def parse_file(self, file: File, fallback_city: str):
logging.info("- Parsing: " + str(file.id) + " (" + file.name + ")")
with minio_client().get_object(minio_file_bucket,
str(file.id)) as file_handle:
recognized_text = get_ocr_text_from_pdf(file_handle.read())
if len(recognized_text) > 0:
file.parsed_text = cleanup_extracted_text(recognized_text)
file.mentioned_persons = extract_persons(file.name + "\n" +
(recognized_text or "") +
"\n")
file.locations.set(
extract_locations(file.parsed_text, fallback_city))
file.save()
else:
logging.warning("Nothing recognized")
def parse_file(self, file: File):
logging.info("- Parsing: " + str(file.id) + " (" + file.name + ")")
file_path = os.path.abspath(os.path.dirname(__name__))
file_path = os.path.join(file_path, settings.MEDIA_ROOT,
file.storage_filename)
recognized_text = get_ocr_text_from_pdf(file_path)
if len(recognized_text) > 0:
file.parsed_text = cleanup_extracted_text(recognized_text)
file.mentioned_persons = extract_persons(file.name + "\n" +
(recognized_text or "") +
"\n")
file.locations = extract_locations(recognized_text)
file.save()
else:
logging.warning("Nothing recognized")
def parse_file(self, file: File, fallback_city: str):
self.stdout.write("Parsing: " + str(file.id) + " (" + file.name + ")")
locations = extract_locations(file.parsed_text, fallback_city)
self.stdout.write("{} locations found".format(len(locations)))
file.locations.set(locations)
file.save()
def parse_file(self, file: File):
logging.info("- Parsing: " + str(file.id) + " (" + file.name + ")")
file.mentioned_persons = extract_persons(file.name + "\n" + (file.parsed_text or "") + "\n")
file.save()