天府市民云APP的消息RSS生成

tianfu.py

import datetime
import time

import PyRSS2Gen
import requests, pytz

class NoOutput:
    def __init__(self):
        pass
    def publish(self, handler):
        pass

class IPhoneRSS2(PyRSS2Gen.RSSItem):
    """给标签内容添加CDATA标记
    """
    def __init__(self, **kwargs):
        PyRSS2Gen.RSSItem.__init__(self, **kwargs)

        # 需要处理的标签
        # Tags that need to be processed 
        self._handle_tag = ['description','title']

    def publish(self, handler):
        for i in self._handle_tag:
            if getattr(self,i):
                setattr(self,'_%s'%i,getattr(self,i)) # self['_%s'%i] = self[i]
                setattr(self,i,NoOutput()) # This disables the Py2GenRSS "Automatic" output of the description, which would be escaped. 

        PyRSS2Gen.RSSItem.publish(self, handler)

    def publish_extensions(self, handler):
        from xml.sax import saxutils
        for i in self._handle_tag:
            if getattr(self,i):
                handler.startElement(i, self.element_attrs)
                # handler._write('<![CDATA[%s]]>' % saxutils.escape(getattr(self,'_%s' % i))) #写入源数据 手动编码处理
                handler._write('<![CDATA[%s]]>' % getattr(self,'_%s' % i)) #写入源数据 避免PyRSS2Gen中handler.characters(obj)的编码处理
                handler.endElement(i)


url = 'https://222.212.139.221/appnews/api/headline?page=1&size=10'
headers = {
    'Host':'tfsmy.chengdu.gov.cn',
    'User-Agent': 'Mozilla/5.0 (iPhone; CPU iPhone OS 12_2 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E148 iOS/12.2 Brand/iPhone Display/320.000000*568.000000 tfsmy/1.6.2'
}
response = requests.get(url, headers=headers, verify =False)
content = response.json()
tz = pytz.utc
if content['code'] == 1:
    items = []
    for item in content['result']['list']:
        url = "http://tfsmy.chengdu.gov.cn/cms/share/news?id=%s" % item['id']

        time_tuple = time.strptime(item['updateTime'], "%Y-%m-%d %H:%M:%S")
        pubData = datetime.datetime.fromtimestamp(int(time.mktime(time_tuple)), tz=pytz.utc)#转换为datetime.datetime对象 默认string会输出GMT时间字符串
        image = item['image']  if 'image' in item else '' # 存在没有图片的数据
        items.append(IPhoneRSS2(
            title=item['title'],
            link=url, # 打开预览的网址
            description='%s'%(item['title']),
            guid=PyRSS2Gen.Guid(image, isPermaLink=1), # 设置为封面图
            pubDate=pubData
        ))

    rss = PyRSS2Gen.RSS2(
        title="天府市民云APP feed",
        link="https://www.tfsmy.com/download/tianfu.html",
        description="用于获取app内「成都头条」的RSS feeds",
        lastBuildDate=datetime.datetime.now(tz=tz),
        items=items,
    )
    # rss.write_xml(open("./tianfu.xml", "w", encoding='utf-8'), encoding='utf-8')
    rss.write_xml(open("/home/wwwroot/rss.hhtjim.com/tianfu.xml", "w", encoding='utf-8'), encoding='utf-8')
    

修改PyRSS2Gen不能处理CDATA的问题