利用者:素手/Bot/ボクサー一覧.py

#coding:utf-8
#!/usr/bin/python -S
import sys
from urllib import urlopen
import linecache
import wikipedia
import sys
import codecs
import re
import linecache
import MySQLdb

yahoo = open("yahoo.txt","w")
#yahoo = codecs.lookup('utf_8')[-1](yahoo)

site = wikipedia.getSite("ja")
pagename = u"男子ボクサー一覧"
page = wikipedia.Page(site,pagename)
wikipedia.output(u"Loading %s..." % pagename)

for i in page.linkedPages():
	try:
		print i
		yahoo.write(i.title())
		if i.isRedirectPage() == True:
			text = i.getRedirectTarget().get()
		else:
			text = i.get(force = False, get_redirect=False, throttle = True, sysop = False, change_edit_time = True)
		for t in text.split("\n"):
			if t.find("{{DEFAULTSORT:") != -1:
				print t
				yahoo.write("\t"+t.replace("{{DEFAULTSORT:","").replace("}}","")+"\n")
				break
		else:
			print u"defaultなし"
			yahoo.write(u"defaultがありません\n")
	except wikipedia.IsRedirectPage():
		print "\a"
		yahoo.write("redirect")
print "\a"