
 FILENAME_FIELD = "filename"
+VERSION_FIELD = "dvc_data_version_info"
 REVISIONS_KEY = "revisions"
 TYPE_KEY = "type"
 SRC_FIELD = "src"

dvc/render/convert.py

@@ -3,8 +3,8 @@ from collections import defaultdict
 from typing import Dict, List, Union
 
 from dvc.render import REVISION_FIELD, REVISIONS_KEY, SRC_FIELD, TYPE_KEY
-from dvc.render.image_converter import ImageConverter
-from dvc.render.vega_converter import VegaConverter
+from dvc.render.converter.image import ImageConverter
+from dvc.render.converter.vega import VegaConverter
 
 
 def _get_converter(
@@ -20,20 +20,6 @@ def _get_converter(
     raise ValueError(f"Invalid renderer class {renderer_class}")
 
 
-def to_datapoints(renderer_class, data: Dict, props: Dict):
-    converter = _get_converter(renderer_class, props)
-    datapoints: List[Dict] = []
-    final_props: Dict = {}
-    for revision, rev_data in data.items():
-        for filename, file_data in rev_data.get("data", {}).items():
-            if "data" in file_data:
-                processed, final_props = converter.convert(
-                    file_data.get("data"), revision, filename
-                )
-                datapoints.extend(processed)
-    return datapoints, final_props
-
-
 def _group_by_rev(datapoints):
     grouped = defaultdict(list)
     for datapoint in datapoints:

dvc/render/converter/__init__.py

@@ -0,0 +1,9 @@
+from typing import Dict, Optional
+
+
+class Converter:
+    def __init__(self, plot_properties: Optional[Dict] = None):
+        self.plot_properties = plot_properties or {}
+
+    def convert(self, data, revision: str, filename: str, **kwargs):
+        raise NotImplementedError

dvc/render/converter/image.py

@@ -1,17 +1,16 @@
 import base64
 import os
-from typing import TYPE_CHECKING, Dict, List, Optional, Tuple
+from typing import TYPE_CHECKING, Dict, List, Tuple
 
 from dvc.render import FILENAME_FIELD, REVISION_FIELD, SRC_FIELD
 
+from . import Converter
+
 if TYPE_CHECKING:
     from dvc.types import StrPath
 
 
-class ImageConverter:
-    def __init__(self, plot_properties: Optional[Dict] = None):
-        self.plot_properties = plot_properties or {}
-
+class ImageConverter(Converter):
     @staticmethod
     def _write_image(
         path: "StrPath",
@@ -20,7 +19,8 @@ class ImageConverter:
         image_data: bytes,
     ) -> "StrPath":
         img_path = os.path.join(
-            path, f"{revision}_{filename.replace(os.sep, '_')}"
+            path,
+            f"{revision}_{filename}".replace(os.sep, "_").replace("/", "_"),
         )
         with open(img_path, "wb") as fd:
             fd.write(image_data)
@@ -35,7 +35,7 @@ class ImageConverter:
         return f"data:image;base64,{base64_str}"
 
     def convert(
-        self, data: bytes, revision, filename
+        self, data, revision: str, filename: str, **kwargs
     ) -> Tuple[List[Dict], Dict]:
         """
         Convert the DVC Plots content to DVC Render datapoints.

dvc/render/converter/vega.py

@@ -0,0 +1,316 @@
+from copy import deepcopy
+from functools import partial
+from typing import Dict, Iterable, List, Optional, Set, Union
+
+from funcy import first, project
+
+from dvc.exceptions import DvcException
+from dvc.render import (
+    FILENAME_FIELD,
+    INDEX_FIELD,
+    REVISION_FIELD,
+    VERSION_FIELD,
+)
+
+from . import Converter
+
+
+class FieldsNotFoundError(DvcException):
+    def __init__(self, expected_fields, found_fields):
+        expected_str = ", ".join(expected_fields)
+        found_str = ", ".join(found_fields)
+        super().__init__(
+            f"Could not find all provided fields ('{expected_str}') "
+            f"in data fields ('{found_str}')."
+        )
+
+
+class PlotDataStructureError(DvcException):
+    def __init__(self):
+        super().__init__(
+            "Plot data extraction failed. Please see "
+            "https://man.dvc.org/plots for supported data formats."
+        )
+
+
+def _filter_fields(
+    datapoints: List[Dict], fields: Set, **kwargs
+) -> List[Dict]:
+    if not fields:
+        return datapoints
+    assert isinstance(fields, set)
+
+    new_data = []
+    for data_point in datapoints:
+        keys = set(data_point.keys())
+        if not fields <= keys:
+            raise FieldsNotFoundError(fields, keys)
+
+        new_data.append(project(data_point, fields))
+
+    return new_data
+
+
+def _lists(dictionary: Dict):
+    for _, value in dictionary.items():
+        if isinstance(value, dict):
+            yield from _lists(value)
+        elif isinstance(value, list):
+            yield value
+
+
+def _find_first_list(
+    data: Union[Dict, List], fields: Set, **kwargs
+) -> List[Dict]:
+    fields = fields or set()
+
+    if not isinstance(data, dict):
+        return data
+
+    for lst in _lists(data):
+        if (
+            all(isinstance(dp, dict) for dp in lst)
+            # if fields is empty, it will match any set
+            and set(first(lst).keys()) & fields == fields
+        ):
+            return lst
+
+    raise PlotDataStructureError()
+
+
+def _append_index(datapoints: List[Dict], **kwargs) -> List[Dict]:
+    if INDEX_FIELD in first(datapoints).keys():
+        return datapoints
+
+    for index, data_point in enumerate(datapoints):
+        data_point[INDEX_FIELD] = index
+    return datapoints
+
+
+class VegaConverter(Converter):
+    """
+    Class that takes care of converting unspecified data blob
+    (Dict or List[Dict]) into datapoints (List[Dict]).
+    If some properties that are required by Template class are missing
+    ('x', 'y') it will attempt to fill in the blanks.
+    """
+
+    def __init__(self, plot_properties: Optional[Dict] = None):
+        super().__init__(plot_properties)
+        self.inferred_properties: Dict = {}
+
+        self.steps = []
+
+        self._infer_x()
+        self._infer_fields()
+
+        self.steps.append(
+            (
+                "find_data",
+                partial(
+                    _find_first_list,
+                    fields=self.inferred_properties.get("fields", set())
+                    - {INDEX_FIELD},
+                ),
+            )
+        )
+
+        if not self.plot_properties.get("x", None):
+            self.steps.append(("append_index", partial(_append_index)))
+
+        self.steps.append(
+            (
+                "filter_fields",
+                partial(
+                    _filter_fields,
+                    fields=self.inferred_properties.get("fields", set()),
+                ),
+            )
+        )
+
+        self.steps.append(
+            (
+                "generate_y",
+                partial(
+                    self._generate_y_values,
+                    y_values=self.plot_properties.get("y", None),
+                ),
+            )
+        )
+
+    def _infer_x(self):
+        if not self.plot_properties.get("x", None):
+            self.inferred_properties["x"] = INDEX_FIELD
+
+    def skip_step(self, name: str):
+        self.steps = [(_name, fn) for _name, fn in self.steps if _name != name]
+
+    def _infer_fields(self):
+        fields = self.plot_properties.get("fields", set())
+        if fields:
+            fields = {
+                *fields,
+                self.plot_properties.get("x", None),
+                self.plot_properties.get("y", None),
+                self.inferred_properties.get("x", None),
+            } - {None}
+            self.inferred_properties["fields"] = fields
+
+    def _infer_y(self, datapoints: List[Dict]):
+        if "y" not in self.plot_properties:
+            data_fields = list(first(datapoints))
+            skip = (
+                REVISION_FIELD,
+                self.plot_properties.get("x", None)
+                or self.inferred_properties.get("x"),
+                FILENAME_FIELD,
+                VERSION_FIELD,
+            )
+            inferred_y = first(
+                f for f in reversed(data_fields) if f not in skip
+            )
+            if "y" in self.inferred_properties:
+                previous_y = self.inferred_properties["y"]
+                if previous_y != inferred_y:
+                    raise DvcException(
+                        f"Inferred y ('{inferred_y}' value does not match"
+                        f"previously matched one ('f{previous_y}')."
+                    )
+            else:
+                self.inferred_properties["y"] = inferred_y
+
+    def convert(
+        self,
+        data,
+        revision: str,
+        filename: str,
+        skip: List = None,
+        **kwargs,
+    ):
+        """
+        Convert the data. Fill necessary fields ('x', 'y') and return both
+        generated datapoints and updated properties.
+        """
+        if not skip:
+            skip = []
+
+        processed = deepcopy(data)
+
+        for step_name, step in self.steps:
+
+            if step_name not in skip:
+                processed = step(  # type: ignore
+                    processed,
+                    revision=revision,
+                    filename=filename,
+                )
+
+        self._infer_y(processed)  # type: ignore
+        return processed, {**self.plot_properties, **self.inferred_properties}
+
+    def _generate_y_values(  # noqa: C901
+        self,
+        datapoints: List[Dict],
+        y_values: Optional[Union[str, List, Dict]],
+        revision: str,
+        filename: str,
+        **kwargs,
+    ) -> List[Dict]:
+
+        result = []
+        properties_update = {}
+
+        def _add_version_info(datapoint, version_info):
+            tmp = datapoint.copy()
+            tmp[VERSION_FIELD] = version_info
+            tmp[REVISION_FIELD] = version_info["revision"]
+            return tmp
+
+        def _version_info(revision, filename=None, field=None):
+            res = {"revision": revision}
+            if filename is not None:
+                res["filename"] = filename
+            if field is not None:
+                res["field"] = field
+            return res
+
+        def _generate_y(datapoint, field):
+            tmp = datapoint.copy()
+            tmp["dvc_inferred_y_value"] = datapoint[field]
+            tmp = _add_version_info(
+                tmp, _version_info(revision, filename, field)
+            )
+            if (
+                "y_label" not in properties_update
+                and "y_label" not in self.plot_properties
+            ):
+                properties_update["y_label"] = "y"
+
+            properties_update["y"] = "dvc_inferred_y_value"
+
+            return tmp
+
+        if not y_values:
+            for dp in datapoints:
+                result.append(
+                    _add_version_info(dp, _version_info(revision, filename))
+                )
+
+        if isinstance(y_values, str):
+            for datapoint in datapoints:
+                result.append(
+                    _add_version_info(
+                        datapoint, _version_info(revision, filename, y_values)
+                    )
+                )
+
+        if isinstance(y_values, list):
+            for datapoint in datapoints:
+                for y_val in y_values:
+                    if y_val in datapoint:
+                        result.append(_generate_y(datapoint, y_val))
+
+        if isinstance(y_values, dict):
+
+            def _to_set(values: Iterable):
+                result = set()
+                for val in values:
+                    if isinstance(val, list):
+                        for elem in val:
+                            result.add(elem)
+                    else:
+                        result.add(val)
+
+                return result
+
+            all_fields = _to_set(y_values.values())
+            if (
+                all([isinstance(field, str) for field in all_fields])
+                and len(all_fields) == 1
+            ):
+                # if we use the same field from all files,
+                # we dont have to generate it
+                for datapoint in datapoints:
+                    result.append(
+                        _add_version_info(
+                            datapoint, _version_info(revision, filename)
+                        )
+                    )
+                properties_update.update({"y": all_fields.pop()})
+            else:
+                for def_filename, val in y_values.items():
+                    if isinstance(val, str):
+                        fields = [val]
+                    if isinstance(val, list):
+                        fields = val
+                    for datapoint in datapoints:
+                        for field in fields:
+                            if field in datapoint and def_filename in filename:
+                                result.append(_generate_y(datapoint, field))
+
+        self.inferred_properties = {
+            **self.inferred_properties,
+            **properties_update,
+        }
+
+        return result

dvc/render/match.py

@@ -1,58 +1,110 @@
-from typing import TYPE_CHECKING, Dict, Optional
+from collections import defaultdict
+from typing import TYPE_CHECKING, Dict, List, Optional
 
+import dpath.options
 import dpath.util
+from funcy import last
 
-from dvc_render import RENDERERS
+from dvc.repo.plots import infer_data_sources
+from dvc.utils.plots import get_plot_id
 
-from .convert import to_datapoints
+from .convert import _get_converter
 
 if TYPE_CHECKING:
     from dvc.types import StrPath
 
-
-def group_by_filename(plots_data: Dict) -> Dict:
-    grouped: Dict[str, Dict] = {}
-
-    for revision in plots_data.keys():
-        data = plots_data[revision].get("data", {})
-        for file in data.keys():
-            content = data.get(file)
-            if content:
-                dpath.util.new(
-                    grouped, [file, revision, "data", file], content
-                )
-
-    return grouped
+dpath.options.ALLOW_EMPTY_STRING_KEYS = True
 
 
-def squash_plots_properties(data: Dict) -> Dict:
-    resolved: Dict[str, str] = {}
-    for rev_data in data.values():
-        for file_data in rev_data.get("data", {}).values():
-            props = file_data.get("props", {})
-            resolved = {**resolved, **props}
+def _squash_plots_properties(data: List) -> Dict:
+    configs = [last(group) for group in data]
+    resolved: Dict = {}
+    for config in reversed(configs):
+        resolved = {**resolved, **config}
     return resolved
 
 
-def match_renderers(
-    plots_data,
-    out: Optional["StrPath"] = None,
+class PlotsData:
+    def __init__(self, data: Dict):
+        self.data = data
+
+    def group_definitions(self):
+        groups = defaultdict(list)
+        for rev, rev_content in self.data.items():
+            for config_file, config_file_content in (
+                rev_content.get("definitions", {}).get("data", {}).items()
+            ):
+                for plot_id, plot_definition in config_file_content.get(
+                    "data", {}
+                ).items():
+                    full_id = get_plot_id(plot_id, config_file)
+                    groups[full_id].append((rev, plot_id, plot_definition))
+        return dict(groups)
+
+    def get_definition_data(self, target_files, rev):
+        result = []
+        for file in target_files:
+            file_content = (
+                self.data.get(rev, {})
+                .get("sources", {})
+                .get("data", {})
+                .get(file, {})
+                .get("data", {})
+            )
+            if file_content:
+                result.append((file, file_content))
+
+        return result
+
+
+def match_defs_renderers(
+    data,
+    out=None,
     templates_dir: Optional["StrPath"] = None,
 ):
+
+    from dvc_render import ImageRenderer, VegaRenderer
+
+    plots_data = PlotsData(data)
     renderers = []
-    for filename, group in group_by_filename(plots_data).items():
-        plot_properties = squash_plots_properties(group)
-        for renderer_class in RENDERERS:
-            if renderer_class.matches(filename, plot_properties):
-                if out is not None:
-                    plot_properties["out"] = out
-                if templates_dir is not None:
-                    plot_properties["template_dir"] = templates_dir
-                datapoints, plot_properties = to_datapoints(
-                    renderer_class, group, plot_properties
+    renderer_cls = None
+    for plot_id, group in plots_data.group_definitions().items():
+        plot_datapoints: List[Dict] = []
+        props = _squash_plots_properties(group)
+        final_props: Dict = {}
+
+        if out is not None:
+            props["out"] = out
+        if templates_dir is not None:
+            props["template_dir"] = templates_dir
+
+        for rev, inner_id, plot_definition in group:
+            plot_sources = infer_data_sources(inner_id, plot_definition)
+            definitions_data = plots_data.get_definition_data(
+                plot_sources, rev
+            )
+
+            if ImageRenderer.matches(inner_id, None):
+                renderer_cls = ImageRenderer
+                renderer_id = inner_id
+            else:
+                renderer_cls = VegaRenderer
+                renderer_id = plot_id
+
+            converter = _get_converter(renderer_cls, props)
+
+            for filename, plot_data in definitions_data:
+                dps, final_props = converter.convert(
+                    revision=rev,
+                    filename=filename,
+                    data=plot_data,
                 )
-                renderers.append(
-                    renderer_class(datapoints, filename, **plot_properties)
-                )
-                break
+                plot_datapoints.extend(dps)
+
+        if "title" not in final_props:
+            final_props["title"] = renderer_id
+        if renderer_cls is not None:
+            renderers.append(
+                renderer_cls(plot_datapoints, renderer_id, **final_props)
+            )
     return renderers

dvc/render/vega_converter.py

@@ -1,183 +0,0 @@
-from copy import deepcopy
-from functools import partial
-from typing import Dict, List, Optional, Set, Union
-
-from funcy import first, project
-
-from dvc.exceptions import DvcException
-from dvc.render import FILENAME_FIELD, INDEX_FIELD, REVISION_FIELD
-
-
-class FieldsNotFoundError(DvcException):
-    def __init__(self, expected_fields, found_fields):
-        expected_str = ", ".join(expected_fields)
-        found_str = ", ".join(found_fields)
-        super().__init__(
-            f"Could not find all provided fields ('{expected_str}') "
-            f"in data fields ('{found_str}')."
-        )
-
-
-class PlotDataStructureError(DvcException):
-    def __init__(self):
-        super().__init__(
-            "Plot data extraction failed. Please see "
-            "https://man.dvc.org/plots for supported data formats."
-        )
-
-
-def _filter_fields(datapoints: List[Dict], fields: Set) -> List[Dict]:
-    if not fields:
-        return datapoints
-    assert isinstance(fields, set)
-
-    new_data = []
-    for data_point in datapoints:
-        keys = set(data_point.keys())
-        if not fields <= keys:
-            raise FieldsNotFoundError(fields, keys)
-
-        new_data.append(project(data_point, fields))
-
-    return new_data
-
-
-def _lists(dictionary: Dict):
-    for _, value in dictionary.items():
-        if isinstance(value, dict):
-            yield from _lists(value)
-        elif isinstance(value, list):
-            yield value
-
-
-def _find_first_list(data: Union[Dict, List], fields: Set) -> List[Dict]:
-    fields = fields or set()
-
-    if not isinstance(data, dict):
-        return data
-
-    for lst in _lists(data):
-        if (
-            all(isinstance(dp, dict) for dp in lst)
-            # if fields is empty, it will match any set
-            and set(first(lst).keys()) & fields == fields
-        ):
-            return lst
-
-    raise PlotDataStructureError()
-
-
-def _append_index(datapoints: List[Dict]) -> List[Dict]:
-    if INDEX_FIELD in first(datapoints).keys():
-        return datapoints
-
-    for index, data_point in enumerate(datapoints):
-        data_point[INDEX_FIELD] = index
-    return datapoints
-
-
-class VegaConverter:
-    """
-    Class that takes care of converting unspecified data blob
-    (Dict or List[Dict]) into datapoints (List[Dict]).
-    If some properties that are required by Template class are missing
-    ('x', 'y') it will attempt to fill in the blanks.
-    """
-
-    def __init__(self, plot_properties: Optional[Dict] = None):
-        plot_properties = plot_properties or {}
-        self.props = deepcopy(plot_properties)
-        self.inferred_props: Dict = {}
-
-        self.steps = []
-
-        self._infer_x()
-        self._infer_fields()
-
-        self.steps.append(
-            (
-                "find_data",
-                partial(
-                    _find_first_list,
-                    fields=self.inferred_props.get("fields", set())
-                    - {INDEX_FIELD},
-                ),
-            )
-        )
-
-        if not self.props.get("x", None):
-            self.steps.append(("append_index", partial(_append_index)))
-
-        self.steps.append(
-            (
-                "filter_fields",
-                partial(
-                    _filter_fields,
-                    fields=self.inferred_props.get("fields", set()),
-                ),
-            )
-        )
-
-    def _infer_x(self):
-        if not self.props.get("x", None):
-            self.inferred_props["x"] = INDEX_FIELD
-
-    def skip_step(self, name: str):
-        self.steps = [(_name, fn) for _name, fn in self.steps if _name != name]
-
-    def _infer_fields(self):
-        fields = self.props.get("fields", set())
-        if fields:
-            fields = {
-                *fields,
-                self.props.get("x", None),
-                self.props.get("y", None),
-                self.inferred_props.get("x", None),
-            } - {None}
-            self.inferred_props["fields"] = fields
-
-    def _infer_y(self, datapoints: List[Dict]):
-        if "y" not in self.props:
-            data_fields = list(first(datapoints))
-            skip = (
-                REVISION_FIELD,
-                self.props.get("x", None) or self.inferred_props.get("x"),
-            )
-            inferred_y = first(
-                f for f in reversed(data_fields) if f not in skip
-            )
-            if "y" in self.inferred_props:
-                previous_y = self.inferred_props["y"]
-                if previous_y != inferred_y:
-                    raise DvcException(
-                        f"Inferred y ('{inferred_y}' value does not match"
-                        f"previously matched one ('f{previous_y}')."
-                    )
-            else:
-                self.inferred_props["y"] = inferred_y
-
-    def convert(
-        self,
-        data: Dict,
-        revision: Optional[str] = None,
-        filename: Optional[str] = None,
-    ):
-        """
-        Convert the data. Fill necessary fields ('x', 'y') and return both
-        generated datapoints and updated properties.
-        """
-        processed = deepcopy(data)
-
-        for _, step in self.steps:
-            processed = step(processed)  # type: ignore
-
-        self._infer_y(processed)  # type: ignore
-
-        if revision:
-            for datapoint in processed:
-                datapoint[REVISION_FIELD] = revision
-        if filename:
-            for datapoint in processed:
-                datapoint[FILENAME_FIELD] = filename
-
-        return processed, {**self.props, **self.inferred_props}

dvc/repo/__init__.py

@@ -3,7 +3,7 @@ import os
 from collections import defaultdict
 from contextlib import contextmanager
 from functools import wraps
-from typing import TYPE_CHECKING, Callable, Optional, Set
+from typing import TYPE_CHECKING, Callable, List, Optional
 
 from funcy import cached_property
 
@@ -18,7 +18,7 @@ if TYPE_CHECKING:
     from dvc.fs import FileSystem
     from dvc.repo.scm_context import SCMContext
     from dvc.scm import Base
-    from dvc_objects.file import HashFile
+    from dvc.stage import Stage
 
 logger = logging.getLogger(__name__)
 
@@ -54,28 +54,30 @@ def locked(f):
 class Repo:
     DVC_DIR = ".dvc"
 
-    from dvc.repo.add import add
-    from dvc.repo.checkout import checkout
-    from dvc.repo.commit import commit
+    from dvc.repo.add import add  # type: ignore[misc]
+    from dvc.repo.checkout import checkout  # type: ignore[misc]
+    from dvc.repo.commit import commit  # type: ignore[misc]
     from dvc.repo.destroy import destroy  # type: ignore[misc]
-    from dvc.repo.diff import diff
-    from dvc.repo.fetch import fetch
+    from dvc.repo.diff import diff  # type: ignore[misc]
+    from dvc.repo.fetch import fetch  # type: ignore[misc]
     from dvc.repo.freeze import freeze, unfreeze  # type: ignore[misc]
-    from dvc.repo.gc import gc
+    from dvc.repo.gc import gc  # type: ignore[misc]
     from dvc.repo.get import get as _get  # type: ignore[misc]
     from dvc.repo.get_url import get_url as _get_url  # type: ignore[misc]
     from dvc.repo.imp import imp  # type: ignore[misc]
-    from dvc.repo.imp_url import imp_url
+    from dvc.repo.imp_url import imp_url  # type: ignore[misc]
     from dvc.repo.install import install  # type: ignore[misc]
     from dvc.repo.ls import ls as _ls  # type: ignore[misc]
-    from dvc.repo.move import move
-    from dvc.repo.pull import pull
-    from dvc.repo.push import push
-    from dvc.repo.remove import remove
-    from dvc.repo.reproduce import reproduce
-    from dvc.repo.run import run
-    from dvc.repo.status import status
-    from dvc.repo.update import update
+    from dvc.repo.move import move  # type: ignore[misc]
+    from dvc.repo.pull import pull  # type: ignore[misc]
+    from dvc.repo.push import push  # type: ignore[misc]
+    from dvc.repo.remove import remove  # type: ignore[misc]
+    from dvc.repo.reproduce import reproduce  # type: ignore[misc]
+    from dvc.repo.run import run  # type: ignore[misc]
+    from dvc.repo.status import status  # type: ignore[misc]
+    from dvc.repo.update import update  # type: ignore[misc]
+
+    from .data import status as data_status  # type: ignore[misc]
 
     ls = staticmethod(_ls)
     get = staticmethod(_get)
@@ -127,19 +129,17 @@ class Repo:
 
         import hashlib
 
-        from dvc.utils.fs import makedirs
-
         root_dir_hash = hashlib.sha224(
             self.root_dir.encode("utf-8")
         ).hexdigest()
 
-        db_dir = os.path.join(
+        db_dir = self.fs.path.join(
             base_db_dir,
             self.DVC_DIR,
-            f"{os.path.basename(self.root_dir)}-{root_dir_hash[0:7]}",
+            f"{self.fs.path.name(self.root_dir)}-{root_dir_hash[0:7]}",
         )
 
-        makedirs(db_dir, exist_ok=True)
+        self.fs.makedirs(db_dir, exist_ok=True)
         return db_dir
 
     def __init__(
@@ -156,7 +156,7 @@ class Repo:
     ):
         from dvc.config import Config
         from dvc.data_cloud import DataCloud
-        from dvc.fs import GitFileSystem, localfs
+        from dvc.fs import GitFileSystem, LocalFileSystem, localfs
         from dvc.lock import LockNoop, make_lock
         from dvc.odbmgr import ODBManager
         from dvc.repo.metrics import Metrics
@@ -165,7 +165,7 @@ class Repo:
         from dvc.repo.stage import StageLoad
         from dvc.scm import SCM
         from dvc.stage.cache import StageCache
-        from dvc_objects.state import State, StateNoop
+        from dvc_data.hashfile.state import State, StateNoop
 
         self.url = url
         self._fs_conf = {"repo_factory": repo_factory}
@@ -199,18 +199,23 @@ class Repo:
             self.odb = ODBManager(self)
             self.tmp_dir = None
         else:
-            from dvc.utils.fs import makedirs
+            self.fs.makedirs(self.tmp_dir, exist_ok=True)
 
-            makedirs(self.tmp_dir, exist_ok=True)
-            self.lock = make_lock(
-                os.path.join(self.tmp_dir, "lock"),
-                tmp_dir=self.tmp_dir,
-                hardlink_lock=self.config["core"].get("hardlink_lock", False),
-                friendly=True,
-            )
+            if isinstance(self.fs, LocalFileSystem):
+                self.lock = make_lock(
+                    self.fs.path.join(self.tmp_dir, "lock"),
+                    tmp_dir=self.tmp_dir,
+                    hardlink_lock=self.config["core"].get(
+                        "hardlink_lock", False
+                    ),
+                    friendly=True,
+                )
+                state_db_dir = self._get_database_dir("state")
+                self.state = State(self.root_dir, state_db_dir, self.dvcignore)
+            else:
+                self.lock = LockNoop()
+                self.state = StateNoop()
 
-            state_db_dir = self._get_database_dir("state")
-            self.state = State(self.root_dir, state_db_dir, self.dvcignore)
             self.odb = ODBManager(self)
 
             self.stage_cache = StageCache(self)
@@ -363,13 +368,13 @@ class Repo:
         )
 
     def unprotect(self, target):
-        return self.odb.local.unprotect(target)
+        return self.odb.repo.unprotect(target)
 
     def _ignore(self):
         flist = [self.config.files["local"], self.tmp_dir]
 
-        if path_isin(self.odb.local.cache_dir, self.root_dir):
-            flist += [self.odb.local.cache_dir]
+        if path_isin(self.odb.repo.path, self.root_dir):
+            flist += [self.odb.repo.path]
 
         for file in flist:
             self.scm_context.ignore(file)
@@ -387,12 +392,14 @@ class Repo:
         all_tags=False,
         all_commits=False,
         all_experiments=False,
+        commit_date: Optional[str] = None,
         remote=None,
         force=False,
         jobs=None,
         recursive=False,
         used_run_cache=None,
         revs=None,
+        num=1,
     ):
         """Get the stages related to the given target and collect
         the `info` of its outputs.
@@ -411,21 +418,14 @@ class Repo:
         """
         used = defaultdict(set)
 
-        def _add_suffix(objs: Set["HashFile"], suffix: str) -> None:
-            from itertools import chain
-
-            from dvc_data import iterobjs
-
-            for obj in chain.from_iterable(map(iterobjs, objs)):
-                if obj.name is not None:
-                    obj.name += suffix
-
-        for branch in self.brancher(
+        for _ in self.brancher(
             revs=revs,
             all_branches=all_branches,
             all_tags=all_tags,
             all_commits=all_commits,
             all_experiments=all_experiments,
+            commit_date=commit_date,
+            num=num,
         ):
             for odb, objs in self.index.used_objs(
                 targets,
@@ -435,8 +435,6 @@ class Repo:
                 recursive=recursive,
                 with_deps=with_deps,
             ).items():
-                if branch:
-                    _add_suffix(objs, f" ({branch})")
                 used[odb].update(objs)
 
         if used_run_cache:
@@ -447,6 +445,44 @@ class Repo:
 
         return used
 
+    def partial_imports(
+        self,
+        targets=None,
+        all_branches=False,
+        all_tags=False,
+        all_commits=False,
+        all_experiments=False,
+        commit_date: Optional[str] = None,
+        recursive=False,
+        revs=None,
+        num=1,
+    ) -> List["Stage"]:
+        """Get the stages related to the given target and collect dependencies
+        which are missing outputs.
+
+        This is useful to retrieve files which have been imported to the repo
+        using --no-download.
+
+        Returns:
+            A list of partially imported stages
+        """
+        from itertools import chain
+
+        partial_imports = chain.from_iterable(
+            self.index.partial_imports(targets, recursive=recursive)
+            for _ in self.brancher(
+                revs=revs,
+                all_branches=all_branches,
+                all_tags=all_tags,
+                all_commits=all_commits,
+                all_experiments=all_experiments,
+                commit_date=commit_date,
+                num=num,
+            )
+        )
+
+        return list(partial_imports)
+
     @property
     def stages(self):  # obsolete, only for backward-compatibility
         return self.index.stages
@@ -486,7 +522,7 @@ class Repo:
     def datafs(self):
         from dvc.fs.data import DataFileSystem
 
-        return DataFileSystem(repo=self)
+        return DataFileSystem(index=self.index.data["repo"])
 
     @cached_property
     def dvcfs(self):
@@ -507,18 +543,23 @@ class Repo:
         from dvc.fs.dvc import DvcFileSystem
 
         if os.path.isabs(path):
-            fs = DataFileSystem(repo=self, workspace="local")
+            fs = DataFileSystem(index=self.index.data["local"])
             fs_path = path
         else:
             fs = DvcFileSystem(repo=self, subrepos=True)
             fs_path = fs.from_os_path(path)
 
         try:
+            if remote:
+                remote_odb = self.cloud.get_remote_odb(name=remote)
+                oid = fs.info(fs_path)["dvc_info"]["md5"]
+                fs = remote_odb.fs
+                fs_path = remote_odb.oid_to_path(oid)
+
             with fs.open(
                 fs_path,
                 mode=mode,
                 encoding=encoding,
-                remote=remote,
             ) as fobj:
                 yield fobj
         except FileNotFoundError as exc:

dvc/repo/add.py

@@ -264,6 +264,7 @@ def create_stages(
             outs=[out],
             external=external,
         )
-        if kwargs.get("desc"):
-            stage.outs[0].desc = kwargs["desc"]
+
+        out_obj = stage.outs[0]
+        out_obj.annot.update(**kwargs)
         yield stage

dvc/repo/brancher.py

@@ -1,3 +1,5 @@
+from typing import Optional
+
 from dvc.scm import iter_revs
 
 
@@ -8,7 +10,9 @@ def brancher(  # noqa: E302
     all_tags=False,
     all_commits=False,
     all_experiments=False,
+    commit_date: Optional[str] = None,
     sha_only=False,
+    num=1,
 ):
     """Generator that iterates over specified revisions.
 
@@ -17,6 +21,9 @@ def brancher(  # noqa: E302
         all_branches (bool): iterate over all available branches.
         all_commits (bool): iterate over all commits.
         all_tags (bool): iterate over all available tags.
+        commit_date (str): Keep experiments from the commits after(include)
+                            a certain date. Date must match the extended
+                            ISO 8601 format (YYYY-MM-DD).
         sha_only (bool): only return git SHA for a revision.
 
     Yields:
@@ -62,6 +69,8 @@ def brancher(  # noqa: E302
         all_tags=all_tags,
         all_commits=all_commits,
         all_experiments=all_experiments,
+        commit_date=commit_date,
+        num=num,
     )
 
     try:
@@ -72,7 +81,9 @@ def brancher(  # noqa: E302
             self.root_dir = self.fs.path.join("/", *repo_root_parts)
 
             if cwd_parts:
-                cwd = self.fs.path.join("/", *cwd_parts)
+                cwd = self.fs.path.join(  # type: ignore[unreachable]
+                    "/", *cwd_parts
+                )
                 self.fs.path.chdir(cwd)
 
             # ignore revs that don't contain repo root

dvc/repo/collect.py

@@ -51,8 +51,8 @@ def _collect_paths(
     return target_paths
 
 
-def _filter_duplicates(
-    outs: Outputs, fs_paths: StrPaths
+def _filter_outs(
+    outs: Outputs, fs_paths: StrPaths, duplicates=False
 ) -> Tuple[Outputs, StrPaths]:
     res_outs: Outputs = []
     fs_res_paths = fs_paths
@@ -61,8 +61,9 @@ def _filter_duplicates(
         fs_path = out.repo.dvcfs.from_os_path(out.fs_path)
         if fs_path in fs_paths:
             res_outs.append(out)
-            # MUTATING THE SAME LIST!!
-            fs_res_paths.remove(fs_path)
+            if not duplicates:
+                # MUTATING THE SAME LIST!!
+                fs_res_paths.remove(fs_path)
 
     return res_outs, fs_res_paths
 
@@ -74,6 +75,7 @@ def collect(
     output_filter: FilterFn = None,
     rev: str = None,
     recursive: bool = False,
+    duplicates: bool = False,
 ) -> Tuple[Outputs, StrPaths]:
     assert targets or output_filter
 
@@ -85,4 +87,4 @@ def collect(
 
     target_paths = _collect_paths(repo, targets, recursive=recursive, rev=rev)
 
-    return _filter_duplicates(outs, target_paths)
+    return _filter_outs(outs, target_paths, duplicates=duplicates)

dvc/repo/data.py

@@ -0,0 +1,324 @@
+import os
+import posixpath
+from collections import defaultdict
+from itertools import chain
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Iterable,
+    Iterator,
+    List,
+    Optional,
+    Tuple,
+    TypedDict,
+    cast,
+)
+
+from dvc.fs.git import GitFileSystem
+from dvc.ui import ui
+
+if TYPE_CHECKING:
+    from scmrepo.base import Base
+
+    from dvc.output import Output
+    from dvc.repo import Repo
+    from dvc_data.hashfile.db import HashFileDB
+    from dvc_data.hashfile.hash_info import HashInfo
+    from dvc_data.hashfile.obj import HashFile
+
+
+def posixpath_to_os_path(path: str) -> str:
+    return path.replace(posixpath.sep, os.path.sep)
+
+
+def _in_cache(obj: "HashInfo", cache: "HashFileDB") -> bool:
+    from dvc_objects.errors import ObjectFormatError
+
+    assert obj.value
+    try:
+        cache.check(obj.value)
+        return True
+    except (FileNotFoundError, ObjectFormatError):
+        return False
+
+
+def _shallow_diff(
+    root: str,
+    old: Optional["HashInfo"],
+    new: Optional["HashInfo"],
+    cache: "HashFileDB",
+) -> Dict[str, List[str]]:
+    d = {}
+
+    old_root = new_root = root
+    if old and old.isdir:
+        old_root = os.path.sep.join([root, ""])
+    if new and new.isdir:
+        new_root = os.path.sep.join([root, ""])
+
+    if old and not _in_cache(old, cache):
+        d["not_in_cache"] = [old_root]
+
+    if not old and not new:
+        return d
+    if not new:
+        return {"deleted": [old_root], **d}
+    if not old:
+        return {"added": [new_root], **d}
+    if old != new:
+        return {"modified": [new_root], **d}
+    return {"unchanged": [new_root], **d}
+
+
+def _granular_diff(
+    root: str,
+    old_obj: Optional["HashFile"],
+    new_obj: Optional["HashFile"],
+    cache: "HashFileDB",
+) -> Dict[str, List[str]]:
+    from dvc_data.hashfile.diff import ROOT
+    from dvc_data.hashfile.diff import diff as odiff
+
+    def path_join(root: str, *paths: str, isdir: bool = False) -> str:
+        if not isdir and paths == ROOT:
+            return root
+        return os.path.sep.join([root, *paths])
+
+    diff_data = odiff(old_obj, new_obj, cache)
+
+    output: Dict[str, List[str]] = defaultdict(list)
+    for state in ("added", "deleted", "modified", "unchanged"):
+        items = getattr(diff_data, state)
+        for item in items:  # pylint: disable=not-an-iterable
+            entry = item.old if state == "deleted" else item.new
+            isdir = entry.oid.isdir if entry.oid else False
+
+            path = path_join(root, *entry.key, isdir=isdir)
+            output[state].append(path)
+            if not item.old.in_cache and state != "added":
+                output["not_in_cache"].append(path)
+    return dict(output)
+
+
+def _get_obj_items(root: str, obj: Optional["HashFile"]) -> List[str]:
+    if not obj:
+        return []
+
+    from dvc_data.hashfile.tree import Tree
+
+    sep = os.path.sep
+    if isinstance(obj, Tree):
+        return [sep.join([root, *key]) for key, _, _ in obj]
+    return [root]
+
+
+def _diff(
+    root: str,
+    old_oid: Optional["HashInfo"],
+    old_obj: Optional["HashFile"],
+    new_oid: Optional["HashInfo"],
+    new_obj: Optional["HashFile"],
+    odb: "HashFileDB",
+    granular: bool = False,
+) -> Dict[str, List[str]]:
+    if not granular:
+        return _shallow_diff(root, old_oid, new_oid, odb)
+    if (old_oid and not old_obj) or (new_oid and not new_obj):
+        # we don't have enough information to give full details
+        unknown = _get_obj_items(root, new_obj)
+        shallow_diff = _shallow_diff(root, old_oid, new_oid, odb)
+        return {**shallow_diff, "unknown": unknown}
+    return _granular_diff(root, old_obj, new_obj, odb)
+
+
+class GitInfo(TypedDict, total=False):
+    staged: Dict[str, List[str]]
+    unstaged: Dict[str, List[str]]
+    untracked: List[str]
+    is_empty: bool
+    is_dirty: bool
+
+
+def _git_info(scm: "Base", untracked_files: str = "all") -> GitInfo:
+    from scmrepo.exceptions import SCMError
+
+    from dvc.scm import NoSCM
+
+    if isinstance(scm, NoSCM):
+        return {}
+
+    try:
+        scm.get_rev()
+    except SCMError:
+        empty_repo = True
+    else:
+        empty_repo = False
+
+    staged, unstaged, untracked = scm.status(untracked_files=untracked_files)
+    if os.name == "nt":
+        untracked = [posixpath_to_os_path(path) for path in untracked]
+    # NOTE: order is important here.
+    return GitInfo(
+        staged=staged,
+        unstaged=unstaged,
+        untracked=untracked,
+        is_empty=empty_repo,
+        is_dirty=any([staged, unstaged, untracked]),
+    )
+
+
+def _diff_index_to_wtree(repo: "Repo", **kwargs: Any) -> Dict[str, List[str]]:
+    from dvc_data.hashfile.build import build
+
+    unstaged_diff = defaultdict(list)
+    for out in repo.index.outs:
+        out = cast("Output", out)
+        if not out.use_cache:
+            continue
+
+        try:
+            _, _, new = build(
+                out.odb,
+                out.fs_path,
+                out.fs,
+                out.fs.PARAM_CHECKSUM,
+                ignore=out.dvcignore,
+                dry_run=True,
+            )
+        except FileNotFoundError:
+            new = None
+
+        cache = repo.odb.repo
+        root = str(out)
+        old = out.get_obj()
+
+        with ui.status(f"Calculating diff for {root} between index/workspace"):
+            d = _diff(
+                root,
+                out.hash_info,
+                old,
+                new.hash_info if new else None,
+                new,
+                cache,
+                **kwargs,
+            )
+
+        for state, items in d.items():
+            if not items:
+                continue
+            unstaged_diff[state].extend(items)
+    return unstaged_diff
+
+
+def _diff_head_to_index(
+    repo: "Repo", head: str = "HEAD", **kwargs: Any
+) -> Dict[str, List[str]]:
+    # we need to store objects from index and the HEAD to diff later
+    objs: Dict[str, Dict[str, Tuple["HashFile", "HashInfo"]]]
+    objs = defaultdict(dict)
+
+    staged_diff = defaultdict(list)
+    for rev in repo.brancher(revs=[head]):
+        for out in repo.index.outs:
+            out = cast("Output", out)
+            if not out.use_cache:
+                continue
+
+            root = str(out)
+            if isinstance(out.fs, GitFileSystem):
+                root = posixpath_to_os_path(root)
+            typ = "index" if rev == "workspace" else head
+            objs[root][typ] = (out.get_obj(), out.hash_info)
+
+    cache = repo.odb.repo
+    for root, obj_d in objs.items():
+        old_obj, old_oid = obj_d.get(head, (None, None))
+        new_obj, new_oid = obj_d.get("index", (None, None))
+        with ui.status(f"Calculating diff for {root} between head/index"):
+            d = _diff(
+                root, old_oid, old_obj, new_oid, new_obj, cache, **kwargs
+            )
+
+        for state, items in d.items():
+            if not items:
+                continue
+            staged_diff[state].extend(items)
+
+    return staged_diff
+
+
+class Status(TypedDict):
+    not_in_cache: List[str]
+    committed: Dict[str, List[str]]
+    uncommitted: Dict[str, List[str]]
+    untracked: List[str]
+    unchanged: List[str]
+    git: GitInfo
+
+
+def _transform_git_paths_to_dvc(
+    repo: "Repo", files: Iterable[str]
+) -> List[str]:
+    """Transform files rel. to Git root to DVC root, and drop outside files."""
+    rel = repo.fs.path.relpath(repo.root_dir, repo.scm.root_dir).rstrip("/")
+
+    # if we have repo root in a different location than scm's root,
+    # i.e. subdir repo, all git_paths need to be transformed rel. to the DVC
+    # repo root and anything outside need to be filtered out.
+    if rel not in (os.curdir, ""):
+        prefix = rel + os.sep
+        length = len(prefix)
+        files = (file[length:] for file in files if file.startswith(prefix))
+
+    start = repo.fs.path.relpath(repo.fs.path.getcwd(), repo.root_dir)
+    if start in (os.curdir, ""):
+        return list(files)
+    # we need to convert repo relative paths to curdir relative.
+    return [repo.fs.path.relpath(file, start) for file in files]
+
+
+def status(repo: "Repo", untracked_files: str = "no", **kwargs: Any) -> Status:
+    from dvc.scm import NoSCMError, SCMError
+
+    head = kwargs.pop("head", "HEAD")
+    uncommitted_diff = _diff_index_to_wtree(repo, **kwargs)
+    not_in_cache = uncommitted_diff.pop("not_in_cache", [])
+    unchanged = set(uncommitted_diff.pop("unchanged", []))
+
+    try:
+        committed_diff = _diff_head_to_index(repo, head=head, **kwargs)
+    except (SCMError, NoSCMError):
+        committed_diff = {}
+    else:
+        # we don't care about not-in-cache between the head and the index.
+        committed_diff.pop("not_in_cache", None)
+        unchanged &= set(committed_diff.pop("unchanged", []))
+
+    git_info = _git_info(repo.scm, untracked_files=untracked_files)
+    untracked = git_info.get("untracked", [])
+    untracked = _transform_git_paths_to_dvc(repo, untracked)
+    # order matters here
+    return Status(
+        not_in_cache=not_in_cache,
+        committed=committed_diff,
+        uncommitted=uncommitted_diff,
+        untracked=untracked,
+        unchanged=list(unchanged),
+        git=git_info,
+    )
+
+
+def ls(
+    repo: "Repo",
+    targets: List[Optional[str]] = None,
+    recursive: bool = False,
+) -> Iterator[Dict[str, Any]]:
+    targets = targets or [None]
+    pairs = chain.from_iterable(
+        repo.stage.collect_granular(target, recursive=recursive)
+        for target in targets
+    )
+    for stage, filter_info in pairs:
+        for out in stage.filter_outs(filter_info):
+            yield {"path": str(out), **out.annot.to_dict()}

dvc/repo/diff.py

@@ -5,6 +5,7 @@ from typing import Dict, List
 
 from dvc.exceptions import PathMissingError
 from dvc.repo import locked
+from dvc.utils.collections import ensure_list
 
 logger = logging.getLogger(__name__)
 
@@ -25,6 +26,8 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
     from dvc.fs.dvc import DvcFileSystem
 
     dvcfs = DvcFileSystem(repo=self)
+    targets = ensure_list(targets)
+    targets = [dvcfs.from_os_path(target) for target in targets]
 
     b_rev = b_rev if b_rev else "workspace"
     results = {}
@@ -36,7 +39,7 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
             continue
 
         targets_paths = None
-        if targets is not None:
+        if targets:
             # convert targets to paths, and capture any missing targets
             targets_paths, missing_targets[rev] = _targets_to_paths(
                 dvcfs, targets
@@ -44,7 +47,7 @@ def diff(self, a_rev="HEAD", b_rev=None, targets=None):
 
         results[rev] = _paths_checksums(self, targets_paths)
 
-    if targets is not None:
+    if targets:
         # check for overlapping missing targets between a_rev and b_rev
         for target in set(missing_targets[a_rev]) & set(
             missing_targets[b_rev]
@@ -112,7 +115,7 @@ def _paths_checksums(repo, targets):
 
 def _output_paths(repo, targets):
     from dvc.fs import LocalFileSystem
-    from dvc_data.stage import stage as ostage
+    from dvc_data.hashfile.build import build
 
     on_working_fs = isinstance(repo.fs, LocalFileSystem)
 
@@ -136,10 +139,10 @@ def _output_paths(repo, targets):
             )
 
             if on_working_fs:
-                _, _, obj = ostage(
-                    repo.odb.local,
+                _, _, obj = build(
+                    repo.odb.repo,
                     output.fs_path,
-                    repo.odb.local.fs,
+                    repo.odb.repo.fs,
                     "md5",
                     dry_run=True,
                     ignore=output.dvcignore,
@@ -183,11 +186,12 @@ def _filter_missing(dvcfs, paths):
         fs_path = dvcfs.from_os_path(path)
         try:
             info = dvcfs.info(fs_path)
-            dvc_info = info.get("dvc_info")
+            dvc_info = info.get("dvc_info", {})
+            entry = dvc_info.get("entry")
             if (
-                dvc_info
+                entry
                 and info["type"] == "directory"
-                and not dvc_info["meta"].obj
+                and not entry.odb.exists(entry.hash_info.value)
             ):
                 yield path
         except FileNotFoundError:

dvc/repo/experiments/__init__.py

@@ -1,74 +1,43 @@
 import logging
 import os
 import re
-from functools import wraps
-from typing import Dict, Iterable, List, Mapping, Optional, Type
+import time
+from typing import Any, Dict, Iterable, Optional
 
 from funcy import cached_property, first
 
-from dvc.dependency.param import MissingParamsError
-from dvc.env import DVCLIVE_RESUME
 from dvc.exceptions import DvcException
+from dvc.ui import ui
 from dvc.utils import relpath
 
-from .base import (
+from .exceptions import (
+    BaselineMismatchError,
+    ExperimentExistsError,
+    InvalidExpRefError,
+    MultipleBranchError,
+)
+from .executor.base import BaseExecutor, ExecutorInfo
+from .queue.base import BaseStashQueue, QueueEntry
+from .queue.celery import LocalCeleryQueue
+from .queue.tempdir import TempDirQueue
+from .queue.workspace import WorkspaceQueue
+from .refs import (
+    CELERY_FAILED_STASH,
+    CELERY_STASH,
     EXEC_APPLY,
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
-    EXPS_STASH,
-    BaselineMismatchError,
-    ExperimentExistsError,
+    WORKSPACE_STASH,
     ExpRefInfo,
-    ExpStashEntry,
-    InvalidExpRefError,
-    MultipleBranchError,
 )
-from .executor.base import (
-    EXEC_PID_DIR,
-    EXEC_TMP_DIR,
-    BaseExecutor,
-    ExecutorInfo,
-)
-from .executor.manager.base import BaseExecutorManager
-from .executor.manager.local import (
-    TempDirExecutorManager,
-    WorkspaceExecutorManager,
-)
-from .executor.manager.ssh import SSHExecutorManager
-from .utils import exp_refs_by_rev
+from .stash import ExpStashEntry
+from .utils import exp_refs_by_rev, scm_locked, unlocked_repo
 
 logger = logging.getLogger(__name__)
 
 
-def scm_locked(f):
-    # Lock the experiments workspace so that we don't try to perform two
-    # different sequences of git operations at once
-    @wraps(f)
-    def wrapper(exp, *args, **kwargs):
-        from dvc.scm import map_scm_exception
-
-        with map_scm_exception(), exp.scm_lock:
-            return f(exp, *args, **kwargs)
-
-    return wrapper
-
-
-def unlocked_repo(f):
-    @wraps(f)
-    def wrapper(exp, *args, **kwargs):
-        exp.repo.lock.unlock()
-        exp.repo._reset()  # pylint: disable=protected-access
-        try:
-            ret = f(exp, *args, **kwargs)
-        finally:
-            exp.repo.lock.lock()
-        return ret
-
-    return wrapper
-
-
 class Experiments:
     """Class that manages experiments in a DVC repo.
 
@@ -76,13 +45,6 @@ class Experiments:
         repo (dvc.repo.Repo): repo instance that these experiments belong to.
     """
 
-    STASH_EXPERIMENT_FORMAT = "dvc-exp:{rev}:{baseline_rev}:{name}"
-    STASH_EXPERIMENT_RE = re.compile(
-        r"(?:commit: )"
-        r"dvc-exp:(?P<rev>[0-9a-f]+):(?P<baseline_rev>[0-9a-f]+)"
-        r":(?P<name>[^~^:\\?\[\]*]*)"
-        r"(:(?P<branch>.+))?$"
-    )
     BRANCH_RE = re.compile(
         r"^(?P<baseline_rev>[a-f0-9]{7})-(?P<exp_sha>[a-f0-9]+)"
         r"(?P<checkpoint>-checkpoint)?$"
@@ -115,289 +77,34 @@ class Experiments:
         return os.path.join(self.repo.tmp_dir, BaseExecutor.PACKED_ARGS_FILE)
 
     @cached_property
-    def stash(self):
-        from scmrepo.git import Stash
+    def workspace_queue(self) -> WorkspaceQueue:
+        return WorkspaceQueue(self.repo, WORKSPACE_STASH)
 
-        return Stash(self.scm, EXPS_STASH)
+    @cached_property
+    def tempdir_queue(self) -> TempDirQueue:
+        # NOTE: tempdir and workspace stash is shared since both
+        # implementations immediately push -> pop (queue length is only 0 or 1)
+        return TempDirQueue(self.repo, WORKSPACE_STASH)
+
+    @cached_property
+    def celery_queue(self) -> LocalCeleryQueue:
+        return LocalCeleryQueue(self.repo, CELERY_STASH, CELERY_FAILED_STASH)
 
     @property
     def stash_revs(self) -> Dict[str, ExpStashEntry]:
         revs = {}
-        for i, entry in enumerate(self.stash):
-            msg = entry.message.decode("utf-8").strip()
-            m = self.STASH_EXPERIMENT_RE.match(msg)
-            if m:
-                revs[entry.new_sha.decode("utf-8")] = ExpStashEntry(
-                    i,
-                    m.group("rev"),
-                    m.group("baseline_rev"),
-                    m.group("branch"),
-                    m.group("name"),
-                )
+        for queue in (self.workspace_queue, self.celery_queue):
+            revs.update(queue.stash.stash_revs)
         return revs
 
-    def _stash_exp(
-        self,
-        *args,
-        params: Optional[dict] = None,
-        resume_rev: Optional[str] = None,
-        baseline_rev: Optional[str] = None,
-        branch: Optional[str] = None,
-        name: Optional[str] = None,
-        **kwargs,
-    ):
-        """Stash changes from the workspace as an experiment.
-
-        Args:
-            params: Optional dictionary of parameter values to be used.
-                Values take priority over any parameters specified in the
-                user's workspace.
-            resume_rev: Optional checkpoint resume rev.
-            baseline_rev: Optional baseline rev for this experiment, defaults
-                to the current SCM rev.
-            branch: Optional experiment branch name. If specified, the
-                experiment will be added to `branch` instead of creating
-                a new branch.
-            name: Optional experiment name. If specified this will be used as
-                the human-readable name in the experiment branch ref. Has no
-                effect of branch is specified.
-        """
-        with self.scm.detach_head(client="dvc") as orig_head:
-            stash_head = orig_head
-            if baseline_rev is None:
-                baseline_rev = orig_head
-
-            with self.scm.stash_workspace() as workspace:
-                try:
-                    if workspace:
-                        self.stash.apply(workspace)
-
-                    if resume_rev:
-                        # move HEAD to the resume rev so that the stashed diff
-                        # only contains changes relative to resume rev
-                        stash_head = resume_rev
-                        self.scm.set_ref(
-                            "HEAD",
-                            resume_rev,
-                            message=f"dvc: resume from HEAD {resume_rev[:7]}",
-                        )
-                        self.scm.reset()
-
-                    # update experiment params from command line
-                    if params:
-                        self._update_params(params)
-
-                    # DVC commit data deps to preserve state across workspace
-                    # & tempdir runs
-                    self._stash_commit_deps(*args, **kwargs)
-
-                    if resume_rev:
-                        if branch:
-                            branch_name = ExpRefInfo.from_ref(branch).name
-                        else:
-                            branch_name = f"{resume_rev[:7]}"
-                        if self.scm.is_dirty(untracked_files=False):
-                            logger.info(
-                                "Modified checkpoint experiment based on "
-                                "'%s' will be created",
-                                branch_name,
-                            )
-                            branch = None
-                        elif (
-                            not branch
-                            or self.scm.get_ref(branch) != resume_rev
-                        ):
-                            msg = [
-                                (
-                                    "Nothing to do for unchanged checkpoint "
-                                    f"'{resume_rev[:7]}'. "
-                                )
-                            ]
-                            if branch:
-                                msg.append(
-                                    "To resume from the head of this "
-                                    "experiment, use "
-                                    f"'dvc exp apply {branch_name}'."
-                                )
-                            else:
-                                names = [
-                                    ref_info.name
-                                    for ref_info in exp_refs_by_rev(
-                                        self.scm, resume_rev
-                                    )
-                                ]
-                                if len(names) > 3:
-                                    names[3:] = [
-                                        f"... ({len(names) - 3} more)"
-                                    ]
-                                msg.append(
-                                    "To resume an experiment containing this "
-                                    "checkpoint, apply one of these heads:\n"
-                                    "\t{}".format(", ".join(names))
-                                )
-                            raise DvcException("".join(msg))
-                        else:
-                            logger.info(
-                                "Existing checkpoint experiment '%s' will be "
-                                "resumed",
-                                branch_name,
-                            )
-                        if name:
-                            logger.warning(
-                                "Ignoring option '--name %s' for resumed "
-                                "experiment. Existing experiment name will"
-                                "be preserved instead.",
-                                name,
-                            )
-
-                    # save additional repro command line arguments
-                    run_env = {DVCLIVE_RESUME: "1"} if resume_rev else {}
-                    self._pack_args(*args, run_env=run_env, **kwargs)
-
-                    # save experiment as a stash commit
-                    msg = self._stash_msg(
-                        stash_head,
-                        baseline_rev=baseline_rev,
-                        branch=branch,
-                        name=name,
-                    )
-                    stash_rev = self.stash.push(message=msg)
-                    logger.debug(
-                        (
-                            "Stashed experiment '%s' with baseline '%s' "
-                            "for future execution."
-                        ),
-                        stash_rev[:7],
-                        baseline_rev[:7],
-                    )
-                finally:
-                    if resume_rev:
-                        # NOTE: this set_ref + reset() is equivalent to
-                        # `git reset orig_head` (our SCM reset() only operates
-                        # on HEAD rather than any arbitrary commit)
-                        self.scm.set_ref(
-                            "HEAD", orig_head, message="dvc: restore HEAD"
-                        )
-                        self.scm.reset()
-                    # Revert any of our changes before prior unstashing
-                    self.scm.reset(hard=True)
-
-        return stash_rev
-
-    def _stash_commit_deps(self, *args, **kwargs):
-        if len(args):
-            targets = args[0]
-        else:
-            targets = kwargs.get("targets")
-        if isinstance(targets, str):
-            targets = [targets]
-        elif not targets:
-            targets = [None]
-        for target in targets:
-            self.repo.commit(
-                target,
-                with_deps=True,
-                recursive=kwargs.get("recursive", False),
-                force=True,
-                allow_missing=True,
-                data_only=True,
-            )
-
-    def _stash_msg(
-        self,
-        rev: str,
-        baseline_rev: str,
-        branch: Optional[str] = None,
-        name: Optional[str] = None,
-    ):
-        if not baseline_rev:
-            baseline_rev = rev
-        msg = self.STASH_EXPERIMENT_FORMAT.format(
-            rev=rev, baseline_rev=baseline_rev, name=name if name else ""
-        )
-        if branch:
-            return f"{msg}:{branch}"
-        return msg
-
-    def _pack_args(self, *args, **kwargs):
-        import pickle
-
-        if os.path.exists(self.args_file) and self.scm.is_tracked(
-            self.args_file
-        ):
-            logger.warning(
-                (
-                    "Temporary DVC file '.dvc/tmp/%s' exists and was "
-                    "likely committed to Git by mistake. It should be removed "
-                    "with:\n"
-                    "\tgit rm .dvc/tmp/%s"
-                ),
-                BaseExecutor.PACKED_ARGS_FILE,
-                BaseExecutor.PACKED_ARGS_FILE,
-            )
-            with open(self.args_file, "rb") as fobj:
-                try:
-                    data = pickle.load(fobj)
-                except Exception:  # pylint: disable=broad-except
-                    data = {}
-            extra = int(data.get("extra", 0)) + 1
-        else:
-            extra = None
-        BaseExecutor.pack_repro_args(
-            self.args_file, *args, extra=extra, **kwargs
-        )
-        self.scm.add(self.args_file)
-
-    def _format_new_params_msg(self, new_params, config_path):
-        """Format an error message for when new parameters are identified"""
-        new_param_count = len(new_params)
-        pluralise = "s are" if new_param_count > 1 else " is"
-        param_list = ", ".join(new_params)
-        return (
-            f"{new_param_count} parameter{pluralise} missing "
-            f"from '{config_path}': {param_list}"
-        )
-
-    def _update_params(self, params: dict):
-        """Update experiment params files with the specified values."""
-        from dvc.utils.collections import NewParamsFound, merge_params
-        from dvc.utils.serialize import MODIFIERS
-
-        logger.debug("Using experiment params '%s'", params)
-
-        for path in params:
-            suffix = self.repo.fs.path.suffix(path).lower()
-            modify_data = MODIFIERS[suffix]
-            with modify_data(path, fs=self.repo.fs) as data:
-                try:
-                    merge_params(data, params[path], allow_new=False)
-                except NewParamsFound as e:
-                    msg = self._format_new_params_msg(e.new_params, path)
-                    raise MissingParamsError(msg)
-
-        # Force params file changes to be staged in git
-        # Otherwise in certain situations the changes to params file may be
-        # ignored when we `git stash` them since mtime is used to determine
-        # whether the file is dirty
-        self.scm.add(list(params.keys()))
-
     def reproduce_one(
         self,
-        queue: bool = False,
         tmp_dir: bool = False,
-        checkpoint_resume: Optional[str] = None,
-        reset: bool = False,
         machine: Optional[str] = None,
         **kwargs,
     ):
-        """Reproduce and checkout a single experiment."""
-        if queue and not checkpoint_resume:
-            reset = True
-
-        if reset:
-            self.reset_checkpoints()
-
-        if not (queue or tmp_dir or machine):
+        """Reproduce and checkout a single (standalone) experiment."""
+        if not (tmp_dir or machine):
             staged, _, _ = self.scm.status(untracked_files="no")
             if staged:
                 logger.warning(
@@ -406,6 +113,31 @@ class Experiments:
                 )
                 self.scm.reset()
 
+        exp_queue: BaseStashQueue = (
+            self.tempdir_queue if tmp_dir else self.workspace_queue
+        )
+        self.queue_one(exp_queue, **kwargs)
+        results = self._reproduce_queue(exp_queue)
+        exp_rev = first(results)
+        if exp_rev is not None:
+            self._log_reproduced(results, tmp_dir=tmp_dir)
+        return results
+
+    def queue_one(
+        self,
+        queue: BaseStashQueue,
+        checkpoint_resume: Optional[str] = None,
+        reset: bool = False,
+        **kwargs,
+    ) -> QueueEntry:
+        """Queue a single experiment."""
+        if reset:
+            self.reset_checkpoints()
+
+        if kwargs.pop("machine", None) is not None:
+            # TODO: decide how to handle queued remote execution
+            raise NotImplementedError
+
         if checkpoint_resume:
             from dvc.scm import resolve_rev
 
@@ -421,30 +153,12 @@ class Experiments:
         else:
             checkpoint_resume = self._workspace_resume_rev()
 
-        stash_rev = self.new(
-            checkpoint_resume=checkpoint_resume, reset=reset, **kwargs
+        return self.new(
+            queue,
+            checkpoint_resume=checkpoint_resume,
+            reset=reset,
+            **kwargs,
         )
-        if queue:
-            logger.info(
-                "Queued experiment '%s' for future execution.", stash_rev[:7]
-            )
-            return [stash_rev]
-        if tmp_dir or queue:
-            manager_cls: Type = TempDirExecutorManager
-        elif machine:
-            manager_cls = SSHExecutorManager
-        else:
-            manager_cls = WorkspaceExecutorManager
-        results = self._reproduce_revs(
-            revs=[stash_rev],
-            keep_stash=False,
-            manager_cls=manager_cls,
-            machine=machine,
-        )
-        exp_rev = first(results)
-        if exp_rev is not None:
-            self._log_reproduced(results, tmp_dir=tmp_dir)
-        return results
 
     def _workspace_resume_rev(self) -> Optional[str]:
         last_checkpoint = self._get_last_checkpoint()
@@ -453,10 +167,46 @@ class Experiments:
             return last_applied
         return None
 
-    def reproduce_queued(self, **kwargs):
-        results = self._reproduce_revs(**kwargs)
+    def reproduce_celery(
+        self, entries: Optional[Iterable[QueueEntry]] = None, **kwargs
+    ) -> Dict[str, str]:
+        results: Dict[str, str] = {}
+        if entries is None:
+            entries = list(self.celery_queue.iter_queued())
+        if not entries:
+            return results
+
+        # TODO: re-enable --jobs concurrency
+        self.celery_queue.spawn_worker()
+        failed = []
+        try:
+            ui.write(
+                "Following logs for all queued experiments. Use Ctrl+C to "
+                "stop following logs (experiment execution will continue).\n"
+            )
+            for entry in entries:
+                # wait for task execution to start
+                while not self.celery_queue.proc.get(entry.stash_rev):
+                    time.sleep(1)
+                self.celery_queue.follow(entry)
+                # wait for task collection to complete
+                result = self.celery_queue.get_result(entry)
+                if result is None or result.exp_hash is None:
+                    name = entry.name or entry.stash_rev[:7]
+                    failed.append(name)
+                elif result.ref_info:
+                    exp_rev = self.scm.get_ref(str(result.ref_info))
+                    results[exp_rev] = result.exp_hash
+        except KeyboardInterrupt:
+            ui.write(
+                "Experiment(s) are still executing in the background. To "
+                "abort execution use 'dvc queue kill' or 'dvc queue stop'."
+            )
+        if failed:
+            names = ", ".join(name for name in failed)
+            ui.error(f"Failed to reproduce experiment(s) '{names}'")
         if results:
-            self._log_reproduced(results, tmp_dir=True)
+            self._log_reproduced((rev for rev in results), True)
         return results
 
     def _log_reproduced(self, revs: Iterable[str], tmp_dir: bool = False):
@@ -464,18 +214,16 @@ class Experiments:
         for rev in revs:
             name = self.get_exact_name(rev)
             names.append(name if name else rev[:7])
-        logger.info("\nReproduced experiment(s): %s", ", ".join(names))
+        ui.write("\nRan experiment(s): {}".format(", ".join(names)))
         if tmp_dir:
-            logger.info(
+            ui.write(
                 "To apply the results of an experiment to your workspace "
                 "run:\n\n"
                 "\tdvc exp apply <exp>"
             )
         else:
-            logger.info(
-                "Experiment results have been applied to your workspace."
-            )
-        logger.info(
+            ui.write("Experiment results have been applied to your workspace.")
+        ui.write(
             "\nTo promote an experiment to a Git branch run:\n\n"
             "\tdvc exp branch <exp> <branch>\n"
         )
@@ -492,15 +240,20 @@ class Experiments:
             raise ExperimentExistsError(exp_ref.name)
 
     @scm_locked
-    def new(self, *args, checkpoint_resume: Optional[str] = None, **kwargs):
-        """Create a new experiment.
+    def new(
+        self,
+        queue: BaseStashQueue,
+        *args,
+        checkpoint_resume: Optional[str] = None,
+        **kwargs,
+    ) -> QueueEntry:
+        """Create and enqueue a new experiment.
 
-        Experiment will be reproduced and checked out into the user's
-        workspace.
+        Experiment will be derived from the current workspace.
         """
         if checkpoint_resume is not None:
             return self._resume_checkpoint(
-                *args, resume_rev=checkpoint_resume, **kwargs
+                queue, *args, resume_rev=checkpoint_resume, **kwargs
             )
 
         name = kwargs.get("name", None)
@@ -513,16 +266,16 @@ class Experiments:
             if not (kwargs.get("force", False) or kwargs.get("reset", False)):
                 raise err
 
-        return self._stash_exp(*args, **kwargs)
+        return queue.put(*args, **kwargs)
 
     def _resume_checkpoint(
-        self, *args, resume_rev: Optional[str] = None, **kwargs
-    ):
-        """Resume an existing (checkpoint) experiment.
-
-        Experiment will be reproduced and checked out into the user's
-        workspace.
-        """
+        self,
+        queue: BaseStashQueue,
+        *args,
+        resume_rev: Optional[str] = None,
+        **kwargs,
+    ) -> QueueEntry:
+        """Create and queue a resumed checkpoint experiment."""
         assert resume_rev
 
         branch: Optional[str] = None
@@ -553,7 +306,7 @@ class Experiments:
             resume_rev[:7],
             baseline_rev,
         )
-        return self._stash_exp(
+        return queue.put(
             *args,
             resume_rev=resume_rev,
             baseline_rev=baseline_rev,
@@ -589,90 +342,25 @@ class Experiments:
         self.scm.remove_ref(EXEC_CHECKPOINT)
         self.scm.remove_ref(EXEC_APPLY)
 
-    @scm_locked
-    def _reproduce_revs(
-        self,
-        revs: Optional[Iterable] = None,
-        keep_stash: Optional[bool] = True,
-        manager_cls: Type = TempDirExecutorManager,
-        machine: Optional[str] = None,
-        **kwargs,
-    ) -> Mapping[str, str]:
-        """Reproduce the specified experiments.
+    @unlocked_repo
+    def _reproduce_queue(
+        self, queue: BaseStashQueue, **kwargs
+    ) -> Dict[str, str]:
+        """Reproduce queued experiments.
 
-        Args:
-            revs: If revs is not specified, all stashed experiments will be
-                reproduced.
-            keep_stash: If True, stashed experiments will be preserved if they
-                fail to reproduce successfully.
+        Arguments:
+            queue: Experiment queue.
 
         Returns:
             dict mapping successfully reproduced experiment revs to their
-            hashes.
+            results.
         """
-        stash_revs = self.stash_revs
+        exec_results = queue.reproduce()
 
-        # to_run contains mapping of:
-        #   input_rev: (stash_index, rev, baseline_rev)
-        # where input_rev contains the changes to execute (usually a stash
-        # commit), rev is the original SCM commit to be checked out, and
-        # baseline_rev is the experiment baseline.
-        if revs is None:
-            to_run = dict(stash_revs)
-        else:
-            to_run = {
-                rev: stash_revs[rev]
-                if rev in stash_revs
-                else ExpStashEntry(None, rev, rev, None, None)
-                for rev in revs
-            }
-
-        logger.debug(
-            "Reproducing experiment revs '%s'",
-            ", ".join(rev[:7] for rev in to_run),
-        )
-
-        manager = manager_cls.from_stash_entries(
-            self.scm,
-            os.path.join(self.repo.tmp_dir, EXEC_TMP_DIR),
-            self.repo,
-            to_run,
-            machine_name=machine,
-        )
-        try:
-            exec_results = {}
-            exec_results.update(self._executors_repro(manager, **kwargs))
-        finally:
-            # only drop successfully run stashed experiments
-            to_drop: List[int] = [
-                entry.stash_index
-                for rev, entry in to_run.items()
-                if (
-                    entry.stash_index is not None
-                    and (not keep_stash or rev in exec_results)
-                )
-            ]
-            for index in sorted(to_drop, reverse=True):
-                self.stash.drop(index)
-
-        result: Dict[str, str] = {}
+        results: Dict[str, str] = {}
         for _, exp_result in exec_results.items():
-            result.update(exp_result)
-        return result
-
-    @unlocked_repo
-    def _executors_repro(
-        self,
-        manager: "BaseExecutorManager",
-        **kwargs,
-    ) -> Dict[str, Dict[str, str]]:
-        """Run dvc repro for the specified BaseExecutors in parallel.
-
-        Returns:
-            dict mapping stash revs to the successfully executed experiments
-            for each stash rev.
-        """
-        return manager.exec_queue(self.repo, **kwargs)
+            results.update(exp_result)
+        return results
 
     def check_baseline(self, exp_rev):
         baseline_sha = self.repo.scm.get_rev()
@@ -689,7 +377,6 @@ class Experiments:
             return exp_baseline
         raise BaselineMismatchError(exp_baseline, baseline_sha)
 
-    @scm_locked
     def get_baseline(self, rev):
         """Return the baseline rev for an experiment rev."""
         return self._get_baseline(rev)
@@ -742,61 +429,68 @@ class Experiments:
             return self.stash_revs[rev].name
         return None
 
-    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, int]:
+    def get_running_exps(self, fetch_refs: bool = True) -> Dict[str, Any]:
         """Return info for running experiments."""
+        result = {}
+        infofile = self.workspace_queue.get_infofile_path("workspace")
+        result.update(
+            self._fetch_running_exp("workspace", infofile, fetch_refs)
+        )
+        for queue in (self.tempdir_queue, self.celery_queue):
+            for entry in queue.iter_active():
+                infofile = queue.get_infofile_path(entry.stash_rev)
+                result.update(
+                    self._fetch_running_exp(
+                        entry.stash_rev, infofile, fetch_refs
+                    )
+                )
+        return result
+
+    def _fetch_running_exp(
+        self, rev: str, infofile: str, fetch_refs: bool
+    ) -> Dict[str, Any]:
         from dvc.scm import InvalidRemoteSCMRepo
         from dvc.utils.serialize import load_json
 
         from .executor.local import TempDirExecutor
 
-        result = {}
-        pid_dir = os.path.join(
-            self.repo.tmp_dir,
-            EXEC_TMP_DIR,
-            EXEC_PID_DIR,
-        )
-        for fname in self.repo.fs.find(pid_dir):
-            rev, ext = os.path.splitext(os.path.basename(fname))
-            if ext != BaseExecutor.INFOFILE_EXT:
-                continue
-
-            try:
-                info = ExecutorInfo.from_dict(load_json(fname))
-                if info.result is not None:
-                    continue
-                if rev == "workspace":
-                    # If we are appending to a checkpoint branch in a workspace
-                    # run, show the latest checkpoint as running.
-                    last_rev = self.scm.get_ref(EXEC_BRANCH)
-                    if last_rev:
-                        result[last_rev] = info.asdict()
-                    else:
-                        result[rev] = info.asdict()
+        result: Dict[str, Any] = {}
+        try:
+            info = ExecutorInfo.from_dict(load_json(infofile))
+        except OSError:
+            return result
+        if info.result is None:
+            if rev == "workspace":
+                # If we are appending to a checkpoint branch in a workspace
+                # run, show the latest checkpoint as running.
+                last_rev = self.scm.get_ref(EXEC_BRANCH)
+                if last_rev:
+                    result[last_rev] = info.asdict()
                 else:
                     result[rev] = info.asdict()
-                    if info.git_url and fetch_refs:
+            else:
+                result[rev] = info.asdict()
+                if info.git_url and fetch_refs:
 
-                        def on_diverged(_ref: str, _checkpoint: bool):
-                            return False
+                    def on_diverged(_ref: str, _checkpoint: bool):
+                        return False
 
-                        executor = TempDirExecutor.from_info(info)
-                        try:
-                            for ref in executor.fetch_exps(
-                                self.scm,
-                                on_diverged=on_diverged,
-                            ):
-                                logger.debug(
-                                    "Updated running experiment '%s'.", ref
-                                )
-                                last_rev = self.scm.get_ref(ref)
-                                result[rev]["last"] = last_rev
-                                if last_rev:
-                                    result[last_rev] = info.asdict()
-                        except InvalidRemoteSCMRepo:
-                            # ignore stale info files
-                            del result[rev]
-            except OSError:
-                pass
+                    executor = TempDirExecutor.from_info(info)
+                    try:
+                        for ref in executor.fetch_exps(
+                            self.scm,
+                            on_diverged=on_diverged,
+                        ):
+                            logger.debug(
+                                "Updated running experiment '%s'.", ref
+                            )
+                            last_rev = self.scm.get_ref(ref)
+                            result[rev]["last"] = last_rev
+                            if last_rev:
+                                result[last_rev] = info.asdict()
+                    except InvalidRemoteSCMRepo:
+                        # ignore stale info files
+                        del result[rev]
         return result
 
     def apply(self, *args, **kwargs):

dvc/repo/experiments/apply.py

@@ -1,25 +1,32 @@
 import logging
 import os
+from contextlib import contextmanager
+from typing import TYPE_CHECKING, Optional
 
 from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
 from dvc.utils.fs import remove
 
-from .base import (
-    EXEC_APPLY,
+from .exceptions import (
     ApplyConflictError,
     BaselineMismatchError,
     InvalidExpRevError,
 )
 from .executor.base import BaseExecutor
+from .refs import EXEC_APPLY
+
+if TYPE_CHECKING:
+    from scmrepo import Git
+
+    from dvc.repo import Repo
 
 logger = logging.getLogger(__name__)
 
 
 @locked
 @scm_context
-def apply(repo, rev, force=True, **kwargs):
-    from scmrepo.exceptions import MergeConflictError
+def apply(repo: "Repo", rev: str, force: bool = True, **kwargs):
+    from scmrepo.exceptions import SCMError as _SCMError
 
     from dvc.repo.checkout import checkout as dvc_checkout
     from dvc.scm import GitMergeError, RevError, resolve_rev
@@ -38,37 +45,14 @@ def apply(repo, rev, force=True, **kwargs):
     ):
         raise InvalidExpRevError(exp_rev)
 
-    # Note that we don't use stash_workspace() here since we need finer control
-    # over the merge behavior when we unstash everything
-    if repo.scm.is_dirty(untracked_files=True):
-        logger.debug("Stashing workspace")
-        workspace = repo.scm.stash.push(include_untracked=True)
-    else:
-        workspace = None
-
-    from scmrepo.exceptions import SCMError as _SCMError
-
-    try:
-        repo.scm.merge(exp_rev, commit=False, squash=True)
-    except _SCMError as exc:
-        raise GitMergeError(str(exc), scm=repo.scm)
-
-    if workspace:
+    # NOTE: we don't use scmrepo's stash_workspace() here since we need
+    # finer control over the merge behavior when we unstash everything
+    with _apply_workspace(repo, rev, force):
         try:
-            repo.scm.stash.apply(workspace)
-        except MergeConflictError as exc:
-            # Applied experiment conflicts with user's workspace changes
-            if force:
-                # prefer applied experiment changes over prior stashed changes
-                repo.scm.checkout_index(ours=True)
-            else:
-                # revert applied changes and restore user's workspace
-                repo.scm.reset(hard=True)
-                repo.scm.stash.pop()
-                raise ApplyConflictError(rev) from exc
+            repo.scm.merge(exp_rev, commit=False, squash=True)
         except _SCMError as exc:
-            raise ApplyConflictError(rev) from exc
-        repo.scm.stash.drop()
+            raise GitMergeError(str(exc), scm=repo.scm)
+
     repo.scm.reset()
 
     if stash_rev:
@@ -84,3 +68,44 @@ def apply(repo, rev, force=True, **kwargs):
         "workspace.",
         rev,
     )
+
+
+@contextmanager
+def _apply_workspace(repo: "Repo", rev: str, force: bool):
+    from scmrepo.exceptions import MergeConflictError
+    from scmrepo.exceptions import SCMError as _SCMError
+
+    if repo.scm.is_dirty(untracked_files=True):
+        logger.debug("Stashing workspace")
+        stash_rev: Optional[str] = repo.scm.stash.push(include_untracked=True)
+    else:
+        stash_rev = None
+    try:
+        yield
+    except Exception:  # pylint: disable=broad-except
+        if stash_rev:
+            _clean_and_pop(repo.scm)
+        raise
+    if not stash_rev:
+        return
+
+    try:
+        repo.scm.reset()
+        repo.scm.stash.apply(stash_rev, skip_conflicts=force)
+        repo.scm.stash.drop()
+    except (MergeConflictError, _SCMError) as exc:
+        _clean_and_pop(repo.scm)
+        raise ApplyConflictError(rev) from exc
+    except Exception:  # pylint: disable=broad-except
+        _clean_and_pop(repo.scm)
+        raise
+
+
+def _clean_and_pop(scm: "Git"):
+    """Revert any changes and pop the last stash entry."""
+    scm.reset(hard=True)
+    if scm.is_dirty(untracked_files=True):
+        # drop any changes to untracked files before popping stash
+        scm.stash.push(include_untracked=True)
+        scm.stash.drop()
+    scm.stash.pop()

dvc/repo/experiments/base.py

@@ -1,161 +0,0 @@
-from typing import NamedTuple, Optional
-
-from dvc.exceptions import DvcException, InvalidArgumentError
-
-# Experiment refs are stored according baseline git SHA:
-#   refs/exps/01/234abcd.../<exp_name>
-EXPS_NAMESPACE = "refs/exps"
-EXPS_STASH = f"{EXPS_NAMESPACE}/stash"
-EXEC_NAMESPACE = f"{EXPS_NAMESPACE}/exec"
-EXEC_APPLY = f"{EXEC_NAMESPACE}/EXEC_APPLY"
-EXEC_CHECKPOINT = f"{EXEC_NAMESPACE}/EXEC_CHECKPOINT"
-EXEC_BRANCH = f"{EXEC_NAMESPACE}/EXEC_BRANCH"
-EXEC_BASELINE = f"{EXEC_NAMESPACE}/EXEC_BASELINE"
-EXEC_HEAD = f"{EXEC_NAMESPACE}/EXEC_HEAD"
-EXEC_MERGE = f"{EXEC_NAMESPACE}/EXEC_MERGE"
-
-
-class UnchangedExperimentError(DvcException):
-    def __init__(self, rev):
-        super().__init__(f"Experiment unchanged from '{rev[:7]}'.")
-        self.rev = rev
-
-
-class BaselineMismatchError(DvcException):
-    def __init__(self, rev, expected):
-        if hasattr(rev, "hexsha"):
-            rev = rev.hexsha
-        rev_str = f"{rev[:7]}" if rev is not None else "invalid commit"
-        super().__init__(
-            f"Experiment derived from '{rev_str}', expected '{expected[:7]}'."
-        )
-        self.rev = rev
-        self.expected_rev = expected
-
-
-class ExperimentExistsError(DvcException):
-    def __init__(self, name: str):
-        msg = (
-            "Experiment conflicts with existing experiment "
-            f"'{name}'. To overwrite the existing experiment run:\n\n"
-            "\tdvc exp run -f ...\n\n"
-            "To run this experiment with a different name run:\n\n"
-            f"\tdvc exp run -n <new_name> ...\n"
-        )
-        super().__init__(msg)
-        self.name = name
-
-
-class CheckpointExistsError(DvcException):
-    def __init__(self, name: str):
-        msg = (
-            "Reproduced checkpoint experiment conflicts with existing "
-            f"experiment '{name}'. To restart (and overwrite) the existing "
-            "experiment run:\n\n"
-            "\tdvc exp run -f ...\n\n"
-            "To resume the existing experiment, run:\n\n"
-            f"\tdvc exp apply {name}\n"
-            f"\tdvc exp run\n"
-        )
-        super().__init__(msg)
-        self.name = name
-
-
-class InvalidExpRefError(DvcException):
-    def __init__(self, ref):
-        super().__init__(f"'{ref}' is not a valid experiment refname.")
-        self.ref = ref
-
-
-class InvalidExpRevError(InvalidArgumentError):
-    def __init__(self, rev):
-        super().__init__(
-            f"'{rev}' does not appear to be an experiment commit."
-        )
-
-
-class MultipleBranchError(DvcException):
-    def __init__(self, rev, ref_infos):
-        super().__init__(
-            f"Ambiguous commit '{rev[:7]}' belongs to multiple experiment "
-            "branches."
-        )
-        self.rev = rev
-        self.ref_infos = ref_infos
-
-
-class ApplyConflictError(InvalidArgumentError):
-    def __init__(self, name):
-        super().__init__(
-            f"Experiment '{name}' cannot be applied to because your current "
-            "workspace contains changes which would be overwritten. Either "
-            "'git stash' your current changes before applying this "
-            "experiment, or re-run this command with '--force' to overwrite "
-            "your current changes."
-        )
-        self.name = name
-
-
-class ExpRefInfo:
-
-    namespace = EXPS_NAMESPACE
-
-    def __init__(
-        self, baseline_sha: Optional[str] = None, name: Optional[str] = None
-    ):
-        self.baseline_sha = baseline_sha
-        self.name: str = name if name else ""
-
-    def __str__(self):
-        return "/".join(self.parts)
-
-    def __repr__(self):
-        baseline = f"'{self.baseline_sha}'" if self.baseline_sha else "None"
-        name = f"'{self.name}'" if self.name else "None"
-        return f"ExpRefInfo(baseline_sha={baseline}, name={name})"
-
-    @property
-    def parts(self):
-        return (
-            (self.namespace,)
-            + (
-                (self.baseline_sha[:2], self.baseline_sha[2:])
-                if self.baseline_sha
-                else ()
-            )
-            + ((self.name,) if self.name else ())
-        )
-
-    @classmethod
-    def from_ref(cls, ref: str):
-        try:
-            parts = ref.split("/")
-            if (
-                len(parts) < 2
-                or len(parts) == 3
-                or len(parts) > 5
-                or "/".join(parts[:2]) != EXPS_NAMESPACE
-            ):
-                raise InvalidExpRefError(ref)
-        except ValueError:
-            raise InvalidExpRefError(ref)
-        baseline_sha = parts[2] + parts[3] if len(parts) >= 4 else None
-        name = parts[4] if len(parts) == 5 else None
-        return cls(baseline_sha, name)
-
-
-class ExpStashEntry(NamedTuple):
-    """Experiment stash entry.
-
-    stash_index: EXPS_STASH index for this entry. Can be None if this commit
-        is not pushed onto the stash ref.
-    head_rev: HEAD Git commit to be checked out for this experiment.
-    baseline_rev: Experiment baseline commit.
-    branch: Optional exp (checkpoint) branch name for this experiment.
-    """
-
-    stash_index: Optional[int]
-    head_rev: str
-    baseline_rev: str
-    branch: Optional[str]
-    name: Optional[str]

dvc/repo/experiments/branch.py

@@ -5,7 +5,7 @@ from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
 from dvc.scm import RevError
 
-from .base import InvalidExpRevError
+from .exceptions import InvalidExpRevError
 from .utils import exp_refs_by_rev
 
 logger = logging.getLogger(__name__)

dvc/repo/experiments/exceptions.py

@@ -1,15 +1,97 @@
-from typing import Collection, Iterable
+from typing import TYPE_CHECKING, Collection, Iterable
 
-from dvc.exceptions import InvalidArgumentError
+from dvc.exceptions import DvcException, InvalidArgumentError
 
-from .base import ExpRefInfo
+if TYPE_CHECKING:
+    from .refs import ExpRefInfo
+
+
+class UnchangedExperimentError(DvcException):
+    def __init__(self, rev):
+        super().__init__(f"Experiment unchanged from '{rev[:7]}'.")
+        self.rev = rev
+
+
+class BaselineMismatchError(DvcException):
+    def __init__(self, rev, expected):
+        if hasattr(rev, "hexsha"):
+            rev = rev.hexsha
+        rev_str = f"{rev[:7]}" if rev is not None else "invalid commit"
+        super().__init__(
+            f"Experiment derived from '{rev_str}', expected '{expected[:7]}'."
+        )
+        self.rev = rev
+        self.expected_rev = expected
+
+
+class ExperimentExistsError(DvcException):
+    def __init__(self, name: str):
+        msg = (
+            "Experiment conflicts with existing experiment "
+            f"'{name}'. To overwrite the existing experiment run:\n\n"
+            "\tdvc exp run -f ...\n\n"
+            "To run this experiment with a different name run:\n\n"
+            f"\tdvc exp run -n <new_name> ...\n"
+        )
+        super().__init__(msg)
+        self.name = name
+
+
+class CheckpointExistsError(DvcException):
+    def __init__(self, name: str):
+        msg = (
+            "Reproduced checkpoint experiment conflicts with existing "
+            f"experiment '{name}'. To restart (and overwrite) the existing "
+            "experiment run:\n\n"
+            "\tdvc exp run -f ...\n\n"
+            "To resume the existing experiment, run:\n\n"
+            f"\tdvc exp apply {name}\n"
+            f"\tdvc exp run\n"
+        )
+        super().__init__(msg)
+        self.name = name
+
+
+class InvalidExpRefError(DvcException):
+    def __init__(self, ref):
+        super().__init__(f"'{ref}' is not a valid experiment refname.")
+        self.ref = ref
+
+
+class InvalidExpRevError(InvalidArgumentError):
+    def __init__(self, rev):
+        super().__init__(
+            f"'{rev}' does not appear to be an experiment commit."
+        )
+
+
+class MultipleBranchError(DvcException):
+    def __init__(self, rev, ref_infos):
+        super().__init__(
+            f"Ambiguous commit '{rev[:7]}' belongs to multiple experiment "
+            "branches."
+        )
+        self.rev = rev
+        self.ref_infos = ref_infos
+
+
+class ApplyConflictError(InvalidArgumentError):
+    def __init__(self, name):
+        super().__init__(
+            f"Experiment '{name}' cannot be applied to because your current "
+            "workspace contains changes which would be overwritten. Either "
+            "'git stash' your current changes before applying this "
+            "experiment, or re-run this command with '--force' to overwrite "
+            "your current changes."
+        )
+        self.name = name
 
 
 class AmbiguousExpRefInfo(InvalidArgumentError):
     def __init__(
         self,
         exp_name: str,
-        exp_ref_list: Iterable[ExpRefInfo],
+        exp_ref_list: Iterable["ExpRefInfo"],
     ):
         msg = [
             (
@@ -23,21 +105,31 @@ class AmbiguousExpRefInfo(InvalidArgumentError):
 
 
 class UnresolvedExpNamesError(InvalidArgumentError):
+    NAME = "experiment name"
+
     def __init__(
         self, unresolved_list: Collection[str], *args, git_remote: str = None
     ):
-        unresolved_names = ";".join(unresolved_list)
+        unresolved_names = "; ".join(unresolved_list)
         if not git_remote:
-            if len(unresolved_names) > 1:
+            if len(unresolved_list) > 1:
                 super().__init__(
-                    f"'{unresolved_names}' are not valid experiment names"
+                    f"'{unresolved_names}' are not valid {self.NAME}s"
                 )
             else:
                 super().__init__(
-                    f"'{unresolved_names}' is not a valid experiment name"
+                    f"'{unresolved_names}' is not a valid {self.NAME}"
                 )
         else:
             super().__init__(
                 f"Experiment '{unresolved_names}' does not exist "
                 f"in '{git_remote}'"
             )
+
+
+class UnresolvedQueueExpNamesError(UnresolvedExpNamesError):
+    NAME = "queued experiment name"
+
+
+class ExpQueueEmptyError(DvcException):
+    pass

dvc/repo/experiments/executor/base.py

@@ -22,24 +22,27 @@ from typing import (
 from dvc.env import DVC_EXP_AUTO_PUSH, DVC_EXP_GIT_REMOTE
 from dvc.exceptions import DvcException
 from dvc.stage.serialize import to_lockfile
+from dvc.ui import ui
 from dvc.utils import dict_sha256, env2bool, relpath
 from dvc.utils.fs import remove
 
-from ..base import (
+from ..exceptions import (
+    CheckpointExistsError,
+    ExperimentExistsError,
+    UnchangedExperimentError,
+)
+from ..refs import (
     EXEC_BASELINE,
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
     EXEC_NAMESPACE,
     EXPS_NAMESPACE,
     EXPS_STASH,
-    CheckpointExistsError,
-    ExperimentExistsError,
     ExpRefInfo,
-    UnchangedExperimentError,
 )
 
 if TYPE_CHECKING:
-    from multiprocessing import Queue
+    from queue import Queue
 
     from scmrepo.git import Git
 
@@ -70,6 +73,7 @@ class ExecutorInfo:
     dvc_dir: str
     name: Optional[str] = None
     wdir: Optional[str] = None
+    collected: bool = False
     result_hash: Optional[str] = None
     result_ref: Optional[str] = None
     result_force: bool = False
@@ -99,6 +103,12 @@ class ExecutorInfo:
         with modify_json(filename) as d:
             d.update(self.asdict())
 
+    @classmethod
+    def load_json(cls, filename: str) -> "ExecutorInfo":
+        from dvc.utils.serialize import load_json
+
+        return cls.from_dict(load_json(filename))
+
 
 _T = TypeVar("_T", bound="BaseExecutor")
 
@@ -348,10 +358,11 @@ class BaseExecutor(ABC):
 
         if git_remote == dvc.root_dir:
             logger.warning(
-                f"'{git_remote}' points to the current Git repo, experiment "
+                "'%s' points to the current Git repo, experiment "
                 "Git refs will not be pushed. But DVC cache and run cache "
                 "will automatically be pushed to the default DVC remote "
-                "(if any) on each experiment commit."
+                "(if any) on each experiment commit.",
+                git_remote,
             )
         try:
             dvc.scm.validate_git_remote(git_remote)
@@ -603,7 +614,9 @@ class BaseExecutor(ABC):
         except BaseException as exc:  # pylint: disable=broad-except
             logger.warning(
                 "Something went wrong while auto pushing experiment "
-                f"to the remote '{git_remote}': {exc}"
+                "to the remote '%s': %s",
+                git_remote,
+                exc,
             )
 
     @classmethod
@@ -625,7 +638,7 @@ class BaseExecutor(ABC):
             if env2bool(DVC_EXP_AUTO_PUSH):
                 git_remote = os.getenv(DVC_EXP_GIT_REMOTE)
                 cls._auto_push(dvc, scm, git_remote)
-            logger.info("Checkpoint experiment iteration '%s'.", exp_rev[:7])
+            ui.write(f"Checkpoint experiment iteration '{exp_rev[:7]}'.")
         except UnchangedExperimentError:
             pass
 

dvc/repo/experiments/executor/local.py

@@ -8,9 +8,9 @@ from funcy import cached_property
 from scmrepo.exceptions import SCMError as _SCMError
 
 from dvc.scm import SCM, GitMergeError
-from dvc.utils.fs import remove
+from dvc.utils.fs import makedirs, remove
 
-from ..base import (
+from ..refs import (
     EXEC_APPLY,
     EXEC_BASELINE,
     EXEC_BRANCH,
@@ -26,7 +26,8 @@ if TYPE_CHECKING:
 
     from dvc.repo import Repo
 
-    from ..base import ExpRefInfo, ExpStashEntry
+    from ..refs import ExpRefInfo
+    from ..stash import ExpStashEntry
 
 logger = logging.getLogger(__name__)
 
@@ -64,7 +65,7 @@ class TempDirExecutor(BaseLocalExecutor):
     # suggestions) that are not applicable outside of workspace runs
     WARN_UNTRACKED = True
     QUIET = True
-    DEFAULT_LOCATION = "temp"
+    DEFAULT_LOCATION = "tempdir"
 
     def init_git(self, scm: "Git", branch: Optional[str] = None):
         from dulwich.repo import Repo as DulwichRepo
@@ -89,6 +90,11 @@ class TempDirExecutor(BaseLocalExecutor):
         head = EXEC_BRANCH if branch else EXEC_HEAD
         self.scm.checkout(head, detach=True)
         merge_rev = self.scm.get_ref(EXEC_MERGE)
+
+        for ref in (EXEC_HEAD, EXEC_MERGE, EXEC_BASELINE):
+            if scm.get_ref(ref):
+                scm.remove_ref(ref)
+
         try:
             self.scm.merge(merge_rev, squash=True, commit=False)
         except _SCMError as exc:
@@ -106,7 +112,7 @@ class TempDirExecutor(BaseLocalExecutor):
 
     def init_cache(self, repo: "Repo", rev: str, run_cache: bool = True):
         """Initialize DVC cache."""
-        self._config(repo.odb.local.cache_dir)
+        self._config(repo.odb.repo.path)
 
     def cleanup(self):
         super().cleanup()
@@ -119,11 +125,16 @@ class TempDirExecutor(BaseLocalExecutor):
         repo: "Repo",
         stash_rev: str,
         entry: "ExpStashEntry",
+        wdir: Optional[str] = None,
         **kwargs,
     ):
-        tmp_dir = mkdtemp(dir=os.path.join(repo.tmp_dir, EXEC_TMP_DIR))
+        parent_dir: str = wdir or os.path.join(repo.tmp_dir, EXEC_TMP_DIR)
+        makedirs(parent_dir, exist_ok=True)
+        tmp_dir = mkdtemp(dir=parent_dir)
         try:
-            executor = cls._from_stash_entry(repo, stash_rev, entry, tmp_dir)
+            executor = cls._from_stash_entry(
+                repo, stash_rev, entry, tmp_dir, **kwargs
+            )
             logger.debug("Init temp dir executor in '%s'", tmp_dir)
             return executor
         except Exception:
@@ -146,7 +157,9 @@ class WorkspaceExecutor(BaseLocalExecutor):
         **kwargs,
     ):
         root_dir = repo.scm.root_dir
-        executor = cls._from_stash_entry(repo, stash_rev, entry, root_dir)
+        executor = cls._from_stash_entry(
+            repo, stash_rev, entry, root_dir, **kwargs
+        )
         logger.debug("Init workspace executor in '%s'", root_dir)
         return executor
 
@@ -174,6 +187,7 @@ class WorkspaceExecutor(BaseLocalExecutor):
     def cleanup(self):
         with self._detach_stack:
             self.scm.remove_ref(EXEC_BASELINE)
+            self.scm.remove_ref(EXEC_MERGE)
             if self.scm.get_ref(EXEC_BRANCH):
                 self.scm.remove_ref(EXEC_BRANCH)
             checkpoint = self.scm.get_ref(EXEC_CHECKPOINT)

dvc/repo/experiments/executor/manager/base.py

@@ -7,15 +7,9 @@ from typing import TYPE_CHECKING, Deque, Dict, Generator, Optional, Tuple, Type
 
 from dvc.proc.manager import ProcessManager
 
-from ...base import (
-    EXEC_BASELINE,
-    EXEC_HEAD,
-    EXEC_MERGE,
-    CheckpointExistsError,
-    ExperimentExistsError,
-    ExpRefInfo,
-    ExpStashEntry,
-)
+from ...exceptions import CheckpointExistsError, ExperimentExistsError
+from ...refs import EXEC_BASELINE, EXEC_HEAD, EXEC_MERGE, ExpRefInfo
+from ...stash import ExpStashEntry
 from ..base import EXEC_PID_DIR, BaseExecutor
 from ..local import TempDirExecutor, WorkspaceExecutor
 
@@ -27,7 +21,7 @@ if TYPE_CHECKING:
 logger = logging.getLogger(__name__)
 
 
-class BaseExecutorManager(ABC, Mapping):
+class BaseExecutorManager(ABC, Mapping):  # noqa: B024
     """Manages executors for a collection of experiments to be run."""
 
     EXECUTOR_CLS: Type = BaseExecutor

dvc/repo/experiments/executor/manager/local.py

@@ -2,13 +2,8 @@ import logging
 from collections import defaultdict
 from typing import TYPE_CHECKING, Dict, Optional
 
-from ...base import (
-    EXEC_BASELINE,
-    EXEC_BRANCH,
-    EXEC_HEAD,
-    EXEC_MERGE,
-    ExpStashEntry,
-)
+from ...refs import EXEC_BASELINE, EXEC_BRANCH, EXEC_HEAD, EXEC_MERGE
+from ...stash import ExpStashEntry
 from ..local import TempDirExecutor, WorkspaceExecutor
 from .base import BaseExecutorManager
 

dvc/repo/experiments/executor/manager/ssh.py

@@ -3,7 +3,7 @@ import posixpath
 from collections import defaultdict
 from typing import TYPE_CHECKING, Callable, Dict, Generator, Optional, Tuple
 
-from ...base import ExpStashEntry
+from ...stash import ExpStashEntry
 from ..base import BaseExecutor
 from ..ssh import SSHExecutor, _sshfs
 from .base import BaseExecutorManager

dvc/repo/experiments/executor/ssh.py

@@ -5,21 +5,20 @@ import sys
 from contextlib import contextmanager
 from typing import TYPE_CHECKING, Callable, Iterable, Optional
 
+from dvc_ssh import SSHFileSystem
 from funcy import first
 
-from dvc.fs import SSHFileSystem
-from dvc.repo.experiments.base import (
+from ..refs import (
     EXEC_BRANCH,
     EXEC_CHECKPOINT,
     EXEC_HEAD,
     EXEC_MERGE,
     EXEC_NAMESPACE,
 )
-
 from .base import BaseExecutor, ExecutorInfo, ExecutorResult
 
 if TYPE_CHECKING:
-    from multiprocessing import Queue
+    from queue import Queue
 
     from scmrepo.git import Git
 

dvc/repo/experiments/gc.py

@@ -15,6 +15,7 @@ def gc(
     all_tags: Optional[bool] = False,
     all_commits: Optional[bool] = False,
     workspace: Optional[bool] = False,
+    commit_date: Optional[str] = None,
     queued: Optional[bool] = False,
 ):
     keep_revs = set(
@@ -22,6 +23,7 @@ def gc(
             all_branches=all_branches,
             all_tags=all_tags,
             all_commits=all_commits,
+            commit_date=commit_date,
             sha_only=True,
         )
     )
@@ -40,11 +42,12 @@ def gc(
     removed = len(to_remove)
 
     delete_stashes = []
-    for _, entry in repo.experiments.stash_revs.items():
+    stash = repo.experiments.celery_queue.stash
+    for stash_rev, entry in stash.stash_revs.items():
         if not queued or entry.baseline_rev not in keep_revs:
-            delete_stashes.append(entry.stash_index)
-    for index in sorted(delete_stashes, reverse=True):
-        repo.experiments.stash.drop(index)
+            delete_stashes.append(stash_rev)
+    if delete_stashes:
+        repo.experiments.celery_queue.remove(delete_stashes)
     removed += len(delete_stashes)
 
     return removed

dvc/repo/experiments/init.py

@@ -253,18 +253,25 @@ def init(
     live_metrics = f"{live_path}.json" if live_path else None
     live_plots = os.path.join(live_path, "scalars") if live_path else None
 
+    if type == "checkpoint":
+        outs_key = "checkpoints"
+        metrics_key = "metrics_persist_no_cache"
+        plots_key = "plots_persist_no_cache"
+    else:
+        outs_key = "outs"
+        metrics_key = "metrics_no_cache"
+        plots_key = "plots_no_cache"
+
     stage = repo.stage.create(
         name=name,
         cmd=context["cmd"],
         deps=compact([context.get("code"), context.get("data")]),
         params=[{params: None}] if params else None,
-        metrics_no_cache=compact([context.get("metrics"), live_metrics]),
-        plots_no_cache=compact([context.get("plots"), live_plots]),
         force=force,
         **{
-            "checkpoints"
-            if type == "checkpoint"
-            else "outs": compact([models])
+            outs_key: compact([models]),
+            metrics_key: compact([context.get("metrics"), live_metrics]),
+            plots_key: compact([context.get("plots"), live_plots]),
         },
     )
 

dvc/repo/experiments/pull.py

@@ -9,8 +9,8 @@ from dvc.repo.scm_context import scm_context
 from dvc.scm import TqdmGit, iter_revs
 from dvc.ui import ui
 
-from .base import ExpRefInfo
 from .exceptions import UnresolvedExpNamesError
+from .refs import ExpRefInfo
 from .utils import exp_commits, exp_refs, exp_refs_by_baseline, resolve_name
 
 logger = logging.getLogger(__name__)
@@ -82,7 +82,7 @@ def _pull(
     force: bool,
 ) -> Mapping[SyncStatus, List["ExpRefInfo"]]:
     refspec_list = [f"{exp_ref}:{exp_ref}" for exp_ref in refs]
-    logger.debug(f"git pull experiment '{git_remote}' -> '{refspec_list}'")
+    logger.debug("git pull experiment '%s' -> '%s'", git_remote, refspec_list)
 
     with TqdmGit(desc="Fetching git refs") as pbar:
         results: Mapping[str, SyncStatus] = repo.scm.fetch_refspecs(
@@ -113,7 +113,7 @@ def _pull_cache(
     if isinstance(refs, ExpRefInfo):
         refs = [refs]
     revs = list(exp_commits(repo.scm, refs))
-    logger.debug(f"dvc fetch experiment '{refs}'")
+    logger.debug("dvc fetch experiment '%s'", refs)
     repo.fetch(
         jobs=jobs, remote=dvc_remote, run_cache=run_cache, revs=revs, odb=odb
     )

dvc/repo/experiments/push.py

@@ -9,8 +9,8 @@ from dvc.repo.scm_context import scm_context
 from dvc.scm import TqdmGit, iter_revs
 from dvc.ui import ui
 
-from .base import ExpRefInfo
 from .exceptions import UnresolvedExpNamesError
+from .refs import ExpRefInfo
 from .utils import exp_commits, exp_refs, exp_refs_by_baseline, resolve_name
 
 logger = logging.getLogger(__name__)
@@ -85,7 +85,7 @@ def _push(
     from ...scm import GitAuthError
 
     refspec_list = [f"{exp_ref}:{exp_ref}" for exp_ref in refs]
-    logger.debug(f"git push experiment '{refs}' -> '{git_remote}'")
+    logger.debug("git push experiment '%s' -> '%s'", refspec_list, git_remote)
 
     with TqdmGit(desc="Pushing git refs") as pbar:
         try:
@@ -118,5 +118,5 @@ def _push_cache(
     if isinstance(refs, ExpRefInfo):
         refs = [refs]
     revs = list(exp_commits(repo.scm, refs))
-    logger.debug(f"dvc push experiment '{refs}'")
+    logger.debug("dvc push experiment '%s'", refs)
     repo.push(jobs=jobs, remote=dvc_remote, run_cache=run_cache, revs=revs)

dvc/repo/experiments/queue/base.py

@@ -0,0 +1,692 @@
+import logging
+import os
+import sys
+from abc import ABC, abstractmethod
+from dataclasses import asdict, dataclass
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Collection,
+    Dict,
+    Generator,
+    Iterable,
+    List,
+    Mapping,
+    NamedTuple,
+    Optional,
+    Type,
+    Union,
+)
+
+from funcy import cached_property
+
+from dvc.dependency import ParamsDependency
+from dvc.env import DVCLIVE_RESUME
+from dvc.exceptions import DvcException
+from dvc.ui import ui
+
+from ..exceptions import CheckpointExistsError, ExperimentExistsError
+from ..executor.base import (
+    EXEC_PID_DIR,
+    EXEC_TMP_DIR,
+    BaseExecutor,
+    ExecutorResult,
+)
+from ..executor.local import WorkspaceExecutor
+from ..refs import EXEC_BASELINE, EXEC_HEAD, EXEC_MERGE, ExpRefInfo
+from ..stash import ExpStash, ExpStashEntry
+from ..utils import exp_refs_by_rev, scm_locked
+
+if TYPE_CHECKING:
+    from scmrepo.git import Git
+
+    from dvc.repo import Repo
+    from dvc.repo.experiments import Experiments
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass(frozen=True)
+class QueueEntry:
+    dvc_root: str
+    scm_root: str
+    stash_ref: str
+    stash_rev: str
+    baseline_rev: str
+    branch: Optional[str]
+    name: Optional[str]
+    head_rev: Optional[str] = None
+
+    def __eq__(self, other: object):
+        return (
+            isinstance(other, QueueEntry)
+            and self.dvc_root == other.dvc_root
+            and self.scm_root == other.scm_root
+            and self.stash_ref == other.stash_ref
+            and self.stash_rev == other.stash_rev
+        )
+
+    def asdict(self) -> Dict[str, Any]:
+        return asdict(self)
+
+    @classmethod
+    def from_dict(cls, d: Dict[str, Any]) -> "QueueEntry":
+        return cls(**d)
+
+
+class QueueGetResult(NamedTuple):
+    entry: QueueEntry
+    executor: BaseExecutor
+
+
+class QueueDoneResult(NamedTuple):
+    entry: QueueEntry
+    result: Optional[ExecutorResult]
+
+
+class BaseStashQueue(ABC):
+    """Naive Git-stash based experiment queue.
+
+    Maps queued experiments to (Git) stash reflog entries.
+    """
+
+    def __init__(
+        self, repo: "Repo", ref: str, failed_ref: Optional[str] = None
+    ):
+        """Construct a queue.
+
+        Arguments:
+            scm: Git SCM instance for this queue.
+            ref: Git stash ref for this queue.
+            failed_ref: Failed run Git stash ref for this queue.
+        """
+        self.repo = repo
+        self.ref = ref
+        self.failed_ref = failed_ref
+
+    @property
+    def scm(self) -> "Git":
+        return self.repo.scm
+
+    @cached_property
+    def stash(self) -> ExpStash:
+        return ExpStash(self.scm, self.ref)
+
+    @cached_property
+    def failed_stash(self) -> Optional[ExpStash]:
+        return ExpStash(self.scm, self.failed_ref) if self.failed_ref else None
+
+    @cached_property
+    def pid_dir(self) -> str:
+        return os.path.join(self.repo.tmp_dir, EXEC_TMP_DIR, EXEC_PID_DIR)
+
+    @cached_property
+    def args_file(self):
+        return os.path.join(self.repo.tmp_dir, BaseExecutor.PACKED_ARGS_FILE)
+
+    @abstractmethod
+    def put(self, *args, **kwargs) -> QueueEntry:
+        """Stash an experiment and add it to the queue."""
+
+    @abstractmethod
+    def get(self) -> QueueGetResult:
+        """Pop and return the first item in the queue."""
+
+    def remove(
+        self,
+        revs: Collection[str],
+        all_: bool = False,
+        queued: bool = False,
+        **kwargs,
+    ) -> List[str]:
+        """Remove the specified entries from the queue.
+
+        Arguments:
+            revs: Stash revisions or queued exp names to be removed.
+            queued: Remove all queued tasks.
+            all: Remove all tasks.
+
+        Returns:
+            Revisions (or names) which were removed.
+        """
+
+        if all_ or queued:
+            return self.clear()
+
+        name_to_remove: List[str] = []
+        entry_to_remove: List[ExpStashEntry] = []
+        queue_entries = self.match_queue_entry_by_name(
+            revs, self.iter_queued()
+        )
+        for name, entry in queue_entries.items():
+            if entry:
+                entry_to_remove.append(self.stash.stash_revs[entry.stash_rev])
+                name_to_remove.append(name)
+
+        self.stash.remove_revs(entry_to_remove)
+        return name_to_remove
+
+    def clear(self, **kwargs) -> List[str]:
+        """Remove all entries from the queue."""
+        stash_revs = self.stash.stash_revs
+        name_to_remove = list(stash_revs)
+        self.stash.remove_revs(list(stash_revs.values()))
+
+        return name_to_remove
+
+    def status(self) -> List[Dict[str, Any]]:
+        """Show the status of exp tasks in queue"""
+        from datetime import datetime
+
+        result: List[Dict[str, Optional[str]]] = []
+
+        def _get_timestamp(rev: str) -> datetime:
+            commit = self.scm.resolve_commit(rev)
+            return datetime.fromtimestamp(commit.commit_time)
+
+        def _format_entry(
+            entry: QueueEntry,
+            exp_result: Optional[ExecutorResult] = None,
+            status: str = "Unknown",
+        ) -> Dict[str, Any]:
+            name = entry.name
+            if not name and exp_result and exp_result.ref_info:
+                name = exp_result.ref_info.name
+            # NOTE: We fallback to Unknown status for experiments
+            # generated in prior (incompatible) DVC versions
+            return {
+                "rev": entry.stash_rev,
+                "name": name,
+                "timestamp": _get_timestamp(entry.stash_rev),
+                "status": status,
+            }
+
+        result.extend(
+            _format_entry(queue_entry, status="Running")
+            for queue_entry in self.iter_active()
+        )
+        result.extend(
+            _format_entry(queue_entry, status="Queued")
+            for queue_entry in self.iter_queued()
+        )
+        result.extend(
+            _format_entry(queue_entry, status="Failed")
+            for queue_entry, _ in self.iter_failed()
+        )
+        result.extend(
+            _format_entry(queue_entry, exp_result=exp_result, status="Success")
+            for queue_entry, exp_result in self.iter_success()
+        )
+        return result
+
+    @abstractmethod
+    def iter_queued(self) -> Generator[QueueEntry, None, None]:
+        """Iterate over items in the queue."""
+
+    @abstractmethod
+    def iter_active(self) -> Generator[QueueEntry, None, None]:
+        """Iterate over items which are being actively processed."""
+
+    @abstractmethod
+    def iter_done(self) -> Generator[QueueDoneResult, None, None]:
+        """Iterate over items which been processed."""
+
+    @abstractmethod
+    def iter_success(self) -> Generator[QueueDoneResult, None, None]:
+        """Iterate over items which been success."""
+
+    @abstractmethod
+    def iter_failed(self) -> Generator[QueueDoneResult, None, None]:
+        """Iterate over items which been failed."""
+
+    @abstractmethod
+    def reproduce(self) -> Mapping[str, Mapping[str, str]]:
+        """Reproduce queued experiments sequentially."""
+
+    @abstractmethod
+    def get_result(self, entry: QueueEntry) -> Optional[ExecutorResult]:
+        """Return result of the specified item.
+
+        This method blocks until the specified item has been collected.
+        """
+
+    @abstractmethod
+    def kill(self, revs: str) -> None:
+        """Kill the specified running entries in the queue.
+
+        Arguments:
+            revs: Stash revs or running exp name to be killed.
+        """
+
+    @abstractmethod
+    def shutdown(self, kill: bool = False):
+        """Shutdown the queue worker.
+
+        Arguments:
+            kill: If True, the any active experiments will be killed and the
+                worker will shutdown immediately. If False, the worker will
+                finish any active experiments before shutting down.
+        """
+
+    def logs(
+        self,
+        rev: str,
+        encoding: Optional[str] = None,
+        follow: bool = False,
+    ):
+        """Print redirected output logs for an exp process.
+
+        Args:
+            rev: Stash rev or exp name.
+            encoding: Text encoding for redirected output. Defaults to
+                `locale.getpreferredencoding()`.
+            follow: Attach to running exp process and follow additional
+                output.
+        """
+
+    def _stash_exp(
+        self,
+        *args,
+        params: Optional[Dict[str, List[str]]] = None,
+        resume_rev: Optional[str] = None,
+        baseline_rev: Optional[str] = None,
+        branch: Optional[str] = None,
+        name: Optional[str] = None,
+        **kwargs,
+    ) -> QueueEntry:
+        """Stash changes from the workspace as an experiment.
+
+        Args:
+            params: Dict mapping paths to `Hydra Override`_ patterns,
+                provided via `exp run --set-param`.
+            resume_rev: Optional checkpoint resume rev.
+            baseline_rev: Optional baseline rev for this experiment, defaults
+                to the current SCM rev.
+            branch: Optional experiment branch name. If specified, the
+                experiment will be added to `branch` instead of creating
+                a new branch.
+            name: Optional experiment name. If specified this will be used as
+                the human-readable name in the experiment branch ref. Has no
+                effect of branch is specified.
+
+        .. _Hydra Override:
+            https://hydra.cc/docs/next/advanced/override_grammar/basic/
+        """
+        with self.scm.detach_head(client="dvc") as orig_head:
+            stash_head = orig_head
+            if baseline_rev is None:
+                baseline_rev = orig_head
+
+            with self.scm.stash_workspace() as workspace:
+                try:
+                    if workspace:
+                        self.stash.apply(workspace)
+
+                    if resume_rev:
+                        # move HEAD to the resume rev so that the stashed diff
+                        # only contains changes relative to resume rev
+                        stash_head = resume_rev
+                        self.scm.set_ref(
+                            "HEAD",
+                            resume_rev,
+                            message=f"dvc: resume from HEAD {resume_rev[:7]}",
+                        )
+                        self.scm.reset()
+
+                    # update experiment params from command line
+                    if params:
+                        self._update_params(params)
+
+                    # DVC commit data deps to preserve state across workspace
+                    # & tempdir runs
+                    self._stash_commit_deps(*args, **kwargs)
+
+                    if resume_rev:
+                        if branch:
+                            branch_name = ExpRefInfo.from_ref(branch).name
+                        else:
+                            branch_name = f"{resume_rev[:7]}"
+                        if self.scm.is_dirty(untracked_files=False):
+                            ui.write(
+                                "Modified checkpoint experiment based on "
+                                f"'{branch_name}' will be created",
+                            )
+                            branch = None
+                        elif (
+                            not branch
+                            or self.scm.get_ref(branch) != resume_rev
+                        ):
+                            err_msg = [
+                                (
+                                    "Nothing to do for unchanged checkpoint "
+                                    f"'{resume_rev[:7]}'. "
+                                )
+                            ]
+                            if branch:
+                                err_msg.append(
+                                    "To resume from the head of this "
+                                    "experiment, use "
+                                    f"'dvc exp apply {branch_name}'."
+                                )
+                            else:
+                                names = [
+                                    ref_info.name
+                                    for ref_info in exp_refs_by_rev(
+                                        self.scm, resume_rev
+                                    )
+                                ]
+                                if len(names) > 3:
+                                    names[3:] = [
+                                        f"... ({len(names) - 3} more)"
+                                    ]
+                                err_msg.append(
+                                    "To resume an experiment containing this "
+                                    "checkpoint, apply one of these heads:\n"
+                                    "\t{}".format(", ".join(names))
+                                )
+                            raise DvcException("".join(err_msg))
+                        else:
+                            ui.write(
+                                "Existing checkpoint experiment "
+                                f"'{branch_name}' will be resumed"
+                            )
+                        if name:
+                            logger.warning(
+                                "Ignoring option '--name %s' for resumed "
+                                "experiment. Existing experiment name will"
+                                "be preserved instead.",
+                                name,
+                            )
+
+                    # save additional repro command line arguments
+                    run_env = {DVCLIVE_RESUME: "1"} if resume_rev else {}
+                    self._pack_args(*args, run_env=run_env, **kwargs)
+
+                    # save experiment as a stash commit
+                    msg = self._stash_msg(
+                        stash_head,
+                        baseline_rev=baseline_rev,
+                        branch=branch,
+                        name=name,
+                    )
+                    stash_rev = self.stash.push(message=msg)
+                    logger.debug(
+                        (
+                            "Stashed experiment '%s' with baseline '%s' "
+                            "for future execution."
+                        ),
+                        stash_rev[:7],
+                        baseline_rev[:7],
+                    )
+                finally:
+                    if resume_rev:
+                        # NOTE: this set_ref + reset() is equivalent to
+                        # `git reset orig_head` (our SCM reset() only operates
+                        # on HEAD rather than any arbitrary commit)
+                        self.scm.set_ref(
+                            "HEAD", orig_head, message="dvc: restore HEAD"
+                        )
+                        self.scm.reset()
+                    # Revert any of our changes before prior unstashing
+                    self.scm.reset(hard=True)
+
+        return QueueEntry(
+            self.repo.root_dir,
+            self.scm.root_dir,
+            self.ref,
+            stash_rev,
+            baseline_rev,
+            branch,
+            name,
+            stash_head,
+        )
+
+    def _stash_commit_deps(self, *args, **kwargs):
+        if len(args):
+            targets = args[0]
+        else:
+            targets = kwargs.get("targets")
+        if isinstance(targets, str):
+            targets = [targets]
+        elif not targets:
+            targets = [None]
+        for target in targets:
+            self.repo.commit(
+                target,
+                with_deps=True,
+                recursive=kwargs.get("recursive", False),
+                force=True,
+                allow_missing=True,
+                data_only=True,
+            )
+
+    @staticmethod
+    def _stash_msg(
+        rev: str,
+        baseline_rev: str,
+        branch: Optional[str] = None,
+        name: Optional[str] = None,
+    ) -> str:
+        if not baseline_rev:
+            baseline_rev = rev
+        msg = ExpStash.format_message(rev, baseline_rev, name)
+        if branch:
+            return f"{msg}:{branch}"
+        return msg
+
+    def _pack_args(self, *args, **kwargs) -> None:
+        import pickle
+
+        if os.path.exists(self.args_file) and self.scm.is_tracked(
+            self.args_file
+        ):
+            logger.warning(
+                (
+                    "Temporary DVC file '.dvc/tmp/%s' exists and was "
+                    "likely committed to Git by mistake. It should be removed "
+                    "with:\n"
+                    "\tgit rm .dvc/tmp/%s"
+                ),
+                BaseExecutor.PACKED_ARGS_FILE,
+                BaseExecutor.PACKED_ARGS_FILE,
+            )
+            with open(self.args_file, "rb") as fobj:
+                try:
+                    data = pickle.load(fobj)
+                except Exception:  # pylint: disable=broad-except
+                    data = {}
+            extra = int(data.get("extra", 0)) + 1
+        else:
+            extra = None
+        BaseExecutor.pack_repro_args(
+            self.args_file, *args, extra=extra, **kwargs
+        )
+        self.scm.add(self.args_file)
+
+    @staticmethod
+    def _format_new_params_msg(new_params, config_path):
+        """Format an error message for when new parameters are identified"""
+        new_param_count = len(new_params)
+        pluralise = "s are" if new_param_count > 1 else " is"
+        param_list = ", ".join(new_params)
+        return (
+            f"{new_param_count} parameter{pluralise} missing "
+            f"from '{config_path}': {param_list}"
+        )
+
+    def _update_params(self, params: Dict[str, List[str]]):
+        """Update param files with the provided `Hydra Override`_ patterns.
+
+        Args:
+            params: Dict mapping paths to `Hydra Override`_ patterns,
+                provided via `exp run --set-param`.
+
+        .. _Hydra Override:
+            https://hydra.cc/docs/advanced/override_grammar/basic/
+        """
+        logger.debug("Using experiment params '%s'", params)
+
+        try:
+            from dvc.utils.hydra import apply_overrides, compose_and_dump
+        except ValueError:
+            if sys.version_info >= (3, 11):
+                raise DvcException(
+                    "--set-param is not supported in Python >= 3.11"
+                )
+            raise
+
+        hydra_config = self.repo.config.get("hydra", {})
+        hydra_enabled = hydra_config.get("enabled", False)
+        hydra_output_file = ParamsDependency.DEFAULT_PARAMS_FILE
+        for path, overrides in params.items():
+            if hydra_enabled and path == hydra_output_file:
+                config_dir = os.path.join(
+                    self.repo.root_dir, hydra_config.get("config_dir", "conf")
+                )
+                config_name = hydra_config.get("config_name", "config")
+                compose_and_dump(
+                    path,
+                    config_dir,
+                    config_name,
+                    overrides,
+                )
+            else:
+                apply_overrides(path, overrides)
+
+        # Force params file changes to be staged in git
+        # Otherwise in certain situations the changes to params file may be
+        # ignored when we `git stash` them since mtime is used to determine
+        # whether the file is dirty
+        self.scm.add(list(params.keys()))
+
+    @staticmethod
+    @scm_locked
+    def setup_executor(
+        exp: "Experiments",
+        queue_entry: QueueEntry,
+        executor_cls: Type[BaseExecutor] = WorkspaceExecutor,
+        **kwargs,
+    ) -> BaseExecutor:
+        scm = exp.scm
+        stash = ExpStash(scm, queue_entry.stash_ref)
+        stash_rev = queue_entry.stash_rev
+        stash_entry = stash.stash_revs.get(
+            stash_rev,
+            ExpStashEntry(None, stash_rev, stash_rev, None, None),
+        )
+        if stash_entry.stash_index is not None:
+            stash.drop(stash_entry.stash_index)
+
+        scm.set_ref(EXEC_HEAD, stash_entry.head_rev)
+        scm.set_ref(EXEC_MERGE, stash_rev)
+        scm.set_ref(EXEC_BASELINE, stash_entry.baseline_rev)
+
+        # Executor will be initialized with an empty git repo that
+        # we populate by pushing:
+        #   EXEC_HEAD - the base commit for this experiment
+        #   EXEC_MERGE - the unmerged changes (from our stash)
+        #       to be reproduced
+        #   EXEC_BASELINE - the baseline commit for this experiment
+        return executor_cls.from_stash_entry(
+            exp.repo, stash_rev, stash_entry, **kwargs
+        )
+
+    def get_infofile_path(self, name: str) -> str:
+        return os.path.join(
+            self.pid_dir,
+            name,
+            f"{name}{BaseExecutor.INFOFILE_EXT}",
+        )
+
+    @staticmethod
+    @scm_locked
+    def collect_executor(
+        exp: "Experiments",
+        executor: BaseExecutor,
+        exec_result: ExecutorResult,
+    ) -> Dict[str, str]:
+        results = {}
+
+        def on_diverged(ref: str, checkpoint: bool):
+            ref_info = ExpRefInfo.from_ref(ref)
+            if checkpoint:
+                raise CheckpointExistsError(ref_info.name)
+            raise ExperimentExistsError(ref_info.name)
+
+        for ref in executor.fetch_exps(
+            exp.scm,
+            force=exec_result.force,
+            on_diverged=on_diverged,
+        ):
+            exp_rev = exp.scm.get_ref(ref)
+            if exp_rev:
+                assert exec_result.exp_hash
+                logger.debug("Collected experiment '%s'.", exp_rev[:7])
+                results[exp_rev] = exec_result.exp_hash
+
+        if exec_result.ref_info is not None:
+            executor.collect_cache(exp.repo, exec_result.ref_info)
+
+        return results
+
+    def match_queue_entry_by_name(
+        self,
+        exp_names: Collection[str],
+        *entries: Iterable[Union[QueueEntry, QueueDoneResult]],
+    ) -> Dict[str, Optional[QueueEntry]]:
+        from funcy import concat
+
+        entry_name_dict: Dict[str, QueueEntry] = {}
+        entry_rev_dict: Dict[str, QueueEntry] = {}
+        for entry in concat(*entries):
+            if isinstance(entry, QueueDoneResult):
+                queue_entry: QueueEntry = entry.entry
+                if (
+                    entry.result is not None
+                    and entry.result.ref_info is not None
+                ):
+                    name: Optional[str] = entry.result.ref_info.name
+                else:
+                    name = queue_entry.name
+            else:
+                queue_entry = entry
+                name = queue_entry.name
+            if name:
+                entry_name_dict[name] = queue_entry
+            entry_rev_dict[queue_entry.stash_rev] = queue_entry
+
+        result: Dict[str, Optional[QueueEntry]] = {}
+        for exp_name in exp_names:
+            result[exp_name] = None
+            if exp_name in entry_name_dict:
+                result[exp_name] = entry_name_dict[exp_name]
+                continue
+            if self.scm.is_sha(exp_name):
+                for rev in entry_rev_dict:
+                    if rev.startswith(exp_name.lower()):
+                        result[exp_name] = entry_rev_dict[rev]
+                        break
+
+        return result
+
+    def stash_failed(self, entry: QueueEntry) -> None:
+        """Add an entry to the failed exp stash.
+
+        Arguments:
+            entry: Failed queue entry to add. ``entry.stash_rev`` must be a
+                valid Git stash commit.
+        """
+        if self.failed_stash is not None:
+            assert entry.head_rev
+            logger.debug("Stashing failed exp '%s'", entry.stash_rev[:7])
+            msg = self.failed_stash.format_message(
+                entry.head_rev,
+                baseline_rev=entry.baseline_rev,
+                name=entry.name,
+                branch=entry.branch,
+            )
+            self.scm.set_ref(
+                self.failed_stash.ref,
+                entry.stash_rev,
+                message=f"commit: {msg}",
+            )