def dispatch(cls, collection_id, meta):
local_path = archive.load_file(meta)
try:
best_cls = cls.auction_file(meta, local_path)
log.debug("Dispatching %r to %r", meta.file_name, best_cls)
best_cls(collection_id).ingest(meta, local_path)
db.session.commit()
except Exception as exc:
cls.handle_exception(meta, collection_id, exc)
finally:
archive.cleanup_file(meta)
def write_document(export_dir, zf, collection, entity):
content_hash = entity.first("contentHash", quiet=True)
if content_hash is None:
return
file_name = entity_filename(entity)
arcname = "{0}-{1}".format(entity.id, file_name)
arcname = os.path.join(collection.get("label"), arcname)
try:
local_path = archive.load_file(content_hash, temp_path=export_dir)
if local_path is not None and os.path.exists(local_path):
zf.write(local_path, arcname=arcname)
finally:
archive.cleanup_file(content_hash, temp_path=export_dir)
def retrieve():
"""Downloads a binary blob from the blob storage archive.
---
get:
summary: Download a blob from the archive
parameters:
- description: Authorization token for an archive blob
in: query
name: claim
schema:
type: string
description: A signed JWT with the object hash.
responses:
'200':
description: OK
content:
'*/*': {}
'404':
description: Object does not exist.
tags:
- Archive
"""
token = request.args.get("token")
token = jwt.decode(token, key=settings.SECRET_KEY, verify=True)
content_hash = token.get("c")
file_name = token.get("f")
mime_type = token.get("m")
expire = datetime.utcfromtimestamp(token["exp"])
tag_request(content_hash=content_hash, file_name=file_name)
url = archive.generate_url(
content_hash,
file_name=file_name,
mime_type=mime_type,
expire=expire,
)
if url is not None:
return redirect(url)
try:
local_path = archive.load_file(content_hash)
if local_path is None:
return Response(status=404)
return send_file(
str(local_path),
as_attachment=True,
conditional=True,
attachment_filename=file_name,
mimetype=mime_type,
)
finally:
archive.cleanup_file(content_hash)
def retrieve():
"""Downloads a binary blob from the blob storage archive.
---
get:
summary: Download a blob from the archive
parameters:
- description: Authorization token for an archive blob
in: query
name: claim
schema:
type: string
description: A signed JWT with the object hash.
responses:
'200':
description: OK
content:
'*/*': {}
'404':
description: Object does not exist.
tags:
- Archive
"""
claim = request.args.get("claim")
role_id, content_hash, file_name, mime_type = archive_claim(claim)
require(request.authz.id == role_id)
tag_request(content_hash=content_hash, file_name=file_name)
url = archive.generate_url(content_hash,
file_name=file_name,
mime_type=mime_type)
if url is not None:
return redirect(url)
try:
local_path = archive.load_file(content_hash)
if local_path is None:
return Response(status=404)
return send_file(
str(local_path),
as_attachment=True,
conditional=True,
attachment_filename=file_name,
mimetype=mime_type,
)
finally:
archive.cleanup_file(content_hash)
def _serve_archive(content_hash, file_name, mime_type):
"""Serve a file from the archive or by generating an external URL."""
url = archive.generate_url(content_hash,
file_name=file_name,
mime_type=mime_type)
if url is not None:
return redirect(url)
try:
local_path = archive.load_file(content_hash, file_name=file_name)
if local_path is None:
return Response(status=404)
return send_file(local_path,
as_attachment=True,
conditional=True,
attachment_filename=file_name,
mimetype=mime_type)
finally:
archive.cleanup_file(content_hash)
def _serve_archive(content_hash, file_name, mime_type):
"""Serve a file from the archive or by generating an external URL."""
url = archive.generate_url(content_hash,
file_name=file_name,
mime_type=mime_type)
if url is not None:
return redirect(url)
enable_cache()
try:
local_path = archive.load_file(content_hash, file_name=file_name)
if local_path is None:
raise NotFound("File does not exist.")
return send_file(open(local_path, 'rb'),
as_attachment=True,
attachment_filename=file_name,
mimetype=mime_type)
finally:
archive.cleanup_file(content_hash)
def dispatch(cls, source_id, meta):
best_score, best_cls = 0, None
local_path = archive.load_file(meta)
try:
for cls in get_ingestors().values():
score = cls.match(meta, local_path)
if score > best_score:
best_score = score
best_cls = cls
if best_cls is None:
log.debug("No ingestor found for: %r", meta.file_name)
return
log.debug("Dispatching %r to %r", meta.file_name,
best_cls.__name__)
best_cls(source_id).ingest(meta, local_path)
except Exception as ex:
log.exception(ex)
db.session.rollback()
finally:
archive.cleanup_file(meta)
def retrieve():
claim = request.args.get('claim')
role_id, content_hash, file_name, mime_type = archive_claim(claim)
require(request.authz.id == role_id)
tag_request(content_hash=content_hash, file_name=file_name)
url = archive.generate_url(content_hash,
file_name=file_name,
mime_type=mime_type)
if url is not None:
return redirect(url)
try:
local_path = archive.load_file(content_hash)
if local_path is None:
return Response(status=404)
return send_file(str(local_path),
as_attachment=True,
conditional=True,
attachment_filename=file_name,
mimetype=mime_type)
finally:
archive.cleanup_file(content_hash)
def retrieve():
claim = request.args.get('claim')
role_id, content_hash, file_name, mime_type = archive_claim(claim)
require(request.authz.id == role_id)
record_audit(Audit.ACT_ARCHIVE, content_hash=content_hash)
tag_request(content_hash=content_hash, file_name=file_name)
url = archive.generate_url(content_hash,
file_name=file_name,
mime_type=mime_type)
if url is not None:
return redirect(url)
try:
local_path = archive.load_file(content_hash)
if local_path is None:
return Response(status=404)
return send_file(local_path,
as_attachment=True,
conditional=True,
attachment_filename=file_name,
mimetype=mime_type)
finally:
archive.cleanup_file(content_hash)
