Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

Aamzon plugin: Workaround broken encoding. Detect and use mobile user agent search results page

Browse Source
This commit is contained in:
Kovid Goyal 2011-04-05 11:35:53 -06:00
parent 0d924d81ef
commit 6e98d78dd7
1 changed files with 16 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
src/calibre/ebooks/metadata/sources/amazon.py
View File

@ -218,6 +218,9 @@ class Worker(Thread): # Get details {{{
' @class="emptyClear" or @href]'): ' @class="emptyClear" or @href]'):
c.getparent().remove(c) c.getparent().remove(c)
desc = tostring(desc, method='html', encoding=unicode).strip() desc = tostring(desc, method='html', encoding=unicode).strip()
# Encoding bug in Amazon data U+fffd (replacement char)
# in some examples it is present in place of '
desc = desc.replace('\ufffd', "'")
# remove all attributes from tags # remove all attributes from tags
desc = re.sub(r'<([a-zA-Z0-9]+)\s[^>]+>', r'<\1>', desc) desc = re.sub(r'<([a-zA-Z0-9]+)\s[^>]+>', r'<\1>', desc)
# Collapse whitespace # Collapse whitespace
@ -410,6 +413,18 @@ class Amazon(Source):
if 'bulk pack' not in title: if 'bulk pack' not in title:
matches.append(a.get('href')) matches.append(a.get('href'))
break break
if not matches:
# This can happen for some user agents that Amazon thinks are
# mobile/less capable
log('Trying alternate results page markup')
for td in root.xpath(
r'//div[@id="Results"]/descendant::td[starts-with(@id, "search:Td:")]'):
for a in td.xpath(r'descendant::td[@class="dataColumn"]/descendant::a[@href]/span[@class="srTitle"]/..'):
title = tostring(a, method='text', encoding=unicode).lower()
if 'bulk pack' not in title:
matches.append(a.get('href'))
break
# Keep only the top 5 matches as the matches are sorted by relevance by # Keep only the top 5 matches as the matches are sorted by relevance by
# Amazon so lower matches are not likely to be very relevant # Amazon so lower matches are not likely to be very relevant