bitforge/orchard
3
0
Fork 0
Code Projects Wiki Activity

Store SHA256 checksums with artifacts and add multiple hash support

Browse Source 
- Add sha256 field to API responses as explicit alias of artifact id
- Add checksum_sha1 and s3_etag fields to artifacts table
- Compute MD5, SHA1, and capture S3 ETag during upload
- Update StorageResult to return all checksums from storage layer
- Add migration 003_checksum_fields.sql for existing databases
- Add Dockerfile.local and docker-compose.local.yml for local development
- Update schemas to include all checksum fields in responses
This commit is contained in:
Mondo Diaz
2025-12-15 13:03:58 -06:00
parent 0eb2deb4ca
commit b124b94b56
9 changed files with 358 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
backend/app/routes.py
View File

@@ -520,40 +520,51 @@ def upload_artifact(
)
# Store file (uses multipart for large files)
sha256_hash, size, s3_key = storage.store(file.file, content_length)
storage_result = storage.store(file.file, content_length)
# Check if this is a deduplicated upload
deduplicated = False
# Create or update artifact record
artifact = db.query(Artifact).filter(Artifact.id == sha256_hash).first()
artifact = db.query(Artifact).filter(Artifact.id == storage_result.sha256).first()
if artifact:
artifact.ref_count += 1
deduplicated = True
# Merge metadata if new metadata was extracted
if file_metadata and artifact.format_metadata:
artifact.format_metadata = {**artifact.format_metadata, **file_metadata}
if file_metadata and artifact.artifact_metadata:
artifact.artifact_metadata = {**artifact.artifact_metadata, **file_metadata}
elif file_metadata:
artifact.format_metadata = file_metadata
artifact.artifact_metadata = file_metadata
# Update checksums if not already set
if not artifact.checksum_md5 and storage_result.md5:
artifact.checksum_md5 = storage_result.md5
if not artifact.checksum_sha1 and storage_result.sha1:
artifact.checksum_sha1 = storage_result.sha1
if not artifact.s3_etag and storage_result.s3_etag:
artifact.s3_etag = storage_result.s3_etag
else:
artifact = Artifact(
id=sha256_hash,
size=size,
id=storage_result.sha256,
size=storage_result.size,
content_type=file.content_type,
original_name=file.filename,
checksum_md5=storage_result.md5,
checksum_sha1=storage_result.sha1,
s3_etag=storage_result.s3_etag,
created_by=user_id,
s3_key=s3_key,
format_metadata=file_metadata or {},
s3_key=storage_result.s3_key,
artifact_metadata=file_metadata or {},
)
db.add(artifact)
# Record upload
upload = Upload(
artifact_id=sha256_hash,
artifact_id=storage_result.sha256,
package_id=package.id,
original_name=file.filename,
uploaded_by=user_id,
source_ip=request.client.host if request.client else None,
deduplicated=deduplicated,
)
db.add(upload)
@@ -561,13 +572,13 @@ def upload_artifact(
if tag:
existing_tag = db.query(Tag).filter(Tag.package_id == package.id, Tag.name == tag).first()
if existing_tag:
existing_tag.artifact_id = sha256_hash
existing_tag.artifact_id = storage_result.sha256
existing_tag.created_by = user_id
else:
new_tag = Tag(
package_id=package.id,
name=tag,
artifact_id=sha256_hash,
artifact_id=storage_result.sha256,
created_by=user_id,
)
db.add(new_tag)
@@ -575,12 +586,16 @@ def upload_artifact(
db.commit()
return UploadResponse(
artifact_id=sha256_hash,
size=size,
artifact_id=storage_result.sha256,
sha256=storage_result.sha256,
size=storage_result.size,
project=project_name,
package=package_name,
tag=tag,
format_metadata=artifact.format_metadata,
checksum_md5=storage_result.md5,
checksum_sha1=storage_result.sha1,
s3_etag=storage_result.s3_etag,
format_metadata=artifact.artifact_metadata,
deduplicated=deduplicated,
)