def parse(headers): name = get_info_by_url('http://www.sahibinden.com/xml_username.php', headers, [("index.php?a=1004'>", "<")])[0] if name is None: return False insert_to_db("Sahibinden", headers, name, "http://www.sahibinden.com/") return True
def parse(headers): name = get_info_by_url("http://" + headers["Host"] + "/", headers, [('<i class="b-user__provider-ico"></i>', "<")])[0] if name is None: return False insert_to_db("Yandex", headers, name, "http://" + headers["Host"] + "/") return True
def parse(headers): name = get_info_by_url("http://www.bing.com/", headers, [('<span id="id_n">', '<')])[0] if name is None: return False insert_to_db("Bing", headers, name, "http://www.bing.com") return True
def parse(headers): name = get_info_by_url('http://my.ebay.com/ws/eBayISAPI.dll?MyeBay', headers, [('="mbg-nw">', '<')])[0] if name is None: return False insert_to_db("Ebay", headers, name, "http://www.ebay.com") return True
def parse(headers): name = get_info_by_url( "http://www.dailymail.co.uk/registration/profile.html", headers, [('<strong class="homeblue">', '<')])[0] if name is None: return False insert_to_db("Dailymail", headers, name, "http://www.dailymail.co.uk/home/index.html") return True
def parse(headers): (name, img) = get_info_by_url("http://www.linkedin.com/", headers, [('alt="', '"'), (' <img class="member-photo" src="', '"')]) if name is None or img is None: return False img_path = "temp/LI_" + name.replace(" ", "_") + ".jpg" open(img_path, 'wb').write(urllib.urlopen(img).read()) insert_to_db("LinkedIn", headers, name, "http://www.linkedin.com/", img_path) return True
def parse(headers): (name, img) = get_info_by_url('http://persianblog.ir/EditAvatar.aspx', headers, [('aspx" style="font-weight:bold;">', '<'), ('id="imgAvatarForChange" src="', '"')]) if name is None: return False img_path = "temp/PB_" + name.replace(" ", "_") + ".jpg" open(img_path, 'wb').write(urllib.urlopen(img).read()) insert_to_db("Persianblog", headers, name, "http://persianblog.ir/", img_path) return True
def parse(headers): (name, img) = get_info_by_url('http://www.tumblr.com/dashboard', headers, [('data-blog-url="http://', '.tumblr.com/"'), ('data-user-avatar-url="', '"')]) if name is None: return False print(name, img) img_path = "temp/TR_" + name.replace(" ", "_") + ".jpg" open(img_path, 'wb').write(urllib.urlopen(img).read()) insert_to_db("Tumblr", headers, name, "http://www.tumblr.com/dashboard", img_path) return True