def GetFileInfo(path, mime=None, ok_to_look_for_hydrus_updates=False):
size = os.path.getsize(path)
if size == 0:
raise HydrusExceptions.FileSizeException('File is of zero length!')
if mime is None:
mime = GetMime(
path, ok_to_look_for_hydrus_updates=ok_to_look_for_hydrus_updates)
if mime not in HC.ALLOWED_MIMES:
if mime == HC.TEXT_HTML:
raise HydrusExceptions.UnsupportedFileException(
'Looks like HTML -- maybe the client needs to be taught how to parse this?'
)
elif mime == HC.APPLICATION_UNKNOWN:
raise HydrusExceptions.UnsupportedFileException(
'Unknown filetype!')
else:
raise HydrusExceptions.UnsupportedFileException(
'Filetype is not permitted!')
width = None
height = None
duration = None
num_frames = None
num_words = None
if mime in (HC.IMAGE_JPEG, HC.IMAGE_PNG, HC.IMAGE_GIF, HC.IMAGE_WEBP,
HC.IMAGE_TIFF, HC.IMAGE_ICON):
((width, height), duration,
num_frames) = HydrusImageHandling.GetImageProperties(path, mime)
elif mime == HC.APPLICATION_ZIP:
temp_dir_path = HydrusPaths.GetTempDir()
try:
subprocess.call(["unzip", path, '-d', temp_dir_path])
cover = sorted(list(Path(temp_dir_path).rglob("*.jpg")) +
list(Path(temp_dir_path).rglob("*.png")),
key=lambda p: p.name)[0].as_posix()
((width, height), duration,
num_frames) = HydrusImageHandling.GetImageProperties(
cover, HC.IMAGE_PNG)
# TODO: delete dir
except Exception as e:
(width, height, duration, num_frames) = (300, 300, 0, 0)
elif mime == HC.APPLICATION_RAR:
temp_dir_path = HydrusPaths.GetTempDir()
try:
subprocess.call(["unrar", path, temp_dir_path])
cover = sorted(list(Path(temp_dir_path).rglob("*.jpg")) +
list(Path(temp_dir_path).rglob("*.png")),
key=lambda p: p.name)[0].as_posix()
((width, height), duration,
num_frames) = HydrusImageHandling.GetImageProperties(
cover, HC.IMAGE_PNG)
except Exception as e:
(width, height, duration, num_frames) = (300, 300, 0, 0)
elif mime == HC.APPLICATION_FLASH:
((width, height), duration,
num_frames) = HydrusFlashHandling.GetFlashProperties(path)
elif mime in (HC.IMAGE_APNG, HC.VIDEO_AVI, HC.VIDEO_FLV, HC.VIDEO_WMV,
HC.VIDEO_MOV, HC.VIDEO_MP4, HC.VIDEO_MKV, HC.VIDEO_REALMEDIA,
HC.VIDEO_WEBM, HC.VIDEO_MPEG):
((width, height), duration,
num_frames) = HydrusVideoHandling.GetFFMPEGVideoProperties(path)
elif mime == HC.APPLICATION_PDF:
num_words = HydrusDocumentHandling.GetPDFNumWords(
path) # this now give None until a better solution can be found
elif mime == HC.APPLICATION_PSD:
(width, height) = HydrusImageHandling.GetPSDResolution(path)
elif mime in HC.AUDIO:
ffmpeg_lines = HydrusVideoHandling.GetFFMPEGInfoLines(path)
(file_duration_in_s, stream_duration_in_s
) = HydrusVideoHandling.ParseFFMPEGDuration(ffmpeg_lines)
duration = int(file_duration_in_s * 1000)
if mime in HC.MIMES_THAT_DEFINITELY_HAVE_AUDIO:
has_audio = True
elif mime in HC.MIMES_THAT_MAY_HAVE_AUDIO:
has_audio = HydrusAudioHandling.VideoHasAudio(path)
else:
has_audio = False
if width is not None and width < 0:
width *= -1
if height is not None and height < 0:
width *= -1
if duration is not None and duration < 0:
duration *= -1
if num_frames is not None and num_frames < 0:
num_frames *= -1
if num_words is not None and num_words < 0:
num_words *= -1
return (size, mime, width, height, duration, num_frames, has_audio,
num_words)
def GetFileInfo(path, mime=None, ok_to_look_for_hydrus_updates=False):
size = os.path.getsize(path)
if size == 0:
raise HydrusExceptions.ZeroSizeFileException('File is of zero length!')
if mime is None:
mime = GetMime(
path, ok_to_look_for_hydrus_updates=ok_to_look_for_hydrus_updates)
if mime not in HC.ALLOWED_MIMES:
if mime == HC.TEXT_HTML:
raise HydrusExceptions.UnsupportedFileException(
'Looks like HTML -- maybe the client needs to be taught how to parse this?'
)
elif mime == HC.APPLICATION_UNKNOWN:
raise HydrusExceptions.UnsupportedFileException(
'Unknown filetype!')
else:
raise HydrusExceptions.UnsupportedFileException(
'Filetype is not permitted!')
width = None
height = None
duration = None
num_frames = None
num_words = None
if mime in HC.MIMES_THAT_DEFINITELY_HAVE_AUDIO:
has_audio = True
else:
has_audio = False
if mime in (HC.IMAGE_JPEG, HC.IMAGE_PNG, HC.IMAGE_GIF, HC.IMAGE_WEBP,
HC.IMAGE_TIFF, HC.IMAGE_ICON):
((width, height), duration,
num_frames) = HydrusImageHandling.GetImageProperties(path, mime)
elif mime == HC.APPLICATION_CLIP:
((width, height), duration,
num_frames) = HydrusClipHandling.GetClipProperties(path)
elif mime == HC.APPLICATION_FLASH:
((width, height), duration,
num_frames) = HydrusFlashHandling.GetFlashProperties(path)
elif mime == HC.IMAGE_APNG:
((width, height), duration, num_frames,
has_audio) = HydrusVideoHandling.GetFFMPEGAPNGProperties(path)
elif mime == HC.APPLICATION_PDF:
num_words = HydrusDocumentHandling.GetPDFNumWords(
path) # this now give None until a better solution can be found
elif mime == HC.APPLICATION_PSD:
(width, height) = HydrusImageHandling.GetPSDResolution(path)
elif mime in HC.VIDEO:
((width, height), duration, num_frames,
has_audio) = HydrusVideoHandling.GetFFMPEGVideoProperties(path)
elif mime in HC.AUDIO:
ffmpeg_lines = HydrusVideoHandling.GetFFMPEGInfoLines(path)
(file_duration_in_s, stream_duration_in_s
) = HydrusVideoHandling.ParseFFMPEGDuration(ffmpeg_lines)
duration = int(file_duration_in_s * 1000)
if width is not None and width < 0:
width *= -1
if height is not None and height < 0:
width *= -1
if duration is not None and duration < 0:
duration *= -1
if num_frames is not None and num_frames < 0:
num_frames *= -1
if num_words is not None and num_words < 0:
num_words *= -1
return (size, mime, width, height, duration, num_frames, has_audio,
num_words)
def VideoHasAudio(path):
info_lines = HydrusVideoHandling.GetFFMPEGInfoLines(path)
(audio_found, audio_format) = ParseFFMPEGAudio(info_lines)
if not audio_found:
return False
# just because video metadata has an audio stream doesn't mean it has audio. some vids have silent audio streams lmao
# so, let's read it as PCM and see if there is any noise
# this obviously only works for single audio stream vids, we'll adapt this if someone discovers a multi-stream mkv with a silent channel that doesn't work here
cmd = [HydrusVideoHandling.FFMPEG_PATH]
# this is perhaps not sensible for eventual playback and I should rather go for wav file-like and feed into python 'wave' in order to maintain stereo/mono and so on and have easy chunk-reading
cmd.extend(['-i', path, '-loglevel', 'quiet', '-f', 's16le', '-'])
sbp_kwargs = HydrusData.GetSubprocessKWArgs()
HydrusData.CheckProgramIsNotShuttingDown()
try:
process = subprocess.Popen(cmd,
bufsize=65536,
stdin=subprocess.PIPE,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
**sbp_kwargs)
except FileNotFoundError as e:
HydrusData.ShowText('Cannot render audio--FFMPEG not found!')
raise
# silent PCM data is just 00 bytes
# every now and then, you'll get a couple ffs for some reason, but this is not legit audio data
try:
chunk_of_pcm_data = process.stdout.read(65536)
while len(chunk_of_pcm_data) > 0:
# iterating over bytes gives you ints, recall
if True in (b != 0 and b != 255 for b in chunk_of_pcm_data):
return True
chunk_of_pcm_data = process.stdout.read(65536)
return False
finally:
process.terminate()
process.stdout.close()
process.stderr.close()