|
@@ -31,4 +31,7 @@ def extract_domain(url):
|
|
|
|
|
|
|
|
def extract_page_title(html):
|
|
def extract_page_title(html):
|
|
|
element = html2element(html)
|
|
element = html2element(html)
|
|
|
- return "".join(element.xpath('/html/head/title/text()')).strip()
|
|
|
|
|
|
|
+ nodes = element.xpath('/html/head/title/text()')
|
|
|
|
|
+ if len(nodes) > 1:
|
|
|
|
|
+ return "".format(nodes[-1]).strip()
|
|
|
|
|
+ return "".join(nodes).strip()
|
