deps: bump dvc-data to 0.0.23

Fixes #7374

.dvcignore

@@ -0,0 +1,4 @@
+/tests
+/dvc
+/.github
+/bin

.git_archival.txt

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

.github/FUNDING.yml

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

.github/mergify.yml

@@ -1,7 +1,7 @@
 pull_request_rules:
   - name: backport patches to 1.11
     conditions:
-      - base=master
+      - base=main
       - label=backport-1.11
     actions:
       backport:

.github/workflows/benchmarks.yaml

@@ -11,13 +11,13 @@ 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@v2
+      - uses: actions/setup-python@v4
         with:
-            python-version: 3.7
+            python-version: 3.8
-      - uses: iterative/dvc-bench@master
+      - uses: iterative/dvc-bench@main
         with:
             pytest_options: "-k 'test_init or test_help'"
 

.github/workflows/packages.yaml

@@ -2,7 +2,7 @@ name: Build packages
 on:
   push:
     branches:
-    - master
+    - main
   release:
     types: [released, prereleased]
   workflow_dispatch:
@@ -10,11 +10,11 @@ jobs:
   deb:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.4.0
+    - uses: actions/checkout@v3
       with:
           fetch-depth: 0
     - name: Set up Python 3.9
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
         python-version: 3.9
     - name: Set up Ruby 2.4
@@ -41,11 +41,11 @@ jobs:
   rpm:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.4.0
+    - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python 3.9
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
         python-version: 3.9
     - name: Set up Ruby 2.4
@@ -72,11 +72,11 @@ jobs:
   osxpkg:
     runs-on: macos-10.15
     steps:
-    - uses: actions/checkout@v2.4.0
+    - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python 3.9
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
         python-version: 3.9
     - name: Set up Ruby 2.4
@@ -106,11 +106,11 @@ jobs:
   exe:
     runs-on: windows-2019
     steps:
-    - uses: actions/checkout@v2.4.0
+    - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python 3.9
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
         python-version: 3.9
     - name: Install deps
@@ -136,13 +136,13 @@ jobs:
   pip:
     runs-on: ubuntu-18.04
     steps:
-    - uses: actions/checkout@v2.4.0
+    - uses: actions/checkout@v3
       with:
         fetch-depth: 0
-    - name: Set up Python 3.7
+    - name: Set up Python 3.8
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
-        python-version: 3.7
+        python-version: 3.8
     - name: Install
       run: pip install ".[all,tests]"
     - name: Build packages

.github/workflows/tests.yaml

@@ -17,23 +17,24 @@ 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@v2.4.0
+    - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python 3.8
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
         python-version: 3.8
     - name: Install requirements
       run: |
         pip install --upgrade pip setuptools wheel
+        pip install collective.checkdocs==0.2
         pip install ".[dev]" pre-commit
     - name: Check README
       run: python setup.py checkdocs
-    - uses: pre-commit/action@v2.0.3
+    - uses: pre-commit/action@v3.0.0
   tests:
     timeout-minutes: 45
     runs-on: ${{ matrix.os }}
@@ -41,13 +42,13 @@ jobs:
       fail-fast: false
       matrix:
         os: [ubuntu-20.04, windows-latest, macos-latest]
-        pyv: ["3.7", "3.8", "3.9", "3.10"]
+        pyv: ["3.8", "3.9", "3.10"]
     steps:
-    - uses: actions/checkout@v2.4.0
+    - uses: actions/checkout@v3
       with:
         fetch-depth: 0
     - name: Set up Python
-      uses: actions/setup-python@v2.2.2
+      uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.pyv }}
     - name: get pip cache dir
@@ -56,7 +57,7 @@ jobs:
         echo "::set-output name=dir::$(pip cache dir)"
     - name: set pip cache
       id: pip-cache
-      uses: actions/cache@v2.1.7
+      uses: actions/cache@v3
       with:
         path: ${{ steps.pip-cache-dir.outputs.dir }}
         key: ${{ runner.os }}-pip-${{ hashFiles('setup.cfg') }}
@@ -67,10 +68,6 @@ jobs:
       run: |
         pip install --upgrade pip setuptools wheel
         pip install -e ".[dev]"
-    - name: setup git
-      run: |
-        git config --global user.email "[email protected]"
-        git config --global user.name "DVC Tester"
     - name: run tests
       timeout-minutes: 40
       run: >-
@@ -78,12 +75,12 @@ jobs:
         --cov-report=xml --cov-report=term
         ${{ env.extra_test_args }}
     - name: upload coverage report
-      uses: codecov/codecov-action@v2.1.0
+      uses: codecov/codecov-action@v3
       with:
         file: ./coverage.xml
         fail_ci_if_error: false
   notify:
-    if: github.ref == 'refs/heads/master' && failure()
+    if: github.ref == 'refs/heads/main' && failure()
     needs:
       - lint
       - tests
@@ -94,6 +91,6 @@ jobs:
       env:
         SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}
         SLACK_COLOR: ${{ job.status }}
-        SLACK_MESSAGE: 'CI Failed on master :boom:'
+        SLACK_MESSAGE: 'CI Failed on main :boom:'
         SLACK_TITLE: CI Status
         SLACK_USERNAME: DVC-CI

.gitignore

@@ -1,10 +1,13 @@
+__pycache__/
 neatlynx.conf
-neatlynx/__pycache__
 .idea
 /cache
 *.pyc
-.env/
+
-.env2.7/
+.env*/
+.venv
+env/
+venv/
 .python-version
 
 .dvc.conf.lock
@@ -26,6 +29,7 @@ innosetup/config.ini
 
 .coverage
 .coverage.*
+coverage.xml
 
 *.sw?
 
@@ -40,3 +44,9 @@ scripts/ci/gcp-creds.json
 
 *~
 /dvc/_dvc_version.py
+
+.mypy_cache/
+.pytest_cache/
+.nox/
+.tox/
+htmlcov/

.pre-commit-config.yaml

@@ -6,9 +6,9 @@ repos:
       - id: black
         language_version: python3
     repo: https://github.com/ambv/black
-    rev: 21.12b0
+    rev: 22.6.0
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.0.1
+    rev: v4.3.0
     hooks:
       - id: check-toml
       - id: check-yaml

README.rst

@@ -2,37 +2,32 @@
 
 `Website <https://dvc.org>`_
 • `Docs <https://dvc.org/doc>`_
+• `VS Code Extension <https://marketplace.visualstudio.com/items?itemName=Iterative.dvc>`_
 • `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>`_
 
-|CI| |Maintainability| |Coverage| |Donate| |DOI|
+|CI| |Maintainability| |Coverage| |VS Code| |DOI|
 
 |PyPI| |Packages| |Brew| |Conda| |Choco| |Snap|
 
 |
 
-**Data Version Control** or **DVC** is an **open-source** tool for data science and machine
+**Data Version Control** or **DVC** is a command line tool and `VS Code Extension <https://marketplace.visualstudio.com/items?itemName=Iterative.dvc>`_ to help you develop reproducible machine learning projects:
-learning projects. Key features:
 
-#. Simple **command line** Git-like experience. Does not require installing and maintaining
+#. **Version** your data and models. Store them in your cloud storage but keep
-   any databases. Does not depend on any proprietary online services.
+   their version info in your Git repo.
 
-#. Management and versioning of **datasets** and **machine learning
+#. **Iterate** fast with lightweight pipelines. When you make changes, only run
-   models**. Data can be saved in S3, Google cloud, Azure, Alibaba 
+   the steps impacted by those changes.
-   cloud, SSH server, HDFS, or even local HDD RAID.
 
-#. Makes projects **reproducible** and **shareable**; helping to answer questions about how
+#. **Track** experiments in your local Git repo (no servers needed).
-   a model was built.
 
-#. Helps manage experiments with Git tags/branches and **metrics** tracking.
+#. **Compare** any data, code, parameters, model, or performance plots
 
-**DVC** aims to replace spreadsheet and document sharing tools (such as Excel or Google Docs)
+#. **Share** experiments and automatically reproduce anyone's experiment.
-frequently used as both knowledge repositories and team ledgers.
-DVC also replaces both ad-hoc scripts to track, move, and deploy different model versions
-and ad-hoc data file suffixes and prefixes.
 
 .. contents:: **Contents**
   :backlinks: none
@@ -58,7 +53,7 @@ Google Cloud, etc) or any on-premise network storage (via SSH, for example).
 
 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
+and output information to be saved. See the quick start sections below or
 the `Get Started <https://dvc.org/doc/get-started>`_ tutorial to learn more.
 
 Quick start
@@ -86,14 +81,27 @@ Please read `Get Started <https://dvc.org/doc/get-started>`_ guide for a full ve
 |                                   | | ``$ dvc push``                                                           |
 +-----------------------------------+----------------------------------------------------------------------------+
 
+Visual Studio Code Extension
+============================
+
+|VS Code|
+
+To get 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|
 
@@ -103,7 +111,7 @@ Snap (Snapcraft/Linux)
 
 This corresponds to the latest tagged release.
 Add ``--beta`` for the latest tagged release candidate,
-or ``--edge`` for the latest ``master`` version.
+or ``--edge`` for the latest ``main`` version.
 
 Choco (Chocolatey/Windows)
 --------------------------
@@ -169,22 +177,24 @@ Ubuntu / Debian (deb)
 .. code-block:: bash
 
    sudo wget https://dvc.org/deb/dvc.list -O /etc/apt/sources.list.d/dvc.list
-   sudo apt-get update
+   wget -qO - https://dvc.org/deb/iterative.asc | sudo apt-key add -
-   sudo apt-get install dvc
+   sudo apt update
+   sudo apt install dvc
 
 Fedora / CentOS (rpm)
 ^^^^^^^^^^^^^^^^^^^^^
 .. code-block:: bash
 
    sudo wget https://dvc.org/rpm/dvc.repo -O /etc/yum.repos.d/dvc.repo
+   sudo rpm --import https://dvc.org/rpm/iterative.asc
    sudo yum update
    sudo yum install dvc
 
 Comparison to related technologies
 ==================================
 
-#. Data Engineering tools such as `AirFlow <https://airflow.apache.org/>`,
+#. Data Engineering tools such as `AirFlow <https://airflow.apache.org/>`_,
-   `Luigi <https://github.com/spotify/luigi>`, and others - in DVC data,
+   `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
@@ -212,14 +222,14 @@ Comparison to related technologies
 #. `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
+   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!
@@ -254,7 +264,10 @@ Barrak, A., Eghan, E.E. and Adams, B. `On the Co-evolution of ML Pipelines and S
    :target: https://dvc.org
    :alt: DVC logo
 
-.. |CI| image:: https://github.com/iterative/dvc/workflows/Tests/badge.svg?branch=master
+.. |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
 
@@ -262,14 +275,10 @@ 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
 
-.. |Coverage| image:: https://codecov.io/gh/iterative/dvc/branch/master/graph/badge.svg
+.. |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
@@ -305,3 +314,7 @@ 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/__main__.py

@@ -1,7 +1,7 @@
 """Main entry point for DVC command line tool."""
 import sys
 
-from dvc.main import main
+from dvc.cli import main
 
 if __name__ == "__main__":
     sys.exit(main(sys.argv[1:]))

dvc/_debug.py

@@ -1,10 +1,32 @@
 from contextlib import ExitStack, contextmanager
+from datetime import datetime
 from typing import TYPE_CHECKING, Callable, Union
 
 if TYPE_CHECKING:
     from argparse import Namespace
 
 
+@contextmanager
+def viztracer_profile(
+    path: Union[Callable[[], str], str],
+    max_stack_depth: int = -1,
+):
+    try:
+        import viztracer  # pylint: disable=import-error
+    except ImportError:
+        print("Failed to run profiler, viztracer is not installed")
+        yield
+        return
+
+    tracer = viztracer.VizTracer(max_stack_depth=max_stack_depth)
+
+    tracer.start()
+    yield
+    tracer.stop()
+
+    tracer.save(path() if callable(path) else path)
+
+
 @contextmanager
 def yappi_profile(
     path: Union[Callable[[], str], str] = None, wall_clock: bool = True
@@ -100,12 +122,18 @@ def debugtools(args: "Namespace" = None, **kwargs):
         if kw.get("instrument") or kw.get("instrument_open"):
             stack.enter_context(instrument(kw.get("instrument_open", False)))
         if kw.get("yappi"):
-            from datetime import datetime
-
             output = "callgrind.dvc-{0:%Y%m%d}_{0:%H%M%S}.out"
             stack.enter_context(
                 yappi_profile(path=lambda: output.format(datetime.now()))
             )
+        if kw.get("viztracer") or kw.get("viztracer_depth"):
+            output = "viztracer.dvc-{0:%Y%m%d}_{0:%H%M%S}.json"
+            stack.enter_context(
+                viztracer_profile(
+                    path=lambda: output.format(datetime.now()),
+                    max_stack_depth=kw.get("viztracer_depth") or -1,
+                )
+            )
         yield
 
 
@@ -118,6 +146,10 @@ def add_debugging_flags(parser):
     parser.add_argument(
         "--yappi", action="store_true", default=False, help=SUPPRESS
     )
+    parser.add_argument(
+        "--viztracer", action="store_true", default=False, help=SUPPRESS
+    )
+    parser.add_argument("--viztracer-depth", type=int, help=SUPPRESS)
     parser.add_argument("--cprofile-dump", help=SUPPRESS)
     parser.add_argument(
         "--pdb", action="store_true", default=False, help=SUPPRESS

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/api.py

@@ -1,122 +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.fs.path.join(_repo.root_dir, path)
-        with reraise(FileNotFoundError, PathMissingError(path, repo)):
-            metadata = _repo.repo_fs.metadata(fs_path)
-
-        if not metadata.is_dvc:
-            raise OutputNotFoundError(path, repo)
-
-        cloud = metadata.repo.cloud
-        md5 = metadata.repo.dvcfs.info(fs_path)["md5"]
-        return 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 +1,274 @@
+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.
+        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

@@ -0,0 +1,218 @@
+"""This module provides an entrypoint to the dvc cli and parsing utils."""
+
+import logging
+import sys
+
+# Workaround for CPython bug. See [1] and [2] for more info.
+# [1] https://github.com/aws/aws-cli/blob/1.16.277/awscli/clidriver.py#L55
+# [2] https://bugs.python.org/issue29288
+from typing import Optional
+
+"".encode("idna")
+
+logger = logging.getLogger("dvc")
+
+
+class DvcParserError(Exception):
+    """Base class for CLI parser errors."""
+
+    def __init__(self):
+        super().__init__("parser error")
+
+
+def parse_args(argv=None):
+    """Parses CLI arguments.
+
+    Args:
+        argv: optional list of arguments to parse. sys.argv is used by default.
+
+    Raises:
+        DvcParserError: raised for argument parsing errors.
+    """
+    from .parser import get_main_parser
+
+    parser = get_main_parser()
+    args = parser.parse_args(argv)
+    args.parser = parser
+    return args
+
+
+def _log_unknown_exceptions() -> None:
+    from dvc.info import get_dvc_info
+    from dvc.logger import FOOTER
+
+    logger.exception("unexpected error")
+    if logger.isEnabledFor(logging.DEBUG):
+        logger.debug("Version info for developers:\n%s", get_dvc_info())
+    logger.info(FOOTER)
+    return None
+
+
+def _log_exceptions(exc: Exception) -> Optional[int]:
+    """Try to log some known exceptions, that are not DVCExceptions."""
+    from dvc.utils import error_link, format_link
+
+    if isinstance(exc, OSError):
+        import errno
+
+        if exc.errno == errno.EMFILE:
+            logger.exception(
+                "too many open files, please visit "
+                "{} to see how to handle this "
+                "problem".format(error_link("many-files")),
+                extra={"tb_only": True},
+            )
+        else:
+            _log_unknown_exceptions()
+        return None
+
+    from dvc.fs import AuthError, ConfigError, RemoteMissingDepsError
+
+    if isinstance(exc, RemoteMissingDepsError):
+        from dvc.utils.pkg import PKG
+
+        proto = exc.protocol
+        by_pkg = {
+            "pip": f"pip install 'dvc[{proto}]'",
+            "conda": f"conda install -c conda-forge dvc-{proto}",
+        }
+
+        cmd = by_pkg.get(PKG)
+        if cmd:
+            link = format_link("https://dvc.org/doc/install")
+            hint = (
+                f"To install dvc with those dependencies, run:\n"
+                "\n"
+                f"\t{cmd}\n"
+                "\n"
+                f"See {link} for more info."
+            )
+        else:
+            link = format_link("https://github.com/iterative/dvc/issues")
+            hint = f"\nPlease report this bug to {link}. Thank you!"
+
+        logger.exception(
+            f"URL '{exc.url}' is supported but requires these missing "
+            f"dependencies: {exc.missing_deps}. {hint}",
+            extra={"tb_only": True},
+        )
+        return None
+
+    if isinstance(exc, (AuthError, ConfigError)):
+        link = format_link("https://man.dvc.org/remote/modify")
+        logger.exception("configuration error")
+        logger.exception(
+            f"{exc!s}\nLearn more about configuration settings at {link}.",
+            extra={"tb_only": True},
+        )
+        return 251
+
+    from dvc_data.hashfile.cache import DiskError
+
+    if isinstance(exc, DiskError):
+        from dvc.utils import relpath
+
+        directory = relpath(exc.directory)
+        logger.exception(
+            f"Could not open pickled '{exc.type}' cache.\n"
+            f"Remove the '{directory}' directory and then retry this command."
+            f"\nSee {error_link('pickle')} for more information.",
+            extra={"tb_only": True},
+        )
+        return None
+
+    from dvc_data.build import IgnoreInCollectedDirError
+
+    if isinstance(exc, IgnoreInCollectedDirError):
+        logger.exception("")
+        return None
+
+    _log_unknown_exceptions()
+    return None
+
+
+def main(argv=None):  # noqa: C901
+    """Main entry point for dvc CLI.
+
+    Args:
+        argv: optional list of arguments to parse. sys.argv is used by default.
+
+    Returns:
+        int: command's return code.
+    """
+    from dvc._debug import debugtools
+    from dvc.config import ConfigError
+    from dvc.exceptions import DvcException, NotDvcRepoError
+    from dvc.logger import disable_other_loggers, set_loggers_level
+
+    # NOTE: stderr/stdout may be closed if we are running from dvc.daemon.
+    # On Linux we directly call cli.main after double forking and closing
+    # the copied parent's standard file descriptors. If we make any logging
+    # calls in this state it will cause an exception due to writing to a closed
+    # file descriptor.
+    if sys.stderr.closed:  # pylint: disable=using-constant-test
+        logging.disable()
+    elif sys.stdout.closed:  # pylint: disable=using-constant-test
+        logging.disable(logging.INFO)
+
+    args = None
+    disable_other_loggers()
+
+    outerLogLevel = logger.level
+    try:
+        args = parse_args(argv)
+
+        level = None
+        if args.quiet:
+            level = logging.CRITICAL
+        elif args.verbose == 1:
+            level = logging.DEBUG
+        elif args.verbose > 1:
+            level = logging.TRACE
+
+        if level is not None:
+            set_loggers_level(level)
+
+        logger.trace(args)
+
+        if not sys.stdout.closed and not args.quiet:
+            from dvc.ui import ui
+
+            ui.enable()
+
+        with debugtools(args):
+            cmd = args.func(args)
+            ret = cmd.do_run()
+    except ConfigError:
+        logger.exception("configuration error")
+        ret = 251
+    except KeyboardInterrupt:
+        logger.exception("interrupted by the user")
+        ret = 252
+    except NotDvcRepoError:
+        logger.exception("")
+        ret = 253
+    except DvcException:
+        ret = 255
+        logger.exception("")
+    except DvcParserError:
+        ret = 254
+    except Exception as exc:  # noqa, pylint: disable=broad-except
+        # pylint: disable=no-member
+        ret = _log_exceptions(exc) or 255
+
+    try:
+        from dvc import analytics
+
+        if analytics.is_enabled():
+            analytics.collect_and_send_report(args, ret)
+
+        return ret
+    finally:
+        logger.setLevel(outerLogLevel)
+
+        from dvc.external_repo import clean_repos
+
+        # Remove cached repos in the end of the call, these are anonymous
+        # so won't be reused by any other subsequent run anyway.
+        clean_repos()

dvc/cli/command.py

@@ -5,29 +5,6 @@ from abc import ABC, abstractmethod
 logger = logging.getLogger(__name__)
 
 
-def fix_subparsers(subparsers):
-    """Workaround for bug in Python 3. See more info at:
-    https://bugs.python.org/issue16308
-    https://github.com/iterative/dvc/issues/769
-
-    Args:
-        subparsers: subparsers to fix.
-    """
-    subparsers.required = True
-    subparsers.dest = "cmd"
-
-
-def append_doc_link(help_message, path):
-    from dvc.utils import format_link
-
-    if not path:
-        return help_message
-    doc_base = "https://man.dvc.org/"
-    return "{message}\nDocumentation: {link}".format(
-        message=help_message, link=format_link(doc_base + path)
-    )
-
-
 class CmdBase(ABC):
     UNINITIALIZED = False
 

dvc/cli/completion.py

@@ -1,17 +1,6 @@
-import argparse
-import logging
-
 import shtab
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+BASH_PREAMBLE = """
-from dvc.ui import ui
-
-logger = logging.getLogger(__name__)
-FILE = shtab.FILE
-DIR = shtab.DIRECTORY
-
-PREAMBLE = {
-    "bash": """
 # $1=COMP_WORDS[1]
 _dvc_compgen_DVCFiles() {
   compgen -d -S '/' -- $1  # recurse into subdirs
@@ -30,11 +19,12 @@ _dvc_compgen_stages_and_files() {
 }
 
 _dvc_compgen_exps() {
-    local _dvc_exps=($(dvc exp list -q --all --names-only))
+    local _dvc_exps=($(dvc exp list -q --all-commits --names-only))
     compgen -W "${_dvc_exps[*]}" -- $1
 }
-    """,
+"""
-    "zsh": """
+
+ZSH_PREAMBLE = """
 _dvc_compadd_DVCFiles() {
     _files -g '(*?.dvc|Dvcfile|dvc.yaml)'
 }
@@ -63,49 +53,21 @@ _dvc_compadd_stages_and_files() {
 }
 
 _dvc_compadd_exps() {
-    _describe 'experiments' "($(dvc exp list -q -a --names-only))"
+    _describe 'experiments' "($(dvc exp list -q --all-commits --names-only))"
 }
-    """,
+"""
+
+PREAMBLE = {
+    "bash": BASH_PREAMBLE,
+    "zsh": ZSH_PREAMBLE,
 }
 
+FILE = shtab.FILE
+DIR = shtab.DIRECTORY
 DVC_FILE = {"bash": "_dvc_compgen_DVCFiles", "zsh": "_dvc_compadd_DVCFiles"}
-
 STAGE = {"bash": "_dvc_compgen_stages", "zsh": "_dvc_compadd_stages"}
-
 DVCFILES_AND_STAGE = {
     "bash": "_dvc_compgen_stages_and_files",
     "zsh": "_dvc_compadd_stages_and_files",
 }
-
 EXPERIMENT = {"bash": "_dvc_compgen_exps", "zsh": "_dvc_compadd_exps"}
-
-
-class CmdCompletion(CmdBaseNoRepo):
-    def run(self):
-        from dvc.cli import get_main_parser
-
-        parser = get_main_parser()
-        shell = self.args.shell
-        script = shtab.complete(parser, shell=shell, preamble=PREAMBLE)
-        ui.write(script, force=True)
-        return 0
-
-
-def add_parser(subparsers, parent_parser):
-    COMPLETION_HELP = "Generate shell tab completion."
-    COMPLETION_DESCRIPTION = "Prints out shell tab completion scripts."
-    completion_parser = subparsers.add_parser(
-        "completion",
-        parents=[parent_parser],
-        description=append_doc_link(COMPLETION_DESCRIPTION, "completion"),
-        help=COMPLETION_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    completion_parser.add_argument(
-        "-s",
-        "--shell",
-        help="Shell syntax for completions.",
-        default="bash",
-        choices=["bash", "zsh"],
-    )
-    completion_parser.set_defaults(func=CmdCompletion)

dvc/cli/parser.py

@@ -1,11 +1,10 @@
-"""DVC command line interface"""
+"""Main parser for the dvc cli."""
 import argparse
 import logging
 import os
 import sys
 
-from ._debug import add_debugging_flags
+from dvc.commands import (
-from .command import (
     add,
     cache,
     check_ignore,
@@ -28,7 +27,6 @@ from .command import (
     imp_url,
     init,
     install,
-    live,
     ls,
     machine,
     metrics,
@@ -45,8 +43,8 @@ from .command import (
     update,
     version,
 )
-from .command.base import fix_subparsers
+
-from .exceptions import DvcParserError
+from . import DvcParserError
 
 logger = logging.getLogger(__name__)
 
@@ -87,7 +85,6 @@ COMMANDS = [
     stage,
     experiments,
     check_ignore,
-    live,
     machine,
 ]
 
@@ -96,7 +93,7 @@ def _find_parser(parser, cmd_cls):
     defaults = parser._defaults  # pylint: disable=protected-access
     if not cmd_cls or cmd_cls == defaults.get("func"):
         parser.print_help()
-        raise DvcParserError()
+        raise DvcParserError
 
     actions = parser._actions  # pylint: disable=protected-access
     for action in actions:
@@ -143,6 +140,8 @@ def get_parent_parser():
     When overwriting `-q` or `-v`, you need to instantiate a new object
     in order to prevent some weird behavior.
     """
+    from dvc._debug import add_debugging_flags
+
     parent_parser = argparse.ArgumentParser(add_help=False)
     add_debugging_flags(parent_parser)
     log_level_group = parent_parser.add_mutually_exclusive_group()
@@ -207,23 +206,11 @@ def get_main_parser():
         help="Use `dvc COMMAND --help` for command-specific help.",
     )
 
+    from .utils import fix_subparsers
+
     fix_subparsers(subparsers)
 
     for cmd in COMMANDS:
         cmd.add_parser(subparsers, parent_parser)
 
     return parser
-
-
-def parse_args(argv=None):
-    """Parses CLI arguments.
-
-    Args:
-        argv: optional list of arguments to parse. sys.argv is used by default.
-
-    Raises:
-        dvc.exceptions.DvcParserError: raised for argument parsing errors.
-    """
-    parser = get_main_parser()
-    args = parser.parse_args(argv)
-    return args

dvc/cli/utils.py

@@ -0,0 +1,31 @@
+def fix_subparsers(subparsers):
+    """Workaround for bug in Python 3. See more info at:
+    https://bugs.python.org/issue16308
+    https://github.com/iterative/dvc/issues/769
+
+    Args:
+        subparsers: subparsers to fix.
+    """
+    subparsers.required = True
+    subparsers.dest = "cmd"
+
+
+def append_doc_link(help_message, path):
+    from dvc.utils import format_link
+
+    if not path:
+        return help_message
+    doc_base = "https://man.dvc.org/"
+    return "{message}\nDocumentation: {link}".format(
+        message=help_message, link=format_link(doc_base + path)
+    )
+
+
+def fix_plumbing_subparsers(subparsers):
+    # metavar needs to be explicitly set in order to hide plumbing subcommands
+    # from the 'positional arguments' choices list
+    # see: https://bugs.python.org/issue22848
+    cmds = [
+        cmd for cmd, parser in subparsers.choices.items() if parser.add_help
+    ]
+    subparsers.metavar = "{{{}}}".format(",".join(cmds))

dvc/command/live.py

@@ -1,93 +0,0 @@
-import argparse
-from pathlib import Path
-
-from dvc.command import completion
-from dvc.command.base import CmdBase, fix_subparsers
-from dvc.ui import ui
-
-
-class CmdLive(CmdBase):
-    UNINITIALIZED = True
-
-    def _run(self, target, revs=None):
-        metrics, plots = self.repo.live.show(target=target, revs=revs)
-
-        if plots:
-            html_path = Path.cwd() / (self.args.target + "_html")
-            from dvc.render.utils import render
-
-            index_path = render(self.repo, plots, metrics, html_path)
-            ui.write(index_path.as_uri())
-            return 0
-        return 1
-
-
-class CmdLiveShow(CmdLive):
-    def run(self):
-        return self._run(self.args.target)
-
-
-class CmdLiveDiff(CmdLive):
-    def run(self):
-        return self._run(self.args.target, self.args.revs)
-
-
-def shared_parent_parser():
-    parent_parser = argparse.ArgumentParser(add_help=False)
-    parent_parser.add_argument(
-        "target", help="Logs dir to produce summary from"
-    ).complete = completion.DIR
-    parent_parser.add_argument(
-        "-o",
-        "--out",
-        default=None,
-        help="Destination path to save plots to",
-        metavar="<path>",
-    ).complete = completion.DIR
-    return parent_parser
-
-
-def add_parser(subparsers, parent_parser):
-    LIVE_DESCRIPTION = (
-        "Commands to visualize and compare dvclive-produced logs."
-    )
-    live_parser = subparsers.add_parser(
-        "live",
-        parents=[parent_parser],
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-        description=LIVE_DESCRIPTION,
-    )
-    live_subparsers = live_parser.add_subparsers(
-        dest="cmd",
-        help="Use `dvc live CMD --help` to display command-specific help.",
-    )
-
-    fix_subparsers(live_subparsers)
-
-    SHOW_HELP = "Visualize dvclive directory content."
-    live_show_parser = live_subparsers.add_parser(
-        "show",
-        parents=[parent_parser, shared_parent_parser()],
-        help=SHOW_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    live_show_parser.set_defaults(func=CmdLiveShow)
-
-    DIFF_HELP = (
-        "Show multiple versions of dvclive data, "
-        "by plotting it in single view."
-    )
-    live_diff_parser = live_subparsers.add_parser(
-        "diff",
-        parents=[parent_parser, shared_parent_parser()],
-        help=DIFF_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    live_diff_parser.add_argument(
-        "--revs",
-        nargs="*",
-        default=None,
-        help="Git revision (e.g. SHA, branch, tag)",
-        metavar="<commit>",
-    )
-    live_diff_parser.set_defaults(func=CmdLiveDiff)

dvc/command/plots.py

@@ -1,265 +0,0 @@
-import argparse
-import logging
-
-from dvc.command import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
-from dvc.exceptions import DvcException
-from dvc.render.utils import find_vega, render
-from dvc.ui import ui
-from dvc.utils import format_link
-
-logger = logging.getLogger(__name__)
-
-
-class CmdPlots(CmdBase):
-    def _func(self, *args, **kwargs):
-        raise NotImplementedError
-
-    def _props(self):
-        from dvc.schema import PLOT_PROPS
-
-        # Pass only props specified by user, to not shadow ones from plot def
-        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 run(self):
-        from pathlib import Path
-
-        if self.args.show_vega:
-            if not self.args.targets:
-                logger.error("please specify a target for `--show-vega`")
-                return 1
-            if len(self.args.targets) > 1:
-                logger.error(
-                    "you can only specify one target for `--show-vega`"
-                )
-                return 1
-
-        try:
-            plots_data = self._func(
-                targets=self.args.targets, props=self._props()
-            )
-
-            if not plots_data:
-                ui.error_write(
-                    "No plots were loaded, "
-                    "visualization file will not be created."
-                )
-
-            if self.args.show_vega:
-                target = self.args.targets[0]
-                plot_json = find_vega(self.repo, plots_data, target)
-                if plot_json:
-                    ui.write(plot_json)
-                return 0
-
-            rel: str = self.args.out or "dvc_plots"
-            path: Path = (Path.cwd() / rel).resolve()
-            index_path = render(
-                self.repo,
-                plots_data,
-                path=path,
-                html_template_path=self.args.html_template,
-            )
-
-            ui.write(index_path.as_uri())
-
-            if self.args.open:
-                return ui.open_browser(index_path)
-
-            return 0
-
-        except DvcException:
-            logger.exception("")
-            return 1
-
-
-class CmdPlotsShow(CmdPlots):
-    UNINITIALIZED = True
-
-    def _func(self, *args, **kwargs):
-        return self.repo.plots.show(*args, **kwargs)
-
-
-class CmdPlotsDiff(CmdPlots):
-    UNINITIALIZED = True
-
-    def _func(self, *args, **kwargs):
-        return self.repo.plots.diff(
-            *args,
-            revs=self.args.revisions,
-            experiment=self.args.experiment,
-            **kwargs,
-        )
-
-
-class CmdPlotsModify(CmdPlots):
-    def run(self):
-        self.repo.plots.modify(
-            self.args.target, props=self._props(), unset=self.args.unset
-        )
-        return 0
-
-
-def add_parser(subparsers, parent_parser):
-    PLOTS_HELP = (
-        "Commands to visualize and compare plot metrics in structured files "
-        "(JSON, YAML, CSV, TSV)."
-    )
-
-    plots_parser = subparsers.add_parser(
-        "plots",
-        parents=[parent_parser],
-        description=append_doc_link(PLOTS_HELP, "plots"),
-        help=PLOTS_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    plots_subparsers = plots_parser.add_subparsers(
-        dest="cmd",
-        help="Use `dvc plots CMD --help` to display command-specific help.",
-    )
-
-    fix_subparsers(plots_subparsers)
-
-    SHOW_HELP = "Generate plots from metrics files."
-    plots_show_parser = plots_subparsers.add_parser(
-        "show",
-        parents=[parent_parser],
-        description=append_doc_link(SHOW_HELP, "plots/show"),
-        help=SHOW_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    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.",
-    ).complete = completion.FILE
-    _add_props_arguments(plots_show_parser)
-    _add_output_arguments(plots_show_parser)
-    plots_show_parser.set_defaults(func=CmdPlotsShow)
-
-    PLOTS_DIFF_HELP = (
-        "Show multiple versions of plot metrics "
-        "by plotting them in a single image."
-    )
-    plots_diff_parser = plots_subparsers.add_parser(
-        "diff",
-        parents=[parent_parser],
-        description=append_doc_link(PLOTS_DIFF_HELP, "plots/diff"),
-        help=PLOTS_DIFF_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    plots_diff_parser.add_argument(
-        "--targets",
-        nargs="*",
-        help=(
-            "Specific plots file(s) to visualize "
-            "(even if not found as `plots` in `dvc.yaml`). "
-            "Shows all tracked plots by default."
-        ),
-        metavar="<paths>",
-    ).complete = completion.FILE
-    plots_diff_parser.add_argument(
-        "-e",
-        "--experiment",
-        action="store_true",
-        default=False,
-        help=argparse.SUPPRESS,
-    )
-    plots_diff_parser.add_argument(
-        "revisions", nargs="*", default=None, help="Git commits to plot from"
-    )
-    _add_props_arguments(plots_diff_parser)
-    _add_output_arguments(plots_diff_parser)
-    plots_diff_parser.set_defaults(func=CmdPlotsDiff)
-
-    PLOTS_MODIFY_HELP = (
-        "Modify display properties of data-series plots "
-        "(has no effect on image-type plots)."
-    )
-    plots_modify_parser = plots_subparsers.add_parser(
-        "modify",
-        parents=[parent_parser],
-        description=append_doc_link(PLOTS_MODIFY_HELP, "plots/modify"),
-        help=PLOTS_MODIFY_HELP,
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-    )
-    plots_modify_parser.add_argument(
-        "target", help="Metric file to set properties to"
-    ).complete = completion.FILE
-    _add_props_arguments(plots_modify_parser)
-    plots_modify_parser.add_argument(
-        "--unset",
-        nargs="*",
-        metavar="<property>",
-        help="Unset one or more display properties.",
-    )
-    plots_modify_parser.set_defaults(func=CmdPlotsModify)
-
-
-def _add_props_arguments(parser):
-    parser.add_argument(
-        "-t",
-        "--template",
-        nargs="?",
-        default=None,
-        help=(
-            "Special JSON or HTML schema file to inject with the data. "
-            "See {}".format(
-                format_link("https://man.dvc.org/plots#plot-templates")
-            )
-        ),
-        metavar="<path>",
-    ).complete = completion.FILE
-    parser.add_argument(
-        "-x", default=None, help="Field name for X axis.", metavar="<field>"
-    )
-    parser.add_argument(
-        "-y", default=None, help="Field name for Y axis.", metavar="<field>"
-    )
-    parser.add_argument(
-        "--no-header",
-        action="store_false",
-        dest="header",
-        default=None,  # Use default None to distinguish when it's not used
-        help="Provided CSV or TSV datafile does not have a header.",
-    )
-    parser.add_argument(
-        "--title", default=None, metavar="<text>", help="Plot title."
-    )
-    parser.add_argument(
-        "--x-label", default=None, help="X axis label", metavar="<text>"
-    )
-    parser.add_argument(
-        "--y-label", default=None, help="Y axis label", metavar="<text>"
-    )
-
-
-def _add_output_arguments(parser):
-    parser.add_argument(
-        "-o",
-        "--out",
-        default=None,
-        help="Destination path to save plots to",
-        metavar="<path>",
-    ).complete = completion.DIR
-    parser.add_argument(
-        "--show-vega",
-        action="store_true",
-        default=False,
-        help="Show output in Vega format.",
-    )
-    parser.add_argument(
-        "--open",
-        action="store_true",
-        default=False,
-        help="Open plot file directly in the browser.",
-    )
-    parser.add_argument(
-        "--html-template",
-        default=None,
-        help="Custom HTML template for VEGA visualization.",
-        metavar="<path>",
-    )

dvc/commands/add.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 
@@ -101,9 +102,10 @@ def add_parser(subparsers, parent_parser):
         "--jobs",
         type=int,
         help=(
-            "Number of jobs to run simultaneously. "
+            "Only used along with '--to-remote'. "
+            "Number of jobs to run simultaneously "
+            "when pushing data to remote."
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/cache.py

@@ -1,8 +1,8 @@
 import argparse
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import append_doc_link, fix_subparsers
+from dvc.cli.utils import append_doc_link, fix_subparsers
-from dvc.command.config import CmdConfig
+from dvc.commands.config import CmdConfig
 from dvc.ui import ui
 
 
@@ -28,7 +28,7 @@ class CmdCacheDir(CmdConfig):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.config import parent_config_parser
+    from dvc.commands.config import parent_config_parser
 
     CACHE_HELP = "Manage cache settings."
 

dvc/commands/check_ignore.py

@@ -1,7 +1,8 @@
 import argparse
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 

dvc/commands/checkout.py

@@ -1,8 +1,9 @@
 import argparse
 import operator
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import CheckoutError
 from dvc.ui import ui
 
@@ -54,6 +55,10 @@ class CmdCheckout(CmdBase):
 
         if exc:
             raise exc
+
+        if self.args.relink:
+            msg = "Relinked successfully"
+            ui.write(msg)
         return 0
 
 

dvc/commands/commit.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 

dvc/commands/completion.py

@@ -0,0 +1,40 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.completion import PREAMBLE
+from dvc.cli.utils import append_doc_link
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdCompletion(CmdBaseNoRepo):
+    def run(self):
+        import shtab
+
+        shell = self.args.shell
+        parser = self.args.parser
+        script = shtab.complete(parser, shell=shell, preamble=PREAMBLE)
+        ui.write(script, force=True)
+        return 0
+
+
+def add_parser(subparsers, parent_parser):
+    COMPLETION_HELP = "Generate shell tab completion."
+    COMPLETION_DESCRIPTION = "Prints out shell tab completion scripts."
+    completion_parser = subparsers.add_parser(
+        "completion",
+        parents=[parent_parser],
+        description=append_doc_link(COMPLETION_DESCRIPTION, "completion"),
+        help=COMPLETION_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    completion_parser.add_argument(
+        "-s",
+        "--shell",
+        help="Shell syntax for completions.",
+        default="bash",
+        choices=["bash", "zsh"],
+    )
+    completion_parser.set_defaults(func=CmdCompletion)

dvc/commands/config.py

@@ -2,7 +2,8 @@ import argparse
 import logging
 import os
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)

dvc/commands/daemon.py

@@ -1,5 +1,6 @@
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBaseNoRepo, fix_subparsers
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import fix_subparsers
 
 
 class CmdDaemonBase(CmdBaseNoRepo):

dvc/commands/dag.py

@@ -1,10 +1,15 @@
 import argparse
+from typing import TYPE_CHECKING
 
-from dvc.command.base import CmdBase, append_doc_link
+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
 
@@ -17,16 +22,54 @@ 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()
-    write_dot(G, dot_file)
+
+    nx.relabel_nodes(G, _quote_label, copy=False)
+    write_dot(G.reverse(), dot_file)
     return dot_file.getvalue()
 
 
+def _show_mermaid(G, markdown: bool = False):
+    from dvc.repo.graph import get_pipelines
+
+    pipelines = get_pipelines(G)
+
+    graph = "flowchart TD"
+
+    total_nodes = 0
+    for pipeline in pipelines:
+        node_ids = {}
+        nodes = sorted(str(x) for x in pipeline.nodes)
+        for node in nodes:
+            total_nodes += 1
+            node_id = f"node{total_nodes}"
+            graph += f'\n\t{node_id}["{node}"]'
+            node_ids[node] = node_id
+        edges = sorted((str(a), str(b)) for b, a in pipeline.edges)
+        for a, b in edges:
+            graph += f"\n\t{node_ids[str(a)]}-->{node_ids[str(b)]}"
+
+    if markdown:
+        return f"```mermaid\n{graph}\n```"
+
+    return graph
+
+
 def _collect_targets(repo, target, outs):
     if not target:
         return []
@@ -43,9 +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
@@ -104,6 +147,8 @@ class CmdDAG(CmdBase):
 
         if self.args.dot:
             ui.write(_show_dot(G))
+        elif self.args.mermaid or self.args.markdown:
+            ui.write(_show_mermaid(G, self.args.markdown))
         else:
             with ui.pager():
                 ui.write(_show_ascii(G))
@@ -126,6 +171,20 @@ def add_parser(subparsers, parent_parser):
         default=False,
         help="Print DAG with .dot format.",
     )
+    dag_parser.add_argument(
+        "--mermaid",
+        action="store_true",
+        default=False,
+        help="Print DAG with mermaid format.",
+    )
+    dag_parser.add_argument(
+        "--md",
+        "--show-md",
+        action="store_true",
+        default=False,
+        dest="markdown",
+        help="Print DAG with mermaid format wrapped in Markdown block.",
+    )
     dag_parser.add_argument(
         "--full",
         action="store_true",

dvc/commands/data_sync.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 
@@ -18,7 +19,7 @@ class CmdDataBase(CmdBase):
 
 class CmdDataPull(CmdDataBase):
     def log_summary(self, stats):
-        from dvc.command.checkout import log_changes
+        from dvc.commands.checkout import log_changes
 
         log_changes(stats)
         super().log_summary(stats)
@@ -97,7 +98,7 @@ class CmdDataFetch(CmdDataBase):
 
 
 def shared_parent_parser():
-    from dvc.cli import get_parent_parser
+    from dvc.cli.parser import get_parent_parser
 
     # Parent parser used in pull/push/status
     parent_parser = argparse.ArgumentParser(
@@ -110,7 +111,6 @@ def shared_parent_parser():
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )
@@ -127,7 +127,7 @@ def shared_parent_parser():
 
 
 def add_parser(subparsers, _parent_parser):
-    from dvc.command.status import CmdDataStatus
+    from dvc.commands.status import CmdDataStatus
 
     # Pull
     PULL_HELP = "Download tracked files or directories from remote storage."

dvc/commands/destroy.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 
 logger = logging.getLogger(__name__)
 

dvc/commands/diff.py

@@ -2,8 +2,9 @@ import argparse
 import logging
 import os
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
@@ -127,7 +128,7 @@ class CmdDiff(CmdBase):
             show_hash = self.args.show_hash
             hide_missing = self.args.b_rev or self.args.hide_missing
             if hide_missing:
-                del diff["not in cache"]
+                diff.pop("not in cache", None)
 
             for key, entries in diff.items():
                 entries = sorted(

dvc/commands/experiments/__init__.py

@@ -0,0 +1,101 @@
+import argparse
+
+from dvc.cli.utils import (
+    append_doc_link,
+    fix_plumbing_subparsers,
+    fix_subparsers,
+)
+from dvc.commands.experiments import (
+    apply,
+    branch,
+    diff,
+    exec_run,
+    gc,
+    init,
+    ls,
+    pull,
+    push,
+    remove,
+    run,
+    show,
+)
+
+SUB_COMMANDS = [
+    apply,
+    branch,
+    diff,
+    exec_run,
+    gc,
+    init,
+    ls,
+    pull,
+    push,
+    remove,
+    run,
+    show,
+]
+
+
+def add_parser(subparsers, parent_parser):
+    EXPERIMENTS_HELP = "Commands to run and compare experiments."
+
+    experiments_parser = subparsers.add_parser(
+        "experiments",
+        parents=[parent_parser],
+        aliases=["exp"],
+        description=append_doc_link(EXPERIMENTS_HELP, "exp"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        help=EXPERIMENTS_HELP,
+    )
+
+    experiments_subparsers = experiments_parser.add_subparsers(
+        dest="cmd",
+        help="Use `dvc experiments CMD --help` to display "
+        "command-specific help.",
+    )
+
+    fix_subparsers(experiments_subparsers)
+    for cmd in SUB_COMMANDS:
+        cmd.add_parser(experiments_subparsers, parent_parser)
+    fix_plumbing_subparsers(experiments_subparsers)
+
+
+def add_rev_selection_flags(
+    experiments_subcmd_parser, command: str, default: bool = True
+):
+    experiments_subcmd_parser.add_argument(
+        "-A",
+        "--all-commits",
+        action="store_true",
+        default=False,
+        help=(
+            f"{command} all experiments in the repository "
+            "(overrides `--rev` and `--num`)."
+        ),
+    )
+    default_msg = " (HEAD by default)" if default else ""
+    msg = (
+        f"{command} experiments derived from the specified `<commit>` as "
+        f"baseline{default_msg}."
+    )
+    experiments_subcmd_parser.add_argument(
+        "--rev",
+        type=str,
+        default=None,
+        help=msg,
+        metavar="<commit>",
+    )
+    experiments_subcmd_parser.add_argument(
+        "-n",
+        "--num",
+        type=int,
+        default=1,
+        dest="num",
+        metavar="<num>",
+        help=(
+            f"{command} experiments from the last `num` commits "
+            "(first parents) starting from the `--rev` baseline. "
+            "Give a negative value to include all first-parent commits "
+            "(similar to `git log -n`)."
+        ),
+    )

dvc/commands/experiments/apply.py

@@ -0,0 +1,42 @@
+import argparse
+import logging
+
+from dvc.cli import completion
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsApply(CmdBase):
+    def run(self):
+
+        self.repo.experiments.apply(
+            self.args.experiment, force=self.args.force
+        )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_APPLY_HELP = (
+        "Apply the changes from an experiment to your workspace."
+    )
+    experiments_apply_parser = experiments_subparsers.add_parser(
+        "apply",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_APPLY_HELP, "exp/apply"),
+        help=EXPERIMENTS_APPLY_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_apply_parser.add_argument(
+        "--no-force",
+        action="store_false",
+        dest="force",
+        help="Fail if this command would overwrite conflicting changes.",
+    )
+    experiments_apply_parser.add_argument(
+        "experiment", help="Experiment to be applied."
+    ).complete = completion.EXPERIMENT
+    experiments_apply_parser.set_defaults(func=CmdExperimentsApply)

dvc/commands/experiments/branch.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 CmdExperimentsBranch(CmdBase):
+    def run(self):
+
+        self.repo.experiments.branch(self.args.experiment, self.args.branch)
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_BRANCH_HELP = "Promote an experiment to a Git branch."
+    experiments_branch_parser = experiments_subparsers.add_parser(
+        "branch",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_BRANCH_HELP, "exp/branch"),
+        help=EXPERIMENTS_BRANCH_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_branch_parser.add_argument(
+        "experiment", help="Experiment to be promoted."
+    )
+    experiments_branch_parser.add_argument(
+        "branch", help="Git branch name to use."
+    )
+    experiments_branch_parser.set_defaults(func=CmdExperimentsBranch)

dvc/commands/experiments/diff.py

@@ -0,0 +1,118 @@
+import argparse
+import logging
+
+from dvc.cli import completion
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands.metrics import DEFAULT_PRECISION
+from dvc.exceptions import DvcException
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsDiff(CmdBase):
+    def run(self):
+
+        try:
+            diff = self.repo.experiments.diff(
+                a_rev=self.args.a_rev,
+                b_rev=self.args.b_rev,
+                all=self.args.all,
+                param_deps=self.args.param_deps,
+            )
+        except DvcException:
+            logger.exception("failed to show experiments diff")
+            return 1
+
+        if self.args.json:
+            ui.write_json(diff)
+        else:
+            from dvc.compare import show_diff
+
+            precision = self.args.precision or DEFAULT_PRECISION
+            diffs = [("metrics", "Metric"), ("params", "Param")]
+            for idx, (key, title) in enumerate(diffs):
+                if idx:
+                    # we are printing tables even in `--quiet` mode
+                    # so we should also be printing the "table" separator
+                    ui.write(force=True)
+
+                show_diff(
+                    diff[key],
+                    title=title,
+                    markdown=self.args.markdown,
+                    no_path=self.args.no_path,
+                    on_empty_diff="diff not supported",
+                    precision=precision if key == "metrics" else None,
+                    a_rev=self.args.a_rev,
+                    b_rev=self.args.b_rev,
+                )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_DIFF_HELP = (
+        "Show changes between experiments in the DVC repository."
+    )
+
+    experiments_diff_parser = experiments_subparsers.add_parser(
+        "diff",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_DIFF_HELP, "exp/diff"),
+        help=EXPERIMENTS_DIFF_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_diff_parser.add_argument(
+        "a_rev", nargs="?", help="Old experiment to compare (defaults to HEAD)"
+    ).complete = completion.EXPERIMENT
+    experiments_diff_parser.add_argument(
+        "b_rev",
+        nargs="?",
+        help="New experiment to compare (defaults to the current workspace)",
+    ).complete = completion.EXPERIMENT
+    experiments_diff_parser.add_argument(
+        "--all",
+        action="store_true",
+        default=False,
+        help="Show unchanged metrics/params as well.",
+    )
+    experiments_diff_parser.add_argument(
+        "--param-deps",
+        action="store_true",
+        default=False,
+        help="Show only params that are stage dependencies.",
+    )
+    experiments_diff_parser.add_argument(
+        "--json",
+        "--show-json",
+        action="store_true",
+        default=False,
+        help="Show output in JSON format.",
+    )
+    experiments_diff_parser.add_argument(
+        "--md",
+        "--show-md",
+        action="store_true",
+        default=False,
+        dest="markdown",
+        help="Show tabulated output in the Markdown format (GFM).",
+    )
+    experiments_diff_parser.add_argument(
+        "--no-path",
+        action="store_true",
+        default=False,
+        help="Don't show metric/param path.",
+    )
+    experiments_diff_parser.add_argument(
+        "--precision",
+        type=int,
+        help=(
+            "Round metrics/params to `n` digits precision after the decimal "
+            f"point. Rounds to {DEFAULT_PRECISION} digits by default."
+        ),
+        metavar="<n>",
+    )
+    experiments_diff_parser.set_defaults(func=CmdExperimentsDiff)

dvc/commands/experiments/exec_run.py

@@ -0,0 +1,42 @@
+import logging
+
+from dvc.cli.command import CmdBaseNoRepo
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExecutorRun(CmdBaseNoRepo):
+    """Run an experiment executor."""
+
+    def run(self):
+        from dvc.repo.experiments.executor.base import (
+            BaseExecutor,
+            ExecutorInfo,
+        )
+        from dvc.utils.serialize import load_json
+
+        info = ExecutorInfo.from_dict(load_json(self.args.infofile))
+        BaseExecutor.reproduce(
+            info=info,
+            rev="",
+            queue=None,
+            log_level=logger.getEffectiveLevel(),
+            infofile=self.args.infofile,
+        )
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    EXEC_RUN_HELP = "Run an experiment executor."
+    exec_run_parser = experiments_subparsers.add_parser(
+        "exec-run",
+        parents=[parent_parser],
+        description=EXEC_RUN_HELP,
+        add_help=False,
+    )
+    exec_run_parser.add_argument(
+        "--infofile",
+        help="Path to executor info file",
+        default=None,
+    )
+    exec_run_parser.set_defaults(func=CmdExecutorRun)

dvc/commands/experiments/gc.py

@@ -0,0 +1,129 @@
+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__)
+
+
+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."
+        )
+
+
+class CmdExperimentsGC(CmdBase):
+    def run(self):
+        _raise_error_if_all_disabled(
+            all_branches=self.args.all_branches,
+            all_tags=self.args.all_tags,
+            all_commits=self.args.all_commits,
+            workspace=self.args.workspace,
+        )
+
+        msg = "This will remove all experiments except those derived from "
+
+        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"
+        msg += " of the current repo."
+        if self.args.queued:
+            msg += " Run queued experiments will be preserved."
+        else:
+            msg += " Run queued experiments will be removed."
+
+        logger.warning(msg)
+
+        msg = "Are you sure you want to proceed?"
+        if not self.args.force and not ui.confirm(msg):
+            return 1
+
+        removed = self.repo.experiments.gc(
+            all_branches=self.args.all_branches,
+            all_tags=self.args.all_tags,
+            all_commits=self.args.all_commits,
+            workspace=self.args.workspace,
+            queued=self.args.queued,
+        )
+
+        if removed:
+            ui.write(
+                f"Removed {removed} experiments.",
+                "To remove unused cache files",
+                "use 'dvc gc'.",
+            )
+        else:
+            ui.write("No experiments to remove.")
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_GC_HELP = "Garbage collect unneeded experiments."
+    EXPERIMENTS_GC_DESCRIPTION = (
+        "Removes all experiments which are not derived from the specified"
+        "Git revisions."
+    )
+    experiments_gc_parser = experiments_subparsers.add_parser(
+        "gc",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_GC_DESCRIPTION, "exp/gc"),
+        help=EXPERIMENTS_GC_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    experiments_gc_parser.add_argument(
+        "-w",
+        "--workspace",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from the current workspace.",
+    )
+    experiments_gc_parser.add_argument(
+        "-a",
+        "--all-branches",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from the tips of all Git branches.",
+    )
+    experiments_gc_parser.add_argument(
+        "-T",
+        "--all-tags",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from all Git tags.",
+    )
+    experiments_gc_parser.add_argument(
+        "-A",
+        "--all-commits",
+        action="store_true",
+        default=False,
+        help="Keep experiments derived from all Git commits.",
+    )
+    experiments_gc_parser.add_argument(
+        "--queued",
+        action="store_true",
+        default=False,
+        help=(
+            "Keep queued experiments (experiments run queue will be cleared "
+            "by default)."
+        ),
+    )
+    experiments_gc_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Force garbage collection - automatically agree to all prompts.",
+    )
+    experiments_gc_parser.set_defaults(func=CmdExperimentsGC)

dvc/commands/experiments/init.py

@@ -0,0 +1,210 @@
+import argparse
+import logging
+from typing import TYPE_CHECKING, List
+
+from funcy import compact
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.exceptions import InvalidArgumentError
+from dvc.ui import ui
+
+if TYPE_CHECKING:
+    from dvc.dependency import Dependency
+    from dvc.stage import PipelineStage
+
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsInit(CmdBase):
+    DEFAULT_NAME = "train"
+    CODE = "src"
+    DATA = "data"
+    MODELS = "models"
+    DEFAULT_METRICS = "metrics.json"
+    DEFAULT_PARAMS = "params.yaml"
+    PLOTS = "plots"
+    DEFAULTS = {
+        "code": CODE,
+        "data": DATA,
+        "models": MODELS,
+        "metrics": DEFAULT_METRICS,
+        "params": DEFAULT_PARAMS,
+        "plots": PLOTS,
+    }
+
+    def run(self):
+        from dvc.commands.stage import parse_cmd
+
+        cmd = parse_cmd(self.args.command)
+        if not self.args.interactive and not cmd:
+            raise InvalidArgumentError("command is not specified")
+
+        from dvc.repo.experiments.init import init
+
+        defaults = {}
+        if not self.args.explicit:
+            config = self.repo.config["exp"]
+            defaults.update({**self.DEFAULTS, **config})
+
+        cli_args = compact(
+            {
+                "cmd": cmd,
+                "code": self.args.code,
+                "data": self.args.data,
+                "models": self.args.models,
+                "metrics": self.args.metrics,
+                "params": self.args.params,
+                "plots": self.args.plots,
+                "live": self.args.live,
+            }
+        )
+
+        initialized_stage, initialized_deps, initialized_out_dirs = init(
+            self.repo,
+            name=self.args.name,
+            type=self.args.type,
+            defaults=defaults,
+            overrides=cli_args,
+            interactive=self.args.interactive,
+            force=self.args.force,
+        )
+        self._post_init_display(
+            initialized_stage, initialized_deps, initialized_out_dirs
+        )
+        if self.args.run:
+            self.repo.experiments.run(targets=[initialized_stage.addressing])
+        return 0
+
+    def _post_init_display(
+        self,
+        stage: "PipelineStage",
+        new_deps: List["Dependency"],
+        new_out_dirs: List[str],
+    ) -> None:
+        from dvc.utils import humanize
+
+        path_fmt = "[green]{}[/green]".format
+        if new_deps:
+            deps_paths = humanize.join(map(path_fmt, new_deps))
+            ui.write(f"Creating dependencies: {deps_paths}", styled=True)
+
+        if new_out_dirs:
+            out_dirs_paths = humanize.join(map(path_fmt, new_out_dirs))
+            ui.write(
+                f"Creating output directories: {out_dirs_paths}", styled=True
+            )
+
+        ui.write(
+            f"Creating [b]{self.args.name}[/b] stage in [green]dvc.yaml[/]",
+            styled=True,
+        )
+        if stage.outs or not self.args.run:
+            # separate the above status-like messages with help/tips section
+            ui.write(styled=True)
+
+        if stage.outs:
+            outs_paths = humanize.join(map(path_fmt, stage.outs))
+            tips = f"Ensure your experiment command creates {outs_paths}."
+            ui.write(tips, styled=True)
+
+        if not self.args.run:
+            ui.write(
+                'You can now run your experiment using [b]"dvc exp run"[/].',
+                styled=True,
+            )
+        else:
+            # separate between `exp.run` output and `dvc exp init` output
+            ui.write(styled=True)
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_INIT_HELP = "Quickly setup any project to use experiments."
+    experiments_init_parser = experiments_subparsers.add_parser(
+        "init",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_INIT_HELP, "exp/init"),
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        help=EXPERIMENTS_INIT_HELP,
+    )
+    experiments_init_parser.add_argument(
+        "command",
+        nargs=argparse.REMAINDER,
+        help="Command to execute.",
+        metavar="command",
+    )
+    experiments_init_parser.add_argument(
+        "--run",
+        action="store_true",
+        help="Run the experiment after initializing it",
+    )
+    experiments_init_parser.add_argument(
+        "--interactive",
+        "-i",
+        action="store_true",
+        help="Prompt for values that are not provided",
+    )
+    experiments_init_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Overwrite existing stage",
+    )
+    experiments_init_parser.add_argument(
+        "--explicit",
+        action="store_true",
+        default=False,
+        help="Only use the path values explicitly provided",
+    )
+    experiments_init_parser.add_argument(
+        "--name",
+        "-n",
+        help="Name of the stage to create (default: %(default)s)",
+        default=CmdExperimentsInit.DEFAULT_NAME,
+    )
+    experiments_init_parser.add_argument(
+        "--code",
+        help="Path to the source file or directory "
+        "which your experiments depend"
+        f" (default: {CmdExperimentsInit.CODE})",
+    )
+    experiments_init_parser.add_argument(
+        "--data",
+        help="Path to the data file or directory "
+        "which your experiments depend"
+        f" (default: {CmdExperimentsInit.DATA})",
+    )
+    experiments_init_parser.add_argument(
+        "--models",
+        help="Path to the model file or directory for your experiments"
+        f" (default: {CmdExperimentsInit.MODELS})",
+    )
+    experiments_init_parser.add_argument(
+        "--params",
+        help="Path to the parameters file for your experiments"
+        f" (default: {CmdExperimentsInit.DEFAULT_PARAMS})",
+    )
+    experiments_init_parser.add_argument(
+        "--metrics",
+        help="Path to the metrics file for your experiments"
+        f" (default: {CmdExperimentsInit.DEFAULT_METRICS})",
+    )
+    experiments_init_parser.add_argument(
+        "--plots",
+        help="Path to the plots file or directory for your experiments"
+        f" (default: {CmdExperimentsInit.PLOTS})",
+    )
+    experiments_init_parser.add_argument(
+        "--live",
+        help="Path to log dvclive outputs for your experiments",
+    )
+    experiments_init_parser.add_argument(
+        "--type",
+        choices=["default", "checkpoint"],
+        default="default",
+        help="Select type of stage to create (default: %(default)s)",
+    )
+    experiments_init_parser.set_defaults(func=CmdExperimentsInit)

dvc/commands/experiments/ls.py

@@ -0,0 +1,64 @@
+import argparse
+import logging
+
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsList(CmdBase):
+    def run(self):
+        name_only = self.args.name_only
+        exps = self.repo.experiments.ls(
+            all_commits=self.args.all_commits,
+            rev=self.args.rev,
+            num=self.args.num,
+            git_remote=self.args.git_remote,
+        )
+        for baseline in exps:
+            tag = self.repo.scm.describe(baseline)
+            if not tag:
+                branch = self.repo.scm.describe(baseline, base="refs/heads")
+                if branch:
+                    tag = branch.split("/")[-1]
+            name = tag if tag else baseline[:7]
+            if not name_only:
+                print(f"{name}:")
+            for exp_name in exps[baseline]:
+                indent = "" if name_only else "\t"
+                print(f"{indent}{exp_name}")
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    from . import add_rev_selection_flags
+
+    EXPERIMENTS_LIST_HELP = "List local and remote experiments."
+    experiments_list_parser = experiments_subparsers.add_parser(
+        "list",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_LIST_HELP, "exp/list"),
+        help=EXPERIMENTS_LIST_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    add_rev_selection_flags(experiments_list_parser, "List")
+    experiments_list_parser.add_argument(
+        "--name-only",
+        "--names-only",
+        action="store_true",
+        help="Only output experiment names (without parent commits).",
+    )
+    experiments_list_parser.add_argument(
+        "git_remote",
+        nargs="?",
+        default=None,
+        help=(
+            "Optional Git remote name or Git URL. "
+            "If provided, experiments from the specified Git repository "
+            " will be listed instead of local ones."
+        ),
+        metavar="<git_remote>",
+    )
+    experiments_list_parser.set_defaults(func=CmdExperimentsList)

dvc/commands/experiments/pull.py

@@ -0,0 +1,117 @@
+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 CmdExperimentsPull(CmdBase):
+    def raise_error_if_all_disabled(self):
+        if not any(
+            [self.args.experiment, self.args.all_commits, self.args.rev]
+        ):
+            raise InvalidArgumentError(
+                "Either provide an `experiment` argument, or use the "
+                "`--rev` or `--all-commits` flag."
+            )
+
+    def run(self):
+        self.raise_error_if_all_disabled()
+
+        pulled_exps = self.repo.experiments.pull(
+            self.args.git_remote,
+            self.args.experiment,
+            all_commits=self.args.all_commits,
+            rev=self.args.rev,
+            num=self.args.num,
+            force=self.args.force,
+            pull_cache=self.args.pull_cache,
+            dvc_remote=self.args.dvc_remote,
+            jobs=self.args.jobs,
+            run_cache=self.args.run_cache,
+        )
+
+        if pulled_exps:
+            ui.write(
+                f"Pulled experiment '{pulled_exps}'",
+                f"from Git remote '{self.args.git_remote}'.",
+            )
+        else:
+            ui.write("No experiments to pull.")
+        if not self.args.pull_cache:
+            ui.write(
+                "To pull cached outputs for this experiment"
+                "from DVC remote storage,"
+                "re-run this command without '--no-cache'."
+            )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    from . import add_rev_selection_flags
+
+    EXPERIMENTS_PULL_HELP = "Pull an experiment from a Git remote."
+    experiments_pull_parser = experiments_subparsers.add_parser(
+        "pull",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_PULL_HELP, "exp/pull"),
+        help=EXPERIMENTS_PULL_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    add_rev_selection_flags(experiments_pull_parser, "Pull", False)
+    experiments_pull_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        help="Replace local experiment already exists.",
+    )
+    experiments_pull_parser.add_argument(
+        "--no-cache",
+        action="store_false",
+        dest="pull_cache",
+        help=(
+            "Do not pull cached outputs for this experiment from DVC remote "
+            "storage."
+        ),
+    )
+    experiments_pull_parser.add_argument(
+        "-r",
+        "--remote",
+        dest="dvc_remote",
+        metavar="<name>",
+        help="Name of the DVC remote to use when pulling cached outputs.",
+    )
+    experiments_pull_parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        metavar="<number>",
+        help=(
+            "Number of jobs to run simultaneously when pulling from DVC "
+            "remote storage."
+        ),
+    )
+    experiments_pull_parser.add_argument(
+        "--run-cache",
+        action="store_true",
+        default=False,
+        help="Pull run history for all stages.",
+    )
+    experiments_pull_parser.add_argument(
+        "git_remote",
+        help="Git remote name or Git URL.",
+        metavar="<git_remote>",
+    )
+    experiments_pull_parser.add_argument(
+        "experiment",
+        nargs="*",
+        default=None,
+        help="Experiments to pull.",
+        metavar="<experiment>",
+    )
+    experiments_pull_parser.set_defaults(func=CmdExperimentsPull)

dvc/commands/experiments/push.py

@@ -0,0 +1,119 @@
+import argparse
+import logging
+
+from dvc.cli import completion
+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 CmdExperimentsPush(CmdBase):
+    def raise_error_if_all_disabled(self):
+        if not any(
+            [self.args.experiment, self.args.all_commits, self.args.rev]
+        ):
+            raise InvalidArgumentError(
+                "Either provide an `experiment` argument, or use the "
+                "`--rev` or `--all-commits` flag."
+            )
+
+    def run(self):
+
+        self.raise_error_if_all_disabled()
+
+        pushed_exps = self.repo.experiments.push(
+            self.args.git_remote,
+            self.args.experiment,
+            all_commits=self.args.all_commits,
+            rev=self.args.rev,
+            num=self.args.num,
+            force=self.args.force,
+            push_cache=self.args.push_cache,
+            dvc_remote=self.args.dvc_remote,
+            jobs=self.args.jobs,
+            run_cache=self.args.run_cache,
+        )
+
+        if pushed_exps:
+            ui.write(
+                f"Pushed experiment '{pushed_exps}'"
+                f"to Git remote '{self.args.git_remote}'."
+            )
+        else:
+            ui.write("No experiments to push.")
+        if not self.args.push_cache:
+            ui.write(
+                "To push cached outputs",
+                "for this experiment to DVC remote storage,"
+                "re-run this command without '--no-cache'.",
+            )
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    from . import add_rev_selection_flags
+
+    EXPERIMENTS_PUSH_HELP = "Push a local experiment to a Git remote."
+    experiments_push_parser = experiments_subparsers.add_parser(
+        "push",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_PUSH_HELP, "exp/push"),
+        help=EXPERIMENTS_PUSH_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    add_rev_selection_flags(experiments_push_parser, "Push", False)
+    experiments_push_parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        help="Replace experiment in the Git remote if it already exists.",
+    )
+    experiments_push_parser.add_argument(
+        "--no-cache",
+        action="store_false",
+        dest="push_cache",
+        help=(
+            "Do not push cached outputs for this experiment to DVC remote "
+            "storage."
+        ),
+    )
+    experiments_push_parser.add_argument(
+        "-r",
+        "--remote",
+        dest="dvc_remote",
+        metavar="<name>",
+        help="Name of the DVC remote to use when pushing cached outputs.",
+    )
+    experiments_push_parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        metavar="<number>",
+        help=(
+            "Number of jobs to run simultaneously when pushing to DVC remote "
+            "storage."
+        ),
+    )
+    experiments_push_parser.add_argument(
+        "--run-cache",
+        action="store_true",
+        default=False,
+        help="Push run history for all stages.",
+    )
+    experiments_push_parser.add_argument(
+        "git_remote",
+        help="Git remote name or Git URL.",
+        metavar="<git_remote>",
+    )
+    experiments_push_parser.add_argument(
+        "experiment",
+        nargs="*",
+        default=None,
+        help="Experiments to push.",
+        metavar="<experiment>",
+    ).complete = completion.EXPERIMENT
+    experiments_push_parser.set_defaults(func=CmdExperimentsPush)

dvc/commands/experiments/remove.py

@@ -0,0 +1,73 @@
+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 CmdExperimentsRemove(CmdBase):
+    def raise_error_if_all_disabled(self):
+        if not any(
+            [
+                self.args.experiment,
+                self.args.all_commits,
+                self.args.rev,
+                self.args.queue,
+            ]
+        ):
+            raise InvalidArgumentError(
+                "Either provide an `experiment` argument, or use the "
+                "`--rev` or `--all-commits` flag."
+            )
+
+    def run(self):
+
+        self.raise_error_if_all_disabled()
+
+        removed_list = self.repo.experiments.remove(
+            exp_names=self.args.experiment,
+            all_commits=self.args.all_commits,
+            rev=self.args.rev,
+            num=self.args.num,
+            queue=self.args.queue,
+            git_remote=self.args.git_remote,
+        )
+        removed = ",".join(removed_list)
+        ui.write(f"Removed experiments: {removed}")
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    from . import add_rev_selection_flags
+
+    EXPERIMENTS_REMOVE_HELP = "Remove experiments."
+    experiments_remove_parser = experiments_subparsers.add_parser(
+        "remove",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_REMOVE_HELP, "exp/remove"),
+        help=EXPERIMENTS_REMOVE_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    remove_group = experiments_remove_parser.add_mutually_exclusive_group()
+    add_rev_selection_flags(experiments_remove_parser, "Remove", False)
+    remove_group.add_argument(
+        "--queue", action="store_true", help="Remove all queued experiments."
+    )
+    remove_group.add_argument(
+        "-g",
+        "--git-remote",
+        metavar="<git_remote>",
+        help="Name or URL of the Git remote to remove the experiment from",
+    )
+    experiments_remove_parser.add_argument(
+        "experiment",
+        nargs="*",
+        help="Experiments to remove.",
+        metavar="<experiment>",
+    )
+    experiments_remove_parser.set_defaults(func=CmdExperimentsRemove)

dvc/commands/experiments/run.py

@@ -0,0 +1,142 @@
+import argparse
+import logging
+
+from dvc.cli import completion
+from dvc.cli.utils import append_doc_link
+from dvc.commands.repro import CmdRepro
+from dvc.commands.repro import add_arguments as add_repro_arguments
+from dvc.exceptions import InvalidArgumentError
+from dvc.ui import ui
+
+logger = logging.getLogger(__name__)
+
+
+class CmdExperimentsRun(CmdRepro):
+    def run(self):
+        from dvc.compare import show_metrics
+
+        if self.args.checkpoint_resume:
+            if self.args.reset:
+                raise InvalidArgumentError(
+                    "--reset and --rev are mutually exclusive."
+                )
+            if not (self.args.queue or self.args.tmp_dir):
+                raise InvalidArgumentError(
+                    "--rev can only be used in conjunction with "
+                    "--queue or --temp."
+                )
+
+        if self.args.reset:
+            ui.write("Any existing checkpoints will be reset and re-run.")
+
+        results = self.repo.experiments.run(
+            name=self.args.name,
+            queue=self.args.queue,
+            run_all=self.args.run_all,
+            jobs=self.args.jobs,
+            params=self.args.set_param,
+            checkpoint_resume=self.args.checkpoint_resume,
+            reset=self.args.reset,
+            tmp_dir=self.args.tmp_dir,
+            machine=self.args.machine,
+            **self._common_kwargs,
+        )
+
+        if self.args.metrics and results:
+            metrics = self.repo.metrics.show(revs=list(results))
+            metrics.pop("workspace", None)
+            show_metrics(metrics)
+
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+
+    EXPERIMENTS_RUN_HELP = "Run or resume an experiment."
+    experiments_run_parser = experiments_subparsers.add_parser(
+        "run",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_RUN_HELP, "exp/run"),
+        help=EXPERIMENTS_RUN_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    _add_run_common(experiments_run_parser)
+    experiments_run_parser.add_argument(
+        "-r",
+        "--rev",
+        type=str,
+        dest="checkpoint_resume",
+        help=(
+            "Continue the specified checkpoint experiment. Can only be used "
+            "in conjunction with --queue or --temp."
+        ),
+        metavar="<experiment_rev>",
+    ).complete = completion.EXPERIMENT
+    experiments_run_parser.add_argument(
+        "--reset",
+        action="store_true",
+        help="Reset existing checkpoints and restart the experiment.",
+    )
+    experiments_run_parser.set_defaults(func=CmdExperimentsRun)
+
+
+def _add_run_common(parser):
+    """Add common args for 'exp run'."""
+    # inherit arguments from `dvc repro`
+    add_repro_arguments(parser)
+    parser.add_argument(
+        "-n",
+        "--name",
+        default=None,
+        help=(
+            "Human-readable experiment name. If not specified, a name will "
+            "be auto-generated."
+        ),
+        metavar="<name>",
+    )
+    parser.add_argument(
+        "-S",
+        "--set-param",
+        action="append",
+        default=[],
+        help="Use the specified param value when reproducing pipelines.",
+        metavar="[<filename>:]<param_name>=<param_value>",
+    )
+    parser.add_argument(
+        "--queue",
+        action="store_true",
+        default=False,
+        help="Stage this experiment in the run queue for future execution.",
+    )
+    parser.add_argument(
+        "--run-all",
+        action="store_true",
+        default=False,
+        help="Execute all experiments in the run queue. Implies --temp.",
+    )
+    parser.add_argument(
+        "-j",
+        "--jobs",
+        type=int,
+        default=1,
+        help="Run the specified number of experiments at a time in parallel.",
+        metavar="<number>",
+    )
+    parser.add_argument(
+        "--temp",
+        action="store_true",
+        dest="tmp_dir",
+        help=(
+            "Run this experiment in a separate temporary directory instead of "
+            "your workspace."
+        ),
+    )
+    parser.add_argument(
+        "--machine",
+        default=None,
+        help=argparse.SUPPRESS,
+        # help=(
+        #     "Run this experiment on the specified 'dvc machine' instance."
+        # )
+        # metavar="<name>",
+    )

dvc/commands/experiments/show.py

@@ -0,0 +1,655 @@
+import argparse
+import logging
+import os
+import re
+from collections import Counter, OrderedDict, defaultdict
+from datetime import date, datetime
+from typing import TYPE_CHECKING
+
+from funcy import lmap
+
+from dvc.cli import completion
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
+from dvc.commands.metrics import DEFAULT_PRECISION
+from dvc.exceptions import DvcException, InvalidArgumentError
+from dvc.ui import ui
+from dvc.utils.flatten import flatten
+from dvc.utils.serialize import encode_exception
+
+if TYPE_CHECKING:
+    from dvc.compare import TabularData
+    from dvc.ui import RichText
+
+FILL_VALUE = "-"
+FILL_VALUE_ERRORED = "!"
+
+
+logger = logging.getLogger(__name__)
+
+
+def _update_names(names, items):
+    for name, item in items:
+        item = item.get("data", {})
+        if isinstance(item, dict):
+            item = flatten(item)
+            names[name].update({key: None for key in item})
+
+
+def _collect_names(all_experiments, **kwargs):
+    metric_names = defaultdict(dict)
+    param_names = defaultdict(dict)
+    deps_names = set()
+
+    for _, experiments in all_experiments.items():
+        for exp_data in experiments.values():
+            exp = exp_data.get("data", {})
+            _update_names(metric_names, exp.get("metrics", {}).items())
+            _update_names(param_names, exp.get("params", {}).items())
+            for dep_name in exp.get("deps", {}):
+                deps_names.add(dep_name)
+
+    return metric_names, param_names, sorted(deps_names)
+
+
+experiment_types = {
+    "checkpoint_tip": "│ ╓",
+    "checkpoint_commit": "│ ╟",
+    "checkpoint_base": "├─╨",
+    "branch_commit": "├──",
+    "branch_base": "└──",
+    "baseline": "",
+}
+
+
+def _collect_rows(
+    base_rev,
+    experiments,
+    all_headers,
+    metric_headers,
+    param_headers,
+    metric_names,
+    param_names,
+    deps_names,
+    precision=DEFAULT_PRECISION,
+    sort_by=None,
+    sort_order=None,
+    fill_value=FILL_VALUE,
+    iso=False,
+):
+    from scmrepo.git import Git
+
+    if sort_by:
+        sort_path, sort_name, sort_type = _sort_column(
+            sort_by, metric_names, param_names
+        )
+        reverse = sort_order == "desc"
+        experiments = _sort_exp(
+            experiments, sort_path, sort_name, sort_type, reverse
+        )
+
+    new_checkpoint = True
+    for i, (rev, results) in enumerate(experiments.items()):
+        fill_value = FILL_VALUE_ERRORED if results.get("error") else fill_value
+        row_dict = {k: fill_value for k in all_headers}
+
+        exp = results.get("data", {})
+
+        if exp.get("running"):
+            state = "Running"
+        elif exp.get("queued"):
+            state = "Queued"
+        else:
+            state = fill_value
+
+        is_baseline = rev == "baseline"
+
+        if is_baseline:
+            name_rev = base_rev[:7] if Git.is_sha(base_rev) else base_rev
+        else:
+            name_rev = rev[:7]
+
+        tip = exp.get("checkpoint_tip")
+        parent_rev = exp.get("checkpoint_parent", "")
+        parent_exp = experiments.get(parent_rev, {}).get("data", {})
+        parent_tip = parent_exp.get("checkpoint_tip")
+
+        parent = ""
+        if is_baseline:
+            typ = "baseline"
+        elif tip:
+            if tip == parent_tip:
+                typ = (
+                    "checkpoint_tip" if new_checkpoint else "checkpoint_commit"
+                )
+            elif parent_rev == base_rev:
+                typ = "checkpoint_base"
+            else:
+                typ = "checkpoint_commit"
+                parent = parent_rev[:7]
+        elif i < len(experiments) - 1:
+            typ = "branch_commit"
+        else:
+            typ = "branch_base"
+
+        if not is_baseline:
+            new_checkpoint = not (tip and tip == parent_tip)
+
+        row_dict["Experiment"] = exp.get("name", "")
+        row_dict["rev"] = name_rev
+        row_dict["typ"] = typ
+        row_dict["Created"] = _format_time(
+            exp.get("timestamp"), fill_value, iso
+        )
+        row_dict["parent"] = parent
+        row_dict["State"] = state
+        row_dict["Executor"] = exp.get("executor", fill_value)
+
+        _extend_row(
+            row_dict,
+            metric_names,
+            metric_headers,
+            exp.get("metrics", {}).items(),
+            precision,
+            fill_value=fill_value,
+        )
+        _extend_row(
+            row_dict,
+            param_names,
+            param_headers,
+            exp.get("params", {}).items(),
+            precision,
+            fill_value=fill_value,
+        )
+        for dep in deps_names:
+            hash_info = exp.get("deps", {}).get(dep, {}).get("hash")
+            if hash_info is not None:
+                hash_info = hash_info[:7]
+            row_dict[dep] = hash_info
+        yield list(row_dict.values())
+
+
+def _sort_column(sort_by, metric_names, param_names):
+    path, _, sort_name = sort_by.rpartition(":")
+    matches = set()
+
+    if path:
+        if path in metric_names and sort_name in metric_names[path]:
+            matches.add((path, sort_name, "metrics"))
+        if path in param_names and sort_name in param_names[path]:
+            matches.add((path, sort_name, "params"))
+    else:
+        for path in metric_names:
+            if sort_name in metric_names[path]:
+                matches.add((path, sort_name, "metrics"))
+        for path in param_names:
+            if sort_name in param_names[path]:
+                matches.add((path, sort_name, "params"))
+
+    if len(matches) == 1:
+        return matches.pop()
+    if len(matches) > 1:
+        raise InvalidArgumentError(
+            "Ambiguous sort column '{}' matched '{}'".format(
+                sort_by,
+                ", ".join([f"{path}:{name}" for path, name, _ in matches]),
+            )
+        )
+    raise InvalidArgumentError(f"Unknown sort column '{sort_by}'")
+
+
+def _sort_exp(experiments, sort_path, sort_name, typ, reverse):
+    def _sort(item):
+        rev, exp = item
+        exp_data = exp.get("data", {})
+        tip = exp_data.get("checkpoint_tip")
+        if tip and tip != rev:
+            # Sort checkpoint experiments by tip commit
+            return _sort((tip, experiments[tip]))
+        data = exp_data.get(typ, {}).get(sort_path, {}).get("data", {})
+        val = flatten(data).get(sort_name)
+        return val is None, val
+
+    ret = OrderedDict()
+    if "baseline" in experiments:
+        ret["baseline"] = experiments.pop("baseline")
+
+    ret.update(sorted(experiments.items(), key=_sort, reverse=reverse))
+    return ret
+
+
+def _format_time(datetime_obj, fill_value=FILL_VALUE, iso=False):
+    if datetime_obj is None:
+        return fill_value
+
+    if iso:
+        return datetime_obj.isoformat()
+
+    if datetime_obj.date() == date.today():
+        fmt = "%I:%M %p"
+    else:
+        fmt = "%b %d, %Y"
+    return datetime_obj.strftime(fmt)
+
+
+def _extend_row(row, names, headers, items, precision, fill_value=FILL_VALUE):
+    from dvc.compare import _format_field, with_value
+
+    for fname, data in items:
+        item = data.get("data", {})
+        item = flatten(item) if isinstance(item, dict) else {fname: item}
+        for name in names[fname]:
+            value = with_value(
+                item.get(name),
+                FILL_VALUE_ERRORED if data.get("error", None) else fill_value,
+            )
+            # wrap field data in ui.rich_text, otherwise rich may
+            # interpret unescaped braces from list/dict types as rich
+            # markup tags
+            value = ui.rich_text(str(_format_field(value, precision)))
+            if name in headers:
+                row[name] = value
+            else:
+                row[f"{fname}:{name}"] = value
+
+
+def experiments_table(
+    all_experiments,
+    headers,
+    metric_headers,
+    metric_names,
+    param_headers,
+    param_names,
+    deps_names,
+    sort_by=None,
+    sort_order=None,
+    precision=DEFAULT_PRECISION,
+    fill_value=FILL_VALUE,
+    iso=False,
+) -> "TabularData":
+    from funcy import lconcat
+
+    from dvc.compare import TabularData
+
+    all_headers = lconcat(headers, metric_headers, param_headers, deps_names)
+    td = TabularData(all_headers, fill_value=fill_value)
+    for base_rev, experiments in all_experiments.items():
+        rows = _collect_rows(
+            base_rev,
+            experiments,
+            all_headers,
+            metric_headers,
+            param_headers,
+            metric_names,
+            param_names,
+            deps_names,
+            sort_by=sort_by,
+            sort_order=sort_order,
+            precision=precision,
+            fill_value=fill_value,
+            iso=iso,
+        )
+        td.extend(rows)
+
+    return td
+
+
+def prepare_exp_id(kwargs) -> "RichText":
+    exp_name = kwargs["Experiment"]
+    rev = kwargs["rev"]
+    typ = kwargs.get("typ", "baseline")
+
+    if typ == "baseline" or not exp_name:
+        text = ui.rich_text(exp_name or rev)
+    else:
+        text = ui.rich_text.assemble(rev, " [", (exp_name, "bold"), "]")
+
+    parent = kwargs.get("parent")
+    suff = f" ({parent})" if parent else ""
+    text.append(suff)
+
+    tree = experiment_types[typ]
+    pref = f"{tree} " if tree else ""
+    return ui.rich_text(pref) + text
+
+
+def baseline_styler(typ):
+    return {"style": "bold"} if typ == "baseline" else {}
+
+
+def show_experiments(
+    all_experiments,
+    keep=None,
+    drop=None,
+    pager=True,
+    csv=False,
+    markdown=False,
+    pcp=False,
+    **kwargs,
+):
+    from funcy.seqs import flatten as flatten_list
+
+    metric_names, param_names, deps_names = _collect_names(all_experiments)
+
+    headers = [
+        "Experiment",
+        "rev",
+        "typ",
+        "Created",
+        "parent",
+        "State",
+        "Executor",
+    ]
+
+    names = {**metric_names, **param_names}
+    counter = Counter(flatten_list([list(a.keys()) for a in names.values()]))
+    counter.update(headers)
+    metric_headers = _normalize_headers(metric_names, counter)
+    param_headers = _normalize_headers(param_names, counter)
+
+    td = experiments_table(
+        all_experiments,
+        headers,
+        metric_headers,
+        metric_names,
+        param_headers,
+        param_names,
+        deps_names,
+        kwargs.get("sort_by"),
+        kwargs.get("sort_order"),
+        kwargs.get("precision"),
+        kwargs.get("fill_value"),
+        kwargs.get("iso"),
+    )
+    if keep:
+        for col in td.keys():
+            if re.match(keep, col):
+                td.protect(col)
+
+    for col in ("State", "Executor"):
+        if td.is_empty(col):
+            td.drop(col)
+
+    row_styles = lmap(baseline_styler, td.column("typ"))
+
+    if not csv:
+        merge_headers = ["Experiment", "rev", "typ", "parent"]
+        td.column("Experiment")[:] = map(
+            prepare_exp_id, td.as_dict(merge_headers)
+        )
+        td.drop(*merge_headers[1:])
+
+    headers = {
+        "metrics": metric_headers,
+        "params": param_headers,
+        "deps": deps_names,
+    }
+    styles = {
+        "Experiment": {"no_wrap": True, "header_style": "black on grey93"},
+        "Created": {"header_style": "black on grey93"},
+        "State": {"header_style": "black on grey93"},
+        "Executor": {"header_style": "black on grey93"},
+    }
+    header_bg_colors = {
+        "metrics": "cornsilk1",
+        "params": "light_cyan1",
+        "deps": "plum2",
+    }
+    styles.update(
+        {
+            header: {
+                "justify": "right" if typ == "metrics" else "left",
+                "header_style": f"black on {header_bg_colors[typ]}",
+                "collapse": idx != 0,
+                "no_wrap": typ == "metrics",
+            }
+            for typ, hs in headers.items()
+            for idx, header in enumerate(hs)
+        }
+    )
+
+    if kwargs.get("only_changed", False) or pcp:
+        td.drop_duplicates("cols", ignore_empty=False)
+
+    cols_to_drop = set()
+    if drop is not None:
+        cols_to_drop = {col for col in td.keys() if re.match(drop, col)}
+    if pcp:
+        cols_to_drop.add("Created")
+    td.drop(*cols_to_drop)
+
+    if pcp:
+        subset = {x for x in td.keys() if x != "Experiment"}
+        td.dropna(
+            "rows",
+            how="all",
+            subset=subset,
+        )
+        td.drop_duplicates("rows", subset=subset)
+        if "Experiment" in td:
+            td.column("Experiment")[:] = [
+                # remove tree characters
+                str(x).encode("ascii", "ignore").strip().decode()
+                for x in td.column("Experiment")
+            ]
+        out = kwargs.get("out") or "dvc_plots"
+        output_file = os.path.join(out, "index.html")
+        ui.write(
+            td.to_parallel_coordinates(
+                output_path=os.path.abspath(output_file),
+                color_by=kwargs.get("sort_by") or "Experiment",
+            )
+        )
+        if kwargs.get("open"):
+            return ui.open_browser(output_file)
+
+    else:
+        td.render(
+            pager=pager,
+            borders="horizontals",
+            rich_table=True,
+            header_styles=styles,
+            row_styles=row_styles,
+            csv=csv,
+            markdown=markdown,
+        )
+
+
+def _normalize_headers(names, count):
+    return [
+        name if count[name] == 1 else f"{path}:{name}"
+        for path in names
+        for name in names[path]
+    ]
+
+
+def _format_json(item):
+    if isinstance(item, (date, datetime)):
+        return item.isoformat()
+    return encode_exception(item)
+
+
+class CmdExperimentsShow(CmdBase):
+    def run(self):
+        try:
+            all_experiments = self.repo.experiments.show(
+                all_branches=self.args.all_branches,
+                all_tags=self.args.all_tags,
+                all_commits=self.args.all_commits,
+                revs=self.args.rev,
+                num=self.args.num,
+                sha_only=self.args.sha,
+                param_deps=self.args.param_deps,
+                fetch_running=self.args.fetch_running,
+            )
+        except DvcException:
+            logger.exception("failed to show experiments")
+            return 1
+
+        if self.args.json:
+            ui.write_json(all_experiments, default=_format_json)
+        else:
+            precision = (
+                self.args.precision or None
+                if self.args.csv
+                else DEFAULT_PRECISION
+            )
+            fill_value = "" if self.args.csv else FILL_VALUE
+            iso = True if self.args.csv else False
+
+            show_experiments(
+                all_experiments,
+                keep=self.args.keep,
+                drop=self.args.drop,
+                sort_by=self.args.sort_by,
+                sort_order=self.args.sort_order,
+                precision=precision,
+                fill_value=fill_value,
+                iso=iso,
+                pager=not self.args.no_pager,
+                csv=self.args.csv,
+                markdown=self.args.markdown,
+                only_changed=self.args.only_changed,
+                pcp=self.args.pcp,
+                out=self.args.out,
+                open=self.args.open,
+            )
+        return 0
+
+
+def add_parser(experiments_subparsers, parent_parser):
+    from . import add_rev_selection_flags
+
+    EXPERIMENTS_SHOW_HELP = "Print experiments."
+    experiments_show_parser = experiments_subparsers.add_parser(
+        "show",
+        parents=[parent_parser],
+        description=append_doc_link(EXPERIMENTS_SHOW_HELP, "exp/show"),
+        help=EXPERIMENTS_SHOW_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    add_rev_selection_flags(experiments_show_parser, "Show")
+    experiments_show_parser.add_argument(
+        "-a",
+        "--all-branches",
+        action="store_true",
+        default=False,
+        help="Show experiments derived from the tip of all Git branches.",
+    )
+    experiments_show_parser.add_argument(
+        "-T",
+        "--all-tags",
+        action="store_true",
+        default=False,
+        help="Show experiments derived from all Git tags.",
+    )
+    experiments_show_parser.add_argument(
+        "--no-pager",
+        action="store_true",
+        default=False,
+        help="Do not pipe output into a pager.",
+    )
+    experiments_show_parser.add_argument(
+        "--only-changed",
+        action="store_true",
+        default=False,
+        help=(
+            "Only show metrics/params with values varying "
+            "across the selected experiments."
+        ),
+    )
+    experiments_show_parser.add_argument(
+        "--drop",
+        help="Remove the columns matching the specified regex pattern.",
+        metavar="<regex_pattern>",
+    )
+    experiments_show_parser.add_argument(
+        "--keep",
+        help="Preserve the columns matching the specified regex pattern.",
+        metavar="<regex_pattern>",
+    )
+    experiments_show_parser.add_argument(
+        "--param-deps",
+        action="store_true",
+        default=False,
+        help="Show only params that are stage dependencies.",
+    )
+    experiments_show_parser.add_argument(
+        "--sort-by",
+        help="Sort related experiments by the specified metric or param.",
+        metavar="<metric/param>",
+    )
+    experiments_show_parser.add_argument(
+        "--sort-order",
+        help=(
+            "Sort order to use with --sort-by."
+            " Defaults to ascending ('asc')."
+        ),
+        choices=("asc", "desc"),
+        default="asc",
+    )
+    experiments_show_parser.add_argument(
+        "--sha",
+        action="store_true",
+        default=False,
+        help="Always show git commit SHAs instead of branch/tag names.",
+    )
+    experiments_show_parser.add_argument(
+        "--json",
+        "--show-json",
+        action="store_true",
+        default=False,
+        help="Print output in JSON format instead of a human-readable table.",
+    )
+    experiments_show_parser.add_argument(
+        "--csv",
+        "--show-csv",
+        action="store_true",
+        default=False,
+        help="Print output in csv format instead of a human-readable table.",
+    )
+    experiments_show_parser.add_argument(
+        "--md",
+        "--show-md",
+        action="store_true",
+        default=False,
+        dest="markdown",
+        help="Show tabulated output in the Markdown format (GFM).",
+    )
+    experiments_show_parser.add_argument(
+        "--precision",
+        type=int,
+        help=(
+            "Round metrics/params to `n` digits precision after the decimal "
+            f"point. Rounds to {DEFAULT_PRECISION} digits by default."
+        ),
+        metavar="<n>",
+    )
+    experiments_show_parser.add_argument(
+        "--parallel-coordinates-plot",
+        "--pcp",
+        dest="pcp",
+        action="store_true",
+        default=False,
+        help="Generate a Parallel Coordinates Plot from the tabulated output.",
+    )
+    experiments_show_parser.add_argument(
+        "-o",
+        "--out",
+        default=None,
+        help="Destination folder to save the Parallel Coordinates Plot to",
+        metavar="<path>",
+    ).complete = completion.DIR
+    experiments_show_parser.add_argument(
+        "--open",
+        action="store_true",
+        default=False,
+        help="Open the Parallel Coordinates Plot directly in the browser.",
+    )
+    experiments_show_parser.add_argument(
+        "--no-fetch",
+        dest="fetch_running",
+        action="store_false",
+        help=argparse.SUPPRESS,
+    )
+    experiments_show_parser.set_defaults(func=CmdExperimentsShow)

dvc/commands/freeze.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)
@@ -15,7 +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

@@ -2,7 +2,8 @@ import argparse
 import logging
 import os
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)
@@ -136,7 +137,6 @@ def add_parser(subparsers, parent_parser):
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/get.py

@@ -1,10 +1,11 @@
 import argparse
 import logging
 
+from dvc.cli import completion
+from dvc.cli.command import CmdBaseNoRepo
 from dvc.exceptions import DvcException
 
-from . import completion
+from ..cli.utils import append_doc_link
-from .base import CmdBaseNoRepo, append_doc_link
 
 logger = logging.getLogger(__name__)
 
@@ -93,7 +94,6 @@ def add_parser(subparsers, parent_parser):
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/get_url.py

@@ -1,10 +1,11 @@
 import argparse
 import logging
 
+from dvc.cli import completion
+from dvc.cli.command import CmdBaseNoRepo
 from dvc.exceptions import DvcException
 
-from . import completion
+from ..cli.utils import append_doc_link
-from .base import CmdBaseNoRepo, append_doc_link
 
 logger = logging.getLogger(__name__)
 
@@ -17,7 +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
 
 
@@ -43,7 +44,6 @@ def add_parser(subparsers, parent_parser):
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/git_hook.py

@@ -1,7 +1,8 @@
 import logging
 import os
 
-from dvc.command.base import CmdBaseNoRepo, fix_subparsers
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import fix_subparsers
 from dvc.exceptions import NotDvcRepoError
 
 logger = logging.getLogger(__name__)
@@ -25,7 +26,7 @@ class CmdHookBase(CmdBaseNoRepo):
 
 class CmdPreCommit(CmdHookBase):
     def _run(self):
-        from dvc.main import main
+        from dvc.cli import main
 
         return main(["status"])
 
@@ -49,14 +50,14 @@ class CmdPostCheckout(CmdHookBase):
         if os.path.isdir(os.path.join(".git", "rebase-merge")):
             return 0
 
-        from dvc.main import main
+        from dvc.cli import main
 
         return main(["checkout"])
 
 
 class CmdPrePush(CmdHookBase):
     def _run(self):
-        from dvc.main import main
+        from dvc.cli import main
 
         return main(["push"])
 

dvc/commands/imp.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)
@@ -90,7 +91,6 @@ def add_parser(subparsers, parent_parser):
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/imp_url.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)
@@ -90,7 +91,6 @@ def add_parser(subparsers, parent_parser):
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/init.py

@@ -4,7 +4,8 @@ import logging
 import colorama
 
 from dvc import analytics
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.utils import boxify
 from dvc.utils import format_link as fmt_link
 

dvc/commands/install.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/ls/__init__.py

@@ -1,9 +1,10 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
-from dvc.command.ls.ls_colors import LsColors
+from dvc.cli.utils import append_doc_link
+from dvc.commands.ls.ls_colors import LsColors
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 
@@ -41,7 +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/machine.py

@@ -1,7 +1,8 @@
 import argparse
 
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.cli.command import CmdBase
-from dvc.command.config import CmdConfig
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.commands.config import CmdConfig
 from dvc.compare import TabularData
 from dvc.config import ConfigError
 from dvc.exceptions import DvcException
@@ -241,6 +242,7 @@ class CmdMachineStatus(CmdBase):
         "instance_hdd_size",
         "instance_gpu",
     ]
+    FILL_VALUE = "-"
 
     def _add_row(
         self,
@@ -250,7 +252,11 @@ class CmdMachineStatus(CmdBase):
     ):
 
         if not all_status:
-            row = [name, None, "offline"]
+            row = [
+                name,
+                self.FILL_VALUE,
+                "offline",
+            ]  # back to `None` after #7167
             td.append(row)
         for i, status in enumerate(all_status, start=1):
             row = [name, f"num_{i}", "running" if status else "offline"]
@@ -264,7 +270,7 @@ class CmdMachineStatus(CmdBase):
             raise MachineDisabledError
 
         td = TabularData(
-            self.INSTANCE_FIELD + self.SHOWN_FIELD, fill_value="-"
+            self.INSTANCE_FIELD + self.SHOWN_FIELD, fill_value=self.FILL_VALUE
         )
 
         if self.args.name:
@@ -304,7 +310,7 @@ class CmdMachineSsh(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.config import parent_config_parser
+    from dvc.commands.config import parent_config_parser
 
     machine_HELP = "Set up and manage cloud machines."
     machine_parser = subparsers.add_parser(

dvc/commands/metrics.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 from dvc.utils.serialize import encode_exception

dvc/commands/move.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/params.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 

dvc/commands/plots.py

@@ -0,0 +1,458 @@
+import argparse
+import json
+import logging
+import os
+
+from funcy import first
+
+from dvc.cli import completion
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
+from dvc.exceptions import DvcException
+from dvc.ui import ui
+from dvc.utils import format_link
+
+logger = logging.getLogger(__name__)
+
+
+def _show_json(renderers, split=False):
+    from dvc.render.convert import to_json
+
+    result = {
+        renderer.name: to_json(renderer, split) for renderer in renderers
+    }
+    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
+
+    def _props(self):
+        from dvc.schema import PLOT_PROPS
+
+        # Pass only props specified by user, to not shadow ones from plot def
+        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_defs_renderers
+        from dvc_render import render_html
+
+        if self.args.show_vega:
+            if not self.args.targets:
+                logger.error("please specify a target for `--show-vega`")
+                return 1
+            if len(self.args.targets) > 1:
+                logger.error(
+                    "you can only specify one target for `--show-vega`"
+                )
+                return 1
+            if self.args.json:
+                logger.error(
+                    "'--show-vega' and '--json' are mutually exclusive "
+                    "options."
+                )
+                return 1
+
+        try:
+            plots_data = self._func(
+                targets=self.args.targets,
+                props=self._props(),
+                config_files=self._config_files(),
+            )
+
+            if not plots_data:
+                ui.error_write(
+                    "No plots were loaded, "
+                    "visualization file will not be created."
+                )
+
+            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_defs_renderers(
+                data=plots_data,
+                out=renderers_out,
+                templates_dir=self.repo.plots.templates_dir,
+            )
+            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:
+                renderers = _filter_unhandled_renderers(renderers)
+                _show_json(renderers, self.args.split)
+                return 0
+
+            _adjust_vega_renderers(renderers)
+
+            output_file: Path = (Path.cwd() / out).resolve() / "index.html"
+
+            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)
+
+            return 0
+
+        except DvcException:
+            logger.exception("")
+            return 1
+
+
+class CmdPlotsShow(CmdPlots):
+    UNINITIALIZED = True
+
+    def _func(self, *args, **kwargs):
+        return self.repo.plots.show(*args, **kwargs)
+
+
+class CmdPlotsDiff(CmdPlots):
+    UNINITIALIZED = True
+
+    def _func(self, *args, **kwargs):
+        return self.repo.plots.diff(
+            *args,
+            revs=self.args.revisions,
+            experiment=self.args.experiment,
+            **kwargs,
+        )
+
+
+class CmdPlotsModify(CmdPlots):
+    def run(self):
+        self.repo.plots.modify(
+            self.args.target, props=self._props(), unset=self.args.unset
+        )
+        return 0
+
+
+class CmdPlotsTemplates(CmdBase):
+    TEMPLATES_CHOICES = [
+        "simple",
+        "linear",
+        "confusion",
+        "confusion_normalized",
+        "scatter",
+        "smooth",
+    ]
+
+    def run(self):
+        from dvc_render.vega_templates import dump_templates
+
+        try:
+            out = (
+                os.path.join(os.getcwd(), self.args.out)
+                if self.args.out
+                else self.repo.plots.templates_dir
+            )
+
+            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}'.")
+
+            return 0
+        except DvcException:
+            logger.exception("")
+            return 1
+
+
+def add_parser(subparsers, parent_parser):
+    PLOTS_HELP = "Commands to visualize and compare plot data."
+
+    plots_parser = subparsers.add_parser(
+        "plots",
+        parents=[parent_parser],
+        description=append_doc_link(PLOTS_HELP, "plots"),
+        help=PLOTS_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    plots_subparsers = plots_parser.add_subparsers(
+        dest="cmd",
+        help="Use `dvc plots CMD --help` to display command-specific help.",
+    )
+
+    fix_subparsers(plots_subparsers)
+
+    SHOW_HELP = (
+        "Generate plots from target files or plots definitions from "
+        "`dvc.yaml` file."
+    )
+    plots_show_parser = plots_subparsers.add_parser(
+        "show",
+        parents=[parent_parser],
+        description=append_doc_link(SHOW_HELP, "plots/show"),
+        help=SHOW_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    plots_show_parser.add_argument(
+        "targets",
+        nargs="*",
+        help="Plots to visualize. Supports any file path, or plot name "
+        "defined in `dvc.yaml`. "
+        "Shows all plots by default.",
+    ).complete = completion.FILE
+    _add_props_arguments(plots_show_parser)
+    _add_output_argument(plots_show_parser)
+    _add_ui_arguments(plots_show_parser)
+    plots_show_parser.set_defaults(func=CmdPlotsShow)
+
+    PLOTS_DIFF_HELP = (
+        "Show multiple versions of plot data "
+        "by plotting them in a single image."
+    )
+    plots_diff_parser = plots_subparsers.add_parser(
+        "diff",
+        parents=[parent_parser],
+        description=append_doc_link(PLOTS_DIFF_HELP, "plots/diff"),
+        help=PLOTS_DIFF_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    plots_diff_parser.add_argument(
+        "--targets",
+        nargs="*",
+        help=(
+            "Specific plots to visualize. "
+            "Accepts any file path or plot name from `dvc.yaml` file. "
+            "Shows all tracked plots by default."
+        ),
+        metavar="<paths>",
+    ).complete = completion.FILE
+    plots_diff_parser.add_argument(
+        "-e",
+        "--experiment",
+        action="store_true",
+        default=False,
+        help=argparse.SUPPRESS,
+    )
+    plots_diff_parser.add_argument(
+        "revisions", nargs="*", default=None, help="Git commits to plot from"
+    )
+    _add_props_arguments(plots_diff_parser)
+    _add_output_argument(plots_diff_parser)
+    _add_ui_arguments(plots_diff_parser)
+    plots_diff_parser.set_defaults(func=CmdPlotsDiff)
+
+    PLOTS_MODIFY_HELP = (
+        "Modify display properties of data-series plot outputs "
+        "(has no effect on image-type plots)."
+    )
+    plots_modify_parser = plots_subparsers.add_parser(
+        "modify",
+        parents=[parent_parser],
+        description=append_doc_link(PLOTS_MODIFY_HELP, "plots/modify"),
+        help=PLOTS_MODIFY_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    plots_modify_parser.add_argument(
+        "target", help="Plot output to set properties to"
+    ).complete = completion.FILE
+    _add_props_arguments(plots_modify_parser)
+    plots_modify_parser.add_argument(
+        "--unset",
+        nargs="*",
+        metavar="<property>",
+        help="Unset one or more display properties.",
+    )
+    plots_modify_parser.set_defaults(func=CmdPlotsModify)
+
+    TEMPLATES_HELP = (
+        "Write built-in plots templates to a directory "
+        "(.dvc/plots by default)."
+    )
+    plots_templates_parser = plots_subparsers.add_parser(
+        "templates",
+        parents=[parent_parser],
+        description=append_doc_link(TEMPLATES_HELP, "plots/templates"),
+        help=TEMPLATES_HELP,
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    plots_templates_parser.add_argument(
+        "target",
+        default=None,
+        nargs="?",
+        choices=CmdPlotsTemplates.TEMPLATES_CHOICES,
+        help="Template to write. Writes all templates by default.",
+    )
+    _add_output_argument(plots_templates_parser, typ="templates")
+    plots_templates_parser.set_defaults(func=CmdPlotsTemplates)
+
+
+def _add_props_arguments(parser):
+    parser.add_argument(
+        "-t",
+        "--template",
+        nargs="?",
+        default=None,
+        help=(
+            "Special JSON or HTML schema file to inject with the data. "
+            "See {}".format(
+                format_link("https://man.dvc.org/plots#plot-templates")
+            )
+        ),
+        metavar="<path>",
+    ).complete = completion.FILE
+    parser.add_argument(
+        "-x", default=None, help="Field name for X axis.", metavar="<field>"
+    )
+    parser.add_argument(
+        "-y", default=None, help="Field name for Y axis.", metavar="<field>"
+    )
+    parser.add_argument(
+        "--no-header",
+        action="store_false",
+        dest="header",
+        default=None,  # Use default None to distinguish when it's not used
+        help="Provided CSV or TSV datafile does not have a header.",
+    )
+    parser.add_argument(
+        "--title", default=None, metavar="<text>", help="Plot title."
+    )
+    parser.add_argument(
+        "--x-label", default=None, help="X axis label", metavar="<text>"
+    )
+    parser.add_argument(
+        "--y-label", default=None, help="Y axis label", metavar="<text>"
+    )
+
+
+def _add_output_argument(parser, typ="plots"):
+    parser.add_argument(
+        "-o",
+        "--out",
+        default=None,
+        help=f"Directory to save {typ} to.",
+        metavar="<path>",
+    ).complete = completion.DIR
+
+
+def _add_ui_arguments(parser):
+    parser.add_argument(
+        "--show-vega",
+        action="store_true",
+        default=False,
+        help="Show output in Vega format.",
+    )
+    parser.add_argument(
+        "--json",
+        "--show-json",
+        action="store_true",
+        default=False,
+        help=argparse.SUPPRESS,
+    )
+    parser.add_argument(
+        "--split", action="store_true", default=False, help=argparse.SUPPRESS
+    )
+    parser.add_argument(
+        "--open",
+        action="store_true",
+        default=False,
+        help="Open plot file directly in the browser.",
+    )
+    parser.add_argument(
+        "--html-template",
+        default=None,
+        help="Custom HTML template for VEGA visualization.",
+        metavar="<path>",
+    )
+    parser.add_argument(
+        "--from-config",
+        default=None,
+        metavar="<path>",
+        help=argparse.SUPPRESS,
+    )

dvc/commands/remote.py

@@ -1,7 +1,7 @@
 import argparse
 
-from dvc.command.base import append_doc_link, fix_subparsers
+from dvc.cli.utils import append_doc_link, fix_subparsers
-from dvc.command.config import CmdConfig
+from dvc.commands.config import CmdConfig
 from dvc.ui import ui
 from dvc.utils import format_link
 
@@ -113,8 +113,8 @@ class CmdRemoteDefault(CmdRemote):
 class CmdRemoteList(CmdRemote):
     def run(self):
         conf = self.config.read(self.args.level)
-        for name, conf in conf["remote"].items():
+        for name, remote_conf in conf["remote"].items():
-            ui.write(name, conf["url"], sep="\t")
+            ui.write(name, remote_conf["url"], sep="\t")
         return 0
 
 
@@ -151,7 +151,7 @@ class CmdRemoteRename(CmdRemote):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.config import parent_config_parser
+    from dvc.commands.config import parent_config_parser
 
     REMOTE_HELP = "Set up and manage data remotes."
     remote_parser = subparsers.add_parser(

dvc/commands/remove.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/repro.py

@@ -1,21 +1,22 @@
 import argparse
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
-from dvc.command.status import CmdDataStatus
+from dvc.cli.utils import append_doc_link
+from dvc.commands.status import CmdDataStatus
 
 
 class CmdRepro(CmdBase):
     def run(self):
         from dvc.ui import ui
 
-        stages = self.repo.reproduce(**self._repro_kwargs)
+        stages = self.repo.reproduce(
+            **self._common_kwargs, **self._repro_kwargs
+        )
         if len(stages) == 0:
             ui.write(CmdDataStatus.UP_TO_DATE_MSG)
         else:
-            ui.write(
+            ui.write("Use `dvc push` to send your updates to remote storage.")
-                "Use `dvc push` to send your updates to " "remote storage."
-            )
 
         if self.args.metrics:
             from dvc.compare import show_metrics
@@ -26,7 +27,7 @@ class CmdRepro(CmdBase):
         return 0
 
     @property
-    def _repro_kwargs(self):
+    def _common_kwargs(self):
         return {
             "targets": self.args.targets,
             "single_item": self.args.single_item,
@@ -35,12 +36,17 @@ class CmdRepro(CmdBase):
             "interactive": self.args.interactive,
             "pipeline": self.args.pipeline,
             "all_pipelines": self.args.all_pipelines,
-            "run_cache": not self.args.no_run_cache,
-            "no_commit": self.args.no_commit,
             "downstream": self.args.downstream,
             "recursive": self.args.recursive,
             "force_downstream": self.args.force_downstream,
             "pull": self.args.pull,
+        }
+
+    @property
+    def _repro_kwargs(self):
+        return {
+            "run_cache": not self.args.no_run_cache,
+            "no_commit": self.args.no_commit,
             "glob": self.args.glob,
         }
 
@@ -58,6 +64,13 @@ def add_arguments(repro_parser):
         default=False,
         help="Reproduce even if dependencies were not changed.",
     )
+    repro_parser.add_argument(
+        "-i",
+        "--interactive",
+        action="store_true",
+        default=False,
+        help="Ask for confirmation before reproducing each stage.",
+    )
     repro_parser.add_argument(
         "-s",
         "--single-item",
@@ -66,27 +79,6 @@ def add_arguments(repro_parser):
         help="Reproduce only single data item without recursive dependencies "
         "check.",
     )
-    repro_parser.add_argument(
-        "-m",
-        "--metrics",
-        action="store_true",
-        default=False,
-        help="Show metrics after reproduction.",
-    )
-    repro_parser.add_argument(
-        "--dry",
-        action="store_true",
-        default=False,
-        help="Only print the commands that would be executed without "
-        "actually executing.",
-    )
-    repro_parser.add_argument(
-        "-i",
-        "--interactive",
-        action="store_true",
-        default=False,
-        help="Ask for confirmation before reproducing each stage.",
-    )
     repro_parser.add_argument(
         "-p",
         "--pipeline",
@@ -110,13 +102,17 @@ def add_arguments(repro_parser):
         help="Reproduce all stages in the specified directory.",
     )
     repro_parser.add_argument(
-        "--no-run-cache",
+        "-m",
+        "--metrics",
         action="store_true",
         default=False,
-        help=(
+        help="Show metrics after reproduction.",
-            "Execute stage commands even if they have already been run with "
+    )
-            "the same command/dependencies/outputs/etc before."
+    repro_parser.add_argument(
-        ),
+        "--downstream",
+        action="store_true",
+        default=False,
+        help="Start from the specified stages when reproducing pipelines.",
     )
     repro_parser.add_argument(
         "--force-downstream",
@@ -125,18 +121,6 @@ def add_arguments(repro_parser):
         help="Reproduce all descendants of a changed stage even if their "
         "direct dependencies didn't change.",
     )
-    repro_parser.add_argument(
-        "--no-commit",
-        action="store_true",
-        default=False,
-        help="Don't put files/directories into cache.",
-    )
-    repro_parser.add_argument(
-        "--downstream",
-        action="store_true",
-        default=False,
-        help="Start from the specified stages when reproducing pipelines.",
-    )
     repro_parser.add_argument(
         "--pull",
         action="store_true",
@@ -147,10 +131,11 @@ def add_arguments(repro_parser):
         ),
     )
     repro_parser.add_argument(
-        "--glob",
+        "--dry",
         action="store_true",
         default=False,
-        help="Allows targets containing shell-style wildcards.",
+        help="Only print the commands that would be executed without "
+        "actually executing.",
     )
 
 
@@ -165,5 +150,28 @@ def add_parser(subparsers, parent_parser):
         help=REPRO_HELP,
         formatter_class=argparse.RawDescriptionHelpFormatter,
     )
+    # repro/exp run shared args
     add_arguments(repro_parser)
+    # repro only args
+    repro_parser.add_argument(
+        "--glob",
+        action="store_true",
+        default=False,
+        help="Allows targets containing shell-style wildcards.",
+    )
+    repro_parser.add_argument(
+        "--no-commit",
+        action="store_true",
+        default=False,
+        help="Don't put files/directories into cache.",
+    )
+    repro_parser.add_argument(
+        "--no-run-cache",
+        action="store_true",
+        default=False,
+        help=(
+            "Execute stage commands even if they have already been run with "
+            "the same command/dependencies/outputs/etc before."
+        ),
+    )
     repro_parser.set_defaults(func=CmdRepro)

dvc/commands/root.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.utils import relpath
 
 logger = logging.getLogger(__name__)

dvc/commands/run.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
-from dvc.command.stage import parse_cmd
+from dvc.cli.utils import append_doc_link
+from dvc.commands.stage import parse_cmd
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)
@@ -52,7 +53,7 @@ class CmdRun(CmdBase):
 
 
 def add_parser(subparsers, parent_parser):
-    from dvc.command.stage import _add_common_args
+    from dvc.commands.stage import _add_common_args
 
     RUN_HELP = (
         "Generate a dvc.yaml file from a command and execute the command."
@@ -68,12 +69,25 @@ def add_parser(subparsers, parent_parser):
     run_parser.add_argument(
         "--file", metavar="<filename>", help=argparse.SUPPRESS
     )
+    run_parser.add_argument(
+        "--single-stage",
+        action="store_true",
+        default=False,
+        help=argparse.SUPPRESS,
+    )
+    _add_common_args(run_parser)
     run_parser.add_argument(
         "--no-exec",
         action="store_true",
         default=False,
         help="Only create dvc.yaml without actually running it.",
     )
+    run_parser.add_argument(
+        "--no-commit",
+        action="store_true",
+        default=False,
+        help="Don't put files/directories into cache.",
+    )
     run_parser.add_argument(
         "--no-run-cache",
         action="store_true",
@@ -83,17 +97,4 @@ def add_parser(subparsers, parent_parser):
             "with the same command/dependencies/outputs/etc before."
         ),
     )
-    run_parser.add_argument(
-        "--no-commit",
-        action="store_true",
-        default=False,
-        help="Don't put files/directories into cache.",
-    )
-    run_parser.add_argument(
-        "--single-stage",
-        action="store_true",
-        default=False,
-        help=argparse.SUPPRESS,
-    )
-    _add_common_args(run_parser)
     run_parser.set_defaults(func=CmdRun)

dvc/commands/stage.py

@@ -3,8 +3,9 @@ import logging
 from itertools import chain, filterfalse
 from typing import TYPE_CHECKING, Dict, Iterable, List
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link, fix_subparsers
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link, fix_subparsers
 from dvc.utils.cli_parse import parse_params
 from dvc.utils.humanize import truncate_text
 
@@ -93,9 +94,7 @@ class CmdStageList(CmdBase):
         self.repo.stage_collection_error_handler = log_error
 
         stages = self._get_stages()
-        names_only = self.args.names_only
+        data = prepare_stages_data(stages, description=not self.args.name_only)
-
-        data = prepare_stages_data(stages, description=not names_only)
         ui.table(data.items())
 
         return 0
@@ -138,6 +137,13 @@ class CmdStageAdd(CmdBase):
 
 
 def _add_common_args(parser):
+    parser.add_argument(
+        "-f",
+        "--force",
+        action="store_true",
+        default=False,
+        help="Overwrite existing stage",
+    )
     parser.add_argument(
         "-d",
         "--deps",
@@ -146,6 +152,14 @@ def _add_common_args(parser):
         help="Declare dependencies for reproducible cmd.",
         metavar="<path>",
     ).complete = completion.FILE
+    parser.add_argument(
+        "-p",
+        "--params",
+        action="append",
+        default=[],
+        help="Declare parameter to use as additional dependency.",
+        metavar="[<filename>:]<params_list>",
+    ).complete = completion.FILE
     parser.add_argument(
         "-o",
         "--outs",
@@ -164,13 +178,36 @@ def _add_common_args(parser):
         metavar="<filename>",
     ).complete = completion.FILE
     parser.add_argument(
-        "-p",
+        "-c",
-        "--params",
+        "--checkpoints",
         action="append",
         default=[],
-        help="Declare parameter to use as additional dependency.",
+        help="Declare checkpoint output file or directory for 'dvc exp run'. "
-        metavar="[<filename>:]<params_list>",
+        "Not compatible with 'dvc repro'.",
+        metavar="<filename>",
     ).complete = completion.FILE
+    parser.add_argument(
+        "--external",
+        action="store_true",
+        default=False,
+        help="Allow outputs that are outside of the DVC repository.",
+    )
+    parser.add_argument(
+        "--outs-persist",
+        action="append",
+        default=[],
+        help="Declare output file or directory that will not be "
+        "removed upon repro.",
+        metavar="<filename>",
+    )
+    parser.add_argument(
+        "--outs-persist-no-cache",
+        action="append",
+        default=[],
+        help="Declare output file or directory that will not be "
+        "removed upon repro (do not put into DVC cache).",
+        metavar="<filename>",
+    )
     parser.add_argument(
         "-m",
         "--metrics",
@@ -213,7 +250,7 @@ def _add_common_args(parser):
         "--live-no-summary",
         action="store_true",
         default=False,
-        help="Signal dvclive logger to not dump latest metrics file.",
+        help=argparse.SUPPRESS,
     )
     parser.add_argument(
         "--live-no-html",
@@ -227,50 +264,12 @@ def _add_common_args(parser):
         help="Directory within your repo to run your command in.",
         metavar="<path>",
     )
-    parser.add_argument(
-        "-f",
-        "--force",
-        action="store_true",
-        default=False,
-        help="Overwrite existing stage",
-    )
-    parser.add_argument(
-        "--outs-persist",
-        action="append",
-        default=[],
-        help="Declare output file or directory that will not be "
-        "removed upon repro.",
-        metavar="<filename>",
-    )
-    parser.add_argument(
-        "--outs-persist-no-cache",
-        action="append",
-        default=[],
-        help="Declare output file or directory that will not be "
-        "removed upon repro (do not put into DVC cache).",
-        metavar="<filename>",
-    )
-    parser.add_argument(
-        "-c",
-        "--checkpoints",
-        action="append",
-        default=[],
-        help="Declare checkpoint output file or directory for 'dvc exp run'. "
-        "Not compatible with 'dvc repro'.",
-        metavar="<filename>",
-    ).complete = completion.FILE
     parser.add_argument(
         "--always-changed",
         action="store_true",
         default=False,
         help="Always consider this DVC-file as changed.",
     )
-    parser.add_argument(
-        "--external",
-        action="store_true",
-        default=False,
-        help="Allow outputs that are outside of the DVC repository.",
-    )
     parser.add_argument(
         "--desc",
         type=str,
@@ -357,6 +356,7 @@ def add_parser(subparsers, parent_parser):
         help="List all stages inside the specified directory.",
     )
     stage_list_parser.add_argument(
+        "--name-only",
         "--names-only",
         action="store_true",
         default=False,

dvc/commands/status.py

@@ -1,6 +1,6 @@
 import logging
 
-from dvc.command.data_sync import CmdDataBase
+from dvc.commands.data_sync import CmdDataBase
 from dvc.exceptions import DvcException
 from dvc.ui import ui
 from dvc.utils import format_link

dvc/commands/unprotect.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)

dvc/commands/update.py

@@ -1,8 +1,9 @@
 import argparse
 import logging
 
-from dvc.command import completion
+from dvc.cli import completion
-from dvc.command.base import CmdBase, append_doc_link
+from dvc.cli.command import CmdBase
+from dvc.cli.utils import append_doc_link
 from dvc.exceptions import DvcException
 
 logger = logging.getLogger(__name__)
@@ -27,7 +28,10 @@ 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],
@@ -70,7 +74,6 @@ def add_parser(subparsers, parent_parser):
         help=(
             "Number of jobs to run simultaneously. "
             "The default value is 4 * cpu_count(). "
-            "For SSH remotes, the default is 4. "
         ),
         metavar="<number>",
     )

dvc/commands/version.py

@@ -1,7 +1,8 @@
 import argparse
 import logging
 
-from dvc.command.base import CmdBaseNoRepo, append_doc_link
+from dvc.cli.command import CmdBaseNoRepo
+from dvc.cli.utils import append_doc_link
 from dvc.ui import ui
 
 logger = logging.getLogger(__name__)

dvc/compare.py

@@ -11,6 +11,7 @@ from typing import (
     List,
     Mapping,
     MutableSequence,
+    Optional,
     Sequence,
     Set,
     Tuple,
@@ -21,6 +22,7 @@ from typing import (
 from funcy import reraise
 
 if TYPE_CHECKING:
+    from dvc.types import StrPath
     from dvc.ui.table import CellT
 
 
@@ -37,11 +39,21 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         self._columns: Dict[str, Column] = {name: Column() for name in columns}
         self._keys: List[str] = list(columns)
         self._fill_value = fill_value
+        self._protected: Set[str] = set()
 
     @property
     def columns(self) -> List[Column]:
         return list(map(self.column, self.keys()))
 
+    def is_protected(self, col_name) -> bool:
+        return col_name in self._protected
+
+    def protect(self, *col_names: str):
+        self._protected.update(col_names)
+
+    def unprotect(self, *col_names: str):
+        self._protected = self._protected.difference(col_names)
+
     def column(self, name: str) -> Column:
         return self._columns[name]
 
@@ -125,9 +137,10 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         return len(self.columns), len(self)
 
     def drop(self, *col_names: str) -> None:
-        for col_name in col_names:
+        for col in col_names:
-            self._keys.remove(col_name)
+            if not self.is_protected(col):
-            self._columns.pop(col_name)
+                self._keys.remove(col)
+                self._columns.pop(col)
 
     def rename(self, from_col_name: str, to_col_name: str) -> None:
         self._columns[to_col_name] = self._columns.pop(from_col_name)
@@ -153,6 +166,24 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             writer.writerow(row)
         return buff.getvalue()
 
+    def to_parallel_coordinates(
+        self, output_path: "StrPath", color_by: str = None
+    ) -> "StrPath":
+        from dvc_render.html import render_html
+        from dvc_render.plotly import ParallelCoordinatesRenderer
+
+        render_html(
+            renderers=[
+                ParallelCoordinatesRenderer(
+                    self.as_dict(),
+                    color_by=color_by,
+                    fill_value=self._fill_value,
+                )
+            ],
+            output_file=output_path,
+        )
+        return output_path
+
     def add_column(self, name: str) -> None:
         self._columns[name] = Column([self._fill_value] * len(self))
         self._keys.append(name)
@@ -184,7 +215,12 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             {k: self._columns[k][i] for k in keys} for i in range(len(self))
         ]
 
-    def dropna(self, axis: str = "rows", how="any"):
+    def dropna(
+        self,
+        axis: str = "rows",
+        how="any",
+        subset: Optional[Iterable[str]] = None,
+    ):
         if axis not in ["rows", "cols"]:
             raise ValueError(
                 f"Invalid 'axis' value {axis}."
@@ -196,13 +232,15 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             )
 
         match_line: Set = set()
-        match = True
+        match_any = True
         if how == "all":
-            match = False
+            match_any = False
 
         for n_row, row in enumerate(self):
             for n_col, col in enumerate(row):
-                if (col == self._fill_value) is match:
+                if subset and self.keys()[n_col] not in subset:
+                    continue
+                if (col == self._fill_value) is match_any:
                     if axis == "rows":
                         match_line.add(n_row)
                         break
@@ -229,7 +267,12 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         else:
             self.drop(*to_drop)
 
-    def drop_duplicates(self, axis: str = "rows"):
+    def drop_duplicates(
+        self,
+        axis: str = "rows",
+        subset: Optional[Iterable[str]] = None,
+        ignore_empty: bool = True,
+    ):
         if axis not in ["rows", "cols"]:
             raise ValueError(
                 f"Invalid 'axis' value {axis}."
@@ -239,8 +282,12 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
         if axis == "cols":
             cols_to_drop: List[str] = []
             for n_col, col in enumerate(self.columns):
+                if subset and self.keys()[n_col] not in subset:
+                    continue
                 # Cast to str because Text is not hashable error
-                unique_vals = {str(x) for x in col if x != self._fill_value}
+                unique_vals = {str(x) for x in col}
+                if ignore_empty and self._fill_value in unique_vals:
+                    unique_vals -= {self._fill_value}
                 if len(unique_vals) == 1:
                     cols_to_drop.append(self.keys()[n_col])
             self.drop(*cols_to_drop)
@@ -249,6 +296,13 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
             unique_rows = []
             rows_to_drop: List[int] = []
             for n_row, row in enumerate(self):
+                if subset:
+                    row = [
+                        col
+                        for n_col, col in enumerate(row)
+                        if self.keys()[n_col] in subset
+                    ]
+
                 tuple_row = tuple(row)
                 if tuple_row in unique_rows:
                     rows_to_drop.append(n_row)

dvc/config.py

@@ -86,22 +86,21 @@ class Config(dict):
     def __init__(
         self, dvc_dir=None, validate=True, fs=None, config=None
     ):  # pylint: disable=super-init-not-called
-        from dvc.fs.local import LocalFileSystem
+        from dvc.fs import LocalFileSystem
 
         self.dvc_dir = dvc_dir
+        self.wfs = LocalFileSystem()
+        self.fs = fs or self.wfs
 
         if not dvc_dir:
             try:
                 from dvc.repo import Repo
 
-                self.dvc_dir = os.path.join(Repo.find_dvc_dir())
+                self.dvc_dir = Repo.find_dvc_dir()
             except NotDvcRepoError:
                 self.dvc_dir = None
         else:
-            self.dvc_dir = os.path.abspath(os.path.realpath(dvc_dir))
+            self.dvc_dir = self.fs.path.abspath(self.fs.path.realpath(dvc_dir))
-
-        self.wfs = LocalFileSystem(url=self.dvc_dir)
-        self.fs = fs or self.wfs
 
         self.load(validate=validate, config=config)
 
@@ -124,8 +123,8 @@ class Config(dict):
         }
 
         if self.dvc_dir is not None:
-            files["repo"] = os.path.join(self.dvc_dir, self.CONFIG)
+            files["repo"] = self.fs.path.join(self.dvc_dir, self.CONFIG)
-            files["local"] = os.path.join(self.dvc_dir, self.CONFIG_LOCAL)
+            files["local"] = self.fs.path.join(self.dvc_dir, self.CONFIG_LOCAL)
 
         return files
 
@@ -188,7 +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))
 
@@ -228,7 +227,7 @@ class Config(dict):
 
     @staticmethod
     def _to_relpath(conf_dir, path):
-        from dvc.fs.local import localfs
+        from dvc.fs import localfs
         from dvc.utils import relpath
 
         from .config_schema import RelPath
@@ -266,7 +265,12 @@ class Config(dict):
                     "key_path": func,
                 }
             },
-            "machine": {str: {"startup_script": func}},
+            "machine": {
+                str: {
+                    "startup_script": func,
+                    "setup_script": func,
+                }
+            },
         }
         return Schema(dirs_schema, extra=ALLOW_EXTRA)(conf)
 

dvc/config_schema.py

@@ -79,13 +79,13 @@ REMOTE_COMMON = {
     "checksum_jobs": All(Coerce(int), Range(1)),
     "jobs": All(Coerce(int), Range(1)),
     Optional("no_traverse"): Bool,  # obsoleted
-    "verify": Bool,
 }
 LOCAL_COMMON = {
     "type": supported_cache_type,
     Optional("protected", default=False): Bool,  # obsoleted
     "shared": All(Lower, Choices("group")),
     Optional("slow_link_warning", default=True): Bool,
+    Optional("verify", default=False): Bool,
 }
 HTTP_COMMON = {
     "auth": All(Lower, Choices("basic", "digest", "custom")),
@@ -95,6 +95,7 @@ HTTP_COMMON = {
     "ask_password": Bool,
     "ssl_verify": Any(Bool, str),
     "method": str,
+    Optional("verify", default=False): Bool,
 }
 WEBDAV_COMMON = {
     "user": str,
@@ -105,6 +106,7 @@ WEBDAV_COMMON = {
     "key_path": str,
     "timeout": Coerce(int),
     "ssl_verify": Any(Bool, str),
+    Optional("verify", default=False): Bool,
 }
 
 SCHEMA = {
@@ -150,6 +152,8 @@ 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,
@@ -158,11 +162,13 @@ SCHEMA = {
                     "cache_regions": bool,
                     "read_timeout": Coerce(int),
                     "connect_timeout": Coerce(int),
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "gs": {
                     "projectname": str,
                     "credentialpath": str,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "ssh": {
@@ -175,6 +181,7 @@ SCHEMA = {
                     "timeout": Coerce(int),
                     "gss_auth": Bool,
                     "allow_agent": Bool,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "hdfs": {"user": str, "kerb_ticket": str, **REMOTE_COMMON},
@@ -185,6 +192,7 @@ SCHEMA = {
                     "ssl_verify": Any(Bool, str),
                     "token": str,
                     "use_https": Bool,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "azure": {
@@ -200,12 +208,14 @@ SCHEMA = {
                     "exclude_visual_studio_code_credential": Bool,
                     "exclude_shared_token_cache_credential": Bool,
                     "exclude_managed_identity_credential": Bool,
+                    Optional("verify", default=False): Bool,
                     **REMOTE_COMMON,
                 },
                 "oss": {
                     "oss_key_id": str,
                     "oss_key_secret": str,
                     "oss_endpoint": str,
+                    Optional("verify", default=True): Bool,
                     **REMOTE_COMMON,
                 },
                 "gdrive": {
@@ -216,6 +226,7 @@ SCHEMA = {
                     "gdrive_service_account_user_email": str,
                     "gdrive_service_account_json_file_path": str,
                     Optional("gdrive_trash_only", default=False): Bool,
+                    Optional("verify", default=True): Bool,
                     **REMOTE_COMMON,
                 },
                 "http": {**HTTP_COMMON, **REMOTE_COMMON},
@@ -248,6 +259,7 @@ SCHEMA = {
             "instance_gpu": Lower,
             "ssh_private": str,
             "startup_script": str,
+            "setup_script": str,
         },
     },
     # section for experimental features
@@ -256,7 +268,11 @@ SCHEMA = {
         # enabled by default. It's of no use, kept for backward compatibility.
         Optional("parametrization", default=True): Bool,
     },
-    "plots": {"html_template": str},
+    "plots": {
+        "html_template": str,
+        Optional("auto_open", default=False): Bool,
+        "out_dir": str,
+    },
     "exp": {
         "code": str,
         "data": str,
@@ -266,4 +282,8 @@ SCHEMA = {
         "plots": str,
         "live": str,
     },
+    "parsing": {
+        "bool": All(Lower, Choices("store_true", "boolean_optional")),
+        "list": All(Lower, Choices("nargs", "append")),
+    },
 }

dvc/daemon.py

@@ -42,7 +42,7 @@ def _spawn_windows(cmd, env):
 
 
 def _spawn_posix(cmd, env):
-    from dvc.main import main
+    from dvc.cli import main
 
     # NOTE: using os._exit instead of sys.exit, because dvc built
     # with PyInstaller has trouble with SystemExit exception and throws
@@ -81,7 +81,7 @@ 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 +90,7 @@ def _spawn(cmd, env):
     else:
         raise NotImplementedError
 
-    logger.debug(f"Spawned '{cmd}'")
+    logger.debug("Spawned '%s'", cmd)
 
 
 def daemon(args):
@@ -106,8 +106,9 @@ def daemon(args):
     cmd = ["daemon", "-q"] + args
 
     env = fix_env()
-    file_path = os.path.abspath(inspect.stack()[0][1])
+    if not is_binary():
-    env["PYTHONPATH"] = os.path.dirname(os.path.dirname(file_path))
+        file_path = os.path.abspath(inspect.stack()[0][1])
+        env["PYTHONPATH"] = os.path.dirname(os.path.dirname(file_path))
     env[DVC_DAEMON] = "1"
 
     _spawn(cmd, env)

dvc/data_cloud.py

@@ -3,11 +3,12 @@
 import logging
 from typing import TYPE_CHECKING, Iterable, Optional
 
-from dvc.objects.db import get_index
+from dvc_data.db import get_index
 
 if TYPE_CHECKING:
-    from dvc.hash_info import HashInfo
+    from dvc_data.hashfile.db import HashFileDB
-    from dvc.objects.db.base import ObjectDB
+    from dvc_data.hashfile.hash_info import HashInfo
+    from dvc_data.status import CompareStatusResult
 
 logger = logging.getLogger(__name__)
 
@@ -30,7 +31,7 @@ class DataCloud:
         self,
         name: Optional[str] = None,
         command: str = "<command>",
-    ) -> "ObjectDB":
+    ) -> "HashFileDB":
         from dvc.config import NoRemoteError
 
         if not name:
@@ -56,18 +57,44 @@ class DataCloud:
 
     def _init_odb(self, name):
         from dvc.fs import get_cloud_fs
-        from dvc.objects.db import get_odb
+        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 _log_missing(self, status: "CompareStatusResult"):
+        if status.missing:
+            missing_desc = "\n".join(
+                f"name: {hash_info.obj_name}, {hash_info}"
+                for hash_info in status.missing
+            )
+            logger.warning(
+                "Some of the cache files do not exist neither locally "
+                f"nor on remote. Missing cache files:\n{missing_desc}"
+            )
+
+    def transfer(
+        self,
+        src_odb: "HashFileDB",
+        dest_odb: "HashFileDB",
+        objs: Iterable["HashInfo"],
+        **kwargs,
+    ):
+        from dvc.exceptions import FileTransferError
+        from dvc_data.transfer import TransferError, transfer
+
+        try:
+            return transfer(src_odb, dest_odb, objs, **kwargs)
+        except TransferError as exc:
+            raise FileTransferError(exc.fails) from exc
+
     def push(
         self,
         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.
 
@@ -78,17 +105,15 @@ class DataCloud:
                 By default remote from core.remote config option is used.
             odb: optional ODB to push to. Overrides remote.
         """
-        from dvc.objects.transfer import transfer
+        odb = odb or self.get_remote_odb(remote, "push")
-
+        return self.transfer(
-        if not odb:
-            odb = self.get_remote_odb(remote, "push")
-        return transfer(
             self.repo.odb.local,
             odb,
             objs,
             jobs=jobs,
             dest_index=get_index(odb),
             cache_odb=self.repo.odb.local,
+            validate_status=self._log_missing,
         )
 
     def pull(
@@ -96,7 +121,7 @@ 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.
 
@@ -107,11 +132,8 @@ class DataCloud:
                 By default remote from core.remote config option is used.
             odb: optional ODB to pull from. Overrides remote.
         """
-        from dvc.objects.transfer import transfer
+        odb = odb or self.get_remote_odb(remote, "pull")
-
+        return self.transfer(
-        if not odb:
-            odb = self.get_remote_odb(remote, "pull")
-        return transfer(
             odb,
             self.repo.odb.local,
             objs,
@@ -119,6 +141,7 @@ class DataCloud:
             src_index=get_index(odb),
             cache_odb=self.repo.odb.local,
             verify=odb.verify,
+            validate_status=self._log_missing,
         )
 
     def status(
@@ -126,7 +149,7 @@ 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.
@@ -141,7 +164,7 @@ class DataCloud:
             log_missing: log warning messages if file doesn't exist
                 neither in cache, neither in cloud.
         """
-        from dvc.objects.status import compare_status
+        from dvc_data.status import compare_status
 
         if not odb:
             odb = self.get_remote_odb(remote, "status")
@@ -157,5 +180,5 @@ 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/__init__.py

@@ -47,16 +47,29 @@ def loads_from(stage, s_list, erepo=None):
 def _merge_params(s_list):
     d = defaultdict(list)
     default_file = ParamsDependency.DEFAULT_PARAMS_FILE
+
+    # figure out completely tracked params file, and ignore specific keys
+    wholly_tracked = set()
+    for key in s_list:
+        if not isinstance(key, dict):
+            continue
+        wholly_tracked.update(k for k, params in key.items() if not params)
+
     for key in s_list:
         if isinstance(key, str):
-            d[default_file].append(key)
+            if default_file not in wholly_tracked:
+                d[default_file].append(key)
             continue
+
         if not isinstance(key, dict):
             msg = "Only list of str/dict is supported. Got: "
             msg += f"'{type(key).__name__}'."
             raise ValueError(msg)
 
         for k, params in key.items():
+            if k in wholly_tracked:
+                d[k] = []
+                continue
             if not isinstance(params, list):
                 msg = "Expected list of params for custom params file "
                 msg += f"'{k}', got '{type(params).__name__}'."

dvc/dependency/param.py

@@ -1,13 +1,15 @@
 import logging
 import os
+import typing
 from collections import defaultdict
+from typing import Dict
 
 import dpath.util
 from voluptuous import Any
 
 from dvc.exceptions import DvcException
-from dvc.hash_info import HashInfo
 from dvc.utils.serialize import LOADERS, ParseError
+from dvc_data.hashfile.hash_info import HashInfo
 
 from .base import Dependency
 
@@ -18,6 +20,14 @@ class MissingParamsError(DvcException):
     pass
 
 
+class MissingParamsFile(DvcException):
+    pass
+
+
+class ParamsIsADirectoryError(DvcException):
+    pass
+
+
 class BadParamFileError(DvcException):
     pass
 
@@ -27,51 +37,83 @@ class ParamsDependency(Dependency):
     PARAM_SCHEMA = {PARAM_PARAMS: Any(dict, list, None)}
     DEFAULT_PARAMS_FILE = "params.yaml"
 
-    def __init__(self, stage, path, params):
+    def __init__(self, stage, path, params=None, repo=None):
-        info = {}
+        self.params = list(params) if params else []
-        self.params = []
+        info = (
-        if params:
+            {self.PARAM_PARAMS: params} if isinstance(params, dict) else None
-            if isinstance(params, list):
-                self.params = params
-            else:
-                assert isinstance(params, dict)
-                self.params = list(params.keys())
-                info = {self.PARAM_PARAMS: params}
-
-        super().__init__(
-            stage,
-            path
-            or os.path.join(stage.repo.root_dir, self.DEFAULT_PARAMS_FILE),
-            info=info,
         )
+        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)
 
     def dumpd(self):
         ret = super().dumpd()
         if not self.hash_info:
-            ret[self.PARAM_PARAMS] = self.params
+            ret[self.PARAM_PARAMS] = self.params or {}
         return ret
 
     def fill_values(self, values=None):
         """Load params values dynamically."""
-        if not values:
+        if values is None:
             return
+
         info = {}
+        if not self.params:
+            info.update(values)
         for param in self.params:
             if param in values:
                 info[param] = values[param]
         self.hash_info = HashInfo(self.PARAM_PARAMS, info)
 
-    def workspace_status(self):
+    def read_params(
-        status = super().workspace_status()
+        self, flatten: bool = True, **kwargs: typing.Any
+    ) -> Dict[str, typing.Any]:
+        try:
+            config = self.read_file()
+        except MissingParamsFile:
+            config = {}
 
-        if status.get(str(self)) == "deleted":
+        if not self.params:
-            return status
+            return config
+
+        ret = {}
+        if flatten:
+            for param in self.params:
+                try:
+                    ret[param] = dpath.util.get(config, param, separator=".")
+                except KeyError:
+                    continue
+            return ret
+
+        from dpath.util import merge
+
+        for param in self.params:
+            merge(
+                ret,
+                dpath.util.search(config, param, separator="."),
+                separator=".",
+            )
+        return ret
+
+    def workspace_status(self):
+        if not self.exists:
+            return {str(self): "deleted"}
+        if self.hash_info.value is None:
+            return {str(self): "new"}
+
+        from funcy import ldistinct
 
         status = defaultdict(dict)
         info = self.hash_info.value if self.hash_info else {}
         actual = self.read_params()
-        for param in self.params:
+
-            if param not in actual.keys():
+        # NOTE: we want to preserve the order of params as specified in the
+        # status. In case of tracking the whole file, the order is top-level
+        # keys in the file and then the keys in the `info` from `dvc.lock`
+        # (which are alphabetically sorted).
+        params = self.params or ldistinct([*actual.keys(), *info.keys()])
+        for param in params:
+            if param not in actual:
                 st = "deleted"
             elif param not in info:
                 st = "new"
@@ -88,12 +130,19 @@ class ParamsDependency(Dependency):
     def status(self):
         return self.workspace_status()
 
-    def _read(self):
+    def validate_filepath(self):
         if not self.exists:
-            return {}
+            raise MissingParamsFile(f"Parameters file '{self}' does not exist")
+        if self.isdir():
+            raise ParamsIsADirectoryError(
+                f"'{self}' is a directory, expected a parameters file"
+            )
 
-        suffix = self.repo.fs.path.suffix(self.fs_path).lower()
+    def read_file(self):
-        loader = LOADERS[suffix]
+        _, ext = os.path.splitext(self.fs_path)
+        loader = LOADERS[ext]
+
+        self.validate_filepath()
         try:
             return loader(self.fs_path, fs=self.repo.fs)
         except ParseError as exc:
@@ -101,29 +150,6 @@ class ParamsDependency(Dependency):
                 f"Unable to read parameters from '{self}'"
             ) from exc
 
-    def read_params_d(self, **kwargs):
-        config = self._read()
-
-        ret = {}
-        for param in self.params:
-            dpath.util.merge(
-                ret,
-                dpath.util.search(config, param, separator="."),
-                separator=".",
-            )
-        return ret
-
-    def read_params(self):
-        config = self._read()
-
-        ret = {}
-        for param in self.params:
-            try:
-                ret[param] = dpath.util.get(config, param, separator=".")
-            except KeyError:
-                pass
-        return ret
-
     def get_hash(self):
         info = self.read_params()
 
@@ -144,12 +170,5 @@ class ParamsDependency(Dependency):
         if not self.isfile and not self.isdir:
             raise self.IsNotFileOrDirError(self)
 
-        if self.is_empty:
-            logger.warning(f"'{self}' is empty.")
-
         self.ignore()
-
-        if self.metric or self.plot:
-            self.verify_metric()
-
         self.hash_info = self.get_hash()

dvc/dependency/repo.py

@@ -5,12 +5,14 @@ from typing import TYPE_CHECKING, Dict, Optional, Set, Tuple
 
 from voluptuous import Required
 
+from dvc.prompt import confirm
+
 from .base import Dependency
 
 if TYPE_CHECKING:
-    from dvc.hash_info import HashInfo
+    from dvc_data.hashfile.file import HashFile
-    from dvc.objects.db.base import ObjectDB
+    from dvc_data.hashfile.hash_info import HashInfo
-    from dvc.objects.file import HashFile
+    from dvc_objects.db import ObjectDB
 
 
 class RepoDependency(Dependency):
@@ -29,7 +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):
@@ -61,7 +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.objects.checkout import checkout
+        from dvc_data.checkout import checkout
 
         for odb, objs in self.get_used_objs().items():
             self.repo.cloud.pull(objs, jobs=jobs, odb=odb)
@@ -72,8 +74,9 @@ class RepoDependency(Dependency):
             to.fs,
             obj,
             self.repo.odb.local,
-            dvcignore=None,
+            ignore=None,
             state=self.repo.state,
+            prompt=confirm,
         )
 
     def update(self, rev=None):
@@ -99,8 +102,9 @@ class RepoDependency(Dependency):
     ) -> Tuple[Dict[Optional["ObjectDB"], Set["HashInfo"]], "HashFile"]:
         from dvc.config import NoRemoteError
         from dvc.exceptions import NoOutputOrStageError, PathMissingError
-        from dvc.objects.stage import stage
+        from dvc.utils import as_posix
-        from dvc.objects.tree import Tree
+        from dvc_data.build import build
+        from dvc_data.objects.tree import Tree, TreeError
 
         local_odb = self.repo.odb.local
         locked = kwargs.pop("locked", True)
@@ -112,11 +116,10 @@ class RepoDependency(Dependency):
             if locked and self.def_repo.get(self.PARAM_REV_LOCK) is None:
                 self.def_repo[self.PARAM_REV_LOCK] = rev
 
-            path = os.path.abspath(os.path.join(repo.root_dir, self.def_path))
             if not obj_only:
                 try:
                     for odb, obj_ids in repo.used_objs(
-                        [path],
+                        [os.path.join(repo.root_dir, self.def_path)],
                         force=True,
                         jobs=kwargs.get("jobs"),
                         recursive=True,
@@ -130,65 +133,65 @@ class RepoDependency(Dependency):
                     pass
 
             try:
-                staging, _, staged_obj = stage(
+                object_store, _, obj = build(
                     local_odb,
-                    path,
+                    as_posix(self.def_path),
-                    repo.repo_fs,
+                    repo.dvcfs,
                     local_odb.fs.PARAM_CHECKSUM,
                 )
-            except FileNotFoundError as exc:
+            except (FileNotFoundError, TreeError) as exc:
                 raise PathMissingError(
                     self.def_path, self.def_repo[self.PARAM_URL]
                 ) from exc
-            staging = copy(staging)
+            object_store = copy(object_store)
-            staging.read_only = True
+            object_store.read_only = True
 
-            self._staged_objs[rev] = staged_obj
+            self._objs[rev] = obj
-            used_obj_ids[staging].add(staged_obj.hash_info)
+            used_obj_ids[object_store].add(obj.hash_info)
-            if isinstance(staged_obj, Tree):
+            if isinstance(obj, Tree):
-                used_obj_ids[staging].update(oid for _, _, oid in staged_obj)
+                used_obj_ids[object_store].update(oid for _, _, oid in obj)
-            return used_obj_ids, staged_obj
+            return used_obj_ids, obj
 
     def _check_circular_import(self, odb, obj_ids):
         from dvc.exceptions import CircularImportError
-        from dvc.fs.repo import RepoFileSystem
+        from dvc.fs.dvc import DvcFileSystem
-        from dvc.objects.db.reference import ReferenceObjectDB
+        from dvc_data.db.reference import ReferenceHashFileDB
-        from dvc.objects.tree import Tree
+        from dvc_data.objects.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():
-            if not isinstance(obj.fs, RepoFileSystem):
+            if not isinstance(obj.fs, DvcFileSystem):
                 continue
             if (
                 obj.fs.repo_url in checked_urls
-                or obj.fs.root_dir in checked_urls
+                or obj.fs.repo.root_dir in checked_urls
             ):
                 continue
             self_url = self.repo.url or self.repo.root_dir
             if (
                 obj.fs.repo_url is not None
                 and obj.fs.repo_url == self_url
-                or obj.fs.root_dir == self.repo.root_dir
+                or obj.fs.repo.root_dir == self.repo.root_dir
             ):
                 raise CircularImportError(self, obj.fs.repo_url, self_url)
-            checked_urls.update([obj.fs.repo_url, obj.fs.root_dir])
+            checked_urls.update([obj.fs.repo_url, obj.fs.repo.root_dir])
 
     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:
+        if rev in self._objs:
-            return self._staged_objs[rev]
+            return self._objs[rev]
         _, obj = self._get_used_and_obj(
             obj_only=True, filter_info=filter_info, **kwargs
         )

dvc/dvcfile.py

@@ -59,7 +59,7 @@ def is_lock_file(path):
 
 
 def is_git_ignored(repo, path):
-    from dvc.fs.local import LocalFileSystem
+    from dvc.fs import LocalFileSystem
     from dvc.scm import NoSCMError
 
     try:
@@ -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)
 

dvc/exceptions.py

@@ -1,4 +1,5 @@
 """Exceptions raised by the dvc."""
+from dvc.utils import format_link
 
 
 class DvcException(Exception):
@@ -128,13 +129,6 @@ class NotDvcRepoError(DvcException):
     """Thrown if a directory is not a DVC repo"""
 
 
-class DvcParserError(DvcException):
-    """Base class for CLI parser errors."""
-
-    def __init__(self):
-        super().__init__("parser error")
-
-
 class CyclicGraphError(DvcException):
     def __init__(self, stages):
         assert isinstance(stages, list)
@@ -157,9 +151,9 @@ class InitError(DvcException):
 
 
 class ReproductionError(DvcException):
-    def __init__(self, dvc_file_name):
+    def __init__(self, name):
-        self.path = dvc_file_name
+        self.name = name
-        super().__init__(f"failed to reproduce '{dvc_file_name}'")
+        super().__init__(f"failed to reproduce '{name}'")
 
 
 class BadMetricError(DvcException):
@@ -208,14 +202,6 @@ class FileMissingError(DvcException):
         )
 
 
-class DvcIgnoreInCollectedDirError(DvcException):
-    def __init__(self, ignore_dirname):
-        super().__init__(
-            ".dvcignore file should not be in collected dir path: "
-            "'{}'".format(ignore_dirname)
-        )
-
-
 class FileTransferError(DvcException):
     _METHOD = "transfer"
 
@@ -294,7 +280,6 @@ class PathMissingError(DvcException):
 
 class RemoteCacheRequiredError(DvcException):
     def __init__(self, scheme, fs_path):
-        from dvc.utils import format_link
 
         super().__init__(
             (
@@ -330,7 +315,6 @@ class MergeError(DvcException):
 
 
 class CacheLinkError(DvcException):
-    from dvc.utils import format_link
 
     SUPPORT_LINK = "See {} for more information.".format(
         format_link(

dvc/external_repo.py

@@ -3,7 +3,7 @@ import os
 import tempfile
 import threading
 from contextlib import contextmanager
-from typing import Dict
+from typing import TYPE_CHECKING, Dict
 
 from funcy import retry, wrap_with
 
@@ -19,6 +19,9 @@ from dvc.repo import Repo
 from dvc.scm import CloneError, map_scm_exception
 from dvc.utils import relpath
 
+if TYPE_CHECKING:
+    from scmrepo import Git
+
 logger = logging.getLogger(__name__)
 
 
@@ -30,6 +33,7 @@ def external_repo(
     from scmrepo.git import Git
 
     from dvc.config import NoRemoteError
+    from dvc.fs import GitFileSystem
 
     logger.debug("Creating external repo %s@%s", url, rev)
     path = _cached_clone(url, rev, for_write=for_write)
@@ -45,14 +49,22 @@ def external_repo(
     config = _get_remote_config(url) if os.path.isdir(url) else {}
     config.update(cache_config)
 
-    root_dir = path if for_write else os.path.realpath(path)
+    if for_write:
+        scm = None
+        root_dir = path
+        fs = None
+    else:
+        scm = Git(path)
+        fs = GitFileSystem(scm=scm, rev=rev)
+        root_dir = "/"
+
     repo_kwargs = dict(
         root_dir=root_dir,
         url=url,
-        scm=None if for_write else Git(root_dir),
+        fs=fs,
-        rev=None if for_write else rev,
         config=config,
-        repo_factory=erepo_factory(url, cache_config),
+        repo_factory=erepo_factory(url, root_dir, cache_config),
+        scm=scm,
         **kwargs,
     )
 
@@ -82,14 +94,16 @@ def external_repo(
             _remove(path)
 
 
-def erepo_factory(url, cache_config, *args, **kwargs):
+def erepo_factory(url, root_dir, cache_config):
-    def make_repo(path, **_kwargs):
+    from dvc.fs import localfs
+
+    def make_repo(path, fs=None, **_kwargs):
         _config = cache_config.copy()
         if os.path.isdir(url):
-            rel = os.path.relpath(path, _kwargs["scm"].root_dir)
+            fs = fs or localfs
-            repo_path = os.path.join(url, rel)
+            repo_path = os.path.join(url, *fs.path.relparts(path, root_dir))
             _config.update(_get_remote_config(repo_path))
-        return Repo(path, config=_config, **_kwargs)
+        return Repo(path, fs=fs, config=_config, **_kwargs)
 
     return make_repo
 
@@ -194,12 +208,12 @@ def _clone_default_branch(url, rev, for_write=False):
                     # given repo URL it is easier/safer for us to work with
                     # full clones in this case.
                     logger.debug("erepo: unshallowing clone for '%s'", url)
-                    _unshallow(git)
+                    _pull(git, unshallow=True)
                     shallow = False
                     CLONES[url] = (clone_path, shallow)
                 else:
                     logger.debug("erepo: git pull '%s'", url)
-                    git.pull()
+                    _pull(git)
         else:
             from dvc.scm import clone
 
@@ -210,12 +224,17 @@ def _clone_default_branch(url, rev, for_write=False):
 
                 try:
                     git = clone(url, clone_path, shallow_branch=rev)
-                    shallow = True
+                    shallow = os.path.exists(
-                    logger.debug(
+                        os.path.join(clone_path, Git.GIT_DIR, "shallow")
-                        "erepo: using shallow clone for branch '%s'", rev
                     )
+                    if shallow:
+                        logger.debug(
+                            "erepo: using shallow clone for branch '%s'", rev
+                        )
                 except CloneError:
-                    pass
+                    git_dir = os.path.join(clone_path, ".git")
+                    if os.path.exists(git_dir):
+                        _remove(git_dir)
             if not git:
                 git = clone(url, clone_path)
                 shallow = False
@@ -227,17 +246,24 @@ def _clone_default_branch(url, rev, for_write=False):
     return clone_path, shallow
 
 
-def _unshallow(git):
+def _pull(git: "Git", unshallow: bool = False):
-    if git.gitpython.repo.head.is_detached:
+    from dvc.repo.experiments.utils import fetch_all_exps
-        # If this is a detached head (i.e. we shallow cloned a tag) switch to
+
-        # the default branch
+    git.fetch(unshallow=unshallow)
-        origin_refs = git.gitpython.repo.remotes["origin"].refs
+    _merge_upstream(git)
-        ref = origin_refs["HEAD"].reference
+    fetch_all_exps(git, "origin")
-        branch_name = ref.name.split("/")[-1]
+
-        branch = git.gitpython.repo.create_head(branch_name, ref)
+
-        branch.set_tracking_branch(ref)
+def _merge_upstream(git: "Git"):
-        branch.checkout()
+    from scmrepo.exceptions import SCMError
-    git.pull(unshallow=True)
+
+    try:
+        branch = git.active_branch()
+        upstream = f"refs/remotes/origin/{branch}"
+        if git.get_ref(upstream):
+            git.merge(upstream)
+    except SCMError:
+        pass
 
 
 def _git_checkout(repo_path, rev):

dvc/fs/__init__.py

@@ -1,41 +1,47 @@
 from urllib.parse import urlparse
 
-from ..scheme import Schemes
+# pylint: disable=unused-import
-from . import utils  # noqa: F401
+from dvc_objects.fs import utils  # noqa: F401
-from .azure import AzureFileSystem
+from dvc_objects.fs import (  # noqa: F401
-from .gdrive import GDriveFileSystem
+    FS_MAP,
-from .gs import GSFileSystem
+    AzureFileSystem,
-from .hdfs import HDFSFileSystem
+    GDriveFileSystem,
-from .http import HTTPFileSystem
+    GSFileSystem,
-from .https import HTTPSFileSystem
+    HDFSFileSystem,
-from .local import LocalFileSystem
+    HTTPFileSystem,
-from .oss import OSSFileSystem
+    HTTPSFileSystem,
-from .s3 import S3FileSystem
+    LocalFileSystem,
-from .ssh import SSHFileSystem
+    MemoryFileSystem,
-from .webdav import WebDAVFileSystem, WebDAVSFileSystem
+    OSSFileSystem,
-from .webhdfs import WebHDFSFileSystem
+    S3FileSystem,
+    Schemes,
+    SSHFileSystem,
+    WebDAVFileSystem,
+    WebDAVSFileSystem,
+    WebHDFSFileSystem,
+    generic,
+    get_fs_cls,
+    system,
+)
+from dvc_objects.fs.base import AnyFSPath, FileSystem  # noqa: F401
+from dvc_objects.fs.errors import (  # noqa: F401
+    AuthError,
+    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
 
-FS_MAP = {
+from .data import DataFileSystem  # noqa: F401
-    Schemes.AZURE: AzureFileSystem,
+from .dvc import DvcFileSystem  # noqa: F401
-    Schemes.GDRIVE: GDriveFileSystem,
+from .git import GitFileSystem  # noqa: F401
-    Schemes.GS: GSFileSystem,
-    Schemes.HDFS: HDFSFileSystem,
-    Schemes.WEBHDFS: WebHDFSFileSystem,
-    Schemes.HTTP: HTTPFileSystem,
-    Schemes.HTTPS: HTTPSFileSystem,
-    Schemes.S3: S3FileSystem,
-    Schemes.SSH: SSHFileSystem,
-    Schemes.OSS: OSSFileSystem,
-    Schemes.WEBDAV: WebDAVFileSystem,
-    Schemes.WEBDAVS: WebDAVSFileSystem,
-    # NOTE: LocalFileSystem is the default
-}
 
-
+# pylint: enable=unused-import
-def get_fs_cls(remote_conf, scheme=None):
-    if not scheme:
-        scheme = urlparse(remote_conf["url"]).scheme
-    return FS_MAP.get(scheme, LocalFileSystem)
 
 
 def get_fs_config(repo, config, **kwargs):
@@ -86,7 +92,7 @@ def _resolve_remote_refs(repo, config, remote_conf):
 
 
 def get_cloud_fs(repo, **kwargs):
-    from dvc.config import ConfigError
+    from dvc.config import ConfigError as RepoConfigError
     from dvc.config_schema import SCHEMA, Invalid
 
     repo_config = repo.config if repo else {}
@@ -96,12 +102,7 @@ def get_cloud_fs(repo, **kwargs):
     try:
         remote_conf = SCHEMA["remote"][str](remote_conf)
     except Invalid as exc:
-        raise ConfigError(str(exc)) from None
+        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")
@@ -114,7 +115,12 @@ def get_cloud_fs(repo, **kwargs):
         remote_conf["gdrive_credentials_tmp_dir"] = repo.tmp_dir
 
     url = remote_conf.pop("url")
-    fs_path = cls._strip_protocol(url)  # pylint:disable=protected-access
+    if issubclass(cls, WebDAVFileSystem):
+        # For WebDAVFileSystem, provided url is the base path itself, so it
+        # should be treated as being a root path.
+        fs_path = cls.root_marker
+    else:
+        fs_path = cls._strip_protocol(url)  # pylint:disable=protected-access
 
     extras = cls._get_kwargs_from_urls(url)  # pylint:disable=protected-access
     conf = {**extras, **remote_conf}  # remote config takes priority

dvc/fs/_metadata.py

@@ -1,73 +0,0 @@
-from dataclasses import dataclass, field
-from typing import TYPE_CHECKING, List
-
-from dvc.output import Output
-
-if TYPE_CHECKING:
-    from dvc.repo import Repo
-
-
-@dataclass
-class Metadata:
-    """
-    Container for storing metadata for a given path, similar in spirit to
-    `os.stat_result`.
-    """
-
-    # required field
-    fs_path: str
-    repo: "Repo"
-
-    # computed fields
-    is_output: bool = field(init=False, default=False)  # is it an output?
-    # is the path part of an output?
-    part_of_output: bool = field(init=False, default=False)
-    # does the path contain outputs?
-    contains_outputs: bool = field(init=False, default=False)
-    # is the path tracked by dvc? equivalent to (is_output or part_of_output)
-    is_dvc: bool = field(init=False, default=False)
-    # does dvc has any output in that path or inside it? equiv. to bool(outs)
-    output_exists: bool = field(init=False)
-
-    # optional fields
-    isdir: bool = False  # is it a directory?
-    is_exec: bool = False  # is it an executable?
-    outs: List[Output] = field(default_factory=list)  # list of outputs
-
-    def __post_init__(self):
-        self.output_exists = bool(self.outs)
-
-        if not self.output_exists:
-            return
-
-        # it can contain multiple outputs
-        if len(self.outs) > 1:
-            self.contains_outputs = True
-        else:
-            out = self.outs[0]
-            # or, the path itself could be an output
-            self.is_output = self.fs_path == out.fs_path
-
-            # or, a directory must have an output somewhere deep inside it
-            if not self.is_output:
-                self.contains_outputs = self.repo.fs.path.isin(
-                    out.fs_path, self.fs_path
-                )
-
-        # or, the path could be a part of an output, i.e. inside of an output
-        self.part_of_output = not self.is_output and not self.contains_outputs
-
-        self.is_dvc = self.is_output or self.part_of_output
-        # if it contains outputs, it must be a dir
-        self.isdir = self.contains_outputs
-
-    @property
-    def isfile(self):
-        # self.isdir might require additional computation, especially if it's
-        # a dir that's inside a dvc-tracked dir
-        # since, `Metadata()` does not support being used as null (i.e. when
-        # path does not exist), `isfile` becomes `isdir`'s complement.
-        return not self.isdir
-
-    def __str__(self):
-        return f"Metadata: {self.fs_path}"

dvc/fs/azure.py

@@ -1,160 +0,0 @@
-import logging
-import os
-import threading
-
-from fsspec.asyn import fsspec_loop
-from fsspec.utils import infer_storage_options
-from funcy import cached_property, memoize, wrap_prop
-
-from dvc.exceptions import DvcException
-from dvc.scheme import Schemes
-from dvc.utils import format_link
-
-from .fsspec_wrapper import CallbackMixin, ObjectFSWrapper
-
-logger = logging.getLogger(__name__)
-_DEFAULT_CREDS_STEPS = (
-    "https://azuresdkdocs.blob.core.windows.net/$web/python/"
-    "azure-identity/1.4.0/azure.identity.html#azure.identity"
-    ".DefaultAzureCredential"
-)
-
-
-class AzureAuthError(DvcException):
-    pass
-
-
-@memoize
-def _az_config():
-    # NOTE: ideally we would've used get_default_cli().config from
-    # azure.cli.core, but azure-cli-core has a lot of conflicts with other
-    # dependencies. So instead we are just use knack directly
-    from knack.config import CLIConfig
-
-    config_dir = os.getenv(
-        "AZURE_CONFIG_DIR", os.path.expanduser(os.path.join("~", ".azure"))
-    )
-    return CLIConfig(config_dir=config_dir, config_env_var_prefix="AZURE")
-
-
-# pylint:disable=abstract-method
-class AzureFileSystem(CallbackMixin, ObjectFSWrapper):
-    scheme = Schemes.AZURE
-    PARAM_CHECKSUM = "etag"
-    REQUIRES = {
-        "adlfs": "adlfs",
-        "knack": "knack",
-        "azure-identity": "azure.identity",
-    }
-
-    @classmethod
-    def _strip_protocol(cls, path: str):
-        opts = infer_storage_options(path)
-        if opts.get("host"):
-            return "{host}{path}".format(**opts)
-
-        return _az_config().get("storage", "container_name", None)
-
-    def unstrip_protocol(self, path: str) -> str:
-        return "azure://" + path.lstrip("/")
-
-    @staticmethod
-    def _get_kwargs_from_urls(urlpath):
-        ops = infer_storage_options(urlpath)
-        if "host" in ops:
-            return {"bucket": ops["host"]}
-        return {}
-
-    def _prepare_credentials(self, **config):
-        from azure.identity.aio import DefaultAzureCredential
-
-        # Disable spam from failed cred types for DefaultAzureCredential
-        logging.getLogger("azure.identity.aio").setLevel(logging.ERROR)
-
-        login_info = {}
-        login_info["connection_string"] = config.get(
-            "connection_string",
-            _az_config().get("storage", "connection_string", None),
-        )
-        login_info["account_name"] = config.get(
-            "account_name", _az_config().get("storage", "account", None)
-        )
-        login_info["account_key"] = config.get(
-            "account_key", _az_config().get("storage", "key", None)
-        )
-        login_info["sas_token"] = config.get(
-            "sas_token", _az_config().get("storage", "sas_token", None)
-        )
-        login_info["tenant_id"] = config.get("tenant_id")
-        login_info["client_id"] = config.get("client_id")
-        login_info["client_secret"] = config.get("client_secret")
-
-        if not (login_info["account_name"] or login_info["connection_string"]):
-            raise AzureAuthError(
-                "Authentication to Azure Blob Storage requires either "
-                "account_name or connection_string.\nLearn more about "
-                "configuration settings at "
-                + format_link("https://man.dvc.org/remote/modify")
-            )
-
-        any_secondary = any(
-            value for key, value in login_info.items() if key != "account_name"
-        )
-        if (
-            login_info["account_name"]
-            and not any_secondary
-            and not config.get("allow_anonymous_login", False)
-        ):
-            with fsspec_loop():
-                login_info["credential"] = DefaultAzureCredential(
-                    exclude_interactive_browser_credential=False,
-                    exclude_environment_credential=config.get(
-                        "exclude_environment_credential", False
-                    ),
-                    exclude_visual_studio_code_credential=config.get(
-                        "exclude_visual_studio_code_credential", False
-                    ),
-                    exclude_shared_token_cache_credential=config.get(
-                        "exclude_shared_token_cache_credential", False
-                    ),
-                    exclude_managed_identity_credential=config.get(
-                        "exclude_managed_identity_credential", False
-                    ),
-                )
-
-        for login_method, required_keys in [  # noqa
-            ("connection string", ["connection_string"]),
-            (
-                "AD service principal",
-                ["tenant_id", "client_id", "client_secret"],
-            ),
-            ("account key", ["account_name", "account_key"]),
-            ("SAS token", ["account_name", "sas_token"]),
-            (
-                f"default credentials ({_DEFAULT_CREDS_STEPS})",
-                ["account_name", "credential"],
-            ),
-            ("anonymous login", ["account_name"]),
-        ]:
-            if all(login_info.get(key) is not None for key in required_keys):
-                break
-        else:
-            login_method = None
-
-        self.login_method = login_method
-        return login_info
-
-    @wrap_prop(threading.Lock())
-    @cached_property
-    def fs(self):
-        from adlfs import AzureBlobFileSystem
-        from azure.core.exceptions import AzureError
-
-        try:
-            return AzureBlobFileSystem(**self.fs_args)
-        except (ValueError, AzureError) as e:
-            raise AzureAuthError(
-                f"Authentication to Azure Blob Storage via {self.login_method}"
-                " failed.\nLearn more about configuration settings at"
-                f" {format_link('https://man.dvc.org/remote/modify')}"
-            ) from e

dvc/fs/base.py

@@ -1,376 +0,0 @@
-import contextlib
-import logging
-import os
-from concurrent.futures import ThreadPoolExecutor, as_completed
-from functools import partialmethod
-from multiprocessing import cpu_count
-from typing import ClassVar, Dict, Optional
-
-from funcy import cached_property
-from tqdm.utils import CallbackIOWrapper
-
-from dvc.exceptions import DvcException
-from dvc.progress import DEFAULT_CALLBACK, FsspecCallback
-from dvc.ui import ui
-from dvc.utils import tmp_fname
-from dvc.utils.fs import makedirs, move
-
-logger = logging.getLogger(__name__)
-
-
-class RemoteCmdError(DvcException):
-    def __init__(self, remote, cmd, ret, err):
-        super().__init__(
-            "{remote} command '{cmd}' finished with non-zero return code"
-            " {ret}': {err}".format(remote=remote, cmd=cmd, ret=ret, err=err)
-        )
-
-
-class RemoteActionNotImplemented(DvcException):
-    def __init__(self, action, scheme):
-        m = f"{action} is not supported for {scheme} remotes"
-        super().__init__(m)
-
-
-class RemoteMissingDepsError(DvcException):
-    pass
-
-
-class FileSystem:
-    sep = "/"
-
-    scheme = "base"
-    REQUIRES: ClassVar[Dict[str, str]] = {}
-    _JOBS = 4 * cpu_count()
-
-    HASH_JOBS = max(1, min(4, cpu_count() // 2))
-    LIST_OBJECT_PAGE_SIZE = 1000
-    TRAVERSE_WEIGHT_MULTIPLIER = 5
-    TRAVERSE_PREFIX_LEN = 3
-    TRAVERSE_THRESHOLD_SIZE = 500000
-    CAN_TRAVERSE = True
-
-    # Needed for some providers, and http open()
-    CHUNK_SIZE = 64 * 1024 * 1024  # 64 MiB
-
-    PARAM_CHECKSUM: ClassVar[Optional[str]] = None
-
-    def __init__(self, **kwargs):
-        self._check_requires(**kwargs)
-
-        self.jobs = kwargs.get("jobs") or self._JOBS
-        self.hash_jobs = kwargs.get("checksum_jobs") or self.HASH_JOBS
-        self._config = kwargs
-
-    @property
-    def config(self):
-        return self._config
-
-    @cached_property
-    def path(self):
-        from .path import Path
-
-        return Path(self.sep)
-
-    @classmethod
-    def _strip_protocol(cls, path: str):
-        return path
-
-    def unstrip_protocol(self, path):
-        return path
-
-    @staticmethod
-    def _get_kwargs_from_urls(urlpath):  # pylint:disable=unused-argument
-        return {}
-
-    @classmethod
-    def get_missing_deps(cls):
-        import importlib
-
-        missing = []
-        for package, module in cls.REQUIRES.items():
-            try:
-                importlib.import_module(module)
-            except ImportError:
-                missing.append(package)
-
-        return missing
-
-    def _check_requires(self, **kwargs):
-        from ..scheme import Schemes
-        from ..utils import format_link
-        from ..utils.pkg import PKG
-
-        missing = self.get_missing_deps()
-        if not missing:
-            return
-
-        url = kwargs.get("url", f"{self.scheme}://")
-
-        scheme = self.scheme
-        if scheme == Schemes.WEBDAVS:
-            scheme = Schemes.WEBDAV
-
-        by_pkg = {
-            "pip": f"pip install 'dvc[{scheme}]'",
-            "conda": f"conda install -c conda-forge dvc-{scheme}",
-        }
-
-        cmd = by_pkg.get(PKG)
-        if cmd:
-            link = format_link("https://dvc.org/doc/install")
-            hint = (
-                f"To install dvc with those dependencies, run:\n"
-                "\n"
-                f"\t{cmd}\n"
-                "\n"
-                f"See {link} for more info."
-            )
-        else:
-            link = format_link("https://github.com/iterative/dvc/issues")
-            hint = f"Please report this bug to {link}. Thank you!"
-
-        raise RemoteMissingDepsError(
-            f"URL '{url}' is supported but requires these missing "
-            f"dependencies: {missing}. {hint}"
-        )
-
-    def checksum(self, path) -> str:
-        raise NotImplementedError
-
-    def open(self, path, mode: str = "r", encoding: str = None, **kwargs):
-        raise RemoteActionNotImplemented("open", self.scheme)
-
-    def exists(self, path) -> bool:
-        raise NotImplementedError
-
-    # pylint: disable=unused-argument
-
-    def isdir(self, path):
-        """Optional: Overwrite only if the remote has a way to distinguish
-        between a directory and a file.
-        """
-        return False
-
-    def isfile(self, path):
-        """Optional: Overwrite only if the remote has a way to distinguish
-        between a directory and a file.
-        """
-        return True
-
-    def isexec(self, path):
-        """Optional: Overwrite only if the remote has a way to distinguish
-        between executable and non-executable file.
-        """
-        return False
-
-    def iscopy(self, path):
-        """Check if this file is an independent copy."""
-        return False  # We can't be sure by default
-
-    def walk(self, top, topdown=True, onerror=None, **kwargs):
-        """Return a generator with (root, dirs, files)."""
-        raise NotImplementedError
-
-    def find(self, path, prefix=None):
-        """Return a generator with `str`s to all the files.
-
-        Optional kwargs:
-            prefix (bool): If true `path` will be treated as a prefix
-                rather than directory path.
-        """
-        raise NotImplementedError
-
-    def is_empty(self, path):
-        return False
-
-    def info(self, path):
-        raise NotImplementedError
-
-    def getsize(self, path):
-        return self.info(path).get("size")
-
-    def remove(self, path):
-        raise RemoteActionNotImplemented("remove", self.scheme)
-
-    def makedirs(self, path, **kwargs):
-        """Optional: Implement only if the remote needs to create
-        directories before copying/linking/moving data
-        """
-
-    def move(self, from_info, to_info):
-        self.copy(from_info, to_info)
-        self.remove(from_info)
-
-    def copy(self, from_info, to_info):
-        raise RemoteActionNotImplemented("copy", self.scheme)
-
-    def symlink(self, from_info, to_info):
-        raise RemoteActionNotImplemented("symlink", self.scheme)
-
-    def hardlink(self, from_info, to_info):
-        raise RemoteActionNotImplemented("hardlink", self.scheme)
-
-    def reflink(self, from_info, to_info):
-        raise RemoteActionNotImplemented("reflink", self.scheme)
-
-    # pylint: enable=unused-argument
-
-    def upload(
-        self,
-        from_info,
-        to_info,
-        total=None,
-        desc=None,
-        callback=None,
-        no_progress_bar=False,
-        **pbar_args,
-    ):
-        is_file_obj = hasattr(from_info, "read")
-        method = "upload_fobj" if is_file_obj else "put_file"
-        if not hasattr(self, method):
-            raise RemoteActionNotImplemented(method, self.scheme)
-
-        if not is_file_obj:
-            from .local import localfs
-
-            desc = desc or localfs.path.name(from_info)
-
-        stack = contextlib.ExitStack()
-        if not callback:
-            pbar = ui.progress(
-                desc=desc,
-                disable=no_progress_bar,
-                bytes=True,
-                total=total or -1,
-                **pbar_args,
-            )
-            stack.enter_context(pbar)
-            callback = pbar.as_callback()
-            if total:
-                callback.set_size(total)
-
-        with stack:
-            if is_file_obj:
-                wrapped = CallbackIOWrapper(
-                    callback.relative_update, from_info
-                )
-                # `size` is used to provide hints to the WebdavFileSystem
-                # for legacy servers.
-                # pylint: disable=no-member
-                return self.upload_fobj(wrapped, to_info, size=total)
-
-            logger.debug("Uploading '%s' to '%s'", from_info, to_info)
-            # pylint: disable=no-member
-            return self.put_file(
-                os.fspath(from_info), to_info, callback=callback
-            )
-
-    def download(
-        self,
-        from_info,
-        to_info,
-        name=None,
-        callback=None,
-        no_progress_bar=False,
-        jobs=None,
-        _only_file=False,
-        **kwargs,
-    ):
-        from .local import localfs
-
-        if not hasattr(self, "get_file"):
-            raise RemoteActionNotImplemented("get_file", self.scheme)
-
-        download_dir = not _only_file and self.isdir(from_info)
-
-        desc = name or localfs.path.name(to_info)
-        stack = contextlib.ExitStack()
-        if not callback:
-            pbar_kwargs = {"unit": "Files"} if download_dir else {}
-            pbar = ui.progress(
-                total=-1,
-                desc="Downloading directory" if download_dir else desc,
-                bytes=not download_dir,
-                disable=no_progress_bar,
-                **pbar_kwargs,
-            )
-            stack.enter_context(pbar)
-            callback = pbar.as_callback()
-
-        with stack:
-            if download_dir:
-                return self._download_dir(
-                    from_info, to_info, callback=callback, jobs=jobs, **kwargs
-                )
-            return self._download_file(from_info, to_info, callback=callback)
-
-    download_file = partialmethod(download, _only_file=True)
-
-    def _download_dir(
-        self,
-        from_info,
-        to_info,
-        callback=DEFAULT_CALLBACK,
-        jobs=None,
-        **kwargs,
-    ):
-        from .local import localfs
-
-        from_infos = list(self.find(from_info, **kwargs))
-        if not from_infos:
-            makedirs(to_info, exist_ok=True)
-            return None
-
-        to_infos = (
-            localfs.path.join(to_info, *self.path.relparts(info, from_info))
-            for info in from_infos
-        )
-        callback.set_size(len(from_infos))
-
-        download_files = FsspecCallback.wrap_fn(callback, self._download_file)
-        max_workers = jobs or self.jobs
-        with ThreadPoolExecutor(max_workers=max_workers) as executor:
-            futures = [
-                executor.submit(download_files, from_info, to_info)
-                for from_info, to_info in zip(from_infos, to_infos)
-            ]
-
-            # NOTE: unlike pulling/fetching cache, where we need to
-            # download everything we can, not raising an error here might
-            # turn very ugly, as the user might think that he has
-            # downloaded a complete directory, while having a partial one,
-            # which might cause unexpected results in his pipeline.
-            for future in as_completed(futures):
-                # NOTE: executor won't let us raise until all futures that
-                # it has are finished, so we need to cancel them ourselves
-                # before re-raising.
-                exc = future.exception()
-                if exc:
-                    for entry in futures:
-                        entry.cancel()
-                    raise exc
-
-    def _download_file(
-        self,
-        from_info,
-        to_info,
-        callback=DEFAULT_CALLBACK,
-    ):
-        from .local import localfs
-
-        makedirs(localfs.path.parent(to_info), exist_ok=True)
-        tmp_file = tmp_fname(to_info)
-
-        logger.debug("Downloading '%s' to '%s'", from_info, to_info)
-        try:
-            # noqa, pylint: disable=no-member
-            self.get_file(from_info, tmp_file, callback=callback)
-        except Exception:  # pylint: disable=broad-except
-            # do we need to rollback makedirs for previously not-existing
-            # directories?
-            with contextlib.suppress(FileNotFoundError):
-                os.unlink(tmp_file)
-            raise
-
-        move(tmp_file, to_info)

dvc/fs/callbacks.py

@@ -0,0 +1,79 @@
+# pylint: disable=unused-import
+from contextlib import ExitStack
+from typing import TYPE_CHECKING, Optional
+
+from funcy import cached_property
+
+from dvc_objects.fs.callbacks import (  # noqa: F401
+    DEFAULT_CALLBACK,
+    Callback,
+    TqdmCallback,
+)
+
+if TYPE_CHECKING:
+    from dvc.ui._rich_progress import RichTransferProgress
+
+
+class RichCallback(Callback):
+    def __init__(
+        self,
+        size: Optional[int] = None,
+        value: int = 0,
+        progress: "RichTransferProgress" = None,
+        desc: str = None,
+        bytes: bool = False,  # pylint: disable=redefined-builtin
+        unit: str = None,
+        disable: bool = False,
+    ) -> None:
+        self._progress = progress
+        self.disable = disable
+        self._task_kwargs = {
+            "description": desc or "",
+            "bytes": bytes,
+            "unit": unit,
+            "total": size or 0,
+            "visible": False,
+            "progress_type": None if bytes else "summary",
+        }
+        self._stack = ExitStack()
+        super().__init__(size=size, value=value)
+
+    @cached_property
+    def progress(self):
+        from dvc.ui import ui
+        from dvc.ui._rich_progress import RichTransferProgress
+
+        if self._progress is not None:
+            return self._progress
+
+        progress = RichTransferProgress(
+            transient=True,
+            disable=self.disable,
+            console=ui.error_console,
+        )
+        return self._stack.enter_context(progress)
+
+    @cached_property
+    def task(self):
+        return self.progress.add_task(**self._task_kwargs)
+
+    def __enter__(self):
+        return self
+
+    def close(self):
+        self.progress.clear_task(self.task)
+        self._stack.close()
+
+    def call(self, hook_name=None, **kwargs):
+        self.progress.update(
+            self.task,
+            completed=self.value,
+            total=self.size,
+            visible=not self.disable,
+        )
+
+    def branch(self, path_1, path_2, kwargs, child: Optional[Callback] = None):
+        child = child or RichCallback(
+            progress=self.progress, desc=path_1, bytes=True
+        )
+        return super().branch(path_1, path_2, kwargs, child=child)

dvc/fs/data.py

@@ -0,0 +1,199 @@
+import logging
+import threading
+import typing
+
+from fsspec import AbstractFileSystem
+from funcy import cached_property, wrap_prop
+
+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.oid_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.oid_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_data.hashfile.meta import Meta
+
+        key = self._get_key(path)
+
+        try:
+            outs = list(self.repo.index.tree.iteritems(key))  # noqa: B301
+        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"
+
+    PARAM_CHECKSUM = "md5"
+
+    def _prepare_credentials(self, **config):
+        return config
+
+    @wrap_prop(threading.Lock())
+    @cached_property
+    def fs(self):
+        return _DataFileSystem(**self.fs_args)
+
+    def isdvc(self, path, **kwargs):
+        return self.fs.isdvc(path, **kwargs)
+
+    @property
+    def repo(self):
+        return self.fs.repo

dvc/fs/dvc.py

@@ -1,275 +1,480 @@
 import logging
+import ntpath
 import os
-import typing
+import posixpath
+import threading
+from contextlib import suppress
+from itertools import takewhile
+from typing import TYPE_CHECKING, Callable, Optional, Tuple, Type, Union
 
-from dvc.exceptions import OutputNotFoundError
+from fsspec.spec import AbstractFileSystem
-from dvc.utils import relpath
+from funcy import cached_property, wrap_prop, wrap_with
 
-from ..progress import DEFAULT_CALLBACK
+from dvc_objects.fs.base import FileSystem
-from ._metadata import Metadata
+from dvc_objects.fs.callbacks import DEFAULT_CALLBACK
-from .base import FileSystem
+from dvc_objects.fs.path import Path
 
-if typing.TYPE_CHECKING:
+from .data import DataFileSystem
-    from dvc.output import Output
+
-    from dvc.types import AnyPath
+if TYPE_CHECKING:
+    from dvc.repo import Repo
 
 logger = logging.getLogger(__name__)
 
+RepoFactory = Union[Callable[[str], "Repo"], Type["Repo"]]
 
-class DvcFileSystem(FileSystem):  # pylint:disable=abstract-method
+
-    """DVC repo fs.
+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
+
+
+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
+
+
+def _merge_info(repo, fs_info, dvc_info):
+    from . import utils
+
+    ret = {"repo": repo}
+
+    if dvc_info:
+        ret["dvc_info"] = dvc_info
+        ret["type"] = dvc_info["type"]
+        ret["size"] = dvc_info["size"]
+        if not fs_info and "md5" in dvc_info:
+            ret["md5"] = dvc_info["md5"]
+
+    if fs_info:
+        ret["type"] = fs_info["type"]
+        ret["size"] = fs_info["size"]
+        isexec = False
+        if fs_info["type"] == "file":
+            isexec = utils.is_exec(fs_info["mode"])
+        ret["isexec"] = isexec
+
+    return ret
+
+
+class _DvcFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
+    """DVC + git-tracked files fs.
 
     Args:
-        repo: DVC repo.
+        repo: DVC or git repo.
+        subrepos: traverse to subrepos (by default, it ignores subrepos)
+        repo_factory: A function to initialize subrepo with, default is Repo.
+        kwargs: Additional keyword arguments passed to the `DataFileSystem()`.
     """
 
-    sep = os.sep
+    root_marker = "/"
 
-    scheme = "local"
+    PARAM_REPO_URL = "repo_url"
-    PARAM_CHECKSUM = "md5"
+    PARAM_REPO_ROOT = "repo_root"
+    PARAM_REV = "rev"
+    PARAM_CACHE_DIR = "cache_dir"
+    PARAM_CACHE_TYPES = "cache_types"
+    PARAM_SUBREPOS = "subrepos"
 
-    def __init__(self, **kwargs):
+    def __init__(
-        super().__init__(**kwargs)
+        self,
-        self.repo = kwargs["repo"]
+        repo: Optional["Repo"] = None,
+        subrepos=False,
+        repo_factory: RepoFactory = None,
+        **kwargs,
+    ):
+        super().__init__()
+
+        from pygtrie import Trie
+
+        if repo is None:
+            repo, repo_factory = self._repo_from_fs_config(
+                subrepos=subrepos, **kwargs
+            )
+
+        if not repo_factory:
+            from dvc.repo import Repo
+
+            self.repo_factory: RepoFactory = Repo
+        else:
+            self.repo_factory = repo_factory
+
+        def _getcwd():
+            relparts = ()
+            if repo.fs.path.isin(repo.fs.path.getcwd(), repo.root_dir):
+                relparts = repo.fs.path.relparts(
+                    repo.fs.path.getcwd(), repo.root_dir
+                )
+            return self.root_marker + self.sep.join(relparts)
+
+        self.path = Path(self.sep, getcwd=_getcwd)
+        self.repo = repo
+        self.hash_jobs = repo.fs.hash_jobs
+        self._traverse_subrepos = subrepos
+
+        self._subrepos_trie = Trie()
+        """Keeps track of each and every path with the corresponding repo."""
+
+        key = self._get_key(self.repo.root_dir)
+        self._subrepos_trie[key] = repo
+
+        self._datafss = {}
+        """Keep a datafs instance of each repo."""
+
+        if hasattr(repo, "dvc_dir"):
+            self._datafss[key] = DataFileSystem(repo=repo)
+
+    def _get_key(self, path):
+        parts = self.repo.fs.path.relparts(path, self.repo.root_dir)
+        if parts == (".",):
+            parts = ()
+        return parts
+
+    @property
+    def repo_url(self):
+        if self.repo is None:
+            return None
+        return self.repo.url
 
     @property
     def config(self):
-        raise NotImplementedError
+        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,
+        }
 
-    def _find_outs(self, path, *args, **kwargs):
+    @classmethod
-        outs = self.repo.find_outs_by_path(path, *args, **kwargs)
+    def _repo_from_fs_config(
+        cls, **config
+    ) -> Tuple["Repo", Optional["RepoFactory"]]:
+        from dvc.external_repo import erepo_factory, external_repo
+        from dvc.repo import Repo
 
-        def _is_cached(out):
+        url = config.get(cls.PARAM_REPO_URL)
-            return out.use_cache
+        root = config.get(cls.PARAM_REPO_ROOT)
+        assert url or root
 
-        outs = list(filter(_is_cached, outs))
+        def _open(*args, **kwargs):
-        if not outs:
+            # NOTE: if original repo was an erepo (and has a URL),
-            raise OutputNotFoundError(path, self.repo)
+            # 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)
 
-        return outs
+        cache_dir = config.get(cls.PARAM_CACHE_DIR)
-
+        cache_config = (
-    def _get_granular_hash(self, path: "AnyPath", out: "Output", remote=None):
+            {}
-        # NOTE: use string paths here for performance reasons
+            if not cache_dir
-        key = tuple(relpath(path, out.fs_path).split(os.sep))
+            else {
-        out.get_dir_cache(remote=remote)
+                "cache": {
-        if out.obj is None:
+                    "dir": cache_dir,
-            raise FileNotFoundError
+                    "type": config.get(cls.PARAM_CACHE_TYPES),
-        (_, oid) = out.obj.trie.get(key) or (None, None)
+                }
-        if oid:
+            }
-            return oid
+        )
-        raise FileNotFoundError
+        repo_kwargs: dict = {
-
+            "rev": config.get(cls.PARAM_REV),
-    def _get_fs_path(self, path: "AnyPath", remote=None):
+            "subrepos": config.get(cls.PARAM_SUBREPOS, False),
-        try:
+            "uninitialized": True,
-            outs = self._find_outs(path, strict=False)
+        }
-        except OutputNotFoundError as exc:
+        factory: Optional["RepoFactory"] = None
-            raise FileNotFoundError from exc
+        if url is None:
-
+            repo_kwargs["config"] = cache_config
-        if len(outs) != 1 or (
-            outs[0].is_dir_checksum and path == outs[0].fs_path
-        ):
-            raise IsADirectoryError
-
-        out = outs[0]
-
-        if not out.hash_info:
-            raise FileNotFoundError
-
-        if out.changed_cache(filter_info=path):
-            from dvc.config import NoRemoteError
-
-            try:
-                remote_odb = self.repo.cloud.get_remote_odb(remote)
-            except NoRemoteError as exc:
-                raise FileNotFoundError from exc
-            if out.is_dir_checksum:
-                checksum = self._get_granular_hash(path, out).value
-            else:
-                checksum = out.hash_info.value
-            remote_fs_path = remote_odb.hash_to_path(checksum)
-            return remote_odb.fs, remote_fs_path
-
-        if out.is_dir_checksum:
-            checksum = self._get_granular_hash(path, out).value
-            cache_path = out.odb.fs.unstrip_protocol(
-                out.odb.hash_to_path(checksum)
-            )
         else:
-            cache_path = out.cache_path
+            repo_kwargs["cache_dir"] = cache_dir
-        return out.odb.fs, cache_path
+            factory = erepo_factory(url, root, cache_config)
 
-    def open(  # type: ignore
+        with _open(
-        self, path: str, mode="r", encoding=None, **kwargs
+            url if url else root,
-    ):  # pylint: disable=arguments-renamed
+            **repo_kwargs,
-        fs, fspath = self._get_fs_path(path, **kwargs)
+        ) as repo:
-        return fs.open(fspath, mode=mode, encoding=encoding)
+            return repo, factory
 
-    def exists(self, path):  # pylint: disable=arguments-renamed
+    def _get_repo(self, path: str) -> "Repo":
-        try:
+        """Returns repo that the path falls in, using prefix.
-            self.metadata(path)
-            return True
-        except FileNotFoundError:
-            return False
 
-    def isdir(self, path):  # pylint: disable=arguments-renamed
+        If the path is already tracked/collected, it just returns the repo.
-        try:
-            meta = self.metadata(path)
-            return meta.isdir
-        except FileNotFoundError:
-            return False
 
-    def check_isdir(self, path, outs):
+        Otherwise, it collects the repos that might be in the path's parents
-        if len(outs) != 1:
+        and then returns the appropriate one.
-            return True
+        """
+        if not self.repo.fs.path.isin_or_eq(path, self.repo.root_dir):
+            # outside of repo
+            return self.repo
 
-        out = outs[0]
+        key = self._get_key(path)
-        if not out.is_dir_checksum:
+        repo = self._subrepos_trie.get(key)
-            return out.fs_path != path
+        if repo:
-        if out.fs_path == path:
+            return repo
-            return True
 
-        try:
+        prefix_key, repo = self._subrepos_trie.longest_prefix(key)
-            self._get_granular_hash(path, out)
+        prefix = self.repo.fs.path.join(
-            return False
+            self.repo.root_dir,
-        except FileNotFoundError:
+            *prefix_key,  # pylint: disable=not-an-iterable
-            return True
-
-    def isfile(self, path):  # pylint: disable=arguments-renamed
-        try:
-            meta = self.metadata(path)
-            return meta.isfile
-        except FileNotFoundError:
-            return False
-
-    def _fetch_dir(self, out, **kwargs):
-        # pull dir cache if needed
-        out.get_dir_cache(**kwargs)
-
-        if not out.obj:
-            raise FileNotFoundError
-
-    def _add_dir(self, trie, out, **kwargs):
-        self._fetch_dir(out, **kwargs)
-
-        base = out.fs.path.parts(out.fs_path)
-        for key, _, _ in out.obj:  # noqa: B301
-            trie[base + key] = None
-
-    def _walk(self, root, trie, topdown=True, **kwargs):
-        dirs = set()
-        files = []
-
-        root_parts = self.path.parts(root)
-        out = trie.get(root_parts)
-        if out and out.is_dir_checksum:
-            self._add_dir(trie, out, **kwargs)
-
-        root_len = len(root_parts)
-        try:
-            for key, out in trie.iteritems(prefix=root_parts):  # noqa: B301
-                if key == root_parts:
-                    continue
-
-                name = key[root_len]
-                if len(key) > root_len + 1 or (out and out.is_dir_checksum):
-                    dirs.add(name)
-                    continue
-
-                files.append(name)
-        except KeyError:
-            pass
-
-        assert topdown
-        dirs = list(dirs)
-        yield root, dirs, files
-
-        for dname in dirs:
-            yield from self._walk(self.path.join(root, dname), trie)
-
-    def walk(self, top, topdown=True, onerror=None, **kwargs):
-        from pygtrie import Trie
-
-        assert topdown
-        root = os.path.abspath(top)
-        try:
-            meta = self.metadata(root)
-        except FileNotFoundError:
-            if onerror is not None:
-                onerror(FileNotFoundError(top))
-            return
-
-        if not meta.isdir:
-            if onerror is not None:
-                onerror(NotADirectoryError(top))
-            return
-
-        trie = Trie()
-        for out in meta.outs:
-            trie[out.fs.path.parts(out.fs_path)] = out
-
-            if out.is_dir_checksum and self.path.isin_or_eq(root, out.fs_path):
-                self._add_dir(trie, out, **kwargs)
-
-        yield from self._walk(root, trie, topdown=topdown, **kwargs)
-
-    def find(self, path, prefix=None):
-        for root, _, files in self.walk(path):
-            for fname in files:
-                # NOTE: os.path.join is ~5.5 times slower
-                yield f"{root}{os.sep}{fname}"
-
-    def isdvc(self, path, recursive=False, strict=True):
-        try:
-            meta = self.metadata(path)
-        except FileNotFoundError:
-            return False
-
-        recurse = recursive or not strict
-        return meta.output_exists if recurse else meta.is_output
-
-    def isexec(self, path):  # pylint: disable=unused-argument
-        return False
-
-    def metadata(self, fs_path):
-        abspath = os.path.abspath(fs_path)
-
-        try:
-            outs = self._find_outs(abspath, strict=False, recursive=True)
-        except OutputNotFoundError as exc:
-            raise FileNotFoundError from exc
-
-        meta = Metadata(fs_path=abspath, outs=outs, repo=self.repo)
-        meta.isdir = meta.isdir or self.check_isdir(meta.fs_path, meta.outs)
-        return meta
-
-    def info(self, path):
-        meta = self.metadata(path)
-        ret = {"type": "directory" if meta.isdir else "file"}
-        if meta.is_output and len(meta.outs) == 1 and meta.outs[0].hash_info:
-            out = meta.outs[0]
-            ret["size"] = out.meta.size
-            ret[out.hash_info.name] = out.hash_info.value
-        elif meta.part_of_output:
-            (out,) = meta.outs
-            key = self.path.parts(self.path.relpath(path, out.fs_path))
-            (obj_meta, oid) = out.obj.trie.get(key) or (None, None)
-            if oid:
-                ret["size"] = obj_meta.size if obj_meta else 0
-                ret[oid.name] = oid.value
-
-        return ret
-
-    def get_file(
-        self, from_info, to_file, callback=DEFAULT_CALLBACK, **kwargs
-    ):
-        fs, path = self._get_fs_path(from_info)
-        fs.get_file(  # pylint: disable=protected-access
-            path, to_file, callback=callback, **kwargs
         )
 
+        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)
+        return self._subrepos_trie.get(key) or self.repo
+
+    @wrap_with(threading.Lock())
+    def _update(self, dirs, starting_repo):
+        """Checks for subrepo in directories and updates them."""
+        repo = starting_repo
+        for d in dirs:
+            key = self._get_key(d)
+            if self._is_dvc_repo(d):
+                repo = self.repo_factory(
+                    d,
+                    fs=self.repo.fs,
+                    scm=self.repo.scm,
+                    repo_factory=self.repo_factory,
+                )
+                self._datafss[key] = DataFileSystem(repo=repo)
+            self._subrepos_trie[key] = repo
+
+    def _is_dvc_repo(self, dir_path):
+        """Check if the directory is a dvc repo."""
+        if not self._traverse_subrepos:
+            return False
+
+        from dvc.repo import Repo
+
+        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],
+    ]:
+        """
+        Returns a pair of fss based on repo the path falls in, using prefix.
+        """
+        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 ""
+        else:
+            dvc_path = None
+
+        return fs, fs_path, dvc_fs, dvc_path
+
+    def open(
+        self, path, mode="r", encoding="utf-8", **kwargs
+    ):  # pylint: disable=arguments-renamed, arguments-differ
+        if "b" in mode:
+            encoding = None
+
+        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
+        try:
+            return fs.open(fs_path, mode=mode, encoding=encoding)
+        except FileNotFoundError:
+            if not dvc_fs:
+                raise
+
+        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)
+        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)
+
+        names = set()
+        if dvc_fs:
+            with suppress(FileNotFoundError):
+                for entry in dvc_fs.ls(dvc_path, detail=False):
+                    names.add(dvc_fs.path.name(entry))
+
+        if not dvc_only and fs:
+            try:
+                for entry in dvcignore.ls(
+                    fs, fs_path, detail=False, ignore_subrepos=ignore_subrepos
+                ):
+                    names.add(fs.path.name(entry))
+            except (FileNotFoundError, NotADirectoryError):
+                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)
+
+        infos = []
+        paths = []
+        for name in names:
+            entry_path = self.path.join(path, name)
+            try:
+                info = self.info(entry_path, ignore_subrepos=ignore_subrepos)
+            except FileNotFoundError:
+                continue
+            infos.append(info)
+            paths.append(entry_path)
+
+        if not detail:
+            return paths
+
+        return infos
+
+    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
+        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
+        ignore_subrepos = kwargs.get("ignore_subrepos", True)
+
+        dvc_info = None
+        if dvc_fs:
+            try:
+                dvc_info = dvc_fs.info(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
+
+        # 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:
+            for parent in fs.path.parents(fs_path):
+                try:
+                    if fs.info(parent)["type"] != "directory":
+                        dvc_info = None
+                        break
+                except FileNotFoundError:
+                    continue
+
+        if not dvc_info and not fs_info:
+            raise FileNotFoundError
+
+        info = _merge_info(repo, fs_info, dvc_info)
+        info["name"] = path
+        return info
+
     def checksum(self, path):
-        info = self.info(path)
+        fs, fs_path, dvc_fs, dvc_path = self._get_fs_pair(path)
-        md5 = info.get("md5")
+
-        if md5:
+        try:
-            return md5
+            return fs.checksum(fs_path)
-        raise NotImplementedError
+        except FileNotFoundError:
+            return dvc_fs.checksum(dvc_path)
+
+
+class DvcFileSystem(FileSystem):
+    protocol = "local"
+    PARAM_CHECKSUM = "md5"
+
+    def _prepare_credentials(self, **config):
+        return config
+
+    @wrap_prop(threading.Lock())
+    @cached_property
+    def fs(self):
+        return _DvcFileSystem(**self.fs_args)
+
+    def isdvc(self, path, **kwargs):
+        return self.fs.isdvc(path, **kwargs)
+
+    @property
+    def path(self):  # pylint: disable=invalid-overridden-method
+        return self.fs.path
+
+    @property
+    def repo(self):
+        return self.fs.repo
+
+    @property
+    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)
+
+        return as_posix(path)

dvc/fs/fsspec_wrapper.py

@@ -1,245 +0,0 @@
-import os
-import shutil
-from typing import IO, TYPE_CHECKING, Any, Dict, Iterator, Optional, overload
-
-from funcy import cached_property
-from tqdm.utils import CallbackIOWrapper
-
-from dvc.progress import DEFAULT_CALLBACK
-
-from .base import FileSystem
-
-FSPath = str
-AnyFSPath = str
-
-if TYPE_CHECKING:
-    from typing_extensions import Literal
-
-
-# An info() entry, might evolve to a TypedDict
-# in the future (e.g for properly type 'size' etc).
-Entry = Dict[str, Any]
-
-
-# pylint: disable=no-member
-class FSSpecWrapper(FileSystem):
-    TRAVERSE_PREFIX_LEN = 2
-
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-        self.fs_args = {"skip_instance_cache": True}
-        self.fs_args.update(self._prepare_credentials(**kwargs))
-
-    @staticmethod
-    def _get_kwargs_from_urls(urlpath: str) -> "Dict[str, Any]":
-        from fsspec.utils import infer_storage_options
-
-        options = infer_storage_options(urlpath)
-        options.pop("path", None)
-        options.pop("protocol", None)
-        return options
-
-    @cached_property
-    def fs(self):
-        raise NotImplementedError
-
-    def _prepare_credentials(
-        self, **config: Dict[str, Any]  # pylint: disable=unused-argument
-    ) -> Dict[str, Any]:
-        """Prepare the arguments for authentication to the
-        host filesystem"""
-        return {}
-
-    def _isdir(self, path: AnyFSPath) -> bool:
-        return self.fs.isdir(path)
-
-    def isdir(self, path: AnyFSPath) -> bool:
-        try:
-            return self._isdir(path)
-        except FileNotFoundError:
-            return False
-
-    def isfile(self, path: AnyFSPath) -> bool:
-        try:
-            return not self._isdir(path)
-        except FileNotFoundError:
-            return False
-
-    def is_empty(self, path: AnyFSPath) -> bool:
-        entry = self.info(path)
-        if entry["type"] == "directory":
-            return not self.fs.ls(path)
-        return entry["size"] == 0
-
-    def open(
-        self,
-        path: AnyFSPath,
-        mode: str = "r",
-        encoding: Optional[str] = None,
-        **kwargs,
-    ) -> "IO":  # pylint: disable=arguments-differ
-        return self.fs.open(path, mode=mode)
-
-    def checksum(self, path: AnyFSPath) -> str:
-        return self.fs.checksum(path)
-
-    def copy(self, from_info: AnyFSPath, to_info: AnyFSPath) -> None:
-        self.makedirs(self.path.parent(to_info))
-        self.fs.copy(from_info, to_info)
-
-    def exists(self, path: AnyFSPath) -> bool:
-        return self.fs.exists(path)
-
-    @overload
-    def ls(
-        self, path: AnyFSPath, detail: "Literal[True]"
-    ) -> "Iterator[Entry]":
-        ...
-
-    @overload
-    def ls(self, path: AnyFSPath, detail: "Literal[False]") -> Iterator[str]:
-        ...
-
-    def ls(self, path, detail=False):
-        yield from self.fs.ls(path, detail=detail)
-
-    def find(self, path, prefix=None):
-        yield from self.fs.find(path)
-
-    def move(self, from_info: AnyFSPath, to_info: AnyFSPath) -> None:
-        self.fs.move(from_info, to_info)
-
-    def remove(self, path: AnyFSPath) -> None:
-        self.fs.rm_file(path)
-
-    def info(self, path: AnyFSPath) -> "Entry":
-        return self.fs.info(path)
-
-    def makedirs(self, path: AnyFSPath, **kwargs) -> None:
-        self.fs.makedirs(path, exist_ok=kwargs.pop("exist_ok", True))
-
-    def put_file(
-        self,
-        from_file: AnyFSPath,
-        to_info: AnyFSPath,
-        callback: Any = DEFAULT_CALLBACK,
-        **kwargs,
-    ) -> None:
-        self.fs.put_file(from_file, to_info, callback=callback, **kwargs)
-        self.fs.invalidate_cache(self.path.parent(to_info))
-
-    def get_file(
-        self,
-        from_info: AnyFSPath,
-        to_info: AnyFSPath,
-        callback: Any = DEFAULT_CALLBACK,
-        **kwargs,
-    ) -> None:
-        self.fs.get_file(from_info, to_info, callback=callback, **kwargs)
-
-    def upload_fobj(self, fobj: IO, to_info: AnyFSPath, **kwargs) -> None:
-        self.makedirs(self.path.parent(to_info))
-        with self.open(to_info, "wb") as fdest:
-            shutil.copyfileobj(
-                fobj, fdest, length=getattr(fdest, "blocksize", None)
-            )
-
-
-# pylint: disable=abstract-method
-class ObjectFSWrapper(FSSpecWrapper):
-    TRAVERSE_PREFIX_LEN = 3
-
-    def makedirs(self, path: AnyFSPath, **kwargs) -> None:
-        # For object storages make this method a no-op. The original
-        # fs.makedirs() method will only check if the bucket exists
-        # and create if it doesn't though we don't want to support
-        # that behavior, and the check will cost some time so we'll
-        # simply ignore all mkdir()/makedirs() calls.
-        return None
-
-    def _isdir(self, path: AnyFSPath) -> bool:
-        # Directory in object storages are interpreted differently
-        # among different fsspec providers, so this logic is a temporary
-        # measure for us to adapt as of now. It checks whether it is a
-        # directory (as in a prefix with contents) or whether it is an empty
-        # file where it's name ends with a forward slash
-
-        entry = self.info(path)
-        return entry["type"] == "directory" or (
-            entry["size"] == 0
-            and entry["type"] == "file"
-            and entry["name"].endswith("/")
-        )
-
-    def find(self, path, prefix=None):
-        if prefix:
-            with_prefix = self.path.parent(path)
-            files = self.fs.find(with_prefix, prefix=self.path.parts(path)[-1])
-        else:
-            with_prefix = path
-            files = self.fs.find(path)
-
-        # When calling find() on a file, it returns the same file in a list.
-        # For object-based storages, the same behavior applies to empty
-        # directories since they are represented as files. This condition
-        # checks whether we should yield an empty list (if it is an empty
-        # directory) or just yield the file itself.
-        if len(files) == 1 and files[0] == with_prefix and self.isdir(path):
-            return None
-
-        yield from files
-
-
-# pylint: disable=arguments-differ
-class NoDirectoriesMixin:
-    def isdir(self, *args, **kwargs):
-        return False
-
-    def isfile(self, *args, **kwargs):
-        return True
-
-    def find(self, *args, **kwargs):
-        raise NotImplementedError
-
-    def walk(self, *args, **kwargs):
-        raise NotImplementedError
-
-    def ls(self, *args, **kwargs):
-        raise NotImplementedError
-
-
-class CallbackMixin:
-    """Provides callback support for the filesystem that don't support yet."""
-
-    def put_file(
-        self,
-        from_file,
-        to_info,
-        callback=DEFAULT_CALLBACK,
-        **kwargs,
-    ):
-        """Add compatibility support for Callback."""
-        # pylint: disable=protected-access
-        self.makedirs(self.path.parent(to_info))
-        size = os.path.getsize(from_file)
-        with open(from_file, "rb") as fobj:
-            callback.set_size(size)
-            wrapped = CallbackIOWrapper(callback.relative_update, fobj)
-            self.upload_fobj(wrapped, to_info)
-            self.fs.invalidate_cache(self.path.parent(to_info))
-
-    def get_file(
-        self,
-        from_info,
-        to_info,
-        callback=DEFAULT_CALLBACK,
-        **kwargs,
-    ):
-        # pylint: disable=protected-access
-        total: int = self.getsize(from_info)
-        if total:
-            callback.set_size(total)
-
-        with self.open(from_info, "rb") as fobj, open(to_info, "wb") as fdest:
-            wrapped = CallbackIOWrapper(callback.relative_update, fobj)
-            shutil.copyfileobj(wrapped, fdest, length=fobj.blocksize)

dvc/fs/gdrive.py

@@ -1,259 +0,0 @@
-import logging
-import os
-import threading
-
-from funcy import cached_property, wrap_prop
-
-from dvc.exceptions import DvcException
-from dvc.scheme import Schemes
-from dvc.utils import format_link, tmp_fname
-
-from .fsspec_wrapper import FSSpecWrapper
-
-logger = logging.getLogger(__name__)
-FOLDER_MIME_TYPE = "application/vnd.google-apps.folder"
-
-
-class GDriveAuthError(DvcException):
-    def __init__(self, cred_location):
-
-        if cred_location:
-            message = (
-                "GDrive remote auth failed with credentials in '{}'.\n"
-                "Backup first, remove or fix them, and run DVC again.\n"
-                "It should do auth again and refresh the credentials.\n\n"
-                "Details:".format(cred_location)
-            )
-        else:
-            message = "Failed to authenticate GDrive remote"
-
-        super().__init__(message)
-
-
-class GDriveFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
-    scheme = Schemes.GDRIVE
-    PARAM_CHECKSUM = "checksum"
-    REQUIRES = {"pydrive2": "pydrive2"}
-    # Always prefer traverse for GDrive since API usage quotas are a concern.
-    TRAVERSE_WEIGHT_MULTIPLIER = 1
-    TRAVERSE_PREFIX_LEN = 2
-
-    GDRIVE_CREDENTIALS_DATA = "GDRIVE_CREDENTIALS_DATA"
-    DEFAULT_USER_CREDENTIALS_FILE = "gdrive-user-credentials.json"
-
-    DEFAULT_GDRIVE_CLIENT_ID = "710796635688-iivsgbgsb6uv1fap6635dhvuei09o66c.apps.googleusercontent.com"  # noqa: E501
-    DEFAULT_GDRIVE_CLIENT_SECRET = "a1Fz59uTpVNeG_VGuSKDLJXv"
-
-    def __init__(self, **config):
-        from fsspec.utils import infer_storage_options
-
-        super().__init__(**config)
-
-        self.url = config["url"]
-        opts = infer_storage_options(self.url)
-
-        if not opts["host"]:
-            raise DvcException(
-                "Empty GDrive URL '{}'. Learn more at {}".format(
-                    config["url"],
-                    format_link("https://man.dvc.org/remote/add"),
-                )
-            )
-
-        self._bucket = opts["host"]
-        self._path = opts["path"].lstrip("/")
-        self._trash_only = config.get("gdrive_trash_only")
-        self._use_service_account = config.get("gdrive_use_service_account")
-        self._service_account_user_email = config.get(
-            "gdrive_service_account_user_email"
-        )
-        self._service_account_json_file_path = config.get(
-            "gdrive_service_account_json_file_path"
-        )
-        self._client_id = config.get("gdrive_client_id")
-        self._client_secret = config.get("gdrive_client_secret")
-        self._validate_config()
-
-        tmp_dir = config["gdrive_credentials_tmp_dir"]
-        assert tmp_dir
-
-        self._gdrive_service_credentials_path = tmp_fname(
-            os.path.join(tmp_dir, "")
-        )
-        self._gdrive_user_credentials_path = (
-            tmp_fname(os.path.join(tmp_dir, ""))
-            if os.getenv(GDriveFileSystem.GDRIVE_CREDENTIALS_DATA)
-            else config.get(
-                "gdrive_user_credentials_file",
-                os.path.join(tmp_dir, self.DEFAULT_USER_CREDENTIALS_FILE),
-            )
-        )
-
-    @classmethod
-    def _strip_protocol(cls, path):
-        from fsspec.utils import infer_storage_options
-
-        opts = infer_storage_options(path)
-
-        return "{host}{path}".format(**opts)
-
-    def unstrip_protocol(self, path):
-        return f"gdrive://{path}"
-
-    @staticmethod
-    def _get_kwargs_from_urls(urlpath):
-        return {"url": urlpath}
-
-    def _validate_config(self):
-        # Validate Service Account configuration
-        if (
-            self._use_service_account
-            and not self._service_account_json_file_path
-        ):
-            raise DvcException(
-                "To use service account, set "
-                "`gdrive_service_account_json_file_path`, and optionally"
-                "`gdrive_service_account_user_email` in DVC config\n"
-                "Learn more at {}".format(
-                    format_link("https://man.dvc.org/remote/modify")
-                )
-            )
-
-        # Validate OAuth 2.0 Client ID configuration
-        if not self._use_service_account:
-            if bool(self._client_id) != bool(self._client_secret):
-                raise DvcException(
-                    "Please specify GDrive's client ID and secret in "
-                    "DVC config or omit both to use the defaults.\n"
-                    "Learn more at {}".format(
-                        format_link("https://man.dvc.org/remote/modify")
-                    )
-                )
-
-    @cached_property
-    def credentials_location(self):
-        """
-        Helper to determine where will GDrive remote read credentials from.
-        Useful for tests, exception messages, etc. Returns either env variable
-        name if it's set or actual path to the credentials file.
-        """
-        if os.getenv(GDriveFileSystem.GDRIVE_CREDENTIALS_DATA):
-            return GDriveFileSystem.GDRIVE_CREDENTIALS_DATA
-        if os.path.exists(self._gdrive_user_credentials_path):
-            return self._gdrive_user_credentials_path
-        return None
-
-    @staticmethod
-    def _validate_credentials(auth, settings):
-        """
-        Detects discrepancy in DVC config and cached credentials file.
-        Usually happens when a second remote is added and it is using
-        the same credentials default file. Or when someones decides to change
-        DVC config client id or secret but forgets to remove the cached
-        credentials file.
-        """
-        if not os.getenv(GDriveFileSystem.GDRIVE_CREDENTIALS_DATA):
-            if (
-                settings["client_config"]["client_id"]
-                != auth.credentials.client_id
-                or settings["client_config"]["client_secret"]
-                != auth.credentials.client_secret
-            ):
-                logger.warning(
-                    "Client ID and secret configured do not match the "
-                    "actual ones used\nto access the remote. Do you "
-                    "use multiple GDrive remotes and forgot to\nset "
-                    "`gdrive_user_credentials_file` for one or more of them? "
-                    "Learn more at\n{}.\n".format(
-                        format_link("https://man.dvc.org/remote/modify")
-                    )
-                )
-
-    @wrap_prop(threading.RLock())
-    @cached_property
-    def fs(self):
-        from pydrive2.auth import GoogleAuth
-        from pydrive2.fs import GDriveFileSystem as _GDriveFileSystem
-
-        temporary_save_path = self._gdrive_user_credentials_path
-        is_credentials_temp = os.getenv(
-            GDriveFileSystem.GDRIVE_CREDENTIALS_DATA
-        )
-        if self._use_service_account:
-            temporary_save_path = self._gdrive_service_credentials_path
-
-        if is_credentials_temp:
-            with open(temporary_save_path, "w", encoding="utf-8") as cred_file:
-                cred_file.write(
-                    os.getenv(GDriveFileSystem.GDRIVE_CREDENTIALS_DATA)
-                )
-
-        auth_settings = {
-            "client_config_backend": "settings",
-            "save_credentials": True,
-            "save_credentials_backend": "file",
-            "save_credentials_file": self._gdrive_user_credentials_path,
-            "get_refresh_token": True,
-            "oauth_scope": [
-                "https://www.googleapis.com/auth/drive",
-                "https://www.googleapis.com/auth/drive.appdata",
-            ],
-        }
-
-        if self._use_service_account:
-            auth_settings["service_config"] = {
-                "client_user_email": self._service_account_user_email,
-                "client_json_file_path": self._service_account_json_file_path,
-            }
-            if is_credentials_temp:
-                auth_settings["service_config"][
-                    "client_json_file_path"
-                ] = temporary_save_path
-
-        else:
-            auth_settings["client_config"] = {
-                "client_id": self._client_id or self.DEFAULT_GDRIVE_CLIENT_ID,
-                "client_secret": self._client_secret
-                or self.DEFAULT_GDRIVE_CLIENT_SECRET,
-                "auth_uri": "https://accounts.google.com/o/oauth2/auth",
-                "token_uri": "https://oauth2.googleapis.com/token",
-                "revoke_uri": "https://oauth2.googleapis.com/revoke",
-                "redirect_uri": "",
-            }
-
-        GoogleAuth.DEFAULT_SETTINGS.update(auth_settings)
-
-        # Pass non existent settings path to force DEFAULT_SETTINGS loadings
-        gauth = GoogleAuth(settings_file="")
-
-        try:
-            logger.debug(
-                "GDrive remote auth with config '{}'.".format(
-                    GoogleAuth.DEFAULT_SETTINGS
-                )
-            )
-            if self._use_service_account:
-                gauth.ServiceAuth()
-            else:
-                gauth.CommandLineAuth()
-                GDriveFileSystem._validate_credentials(gauth, auth_settings)
-
-        # Handle AuthenticationError, RefreshError and other auth failures
-        # It's hard to come up with a narrow exception, since PyDrive throws
-        # a lot of different errors - broken credentials file, refresh token
-        # expired, flow failed, etc.
-        except Exception as exc:
-            raise GDriveAuthError(self.credentials_location) from exc
-        finally:
-            if is_credentials_temp:
-                os.remove(temporary_save_path)
-
-        return _GDriveFileSystem(
-            f"{self._bucket}/{self._path}",
-            gauth,
-            trash_only=self._trash_only,
-        )
-
-    def upload_fobj(self, fobj, to_info, **kwargs):
-        self.makedirs(os.path.dirname(to_info))
-        return self.fs.upload_fobj(fobj, to_info, **kwargs)

dvc/fs/git.py

@@ -1,10 +1,9 @@
-import os
 import threading
-from typing import TYPE_CHECKING, Any, Callable
+from typing import TYPE_CHECKING, Any
 
 from funcy import cached_property, wrap_prop
 
-from .fsspec_wrapper import AnyFSPath, FSSpecWrapper
+from . import FileSystem
 
 if TYPE_CHECKING:
     from scmrepo.fs import GitFileSystem as FsspecGitFileSystem
@@ -12,11 +11,10 @@ if TYPE_CHECKING:
     from scmrepo.git.objects import GitTrie
 
 
-class GitFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
+class GitFileSystem(FileSystem):  # pylint:disable=abstract-method
     """Proxies the repo file access methods to Git objects"""
 
-    sep = os.sep
+    protocol = "local"
-    scheme = "local"
 
     def __init__(
         self,
@@ -47,26 +45,13 @@ class GitFileSystem(FSSpecWrapper):  # pylint:disable=abstract-method
 
         return FsspecGitFileSystem(**self.fs_args)
 
+    @cached_property
+    def path(self):
+        return self.fs.path
+
     @property
     def rev(self) -> str:
         return self.fs.rev
 
-    def isexec(self, path: AnyFSPath) -> bool:
+    def ls(self, path, detail=True, **kwargs):
-        from dvc.utils import is_exec
+        return self.fs.ls(path, detail=detail, **kwargs) or []
-
-        try:
-            return is_exec(self.info(path)["mode"])
-        except FileNotFoundError:
-            return False
-
-    def isfile(self, path: AnyFSPath) -> bool:
-        return self.fs.isfile(path)
-
-    def walk(
-        self,
-        top: AnyFSPath,
-        topdown: bool = True,
-        onerror: Callable[[OSError], None] = None,
-        **kwargs: Any,
-    ):
-        return self.fs.walk(top, topdown=topdown, onerror=onerror, **kwargs)

dvc/fs/gs.py

@@ -1,27 +0,0 @@
-import threading
-
-from funcy import cached_property, wrap_prop
-
-from dvc.scheme import Schemes
-
-# pylint:disable=abstract-method
-from .fsspec_wrapper import CallbackMixin, ObjectFSWrapper
-
-
-class GSFileSystem(CallbackMixin, ObjectFSWrapper):
-    scheme = Schemes.GS
-    REQUIRES = {"gcsfs": "gcsfs"}
-    PARAM_CHECKSUM = "etag"
-
-    def _prepare_credentials(self, **config):
-        login_info = {"consistency": None}
-        login_info["project"] = config.get("projectname")
-        login_info["token"] = config.get("credentialpath")
-        return login_info
-
-    @wrap_prop(threading.Lock())
-    @cached_property
-    def fs(self):
-        from gcsfs import GCSFileSystem
-
-        return GCSFileSystem(**self.fs_args)