fileUtil=FileUtil(); count=0; for url,name in answerPageUrls: answerPage=doPost(urlRoot+url); fileUrl=getAnswerFileUrl(answerPage) # print fileUrl; saveName=name+"."+fileUrl.split('.')[1]; # saveName=str(count)+"."+fileUrl.split('.')[1]; count+=1; saveName=unicode(saveName,"utf-8"); fileUrl=urlHomeRoot+fileUrl; # print fileUrl; fileContent=HtmlGraber().doGrab(fileUrl); filePath=savePath+saveName; fileUtil.binaryWrite(filePath, fileContent); print saveName+" saved\n";
maxCount=50; firstIndex=249317678-50; fileUtil=FileUtil(); homeSavePath="E:\\temp\\" def grabImageUrl(picUrl): patt=re.compile(r'http://fmn.rrimg.com/.*'); htmlContent=htmlGraber.doGrab(picUrl); # print htmlContent; soup=BeautifulSoup(htmlContent); imgurls=soup.findAll('img',id="photo");#re.compile(patt)); #print str(imgurl[0].src); # print "\n".join([str(item) for item in imgurls]); if(len(imgurls)>0): return imgurls[0]["src"]; else: return "" fileUtil.makeDirWhenNeeded("E:\\temp\\", "renren"); for index in range(maxCount): picId=firstIndex-index; picUrl = PicUrlHead + str(picId); print picUrl+"\n"; imgurl = grabImageUrl(picUrl); if(imgurl != ""): imgContent=htmlGraber.doGrab(imgurl); fileUtil.binaryWrite("E:\\temp\\renren\\"+str(index)+".jpg", imgContent);