PsychoticNinja/JLU5-SupyPlugins
1
0
Fork 0
mirror of https://github.com/jlu5/SupyPlugins.git synced 2025-04-26 21:11:14 -05:00
Code Issues Projects Releases Wiki Activity

Wikifetch: prefer <link rel="canonical"> links again when available

Browse Source
This commit is contained in:
James Lu 2017-03-24 19:08:25 -07:00
parent b9481184cb
commit 001b49b6c3
1 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
Wikifetch/plugin.py
View File

@ -171,11 +171,16 @@ class Wikifetch(callbacks.Plugin):
# extract the address we got it from - most sites have the perm link # extract the address we got it from - most sites have the perm link
# inside the page itself # inside the page itself
try: try:
addr = tree.find(".//div[@class='printfooter']/a").attrib['href'] addr = tree.find(".//link[@rel='canonical']").attrib['href']
addr = re.sub('([&?]|(amp;)?)oldid=\d+$', '', addr) except (ValueError, AttributeError):
except: self.log.debug('Wikifetch: failed <link rel="canonical"> link extraction, skipping')
# If any of the above post-processing tricks fail, just ignore try:
pass addr = tree.find(".//div[@class='printfooter']/a").attrib['href']
addr = re.sub('([&?]|(amp;)?)oldid=\d+$', '', addr)
except (ValueError, AttributeError):
self.log.debug('Wikifetch: failed printfooter link extraction, skipping')
# If any of the above post-processing tricks fail, just ignore
pass
# check if it's a disambiguation page # check if it's a disambiguation page
disambig = tree.xpath('//table[@id="disambigbox"]') or \ disambig = tree.xpath('//table[@id="disambigbox"]') or \