def distributed_collage(self, callback, orientation, alignment, urls):
logger.debug('filters.distributed_collage: distributed_collage invoked')
self.storage = self.context.modules.storage
self.callback = callback
self.orientation = orientation
self.alignment = alignment
self.urls = urls.split('|')
self.images = {}
total = len(self.urls)
if total > self.MAX_IMAGES:
logger.error('filters.distributed_collage: Too many images to join')
callback()
elif total == 0:
logger.error('filters.distributed_collage: No images to join')
callback()
else:
self.urls = self.urls[:self.MAX_IMAGES]
for url in self.urls:
self.images[url] = Picture(url, self)
# second loop needed to ensure that all images are in self.images
# otherwise, self.on_image_fetch can call the self.assembly()
# without that all images had being loaded
for url in self.urls:
buffer = yield tornado.gen.maybe_future(self.storage.get(url))
pic = self.images[url]
if buffer is not None:
pic.fill_buffer(buffer)
self.on_image_fetch()
else:
pic.request()
def dispatch(self, file_key):
""" Callback method for getObject from s3 """
if not file_key or 'Error' in file_key or 'Body' not in file_key:
logger.error(
"ERROR retrieving image from S3 {0}: {1}".
format(self.key, str(file_key)))
# If we got here, there was a failure.
# We will return 404 if S3 returned a 404, otherwise 502.
result = LoaderResult()
result.successful = False
if not file_key:
result.error = LoaderResult.ERROR_UPSTREAM
self.callback(result)
return
response_metadata = file_key.get('ResponseMetadata', {})
status_code = response_metadata.get('HTTPStatusCode')
if status_code == 404:
result.error = LoaderResult.ERROR_NOT_FOUND
self.callback(result)
return
if self.retries_counter < self.max_retry:
self.__increment_retry_counter()
self.bucket_loader.get(self.key,
callback=self.dispatch)
else:
result.error = LoaderResult.ERROR_UPSTREAM
self.callback(result)
else:
self.callback(file_key['Body'].read())
def load(self, buffer, extension):
self.extension = extension
if extension is None:
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, '.jpg')
if self.extension == '.svg':
buffer = self.convert_svg_to_png(buffer)
image_or_frames = self.create_image(buffer)
if METADATA_AVAILABLE:
try:
self.metadata = ImageMetadata.from_buffer(buffer)
self.metadata.read()
except Exception as e:
logger.error('Error reading image metadata: %s' % e)
if self.context.config.ALLOW_ANIMATED_GIFS and isinstance(
image_or_frames, (list, tuple)):
self.image = image_or_frames[0]
if len(image_or_frames) > 1:
self.multiple_engine = MultipleEngine(self)
for frame in image_or_frames:
self.multiple_engine.add_frame(frame)
self.wrap(self.multiple_engine)
else:
self.image = image_or_frames
if self.source_width is None:
self.source_width = self.size[0]
if self.source_height is None:
self.source_height = self.size[1]
def command(
cls,
context,
pre=[],
post=[],
buffer='',
input_temp_file=None
):
if not input_temp_file:
input_temp_file = NamedTemporaryFile()
input_temp_file.write(buffer)
input_temp_file.flush()
command = [context.config.EXIFTOOL_PATH]
command += pre
command.append(input_temp_file.name)
command += post
logger.debug('[ExiftoolRunner] command: %r' % command, extra=log_extra(context))
code, stderr, stdout = ShellRunner.command(command, context)
input_temp_file.close()
if stderr:
logger.error('[ExiftoolRunner] error: %r' % stderr, extra=log_extra(context))
return stdout
def post(self, **kwargs):
self.should_return_image = False
# URL can be passed as a URL argument or in the body
url = kwargs['url'] if 'url' in kwargs else kwargs['key']
if not url:
logger.error("Couldn't find url param in body or key in URL...")
raise tornado.web.HTTPError(404)
options = RequestParser.path_to_parameters(url)
yield self.check_image(options)
# We check the status code, if != 200 the image is incorrect, and we shouldn't store the key
if self.get_status() == 200:
logger.debug("Image is checked, clearing the response before trying to store...")
self.clear()
try:
shortener = Shortener(self.context)
key = shortener.generate(url)
shortener.put(key, url)
self.write(json.dumps({'key': key}))
self.set_header("Content-Type", "application/json")
except Exception as e:
logger.error("An error occurred while trying to store shortened URL: {error}.".format(error=e.message))
self.set_status(500)
self.write(json.dumps({'error': e.message}))
def get_image(self):
try:
result = yield self._fetch(self.context.request.image_url)
if not result.successful:
if result.loader_error == LoaderResult.ERROR_NOT_FOUND:
self._error(404)
return
elif result.loader_error == LoaderResult.ERROR_UPSTREAM:
# Return a Bad Gateway status if the error came from upstream
self._error(502)
return
elif result.loader_error == LoaderResult.ERROR_TIMEOUT:
# Return a Gateway Timeout status if upstream timed out (i.e. 599)
self._error(504)
return
else:
self._error(500)
return
except Exception as e:
msg = "[BaseHandler] get_image failed for url `{url}`. error: `{error}`".format(
url=self.context.request.image_url, error=e
)
self.log_exception(*sys.exc_info())
if "cannot identify image file" in e.message:
logger.warning(msg)
self._error(400)
else:
logger.error(msg)
self._error(500)
return
normalized = result.normalized
buffer = result.buffer
engine = result.engine
req = self.context.request
if engine is None:
if buffer is None:
self._error(504)
return
engine = self.context.request.engine
engine.load(buffer, self.context.request.extension)
def transform():
self.normalize_crops(normalized, req, engine)
if req.meta:
self.context.request.engine = JSONEngine(engine, req.image_url, req.meta_callback)
after_transform_cb = functools.partial(self.after_transform, self.context)
Transformer(self.context).transform(after_transform_cb)
self.filters_runner.apply_filters(thumbor.filters.PHASE_AFTER_LOAD, transform)
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.pngcrush_path = self.context.config.PNGCRUSH_PATH
if not (os.path.isfile(self.pngcrush_path) and os.access(self.pngcrush_path, os.X_OK)):
logger.error("ERROR pngcrush path '{0}' is not accessible".format(self.pngcrush_path))
self.runnable = False
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.imgmin_path = self.context.config.IMGMIN_PATH
if not ( os.path.isfile(self.imgmin_path) ):
logger.error("ERROR path '{0}' is not accessible".format(self.imgmin_path))
self.runnable = False
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.zopflipng_path = self.context.config.ZOPFLIPNG_PATH
if not (os.path.isfile(self.zopflipng_path) and os.access(self.zopflipng_path, os.X_OK)):
logger.error("ERROR zopflipng path '{0}' is not accessible".format(self.zopflipng_path))
self.runnable = False
def return_contents(response, url, callback):
if response.error:
logger.error("ERROR retrieving image {0}: {1}".format(url, str(response.error)))
callback(None)
elif response.body is None or len(response.body) == 0:
logger.error("ERROR retrieving image {0}: Empty response.".format(url))
callback(None)
else:
callback(response.body)
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.jpegrecompress_path = self.context.config.JPEGRECOMPRESS_PATH
if not (os.path.isfile(self.jpegrecompress_path) and os.access(self.jpegrecompress_path, os.X_OK)):
logger.error("ERROR jpeg-recompress path '{0}' is not accessible".format(self.jpegrecompress_path))
self.runnable = False
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.mozjpeg_path = self.context.config.MOZJPEG_PATH
self.mozjpeg_level = self.context.config.MOZJPEG_QUALITY or '75'
if not (os.path.isfile(self.mozjpeg_path) and os.access(self.mozjpeg_path, os.X_OK)):
logger.error("ERROR mozjpeg path '{0}' is not accessible".format(self.mozjpeg_path))
self.runnable = False
def put(self, bytes):
normalized_path = self.normalize_path(self.context.request.url)
uri = self.context.config.get('RESULT_STORAGE_WEBDAV_URI') + normalized_path
logger.debug("[RESULT_STORAGE] Making PUT request to: %s", uri)
http_client = HTTPClient()
try:
response = http_client.fetch(uri, method='PUT', body=bytes)
logger.debug("[RESULT_STORAGE] Success on PUT request!")
except HTTPError as e:
logger.error("[RESULT_STORAGE] Error on PUT request: %s", e)
def handle_error(self, context, handler, exception):
ex_type, value, tb = exception
ex_msg = traceback.format_exception_only(ex_type, value)
tb_msg = traceback.format_tb(tb)
extra = log_extra(context)
extra['traceback'] = ''.join(tb_msg)
logger.error(''.join(ex_msg), extra=extra)
def validate(self, path):
if not hasattr(self.loader, 'validate'):
return True
is_valid = self.loader.validate(path)
if not is_valid:
logger.error('Request denied because the specified path "%s" was not identified by the loader as a valid path' % path)
return is_valid
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.pngquant_path = self.context.config.PNGQUANT_PATH
self.pngquant_quality = self.context.config.PNGQUANT_QUALITY or '65-80'
self.pngquant_speed = self.context.config.PNGQUANT_SPEED or '3'
if not (os.path.isfile(self.pngquant_path) and os.access(self.pngquant_path, os.X_OK)):
logger.error("ERROR pnqquant path '{0}' is not accessible".format(self.pngquant_path))
self.runnable = False
def _handle_request_exception(self, e):
try:
exc_info = sys.exc_info()
msg = traceback.format_exception(exc_info[0], exc_info[1], exc_info[2])
if self.context.config.USE_CUSTOM_ERROR_HANDLING:
self.context.modules.importer.error_handler.handle_error(context=self.context, handler=self, exception=exc_info)
finally:
del exc_info
logger.error('ERROR: %s' % "".join(msg))
self.send_error(500)
def convert_svg_to_png(self, buffer):
if not cairosvg:
msg = """[BaseEngine] convert_svg_to_png failed cairosvg not
imported (if you want svg conversion to png please install cairosvg)
"""
logger.error(msg)
return buffer
buffer = cairosvg.svg2png(bytestring=buffer, dpi=self.context.config.SVG_DPI)
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, '.jpg')
return buffer
def convert_tif_to_png(self, buffer):
if not numpy:
msg = """[BaseEngin] convert_tif_to_png failed numpy not imported"""
logger.error(msg)
return buffer
if not cv2:
msg = """[BaseEngin] convert_tif_to_png failed opencv not imported"""
logger.error(msg)
return buffer
img = cv2.imdecode(numpy.fromstring(buffer), -1)
img_str = cv2.imencode('.png', img)[1].tostring()
return img_str
def log_exception(self, *exc_info):
if isinstance(exc_info[1], tornado.web.HTTPError):
# Delegate HTTPError's to the base class
# We don't want these through normal exception handling
return super(ContextHandler, self).log_exception(*exc_info)
msg = traceback.format_exception(*exc_info)
try:
if self.context.config.USE_CUSTOM_ERROR_HANDLING:
self.context.modules.importer.error_handler.handle_error(context=self.context, handler=self, exception=exc_info)
finally:
del exc_info
logger.error('ERROR: %s' % "".join(msg))
def _read(self, update_image=True):
buffer = self.flush_operations(update_image)
# Make sure gifsicle produced a valid gif.
try:
with BytesIO(buffer) as buff:
Image.open(buff).verify()
except Exception:
self.context.metrics.incr('gif_engine.no_output')
logger.error("[GIF_ENGINE] invalid gif engine result for url `{url}`.".format(
url=self.context.request.url
))
raise
return buffer
def read(self, extension=None, quality=None):
self.flush_operations()
# Make sure gifsicle produced a valid gif.
try:
with Image.open(BytesIO(self.buffer)) as image:
image.verify()
except Exception:
self.context.statsd_client.incr('gif_engine.no_output')
logger.error("[GIF_ENGINE] invalid gif engine result for url `{url}`.".format(
url=self.context.request.url
))
raise
return self.buffer
def convert_tif_to_png(self, buffer):
if not cv2:
msg = """[PILEngine] convert_tif_to_png failed: opencv not imported"""
logger.error(msg)
return buffer
if not numpy:
msg = """[PILEngine] convert_tif_to_png failed: opencv not imported"""
logger.error(msg)
return buffer
img = cv2.imdecode(numpy.fromstring(buffer, dtype='uint16'), -1)
buffer = cv2.imencode('.png', img)[1].tostring()
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, '.jpg')
return buffer
def detect(self, callback):
self.context.request.prevent_result_storage = True
try:
if not self.pyremotecv:
self.pyremotecv = PyRemoteCV(host=self.context.config.REDIS_QUEUE_SERVER_HOST,
port=self.context.config.REDIS_QUEUE_SERVER_PORT,
db=self.context.config.REDIS_QUEUE_SERVER_DB,
password=self.context.config.REDIS_QUEUE_SERVER_PASSWORD)
self.pyremotecv.async_detect('remotecv.pyres_tasks.DetectTask', 'Detect',
args=[self.detection_type, self.context.request.image_url],
key=self.context.request.image_url)
except RedisError:
self.context.request.detection_error = True
logger.error(traceback.format_exc())
finally:
callback([])
def save_on_disc(self):
if self.fetched:
try:
self.engine.load(self.buffer, self.extension)
except Exception as err:
self.failed = True
logger.exception(err)
try:
self.thumbor_filter.storage.put(self.url, self.engine.read())
self.thumbor_filter.storage.put_crypto(self.url)
except Exception as err:
self.failed = True
logger.exception(err)
else:
self.failed = True
logger.error("filters.distributed_collage: Can't save unfetched image")
def convert_tif_to_png(self, buffer):
if not cv:
msg = """[PILEngine] convert_tif_to_png failed: opencv not imported"""
logger.error(msg)
return buffer
# can not use cv2 here, because ubuntu precise shipped with python-opencv 2.3 which has bug with imencode
# requires 3rd parameter buf which could not be created in python. Could be replaced with these lines:
# img = cv2.imdecode(numpy.fromstring(buffer, dtype='uint16'), -1)
# buffer = cv2.imencode('.png', img)[1].tostring()
mat_data = cv.CreateMatHeader(1, len(buffer), cv.CV_8UC1)
cv.SetData(mat_data, buffer, len(buffer))
img = cv.DecodeImage(mat_data, -1)
buffer = cv.EncodeImage(".png", img).tostring()
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, '.jpg')
return buffer
def __init__(self, context):
super(Optimizer, self).__init__(context)
self.runnable = True
self.imgmin_path = self.context.config.IMGMIN_PATH
# All defaults are from imgmin unless noted otherwise
self.error_threshold = self.context.config.AUTO_ERROR_THRESHOLD or "1.0"
self.color_density_ratio = self.context.config.AUTO_COLOR_DENSITY_RATIO or "0.11"
self.min_unique_colors = self.context.config.AUTO_MIN_UNIQUE_COLORS or "4096"
self.quality_out_max = self.context.config.AUTO_QUALITY_OUT_MAX or "100" # imgming default 95
self.quality_out_min = self.context.config.AUTO_QUALITY_OUT_MIN or "95" # imgming default 70
self.quality_in_min = self.context.config.AUTO_QUALITY_IN_MIN or "82"
self.max_steps = self.context.config.AUTO_MAX_STEPS or "5"
if not (os.path.isfile(self.imgmin_path)):
logger.error("ERROR path '{0}' or '{1}' is not accessible".format(self.imgmin_path))
self.runnable = False
def run_gifsicle(self, command):
p = Popen([self.context.server.gifsicle_path] + command.split(' '), stdout=PIPE, stdin=PIPE, stderr=PIPE)
stdout_data, stderr_data = p.communicate(input=self.buffer)
if p.returncode != 0:
logger.error(stderr_data)
if stdout_data is None:
raise GifSicleError(
'gifsicle command returned errorlevel {0} for command "{1}" (image maybe corrupted?)'.format(
p.returncode, ' '.join(
[self.context.server.gifsicle_path] +
command.split(' ') +
[self.context.request.url]
)
)
)
return stdout_data
def normalize_color_to_hex(self, color_string):
try:
return webcolors.normalize_hex("#" + color_string)
except ValueError:
pass
try:
return webcolors.name_to_hex(color_string)
except ValueError:
pass
try:
return webcolors.normalize_hex(color_string)
except ValueError:
pass
if color_string:
logger.error('background_color value could not be parsed')
def put(self, path, bytes):
start = datetime.datetime.now()
normalized_path = self._normalize_path(path)
content_type = 'text/plain'
if bytes:
try:
mime = BaseEngine.get_mimetype(bytes)
content_type = mime
except:
logger.error("[GoogleCloudStorage] Couldn't determine mimetype for %s"
% path)
blob = self._get_bucket().blob(normalized_path)
blob.upload_from_string(bytes, content_type=content_type)
finish = datetime.datetime.now()
self.context.metrics.timing('gcs.put.{0}'.format(normalized_path),
(finish - start).total_seconds()
* 1000)
def _process_done(
callback,
process,
context,
normalized_url,
seek,
output_file,
status
):
# T183907 Sometimes ffmpeg returns status 0 and actually fails to
# generate a thumbnail. We double-check the existence of the thumbnail
# in case of apparent success
if status == 0:
if os.stat(output_file.name).st_size == 0:
status = -1
# If rendering the desired frame fails, attempt to render the
# first frame instead
if status != 0 and seek > 0:
seek_and_screenshot(callback, context, normalized_url, 0)
return
result = LoaderResult()
if status != 0: # pragma: no cover
_http_code_from_stderr(context, process, result, normalized_url)
else:
result.successful = True
result.buffer = output_file.read()
process.stdout.close()
process.stderr.close()
output_file.close()
try:
os.unlink(output_file.name)
except OSError as e: # pragma: no cover
if e.errno != errno.ENOENT:
logger.error('[Video] Unable to unlink output file', extra=log_extra(context))
raise
callback(result)
def on_redis_error(self, fname, exc_type, exc_value):
'''Callback executed when there is a redis error.
:param string fname: Function name that was being called.
:param type exc_type: Exception type
:param Exception exc_value: The current exception
:returns: Default value or raise the current exception
'''
if self.shared_client:
Storage.storage = None
else:
self.storage = None
if self.context.config.REDIS_RESULT_STORAGE_IGNORE_ERRORS is True:
logger.error("Redis result storage failure: %s" % exc_value)
return None
else:
raise exc_value
def reorientate(self, override_exif=True):
"""
Rotates the image in the buffer so that it is oriented correctly.
If override_exif is True (default) then the metadata
orientation is adjusted as well.
:param override_exif: If the metadata should be adjusted as well.
:type override_exif: Boolean
"""
exif = self._get_exif_object()
if exif is None:
return None
orientation = exif.get_orientation()
if orientation is None:
return
if orientation == 2:
self.flip_horizontally()
elif orientation == 3:
self.rotate(180)
elif orientation == 4:
self.flip_vertically()
elif orientation == 5:
# Horizontal Mirror + Rotation 270 CCW
self.flip_vertically()
self.rotate(270)
elif orientation == 6:
self.rotate(270)
elif orientation == 7:
# Vertical Mirror + Rotation 270 CCW
self.flip_horizontally()
self.rotate(270)
elif orientation == 8:
self.rotate(90)
if orientation != 1 and override_exif:
try:
exif.set_orientation(1)
self.exif = exif.tobytes()
except Exception as e:
msg = """[exif] %s""" % e
logger.error(msg)
def load(context, path, callback):
result = LoaderResult()
for idx, next_dir in enumerate(context.config.TC_MULTIDIR_PATHS):
file_path = join(next_dir.rstrip('/'), path.lstrip('/'))
file_path = abspath(file_path)
inside_root_path = file_path.startswith(abspath(next_dir))
if inside_root_path:
# keep backwards compatibility, try the actual path first
# if not found, unquote it and try again
found = exists(file_path)
if not found:
file_path = unquote(file_path)
found = exists(file_path)
if found:
with open(file_path, 'rb') as f:
stats = fstat(f.fileno())
result.successful = True
result.buffer = f.read()
result.metadata.update(
size=stats.st_size,
updated_at=datetime.utcfromtimestamp(stats.st_mtime))
callback(result)
return
logger.debug('TC_MULTIDIR: File {0} not found in {1}'.format(path, next_dir))
# else loop and try next directory
if not context.config.TC_MULTIDIR_PATHS:
logger.error('TC_MULTIDIR: No paths set in configuration TC_MULTIDIR_PATHS')
# no file found
result.error = LoaderResult.ERROR_NOT_FOUND
result.successful = False
callback(result)
def convert_svg_to_png(self, buffer):
if not cairosvg:
msg = """[BaseEngine] convert_svg_to_png failed cairosvg not
imported (if you want svg conversion to png please install cairosvg)
"""
logger.error(msg)
return buffer
try:
buffer = cairosvg.svg2png(bytestring=buffer, dpi=self.context.config.SVG_DPI)
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, '.jpg')
except ParseError:
mime = self.get_mimetype(buffer)
extension = EXTENSION.get(mime)
if extension is None or extension == '.svg':
raise
self.extension = extension
return buffer
def detect(self, callback):
self.context.request.prevent_result_storage = True
try:
if not self.pyremotecv:
self.pyremotecv = PyRemoteCV(
host=self.context.config.REDIS_QUEUE_SERVER_HOST,
port=self.context.config.REDIS_QUEUE_SERVER_PORT,
db=self.context.config.REDIS_QUEUE_SERVER_DB,
password=self.context.config.REDIS_QUEUE_SERVER_PASSWORD)
self.pyremotecv.async_detect(
'remotecv.pyres_tasks.DetectTask',
'Detect',
args=[self.detection_type, self.context.request.image_url],
key=self.context.request.image_url)
except RedisError:
self.context.request.detection_error = True
logger.error(traceback.format_exc())
finally:
callback([])
def reorientate(self, override_exif=True):
"""
Rotates the image in the buffer so that it is oriented correctly.
If override_exif is True (default) then the metadata
orientation is adjusted as well.
:param override_exif: If the metadata should be adjusted as well.
:type override_exif: Boolean
"""
orientation = self.get_orientation()
if orientation is None:
return
if orientation == 2:
self.flip_horizontally()
elif orientation == 3:
self.rotate(180)
elif orientation == 4:
self.flip_vertically()
elif orientation == 5:
# Horizontal Mirror + Rotation 270 CCW
self.flip_vertically()
self.rotate(270)
elif orientation == 6:
self.rotate(270)
elif orientation == 7:
# Vertical Mirror + Rotation 270 CCW
self.flip_horizontally()
self.rotate(270)
elif orientation == 8:
self.rotate(90)
if orientation != 1 and override_exif:
exif_dict = self._get_exif_segment()
if exif_dict and piexif.ImageIFD.Orientation in exif_dict["0th"]:
exif_dict["0th"][piexif.ImageIFD.Orientation] = 1
try:
self.exif = piexif.dump(exif_dict)
except Exception as e:
msg = """[piexif] %s""" % e
logger.error(msg)
def get_image(self):
try:
normalized, buffer, engine = yield self._fetch(
self.context.request.image_url)
except Exception as e:
msg = '[BaseHandler] get_image failed for url `{url}`. error: `{error}`'.format(
url=self.context.request.image_url, error=e)
self.log_exception(*sys.exc_info())
if 'cannot identify image file' in e.message:
logger.warning(msg)
self._error(400)
else:
logger.error(msg)
self._error(500)
return
req = self.context.request
if engine is None:
if buffer is None:
self._error(504)
return
engine = self.context.request.engine
engine.load(buffer, self.context.request.extension)
def transform():
self.normalize_crops(normalized, req, engine)
if req.meta:
self.context.request.engine = JSONEngine(
engine, req.image_url, req.meta_callback)
after_transform_cb = functools.partial(self.after_transform,
self.context)
Transformer(self.context).transform(after_transform_cb)
self.filters_runner.apply_filters(thumbor.filters.PHASE_AFTER_LOAD,
transform)
def run_gifsicle(self, command):
process = Popen(
[self.context.server.gifsicle_path] + command.split(" "),
stdout=PIPE,
stdin=PIPE,
stderr=PIPE,
)
stdout_data, stderr_data = process.communicate(input=self.buffer)
if process.returncode != 0:
logger.error(stderr_data)
if stdout_data is None:
raise GifSicleError(
("gifsicle command returned errorlevel {0} for "
'command "{1}" (image maybe corrupted?)').format(
process.returncode,
" ".join([self.context.server.gifsicle_path] +
command.split(" ") + [self.context.request.url]),
))
return stdout_data
def _get(self, key, col):
try:
key = md5(key).hexdigest() + '-' + key
except UnicodeEncodeError:
key = md5(
key.encode('utf-8')).hexdigest() + '-' + key.encode('utf-8')
if self.storage is None:
self._connect()
try:
r = self.storage.get(self.table, key, self.data_fam + ':' + col)[0]
except IndexError:
r = None
except:
r = None
logger.error("Error retrieving image from HBase; key " + key)
self.hbase_server_offset = self.hbase_server_offset + 1
return r
def load(self, buffer, extension):
self.extension = extension
if extension is None:
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, '.jpg')
if self.extension == '.svg':
buffer = self.convert_svg_to_png(buffer)
# added by zhaorong, if it is a tif picture, convert it to png
if self.extension == '.tif':
buffer = self.convert_tif_to_png(buffer)
image_or_frames = self.create_image(buffer)
if image_or_frames is None:
return
if METADATA_AVAILABLE:
try:
self.metadata = ImageMetadata.from_buffer(buffer)
self.metadata.read()
except Exception as e:
logger.error('Error reading image metadata: %s' % e)
if self.context.config.ALLOW_ANIMATED_GIFS and isinstance(
image_or_frames, (list, tuple)):
self.image = image_or_frames[0]
if len(image_or_frames) > 1:
self.multiple_engine = MultipleEngine(self)
for frame in image_or_frames:
self.multiple_engine.add_frame(frame)
self.wrap(self.multiple_engine)
else:
self.image = image_or_frames
if self.source_width is None:
self.source_width = self.size[0]
if self.source_height is None:
self.source_height = self.size[1]
def post(self, **kwargs):
self.should_return_image = False
content_type = self.request.headers.get("Content-Type", '')
if 'key' in kwargs and kwargs['key']:
url = kwargs['key']
elif content_type.startswith("application/json"):
data = json.loads(self.request.body)
url = data['url'] if 'url' in data else None
else:
url = self.get_body_argument('url', None)
if not url:
logger.error("Couldn't find url param in body or key in URL...")
raise tornado.web.HTTPError(400)
options = RequestParser.path_to_parameters(url)
yield self.check_image(options)
# We check the status code, if != 200 the image is incorrect, and we shouldn't store the key
if self.get_status() == 200:
logger.debug(
"Image is checked, clearing the response before trying to store..."
)
self.clear()
try:
shortener = Shortener(self.context)
key = shortener.generate(url)
shortener.put(key, url)
self.write(json.dumps({'key': key}))
self.set_header("Content-Type", "application/json")
except Exception as e:
logger.error(
"An error occurred while trying to store shortened URL: {error}."
.format(error=e.message))
self.set_status(500)
self.write(json.dumps({'error': e.message}))
def put(self, path, bytes):
file_abspath = self._normalize_path(path)
logger.debug("[STORAGE] putting at %s" % file_abspath)
bucket = self._get_bucket()
blob = bucket.blob(file_abspath)
blob.upload_from_string(bytes)
max_age = self.context.config.CLOUD_STORAGE_MAX_AGE
blob.cache_control = "public,max-age=%s" % max_age
if bytes:
try:
mime = BaseEngine.get_mimetype(bytes)
blob.content_type = mime
except Exception as ex:
logger.debug("[STORAGE] Couldn't determine mimetype: %s" % ex)
try:
blob.patch()
except Exception as ex:
logger.error("[STORAGE] Couldn't patch blob: %s" % ex)
async def distributed_collage(self, orientation, alignment, urls):
self.orientation = orientation
self.alignment = alignment
self.urls = urls.split("|")
self.images = {}
total = len(self.urls)
if total > self.MAX_IMAGES:
logger.error("filters.distributed_collage: Too many images to join")
return
elif total == 0:
logger.error("filters.distributed_collage: No images to join")
return
else:
self.urls = self.urls[: self.MAX_IMAGES]
self.max_age = self.context.config.MAX_AGE
self._calculate_dimensions()
await self._fetch_images()
self.context.request.max_age = self.max_age
def load(self, buffer, extension):
self.extension = extension
if extension is None:
mime = self.get_mimetype(buffer)
self.extension = EXTENSION.get(mime, ".jpg")
if self.extension == ".svg":
buffer = self.convert_svg_to_png(buffer)
image_or_frames = self.create_image(buffer)
if image_or_frames is None:
return
if METADATA_AVAILABLE:
try:
self.metadata = ImageMetadata.from_buffer(buffer)
self.metadata.read()
except Exception as error: # pylint: disable=broad-except
logger.error("Error reading image metadata: %s", error)
if self.context.config.ALLOW_ANIMATED_GIFS and isinstance(
image_or_frames, (list, tuple)
):
self.image = image_or_frames[0]
if len(image_or_frames) > 1:
self.multiple_engine = MultipleEngine(self)
for frame in image_or_frames:
self.multiple_engine.add_frame(frame)
self.wrap(self.multiple_engine)
else:
self.image = image_or_frames
if self.source_width is None:
self.source_width = self.size[0]
if self.source_height is None:
self.source_height = self.size[1]
def get_pdf_page(context, file_path):
"""
A context manager that extracts a single page out of a pdf file and
stores it in a temporary file. Returns the path of the temporary file
or None in case of failure.
"""
import subprocess
import tempfile
import os
gspath = '/usr/bin/gs'
# Fail nicely when ffmpeg cannot be found
if not os.path.exists(gspath):
logger.error('%s does not exist, please configure Ghostscript', gspath)
yield None
return
# Prepare temporary file
f, image_path = tempfile.mkstemp('.jpg')
os.close(f)
# Extract image
try:
cmd = [
gspath, '-sDEVICE=jpeg', '-dDOINTERPOLATE', '-dCOLORSCREEN',
'-dFirstPage=1', '-dLastPage=1', '-dPDFFitPage', '-dBATCH',
'-dJPEGQ=85', '-r120x120', '-dNOPAUSE',
'-sOutputFile="' + image_path + '"', file_path
]
subprocess.check_call(cmd)
yield image_path
except:
logger.exception('Cannot extract image frame from %s', file_path)
yield None
finally:
# Cleanup
try_to_delete(image_path)
def distributed_collage(self, callback, orientation, alignment, urls):
self.callback = callback
self.orientation = orientation
self.alignment = alignment
self.urls = urls.split('|')
self.images = {}
total = len(self.urls)
if total > self.MAX_IMAGES:
logger.error(
'filters.distributed_collage: Too many images to join')
callback()
elif total == 0:
logger.error('filters.distributed_collage: No images to join')
callback()
else:
self.urls = self.urls[:self.MAX_IMAGES]
self.max_age = self.context.config.MAX_AGE
self._calculate_dimensions()
yield self._fetch_images()
self.context.request.max_age = self.max_age
def create_preview(self, url, resolution=200):
out_io = BytesIO()
try:
http_client = httpclient.AsyncHTTPClient()
response = yield http_client.fetch(url)
if not response.error:
try:
with (Image(blob=response.body,
resolution=resolution)) as source:
single_image = source.sequence[
0] # Just work on first page
with Image(single_image) as i:
i.format = 'png'
i.background_color = Color(
'white') # Set white background.
i.alpha_channel = 'remove' # Remove transparency and replace with bg.
i.save(file=out_io)
raise gen.Return(out_io.getvalue())
except WandException as e:
logger.exception('[PDFHander.create_preview] %s', e)
raise gen.Return(None)
else:
logger.error('STATUS: %s - Failed to get pdf from url %s' %
(str(400), url))
raise tornado.web.HTTPError(400)
except httpclient.HTTPError as e:
logger.error('STATUS: %s - Failed to get pdf from url %s' %
(str(e.code), url))
valid_status_code = httputil.responses.get(e.code)
if valid_status_code:
self._error(e.code)
else:
raise tornado.web.HTTPError(400)
finally:
out_io.close()
def get_video_frame(context, url, normalize_url_func):
"""
A context manager that extracts a single frame out of a video file and
stores it in a temporary file. Returns the path of the temporary file
or None in case of failure.
Depends on FFMPEG_PATH from Thumbor's configuration.
"""
url = normalize_url_func(url)
# Fail nicely when ffmpeg cannot be found
if not os.path.exists(context.config.FFMPEG_PATH):
logger.error('%s does not exist, please configure FFMPEG_PATH', context.config.FFMPEG_PATH)
yield None
return
# Prepare temporary file
f, image_path = tempfile.mkstemp('.jpg')
os.close(f)
# Extract image
try:
cmd = [
context.config.FFMPEG_PATH,
'-i', url,
'-ss', '00:00:01.000',
'-vframes', '1',
'-y',
'-nostats',
'-loglevel', 'error',
image_path
]
subprocess.check_call(cmd)
yield image_path
except:
logger.exception('Cannot extract image frame from %s', url)
yield None
finally:
# Cleanup
try_to_delete(image_path)
def __init__(self, context, importer):
'''
:param context:
:param importer:
'''
ThumborContextImporter.__init__(self, context, importer)
# Dynamically load registered modules
for name in self._community_modules:
if hasattr(importer, name):
init = getattr(importer, name)
if not hasattr(init, '__call__'):
logger.error(
"Attr {attr} of object {obj} is not callable".format(
attr=name,
obj=importer,
))
instance = getattr(importer, name)(context)
setattr(self, name, instance)
else:
logger.warning("Module {name} is not configured.".format(
name=name.upper()))
setattr(self, name, None)
def on_image_fetch(self):
if (self.is_any_failed()):
logger.error('filters.distributed_collage: Some images failed')
self.callback()
elif self.is_all_fetched():
self.assembly()
def get_image(self):
"""
This function is called after the PRE_LOAD filters have been applied.
It applies the AFTER_LOAD filters on the result, then crops the image.
"""
try:
result = yield self._fetch(self.context.request.image_url)
if not result.successful:
if result.loader_error == LoaderResult.ERROR_NOT_FOUND:
self._error(404)
return
elif result.loader_error == LoaderResult.ERROR_UPSTREAM:
# Return a Bad Gateway status if the error came from upstream
self._error(502)
return
elif result.loader_error == LoaderResult.ERROR_TIMEOUT:
# Return a Gateway Timeout status if upstream timed out (i.e. 599)
self._error(504)
return
else:
self._error(500)
return
except Exception as e:
msg = '[BaseHandler] get_image failed for url `{url}`. error: `{error}`'.format(
url=self.context.request.image_url, error=e)
self.log_exception(*sys.exc_info())
if 'cannot identify image file' in e.message:
logger.warning(msg)
self._error(400)
else:
logger.error(msg)
self._error(500)
return
normalized = result.normalized
buffer = result.buffer
engine = result.engine
req = self.context.request
if engine is None:
if buffer is None:
self._error(504)
return
engine = self.context.request.engine
try:
engine.load(buffer, self.context.request.extension)
except Exception:
self._error(504)
return
def transform():
self.normalize_crops(normalized, req, engine)
if req.meta:
self.context.request.engine = JSONEngine(
engine, req.image_url, req.meta_callback)
after_transform_cb = functools.partial(self.after_transform,
self.context)
Transformer(self.context).transform(after_transform_cb)
self.filters_runner.apply_filters(thumbor.filters.PHASE_AFTER_LOAD,
transform)
async def _fetch_images(self):
crypto = CryptoURL(key=self.context.server.security_key)
image_ops = []
if not hasattr(self.context.config, "DISTRIBUTED_COLLAGE_FILTER_HTTP_LOADER"):
self.context.config.DISTRIBUTED_COLLAGE_FILTER_HTTP_LOADER = (
"thumbor.loaders.http_loader"
)
self.context.modules.importer.import_item(
"DISTRIBUTED_COLLAGE_FILTER_HTTP_LOADER"
)
loader = self.context.modules.importer.distributed_collage_filter_http_loader
for i, url in enumerate(self.urls):
width = (
self.image_width if i < len(self.urls) - 1 else self.last_image_width
)
height = (
self.context.request.height
or self.context.transformer.get_target_dimensions()[1]
)
params = {
"width": int(width),
"height": int(height),
"image_url": url,
"smart": True,
"halign": "center",
"valign": "middle",
"filters": ["quality(100)"],
}
thumbor_host = getattr(
self.context.config,
"DISTRIBUTED_COLLAGE_FILTER_THUMBOR_SERVER_URL",
"%s://%s"
% (
self.context.request_handler.request.protocol,
self.context.request_handler.request.host,
),
)
encrypted_url = "%s%s" % (thumbor_host, crypto.generate(**params))
image_ops.append(await loader.load(self.context, encrypted_url))
successful = all([image.successful for image in image_ops])
if not successful:
logger.error(
"Retrieving at least one of the collaged images failed: %s"
% (
", ".join(
[image.error for image in image_ops if not image.successful]
),
)
)
return
max_age = min(
[
self.get_max_age(image.metadata.get("Cache-Control"), self.max_age)
for image in image_ops
]
)
self.assembly_images(image_ops)
def _error(self, status, msg=None):
self.set_status(status)
if msg is not None:
logger.error(msg)
self.finish()
def run_and_check_ssim_and_size(
self,
url,
mediawiki_reference_thumbnail,
perfect_reference_thumbnail,
expected_width,
expected_height,
expected_ssim,
size_tolerance,
):
"""Request URL and check ssim and size.
Arguments:
url -- thumbnail URL
mediawiki_reference_thumbnail -- reference thumbnail file
expected_width -- expected thumbnail width
expected_height -- expected thumbnail height
expected_ssim -- minimum SSIM score
size_tolerance -- maximum file size ratio between reference and result
perfect_reference_thumbnail -- perfect lossless version of the target thumbnail, for visual comparison
"""
try:
result = self.fetch(url)
except Exception as e:
assert False, 'Exception occured: %r' % e
assert result is not None, 'No result'
assert result.code == 200, 'Response code: %s' % result.code
result.buffer.seek(0)
generated = Image.open(result.buffer)
expected_path = os.path.join(
os.path.dirname(__file__),
'thumbnails',
mediawiki_reference_thumbnail
)
visual_expected_path = os.path.join(
os.path.dirname(__file__),
'thumbnails',
perfect_reference_thumbnail
)
visual_expected = Image.open(visual_expected_path).convert(generated.mode)
assert generated.size[0] == expected_width, \
'Width differs: %d (should be == %d)\n' % (generated.size[0], expected_width)
assert generated.size[1] == expected_height, \
'Height differs: %d (should be == %d)\n' % (generated.size[1], expected_height)
ssim = compute_ssim(generated, visual_expected)
try:
assert ssim >= expected_ssim, 'Images too dissimilar: %f (should be >= %f)\n' % (ssim, expected_ssim)
except AssertionError as e:
output_file = NamedTemporaryFile(delete=False)
output_file.write(result.buffer.getvalue())
output_file.close()
logger.error('Dumped generated test image for debugging purposes: %s' % output_file.name)
raise e
expected_filesize = float(os.path.getsize(expected_path))
generated_filesize = float(len(result.buffer.getvalue()))
ratio = generated_filesize / expected_filesize
assert ratio <= size_tolerance, \
'Generated file bigger than size tolerance: %f (should be <= %f)' % (ratio, size_tolerance)
return result.buffer
async def get_image(self):
"""
This function is called after the PRE_LOAD filters have been applied.
It applies the AFTER_LOAD filters on the result, then crops the image.
"""
try:
result = await self._fetch(self.context.request.image_url)
if not result.successful:
if result.loader_error == LoaderResult.ERROR_NOT_FOUND:
self._error(404)
return
if result.loader_error == LoaderResult.ERROR_UPSTREAM:
# Return a Bad Gateway status if the error
# came from upstream
self._error(502)
return
if result.loader_error == LoaderResult.ERROR_TIMEOUT:
# Return a Gateway Timeout status if upstream
# timed out (i.e. 599)
self._error(504)
return
if isinstance(result.loader_error, int):
self._error(result.loader_error)
return
if (hasattr(result, "engine_error") and result.engine_error
== EngineResult.COULD_NOT_LOAD_IMAGE):
self._error(400)
return
self._error(500)
return
except Exception as error:
msg = (
"[BaseHandler] get_image failed for url `{url}`. error: `{error}`"
).format(url=self.context.request.image_url, error=error)
self.log_exception(*sys.exc_info())
if "cannot identify image file" in str(error):
logger.warning(msg)
self._error(400)
else:
logger.error(msg)
self._error(500)
return
normalized = result.normalized
buffer = result.buffer
engine = result.engine
req = self.context.request
if engine is None:
if buffer is None:
self._error(504)
return
engine = self.context.request.engine
try:
engine.load(buffer, self.context.request.extension)
except Exception as error:
logger.exception("Loading image failed with %s", error)
self._error(504)
return
self.context.transformer = Transformer(self.context)
await self.filters_runner.apply_filters(
thumbor.filters.PHASE_AFTER_LOAD)
self.normalize_crops(normalized, req, engine)
if req.meta:
self.context.transformer.engine = self.context.request.engine = JSONEngine(
engine, req.image_url, req.meta_callback)
await self.context.transformer.transform()
await self.after_transform()
def get_image(self):
"""
This function is called after the PRE_LOAD filters have been applied.
It applies the AFTER_LOAD filters on the result, then crops the image.
"""
if self.context.request.mgnlogin:
mgnl_auth = ""
if self.context.config.MGNLOGIN_USER and self.context.config.MGNLOGIN_PASS:
mgnl_auth = "?mgnlUserId=%s&mgnlUserPSWD=%s" % (
self.context.config.MGNLOGIN_USER,
self.context.config.MGNLOGIN_PASS)
req_image_url = self.context.request.image_url + mgnl_auth
else:
req_image_url = self.context.request.image_url
try:
result = yield self._fetch(req_image_url)
if not result.successful:
if result.loader_error == LoaderResult.ERROR_NOT_FOUND:
self._error(404)
return
elif result.loader_error == LoaderResult.ERROR_UPSTREAM:
# Return a Bad Gateway status if the error came from upstream
self._error(502)
return
elif result.loader_error == LoaderResult.ERROR_TIMEOUT:
# Return a Gateway Timeout status if upstream timed out (i.e. 599)
self._error(504)
return
elif isinstance(result.loader_error, int):
self._error(result.loader_error)
return
elif hasattr(
result, 'engine_error'
) and result.engine_error == EngineResult.COULD_NOT_LOAD_IMAGE:
self._error(400)
return
else:
self._error(500)
return
except Exception as e:
msg = '[BaseHandler] get_image failed for url `{url}`. error: `{error}`'.format(
url=self.context.request.image_url, error=e)
self.log_exception(*sys.exc_info())
if 'cannot identify image file' in e.message:
logger.warning(msg)
self._error(400)
else:
logger.error(msg)
self._error(500)
return
normalized = result.normalized
buffer = result.buffer
engine = result.engine
req = self.context.request
if engine is None:
if buffer is None:
self._error(504)
return
engine = self.context.request.engine
try:
engine.load(buffer, self.context.request.extension)
except Exception:
self._error(504)
return
self.context.transformer = Transformer(self.context)
def transform():
self.normalize_crops(normalized, req, engine)
if req.meta:
self.context.transformer.engine = \
self.context.request.engine = \
JSONEngine(engine, req.image_url, req.meta_callback)
self.context.transformer.transform(self.after_transform)
self.filters_runner.apply_filters(thumbor.filters.PHASE_AFTER_LOAD,
transform)
async def red_eye(self) -> None:
if not OPENCV_AVAILABLE:
logger.error(
"Can't use red eye removal filter if OpenCV and NumPy are not available."
)
return
faces = [
face for face in self.context.request.focal_points
if face.origin == "Face Detection"
]
if not faces:
return
mode, data = self.engine.image_data_as_rgb()
mode = mode.lower()
size = self.engine.size
image = np.ndarray(
shape=(size[1], size[0], 4 if mode == "rgba" else 3),
dtype="|u1",
buffer=data,
).copy()
for face in faces:
face_x = int(face.x - face.width / 2)
face_y = int(face.y - face.height / 2)
face_image = image[face_y:face_y + face.height,
face_x:face_x + face.width]
eye_rects = self.cascade.detectMultiScale(
face_image,
scaleFactor=HAAR_SCALE,
minNeighbors=MIN_NEIGHBORS,
minSize=MIN_SIZE,
)
for pos_x, pos_y, width, height in eye_rects:
# Crop the eye region
eye_image = face_image[pos_y:pos_y + height,
pos_x:pos_x + width]
# split the images into 3 channels
red, green, blue = cv2.split(eye_image)
# Add blue and green channels
blue_green = cv2.add(blue, green)
mean = blue_green // 2
# threshold the mask based on red color and combination of blue and green color
mask = ((red > RED_THRESHOLD * mean) &
(red > 60)).astype(np.uint8) * 255
# Some extra region may also get detected , we find the largest region
# find all contours
contours_return = cv2.findContours(
mask.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_NONE) # It return contours and Hierarchy
if len(contours_return) == 2:
contours, _ = contours_return
else:
_, contours, _ = contours_return
# find contour with max area
max_area = 0
max_cont = None
for cont in contours:
area = cv2.contourArea(cont)
if area > max_area:
max_area = area
max_cont = cont
if max_cont is None:
continue
mask = mask * 0 # Reset the mask image to complete black image
# draw the biggest contour on mask
cv2.drawContours(mask, [max_cont], 0, (255), -1)
# Close the holes to make a smooth region
mask = cv2.morphologyEx(
mask,
cv2.MORPH_CLOSE,
cv2.getStructuringElement(cv2.MORPH_DILATE, (5, 5)),
)
mask = cv2.dilate(mask, (3, 3), iterations=3)
# The information of only red color is lost,
# So we fill the mean of blue and green color in all
# three channels(BGR) to maintain the texture
# Fill this black mean value to masked image
mean = cv2.bitwise_and(mean, mask) # mask the mean image
mean = cv2.cvtColor(
mean, cv2.COLOR_GRAY2RGB) # convert mean to 3 channel
mask = cv2.cvtColor(
mask, cv2.COLOR_GRAY2RGB) # convert mask to 3 channel
eye = (cv2.bitwise_and(~mask, eye_image) + mean
) # Copy the mean color to masked region to color image
face_image[pos_y:pos_y + height, pos_x:pos_x + width] = eye
self.engine.set_image_data(image.tobytes())
def get_image(self):
"""
This function is called after the PRE_LOAD filters have been applied.
It applies the AFTER_LOAD filters on the result, then crops the image.
"""
try:
result = yield self._fetch(
self.context.request.image_url
)
if not result.successful:
if result.loader_error == LoaderResult.ERROR_NOT_FOUND:
self._error(404)
return
elif result.loader_error == LoaderResult.ERROR_FORBIDDEN:
blacklist = yield self.get_blacklist_contents()
blacklist += self.context.request.image_url + "\n"
logger.warning('403 Adding to blacklist: %s' % self.context.request.image_url)
self.context.modules.storage.put('blacklist.txt', blacklist)
self._error(404)
return
elif result.loader_error == LoaderResult.ERROR_UPSTREAM:
# Return a Bad Gateway status if the error came from upstream
self._error(502)
return
elif result.loader_error == LoaderResult.ERROR_TIMEOUT:
# Return a Gateway Timeout status if upstream timed out (i.e. 599)
self._error(504)
return
elif isinstance(result.loader_error, int):
self._error(result.loader_error)
return
elif hasattr(result, 'engine_error') and result.engine_error == EngineResult.COULD_NOT_LOAD_IMAGE:
self._error(400)
return
else:
self._error(500)
return
except Exception as e:
url = self.context.request.image_url if self.context.request else '-|-'
msg = '[BaseHandler] get_image failed for url `{url}`. error: `{error}`'.format(
url=url,
error=e
)
self.log_exception(*sys.exc_info())
if 'cannot identify image file' in e.message:
logger.warning(msg)
self._error(400)
else:
logger.error(msg)
self._error(500)
return
normalized = result.normalized
buffer = result.buffer
engine = result.engine
req = self.context.request
if engine is None:
if buffer is None:
self._error(504)
return
engine = self.context.request.engine
try:
engine.load(buffer, self.context.request.extension)
except Exception:
self._error(504)
return
self.context.transformer = Transformer(self.context)
def transform():
self.normalize_crops(normalized, req, engine)
if req.meta:
self.context.transformer.engine = \
self.context.request.engine = \
JSONEngine(engine, req.image_url, req.meta_callback)
self.context.transformer.transform(self.after_transform)
self.filters_runner.apply_filters(thumbor.filters.PHASE_AFTER_LOAD, transform)
