def execute(self, *args, **kwargs):
"""
Execute the command line binary of tesseract
"""
super(Tesseract, self).execute(*args, **kwargs)
if self.command_tesseract:
image = self.converter.get_page()
try:
temporary_image_file = TemporaryFile()
shutil.copyfileobj(image, temporary_image_file)
temporary_image_file.seek(0)
arguments = ['-', '-']
keyword_arguments = {
'_in': temporary_image_file,
'_timeout': self.command_timeout
}
if self.language:
keyword_arguments['l'] = self.language
environment = os.environ.copy()
environment.update(self.environment)
keyword_arguments['_env'] = environment
try:
result = self.command_tesseract(*arguments,
**keyword_arguments)
return force_text(result.stdout)
except Exception as exception:
error_message = (
'Exception calling Tesseract with language option: {}; {}'
).format(self.language, exception)
if self.language not in self.languages:
error_message = (
'{}\nThe requested OCR language "{}" is not '
'available and needs to be installed.\n').format(
error_message, self.language)
logger.error(error_message)
raise OCRError(error_message)
else:
return result
finally:
temporary_image_file.close()