def test_commit_without_manifest(self):
"""Tests a Release can still be produced when the manifest is not understood
by the custom parser in dulwich_commit_to_revision."""
target = b"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"
message = b"some commit message"
author = Person(fullname=b"Foo <*****@*****.**>",
name=b"Foo",
email=b"*****@*****.**")
commit = dulwich.objects.Commit()
commit.tree = target
commit.message = message
commit.author = commit.committer = b"Foo <*****@*****.**>"
commit.author_time = commit.commit_time = 1641980946
commit.author_timezone = commit.commit_timezone = 3600
assert converters.dulwich_commit_to_revision(commit) == Revision(
message=b"some commit message",
author=author,
committer=author,
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1641980946, microseconds=0),
offset_bytes=b"+0100",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1641980946, microseconds=0),
offset_bytes=b"+0100",
),
type=RevisionType.GIT,
directory=hash_to_bytes(target.decode()),
synthetic=False,
metadata=None,
parents=(),
)
def test_svn_date_to_swh_date():
"""The timestamp should not be tampered with and include the
decimals.
"""
assert converters.svn_date_to_swh_date(
b"2011-05-31T06:04:39.500900Z") == TimestampWithTimezone(
timestamp=Timestamp(seconds=1306821879, microseconds=500900),
offset_bytes=b"+0000",
)
assert converters.svn_date_to_swh_date(
b"2011-05-31T06:04:39.800722Z") == TimestampWithTimezone(
timestamp=Timestamp(seconds=1306821879, microseconds=800722),
offset_bytes=b"+0000",
)
def db_to_date(
date: Optional[datetime.datetime],
offset_bytes: bytes,
) -> Optional[TimestampWithTimezone]:
"""Convert the DB representation of a date to a swh-model compatible date.
Args:
date: a date pulled out of the database
offset_bytes: a byte representation of the latter two, usually as "+HHMM"
or "-HHMM"
Returns:
a TimestampWithTimezone, or None if the date is None.
"""
if date is None:
return None
return TimestampWithTimezone(
timestamp=Timestamp(
# we use floor() instead of int() to round down, because of negative dates
seconds=math.floor(date.timestamp()),
microseconds=date.microsecond,
),
offset_bytes=offset_bytes,
)
def test_commit_to_revision_with_extra_headers_mergetag(self):
sha1 = b"3ab3da4bf0f81407be16969df09cd1c8af9ac703"
revision = converters.dulwich_commit_to_revision(self.repo[sha1])
expected_revision = Revision(
id=hash_to_bytes(sha1.decode()),
directory=bytes.fromhex(
"faa4b64a841ca3e3f07d6501caebda2e3e8e544e"),
type=RevisionType.GIT,
committer=Person(
name=b"David Douard",
fullname=b"David Douard <*****@*****.**>",
email=b"*****@*****.**",
),
author=Person(
name=b"David Douard",
fullname=b"David Douard <*****@*****.**>",
email=b"*****@*****.**",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1594138183,
microseconds=0,
),
offset_bytes=b"+0200",
),
message=b"Merge tag 'v0.0.1' into readme\n\nv0.0.1\n",
metadata=None,
extra_headers=((b"encoding", b"ISO-8859-15"), (b"mergetag",
MERGETAG)),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1594138183,
microseconds=0,
),
offset_bytes=b"+0200",
),
parents=(
bytes.fromhex("322f5bc915e50fc25e85226b5a182bded0e98e4b"),
bytes.fromhex("9768d0b576dbaaecd80abedad6dfd0d72f1476da"),
),
synthetic=False,
)
assert revision == expected_revision
def test_svn_date_to_swh_date_epoch():
"""Empty date should be EPOCH (timestamp and offset at 0)."""
# It should return 0, epoch
default_tstz = TimestampWithTimezone(timestamp=Timestamp(seconds=0,
microseconds=0),
offset_bytes=b"+0000")
assert converters.svn_date_to_swh_date("") == default_tstz
assert converters.svn_date_to_swh_date(None) == default_tstz
def test_build_swh_revision_default():
"""This should build the swh revision with the swh revision's extra
headers about the repository.
"""
dir_id = hash_to_bytes("d6e08e19159f77983242877c373c75222d5ae9dd")
date = TimestampWithTimezone(timestamp=Timestamp(seconds=1088108379,
microseconds=0),
offset_bytes=b"+0000")
actual_rev = converters.build_swh_revision(
repo_uuid=b"uuid",
dir_id=dir_id,
commit={
"author_name":
Person(name=b"theo",
email=b"theo@uuid",
fullname=b"theo <theo@uuid>"),
"message":
b"commit message",
"author_date":
date,
},
rev=10,
parents=(),
)
expected_rev = Revision.from_dict({
"date":
date.to_dict(),
"committer_date":
date.to_dict(),
"type":
"svn",
"directory":
dir_id,
"message":
b"commit message",
"author": {
"name": b"theo",
"email": b"theo@uuid",
"fullname": b"theo <theo@uuid>",
},
"committer": {
"name": b"theo",
"email": b"theo@uuid",
"fullname": b"theo <theo@uuid>",
},
"synthetic":
True,
"extra_headers": (
(b"svn_repo_uuid", b"uuid"),
(b"svn_revision", b"10"),
),
"parents": (),
})
assert actual_rev == expected_rev
def test_commit_to_revision_with_extra_headers(self):
sha1 = b"322f5bc915e50fc25e85226b5a182bded0e98e4b"
revision = converters.dulwich_commit_to_revision(self.repo[sha1])
expected_revision = Revision(
id=hash_to_bytes(sha1.decode()),
directory=bytes.fromhex(
"f8ec06e4ed7b9fff4918a0241a48023143f30000"),
type=RevisionType.GIT,
committer=Person(
name=b"David Douard",
fullname=b"David Douard <*****@*****.**>",
email=b"*****@*****.**",
),
author=Person(
name=b"David Douard",
fullname=b"David Douard <*****@*****.**>",
email=b"*****@*****.**",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1594137902,
microseconds=0,
),
offset_bytes=b"+0200",
),
message=b"Am\xe9lioration du fichier READM\xa4\n",
metadata=None,
extra_headers=((b"encoding", b"ISO-8859-15"), (b"gpgsig", GPGSIG)),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1594136900,
microseconds=0,
),
offset_bytes=b"+0200",
),
parents=(
bytes.fromhex("c730509025c6e81947102b2d77bc4dc1cade9489"), ),
synthetic=False,
)
assert revision == expected_revision
def test_commit_to_revision(self):
sha1 = b"9768d0b576dbaaecd80abedad6dfd0d72f1476da"
revision = converters.dulwich_commit_to_revision(self.repo[sha1])
expected_revision = Revision(
id=hash_to_bytes("9768d0b576dbaaecd80abedad6dfd0d72f1476da"),
directory=b"\xf0i\\./\xa7\xce\x9dW@#\xc3A7a\xa4s\xe5\x00\xca",
type=RevisionType.GIT,
committer=Person(
name=b"Stefano Zacchiroli",
fullname=b"Stefano Zacchiroli <*****@*****.**>",
email=b"*****@*****.**",
),
author=Person(
name=b"Stefano Zacchiroli",
fullname=b"Stefano Zacchiroli <*****@*****.**>",
email=b"*****@*****.**",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1443083765,
microseconds=0,
),
offset_bytes=b"+0200",
),
message=b"add submodule dependency\n",
metadata=None,
extra_headers=(),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1443083765,
microseconds=0,
),
offset_bytes=b"+0200",
),
parents=(
b"\xc3\xc5\x88q23`\x9f[\xbb\xb2\xd9\xe7\xf3\xfbJf\x0f?r", ),
synthetic=False,
)
assert revision == expected_revision
def test_dulwich_tag_to_release_author_and_date(self):
sha = hash_to_bytes("fc1e6a4f1e37e93e28e78560e73efd0b12f616ef")
tagger = b"hey dude <*****@*****.**>"
target = b"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"
message = b"some release message"
date = int(
datetime.datetime(2007, 12, 5,
tzinfo=datetime.timezone.utc).timestamp())
tag = dulwich.objects.Tag()
tag.name = b"blah"
tag.object = (dulwich.objects.Commit, target)
tag.message = message
tag.signature = None
tag.tagger = tagger
tag.tag_time = date
tag.tag_timezone = 0
assert tag.sha().digest() == sha
# when
actual_release = converters.dulwich_tag_to_release(tag)
# then
expected_release = Release(
author=Person(
email=b"*****@*****.**",
fullname=b"hey dude <*****@*****.**>",
name=b"hey dude",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1196812800,
microseconds=0,
),
offset_bytes=b"+0000",
),
id=sha,
message=message,
metadata=None,
name=b"blah",
synthetic=False,
target=hash_to_bytes(target.decode()),
target_type=ObjectType.REVISION,
)
assert actual_release == expected_release
def test_dulwich_tag_to_release_author_zero_date(self):
# to reproduce bug T815 (fixed)
sha = hash_to_bytes("6cc1deff5cdcd853428bb63b937f43dd2566c36f")
tagger = b"hey dude <*****@*****.**>"
target = b"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"
message = b"some release message"
date = int(
datetime.datetime(1970, 1, 1,
tzinfo=datetime.timezone.utc).timestamp())
tag = dulwich.objects.Tag()
tag.name = b"blah"
tag.object = (dulwich.objects.Commit, target)
tag.message = message
tag.signature = None
tag.tagger = tagger
tag.tag_time = date
tag.tag_timezone = 0
assert tag.sha().digest() == sha
# when
actual_release = converters.dulwich_tag_to_release(tag)
# then
expected_release = Release(
author=Person(
email=b"*****@*****.**",
fullname=b"hey dude <*****@*****.**>",
name=b"hey dude",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=0,
microseconds=0,
),
offset_bytes=b"+0000",
),
id=sha,
message=message,
metadata=None,
name=b"blah",
synthetic=False,
target=hash_to_bytes(target.decode()),
target_type=ObjectType.REVISION,
)
assert actual_release == expected_release
def test_from_release():
"""Convert release model object to a dict should be ok"""
ts = int(
datetime.datetime(2015, 1, 1, 22, 0, 0,
tzinfo=datetime.timezone.utc).timestamp())
release_input = Release(
id=hashutil.hash_to_bytes("aad23fa492a0c5fed0708a6703be875448c86884"),
target=hashutil.hash_to_bytes(
"5e46d564378afc44b31bb89f99d5675195fbdf67"),
target_type=ObjectType.REVISION,
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=ts, microseconds=0),
offset=0,
negative_utc=False,
),
author=Person(
name=b"author name",
fullname=b"Author Name author@email",
email=b"author@email",
),
name=b"v0.0.1",
message=b"some comment on release",
synthetic=True,
)
expected_release = {
"id": "aad23fa492a0c5fed0708a6703be875448c86884",
"target": "5e46d564378afc44b31bb89f99d5675195fbdf67",
"target_type": "revision",
"date": "2015-01-01T22:00:00+00:00",
"author": {
"name": "author name",
"fullname": "Author Name author@email",
"email": "author@email",
},
"name": "v0.0.1",
"message": "some comment on release",
"target_type": "revision",
"synthetic": True,
}
actual_release = converters.from_release(release_input)
assert actual_release == expected_release
def dulwich_tsinfo_to_timestamp(
timestamp,
timezone: int,
timezone_neg_utc: bool,
timezone_bytes: Optional[bytes],
) -> TimestampWithTimezone:
"""Convert the dulwich timestamp information to a structure compatible with
Software Heritage."""
ts = Timestamp(
seconds=int(timestamp),
microseconds=0,
)
if timezone_bytes is None:
# Failed to parse from the raw manifest, fallback to what Dulwich managed to
# parse.
return TimestampWithTimezone.from_numeric_offset(
timestamp=ts,
offset=timezone // 60,
negative_utc=timezone_neg_utc,
)
else:
return TimestampWithTimezone(timestamp=ts, offset_bytes=timezone_bytes)
def test_weird_commit(self):
"""Checks raw_manifest is set when the commit cannot fit the data model"""
# Well-formed manifest
raw_manifest = (b"tree 641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce\n"
b"author Foo <*****@*****.**> 1640191028 +0200\n"
b"committer Foo <*****@*****.**> 1640191028 +0200\n\n"
b"some commit message")
commit = dulwich.objects.Commit.from_raw_string(
b"commit", raw_manifest)
date = TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191028, microseconds=0),
offset_bytes=b"+0200",
)
assert converters.dulwich_commit_to_revision(commit) == Revision(
message=b"some commit message",
directory=hash_to_bytes(
"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
committer=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=date,
committer_date=date,
type=RevisionType.GIT,
raw_manifest=None,
)
# Mess with the offset
raw_manifest2 = raw_manifest.replace(b"+0200", b"+200")
commit = dulwich.objects.Commit.from_raw_string(
b"commit", raw_manifest2)
date = TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191028, microseconds=0),
offset_bytes=b"+200",
)
assert converters.dulwich_commit_to_revision(commit) == Revision(
message=b"some commit message",
directory=hash_to_bytes(
"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
committer=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=date,
committer_date=date,
type=RevisionType.GIT,
raw_manifest=None,
)
# Mess with the rest of the manifest
raw_manifest2 = raw_manifest.replace(
b"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce",
b"641FB6E08DDB2E4FD096DCF18E80B894BF7E25CE",
)
commit = dulwich.objects.Commit.from_raw_string(
b"commit", raw_manifest2)
date = TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191028, microseconds=0),
offset_bytes=b"+0200",
)
assert converters.dulwich_commit_to_revision(commit) == Revision(
message=b"some commit message",
directory=hash_to_bytes(
"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
committer=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=date,
committer_date=date,
type=RevisionType.GIT,
raw_manifest=b"commit 161\x00" + raw_manifest2,
)
"model_date,db_date",
[
(
None,
{
"timestamp": None,
"offset": 0,
"neg_utc_offset": None,
"offset_bytes": None,
},
),
(
TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567890,
microseconds=0,
),
offset_bytes=b"+0200",
),
{
"timestamp": "2009-02-13T23:31:30+00:00",
"offset": 120,
"neg_utc_offset": False,
"offset_bytes": b"+0200",
},
),
(
TimestampWithTimezone(
timestamp=Timestamp(
seconds=1123456789,
microseconds=0,
type="file",
target=CONTENT.sha1_git,
perms=DentryPerms.content,
)
]),
)
REVISION = Revision(
id=hash_to_bytes("066b1b62dbfa033362092af468bf6cfabec230e7"),
message=b"hello",
author=Person(
name=b"Nicolas Dandrimont",
email=b"*****@*****.**",
fullname=b"Nicolas Dandrimont <*****@*****.**> ",
),
date=TimestampWithTimezone(Timestamp(1234567890, 0),
offset_bytes=b"+0200"),
committer=Person(
name=b"St\xc3fano Zacchiroli",
email=b"*****@*****.**",
fullname=b"St\xc3fano Zacchiroli <*****@*****.**>",
),
committer_date=TimestampWithTimezone(Timestamp(1123456789, 0),
offset_bytes=b"-0000"),
parents=(),
type=RevisionType.GIT,
directory=DIRECTORY.id,
metadata={
"checksums": {
"sha1": "tarball-sha1",
"sha256": "tarball-sha256",
},
def test_debian_origins_from_row__check_revisions():
"""Tests debian_origins_from_row errors when the revision at the head
of a branch is a DSC and has no parents
"""
storage = get_storage("memory")
origin_url = "deb://Debian/packages/kalgebra"
revision_id = b"21" * 10
storage.origin_add([Origin(url=origin_url)])
revision_row = {
"id": b"\x00\x00\x03l1\x1e\xf3:(\x1b\x05h\x8fn\xad\xcf\xc0\x94:\xee",
"directory": DIRECTORY_ID,
"metadata": {
"original_artifact": [
{
"filename": "kalgebra_19.12.1-1.dsc",
},
]
},
}
storage.origin_visit_add([
OriginVisit(
origin=origin_url,
date=datetime.datetime.now(tz=datetime.timezone.utc),
type="deb",
visit=280,
)
])
storage.origin_visit_status_add([
OriginVisitStatus(
origin=origin_url,
visit=280,
date=datetime.datetime.now(tz=datetime.timezone.utc),
status="full",
snapshot=b"42" * 10,
metadata=None,
)
])
storage.snapshot_add([
Snapshot(
id=b"42" * 10,
branches={
b"foo":
SnapshotBranch(target_type=TargetType.REVISION,
target=revision_id)
},
)
])
storage_before_revision = copy.deepcopy(storage)
revision = Revision(
id=revision_id,
message=b"foo",
author=Person.from_fullname(b"foo"),
committer=Person.from_fullname(b"foo"),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1580076204, microseconds=0),
offset_bytes=b"+0100",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1580076204, microseconds=0),
offset_bytes=b"+0100",
),
type=RevisionType.DSC,
directory=
b"\xd5\x9a\x1f\x9c\x80\x9d\x8c}19P\xf6\xc8\xa2\x0f^%H\xcd\xdb",
synthetic=True,
metadata=None,
parents=(b"parent " * 2, ),
extra_headers=(),
)
storage.revision_add([revision])
with pytest.raises(AssertionError, match="revision with parents"):
debian_origins_from_row(revision_row, storage)
def test_weird_tag(self):
"""Checks raw_manifest is set when the tag cannot fit the data model"""
# Well-formed manifest
raw_manifest = (b"object 641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce\n"
b"type commit\n"
b"tag blah\n"
b"tagger Foo <*****@*****.**> 1640191027 +0200\n\n"
b"some release message")
tag = dulwich.objects.Tag.from_raw_string(b"tag", raw_manifest)
assert converters.dulwich_tag_to_release(tag) == Release(
name=b"blah",
message=b"some release message",
target=hash_to_bytes("641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
target_type=ObjectType.REVISION,
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191027, microseconds=0),
offset_bytes=b"+0200",
),
raw_manifest=None,
)
# Mess with the offset (negative UTC)
raw_manifest2 = raw_manifest.replace(b"+0200", b"-0000")
tag = dulwich.objects.Tag.from_raw_string(b"tag", raw_manifest2)
assert converters.dulwich_tag_to_release(tag) == Release(
name=b"blah",
message=b"some release message",
target=hash_to_bytes("641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
target_type=ObjectType.REVISION,
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191027, microseconds=0),
offset_bytes=b"-0000",
),
)
# Mess with the offset (other)
raw_manifest2 = raw_manifest.replace(b"+0200", b"+200")
tag = dulwich.objects.Tag.from_raw_string(b"tag", raw_manifest2)
assert converters.dulwich_tag_to_release(tag) == Release(
name=b"blah",
message=b"some release message",
target=hash_to_bytes("641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
target_type=ObjectType.REVISION,
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191027, microseconds=0),
offset_bytes=b"+200",
),
)
# Mess with the rest of the manifest
raw_manifest2 = raw_manifest.replace(
b"641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce",
b"641FB6E08DDB2E4FD096DCF18E80B894BF7E25CE",
)
tag = dulwich.objects.Tag.from_raw_string(b"tag", raw_manifest2)
assert converters.dulwich_tag_to_release(tag) == Release(
name=b"blah",
message=b"some release message",
target=hash_to_bytes("641fb6e08ddb2e4fd096dcf18e80b894bf7e25ce"),
target_type=ObjectType.REVISION,
synthetic=False,
author=Person.from_fullname(b"Foo <*****@*****.**>", ),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1640191027, microseconds=0),
offset_bytes=b"+0200",
),
raw_manifest=b"tag 136\x00" + raw_manifest2,
)
def test_from_revision_model_object():
ts = int(
datetime.datetime(2000,
1,
17,
11,
23,
54,
tzinfo=datetime.timezone.utc).timestamp())
revision_input = Revision(
directory=hashutil.hash_to_bytes(
"7834ef7e7c357ce2af928115c6c6a42b7e2a44e6"),
author=Person(
name=b"Software Heritage",
fullname=b"robot [email protected]",
email=b"*****@*****.**",
),
committer=Person(
name=b"Software Heritage",
fullname=b"robot [email protected]",
email=b"*****@*****.**",
),
message=b"synthetic revision message",
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=ts, microseconds=0),
offset=0,
negative_utc=False,
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(seconds=ts, microseconds=0),
offset=0,
negative_utc=False,
),
synthetic=True,
type=RevisionType.TAR,
parents=tuple([
hashutil.hash_to_bytes("29d8be353ed3480476f032475e7c244eff7371d5"),
hashutil.hash_to_bytes("30d8be353ed3480476f032475e7c244eff7371d5"),
]),
extra_headers=((b"gpgsig", b"some-signature"), ),
metadata={
"original_artifact": [{
"archive_type":
"tar",
"name":
"webbase-5.7.0.tar.gz",
"sha1":
"147f73f369733d088b7a6fa9c4e0273dcd3c7ccd",
"sha1_git":
"6a15ea8b881069adedf11feceec35588f2cfe8f1",
"sha256":
"401d0df797110bea805d358b85bcc1ced29549d3d73f"
"309d36484e7edf7bb912",
}],
},
)
expected_revision = {
"id":
"a001358278a0d811fe7072463f805da601121c2a",
"directory":
"7834ef7e7c357ce2af928115c6c6a42b7e2a44e6",
"author": {
"name": "Software Heritage",
"fullname": "robot [email protected]",
"email": "*****@*****.**",
},
"committer": {
"name": "Software Heritage",
"fullname": "robot [email protected]",
"email": "*****@*****.**",
},
"message":
"synthetic revision message",
"date":
"2000-01-17T11:23:54+00:00",
"committer_date":
"2000-01-17T11:23:54+00:00",
"parents":
tuple([
"29d8be353ed3480476f032475e7c244eff7371d5",
"30d8be353ed3480476f032475e7c244eff7371d5",
]),
"type":
"tar",
"synthetic":
True,
"extra_headers": (("gpgsig", "some-signature"), ),
"metadata": {
"original_artifact": [{
"archive_type":
"tar",
"name":
"webbase-5.7.0.tar.gz",
"sha1":
"147f73f369733d088b7a6fa9c4e0273dcd3c7ccd",
"sha1_git":
"6a15ea8b881069adedf11feceec35588f2cfe8f1",
"sha256":
"401d0df797110bea805d358b85bcc1ced29549d3d73f"
"309d36484e7edf7bb912",
}],
},
"merge":
True,
}
actual_revision = converters.from_revision(revision_input)
assert actual_revision == expected_revision
sha1_array[0] += 1
duplicate_content2 = attr.evolve(duplicate_content1,
sha1_git=bytes(sha1_array))
DUPLICATE_CONTENTS = [duplicate_content1, duplicate_content2]
COMMITTERS = [
Person(fullname=b"foo", name=b"foo", email=b""),
Person(fullname=b"bar", name=b"bar", email=b""),
]
DATES = [
TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567891,
microseconds=0,
),
offset_bytes=b"+0200",
),
TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567892,
microseconds=0,
),
offset_bytes=b"+0200",
),
]
REVISIONS = [
Revision(
id=hash_to_bytes("66c7c1cd9673275037140f2abff7b7b11fc9439c"),
class StorageData:
"""Data model objects to use within tests."""
content = Content(
data=b"42\n",
length=3,
sha1=hash_to_bytes("34973274ccef6ab4dfaaf86599792fa9c3fe4689"),
sha1_git=hash_to_bytes("d81cc0710eb6cf9efd5b920a8453e1e07157b6cd"),
sha256=hash_to_bytes(
"084c799cd551dd1d8d5c5f9a5d593b2e931f5e36122ee5c793c1d08a19839cc0"
),
blake2s256=hash_to_bytes(
"d5fe1939576527e42cfd76a9455a2432fe7f56669564577dd93c4280e76d661d"
),
status="visible",
)
content2 = Content(
data=b"4242\n",
length=5,
sha1=hash_to_bytes("61c2b3a30496d329e21af70dd2d7e097046d07b7"),
sha1_git=hash_to_bytes("36fade77193cb6d2bd826161a0979d64c28ab4fa"),
sha256=hash_to_bytes(
"859f0b154fdb2d630f45e1ecae4a862915435e663248bb8461d914696fc047cd"
),
blake2s256=hash_to_bytes(
"849c20fad132b7c2d62c15de310adfe87be94a379941bed295e8141c6219810d"
),
status="visible",
)
content3 = Content(
data=b"424242\n",
length=7,
sha1=hash_to_bytes("3e21cc4942a4234c9e5edd8a9cacd1670fe59f13"),
sha1_git=hash_to_bytes("c932c7649c6dfa4b82327d121215116909eb3bea"),
sha256=hash_to_bytes(
"92fb72daf8c6818288a35137b72155f507e5de8d892712ab96277aaed8cf8a36"
),
blake2s256=hash_to_bytes(
"76d0346f44e5a27f6bafdd9c2befd304aff83780f93121d801ab6a1d4769db11"
),
status="visible",
ctime=datetime.datetime(2019, 12, 1, tzinfo=datetime.timezone.utc),
)
contents: Tuple[Content, ...] = (content, content2, content3)
skipped_content = SkippedContent(
length=1024 * 1024 * 200,
sha1_git=hash_to_bytes("33e45d56f88993aae6a0198013efa80716fd8920"),
sha1=hash_to_bytes("43e45d56f88993aae6a0198013efa80716fd8920"),
sha256=hash_to_bytes(
"7bbd052ab054ef222c1c87be60cd191addedd24cc882d1f5f7f7be61dc61bb3a"
),
blake2s256=hash_to_bytes(
"ade18b1adecb33f891ca36664da676e12c772cc193778aac9a137b8dc5834b9b"
),
reason="Content too long",
status="absent",
origin="file:///dev/zero",
)
skipped_content2 = SkippedContent(
length=1024 * 1024 * 300,
sha1_git=hash_to_bytes("44e45d56f88993aae6a0198013efa80716fd8921"),
sha1=hash_to_bytes("54e45d56f88993aae6a0198013efa80716fd8920"),
sha256=hash_to_bytes(
"8cbd052ab054ef222c1c87be60cd191addedd24cc882d1f5f7f7be61dc61bb3a"
),
blake2s256=hash_to_bytes(
"9ce18b1adecb33f891ca36664da676e12c772cc193778aac9a137b8dc5834b9b"
),
reason="Content too long",
status="absent",
)
skipped_contents: Tuple[SkippedContent,
...] = (skipped_content, skipped_content2)
directory5 = Directory(
id=hash_to_bytes("4b825dc642cb6eb9a060e54bf8d69288fbee4904"),
entries=(),
)
directory = Directory(
id=hash_to_bytes("5256e856a0a0898966d6ba14feb4388b8b82d302"),
entries=tuple([
DirectoryEntry(
name=b"foo",
type="file",
target=content.sha1_git,
perms=from_disk.DentryPerms.content,
),
DirectoryEntry(
name=b"bar\xc3",
type="dir",
target=directory5.id,
perms=from_disk.DentryPerms.directory,
),
], ),
)
directory2 = Directory(
id=hash_to_bytes("8505808532953da7d2581741f01b29c04b1cb9ab"),
entries=tuple([
DirectoryEntry(
name=b"oof",
type="file",
target=content2.sha1_git,
perms=from_disk.DentryPerms.content,
)
], ),
)
directory3 = Directory(
id=hash_to_bytes("13089e6e544f78df7c9a40a3059050d10dee686a"),
entries=tuple([
DirectoryEntry(
name=b"foo",
type="file",
target=content.sha1_git,
perms=from_disk.DentryPerms.content,
),
DirectoryEntry(
name=b"subdir",
type="dir",
target=directory.id,
perms=from_disk.DentryPerms.directory,
),
DirectoryEntry(
name=b"hello",
type="file",
target=content2.sha1_git,
perms=from_disk.DentryPerms.content,
),
], ),
)
directory4 = Directory(
id=hash_to_bytes("cd5dfd9c09d9e99ed123bc7937a0d5fddc3cd531"),
entries=tuple([
DirectoryEntry(
name=b"subdir1",
type="dir",
target=directory3.id,
perms=from_disk.DentryPerms.directory,
)
], ),
)
directory6 = Directory(
id=hash_to_bytes("afa0105cfcaa14fdbacee344e96659170bb1bda5"),
entries=tuple([
DirectoryEntry(
name=b"foo",
type="file",
target=b"\x00" * 20,
perms=from_disk.DentryPerms.content,
),
DirectoryEntry(
name=b"bar",
type="dir",
target=b"\x01" * 20,
perms=from_disk.DentryPerms.directory,
),
], ),
raw_manifest=(
b"tree 61\x00"
b"100644 foo\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00" # noqa
b"40000 bar\x00\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01\x01" # noqa
),
)
directories: Tuple[Directory, ...] = (
directory2,
directory,
directory3,
directory4,
directory5,
directory6,
)
revision = Revision(
id=hash_to_bytes("01a7114f36fddd5ef2511b2cadda237a68adbb12"),
message=b"hello",
author=Person(
name=b"Nicolas Dandrimont",
email=b"*****@*****.**",
fullname=b"Nicolas Dandrimont <*****@*****.**> ",
),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1234567890, microseconds=0),
offset_bytes=b"+0200",
),
committer=Person(
name=b"St\xc3fano Zacchiroli",
email=b"*****@*****.**",
fullname=b"St\xc3fano Zacchiroli <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1123456789, microseconds=0),
offset_bytes=b"+0200",
),
parents=(),
type=RevisionType.GIT,
directory=directory.id,
metadata={
"checksums": {
"sha1": "tarball-sha1",
"sha256": "tarball-sha256",
},
"signed-off-by": "some-dude",
},
extra_headers=(
(b"gpgsig", b"test123"),
(b"mergetag", b"foo\\bar"),
(b"mergetag", b"\x22\xaf\x89\x80\x01\x00"),
),
synthetic=True,
)
revision2 = Revision(
id=hash_to_bytes("a646dd94c912829659b22a1e7e143d2fa5ebde1b"),
message=b"hello again",
author=Person(
name=b"Roberto Dicosmo",
email=b"*****@*****.**",
fullname=b"Roberto Dicosmo <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
committer=Person(
name=b"tony",
email=b"*****@*****.**",
fullname=b"tony <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1123456789,
microseconds=220000,
),
offset_bytes=b"+0000",
),
parents=tuple([revision.id]),
type=RevisionType.GIT,
directory=directory2.id,
metadata=None,
extra_headers=(),
synthetic=False,
)
revision3 = Revision(
id=hash_to_bytes("beb2844dff30658e27573cb46eb55980e974b391"),
message=b"a simple revision with no parents this time",
author=Person(
name=b"Roberto Dicosmo",
email=b"*****@*****.**",
fullname=b"Roberto Dicosmo <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
committer=Person(
name=b"tony",
email=b"*****@*****.**",
fullname=b"tony <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1127351742,
microseconds=220000,
),
offset_bytes=b"+0000",
),
parents=tuple([revision.id, revision2.id]),
type=RevisionType.GIT,
directory=directory2.id,
metadata=None,
extra_headers=(),
synthetic=True,
)
revision4 = Revision(
id=hash_to_bytes("ae860aec43700c7f5a295e2ef47e2ae41b535dfe"),
message=b"parent of self.revision2",
author=Person(
name=b"me",
email=b"*****@*****.**",
fullname=b"me <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
committer=Person(
name=b"committer-dude",
email=b"*****@*****.**",
fullname=b"committer-dude <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1244567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
parents=tuple([revision3.id]),
type=RevisionType.GIT,
directory=directory.id,
metadata=None,
extra_headers=(),
synthetic=False,
)
git_revisions: Tuple[Revision,
...] = (revision, revision2, revision3, revision4)
hg_revision = Revision(
id=hash_to_bytes("951c9503541e7beaf002d7aebf2abd1629084c68"),
message=b"hello",
author=Person(
name=b"Nicolas Dandrimont",
email=b"*****@*****.**",
fullname=b"Nicolas Dandrimont <*****@*****.**> ",
),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1234567890, microseconds=0),
offset_bytes=b"+0200",
),
committer=Person(
name=b"St\xc3fano Zacchiroli",
email=b"*****@*****.**",
fullname=b"St\xc3fano Zacchiroli <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1123456789, microseconds=0),
offset_bytes=b"+0200",
),
parents=(),
type=RevisionType.MERCURIAL,
directory=directory.id,
metadata={
"checksums": {
"sha1": "tarball-sha1",
"sha256": "tarball-sha256",
},
"signed-off-by": "some-dude",
"node": "a316dfb434af2b451c1f393496b7eaeda343f543",
},
extra_headers=(),
synthetic=True,
)
hg_revision2 = Revision(
id=hash_to_bytes("df4afb063236300eb13b96a0d7fff03f7b7cbbaf"),
message=b"hello again",
author=Person(
name=b"Roberto Dicosmo",
email=b"*****@*****.**",
fullname=b"Roberto Dicosmo <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
committer=Person(
name=b"tony",
email=b"*****@*****.**",
fullname=b"tony <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1123456789,
microseconds=220000,
),
offset_bytes=b"+0000",
),
parents=tuple([hg_revision.id]),
type=RevisionType.MERCURIAL,
directory=directory2.id,
metadata=None,
extra_headers=(
(b"node",
hash_to_bytes("fa1b7c84a9b40605b67653700f268349a6d6aca1")), ),
synthetic=False,
)
hg_revision3 = Revision(
id=hash_to_bytes("84d8e7081b47ebb88cad9fa1f25de5f330872a37"),
message=b"a simple revision with no parents this time",
author=Person(
name=b"Roberto Dicosmo",
email=b"*****@*****.**",
fullname=b"Roberto Dicosmo <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
committer=Person(
name=b"tony",
email=b"*****@*****.**",
fullname=b"tony <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1127351742,
microseconds=220000,
),
offset_bytes=b"+0000",
),
parents=tuple([hg_revision.id, hg_revision2.id]),
type=RevisionType.MERCURIAL,
directory=directory2.id,
metadata=None,
extra_headers=(
(b"node",
hash_to_bytes("7f294a01c49065a90b3fe8b4ad49f08ce9656ef6")), ),
synthetic=True,
)
hg_revision4 = Revision(
id=hash_to_bytes("4683324ba26dfe941a72cc7552e86eaaf7c27fe3"),
message=b"parent of self.revision2",
author=Person(
name=b"me",
email=b"*****@*****.**",
fullname=b"me <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1234567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
committer=Person(
name=b"committer-dude",
email=b"*****@*****.**",
fullname=b"committer-dude <*****@*****.**>",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(
seconds=1244567843,
microseconds=220000,
),
offset_bytes=b"-1200",
),
parents=tuple([hg_revision3.id]),
type=RevisionType.MERCURIAL,
directory=directory.id,
metadata=None,
extra_headers=(
(b"node",
hash_to_bytes("f4160af0485c85823d9e829bae2c00b00a2e6297")), ),
synthetic=False,
)
hg_revisions: Tuple[Revision, ...] = (
hg_revision,
hg_revision2,
hg_revision3,
hg_revision4,
)
revisions: Tuple[Revision, ...] = git_revisions + hg_revisions
origins: Tuple[Origin, ...] = (
Origin(url="https://github.com/user1/repo1"),
Origin(url="https://github.com/user2/repo1"),
Origin(url="https://github.com/user3/repo1"),
Origin(url="https://gitlab.com/user1/repo1"),
Origin(url="https://gitlab.com/user2/repo1"),
Origin(url="https://forge.softwareheritage.org/source/repo1"),
Origin(url="https://example.рф/🏛️.txt"),
)
origin, origin2 = origins[:2]
metadata_authority = MetadataAuthority(
type=MetadataAuthorityType.DEPOSIT_CLIENT,
url="http://hal.inria.example.com/",
)
metadata_authority2 = MetadataAuthority(
type=MetadataAuthorityType.REGISTRY,
url="http://wikidata.example.com/",
)
authorities: Tuple[MetadataAuthority, ...] = (
metadata_authority,
metadata_authority2,
)
metadata_fetcher = MetadataFetcher(
name="swh-deposit",
version="0.0.1",
)
metadata_fetcher2 = MetadataFetcher(
name="swh-example",
version="0.0.1",
)
fetchers: Tuple[MetadataFetcher,
...] = (metadata_fetcher, metadata_fetcher2)
date_visit1 = datetime.datetime(2015,
1,
1,
23,
0,
0,
tzinfo=datetime.timezone.utc)
date_visit2 = datetime.datetime(2017,
1,
1,
23,
0,
0,
tzinfo=datetime.timezone.utc)
date_visit3 = datetime.datetime(2018,
1,
1,
23,
0,
0,
tzinfo=datetime.timezone.utc)
type_visit1 = "git"
type_visit2 = "hg"
type_visit3 = "deb"
origin_visit = OriginVisit(
origin=origin.url,
visit=1,
date=date_visit1,
type=type_visit1,
)
origin_visit2 = OriginVisit(
origin=origin.url,
visit=2,
date=date_visit2,
type=type_visit1,
)
origin_visit3 = OriginVisit(
origin=origin2.url,
visit=1,
date=date_visit1,
type=type_visit2,
)
origin_visits: Tuple[OriginVisit, ...] = (
origin_visit,
origin_visit2,
origin_visit3,
)
release = Release(
id=hash_to_bytes("f7f222093a18ec60d781070abec4a630c850b837"),
name=b"v0.0.1",
author=Person(
name=b"olasd",
email=b"*****@*****.**",
fullname=b"olasd <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1234567890, microseconds=0),
offset_bytes=b"+0042",
),
target=revision.id,
target_type=ObjectType.REVISION,
message=b"synthetic release",
synthetic=True,
)
release2 = Release(
id=hash_to_bytes("db81a26783a3f4a9db07b4759ffc37621f159bb2"),
name=b"v0.0.2",
author=Person(
name=b"tony",
email=b"*****@*****.**",
fullname=b"tony <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1634366813, microseconds=0),
offset_bytes=b"-0200",
),
target=revision2.id,
target_type=ObjectType.REVISION,
message=b"v0.0.2\nMisc performance improvements + bug fixes",
synthetic=False,
)
release3 = Release(
id=hash_to_bytes("1c5d42e603ce2eea44917fadca76c78bad76aeb9"),
name=b"v0.0.2",
author=Person(
name=b"tony",
email=b"*****@*****.**",
fullname=b"tony <*****@*****.**>",
),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1634366813, microseconds=0),
offset_bytes=b"-0200",
),
target=revision3.id,
target_type=ObjectType.REVISION,
message=b"yet another synthetic release",
synthetic=True,
)
releases: Tuple[Release, ...] = (release, release2, release3)
snapshot = Snapshot(
id=hash_to_bytes("9b922e6d8d5b803c1582aabe5525b7b91150788e"),
branches={
b"master":
SnapshotBranch(
target=revision.id,
target_type=TargetType.REVISION,
),
},
)
empty_snapshot = Snapshot(
id=hash_to_bytes("1a8893e6a86f444e8be8e7bda6cb34fb1735a00e"),
branches={},
)
complete_snapshot = Snapshot(
id=hash_to_bytes("db99fda25b43dc5cd90625ee4b0744751799c917"),
branches={
b"directory":
SnapshotBranch(
target=directory.id,
target_type=TargetType.DIRECTORY,
),
b"directory2":
SnapshotBranch(
target=directory2.id,
target_type=TargetType.DIRECTORY,
),
b"content":
SnapshotBranch(
target=content.sha1_git,
target_type=TargetType.CONTENT,
),
b"alias":
SnapshotBranch(
target=b"revision",
target_type=TargetType.ALIAS,
),
b"revision":
SnapshotBranch(
target=revision.id,
target_type=TargetType.REVISION,
),
b"release":
SnapshotBranch(
target=release.id,
target_type=TargetType.RELEASE,
),
b"snapshot":
SnapshotBranch(
target=empty_snapshot.id,
target_type=TargetType.SNAPSHOT,
),
b"dangling":
None,
},
)
snapshots: Tuple[Snapshot,
...] = (snapshot, empty_snapshot, complete_snapshot)
content_metadata1 = RawExtrinsicMetadata(
target=ExtendedSWHID(object_type=ExtendedObjectType.CONTENT,
object_id=content.sha1_git),
origin=origin.url,
discovery_date=datetime.datetime(2015,
1,
1,
21,
0,
0,
tzinfo=datetime.timezone.utc),
authority=metadata_authority,
fetcher=metadata_fetcher,
format="json",
metadata=b'{"foo": "bar"}',
)
content_metadata2 = RawExtrinsicMetadata(
target=ExtendedSWHID(object_type=ExtendedObjectType.CONTENT,
object_id=content.sha1_git),
origin=origin2.url,
discovery_date=datetime.datetime(2017,
1,
1,
22,
0,
0,
tzinfo=datetime.timezone.utc),
authority=metadata_authority,
fetcher=metadata_fetcher,
format="yaml",
metadata=b"foo: bar",
)
content_metadata3 = RawExtrinsicMetadata(
target=ExtendedSWHID(object_type=ExtendedObjectType.CONTENT,
object_id=content.sha1_git),
discovery_date=datetime.datetime(2017,
1,
1,
22,
0,
0,
tzinfo=datetime.timezone.utc),
authority=attr.evolve(metadata_authority2, metadata=None),
fetcher=attr.evolve(metadata_fetcher2, metadata=None),
format="yaml",
metadata=b"foo: bar",
origin=origin.url,
visit=42,
snapshot=snapshot.swhid(),
release=release.swhid(),
revision=revision.swhid(),
directory=directory.swhid(),
path=b"/foo/bar",
)
content_metadata: Tuple[RawExtrinsicMetadata, ...] = (
content_metadata1,
content_metadata2,
content_metadata3,
)
origin_metadata1 = RawExtrinsicMetadata(
target=Origin(origin.url).swhid(),
discovery_date=datetime.datetime(2015,
1,
1,
21,
0,
0,
tzinfo=datetime.timezone.utc),
authority=attr.evolve(metadata_authority, metadata=None),
fetcher=attr.evolve(metadata_fetcher, metadata=None),
format="json",
metadata=b'{"foo": "bar"}',
)
origin_metadata2 = RawExtrinsicMetadata(
target=Origin(origin.url).swhid(),
discovery_date=datetime.datetime(2017,
1,
1,
22,
0,
0,
tzinfo=datetime.timezone.utc),
authority=attr.evolve(metadata_authority, metadata=None),
fetcher=attr.evolve(metadata_fetcher, metadata=None),
format="yaml",
metadata=b"foo: bar",
)
origin_metadata3 = RawExtrinsicMetadata(
target=Origin(origin.url).swhid(),
discovery_date=datetime.datetime(2017,
1,
1,
22,
0,
0,
tzinfo=datetime.timezone.utc),
authority=attr.evolve(metadata_authority2, metadata=None),
fetcher=attr.evolve(metadata_fetcher2, metadata=None),
format="yaml",
metadata=b"foo: bar",
)
origin_metadata: Tuple[RawExtrinsicMetadata, ...] = (
origin_metadata1,
origin_metadata2,
origin_metadata3,
)
extid1 = ExtID(
target=CoreSWHID(object_type=SwhidObjectType.REVISION,
object_id=revision.id),
extid_type="git",
extid=revision.id,
)
extid2 = ExtID(
target=CoreSWHID(object_type=SwhidObjectType.REVISION,
object_id=hg_revision.id),
extid_type="mercurial",
extid=hash_to_bytes("a316dfb434af2b451c1f393496b7eaeda343f543"),
)
extid3 = ExtID(
target=CoreSWHID(object_type=SwhidObjectType.DIRECTORY,
object_id=directory.id),
extid_type="directory",
extid=b"something",
)
extid4 = ExtID(
target=CoreSWHID(object_type=SwhidObjectType.DIRECTORY,
object_id=directory2.id),
extid_type="directory",
extid=b"something",
extid_version=2,
)
extids: Tuple[ExtID, ...] = (
extid1,
extid2,
extid3,
extid4,
)
def test_debian_origins_from_row__no_result():
"""Tests debian_origins_from_row when there's no origin, visit, status,
snapshot, branch, or matching branch.
"""
storage = get_storage("memory")
origin_url = "deb://Debian/packages/kalgebra"
snapshot_id = b"42424242424242424242"
revision_id = b"21212121212121212121"
storage.origin_add([Origin(url=origin_url)])
revision_row = {
"id": b"\x00\x00\x03l1\x1e\xf3:(\x1b\x05h\x8fn\xad\xcf\xc0\x94:\xee",
"directory": DIRECTORY_ID,
"metadata": {
"original_artifact": [
{
"filename": "kalgebra_19.12.1-1.dsc",
},
]
},
}
# no visit
assert debian_origins_from_row(revision_row, storage) == []
storage.origin_visit_add(
[OriginVisit(
origin=origin_url,
date=now(),
type="deb",
visit=280,
)])
# no status
assert debian_origins_from_row(revision_row, storage) == []
status = OriginVisitStatus(
origin=origin_url,
visit=280,
date=now(),
status="full",
snapshot=None,
metadata=None,
)
storage.origin_visit_status_add([status])
# no snapshot
assert debian_origins_from_row(revision_row, storage) == []
status = attr.evolve(status, snapshot=snapshot_id, date=now())
storage.origin_visit_status_add([status])
storage_before_snapshot = copy.deepcopy(storage)
snapshot = Snapshot(id=snapshot_id, branches={})
storage.snapshot_add([snapshot])
# no branch
assert debian_origins_from_row(revision_row, storage) == []
# "remove" the snapshot, so we can add a new one with the same id
storage = copy.deepcopy(storage_before_snapshot)
snapshot = attr.evolve(
snapshot,
branches={
b"foo": None,
},
)
storage.snapshot_add([snapshot])
# dangling branch
assert debian_origins_from_row(revision_row, storage) == []
# "remove" the snapshot again
storage = copy.deepcopy(storage_before_snapshot)
snapshot = attr.evolve(
snapshot,
branches={
b"foo":
SnapshotBranch(
target_type=TargetType.REVISION,
target=revision_id,
)
},
)
storage.snapshot_add([snapshot])
# branch points to unknown revision
assert debian_origins_from_row(revision_row, storage) == []
revision = Revision(
id=revision_id,
message=b"foo",
author=Person.from_fullname(b"foo"),
committer=Person.from_fullname(b"foo"),
date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1580076204, microseconds=0),
offset_bytes=b"+0100",
),
committer_date=TimestampWithTimezone(
timestamp=Timestamp(seconds=1580076204, microseconds=0),
offset_bytes=b"+0100",
),
type=RevisionType.DSC,
directory=
b"\xd5\x9a\x1f\x9c\x80\x9d\x8c}19P\xf6\xc8\xa2\x0f^%H\xcd\xdb",
synthetic=True,
metadata=None,
parents=(),
extra_headers=(),
)
storage.revision_add([revision])
# no matching branch
assert debian_origins_from_row(revision_row, storage) == []