pdf_title_image/extract_price.py

# -*- coding: utf-8 -*-
# @Author: privacy
# @Date:   2024-06-11 13:43:14
# @Last Modified by:   privacy
# @Last Modified time: 2024-09-05 10:37:12
from pprint import pprint

from celery_tasks.text_extractor import get_instance
from celery_tasks.tools import match_price_zhs, match_price_num, match_duration, match_quality


if __name__ == '__main__':
    import json
    pdf_path = r'./data/projects/三峡左岸及地下电站地坪整治/投标/湖北建新建设工程有限公司_T221100130348%2F01整本文件/投标文件-修改版9-5-1-1.pdf'

    with open(pdf_path[:-4] + '-title.json', 'r', encoding='utf-8') as fp:
        titles = json.load(fp)
    with open(pdf_path[:-4] + '-text.json', 'r', encoding='utf-8') as fp:
        texts = json.load(fp)

    price_zhs = get_instance(
        title_instances=['投标函', '开标一览表'],
        content_instances=['人民币投标总报价'],
        titles_list=titles,
        texts_list=texts,
        extractor=match_price_zhs
    )
    price_num = get_instance(
        title_instances=['投标函', '开标一览表'],
        content_instances=['人民币投标总报价'],
        titles_list=titles,
        texts_list=texts,
        extractor=match_price_num
    )
    duration = get_instance(
        title_instances=['投标函', '开标一览表'],
        content_instances=['工期日历天'],
        titles_list=titles,
        texts_list=texts,
        extractor=match_duration
    )
    quality = get_instance(
        title_instances=['投标函', '开标一览表'],
        content_instances=['工程质量'],
        titles_list=titles,
        texts_list=texts,
        extractor=match_quality
    )
    # valid = rmb_to_digit(price_zhs[0][0][0]) == price_num[0][0][0][1:]

    pprint({
        "price_zhs": price_zhs,
        "price_num": price_num,
        "duration": duration,
        "quality": quality,
        # "valid": valid
    })