抓些playstore数据

#for win
#python + curl
import urllib2
import os
local = "D:\\apks\\"
logoWidth = 124 #MAX 512
imgNum = 2 #Based on app
url = raw_input("play Uri\n")

def getPkgName(url):
    pkgNameB = url.index("=")+1
    if '&' in url:
        pkgNameE = url.index('&')
    else:
        pkgNameE = len(url)
    return url[pkgNameB:pkgNameE]
def getUrl(pkg):
    return "http://play.google.com/store/apps/details?id=" + pkg + "&hl=en"
def getContent(url):
    return urllib2.urlopen(url).read()
def getVer(content):
    verB = content.index("softwareVersion")+17
    verE = content.find("<",verB)
    return content[verB:verE]
def getLogo(content):
    logoB = content.index("doc-banner-icon")+27
    logoE = content.find("=",logoB)
    return (content[logoB:logoE] + "=w" + repr(logoWidth)).replace("https","http")
def getSys(content):
    sysB = content.find("<dd>",content.index("operatingSystems"))+4
    sysE = content.find("</dd>",sysB)
    return content[sysB:sysE]
def getCat(content):
    catB = content.find("<dd>",content.index("Category:"))+4
    catB = content.find(">",catB)+1
    catE = content.find("</a>",catB)
    return content[catB:catE].replace("&amp;","&")
def getPrice(content):
    priceB = content.index("buy-button-price")+16
    priceB = content.find(">",priceB)+1
    priceE = content.find("span>",priceB)-2
    return content[priceB:priceE].replace(" Buy","").replace("Install","$0")
def getImg(content):
    sliceB = content.index("screenshot-carousel-content-container")+39
    sliceE = content.find("screenshot-carousel-right-fade",sliceB)-78
    imgContainer = content[sliceB:sliceE]
    imgContainer = imgContainer.split("=h")
    result = []
    i = 0
    while( i < len(imgContainer)):
        if "data-baseUrl=" in imgContainer[i]:
            posB = imgContainer[i].index("data-baseUrl=")+14
            posE = imgContainer[i].index("title")-2
            result.append((imgContainer[i][posB:posE]).replace("https","http"))
        i = i+1
    return result
def getTitle(content):
    titleB = content.index("<h1 class=\"doc-banner-title\">")+29
    titleE = content.find("</h1>",titleB)
    return content[titleB:titleE]

pkgName = getPkgName(url)
content = getContent(getUrl(pkgName))

title = getTitle(content)
apkdir = local + title
os.system("mkdir " +"\""+apkdir +"\"")

logo = getLogo(content)
os.system(local+"curl.exe -o \"" + apkdir +"\\logo.png\" "+ logo)

img = getImg(content)
i = 0
while( i < len(img)):
    os.system(local+"curl.exe -o \"" + apkdir +"\\pic"+ repr(i) + ".png\" "+ img[i])
    i = i+1
#os.system('cls')

cat = getCat(content)
sys = getSys(content)
ver = getVer(content)
price = getPrice(content)
print title
print cat
print sys
print pkgName
print ver
print price