def __init__(self, create_table_if_not_exists: bool) -> None:
self.cache_dir = get_program_cache_directory(create_if_not_exist=True)
self.valid = self.cache_dir is not None
if not self.valid:
return
assert self.cache_dir is not None
self.cache_dir = os.path.join(self.cache_dir, 'thumbnails')
try:
if not os.path.exists(self.cache_dir):
os.makedirs(self.cache_dir, 0o700)
logging.debug("Created thumbnails cache %s", self.cache_dir)
elif not os.path.isdir(self.cache_dir):
os.remove(self.cache_dir)
logging.warning("Removed file %s", self.cache_dir)
os.makedirs(self.cache_dir, 0o700)
logging.debug("Created thumbnails cache %s", self.cache_dir)
except:
logging.error(
"Failed to create Rapid Photo Downloader Thumbnail Cache at %s",
self.cache_dir)
self.valid = False
self.cache_dir = None
self.random_filename = None
self.fs_encoding = None
else:
self.random_filename = GenerateRandomFileName()
self.md5 = MD5Name()
self.thumb_db = CacheSQL(self.cache_dir,
create_table_if_not_exists)
def __init__(self, cache_dir: str, failure_dir: Optional[str]) -> None:
"""
Create cache if it doesn't exist; checks validity.
:param cache_dir: full path of the directory into which
thumbnails will be saved / read.
:param failure_dir: full path of the directory into which
failed thumbnails will be saved / read (thumbnails that could
not be generated)
"""
assert sys.platform.startswith('linux')
self.cache_dir = cache_dir
self.failure_dir = failure_dir
assert self.cache_dir
self.valid = self._create_directory(self.cache_dir,
"Freedesktop.org thumbnail")
if self.valid:
self.random_filename = GenerateRandomFileName()
self.md5 = MD5Name()
if self.failure_dir is not None:
self.valid = self._create_directory(self.failure_dir,
"thumbnails failure")
if not self.valid:
self.random_filename = self.fs_encoding = None
def __init__(self) -> None:
self.random_file_name = GenerateRandomFileName()
super().__init__("Thumbnails")
class GenerateThumbnails(WorkerInPublishPullPipeline):
def __init__(self) -> None:
self.random_file_name = GenerateRandomFileName()
super().__init__("Thumbnails")
def cache_full_size_file_from_camera(self, rpd_file: RPDFile) -> bool:
"""
Get the file from the camera chunk by chunk and cache it.
:return: True if operation succeeded, False otherwise
"""
if rpd_file.file_type == FileType.photo:
cache_dir = self.photo_cache_dir
else:
cache_dir = self.video_cache_dir
cache_full_file_name = os.path.join(
cache_dir,
self.random_file_name.name(extension=rpd_file.extension))
try:
self.camera.save_file_by_chunks(
dir_name=rpd_file.path,
file_name=rpd_file.name,
size=rpd_file.size,
dest_full_filename=cache_full_file_name,
progress_callback=None,
check_for_command=self.check_for_controller_directive,
return_file_bytes=False,
)
except CameraProblemEx as e:
# TODO report error
return False
else:
rpd_file.cache_full_file_name = cache_full_file_name
return True
def cache_file_chunk_from_camera(self, rpd_file: RPDFile,
offset: int) -> bool:
if rpd_file.file_type == FileType.photo:
cache_dir = self.photo_cache_dir
else:
cache_dir = self.video_cache_dir
cache_full_file_name = os.path.join(
cache_dir,
self.random_file_name.name(extension=rpd_file.extension))
try:
self.camera.save_file_chunk(
dir_name=rpd_file.path,
file_name=rpd_file.name,
chunk_size_in_bytes=min(offset, rpd_file.size),
dest_full_filename=cache_full_file_name,
)
rpd_file.temp_cache_full_file_chunk = cache_full_file_name
return True
except CameraProblemEx as e:
# TODO problem reporting
return False
def extract_photo_video_from_camera(
self,
rpd_file: RPDFile,
entire_file_required: bool,
full_file_name_to_work_on,
using_exiftool: bool,
) -> Tuple[ExtractionTask, str, bool]:
"""
Extract part of a photo of video to be able to get the orientation
and date time metadata, if and only if we know how much of the file
is needed to get the thumbnail.
Otherwise, download the entire photo or video from the camera to be able
to generate the thumbnail and cache it.
:param rpd_file: photo or video
:param entire_file_required: whether we already know (from scanning) that
the entire file is required
:param full_file_name_to_work_on: file name and path of the photo or video
:param using_exiftool: if all the metadata extraction is done using ExifTool
:return: extraction task, full file name, and whether the full file name
refers to a temporary file that should be deleted
"""
task = ExtractionTask.undetermined
file_to_work_on_is_temporary = False
if rpd_file.is_mtp_device and rpd_file.file_type == FileType.video:
entire_file_required = True
if not entire_file_required:
# For many photos videos, extract a small part of the file and use
# that to get the metadata
if using_exiftool:
offset = thumbnail_offset_exiftool.get(rpd_file.extension)
else:
offset = thumbnail_offset.get(rpd_file.extension)
if offset:
if using_exiftool:
offset = max(
offset,
datetime_offset_exiftool.get(rpd_file.extension))
else:
offset = max(offset,
datetime_offset.get(rpd_file.extension))
if offset and self.cache_file_chunk_from_camera(rpd_file, offset):
if rpd_file.file_type == FileType.photo:
task = ExtractionTask.load_from_bytes_metadata_from_temp_extract
else:
task = ExtractionTask.extract_from_file_and_load_metadata
file_to_work_on_is_temporary = True
full_file_name_to_work_on = rpd_file.temp_cache_full_file_chunk
if task == ExtractionTask.undetermined:
if self.cache_full_size_file_from_camera(rpd_file):
task = ExtractionTask.extract_from_file_and_load_metadata
full_file_name_to_work_on = rpd_file.cache_full_file_name
else:
# Failed to generate thumbnail
task = ExtractionTask.bypass
return task, full_file_name_to_work_on, file_to_work_on_is_temporary
def do_work(self) -> None:
try:
self.generate_thumbnails()
except SystemExit as e:
sys.exit(e)
except Exception:
if hasattr(self, "device_name"):
logging.error("Exception generating thumbnails for %s",
self.device_name)
else:
logging.error("Exception generating thumbnails")
logging.exception("Traceback:")
def generate_thumbnails(self) -> None:
self.camera = None
arguments = pickle.loads(
self.content) # type: GenerateThumbnailsArguments
self.device_name = arguments.name
logging.info("Generating %s thumbnails for %s",
len(arguments.rpd_files), arguments.name)
if arguments.log_gphoto2:
self.gphoto2_logging = gphoto2_python_logging()
self.frontend = self.context.socket(zmq.PUSH)
self.frontend.connect("tcp://localhost:{}".format(
arguments.frontend_port))
self.prefs = Preferences()
# Whether we must use ExifTool to read photo metadata
force_exiftool = self.prefs.force_exiftool
# If the entire photo or video is required to extract the thumbnail, which is
# determined when extracting sample metadata from a photo or video during the
# device scan
entire_photo_required = arguments.entire_photo_required
entire_video_required = arguments.entire_video_required
# Access and generate Rapid Photo Downloader thumbnail cache
use_thumbnail_cache = self.prefs.use_thumbnail_cache
thumbnail_caches = GetThumbnailFromCache(
use_thumbnail_cache=use_thumbnail_cache)
photo_cache_dir = video_cache_dir = None
cache_file_from_camera = force_exiftool
rpd_files = arguments.rpd_files
# with open('tests/thumbnail_data_medium_no_tiff', 'wb') as f:
# pickle.dump(rpd_files, f)
# Must sort files by modification time prior to temporal analysis needed to figure out
# which thumbnails to prioritize
rpd_files = sorted(rpd_files, key=attrgetter("modification_time"))
time_span = arguments.proximity_seconds
rpd_files2 = []
if rpd_files:
gaps, sequences = get_temporal_gaps_and_sequences(
rpd_files, time_span)
rpd_files2.extend(gaps)
indexes = split_indexes(len(sequences))
rpd_files2.extend([sequences[idx] for idx in indexes])
assert len(rpd_files) == len(rpd_files2)
rpd_files = rpd_files2
if arguments.camera is not None:
self.camera = Camera(
model=arguments.camera,
port=arguments.port,
is_mtp_device=arguments.is_mtp_device,
specific_folders=self.prefs.folders_to_scan,
)
if not self.camera.camera_initialized:
# There is nothing to do here: exit!
logging.debug(
"Prematurely exiting thumbnail generation due to lack of access to "
"camera %s",
arguments.camera,
)
self.content = pickle.dumps(
GenerateThumbnailsResults(
scan_id=arguments.scan_id,
camera_removed=True,
),
pickle.HIGHEST_PROTOCOL,
)
self.send_message_to_sink()
self.disconnect_logging()
self.send_finished_command()
sys.exit(0)
if not cache_file_from_camera:
for rpd_file in rpd_files:
if use_exiftool_on_photo(
rpd_file.extension,
preview_extraction_irrelevant=False):
cache_file_from_camera = True
break
must_make_cache_dirs = (not self.camera.can_fetch_thumbnails
or cache_file_from_camera)
if (must_make_cache_dirs or arguments.need_video_cache_dir
or arguments.need_photo_cache_dir):
# If downloading complete copy of the files to
# generate previews, then may as well cache them to speed up
# the download process
self.photo_cache_dir = create_temp_dir(
folder=arguments.cache_dirs.photo_cache_dir,
prefix=cache_dir_name(self.device_name),
)
self.video_cache_dir = create_temp_dir(
folder=arguments.cache_dirs.video_cache_dir,
prefix=cache_dir_name(self.device_name),
)
cache_dirs = CacheDirs(self.photo_cache_dir,
self.video_cache_dir)
self.content = pickle.dumps(
GenerateThumbnailsResults(scan_id=arguments.scan_id,
cache_dirs=cache_dirs),
pickle.HIGHEST_PROTOCOL,
)
self.send_message_to_sink()
from_thumb_cache = 0
from_fdo_cache = 0
if self.camera:
rescan = RescanCamera(camera=self.camera, prefs=self.prefs)
rescan.rescan_camera(rpd_files)
rpd_files = rescan.rpd_files
if rescan.missing_rpd_files:
logging.error(
"%s files could not be relocated on %s",
len(rescan.missing_rpd_files),
self.camera.display_name,
)
for rpd_file in rescan.missing_rpd_files: # type: RPDFile
self.content = pickle.dumps(
GenerateThumbnailsResults(rpd_file=rpd_file,
thumbnail_bytes=None),
pickle.HIGHEST_PROTOCOL,
)
self.send_message_to_sink()
for rpd_file in rpd_files: # type: RPDFile
# Check to see if the process has received a command
self.check_for_controller_directive()
exif_buffer = None
file_to_work_on_is_temporary = False
secondary_full_file_name = ""
processing = set() # type: Set[ExtractionProcessing]
# Attempt to get thumbnail from Thumbnail Cache
# (see cache.py for definitions of various caches)
cache_search = thumbnail_caches.get_from_cache(rpd_file)
task, thumbnail_bytes, full_file_name_to_work_on, origin = cache_search
if task != ExtractionTask.undetermined:
if origin == ThumbnailCacheOrigin.thumbnail_cache:
from_thumb_cache += 1
else:
assert origin == ThumbnailCacheOrigin.fdo_cache
logging.debug(
"Thumbnail for %s found in large FDO cache",
rpd_file.full_file_name,
)
from_fdo_cache += 1
processing.add(ExtractionProcessing.resize)
if not rpd_file.mdatatime:
# Since we're extracting the thumbnail from the FDO cache,
# need to grab its metadata too.
# Reassign the task
task = ExtractionTask.load_file_directly_metadata_from_secondary
# It's not being downloaded from a camera, so nothing
# special to do except assign the name of the file from which
# to extract the metadata
secondary_full_file_name = rpd_file.full_file_name
logging.debug(
"Although thumbnail found in the cache, tasked to extract "
"file time recorded in metadata from %s",
secondary_full_file_name,
)
if task == ExtractionTask.undetermined:
# Thumbnail was not found in any cache: extract it
if self.camera: # type: Camera
if rpd_file.file_type == FileType.photo:
if rpd_file.is_heif():
# Load HEIF / HEIC using entire file.
# We are assuming that there is no tool to extract a
# preview image from an HEIF / HEIC, or the file simply
# does not have one to extract.
if self.cache_full_size_file_from_camera(rpd_file):
task = ExtractionTask.load_heif_and_exif_directly
processing.add(ExtractionProcessing.resize)
full_file_name_to_work_on = (
rpd_file.cache_full_file_name)
# For now, do not orient, as it seems pyheif or libheif does
# that automatically.
# processing.add(ExtractionProcessing.orient)
elif self.camera.can_fetch_thumbnails:
task = ExtractionTask.load_from_bytes
if rpd_file.is_jpeg_type():
# gPhoto2 knows how to get jpeg thumbnails
try:
thumbnail_bytes = self.camera.get_thumbnail(
rpd_file.path, rpd_file.name)
except CameraProblemEx as e:
# TODO handle error?
thumbnail_bytes = None
else:
if force_exiftool or use_exiftool_on_photo(
rpd_file.extension,
preview_extraction_irrelevant=False,
):
(
task,
full_file_name_to_work_on,
file_to_work_on_is_temporary,
) = self.extract_photo_video_from_camera(
rpd_file,
entire_photo_required,
full_file_name_to_work_on,
True,
)
if (task == ExtractionTask.
load_from_bytes_metadata_from_temp_extract
):
secondary_full_file_name = (
full_file_name_to_work_on)
file_to_work_on_is_temporary = False
else:
# gPhoto2 does not know how to get RAW thumbnails,
# so we do that part ourselves
if rpd_file.extension == "crw":
# Could cache this file, since reading its
# entirety But does anyone download a CRW file
# from the camera these days?!
bytes_to_read = rpd_file.size
else:
bytes_to_read = min(
rpd_file.size,
orientation_offset.get(
rpd_file.extension, 500),
)
exif_buffer = self.camera.get_exif_extract(
rpd_file.path, rpd_file.name,
bytes_to_read)
try:
thumbnail_bytes = self.camera.get_thumbnail(
rpd_file.path, rpd_file.name)
except CameraProblemEx as e:
# TODO report error
thumbnail_bytes = None
processing.add(
ExtractionProcessing.strip_bars_photo)
processing.add(ExtractionProcessing.orient)
else:
# Many (all?) jpegs from phones don't include jpeg previews,
# so need to render from the entire jpeg itself. Slow!
# For raw, extract merely a part of phone's raw format, and
# try to extract the jpeg preview from it (which probably
# doesn't exist!). This is fast.
if not rpd_file.is_jpeg():
bytes_to_read = thumbnail_offset.get(
rpd_file.extension)
if bytes_to_read:
exif_buffer = self.camera.get_exif_extract(
rpd_file.path, rpd_file.name,
bytes_to_read)
task = ExtractionTask.load_from_exif_buffer
processing.add(ExtractionProcessing.orient)
if (task == ExtractionTask.undetermined
and self.cache_full_size_file_from_camera(
rpd_file)):
if rpd_file.is_jpeg():
task = ExtractionTask.load_file_and_exif_directly
processing.add(ExtractionProcessing.resize)
processing.add(ExtractionProcessing.orient)
else:
task = ExtractionTask.load_from_exif
processing.add(ExtractionProcessing.resize)
processing.add(ExtractionProcessing.orient)
full_file_name_to_work_on = (
rpd_file.cache_full_file_name)
else:
# Failed to generate thumbnail
task = ExtractionTask.bypass
else:
# video from camera
if rpd_file.thm_full_name is not None:
# Fortunately, we have a special video thumbnail file
# Still need to get metadata time, however.
if entire_video_required:
offset = rpd_file.size
else:
offset = datetime_offset.get(
rpd_file.extension)
# If there is no offset, there is no point trying to
# extract the metadata time from part of the video. It's
# not ideal, but if this is from a camera on which there
# were any other files we can assume we've got a
# somewhat accurate date time for it from the
# modification time. The only exception is if the video
# file is not that big, in which case it's worth reading
# in its entirety:
if offset is None and rpd_file.size < 4000000:
offset = rpd_file.size
if rpd_file.mdatatime or not offset:
task = ExtractionTask.load_from_bytes
elif self.cache_file_chunk_from_camera(
rpd_file, offset):
task = (
ExtractionTask.
load_from_bytes_metadata_from_temp_extract)
secondary_full_file_name = (
rpd_file.temp_cache_full_file_chunk)
else:
# For some reason was unable to download part of the
# video file
task = ExtractionTask.load_from_bytes
try:
thumbnail_bytes = self.camera.get_THM_file(
rpd_file.thm_full_name)
except CameraProblemEx as e:
# TODO report error
thumbnail_bytes = None
processing.add(
ExtractionProcessing.strip_bars_video)
processing.add(ExtractionProcessing.add_film_strip)
else:
(
task,
full_file_name_to_work_on,
file_to_work_on_is_temporary,
) = self.extract_photo_video_from_camera(
rpd_file,
entire_video_required,
full_file_name_to_work_on,
False,
)
else:
# File is not on a camera
task = preprocess_thumbnail_from_disk(
rpd_file=rpd_file, processing=processing)
if task != ExtractionTask.bypass:
if rpd_file.thm_full_name is not None:
full_file_name_to_work_on = rpd_file.thm_full_name
if (task == ExtractionTask.
load_file_directly_metadata_from_secondary
):
secondary_full_file_name = rpd_file.full_file_name
else:
full_file_name_to_work_on = rpd_file.full_file_name
if task == ExtractionTask.bypass:
self.content = pickle.dumps(
GenerateThumbnailsResults(rpd_file=rpd_file,
thumbnail_bytes=thumbnail_bytes),
pickle.HIGHEST_PROTOCOL,
)
self.send_message_to_sink()
elif task != ExtractionTask.undetermined:
# Send data to load balancer, which will send to one of its
# workers
self.content = pickle.dumps(
ThumbnailExtractorArgument(
rpd_file=rpd_file,
task=task,
processing=processing,
full_file_name_to_work_on=full_file_name_to_work_on,
secondary_full_file_name=secondary_full_file_name,
exif_buffer=exif_buffer,
thumbnail_bytes=thumbnail_bytes,
use_thumbnail_cache=use_thumbnail_cache,
file_to_work_on_is_temporary=
file_to_work_on_is_temporary,
write_fdo_thumbnail=False,
send_thumb_to_main=True,
force_exiftool=force_exiftool,
),
pickle.HIGHEST_PROTOCOL,
)
self.frontend.send_multipart([b"data", self.content])
if arguments.camera:
self.camera.free_camera()
# Delete our temporary cache directories if they are empty
if photo_cache_dir is not None:
if not os.listdir(self.photo_cache_dir):
os.rmdir(self.photo_cache_dir)
if video_cache_dir is not None:
if not os.listdir(self.video_cache_dir):
os.rmdir(self.video_cache_dir)
logging.debug("Finished phase 1 of thumbnail generation for %s",
self.device_name)
if from_thumb_cache:
logging.info(
"{} of {} thumbnails for {} came from thumbnail cache".format(
from_thumb_cache, len(rpd_files), self.device_name))
if from_fdo_cache:
logging.info(
"{} of {} thumbnails of for {} came from Free Desktop cache".
format(from_fdo_cache, len(rpd_files), self.device_name))
self.disconnect_logging()
self.send_finished_command()
def cleanup_pre_stop(self):
if self.camera is not None:
self.camera.free_camera()
class ThumbnailCacheSql:
not_found = GetThumbnailPath(ThumbnailCacheDiskStatus.not_found, None,
None, None)
def __init__(self, create_table_if_not_exists: bool) -> None:
self.cache_dir = get_program_cache_directory(create_if_not_exist=True)
self.valid = self.cache_dir is not None
if not self.valid:
return
assert self.cache_dir is not None
self.cache_dir = os.path.join(self.cache_dir, 'thumbnails')
try:
if not os.path.exists(self.cache_dir):
os.makedirs(self.cache_dir, 0o700)
logging.debug("Created thumbnails cache %s", self.cache_dir)
elif not os.path.isdir(self.cache_dir):
os.remove(self.cache_dir)
logging.warning("Removed file %s", self.cache_dir)
os.makedirs(self.cache_dir, 0o700)
logging.debug("Created thumbnails cache %s", self.cache_dir)
except:
logging.error(
"Failed to create Rapid Photo Downloader Thumbnail Cache at %s",
self.cache_dir)
self.valid = False
self.cache_dir = None
self.random_filename = None
self.fs_encoding = None
else:
self.random_filename = GenerateRandomFileName()
self.md5 = MD5Name()
self.thumb_db = CacheSQL(self.cache_dir,
create_table_if_not_exists)
def save_thumbnail(self,
full_file_name: str,
size: int,
mtime: float,
mdatatime: float,
generation_failed: bool,
orientation_unknown: bool,
thumbnail: Optional[QImage],
camera_model: Optional[str] = None) -> Optional[str]:
"""
Save in the thumbnail cache using jpeg 75% compression.
:param full_file_name: full path of the file (including file
name). Will be turned into an absolute path if it is a file
system path
:param size: size of the file in bytes
:param mtime: file modification time
:param mdatatime: file time recorded in metadata
:param generation_failed: True if the thumbnail is meant to
signify the application failed to generate the thumbnail. If
so, it will be saved as an empty PNG in the application
subdirectory in the fail cache directory.
:param thumbnail: the thumbnail to be saved. Will not be
resized. Will be ignored if generation_failed is True.
:param camera_model: optional camera model. If the thumbnail is
not from a camera, then should be None.
:return the path of the saved file, else None if operation
failed
"""
if not self.valid:
return None
md5_name, uri = self.md5.md5_hash_name(full_file_name=full_file_name,
camera_model=camera_model,
extension='jpg')
if generation_failed:
logging.debug("Marking thumbnail for %s as 'generation failed'",
uri)
else:
logging.debug("Saving thumbnail for %s in RPD thumbnail cache",
uri)
try:
self.thumb_db.add_thumbnail(
uri=uri,
size=size,
mtime=mtime,
mdatatime=mdatatime,
md5_name=md5_name,
orientation_unknown=orientation_unknown,
failure=generation_failed)
except sqlite3.OperationalError as e:
logging.error(
"Database error adding thumbnail for %s: %s. Will not retry.",
uri, e)
return None
if generation_failed:
return None
md5_full_name = os.path.join(self.cache_dir, md5_name)
temp_path = os.path.join(self.cache_dir,
self.random_filename.name(extension='jpg'))
if thumbnail.save(temp_path, format='jpg', quality=75):
try:
os.rename(temp_path, md5_full_name)
os.chmod(md5_full_name, 0o600)
except OSError:
return None
return md5_full_name
return None
def get_thumbnail_path(self,
full_file_name: str,
mtime,
size: int,
camera_model: str = None) -> GetThumbnailPath:
"""
Attempt to get a thumbnail's path from the thumbnail cache.
:param full_file_name: full path of the file (including file
name). Will be turned into an absolute path if it is a file
system path
:param size: size of the file in bytes
:param mtime: file modification time, to be turned
into a float if it's not already
:param camera_model: optional camera model. If the thumbnail is
not from a camera, then should be None.
:return a GetThumbnailPath tuple of (1) ThumbnailCacheDiskStatus,
to indicate whether the thumbnail was found, a failure, or
missing, (2) the path (including the md5 name), else None,
(3) the file's metadata time, and (4) a bool indicating whether
the orientation of the thumbnail is unknown
"""
if not self.valid:
return self.not_found
uri = self.md5.get_uri(full_file_name, camera_model)
in_cache = self.thumb_db.have_thumbnail(uri, size, mtime)
if in_cache is None:
return self.not_found
if in_cache.failure:
return GetThumbnailPath(ThumbnailCacheDiskStatus.failure, None,
in_cache.mdatatime, None)
path = os.path.join(self.cache_dir, in_cache.md5_name)
if not os.path.exists(path):
self.thumb_db.delete_thumbnails([in_cache.md5_name])
return self.not_found
return GetThumbnailPath(ThumbnailCacheDiskStatus.found, path,
in_cache.mdatatime,
in_cache.orientation_unknown)
def cleanup_cache(self, days: int = 30) -> None:
"""
Remove all thumbnails that have not been accessed for x days
:param how many days to remove from
"""
time_period = 60 * 60 * 24 * days
if self.valid:
i = 0
now = time.time()
deleted_thumbnails = []
for name in os.listdir(self.cache_dir):
thumbnail = os.path.join(self.cache_dir, name)
if (os.path.isfile(thumbnail)
and os.path.getatime(thumbnail) < now - time_period):
os.remove(thumbnail)
deleted_thumbnails.append(name)
if len(deleted_thumbnails):
if self.thumb_db.cache_exists():
self.thumb_db.delete_thumbnails(deleted_thumbnails)
logging.debug(
'Deleted {} thumbnail files that had not been accessed for {} or more days'
.format(len(deleted_thumbnails), days))
def purge_cache(self) -> None:
"""
Delete the entire cache of all contents and remove the
directory
"""
if self.valid:
if self.cache_dir is not None and os.path.isdir(self.cache_dir):
# Delete the sqlite3 database too
shutil.rmtree(self.cache_dir)
def no_thumbnails(self) -> int:
"""
:return: how many thumbnails there are in the thumbnail database
"""
if not self.valid:
return 0
return self.thumb_db.no_thumbnails()
def cache_size(self) -> int:
"""
:return: the size of the entire cache (include the database) in bytes
"""
if not self.valid:
return 0
cwd = os.getcwd()
os.chdir(self.cache_dir)
s = sum(
os.path.getsize(f) for f in os.listdir('.') if os.path.isfile(f))
os.chdir(cwd)
return s
def db_size(self) -> int:
"""
:return: the size in bytes of the sql database file
"""
if not self.valid:
return 0
return os.path.getsize(self.thumb_db.db)
def optimize(self) -> Tuple[int, int, int]:
"""
Check for any thumbnails in the db that are not in the file system
Check for any thumbnails exist on the file system that are not in the db
Vacuum the db
:return db rows removed, file system photos removed, db size reduction in bytes
"""
rows = self.thumb_db.md5_names()
rows = {row[0] for row in rows}
cwd = os.getcwd()
os.chdir(self.cache_dir)
to_delete_from_db = {md5 for md5 in rows if not os.path.exists(md5)}
if len(to_delete_from_db):
self.thumb_db.delete_thumbnails(list(to_delete_from_db))
md5s = {md5 for md5 in os.listdir('.')} - {self.thumb_db.db_fs_name()}
to_delete_from_fs = md5s - rows
if len(to_delete_from_fs):
for md5 in to_delete_from_fs:
os.remove(md5)
os.chdir(cwd)
size = self.db_size()
self.thumb_db.vacuum()
return len(to_delete_from_db), len(
to_delete_from_fs), size - self.db_size()
class Cache:
"""
Base class with which to write and read cache thumbnails.
Create cache if it doesn't exist; checks validity.
"""
def __init__(self, cache_dir: str, failure_dir: Optional[str]) -> None:
"""
Create cache if it doesn't exist; checks validity.
:param cache_dir: full path of the directory into which
thumbnails will be saved / read.
:param failure_dir: full path of the directory into which
failed thumbnails will be saved / read (thumbnails that could
not be generated)
"""
assert sys.platform.startswith('linux')
self.cache_dir = cache_dir
self.failure_dir = failure_dir
assert self.cache_dir
self.valid = self._create_directory(self.cache_dir,
"Freedesktop.org thumbnail")
if self.valid:
self.random_filename = GenerateRandomFileName()
self.md5 = MD5Name()
if self.failure_dir is not None:
self.valid = self._create_directory(self.failure_dir,
"thumbnails failure")
if not self.valid:
self.random_filename = self.fs_encoding = None
def _create_directory(self, dir: str, descrtiption: str) -> None:
try:
if not os.path.exists(dir):
os.makedirs(dir, 0o700)
logging.debug("Created %s cache at %s", descrtiption, dir)
elif not os.path.isdir(dir):
os.remove(dir)
logging.warning("Removed file %s", dir)
os.makedirs(dir, 0o700)
logging.debug("Created %s cache at %s", descrtiption, dir)
except OSError:
logging.error("Failed to create %s cache at %s", descrtiption, dir)
return False
return True
def save_thumbnail(self,
full_file_name: str,
size: int,
modification_time: Union[float, int],
generation_failed: bool,
thumbnail: QImage,
camera_model: str = None,
free_desktop_org: bool = True) -> str:
"""
Save a thumbnail in the thumbnail cache.
:param full_file_name: full path of the file (including file
name). If the path contains symbolic links, two thumbnails will be
saved: the canonical path (without symlinks), and the path as
passed.
:param size: size of the file in bytes
:param modification_time: file modification time, to be turned
into a float if it's not already
:param generation_failed: True if the thumbnail is meant to
signify the application failed to generate the thumbnail. If
so, it will be saved as an empty PNG in the application
subdirectory in the fail cache directory.
:param thumbnail: the thumbnail to be saved. Will not be
resized. Will be ignored if generation_failed is True.
:param camera_model: optional camera model. If the thumbnail is
not from a camera, then should be None.
:param free_desktop_org: if True, then image will be convereted
to 8bit mode if necessary
:return the md5_name of the saved file, else None if operation
failed
"""
if not self.valid:
return None
# Save to both the real path and the path passed, which may include
# symbolic links
full_file_name_real_path = os.path.realpath(full_file_name)
if full_file_name_real_path != full_file_name:
self.save_thumbnail(full_file_name_real_path, size,
modification_time, generation_failed,
thumbnail, camera_model, free_desktop_org)
md5_name, uri = self.md5.md5_hash_name(full_file_name, camera_model)
if generation_failed:
thumbnail = QImage(QSize(1, 1), QImage.Format_Indexed8)
save_dir = self.failure_dir
else:
save_dir = self.cache_dir
path = os.path.join(save_dir, md5_name)
thumbnail.setText('Thumb::URI', uri)
thumbnail.setText('Thumb::MTime', str(float(modification_time)))
thumbnail.setText('Thumb::Size', str(size))
if free_desktop_org and not generation_failed:
if thumbnail.depth() != 8:
thumbnail = thumbnail.convertToFormat(QImage.Format_Indexed8)
temp_path = os.path.join(save_dir,
self.random_filename.name(extension='png'))
if thumbnail.save(temp_path):
os.rename(temp_path, path)
os.chmod(path, 0o600)
if generation_failed:
logging.debug("Wrote {}x{} thumbnail {} for {}".format(
thumbnail.width(), thumbnail.height(), path, uri))
return md5_name
else:
return None
def _get_thumbnail(self, path: str, modification_time: float,
size: int) -> Optional[bytes]:
if os.path.exists(path):
png = QImage(path)
if not png.isNull():
try:
mtime = float(png.text('Thumb::MTime'))
thumb_size = int(png.text('Thumb::Size'))
except ValueError:
return None
if mtime == float(modification_time) and thumb_size == size:
return png
return None
def get_thumbnail_md5_name(self,
full_file_name: str,
camera_model: Optional[str] = None) -> str:
"""
Returns the md5 name for the photo or video. Does not check if the file exists
on the file system in the cache.
:param full_file_name: full_file_name: full path of the file (including file
name). Will be turned into an absolute path if it is a file
system path
:param camera_model: optional camera model. If the thumbnail is
not from a camera, then should be None.
:return: the md5 name
"""
return self.md5.md5_hash_name(full_file_name=full_file_name,
camera_model=camera_model)[0]
def get_thumbnail(self,
full_file_name: str,
modification_time,
size: int,
camera_model: Optional[str] = None) -> GetThumbnail:
"""
Attempt to retrieve a thumbnail from the thumbnail cache.
:param full_file_name: full path of the file (including file
name). Will be turned into an absolute path if it is a file
system path
:param size: size of the file in bytes
:param modification_time: file modification time, to be turned
into a float if it's not already
:param camera_model: optional camera model. If the thumbnail is
not from a camera, then should be None.
:return a GetThumbnail tuple of (1) ThumbnailCacheDiskStatus,
to indicate whether the thumbnail was found, a failure, or
missing (2) the thumbnail as QImage, if found (or None), and
(3) the path (including the md5 name), else None,
"""
if not self.valid:
return GetThumbnail(ThumbnailCacheDiskStatus.not_found, None, None)
md5_name, uri = self.md5.md5_hash_name(full_file_name=full_file_name,
camera_model=camera_model)
path = os.path.join(self.cache_dir, md5_name)
png = self._get_thumbnail(path, modification_time, size)
if png is not None:
return GetThumbnail(ThumbnailCacheDiskStatus.found, png, path)
if self.failure_dir is not None:
path = os.path.join(self.failure_dir, md5_name)
png = self._get_thumbnail(path, modification_time, size)
if png is not None:
return GetThumbnail(ThumbnailCacheDiskStatus.failure, None,
None)
return GetThumbnail(ThumbnailCacheDiskStatus.not_found, None, None)
def modify_existing_thumbnail_and_save_copy(
self, existing_cache_thumbnail: str, full_file_name: str,
modification_time, size: int,
error_on_missing_thumbnail: bool) -> str:
"""
:param existing_cache_thumbnail: the md5 name of the cache thumbnail,
without the path to the cache
:param full_file_name: full path of the file (including file
name). Will be turned into an absolute path if need be
:param size: size of the file in bytes
:param modification_time: file modification time, to be turned
into a float if it's not already
:param error_on_missing_thumbnail: if True, issue error if thumbnail is
not located (useful when dealing with FDO 128 cache, but not helpful
with FDO 256 cache as not all RAW files have thumbnails large enough)
:return: the path of the saved file, else None if operation
failed
"""
existing_cache_thumbnail_full_path = os.path.join(
self.cache_dir, existing_cache_thumbnail)
if not os.path.isfile(existing_cache_thumbnail_full_path):
if error_on_missing_thumbnail:
logging.error("No FDO thumbnail to copy for %s",
full_file_name)
return None
thumbnail = QImage(existing_cache_thumbnail_full_path)
if not thumbnail.isNull():
return self.save_thumbnail(full_file_name=full_file_name,
size=size,
modification_time=modification_time,
generation_failed=False,
thumbnail=thumbnail,
camera_model=None,
free_desktop_org=False)
else:
return None
def delete_thumbnail(self,
full_file_name: str,
camera_model: str = None) -> None:
"""
Delete the thumbnail associated with the file if it exists
"""
if not self.valid:
return None
md5_name, uri = self.md5_hash_name(full_file_name, camera_model)
path = os.path.join(self.cache_dir, md5_name)
if os.path.isfile(path):
os.remove(path)
else:
path = os.path.join(self.failure_dir, md5_name)
if os.path.isfile(path):
os.remove(path)
def do_work(self):
self.problems = CopyingProblems()
args = pickle.loads(self.content) # type: CopyFilesArguments
if args.log_gphoto2:
self.gphoto2_logging = gphoto2_python_logging()
self.scan_id = args.scan_id
self.verify_file = args.verify_file
self.camera = None
# To workaround a bug in iOS and possibly other devices, check if need to rescan the files
# on the device
rescan_check = [
rpd_file for rpd_file in args.files
if rpd_file.from_camera and not rpd_file.cache_full_file_name
]
no_rescan = [
rpd_file for rpd_file in args.files
if not rpd_file.from_camera or rpd_file.cache_full_file_name
]
if rescan_check:
prefs = Preferences()
# Initialize camera
try:
self.camera = Camera(args.device.camera_model,
args.device.camera_port,
raise_errors=True,
specific_folders=prefs.folders_to_scan)
except CameraProblemEx as e:
self.problems.append(
CameraInitializationProblem(gp_code=e.gp_code))
logging.error("Could not initialize camera %s",
self.display_name)
self.terminate_camera_removed()
else:
rescan = RescanCamera(camera=self.camera, prefs=prefs)
rescan.rescan_camera(rpd_files=rescan_check)
rescan_check = rescan.rpd_files
if rescan.missing_rpd_files:
logging.error("%s files could not be relocated on %s",
len(rescan.missing_rpd_files),
self.camera.display_name)
rescan_check = list(
chain(rescan_check, rescan.missing_rpd_files))
rpd_files = list(chain(rescan_check, no_rescan))
random_filename = GenerateRandomFileName()
rpd_cache_same_device = defaultdict(
lambda: None) # type: Dict[FileType, Optional[bool]]
photo_temp_dir, video_temp_dir = create_temp_dirs(
args.photo_download_folder, args.video_download_folder)
# Notify main process of temp directory names
self.content = pickle.dumps(
CopyFilesResults(scan_id=args.scan_id,
photo_temp_dir=photo_temp_dir or '',
video_temp_dir=video_temp_dir or ''),
pickle.HIGHEST_PROTOCOL)
self.send_message_to_sink()
# Sort the files to be copied by modification time
# Important to do this with respect to sequence numbers, or else
# they'll be downloaded in what looks like a random order
rpd_files = sorted(rpd_files, key=attrgetter('modification_time'))
self.display_name = args.device.display_name
for idx, rpd_file in enumerate(rpd_files):
self.dest = self.src = None
if rpd_file.file_type == FileType.photo:
dest_dir = photo_temp_dir
else:
dest_dir = video_temp_dir
# Three scenarios:
# 1. Downloading from device with file system we can directly
# access
# 2. Downloading from camera using libgphoto2
# 3. Downloading from camera where we've already cached at
# least some of the files in the Download Cache
self.init_copy_progress()
if rpd_file.cache_full_file_name and os.path.isfile(
rpd_file.cache_full_file_name):
# Scenario 3
temp_file_name = os.path.basename(
rpd_file.cache_full_file_name)
temp_name = os.path.splitext(temp_file_name)[0]
temp_full_file_name = os.path.join(dest_dir, temp_file_name)
if rpd_cache_same_device[rpd_file.file_type] is None:
rpd_cache_same_device[rpd_file.file_type] = same_device(
rpd_file.cache_full_file_name, dest_dir)
if rpd_cache_same_device[rpd_file.file_type]:
try:
shutil.move(rpd_file.cache_full_file_name,
temp_full_file_name)
copy_succeeded = True
except (OSError, PermissionError,
FileNotFoundError) as inst:
copy_succeeded = False
logging.error(
"Could not move cached file %s to temporary file %s. Error "
"code: %s", rpd_file.cache_full_file_name,
temp_full_file_name, inst.errno)
self.problems.append(
FileMoveProblem(name=rpd_file.name,
uri=rpd_file.get_uri(),
exception=inst))
if self.verify_file:
rpd_file.md5 = hashlib.md5(
open(temp_full_file_name,
'rb').read()).hexdigest()
self.update_progress(rpd_file.size, rpd_file.size)
else:
# The download folder changed since the scan occurred, and is now
# on a different file system compared to that where the devices
# files were cached. Or the file was downloaded in full by the scan
# stage and saved, e.g. a sample video.
source = rpd_file.cache_full_file_name
destination = temp_full_file_name
copy_succeeded = self.copy_from_filesystem(
source, destination, rpd_file)
try:
os.remove(source)
except (OSError, PermissionError, FileNotFoundError) as e:
logging.error(
"Error removing RPD Cache file %s while copying %s. Error code: %s",
source, rpd_file.full_file_name, e.errno)
self.problems.append(
FileDeleteProblem(name=os.path.basename(source),
uri=get_uri(source),
exception=e))
else:
# Scenario 1 or 2
# Generate temporary name 5 digits long, because we cannot
# guarantee the source does not have duplicate file names in
# different directories, and here we are copying the files into
# a single directory
temp_name = random_filename.name()
temp_name_ext = '{}.{}'.format(temp_name, rpd_file.extension)
temp_full_file_name = os.path.join(dest_dir, temp_name_ext)
rpd_file.temp_full_file_name = temp_full_file_name
if not rpd_file.cache_full_file_name:
if rpd_file.from_camera:
# Scenario 2
if not self.camera:
copy_succeeded = False
logging.error("Could not copy %s from the %s",
rpd_file.full_file_name,
self.display_name)
# self.problems.append(CameraFileReadProblem(name=rpd_file.name,
# uri=rpd_file.get_uri()))
self.update_progress(rpd_file.size, rpd_file.size)
else:
copy_succeeded = self.copy_from_camera(rpd_file)
else:
# Scenario 1
source = rpd_file.full_file_name
destination = rpd_file.temp_full_file_name
copy_succeeded = self.copy_from_filesystem(
source, destination, rpd_file)
# increment this amount regardless of whether the copy actually
# succeeded or not. It's necessary to keep the user informed.
self.total_downloaded += rpd_file.size
mdata_exceptions = None
if not copy_succeeded:
rpd_file.status = DownloadStatus.download_failed
logging.debug("Download failed for %s",
rpd_file.full_file_name)
else:
if rpd_file.from_camera:
mdata_exceptions = copy_camera_file_metadata(
float(rpd_file.modification_time), temp_full_file_name)
else:
mdata_exceptions = copy_file_metadata(
rpd_file.full_file_name, temp_full_file_name)
# copy THM (video thumbnail file) if there is one
if rpd_file.thm_full_name:
rpd_file.temp_thm_full_name = self.copy_associate_file(
# translators: refers to the video thumbnail file that some
# cameras generate -- it has a .THM file extension
rpd_file,
temp_name,
dest_dir,
rpd_file.thm_full_name,
_('video THM'))
# copy audio file if there is one
if rpd_file.audio_file_full_name:
rpd_file.temp_audio_full_name = self.copy_associate_file(
rpd_file, temp_name, dest_dir,
rpd_file.audio_file_full_name, _('audio'))
# copy XMP file if there is one
if rpd_file.xmp_file_full_name:
rpd_file.temp_xmp_full_name = self.copy_associate_file(
rpd_file, temp_name, dest_dir,
rpd_file.xmp_file_full_name, 'XMP')
# copy Magic Lantern LOG file if there is one
if rpd_file.log_file_full_name:
rpd_file.temp_log_full_name = self.copy_associate_file(
rpd_file, temp_name, dest_dir,
rpd_file.log_file_full_name, 'LOG')
download_count = idx + 1
self.content = pickle.dumps(
CopyFilesResults(copy_succeeded=copy_succeeded,
rpd_file=rpd_file,
download_count=download_count,
mdata_exceptions=mdata_exceptions),
pickle.HIGHEST_PROTOCOL)
self.send_message_to_sink()
if len(self.problems):
logging.debug('Encountered %s problems while copying from %s',
len(self.problems), self.display_name)
self.send_problems()
if self.camera is not None:
self.camera.free_camera()
self.disconnect_logging()
self.send_finished_command()