def test_get_hash_dirty_file(tmp_dir, dvc):
from dvc_data import check
from dvc_data.hashfile.hash import hash_file
tmp_dir.dvc_gen("file", "file")
file_hash_info = HashInfo("md5", "8c7dd922ad47494fc02c388e12c00eac")
(tmp_dir / "file").write_text("something")
something_hash_info = HashInfo("md5", "437b930db84b8079c2dd804a71936b5f")
# file is modified in workspace
# hash_file(file) should return workspace hash, not DVC cached hash
fs = DvcFileSystem(repo=dvc)
assert fs.info("file").get("md5") is None
staging, _, obj = stage(dvc.odb.local, "file", fs, "md5")
assert obj.hash_info == something_hash_info
check(staging, obj)
# hash_file(file) should return DVC cached hash
(tmp_dir / "file").unlink()
assert fs.info("file")["md5"] == file_hash_info.value
_, hash_info = hash_file("file", fs, "md5", state=dvc.state)
assert hash_info == file_hash_info
# tmp_dir/file can be staged even though it is missing in workspace since
# repofs will use the DVC cached hash (and refer to the local cache object)
_, _, obj = stage(dvc.odb.local, "file", fs, "md5")
assert obj.hash_info == file_hash_info
def test_get_hash_cached_file(tmp_dir, dvc, mocker):
tmp_dir.dvc_gen({"foo": "foo"})
fs = DvcFileSystem(repo=dvc)
expected = "acbd18db4cc2f85cedef654fccc4a4d8"
assert fs.info("foo").get("md5") is None
_, _, obj = stage(dvc.odb.local, "foo", fs, "md5")
assert obj.hash_info == HashInfo("md5", expected)
(tmp_dir / "foo").unlink()
assert fs.info("foo")["md5"] == expected
def test_get_hash_granular(tmp_dir, dvc):
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DvcFileSystem(repo=dvc)
subdir = "dir/subdir"
assert fs.info(subdir).get("md5") is None
_, _, obj = stage(dvc.odb.local, subdir, fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo(
"md5", "af314506f1622d107e0ed3f14ec1a3b5.dir"
)
data = posixpath.join(subdir, "data")
assert fs.info(data)["md5"] == "8d777f385d3dfec8815d20f7496026dc"
_, _, obj = stage(dvc.odb.local, data, fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo("md5", "8d777f385d3dfec8815d20f7496026dc")
def test_get_hash_file(tmp_dir, dvc):
tmp_dir.dvc_gen({"foo": "foo"})
fs = DvcFileSystem(repo=dvc)
assert (
fs.info((tmp_dir / "foo").fs_path)["md5"]
== "acbd18db4cc2f85cedef654fccc4a4d8"
)
def test_get_hash_granular(tmp_dir, dvc):
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DvcFileSystem(repo=dvc)
subdir = PathInfo(tmp_dir) / "dir" / "subdir"
assert fs.info(subdir).get("md5") is None
assert stage(dvc.odb.local, subdir, fs, "md5").hash_info == HashInfo(
"md5", "af314506f1622d107e0ed3f14ec1a3b5.dir",
)
assert (
fs.info(subdir / "data")["md5"] == "8d777f385d3dfec8815d20f7496026dc"
)
assert stage(
dvc.odb.local, subdir / "data", fs, "md5"
).hash_info == HashInfo("md5", "8d777f385d3dfec8815d20f7496026dc",)
def test_get_hash_dirty_dir(tmp_dir, dvc):
tmp_dir.dvc_gen({"dir": {"foo": "foo", "bar": "bar"}})
(tmp_dir / "dir" / "baz").write_text("baz")
fs = DvcFileSystem(repo=dvc)
expected = "5ea40360f5b4ec688df672a4db9c17d1.dir"
assert fs.info("dir").get("md5") == expected
_, _, obj = stage(dvc.odb.local, "dir", fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo("md5", expected)
def test_get_hash_dirty_file(tmp_dir, dvc):
tmp_dir.dvc_gen("file", "file")
(tmp_dir / "file").write_text("something")
fs = DvcFileSystem(repo=dvc)
expected = "8c7dd922ad47494fc02c388e12c00eac"
assert fs.info("file").get("md5") == expected
_, _, obj = stage(dvc.odb.local, "file", fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo("md5", expected)
def test_get_hash_dirty_dir(tmp_dir, dvc):
tmp_dir.dvc_gen({"dir": {"foo": "foo", "bar": "bar"}})
(tmp_dir / "dir" / "baz").write_text("baz")
fs = DvcFileSystem(dvc)
expected = "5ea40360f5b4ec688df672a4db9c17d1.dir"
assert fs.info(PathInfo(tmp_dir) / "dir").get("md5") == expected
assert get_hash(PathInfo(tmp_dir) / "dir", fs,
"md5") == HashInfo("md5", expected)
def test_get_hash_dirty_file(tmp_dir, dvc):
tmp_dir.dvc_gen("file", "file")
(tmp_dir / "file").write_text("something")
fs = DvcFileSystem(dvc)
expected = "8c7dd922ad47494fc02c388e12c00eac"
assert fs.info(PathInfo(tmp_dir) / "file").get("md5") == expected
assert get_hash(PathInfo(tmp_dir) / "file", fs,
"md5") == HashInfo("md5", expected)
def test_get_hash_cached_dir(tmp_dir, dvc, mocker):
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DvcFileSystem(repo=dvc)
expected = "8761c4e9acad696bee718615e23e22db.dir"
assert fs.info("dir").get("md5") is None
_, _, obj = stage(dvc.odb.local, "dir", fs, "md5")
assert obj.hash_info == HashInfo(
"md5", "8761c4e9acad696bee718615e23e22db.dir"
)
shutil.rmtree(tmp_dir / "dir")
assert fs.info("dir")["md5"] == expected
_, _, obj = stage(dvc.odb.local, "dir", fs, "md5")
assert obj.hash_info == HashInfo(
"md5", "8761c4e9acad696bee718615e23e22db.dir"
)
def test_get_hash_dir(tmp_dir, dvc, mocker):
import dvc as dvc_module
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DvcFileSystem(repo=dvc)
get_file_hash_spy = mocker.spy(dvc_module.data.stage, "get_file_hash")
assert fs.info("dir")["md5"] == "8761c4e9acad696bee718615e23e22db.dir"
assert not get_file_hash_spy.called
def test_get_hash_cached_granular(tmp_dir, dvc, mocker):
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DvcFileSystem(repo=dvc)
subdir = "dir/subdir"
assert fs.info(subdir).get("md5") is None
_, _, obj = stage(dvc.odb.local, subdir, fs, "md5")
assert obj.hash_info == HashInfo(
"md5", "af314506f1622d107e0ed3f14ec1a3b5.dir"
)
assert fs.info(posixpath.join(subdir, "data")).get("md5") is None
_, _, obj = stage(dvc.odb.local, posixpath.join(subdir, "data"), fs, "md5")
assert obj.hash_info == HashInfo("md5", "8d777f385d3dfec8815d20f7496026dc")
(tmp_dir / "dir" / "subdir" / "data").unlink()
assert (
fs.info(posixpath.join(subdir, "data"))["md5"]
== "8d777f385d3dfec8815d20f7496026dc"
)
def test_get_hash_file(tmp_dir, dvc):
tmp_dir.dvc_gen({"foo": "foo"})
fs = DvcFileSystem(dvc)
assert (fs.info(PathInfo(tmp_dir) /
"foo")["md5"] == "acbd18db4cc2f85cedef654fccc4a4d8")
