Files
dvc/tests/func/test_import.py
T
Saugat PachhaiandRuslan Kuprieiev 65854c4946 refactor: extract stage's relationship with Dvcfile out from Stage (#3619)
* refactor stage into dvcfile

* Split create_stage out of dvcfile to Stage::create() and overwrite

* Remove logging from Repo::run

* tests: stop using --remove-outs

Co-authored-by: Ruslan Kuprieiev <[email protected]>
2020-04-12 23:09:29 +03:00

325 lines
10 KiB
Python

import filecmp
import os
from dvc.compat import fspath
import pytest
from mock import patch
from dvc.cache import Cache
from dvc.exceptions import DownloadError, PathMissingError
from dvc.config import NoRemoteError
from dvc.dvcfile import Dvcfile
from dvc.system import System
from dvc.utils.fs import makedirs, remove
import dvc.data_cloud as cloud
from tests.utils import trees_equal
def test_import(tmp_dir, scm, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
stage = dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
assert os.path.isfile("foo_imported")
assert (tmp_dir / "foo_imported").read_text() == "foo content"
assert scm.repo.git.check_ignore("foo_imported")
assert stage.deps[0].def_repo == {
"url": fspath(erepo_dir),
"rev_lock": erepo_dir.scm.get_rev(),
}
@pytest.mark.parametrize("src_is_dvc", [True, False])
def test_import_git_file(tmp_dir, scm, dvc, git_dir, src_is_dvc):
if src_is_dvc:
git_dir.init(dvc=True)
git_dir.scm_gen("src", "hello", commit="add a git file")
stage = tmp_dir.dvc.imp(fspath(git_dir), "src", "dst")
assert (tmp_dir / "dst").read_text() == "hello"
assert tmp_dir.scm.repo.git.check_ignore(fspath(tmp_dir / "dst"))
assert stage.deps[0].def_repo == {
"url": fspath(git_dir),
"rev_lock": git_dir.scm.get_rev(),
}
def test_import_cached_file(erepo_dir, tmp_dir, dvc, scm, monkeypatch):
src = "some_file"
dst = "some_file_imported"
with erepo_dir.chdir():
erepo_dir.dvc_gen({src: "hello"}, commit="add a regular file")
tmp_dir.dvc_gen({dst: "hello"})
(tmp_dir / dst).unlink()
remote_exception = NoRemoteError("dvc import")
with patch.object(cloud.DataCloud, "pull", side_effect=remote_exception):
tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
assert (tmp_dir / dst).is_file()
assert filecmp.cmp(
fspath(erepo_dir / src), fspath(tmp_dir / dst), shallow=False
)
@pytest.mark.parametrize("src_is_dvc", [True, False])
def test_import_git_dir(tmp_dir, scm, dvc, git_dir, src_is_dvc):
if src_is_dvc:
git_dir.init(dvc=True)
git_dir.scm_gen({"src": {"file.txt": "hello"}}, commit="add a dir")
stage = dvc.imp(fspath(git_dir), "src", "dst")
assert (tmp_dir / "dst").is_dir()
trees_equal(fspath(git_dir / "src"), fspath(tmp_dir / "dst"))
assert tmp_dir.scm.repo.git.check_ignore(fspath(tmp_dir / "dst"))
assert stage.deps[0].def_repo == {
"url": fspath(git_dir),
"rev_lock": git_dir.scm.get_rev(),
}
def test_import_dir(tmp_dir, scm, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen({"dir": {"foo": "foo content"}}, commit="create dir")
stage = dvc.imp(fspath(erepo_dir), "dir", "dir_imported")
assert os.path.isdir("dir_imported")
trees_equal(fspath(erepo_dir / "dir"), "dir_imported")
assert scm.repo.git.check_ignore("dir_imported")
assert stage.deps[0].def_repo == {
"url": fspath(erepo_dir),
"rev_lock": erepo_dir.scm.get_rev(),
}
def test_import_file_from_dir(tmp_dir, scm, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen(
{
"dir": {
"1": "1",
"2": "2",
"subdir": {"foo": "foo", "bar": "bar"},
}
},
commit="create dir",
)
stage = dvc.imp(fspath(erepo_dir), os.path.join("dir", "1"))
assert (tmp_dir / "1").read_text() == "1"
assert scm.repo.git.check_ignore("1")
assert stage.deps[0].def_repo == {
"url": fspath(erepo_dir),
"rev_lock": erepo_dir.scm.get_rev(),
}
dvc.imp(fspath(erepo_dir), os.path.join("dir", "2"), out="file")
assert (tmp_dir / "file").read_text() == "2"
assert (tmp_dir / "file.dvc").exists()
dvc.imp(fspath(erepo_dir), os.path.join("dir", "subdir"))
assert (tmp_dir / "subdir" / "foo").read_text() == "foo"
assert (tmp_dir / "subdir" / "bar").read_text() == "bar"
assert (tmp_dir / "subdir.dvc").exists()
dvc.imp(fspath(erepo_dir), os.path.join("dir", "subdir", "foo"), out="X")
assert (tmp_dir / "X").read_text() == "foo"
assert (tmp_dir / "X.dvc").exists()
def test_import_non_cached(erepo_dir, tmp_dir, dvc, scm):
src = "non_cached_output"
dst = src + "_imported"
with erepo_dir.chdir():
erepo_dir.dvc.run(
cmd="echo hello > {}".format(src), outs_no_cache=[src],
)
erepo_dir.scm_add([fspath(erepo_dir / src)], commit="add a non-cached out")
stage = tmp_dir.dvc.imp(fspath(erepo_dir), src, dst)
assert (tmp_dir / dst).is_file()
assert filecmp.cmp(
fspath(erepo_dir / src), fspath(tmp_dir / dst), shallow=False
)
assert tmp_dir.scm.repo.git.check_ignore(dst)
assert stage.deps[0].def_repo == {
"url": fspath(erepo_dir),
"rev_lock": erepo_dir.scm.get_rev(),
}
def test_import_rev(tmp_dir, scm, dvc, erepo_dir):
rev = None
with erepo_dir.chdir(), erepo_dir.branch("branch", new=True):
erepo_dir.dvc_gen("foo", "foo content", commit="create foo on branch")
rev = erepo_dir.scm.get_rev()
stage = dvc.imp(fspath(erepo_dir), "foo", "foo_imported", rev="branch")
assert (tmp_dir / "foo_imported").read_text() == "foo content"
assert scm.repo.git.check_ignore("foo_imported")
assert stage.deps[0].def_repo == {
"url": fspath(erepo_dir),
"rev": "branch",
"rev_lock": rev,
}
def test_pull_imported_stage(tmp_dir, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
dst_stage = Dvcfile(dvc, "foo_imported.dvc").load()
dst_cache = dst_stage.outs[0].cache_path
remove("foo_imported")
remove(dst_cache)
dvc.pull(["foo_imported.dvc"])
assert os.path.isfile("foo_imported")
assert os.path.isfile(dst_cache)
def test_cache_type_is_properly_overridden(tmp_dir, scm, dvc, erepo_dir):
with erepo_dir.chdir():
with erepo_dir.dvc.config.edit() as conf:
conf["cache"]["type"] = "symlink"
erepo_dir.dvc.cache = Cache(erepo_dir.dvc)
erepo_dir.scm_add(
[erepo_dir.dvc.config.files["repo"]],
"set source repo cache type to symlink",
)
erepo_dir.dvc_gen("foo", "foo content", "create foo")
assert System.is_symlink(erepo_dir / "foo")
dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
assert not System.is_symlink("foo_imported")
assert (tmp_dir / "foo_imported").read_text() == "foo content"
assert scm.repo.git.check_ignore("foo_imported")
def test_pull_imported_directory_stage(tmp_dir, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen({"dir": {"foo": "foo content"}}, commit="create dir")
dvc.imp(fspath(erepo_dir), "dir", "dir_imported")
remove("dir_imported")
remove(dvc.cache.local.cache_dir)
dvc.pull(["dir_imported.dvc"])
assert os.path.isdir("dir_imported")
trees_equal(fspath(erepo_dir / "dir"), "dir_imported")
def test_download_error_pulling_imported_stage(tmp_dir, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
dst_stage = Dvcfile(dvc, "foo_imported.dvc").load()
dst_cache = dst_stage.outs[0].cache_path
remove("foo_imported")
remove(dst_cache)
with patch(
"dvc.remote.RemoteLOCAL._download", side_effect=Exception
), pytest.raises(DownloadError):
dvc.pull(["foo_imported.dvc"])
@pytest.mark.parametrize("dname", [".", "dir", "dir/subdir"])
def test_import_to_dir(dname, tmp_dir, dvc, erepo_dir):
makedirs(dname, exist_ok=True)
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
stage = dvc.imp(fspath(erepo_dir), "foo", dname)
dst = os.path.join(dname, "foo")
assert stage.outs[0].fspath == os.path.abspath(dst)
assert os.path.isdir(dname)
assert (tmp_dir / dst).read_text() == "foo content"
def test_pull_non_workspace(tmp_dir, scm, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "master content", commit="create foo")
with erepo_dir.branch("branch", new=True):
erepo_dir.dvc_gen("foo", "branch content", commit="modify foo")
stage = dvc.imp(fspath(erepo_dir), "foo", "foo_imported", rev="branch")
tmp_dir.scm_add([stage.relpath], commit="imported branch")
scm.tag("ref-to-branch")
# Overwrite via import
dvc.imp(fspath(erepo_dir), "foo", "foo_imported", rev="master")
remove(stage.outs[0].cache_path)
dvc.fetch(all_tags=True)
assert os.path.exists(stage.outs[0].cache_path)
def test_import_non_existing(erepo_dir, tmp_dir, dvc):
with pytest.raises(PathMissingError):
tmp_dir.dvc.imp(fspath(erepo_dir), "invalid_output")
# https://github.com/iterative/dvc/pull/2837#discussion_r352123053
with pytest.raises(PathMissingError):
tmp_dir.dvc.imp(fspath(erepo_dir), "/root/", "root")
def test_pull_no_rev_lock(erepo_dir, tmp_dir, dvc):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "contents", commit="create foo")
stage = dvc.imp(fspath(erepo_dir), "foo", "foo_imported")
assert "rev" not in stage.deps[0].def_repo
stage.deps[0].def_repo.pop("rev_lock")
Dvcfile(dvc, stage.path).dump(stage)
remove(stage.outs[0].cache_path)
(tmp_dir / "foo_imported").unlink()
dvc.pull([stage.path])
assert (tmp_dir / "foo_imported").is_file()
assert (tmp_dir / "foo_imported").read_text() == "contents"
def test_import_from_bare_git_repo(tmp_dir, make_tmp_dir, erepo_dir):
import git
git.Repo.init(fspath(tmp_dir), bare=True)
with erepo_dir.chdir():
erepo_dir.dvc_gen({"foo": "foo"}, commit="initial")
erepo_dir.dvc.push()
erepo_dir.scm.repo.create_remote("origin", fspath(tmp_dir))
erepo_dir.scm.repo.remote("origin").push("master")
dvc_repo = make_tmp_dir("dvc-repo", scm=True, dvc=True)
with dvc_repo.chdir():
dvc_repo.dvc.imp(fspath(tmp_dir), "foo")