PyQueryでYahoo天気をスクレイピングする

yahootenki.py

# -*- coding: utf-8 -*-

import re
import pyquery
import sys

reload(sys)
sys.setdefaultencoding('utf-8')

def place2weather(place, day="today") :
  """
    場所を受け取って,天気を返す関数.
    引数:場所の文字列,郵便番号が良い.地名だと失敗すること多し.
    返り値:実際の返してくる値を見てください.
  """
  
  urls = []
  d = pyquery.PyQuery("http://search.weather.yahoo.co.jp/bin/search?p=%s&s.x=0&s.y=0"%place)
  for a in d(".serch-table table tr td a") :
    try :
      if re.match("http://weather.yahoo.co.jp/weather/(.*)\.html$", a.attrib["href"]) :
        urls.append(a.attrib["href"])
    except :
      pass
  
  weather = {}
  d = pyquery.PyQuery(urls[0])
  for tr in d("#yjw_pinpoint_today table tr")[1:] :
    kind = re.sub(r"\s", "", d(tr)("td")[0].text_content())
    weather[kind] = {}
  
    for k,v in zip(d(d("#yjw_pinpoint_%s table tr"%day)[0])("td")[1:],d(tr)("td")[1:]) :
      m = re.match(u"^([0-9]*)時$", re.sub(r"\s", "", k.text_content()))
      time = u"%02d時"%int(m.group(1))
      value = re.sub(r"\s", "", v.text_content())
      weather[kind][time] = value
  
  return weather 

if __name__ == "__main__" :
  post_code = "2900062" #千葉県市原市八幡の郵便番号
  print "---今日の天気---"
  tenki = place2weather(post_code, "today")
  for v  in tenki.keys() :
    print v
    for v2 in sorted(tenki[v].keys()) :
      print "  %s:%6s"%(v2, tenki[v][v2])
  
  print "---明日の天気---"
  tenki = place2weather(post_code, "tomorrow")
  for v  in tenki.keys() :
    print v
    for v2 in sorted(tenki[v].keys()) :
      print "  %s:%6s"%(v2, tenki[v][v2])