def extract_name(province, html, text, lines, table_line, table): res = get_data_from_table(table, '姓名') if not res: res = get_data_from_table(table, '项目负责人') if not res: res = bt_get_value_by_next_td(html, '项目经理') return res
def bt_cgw_cggg(province, html, text, lines, table_line, table): res = {} cgnr = get_data_from_table(table, '名称') if cgnr: res['cgnr'] = ','.join(cgnr) res['zgyq'] = get_chapter_content(lines, '资格条件') return res
def extract_tbbj(province, html, text, lines, table_line, table): res = get_data_from_table(table, '报价') if not res: try: res = re.findall('中标候选人.*?投标报价:(.*?)(\s|()', text, re.S)[0][0] except: pass return res
def extract_htjgw(province, html, text, lines, table_line, table): res = 0 try: res = get_data_from_table(table, '万元') if not res: res = 0 except: pass return res
def extract_htjg(province, html, text, lines, table_line, table): res = None try: res = get_data_from_table(table, '元') except: pass if not res: res = get_value_by_split_maohao('中标标价', text) return res
def sd_jyw_gcjg(province, html, text, lines, table_line, table): res = {} # res['xmmc'] = get_value_by_table_line(table_line, '项目名称') res['xmbh'] = get_value_by_table_line(table_line, '项目编号') res['kbsj'] = get_value_by_table_line(table_line, '开标时间') res['zbdl'] = get_value_by_table_line(table_line, '代理机构') res['pscy'] = get_value_by_table_line(table_line, '评标委员会') res['company'] = get_data_from_table(table, '中标单位') return res
def bt_jyw_zbjggg(province, html, text, lines, table_line, table): res = {} res['xmbh'] = bt_get_value_by_next_td(html, '项目编号') res['type'] = bt_get_value_by_next_td(html, '项目类别') res['zbr'] = bt_get_value_by_next_td(html, '招标人') # res['xmmc'] = bt_get_value_by_next_td(html, '项目名称') company = get_data_from_table(table, '中标单位') if len(company) == 1: company = company[0] res['company'] = company res['gsmc'] = company # jhgq = get_data_from_table(table, '工期') # res['jhgq'] = jhgq if jhgq else None res['name'] = get_data_from_table(table, '项目经理') gsq_start = get_value_by_next_td(html, '公示开始时间') gsq_end = get_value_by_next_td(html, '公示结束时间') res['gsq'] = gsq_start + '-' + gsq_end if gsq_start else None return res
def extract_xmbh(province, html, text, lines, table_line, table): if province == 'bt': try: res = get_data_from_table(table, '标段编号')[0] return res except: return '' pattern = r'项目.*?编号:?([0-9a-zA-Z_-]+)' res = re.findall(pattern, text) if res: return res[0] else: return ''
def ln_jyw_zbhxr(province, html, text, lines, table_line, table): res = {} # res['xmmc'] = get_value_by_next_td(html, '工程名称') res['type'] = get_value_by_next_td(html, '工程类别') res['xmbh'] = get_value_by_next_td(html, '编号') res['zbr'] = get_value_by_next_td(html, '建设单位') res['gsmc'] = get_data_from_table(table, '单位名称') # jhgq = get_data_from_table(table, '工期') # if not jhgq: # jhgq = None # res['jhgq'] = jhgq res['tbbj'] = get_data_from_table(table, '报价') res['name'] = get_data_from_table(table, '项目负责人') text_list = ['注册资格', '证书'] res['zsmc'] = get_data_from_table2(table, text_list) res['zsbh'] = get_data_from_table(table, '证书编号') gsq_start = get_value_by_next_td(html, '公示开始时间') gsq_end = get_value_by_next_td(html, '公示截止时间') if not gsq_end: gsq_end = get_value_by_next_td(html, '公示结束时间') res['gsq'] = gsq_start + '-' + gsq_end return res
def extract_jhgq(province, html, text, lines, table_line, table): result = [] get_value_by_split_maohao(lines, '工期',result) result3 = get_data_from_table(table, '工期') if result3: result.append(result3[0]) res = get_finally_result(result) if res: try: res = int(re.findall('\d+', res)[0]) except: pass return res
def extract_xmbh(province, html, text, lines, table_line, table): result = [] if province == 'bt': try: res = get_data_from_table(table, '标段编号')[0] if res:result.append(res) except: pass pattern = r'(标段|项目|招标).*?编号:?([0-9a-zA-Z_-]+)' res = re.findall(pattern, text) if res: result.append(res[0][1]) get_value_by_next_td(html, '工程编号',result) bt_get_value_by_next_td(html, '项目编号',result) get_value_by_split_maohao(lines,'项目编号',result) res = get_finally_result(result) return res
def bt_jyw_zbhxr(province, html, text, lines, table_line, table): res = {} # res['xmmc'] = bt_get_value_by_next_td(html, '工程名称') res['zbr'] = bt_get_value_by_next_td(html, '建设单位') # jhgq = bt_get_value_by_next_td(html, '工期') # res['jhgq'] = jhgq if jhgq else None company = get_data_from_table(table, '中标单位') if len(company) == 1: company = company[0] res['company'] = company res['gsmc'] = company res['tbbj'] = bt_get_value_by_next_td(html, '小写') res['name'] = bt_get_value_by_next_td(html, '建造师姓名') res['jb'] = bt_get_value_by_next_td(html, '注册级别') res['zsbh'] = bt_get_value_by_next_td(html, '证书编号') return res
def extract_zczy(province, html, text, lines, table_line, table): res = get_data_from_table(table, '专业') return res
def extract_gsmc(province, html, text, lines, table_line, table): # 关联公司名称 res = get_data_from_table(table, '中标单位') return res
def extract_zhpf(province, html, text, lines, table_line, table): # 得分 res = get_data_from_table(table, '得分') return res
def extract_tbbjw(province, html, text, lines, table_line, table): if '万元' in text: res = get_data_from_table(table, '报价') return res else: return 0