def _resized_image(self, file):
(max_width, max_height) = self._max_dimensions
file_ext = filename_extension(file)
say = self._say
if file.find('-reduced') > 0:
new_file = file
else:
new_file = filename_basename(file) + '-reduced' + file_ext
if path.exists(new_file) and readable(new_file):
(image_width, image_height) = image_dimensions(new_file)
if image_width < max_width and image_height < max_height:
say.info('Using reduced image found in {}'.format(
relative(new_file)))
return new_file
else:
# We found a "-reduced" file, perhaps from a previous run, but
# for the current set of services, dimension are too large.
if __debug__:
log('existing resized file larger than {}x{}: {}',
max_width, max_height, new_file)
say.info('Dimensions too large; reducing dimensions: {}'.format(
relative(file)))
(resized, error) = reduced_image_dimensions(file, new_file, max_width,
max_height)
if error:
say.error('Failed to re-dimension {}: {}'.format(
relative(file), error))
return None
return resized
def annotated_image(file, text_boxes, service):
service_name = service.name()
fig, axes = plt.subplots(nrows=1, ncols=1, figsize=(20, 20))
axes.get_xaxis().set_visible(False)
axes.get_yaxis().set_visible(False)
axes.set_title(service_name, color='r', fontweight='bold', fontsize=22)
if __debug__:
log('reading image file for {}: {}', service_name, relative(file))
img = mpimg.imread(file)
axes.imshow(img, cmap="gray")
props = dict(facecolor='white', alpha=0.7)
if text_boxes:
if __debug__:
log('adding {} annotations for {}', len(text_boxes), service_name)
polygons = [(item.boundingBox, item.text) for item in text_boxes]
for polygon in polygons:
vertices = [(polygon[0][i], polygon[0][i + 1])
for i in range(0, len(polygon[0]), 2)]
x = max(0, vertices[0][0] - 4)
y = max(0, vertices[0][1] - 8)
text = polygon[1]
plt.text(x, y, text, color='r', fontsize=11, va="top", bbox=props)
if __debug__:
log('generating png for {} for {}', service_name, relative(file))
buf = io.BytesIO()
fig.savefig(buf, format='png', dpi=300, bbox_inches='tight', pad_inches=0)
buf.flush()
buf.seek(0)
plt.close(fig)
return buf
def _smaller_file(self, file):
if not file:
return None
say = self._say
file_ext = filename_extension(file)
if file.find('-reduced') > 0:
new_file = file
else:
new_file = filename_basename(file) + '-reduced' + file_ext
if path.exists(new_file):
if image_size(new_file) < self._max_size:
say.info('Reusing resized image found in {}'.format(
relative(new_file)))
return new_file
else:
# We found a "-reduced" file, perhaps from a previous run, but
# for the current set of services, it's larger than allowed.
if __debug__:
log('existing resized file larger than {}b: {}',
humanize.intcomma(self._max_size), new_file)
say.info('Size too large; reducing size: {}'.format(relative(file)))
(resized, error) = reduced_image_size(file, new_file, self._max_size)
if error:
say.error('Failed to resize {}: {}'.format(relative(file), error))
return None
return resized
def file_after_converting(file, to_format, tool, spinner):
new_file = filename_basename(file) + '.' + to_format
if path.exists(new_file):
spinner.update('Using converted image found in {}'.format(
relative(new_file)))
return new_file
else:
spinner.update('Converting to {} format: {}'.format(
to_format, relative(file)))
(converted, error) = converted_image(file, to_format)
if not converted:
spinner.fail('Failed to convert {}: {}'.format(
relative(file), error))
return None
return converted
def file_after_resizing(file, tool, spinner):
file_ext = filename_extension(file)
new_file = filename_basename(file) + '-reduced.' + file_ext
if path.exists(new_file):
spinner.update('Using reduced image found in {}'.format(
relative(new_file)))
return new_file
else:
spinner.update('Original image too large; reducing size')
(resized, error) = reduced_image(file, tool.max_dimensions())
if not resized:
spinner.fail('Failed to resize {}: {}'.format(relative(
file, error)))
return None
return resized
def _save_output(self, result, file):
say = self._say
# First perform some sanity checks.
if result is None:
say.warn('No data for {}'.format(file))
return
if isinstance(result, tuple):
# Assumes 2 elements: data, and error
(data, error) = result
if error:
say.error('Error: {}'.format(error))
say.warn('Unable to write {}'.format(file))
return
else:
result = data
if __debug__: log('writing output to file {}', relative(file))
if isinstance(result, str):
with open(file, 'w') as f:
f.write(result)
elif isinstance(result, io.BytesIO):
with open(file, 'wb') as f:
shutil.copyfileobj(result, f)
else:
# There's no other type in the code, so if we get here ...
raise InternalError(
'Unexpected data in save_output() -- please report this.')
def _converted_file(self, file, to_format, dest_dir):
basename = path.basename(filename_basename(file))
new_file = path.join(dest_dir, basename + '.' + to_format)
say = self._say
if path.exists(new_file):
say.info('Using already converted image in {}'.format(
relative(new_file)))
return new_file
else:
say.info('Converting to {} format: {}'.format(
to_format, relative(file)))
(converted, error) = converted_image(file, to_format, new_file)
if error:
say.error('Failed to convert {}: {}'.format(
relative(file), error))
return None
return converted
def _get(self, item, base_name, index):
# Shortcuts to make the code more readable.
output_dir = self._output_dir
say = self._say
# For URLs, we download the corresponding files and name them with
# the base_name.
if is_url(item):
# First make sure the URL actually points to an image.
if __debug__: log('testing if URL contains an image: {}', item)
try:
response = urllib.request.urlopen(item)
except Exception as ex:
say.warn('Skipping URL due to error: {}'.format(ex))
return (None, None)
if response.headers.get_content_maintype() != 'image':
say.warn('Did not find an image at {}'.format(item))
return (None, None)
orig_fmt = response.headers.get_content_subtype()
base = '{}-{}'.format(base_name, index)
# If we weren't given an output dir, then for URLs, we have no
# choice but to use the current dir to download the file.
# Important: don't change self._output_dir because if other
# inputs *are* files, then those files will need other output dirs.
if not output_dir:
output_dir = os.getcwd()
file = path.realpath(path.join(output_dir, base + '.' + orig_fmt))
if not download_file(item, file, say):
say.warn('Unable to download {}'.format(item))
return (None, None)
url_file = path.realpath(path.join(output_dir, base + '.url'))
with open(url_file, 'w') as f:
f.write(url_file_content(item))
say.info('Wrote URL to {}'.format(relative(url_file)))
else:
file = path.realpath(path.join(os.getcwd(), item))
orig_fmt = filename_extension(file)[1:]
if __debug__:
log('{} has original format {}', relative(file), orig_fmt)
return (file, orig_fmt)
def _send(self, file, service, dest_dir):
'''Send the "file" to the service named "service" and write output in
directory "dest_dir".
'''
say = self._say
use_color = say.use_color()
color = service.name_color()
service_name = styled(service.name(),
color) if use_color else service.name()
say.info(
'Sending to {} and waiting for response ...'.format(service_name))
last_time = timer()
try:
result = service.result(file)
except AuthFailure as ex:
raise AuthFailure('Unable to use {}: {}'.format(service, ex))
except RateLimitExceeded as ex:
time_passed = timer() - last_time
if time_passed < 1 / service.max_rate():
say.warn('Pausing {} due to rate limits'.format(service_name))
time.sleep(1 / service.max_rate() - time_passed)
# FIXME resend after pause
if result.error:
say.error('{} failed: {}'.format(service_name, result.error))
say.warn('No result from {} for {}'.format(service_name,
relative(file)))
return None
say.info('Got result from {}.'.format(service_name))
file_name = path.basename(file)
base_path = path.join(dest_dir, file_name)
annot_path = alt_extension(base_path, str(service) + '.png')
say.info('Creating annotated image for {}.'.format(service_name))
self._save_output(annotated_image(file, result.boxes, service),
annot_path)
if self._extended_results:
txt_file = alt_extension(base_path, str(service) + '.txt')
json_file = alt_extension(base_path, str(service) + '.json')
say.info('Saving all data for {}.'.format(service_name))
self._save_output(json.dumps(result.data), json_file)
say.info('Saving extracted text for {}.'.format(service_name))
self._save_output(result.text, txt_file)
# Return the annotated image file b/c we use it for the summary grid.
return annot_path
def run(classes, item, index, base_name, output_dir, creds_dir, annotate, say):
spinner = ProgressIndicator(say.use_color(), say.be_quiet())
try:
spinner.start('Starting on {}'.format(relative(item)))
if is_url(item):
# Make sure the URLs point to images.
if __debug__: log('Testing if URL contains an image: {}', item)
try:
response = request.urlopen(item)
except Exception as err:
if __debug__:
log('Network access resulted in error: {}', str(err))
spinner.fail('Skipping URL due to error: {}'.format(err))
return
if response.headers.get_content_maintype() != 'image':
spinner.fail('Did not find an image at {}'.format(item))
return
fmt = response.headers.get_content_subtype()
base = '{}-{}'.format(base_name, index)
file = path.realpath(path.join(output_dir, base + '.' + fmt))
error = download(item, file)
if not error:
spinner.update('Wrote contents to {}'.format(relative(file)))
else:
spinner.fail('Failed to download {}: {}'.format(item, error))
return
url_file = path.realpath(path.join(output_dir, base + '.url'))
with open(url_file, 'w') as f:
f.write(url_file_content(item))
spinner.update('Wrote URL to {}'.format(relative(url_file)))
else:
file = path.realpath(path.join(os.getcwd(), item))
fmt = filename_extension(file)
dest_dir = output_dir if output_dir else path.dirname(file)
if not writable(dest_dir):
say.fatal('Cannot write output in {}.'.format(dest_dir))
return
# Iterate over the methods.
for method_class in classes:
method = method_class()
method.init_credentials(creds_dir)
last_time = timer()
# If need to convert format, best do it after resizing original fmt.
need_convert = fmt not in method.accepted_formats()
# Test the dimensions, not bytes, because of compression.
if image_dimensions(file) > method.max_dimensions():
file = file_after_resizing(file, method, spinner)
if file and need_convert:
file = file_after_converting(file, 'jpg', method, spinner)
if not file:
return
spinner.update('Sending to {} {}'.format(
color(method, 'white', say.use_color()),
# Need explicit color research or colorization goes wrong.
color('and waiting for response', 'info', say.use_color())))
try:
result = method.result(file)
except RateLimitExceeded as err:
time_passed = timer() - last_time
if time_passed < 1 / method.max_rate():
spinner.warn('Pausing due to rate limits')
time.sleep(1 / method.max_rate() - time_passed)
if result.error:
spinner.fail(result.error)
return
file_name = path.basename(file)
base_path = path.join(dest_dir, file_name)
txt_file = alt_extension(base_path, str(method) + '.txt')
json_file = alt_extension(base_path, str(method) + '.json')
annot_file = alt_extension(base_path, str(method) + '.jpg')
spinner.update('Text -> {}'.format(relative(txt_file)))
save_output(result.text, txt_file)
spinner.update('All data -> {}'.format(relative(json_file)))
save_output(json.dumps(result.data), json_file)
if annotate:
spinner.update('Annotated image -> {}'.format(
relative(annot_file)))
save_output(annotated_image(file, result.boxes), annot_file)
spinner.stop('Done with {}'.format(relative(item)))
except (KeyboardInterrupt, UserCancelled) as err:
spinner.warn('Interrupted')
raise
except AuthenticationFailure as err:
spinner.fail('Unable to continue using {}: {}'.format(method, err))
return
except Exception as err:
spinner.fail(say.error_text('Stopping due to a problem'))
raise
def run_services(self, item, index, base_name):
'''Run all requested services on the image indicated by "item", using
"index" and "base_name" to construct a download copy of the item if
it has to be downloaded from a URL first.
'''
# Shortcuts to make the code more readable.
services = self._services
output_dir = self._output_dir
say = self._say
try:
say.info('Starting on {}'.format(
styled(item, 'white') if say.use_color() else item))
(file, orig_fmt) = self._get(item, base_name, index)
if not file:
return
dest_dir = output_dir if output_dir else path.dirname(file)
if not writable(dest_dir):
say.error('Cannot write output in {}.'.format(dest_dir))
return
# Sanity check
if not path.getsize(file) > 0:
say.warn('Skipping zero-length file {}'.format(relative(file)))
return
# Save grid file name now, because it's based on the original file.
basename = path.basename(filename_basename(file))
grid_file = path.realpath(
path.join(dest_dir, basename + '.all-results.png'))
# We will usually delete temporary files we create.
to_delete = set()
# Normalize to the lowest common denominator.
(new_file,
intermediate_files) = self._normalized(file, orig_fmt, dest_dir)
if not new_file:
say.warn('Skipping {}'.format(relative(file)))
return
file = new_file
if intermediate_files:
to_delete.update(intermediate_files)
# Send the file to the services. If the number of threads is set
# to 1, we force non-thread-pool execution to make debugging easier.
results = []
if self._num_threads == 1:
results = [self._send(file, s, dest_dir) for s in services]
else:
with ThreadPoolExecutor(
max_workers=self._num_threads) as executor:
results = list(
executor.map(self._send, repeat(file), iter(services),
repeat(dest_dir)))
# If a service failed for some reason (e.g., a network glitch), we
# get no result back. Remove empty results & go on with the rest.
results = [x for x in results if x is not None]
to_delete.update(results)
# Create grid file if requested.
if self._make_grid:
say.info('Creating results grid image: {}'.format(
relative(grid_file)))
create_image_grid(results, grid_file, max_horizontal=2)
# Clean up after ourselves.
if self._make_grid and not self._extended_results:
for image_file in to_delete:
delete_existing(image_file)
say.info('Done with {}'.format(relative(item)))
except (KeyboardInterrupt, UserCancelled) as ex:
say.warn('Interrupted')
raise
except Exception as ex:
say.error('Stopping due to a problem')
raise