扇贝打卡 RSS 源

shanbay2rss.py

#!/usr/bin/env python
# encoding: utf-8
from __future__ import absolute_import, unicode_literals

import tornado.ioloop
import tornado.httpserver
from tornado.options import parse_command_line, define, options
import tornado.httpclient
import tornado.web
import tornado.gen
import logging
from bs4 import BeautifulSoup
import PyRSS2Gen
import datetime
import locale


define('debug', type=bool, default=True)
define('port', type=int, default=8088)
define('host', type=str, default="127.0.0.1")

class ShanbayRss(object):
    URL = "http://shanbay.com/checkin/record/{record_id}/"
    DEFAULT_HEADERS = {
        "User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_10_5) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/45.0.2454.99 Safari/537.36",
    }

    def __init__(self, record_id):
        self.record_id = record_id
        self.http_client = tornado.httpclient.AsyncHTTPClient()

    @tornado.gen.coroutine
    def get_latest_chekcins(self):
        req = tornado.httpclient.HTTPRequest(
                url=self.URL.format(record_id=self.record_id),
                headers=self.DEFAULT_HEADERS,
        )
        resp = yield self.http_client.fetch(req)
        checkins = self.parse_checkins(resp.body)
        raise tornado.gen.Return(checkins)

    @tornado.gen.coroutine
    def to_xml(self):
        checkins = yield self.get_latest_chekcins()
        items = list()
        items = [PyRSS2Gen.RSSItem(
            title="{date} 第{number}天打卡".format(
                date=item["date"],
                number=item["day"],
                ),
            link=item["link"],
            description=item["note"],
            guid=PyRSS2Gen.Guid(item["link"]),
        ) for item in checkins["checkins"]]
        rss = PyRSS2Gen.RSS2(
            title=checkins["title"],
            link="http://hui.lu",
            description="",
            lastBuildDate=datetime.datetime.now(),
            items=items
        )

        xml = rss.to_xml(encoding="utf-8")
        raise tornado.gen.Return(xml)


    def parse_checkins(self, content):

        soup = BeautifulSoup(content, 'html.parser')
        checkin_content = soup.find(attrs={"class":"checkins"}).\
                find_all(attrs={"class": "checkin"})

        checkins = list()

        for i in checkin_content:
            note = i.find(attrs={"class": "note"}).text.strip()
            target = i.find(attrs={"class": "target"}).text.strip()
            locale.setlocale(locale.LC_ALL, str('zh_CN.UTF-8'))
            date = datetime.datetime.strptime(target.encode('utf-8'), '%B %d, %Y')
            checkins.append({
                "day": i.find(attrs={"class": "number"}).text.strip(),
                "note": "".join(note.split()),
                "date": date.strftime('%Y年%m月%d天'.encode('utf-8')).decode('utf-8'),
                "link": "http://shanbay.com" + i.find("a",
                    attrs={"class": "target"})["href"]
            })

        title = soup.title.text.strip()

        return {
            "title": title,
            "checkins": checkins,
        }


class MainHandler(tornado.web.RequestHandler):

    @tornado.gen.coroutine
    def get(self):
        self.write("hi")


class ShanbayRssHandler(tornado.web.RequestHandler):
    @tornado.gen.coroutine
    def get(self, record_id):
        """
        由于个人打卡页面必须登录
        所以曲线救国，通过单条打卡记录，可以查看最新的打卡记录
        """
        try:
            shanby_rss = ShanbayRss(record_id)
            res = yield shanby_rss.to_xml()
            self.set_header("Content-Type", "text/xml;charset=UTF-8")
            self.write(res)
        except Exception as e:
            self.application.logger.error(e)
            self.set_status(404)
            self.write('404')


class Application(tornado.web.Application):

    def __init__(self):
        handlers = [
            (r'/shanbay2rss/(\d+)', ShanbayRssHandler),
            (r'/', MainHandler),
        ]

        settings = dict()
        settings["debug"] = options.debug
        self.logger = logging.getLogger('application')

        super(Application, self).__init__(handlers, **settings)


def main():
    http_server = tornado.httpserver.HTTPServer(Application())
    parse_command_line()
    http_server.listen(options.port, address=options.host)
    logger = logging.getLogger('application')
    logger.info("Server listen on http://{host}:{port}".format(host=options.host,
        port=options.port))

    tornado.ioloop.IOLoop.current().start()


if __name__ == '__main__':
    main()