|
|
@@ -230,11 +230,11 @@ def njpc_fields_extract_special(html, data_item):
|
|
|
html = "".join(soup.get_text().split()).strip()
|
|
|
# 抽取字段
|
|
|
data_item.title = data_item.projectname
|
|
|
- projectname = re.findall('项目名称(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
|
|
+ projectname = re.findall('项目名称(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]+[，,.。；;、：:]', html, re.S)
|
|
|
approvecode = re.findall('项目(?:代码|编码)(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
|
|
- approvecontent = re.findall('(?:事项名称|审批事项)(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
|
|
+ approvecontent = re.findall('(?:事项名称|审批事项)(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]+[，,.。；;、：:]', html, re.S)
|
|
|
owner = re.findall('[建设|项目](?:单位|单位名称|业主)(?:[\u4e00-\u9fa5]+|)[:|：](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
|
|
- projectaddr = re.findall('建设(?:地点|地址)(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
|
|
+ projectaddr = re.findall('建设(?:地点|地址)(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]+[，,.。；;、：:]', html, re.S)
|
|
|
total_investment = re.findall('总投资(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[。；;、：:]', html, re.S)
|
|
|
project_person = re.findall('联系人(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
|
|
project_phone = re.findall('联系(?:电话|方式)(?:[\u4e00-\u9fa5]+|)[：|:](.*?)[一二三四五六七八九十\d]?[，,.。；;、：:]', html, re.S)
|
