Skip to content

Commit 30f52ce

Browse files
committed
Support on-demand content in repair_metadata
1 parent 134afaa commit 30f52ce

File tree

3 files changed

+178
-5
lines changed

3 files changed

+178
-5
lines changed

pulp_python/app/tasks/repair.py

Lines changed: 68 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -2,12 +2,17 @@
22
import uuid
33
from gettext import gettext as _
44

5+
from requests.exceptions import RequestException
56
from django.db.models.query import QuerySet
67
from pulpcore.plugin.models import ProgressReport
78
from pulpcore.plugin.util import get_domain
89

910
from pulp_python.app.models import PythonPackageContent, PythonRepository
10-
from pulp_python.app.utils import artifact_to_python_content_data
11+
from pulp_python.app.utils import (
12+
artifact_to_python_content_data,
13+
fetch_json_release_metadata,
14+
parse_metadata,
15+
)
1116

1217
log = logging.getLogger(__name__)
1318

@@ -47,8 +52,16 @@ def repair_metadata(content: QuerySet[PythonPackageContent]) -> int:
4752
Returns:
4853
int: The number of packages that were repaired.
4954
"""
50-
# TODO: Add on_demand content repair
51-
immediate_content = content.filter(contentartifact__artifact__isnull=False)
55+
immediate_content = (
56+
content.filter(contentartifact__artifact__isnull=False)
57+
.distinct()
58+
.prefetch_related("_artifacts")
59+
)
60+
on_demand_content = (
61+
content.filter(contentartifact__artifact__isnull=True)
62+
.distinct()
63+
.prefetch_related("contentartifact_set__remoteartifact_set")
64+
)
5265
domain = get_domain()
5366

5467
batch = []
@@ -58,12 +71,12 @@ def repair_metadata(content: QuerySet[PythonPackageContent]) -> int:
5871
progress_report = ProgressReport(
5972
message="Repairing packages' metadata",
6073
code="repair.metadata",
61-
total=immediate_content.count(),
74+
total=content.count(),
6275
)
6376
progress_report.save()
6477
with progress_report:
6578
for package in progress_report.iter(
66-
immediate_content.prefetch_related("_artifacts").iterator(chunk_size=1000)
79+
immediate_content.iterator(chunk_size=1000)
6780
):
6881
new_data = artifact_to_python_content_data(
6982
package.filename, package._artifacts.get(), domain
@@ -82,6 +95,56 @@ def repair_metadata(content: QuerySet[PythonPackageContent]) -> int:
8295
batch = []
8396
set_of_update_fields.clear()
8497

98+
for package in progress_report.iter(
99+
on_demand_content.iterator(chunk_size=1000)
100+
):
101+
remote_artifacts = (
102+
package.contentartifact_set.get().remoteartifact_set.all()
103+
)
104+
# We expect that PythonPackageContent always has correct name and version
105+
try:
106+
json_data = fetch_json_release_metadata(
107+
package.name, package.version, remote_artifacts.get().remote)
108+
except RequestException as exc:
109+
log.warning(
110+
_("Could not fetch metadata for {} {} from PyPI. Error: {}").format(
111+
package.name, package.version, exc
112+
)
113+
)
114+
continue
115+
# Extract data only for the specific distribution we are currently checking
116+
# We expect that RemoteArtifact always has correct sha256
117+
dist_data = next(
118+
(
119+
dist
120+
for ra in remote_artifacts
121+
for dist in json_data["urls"]
122+
if ra.sha256 == dist["digests"]["sha256"]
123+
),
124+
None,
125+
)
126+
if not dist_data:
127+
log.warning(
128+
_("No matching distribution for {} was found.").format(package.name)
129+
)
130+
continue
131+
132+
new_data = parse_metadata(json_data["info"], package.version, dist_data)
133+
new_data.pop("url") # belongs to RemoteArtifact, not PythonPackageContent
134+
changed = False
135+
for field, value in new_data.items():
136+
if getattr(package, field) != value:
137+
setattr(package, field, value)
138+
set_of_update_fields.add(field)
139+
changed = True
140+
if changed:
141+
batch.append(package)
142+
if len(batch) == 1000:
143+
total_repaired += len(batch)
144+
PythonPackageContent.objects.bulk_update(batch, set_of_update_fields)
145+
batch = []
146+
set_of_update_fields.clear()
147+
85148
if batch:
86149
total_repaired += len(batch)
87150
PythonPackageContent.objects.bulk_update(batch, set_of_update_fields)

pulp_python/app/utils.py

Lines changed: 15 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,6 @@
11
import pkginfo
22
import re
3+
import requests
34
import shutil
45
import tempfile
56
import json
@@ -9,6 +10,7 @@
910
from packaging.utils import canonicalize_name
1011
from packaging.requirements import Requirement
1112
from packaging.version import parse, InvalidVersion
13+
from pulpcore.plugin.models import Remote
1214

1315

1416
PYPI_LAST_SERIAL = "X-PYPI-LAST-SERIAL"
@@ -189,6 +191,19 @@ def artifact_to_python_content_data(filename, artifact, domain=None):
189191
return data
190192

191193

194+
def fetch_json_release_metadata( name: str, version: str, remote: Remote) -> dict:
195+
"""
196+
Fetches metadata for a specific release from PyPI's JSON API. A release can contain
197+
multiple distributions. See https://docs.pypi.org/api/json/#get-a-release for more details.
198+
199+
Returns dict containing "info", "last_serial", "urls", and "vulnerabilities" keys.
200+
"""
201+
url = f"{remote.url}pypi/{name}/{version}/json"
202+
response = requests.get(url, timeout=10)
203+
response.raise_for_status()
204+
return response.json()
205+
206+
192207
def python_content_to_json(base_path, content_query, version=None, domain=None):
193208
"""
194209
Converts a QuerySet of PythonPackageContent into the PyPi JSON format

pulp_python/tests/functional/api/test_repair.py

Lines changed: 95 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -32,6 +32,49 @@ def _create(artifact_filename, filename, content_data):
3232
return _create
3333

3434

35+
@pytest.fixture
36+
def create_content_remote(python_bindings):
37+
def _create(filename, ra_url, ra_sha256, content_data, remote):
38+
commands = (
39+
"from pulpcore.plugin.models import ContentArtifact, RemoteArtifact; "
40+
"from pulpcore.plugin.util import extract_pk, get_url; "
41+
"from pulp_python.app.models import PythonPackageContent, PythonRemote; "
42+
f"c = PythonPackageContent(filename={filename!r}, **{content_data!r}); "
43+
"c.save(); "
44+
f"ca = ContentArtifact(artifact=None, content=c, relative_path={filename!r}); "
45+
"ca.save(); "
46+
f"r = PythonRemote.objects.get(pk=extract_pk({remote.pulp_href!r})); "
47+
f"ra = RemoteArtifact(content_artifact=ca, remote=r, sha256={ra_sha256!r}, url={ra_url!r}); " # noqa: E501
48+
"ra.save(); "
49+
"print(get_url(c))"
50+
)
51+
process = subprocess.run(
52+
["pulpcore-manager", "shell", "-c", commands], capture_output=True
53+
)
54+
55+
assert process.returncode == 0
56+
content_href = process.stdout.decode().strip()
57+
return python_bindings.ContentPackagesApi.read(content_href)
58+
59+
return _create
60+
61+
62+
@pytest.mark.django_db
63+
@pytest.fixture
64+
def delete_content():
65+
def _delete(content_href):
66+
from pulpcore.plugin.util import extract_pk
67+
from pulp_python.app.models import PythonPackageContent
68+
69+
content = PythonPackageContent.objects.get(pk=extract_pk(content_href))
70+
content.version_memberships.all().delete()
71+
artifacts = content._artifacts.all()
72+
content.delete()
73+
artifacts.delete()
74+
75+
return _delete
76+
77+
3578
@pytest.fixture
3679
def move_to_repository(python_bindings, monitor_task):
3780
def _move(repo_href, content_hrefs):
@@ -84,6 +127,7 @@ def test_metadata_repair_command(
84127

85128
def test_metadata_repair_endpoint(
86129
create_content_direct,
130+
delete_content,
87131
download_python_file,
88132
monitor_task,
89133
move_to_repository,
@@ -124,3 +168,54 @@ def test_metadata_repair_endpoint(
124168
assert content.packagetype == "sdist"
125169
assert content.requires_python == ">=2.7,!=3.0.*,!=3.1.*,!=3.2.*,!=3.3.*"
126170
assert content.author == ""
171+
delete_content(content.pulp_href)
172+
173+
174+
def test_metadata_repair_endpoint_on_demand(
175+
create_content_remote,
176+
delete_content,
177+
monitor_task,
178+
move_to_repository,
179+
python_bindings,
180+
python_remote_factory,
181+
python_repo_factory,
182+
):
183+
"""
184+
Test repairing of package metadata via `Repositories.repair_metadata` endpoint
185+
when only RemoteArtifact is present.
186+
"""
187+
python_egg_filename = "scipy-1.1.0.tar.gz"
188+
python_egg_url = urljoin(
189+
urljoin(PYTHON_FIXTURES_URL, "packages/"), python_egg_filename
190+
)
191+
python_egg_sha256 = (
192+
"878352408424dffaa695ffedf2f9f92844e116686923ed9aa8626fc30d32cfd1"
193+
)
194+
data = {
195+
"name": "scipy",
196+
"version": "1.1.0",
197+
# Wrong metadata
198+
"author": "ME",
199+
"packagetype": "bdist",
200+
"requires_python": ">=3.8",
201+
}
202+
remote = python_remote_factory(includes=["scipy"])
203+
repo = python_repo_factory(remote=remote)
204+
205+
content = create_content_remote(
206+
python_egg_filename, python_egg_url, python_egg_sha256, data, remote
207+
)
208+
for field, test_value in data.items():
209+
assert getattr(content, field) == test_value
210+
move_to_repository(repo.pulp_href, [content.pulp_href])
211+
212+
response = python_bindings.RepositoriesPythonApi.repair_metadata(repo.pulp_href)
213+
monitor_task(response.task)
214+
215+
new_content = python_bindings.ContentPackagesApi.read(content.pulp_href)
216+
assert new_content.author == ""
217+
assert new_content.name == "scipy"
218+
assert new_content.packagetype == "sdist"
219+
assert new_content.requires_python == ">=2.7,!=3.0.*,!=3.1.*,!=3.2.*,!=3.3.*"
220+
assert new_content.version == "1.1.0"
221+
delete_content(content.pulp_href)

0 commit comments

Comments
 (0)