import os
import re
import time
import json
import shutil
import base64
import signal
import requests
import traceback
from PIL import Image
from datetime import datetime
from django.core.management import BaseCommand
from multiprocessing import Process, Queue
from openpyxl import load_workbook, Workbook
from settings import conf
from common.mixins import LoggerMixin
from common.tools.pdf_to_img import PDFHandler
from apps.doc import consts
from apps.doc.exceptions import OCR1Exception, OCR2Exception, LTGTException
from apps.doc.ocr.wb import BSWorkbook
class TIFFHandler:
def __init__(self, path, img_save_path):
self.path = path
self.img_save_path = img_save_path
self.img_path_list = []
def extract_image(self):
os.makedirs(self.img_save_path, exist_ok=True)
tiff = Image.open(self.path)
tiff.load()
for i in range(tiff.n_frames):
try:
save_path = os.path.join(self.img_save_path, 'page_{0}.jpeg'.format(i))
tiff.seek(i)
tiff.save(save_path)
self.img_path_list.append(save_path)
except EOFError:
break
class Command(BaseCommand, LoggerMixin):
def __init__(self):
super().__init__()
self.log_base = '[folder ltgt process]'
# 处理文件开关
self.switch = True
self.amount_key_set = {'债权金额', '贷款本金', '罚息', '律师费', '案件受理费', '支付金额'}
self.ltgt_classify_mapping = {
128: '执行裁定书',
129: '民事判决书',
130: '民事调解书'
}
self.sheet_content = {
128: ('执行裁定书', ('承办法院', '案号/标号', '被执行人', '债权金额', '诉讼时间')),
129: ('民事判决书', ('承办法院', '案号/标号', '被告', '判决结果: 贷款本金', '判决结果: 罚息', '判决结果: 律师费', '判决结果: 案件受理费', '诉讼时间')),
130: ('民事调解书', ('承办法院', '案号/标号', '被告', '协议内容: 支付金额', '协议内容: 案件受理费', '诉讼时间')),
}
self.DATE_KEY = 'date'
self.CLASSIFY_KEY = 'classify'
self.RESULT_KEY = 'result'
self.daily_wb_name = 'Output_{0}.xlsx'
self.short_sleep_time = 10
self.long_sleep_time = 3600
# 睡眠时间
self.sleep_time = float(conf.SLEEP_SECOND_FOLDER)
# input folder
self.input_dirs = conf.get_namespace('LTGT_DIR_')
# seperate folder name
self.seperate_map = {
consts.IC_CLASSIFY: 'IDCard',
consts.BC_CLASSIFY: 'BankCard'
}
self.field_map = {
consts.VAT_CLASSIFY: (consts.VAT_CN_NAME, None, None, consts.VATS_FIELD_ORDER),
consts.IC_CLASSIFY: (consts.IC_CN_NAME, '有效期限', consts.IC_FIELD_ORDER_3, consts.IC_FIELD_ORDER_2),
consts.BC_CLASSIFY: (consts.BC_CN_NAME, None, None, consts.BC_FIELD_ORDER_2)
}
# ocr相关
self.ocr_url = conf.OCR_URL_FOLDER
self.ocr_url_2 = conf.OCR2_URL_FOLDER
# self.ocr_url_4 = conf.IC_URL
self.ltgt_ocr_url = conf.LTGT_URL
# 优雅退出信号：15
signal.signal(signal.SIGTERM, self.signal_handler)
def signal_handler(self, sig, frame):
self.switch = False # 停止处理文件
def license1_process(self, ocr_data, all_res, classify):
# 类别：'0'身份证， '1'居住证
license_data = ocr_data.get('data', [])
if not license_data:
return
if classify == consts.IC_CLASSIFY:
for id_card_dict in license_data:
try:
id_card_dict.pop('base64_img')
except Exception as e:
continue
all_res.extend(license_data)
def license2_process(self, ocr_data, all_res, classify, img_path):
pid, _, _, _, _, _ = consts.LICENSE_CLASSIFY_MAPPING.get(classify)
file_data = ocr_data.get('section_img')
if file_data is None:
with open(img_path, 'rb') as f:
base64_data = base64.b64encode(f.read())
# 获取解码后的base64值
file_data = base64_data.decode()
json_data_2 = {
"pid": str(pid),
"filedata": file_data
}
for times in range(consts.RETRY_TIMES):
try:
start_time = time.time()
ocr_2_response = requests.post(self.ocr_url_2, data=json_data_2)
if ocr_2_response.status_code != 200:
raise OCR2Exception('ocr_2 status code: {0}'.format(ocr_2_response.status_code))
except Exception as e:
self.folder_log.warn(
'{0} [ocr_2 failed] [times={1}] [img_path={2}] [error={3}]'.format(
self.log_base, times, img_path, traceback.format_exc()))
else:
ocr_res_2 = json.loads(ocr_2_response.text)
end_time = time.time()
speed_time = int(end_time - start_time)
self.folder_log.info(
'{0} [ocr_2 success] [img={1}] [speed_time={2}]'.format(
self.log_base, img_path, speed_time))
if ocr_res_2.get('ErrorCode') in consts.SUCCESS_CODE_SET:
if pid == consts.BC_PID:
all_res.append(ocr_res_2)
else:
# 营业执照等
for result_dict in ocr_res_2.get('ResultList', []):
res_dict = {}
for field_dict in result_dict.get('FieldList', []):
res_dict[field_dict.get('chn_key', '')] = field_dict.get('value', '')
all_res.append(res_dict)
break
@staticmethod
def parse_img_path(img_path):
# 'page_{0}_img_{1}.{2}'.format(pno, img_index, ext)
img_name, _ = os.path.splitext(os.path.basename(img_path))
if re.match(r'page_\d+_img_\d+', img_name):
part_list = img_name.split('_')
return img_name, int(part_list[1])+1, int(part_list[3])+1
else:
return img_name, 1, 1
@staticmethod
def get_path(name, img_output_dir, wb_output_dir, pdf_output_dir, seperate_dir):
time_stamp = datetime.now().strftime('%Y-%m-%d_%H:%M:%S')
new_name = '{0}_{1}'.format(time_stamp, name)
img_save_path = os.path.join(img_output_dir, new_name)
pdf_save_path = os.path.join(pdf_output_dir, new_name)
excel_name = '{0}.xlsx'.format(os.path.splitext(new_name)[0])
excel_path = os.path.join(wb_output_dir, excel_name)
seperate_path = None if seperate_dir is None else os.path.join(seperate_dir, new_name)
return img_save_path, excel_path, pdf_save_path, seperate_path
def res_process(self, all_res, excel_path, classify):
try:
wb = BSWorkbook(set(), set(), set(), set(), set())
sheet_name, key_field, side_field_order, src_field_order = self.field_map.get(classify)
ws = wb.create_sheet(sheet_name)
for res in all_res:
if key_field is not None and key_field in res:
field_order = side_field_order
else:
field_order = src_field_order
for search_field, write_field in field_order:
field_value = res.get(search_field, '')
if isinstance(field_value, list):
ws.append((write_field, *field_value))
else:
ws.append((write_field, field_value))
ws.append((None,))
wb.remove_base_sheet()
wb.save(excel_path)
except Exception as e:
self.folder_log.error('{0} [wb build error] [path={1}] [error={2}]'.format(
self.log_base, excel_path, traceback.format_exc()))
def basename(self, path):
# A basename() variant which first strips the trailing slash, if present.
# Thus we always get the last component of the path, even for directories.
sep = os.path.sep + (os.path.altsep or '')
return os.path.basename(path.rstrip(sep))
def ocr_process(self, img_path, classify, all_res, seperate_dir):
if os.path.exists(img_path):
# TODO 图片验证
with open(img_path, 'rb') as f:
base64_data = base64.b64encode(f.read())
# 获取解码后的base64值
file_data = base64_data.decode()
json_data = {
"file": file_data,
}
if seperate_dir is None:
json_data["classify"] = classify
for times in range(consts.RETRY_TIMES):
try:
start_time = time.time()
ocr_response = requests.post(self.ocr_url, json=json_data)
if ocr_response.status_code != 200:
raise OCR1Exception('{0} ocr status code: {1}'.format(self.log_base, ocr_response.status_code))
except Exception as e:
self.folder_log.warn('{0} [ocr failed] [times={1}] [img_path={2}] [error={3}]'.format(
self.log_base, times, img_path, traceback.format_exc()))
else:
ocr_res = ocr_response.json()
end_time = time.time()
speed_time = int(end_time - start_time)
self.folder_log.info('{0} [ocr success] [img={1}] [res={2}] [speed_time={3}]'.format(
self.log_base, img_path, ocr_res, speed_time))
if isinstance(ocr_res, dict):
if ocr_res.get('code') == 1:
data_list = ocr_res.get('data', [])
if isinstance(data_list, list):
for ocr_data in data_list:
if ocr_data.get('classify') == classify:
if seperate_dir is not None:
os.makedirs(seperate_dir, exist_ok=True)
real_dst = os.path.join(seperate_dir, self.basename(img_path))
if not os.path.exists(real_dst):
shutil.move(img_path, seperate_dir)
if classify in consts.LICENSE_CLASSIFY_SET_1:
self.license1_process(ocr_data, all_res, classify)
elif classify in consts.LICENSE_CLASSIFY_SET_2:
self.license2_process(ocr_data, all_res, classify, img_path)
break
else:
self.folder_log.warn('{0} [ocr failed] [img_path={1}]'.format(self.log_base, img_path))
def ltgt_ocr_process(self, img_path_list, label, path):
img_data_list = []
for img_path in img_path_list:
if os.path.exists(img_path):
with open(img_path, 'rb') as f:
base64_data = base64.b64encode(f.read())
# 获取解码后的base64值
file_data = base64_data.decode()
img_data_list.append(file_data)
json_data = {
"label": label,
"img_data_list": img_data_list
}
for times in range(consts.RETRY_TIMES):
try:
start_time = time.time()
ocr_response = requests.post(self.ltgt_ocr_url, json=json_data)
if ocr_response.status_code != 200:
raise LTGTException('{0} ltgt ocr status code: {1}'.format(self.log_base, ocr_response.status_code))
except Exception as e:
self.folder_log.warn('{0} [ltgt ocr failed] [times={1}] [path={2}] [error={3}]'.format(
self.log_base, times, path, traceback.format_exc()))
else:
ocr_res = ocr_response.json()
end_time = time.time()
speed_time = int(end_time - start_time)
self.folder_log.info('{0} [ltgt ocr success] [path={1}] [res={2}] [speed_time={3}]'.format(
self.log_base, path, ocr_res, speed_time))
return ocr_res
else:
self.folder_log.warn('{0} [ltgt ocr failed] [path={1}]'.format(self.log_base, path))
def ltgt_res_process(self, ocr_res, label, excel_path):
try:
if isinstance(ocr_res, dict):
if ocr_res.get('code') == 1:
result_dict = ocr_res.get('data', {})
wb = BSWorkbook(set(), set(), set(), set(), set())
rebuild_res = wb.ltgt_build(label, result_dict, self.amount_key_set)
wb.remove_base_sheet()
wb.save(excel_path)
return rebuild_res
except Exception as e:
self.folder_log.error('{0} [wb build error] [path={1}] [error={2}]'.format(
self.log_base, excel_path, traceback.format_exc()))
def ltgt_process(self, img_path_list, label, excel_path, path):
ocr_res = self.ltgt_ocr_process(img_path_list, label, path)
rebuild_res = self.ltgt_res_process(ocr_res, label, excel_path)
return rebuild_res
def images_process(self, img_path_list, classify, excel_path, seperate_dir):
all_res = []
for img_path in img_path_list:
self.ocr_process(img_path, classify, all_res, seperate_dir)
# if len(all_res) > 0:
self.res_process(all_res, excel_path, classify)
return all_res
def pdf_process(self, name, path, classify, img_output_dir, wb_output_dir, pdf_output_dir, seperate_dir):
if os.path.exists(path):
rebuild_res = None
try:
img_save_path, excel_path, pdf_save_path, seperate_path = self.get_path(
name, img_output_dir, wb_output_dir, pdf_output_dir, seperate_dir)
self.folder_log.info('{0} [pdf to img start] [path={1}]'.format(self.log_base, path))
pdf_handler = PDFHandler(path, img_save_path)
if classify in self.ltgt_classify_mapping:
pdf_handler.extract_page_image()
else:
pdf_handler.extract_image()
self.folder_log.info('{0} [pdf to img end] [path={1}]'.format(self.log_base, path))
except Exception as e:
self.folder_log.error('{0} [pdf to img error] [path={1}] [error={2}]'.format(
self.log_base, path, traceback.format_exc()))
raise e
else:
if classify in self.ltgt_classify_mapping:
rebuild_res = self.ltgt_process(pdf_handler.img_path_list, self.ltgt_classify_mapping[classify],
excel_path, path)
else:
rebuild_res = self.images_process(pdf_handler.img_path_list, classify, excel_path, seperate_path)
shutil.move(path, pdf_save_path)
return rebuild_res
def tif_process(self, name, path, classify, img_output_dir, wb_output_dir, tiff_output_dir, seperate_dir):
if os.path.exists(path):
rebuild_res = None
try:
img_save_path, excel_path, tiff_save_path, seperate_path = self.get_path(
name, img_output_dir, wb_output_dir, tiff_output_dir, seperate_dir)
self.folder_log.info('{0} [tiff to img start] [path={1}]'.format(self.log_base, path))
tiff_handler = TIFFHandler(path, img_save_path)
tiff_handler.extract_image()
self.folder_log.info('{0} [tiff to img end] [path={1}]'.format(self.log_base, path))
except Exception as e:
self.folder_log.error('{0} [tiff to img error] [path={1}] [error={2}]'.format(
self.log_base, path, traceback.format_exc()))
raise e
else:
if classify in self.ltgt_classify_mapping:
rebuild_res = self.ltgt_process(tiff_handler.img_path_list, self.ltgt_classify_mapping[classify],
excel_path, path)
else:
rebuild_res = self.images_process(tiff_handler.img_path_list, classify, excel_path, seperate_path)
shutil.move(path, tiff_save_path)
return rebuild_res
def img_process(self, name, path, classify, wb_output_dir, img_output_dir, pdf_output_dir, seperate_dir):
rebuild_res = None
try:
img_save_path, excel_path, _, seperate_path = self.get_path(
name, img_output_dir, wb_output_dir, pdf_output_dir, seperate_dir)
except Exception as e:
self.folder_log.error('{0} [get path error] [path={1}] [error={2}]'.format(
self.log_base, path, traceback.format_exc()))
else:
if classify in self.ltgt_classify_mapping:
rebuild_res = self.ltgt_process([path], self.ltgt_classify_mapping[classify], excel_path, path)
else:
rebuild_res = self.images_process([path], classify, excel_path, seperate_path)
shutil.move(path, img_save_path)
return rebuild_res
def wb_process(self, wb_dir, result_queue):
while self.switch:
result_list = []
date_str = None
for i in range(100):
try:
result = result_queue.get(block=False)
except Exception as e:
time.sleep(self.short_sleep_time)
else:
if date_str is None:
date_str = result[self.DATE_KEY]
result_list.append(result)
elif result[self.DATE_KEY] == date_str:
result_list.append(result)
else:
break
if date_str is None:
time.sleep(self.long_sleep_time)
continue
else:
wb_name = self.daily_wb_name.format(date_str)
wb_path = os.path.join(wb_dir, wb_name)
if os.path.isfile(wb_path):
wb = load_workbook(wb_path)
else:
wb = Workbook()
for result in result_list:
try:
if result[self.CLASSIFY_KEY] in self.sheet_content:
sheet_name, head_fields = self.sheet_content[result[self.CLASSIFY_KEY]]
else:
sheet_name, key_field, side_field_order, field_order = self.field_map[result[self.CLASSIFY_KEY]]
if key_field is not None and key_field in result[self.RESULT_KEY]:
head_fields = [a for a, _ in side_field_order]
else:
head_fields = [a for a, _ in field_order]
row = []
for field in head_fields:
row.append(result[self.RESULT_KEY].get(field))
if sheet_name in wb.sheetnames:
ws = wb.get_sheet_by_name(sheet_name)
else:
ws = wb.create_sheet(sheet_name)
ws.append(head_fields)
ws.append(row)
except Exception as e:
self.folder_log.info('{0} [daily wb failed] [result={1}] [error={2}]'.format(
self.log_base, result, traceback.format_exc()))
wb.save(wb_path)
def folder_process(self, input_dir, classify, is_combined, result_queue):
while not os.path.isdir(input_dir):
self.folder_log.info('{0} [input dir is not dir] [input_dir={1}]'.format(self.log_base, input_dir))
if self.switch:
time.sleep(self.sleep_time)
continue
else:
return
output_dir = os.path.join(os.path.dirname(input_dir), 'Output')
seperate_dir = os.path.join(output_dir, self.seperate_map.get(classify, 'Unknown')) if is_combined else None
img_output_dir = os.path.join(output_dir, 'image')
wb_output_dir = os.path.join(output_dir, 'excel')
pdf_output_dir = os.path.join(output_dir, 'pdf')
tiff_output_dir = os.path.join(output_dir, 'tiff')
failed_output_dir = os.path.join(output_dir, 'failed')
os.makedirs(output_dir, exist_ok=True)
os.makedirs(img_output_dir, exist_ok=True)
os.makedirs(wb_output_dir, exist_ok=True)
os.makedirs(pdf_output_dir, exist_ok=True)
os.makedirs(tiff_output_dir, exist_ok=True)
os.makedirs(failed_output_dir, exist_ok=True)
if seperate_dir is not None:
os.makedirs(seperate_dir, exist_ok=True)
os_error_filename_set = set()
while self.switch:
# if not os.path.isdir(input_dir):
# self.folder_log.info('{0} [input dir is not dir] [input_dir={1}]'.format(self.log_base, input_dir))
# time.sleep(self.sleep_time)
# continue
# 1. 从input dir获取pdf or image
list_dir = os.listdir(input_dir)
if not list_dir and len(os_error_filename_set) == 0:
self.folder_log.info('{0} [input dir empty] [input_dir={1}]'.format(self.log_base, input_dir))
time.sleep(self.sleep_time)
continue
all_file_set = set(list_dir)
true_file_set = all_file_set - os_error_filename_set
if len(true_file_set) == 0 and len(os_error_filename_set) > 0:
true_file_set.add(os_error_filename_set.pop())
for name in true_file_set:
path = os.path.join(input_dir, name)
try:
if os.path.isfile(path):
self.folder_log.info('{0} [file start] [path={1}]'.format(self.log_base, path))
if name.endswith('.pdf') or name.endswith('.PDF'):
result = self.pdf_process(name, path, classify, img_output_dir, wb_output_dir,
pdf_output_dir, seperate_dir)
elif name.endswith('.tif') or name.endswith('.TIF'):
result = self.tif_process(name, path, classify, img_output_dir, wb_output_dir,
tiff_output_dir, seperate_dir)
else:
result = self.img_process(name, path, classify, wb_output_dir, img_output_dir,
pdf_output_dir, seperate_dir)
self.folder_log.info('{0} [file end] [path={1}]'.format(self.log_base, path))
else:
result = None
self.folder_log.info('{0} [path is dir] [path={1}]'.format(self.log_base, input_dir))
failed_path = os.path.join(failed_output_dir, '{0}_{1}'.format(time.time(), name))
shutil.move(path, failed_path)
except OSError:
os_error_filename_set.add(name)
self.folder_log.error('{0} [os error] [path={1}] [error={2}]'.format(
self.log_base, path, traceback.format_exc()))
except Exception as e:
try:
self.folder_log.error('{0} [file error] [path={1}] [error={2}]'.format(self.log_base, path,
traceback.format_exc()))
failed_path = os.path.join(failed_output_dir, '{0}_{1}'.format(time.time(), name))
shutil.move(path, failed_path)
except Exception as e:
os_error_filename_set.add(name)
self.folder_log.error('{0} [file move error] [path={1}] [error={2}]'.format(
self.log_base, path, traceback.format_exc()))
else:
if isinstance(result, dict) and len(result) > 0:
date_str = time.strftime("%Y-%m-%d")
result_queue.put(
{
self.CLASSIFY_KEY: classify,
self.RESULT_KEY: result,
self.DATE_KEY: date_str
}
)
elif isinstance(result, list) and len(result) > 0:
date_str = time.strftime("%Y-%m-%d")
for res in result:
result_queue.put(
{
self.CLASSIFY_KEY: classify,
self.RESULT_KEY: res,
self.DATE_KEY: date_str
}
)
def handle(self, *args, **kwargs):
if len(self.input_dirs) == 0:
return
result_queue = Queue()
process_list = []
one_input_dir = None
for classify_idx, input_dir in self.input_dirs.items():
if one_input_dir is None:
one_input_dir = input_dir
classify = int(classify_idx.split('_')[0])
is_combined = True if int(classify_idx.split('_')[2]) == 1 else False
process = Process(target=self.folder_process, args=(input_dir, classify, is_combined, result_queue))
process_list.append(process)
wb_dir = os.path.dirname(os.path.dirname(one_input_dir))
wb_process = Process(target=self.wb_process, args=(wb_dir, result_queue, ))
process_list.append(wb_process)
for p in process_list:
p.start()
for p in process_list:
p.join()
self.folder_log.info('{0} [stop safely]'.format(self.log_base))