def GenerateNumPyImageFromPILImage(pil_image: PILImage.Image) -> numpy.array:
(w, h) = pil_image.size
try:
s = pil_image.tobytes()
except OSError as e: # e.g. OSError: unrecognized data stream contents when reading image file
raise HydrusExceptions.UnsupportedFileException(str(e))
depth = len(s) // (w * h)
return numpy.fromstring(s, dtype='uint8').reshape((h, w, depth))
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 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 GenerateNumPyImage(path, mime, force_pil=False) -> numpy.array:
if HG.media_load_report_mode:
HydrusData.ShowText('Loading media: ' + path)
if not OPENCV_OK:
force_pil = True
if not force_pil:
try:
pil_image = RawOpenPILImage(path)
try:
pil_image.verify()
except:
raise HydrusExceptions.UnsupportedFileException()
# I and F are some sort of 32-bit monochrome or whatever, doesn't seem to work in PIL well, with or without ICC
if pil_image.mode not in ('I', 'F'):
if pil_image.mode == 'LAB':
force_pil = True
if HasICCProfile(pil_image):
if HG.media_load_report_mode:
HydrusData.ShowText(
'Image has ICC, so switching to PIL')
force_pil = True
except HydrusExceptions.UnsupportedFileException:
# pil had trouble, let's cross our fingers cv can do it
pass
if mime in PIL_ONLY_MIMETYPES or force_pil:
if HG.media_load_report_mode:
HydrusData.ShowText('Loading with PIL')
pil_image = GeneratePILImage(path)
numpy_image = GenerateNumPyImageFromPILImage(pil_image)
else:
if HG.media_load_report_mode:
HydrusData.ShowText('Loading with OpenCV')
if mime in (HC.IMAGE_JPEG, HC.IMAGE_TIFF):
flags = CV_IMREAD_FLAGS_JPEG
elif mime == HC.IMAGE_PNG:
flags = CV_IMREAD_FLAGS_PNG
else:
flags = CV_IMREAD_FLAGS_WEIRD
numpy_image = cv2.imread(path, flags=flags)
if numpy_image is None: # doesn't support some random stuff
if HG.media_load_report_mode:
HydrusData.ShowText('OpenCV Failed, loading with PIL')
pil_image = GeneratePILImage(path)
numpy_image = GenerateNumPyImageFromPILImage(pil_image)
else:
numpy_image = DequantizeNumPyImage(numpy_image)
if NumPyImageHasOpaqueAlphaChannel(numpy_image):
convert = cv2.COLOR_RGBA2RGB
numpy_image = cv2.cvtColor(numpy_image, convert)
return numpy_image
