def __init__(self, methodName='runTest'):
super().__init__(methodName)
self.port = 8081
self.name = None
self.data_folder = None
self.server_address = "localhost:{}".format(self.port)
self.server = None
self.server_thread = None
self.server_process = None
self.client = None
self.converters = TExternalConverters(enable_smart_parser=False,
enable_calibre=False,
enable_cat_doc=False,
enable_xls2csv=False,
enable_office_2_txt=False)
self.pdf_ocr_folder = os.path.join(os.path.dirname(__file__),
"pdf.ocr")
self.pdf_ocr_out_folder = os.path.join(os.path.dirname(__file__),
"pdf.ocr.out")
if not os.path.exists(self.pdf_ocr_folder) or not os.path.exists(
self.pdf_ocr_out_folder):
raise Exception(
"run python update_finereader_task.py, and upload test.hft to finreader hot folder"
)
self.project_file = "converted_file_storage.json"
self.client = None
self.server_args = None
self.client_count = 0
from source_doc_http.source_doc_client import TSourceDocClient
from DeclDocRecognizer.external_convertors import TExternalConverters
from smart_parser_http.smart_parser_client import TSmartParserCacheClient
from smart_parser_http.smart_parser_server import TSmartParserHTTPServer
from common.logging_wrapper import setup_logging
import sys
import os
if __name__ == "__main__":
converters = TExternalConverters()
INPUT_SHA256 = sys.argv[1]
log_file_name = INPUT_SHA256 + ".log"
logger = setup_logging(log_file_name=log_file_name)
assert len(INPUT_SHA256) == 64
source_doc_client = TSourceDocClient(TSourceDocClient.parse_args(
["--disable-first-ping"]),
logger=logger)
file_data, file_extension = source_doc_client.retrieve_file_data_by_sha256(
INPUT_SHA256)
if file_data is None:
logger.error("cannot find source document {}".format(INPUT_SHA256))
sys.exit(1)
file_path = "{}{}".format(INPUT_SHA256, file_extension)
with open(file_path, "wb") as outp:
outp.write(file_data)
sha256, json_data = converters.run_smart_parser_official(
file_path,
logger=logger,
default_value=TSmartParserHTTPServer.SMART_PARSE_FAIL_CONSTANT)