Merge pull request #2318 from camilleroux/patch-3

Update the Amazon links regexp
2014-05-06 15:15:33 -04:00 · 2014-05-06 15:15:33 -04:00 · 4cdf25e39d
parent 2cf6f65660 f14c71b9d4
commit 4cdf25e39d
1 changed files with 1 additions and 1 deletions
--- a/app/jobs/regular/crawl_topic_link.rb
+++ b/app/jobs/regular/crawl_topic_link.rb
@ -50,7 +50,7 @@ module Jobs
    def self.max_chunk_size(uri)
      # Amazon leaves the title until very late. Normally it's a bad idea to make an exception for
      # one host but amazon is a big one.
-      return 80 if uri.host =~ /amazon\.(com|ca|co.uk)$/
+      return 80 if uri.host =~ /amazon\.(com|ca|co\.uk|es|fr|de|it|com\.au|com\.br|cn|in|co\.jp|com\.mx)$/

      # Default is 10k
      10