30 lines
736 B
Python
30 lines
736 B
Python
from paddlenlp.utils.doc_parser import DocParser
|
|
|
|
"""
|
|
项目配置
|
|
"""
|
|
# 每次从数据库获取的案子数量
|
|
PHHD_BATCH_SIZE = 20
|
|
# 没有查询到案子的等待时间(分钟)
|
|
SLEEP_MINUTES = 5
|
|
# 是否发送异常提醒邮件
|
|
SEND_ERROR_EMAIL = True
|
|
# 备份原图的尝试次数
|
|
COPY_TRY_TIMES = 3
|
|
# 上传新图的尝试次数
|
|
UPLOAD_TRY_TIMES = 3
|
|
|
|
"""
|
|
关键词配置
|
|
"""
|
|
NAME_KEYS = [
|
|
{"key": "姓名", "length": 4},
|
|
{"key": "交款人", "length": 4},
|
|
{"key": "文款人", "length": 4},
|
|
{"key": "购买方名称", "length": 4},
|
|
]
|
|
ID_CARD_NUM_KEYS = [{"key": "身份证号", "length": 19}, ]
|
|
|
|
# OCR = PaddleOCR(use_angle_cls=False, show_log=False, gpu_id=1)
|
|
OCR = DocParser(use_gpu=True, layout_analysis=False)
|