return (urls,body,title,desc,keywords,error_code,error_reason,content_type,last_modified,err) err=0 if output > 0: print info if info.has_key("content-type"): content_type = str(info["content-type"]) if info.has_key("last-modified"): last_modified = str(info["last-modified"]) soup=BeautifulSoup(data) try: title=cleanHTML(soup.html.head.title.string) title=convertAccents(title) except: title = '' try: for meta in soup.head('meta'): ctxt = str(meta) pat = re.compile(r"meta[ ]*name[ ]*=[ ]*[\"]*key").findall(ctxt.lower()) if pat: temp=re.compile(r"ontent[ ]*=[ ]*[\"]*").split(ctxt) if len(temp) > 1: keywords=temp[1] keywords=re.compile(r"[ ]*[\"]*[ ]*[/]*[>]").sub(' ',keywords) keywords=cleanHTML(keywords) keywords=convertAccents(keywords) keywords=keywords.strip() pat = re.compile(r"meta[ ]*name[ ]*=[ ]*[\"]*descrip").findall(ctxt.lower()) if pat: temp=re.compile(r"ontent[ ]*=[ ]*[\"]*").split(ctxt) if len(temp) > 1: desc=temp[1]