Fix httpx.Timeout configuration in PyPI proxy

This commit is contained in:
Mondo Diaz
2026-02-05 10:31:04 -06:00
parent 11c5aee0f1
commit dfa089376a
76 changed files with 9384 additions and 4407 deletions

View File

@@ -6,9 +6,8 @@ from typing import List, Optional, Tuple
from sqlalchemy.orm import Session
import logging
from ..models import Artifact, Tag
from ..models import Artifact, PackageVersion
from ..repositories.artifact import ArtifactRepository
from ..repositories.tag import TagRepository
from ..storage import S3Storage
logger = logging.getLogger(__name__)
@@ -21,8 +20,8 @@ class ArtifactCleanupService:
Reference counting rules:
- ref_count starts at 1 when artifact is first uploaded
- ref_count increments when the same artifact is uploaded again (deduplication)
- ref_count decrements when a tag is deleted or updated to point elsewhere
- ref_count decrements when a package is deleted (for each tag pointing to artifact)
- ref_count decrements when a version is deleted or updated to point elsewhere
- ref_count decrements when a package is deleted (for each version pointing to artifact)
- When ref_count reaches 0, artifact is a candidate for deletion from S3
"""
@@ -30,12 +29,11 @@ class ArtifactCleanupService:
self.db = db
self.storage = storage
self.artifact_repo = ArtifactRepository(db)
self.tag_repo = TagRepository(db)
def on_tag_deleted(self, artifact_id: str) -> Artifact:
def on_version_deleted(self, artifact_id: str) -> Artifact:
"""
Called when a tag is deleted.
Decrements ref_count for the artifact the tag was pointing to.
Called when a version is deleted.
Decrements ref_count for the artifact the version was pointing to.
"""
artifact = self.artifact_repo.get_by_sha256(artifact_id)
if artifact:
@@ -45,11 +43,11 @@ class ArtifactCleanupService:
)
return artifact
def on_tag_updated(
def on_version_updated(
self, old_artifact_id: str, new_artifact_id: str
) -> Tuple[Optional[Artifact], Optional[Artifact]]:
"""
Called when a tag is updated to point to a different artifact.
Called when a version is updated to point to a different artifact.
Decrements ref_count for old artifact, increments for new (if different).
Returns (old_artifact, new_artifact) tuple.
@@ -79,21 +77,21 @@ class ArtifactCleanupService:
def on_package_deleted(self, package_id) -> List[str]:
"""
Called when a package is deleted.
Decrements ref_count for all artifacts that had tags in the package.
Decrements ref_count for all artifacts that had versions in the package.
Returns list of artifact IDs that were affected.
"""
# Get all tags in the package before deletion
tags = self.db.query(Tag).filter(Tag.package_id == package_id).all()
# Get all versions in the package before deletion
versions = self.db.query(PackageVersion).filter(PackageVersion.package_id == package_id).all()
affected_artifacts = []
for tag in tags:
artifact = self.artifact_repo.get_by_sha256(tag.artifact_id)
for version in versions:
artifact = self.artifact_repo.get_by_sha256(version.artifact_id)
if artifact:
self.artifact_repo.decrement_ref_count(artifact)
affected_artifacts.append(tag.artifact_id)
affected_artifacts.append(version.artifact_id)
logger.info(
f"Decremented ref_count for artifact {tag.artifact_id} (package delete)"
f"Decremented ref_count for artifact {version.artifact_id} (package delete)"
)
return affected_artifacts
@@ -152,7 +150,7 @@ class ArtifactCleanupService:
def verify_ref_counts(self, fix: bool = False) -> List[dict]:
"""
Verify that ref_counts match actual tag references.
Verify that ref_counts match actual version references.
Args:
fix: If True, fix any mismatched ref_counts
@@ -162,28 +160,28 @@ class ArtifactCleanupService:
"""
from sqlalchemy import func
# Get actual tag counts per artifact
tag_counts = (
self.db.query(Tag.artifact_id, func.count(Tag.id).label("tag_count"))
.group_by(Tag.artifact_id)
# Get actual version counts per artifact
version_counts = (
self.db.query(PackageVersion.artifact_id, func.count(PackageVersion.id).label("version_count"))
.group_by(PackageVersion.artifact_id)
.all()
)
tag_count_map = {artifact_id: count for artifact_id, count in tag_counts}
version_count_map = {artifact_id: count for artifact_id, count in version_counts}
# Check all artifacts
artifacts = self.db.query(Artifact).all()
mismatches = []
for artifact in artifacts:
actual_count = tag_count_map.get(artifact.id, 0)
actual_count = version_count_map.get(artifact.id, 0)
# ref_count should be at least 1 (initial upload) + additional uploads
# But tags are the primary reference, so we check against tag count
# But versions are the primary reference, so we check against version count
if artifact.ref_count < actual_count:
mismatch = {
"artifact_id": artifact.id,
"stored_ref_count": artifact.ref_count,
"actual_tag_count": actual_count,
"actual_version_count": actual_count,
}
mismatches.append(mismatch)