备份饭否消息。 非API

fetch.py

# -*- coding: utf-8 -*- 
# AUTHOR: Zeray Rice <[email protected]>
# FILE: main.py
# CREATED: 01:15:37 06/04/2012
# MODIFIED: 01:57:15 06/04/2012

import requests
import datetime
from jinja2 import Environment, FileSystemLoader
from BeautifulSoup import BeautifulSoup

cookies = dict(ai='', u='', SID='', uuid='')  # 在这里填写 Cookies 信息
jinja = Environment(loader = FileSystemLoader('.'))
tpl = jinja.get_template("message.xml")

class Status(object):
    def __init__(self, status):
        # parse status HTML
        self.text = status.findAll("span", {'class' : "content"})[0].text
        self.time = status.findAll("a", {'class' : "time"})[0].attrMap['stime']
        self.time = datetime.datetime.strptime(self.time, "%a %b %d %H:%M:%S +0000 %Y")

def parseHTML(html, result):
    bs = BeautifulSoup(html)
    stream = bs.findAll("div", id = "stream")[0].findAll("li")
    for status in stream:
        st = Status(status)
        result.append(st)

def renderMSG(result, filenameCount):
    print "Saving to %d.xml" % filenameCount
    with open("treeholes/%d.xml" % filenameCount, "w") as fp:
        fp.write(tpl.render(status = result).encode("utf-8"))

def getStatus():
    filenameCount = 1
    result = []
    for p in range(1, 3512):
        r = requests.get("http://fanfou.com/treeholes/p.%d" % p, cookies=cookies)
        parseHTML(r.text, result)
        print "Parsing Page %d.." % p
        if len(result) >= 100:
            renderMSG(result, filenameCount)
            filenameCount = filenameCount + 1
            result = []
    renderMSG(result, filenameCount)

getStatus()

message.xml

<?xml version="1.0" encoding="UTF-8"?>
<timelime id="treeholes">{% for st in status %}
    <status>
        <text><![CDATA[{{ st.text }}]]></text>
        <time><![CDATA[{{ st.time }}]]></time>
    </status>
{% endfor %}</timelime>