def _import_and_find_dupes( self ):
phash = os.urandom( 8 )
# fake-import the files with the phash
( size, mime, width, height, duration, num_frames, has_audio, num_words ) = ( 65535, HC.IMAGE_JPEG, 640, 480, None, None, False, None )
file_import_options = HG.client_controller.new_options.GetDefaultFileImportOptions( 'loud' )
for hash in self._all_hashes:
fake_file_import_job = ClientImportFiles.FileImportJob( 'fake path', file_import_options )
fake_file_import_job._pre_import_file_status = ClientImportFiles.FileImportStatus( CC.STATUS_UNKNOWN, hash )
fake_file_import_job._file_info = ( size, mime, width, height, duration, num_frames, has_audio, num_words )
fake_file_import_job._extra_hashes = ( b'abcd', b'abcd', b'abcd' )
fake_file_import_job._phashes = [ phash ]
fake_file_import_job._file_import_options = FileImportOptions.FileImportOptions()
self._write( 'import_file', fake_file_import_job )
# run search maintenance
self._write( 'maintain_similar_files_tree' )
self._write( 'maintain_similar_files_search_for_potential_duplicates', 0 )
def _do_fake_imports(self):
self._md5_to_sha256 = {}
self._sha256_to_md5 = {}
self._sha256_to_sha1 = {}
self._my_files_sha256 = set()
self._hashes_to_current_tags = {}
self._hashes_to_pending_tags = {}
self._hashes_to_deleted_tags = {}
(size, mime, width, height, duration, num_frames, has_audio,
num_words) = (65535, HC.IMAGE_JPEG, 640, 480, None, None, False, None)
file_import_options = HG.client_controller.new_options.GetDefaultFileImportOptions(
'loud')
for i in range(100):
hash = HydrusData.GenerateKey()
md5 = os.urandom(16)
sha1 = os.urandom(20)
sha512 = os.urandom(64)
self._md5_to_sha256[md5] = hash
self._sha256_to_md5[hash] = md5
self._sha256_to_sha1[hash] = sha1
self._hashes_to_current_tags[hash] = set(
random.sample(current_tag_pool, 3))
self._hashes_to_pending_tags[hash] = set(
random.sample(pending_tag_pool, 3))
self._hashes_to_deleted_tags[hash] = set(
random.sample(deleted_tag_pool, 3))
if i < 50:
fake_file_import_job = ClientImportFiles.FileImportJob(
'fake path', file_import_options)
fake_file_import_job._pre_import_file_status = ClientImportFiles.FileImportStatus(
CC.STATUS_UNKNOWN, hash)
fake_file_import_job._file_info = (size, mime, width, height,
duration, num_frames,
has_audio, num_words)
fake_file_import_job._extra_hashes = (md5, sha1, sha512)
fake_file_import_job._perceptual_hashes = [os.urandom(8)]
fake_file_import_job._file_import_options = FileImportOptions.FileImportOptions(
)
self.WriteSynchronous('import_file', fake_file_import_job)
self._my_files_sha256.add(hash)
def WriteSynchronous(self, name, *args, **kwargs):
self._write_call_args[name].append((args, kwargs))
if name == 'import_file':
(file_import_job, ) = args
if file_import_job.GetHash().hex(
) == 'a593942cb7ea9ffcd8ccf2f0fa23c338e23bfecd9a3e508dfc0bcf07501ead08': # 'blarg' in sha256 hex
raise Exception('File failed to import for some reason!')
else:
h = file_import_job.GetHash()
if h is None:
h = os.urandom(32)
return ClientImportFiles.FileImportStatus(
CC.STATUS_SUCCESSFUL_AND_NEW, h, note='test note')
def GetHashIdStatus(self,
hash_id,
prefix='') -> ClientImportFiles.FileImportStatus:
if prefix != '':
prefix += ': '
hash = self.modules_hashes_local_cache.GetHash(hash_id)
(is_deleted, timestamp,
file_deletion_reason) = self.modules_files_storage.GetDeletionStatus(
self.modules_services.combined_local_file_service_id, hash_id)
if is_deleted:
if timestamp is None:
note = 'Deleted from the client before delete times were tracked ({}).'.format(
file_deletion_reason)
else:
note = 'Deleted from the client {} ({}), which was {} before this check.'.format(
HydrusData.ConvertTimestampToPrettyTime(timestamp),
file_deletion_reason,
HydrusData.BaseTimestampToPrettyTimeDelta(timestamp))
return ClientImportFiles.FileImportStatus(CC.STATUS_DELETED,
hash,
note=prefix + note)
result = self.modules_files_storage.GetCurrentTimestamp(
self.modules_services.trash_service_id, hash_id)
if result is not None:
timestamp = result
note = 'Currently in trash ({}). Sent there at {}, which was {} before this check.'.format(
file_deletion_reason,
HydrusData.ConvertTimestampToPrettyTime(timestamp),
HydrusData.BaseTimestampToPrettyTimeDelta(
timestamp, just_now_threshold=0))
return ClientImportFiles.FileImportStatus(CC.STATUS_DELETED,
hash,
note=prefix + note)
result = self.modules_files_storage.GetCurrentTimestamp(
self.modules_services.combined_local_file_service_id, hash_id)
if result is not None:
timestamp = result
mime = self.modules_files_metadata_basic.GetMime(hash_id)
note = 'Imported at {}, which was {} before this check.'.format(
HydrusData.ConvertTimestampToPrettyTime(timestamp),
HydrusData.BaseTimestampToPrettyTimeDelta(
timestamp, just_now_threshold=0))
return ClientImportFiles.FileImportStatus(
CC.STATUS_SUCCESSFUL_BUT_REDUNDANT,
hash,
mime=mime,
note=prefix + note)
return ClientImportFiles.FileImportStatus(CC.STATUS_UNKNOWN, hash)