def _migrate_resource(self, contribution, revision, resource, created_dt,
ignored_checksums):
storage_backend, storage_path, size, md5 = self._get_local_file_info(
resource, force_access=True)
content_type = mimetypes.guess_type(
resource.fileName)[0] or 'application/octet-stream'
if not storage_path:
self.print_error("%[red!]File not accessible [{}]".format(
convert_to_unicode(resource.fileName)))
return
paper_file = PaperFile(filename=resource.fileName,
content_type=content_type,
size=size,
md5=md5,
storage_backend=storage_backend,
storage_file_id=storage_path,
created_dt=created_dt)
# check whether the same file has been uploaded to a subsequent revision
if md5:
self.checksum_map[md5] = paper_file
collision = self.file_checksums.get(md5)
if collision:
ignored_checksums.add(md5)
self.print_warning(
'%[yellow!]File {} (rev. {}) already in revision {}'.
format(convert_to_unicode(resource.fileName),
revision.id if revision else None, collision.id))
return
else:
self.file_checksums[md5] = revision
else:
self.print_error(
"%[red!]File not accessible; can't MD5 it [{}]".format(
convert_to_unicode(paper_file.filename)))
paper_file._contribution = contribution
paper_file.paper_revision = revision
db.session.add(paper_file)
return paper_file
def _migrate_resource(self, old_contrib, contribution, revision, resource,
created_dt, ignored_checksums):
storage_backend, storage_path, size = self._get_local_file_info(
resource)
content_type = mimetypes.guess_type(
resource.fileName)[0] or 'application/octet-stream'
paper_file = PaperFile(filename=resource.fileName,
content_type=content_type,
size=size,
storage_backend=storage_backend,
storage_file_id=storage_path,
created_dt=created_dt,
paper_revision=revision)
# check whether the same file has been uploaded to a subsequent revision
try:
with paper_file.open() as f:
checksum = crc32(f.read())
self.checksum_map[checksum] = paper_file
collision = self.file_checksums.get(checksum)
if collision:
ignored_checksums.add(checksum)
self.print_warning(
'%[yellow!]File {} (rev. {}) already in revision {}'.
format(resource.filename,
revision.id if revision else None, collision.id))
return
else:
self.file_checksums[checksum] = revision
except (RuntimeError, StorageError):
self.print_error(
"%[red!]File not accessible; can't CRC it [{}]".format(
convert_to_unicode(paper_file.filename)))
paper_file._contribution = contribution
db.session.add(paper_file)
return paper_file