WBS「トレたま」で放送された商品をAmazonにあるかをpythonで調べる

WBS「トレたま」でされた商品が気になって調べることがあります。

というわけで過去のものも一気に調べてみようかなと思い立ちましたのでpythonスクリプトを作ってみました。

適当なディレクトリに以下の
toretama.py
amazon_helper.py
を配置して

python toretama.py 201104
みたいな感じで実行します。
実際は関係ない商品も出てくるので最終的には目視確認ですけどね

それぞれのファイルは以下の通りです。

toretama.py

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

import os
import sys
import urllib2
import re
import xml.etree.ElementTree as etree

from amazon_helper import AmazonHelper

print sys.stdout.encoding, sys.getdefaultencoding() 

yyyy = sys.argv[1][0:4]
mm = sys.argv[1][4:6]

#指定月に放送されたリストを取得
url = "http://www.tv-tokyo.co.jp/wbs/trend_tamago/%s/%s/writeThisMonth.js"  % (yyyy,mm)
req_l = urllib2.Request(url)
cont_l = urllib2.urlopen(req_l).read()

pat_l = "<a href=\"(.*)\">(.*)</a> <span class=\"date\">\((.*)\)</span>"
for m in re.finditer(pat_l, cont_l, re.MULTILINE):
    #個別ページから商品名を取得
    req  = urllib2.Request(m.group(1))
    cont = urllib2.urlopen(req).read()
    cont = cont.replace("\n","")
    pat  = "<p><strong>商品名</strong></p><p>(.*)</p><p><strong>商品の特徴</strong></p>"
    mlist = re.search(pat,cont) 
    if mlist == None :
        print "not find"
        continue
    
    #amazonのAPIで商品検索
    ackey = "amazonのアクセスキー"
    sckey = "シークレットキー"
    askey = "アソシエートタグ"
    options = {}
    options['SearchIndex'] = "Blended"
    options['ResponseGroup'] = "Small,Images"
    options['Keywords'] = mlist.group(1)
    ah = AmazonHelper(ackey,sckey,askey)
    url_a = ah.make_item_search(options)
    req_a = urllib2.Request(url_a)
    cont_a = urllib2.urlopen(req_a).read()
    dom = etree.fromstring(cont_a)
    namespace ='http://webservices.amazon.com/AWSECommerceService/2010-06-01'
    items = dom.find('.//{%s}Items' % namespace)
    valid = items.find('.//{%s}IsValid' % namespace)
    if valid.text != "True" :
        print "not find"
        continue
    if items.find('.//{%s}Errors' % namespace) :
        print "not find"
        continue
    for item in items.findall('.//{%s}Item' % namespace):
        print item.find('.//{%s}Title' % namespace).text
        print item.find('.//{%s}DetailPageURL' % namespace).text

amazon_helper.py

import urllib,hmac, hashlib, base64
from datetime import datetime

class AmazonHelper:
    def __init__(self, access_key_id, secret_key, associate_tag=None, version='2010-06-01'):
        self.service = 'AWSECommerceService'
        self.access_key_id = access_key_id
        self.secret_key = secret_key
        self.associate_tag = associate_tag
        self.version = version
        self.uri = 'ecs.amazonaws.jp'
        self.end_point = '/onca/xml'
        self.dummy_timestamp = None


    def set_dummy_timestamp(self,dt):
        self.dummy_timestamp = dt
        
    def make_item_lookup(self, options):
        options['Operation'] = 'ItemLookup'
        return self.make_request(options)

    def make_item_search(self, options):
        options['Operation'] = 'ItemSearch'
        return self.make_request(options)

    def make_request(self, options):
        options['Service'] = self.service
        options['AWSAccessKeyId'] = self.access_key_id
        options['Version'] = self.version
        
        if self.dummy_timestamp:
            options['Timestamp'] = self.dummy_timestamp
        else:
            options['Timestamp'] = datetime.utcnow().isoformat()

        if self.associate_tag:
            options['AssociateTag'] = self.associate_tag

        payload = ""
        for v in sorted(options.items()):
            payload += '&%s=%s' % (v[0], urllib.quote(str(v[1])))
        payload = payload[1:]

        strings = ['GET', self.uri, self.end_point, payload]

        digest = hmac.new(self.secret_key, '\n'.join(strings), hashlib.sha256).digest()
        signature = base64.b64encode(digest)

        url = "http://%s%s?%s&Signature=%s" % (self.uri, self.end_point, payload, urllib.quote_plus(signature))
        return url