summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorGiuseppe Bilotta <giuseppe.bilotta@gmail.com>2008-07-07 23:14:24 +0200
committerGiuseppe Bilotta <giuseppe.bilotta@gmail.com>2008-07-07 23:17:40 +0200
commit61a22b729532316193612eac4f948ff11aa57f50 (patch)
tree440fac36aa27f62c70b28e354eff9f4b97a75d3b
parentbe2e4072dcc4a2324067a571096592bd075c63e4 (diff)
httputil: don't fail when b0rked servers put the charset in the content-encoding
-rw-r--r--lib/rbot/core/utils/httputil.rb9
1 files changed, 9 insertions, 0 deletions
diff --git a/lib/rbot/core/utils/httputil.rb b/lib/rbot/core/utils/httputil.rb
index 9df8fc2c..ae16416d 100644
--- a/lib/rbot/core/utils/httputil.rb
+++ b/lib/rbot/core/utils/httputil.rb
@@ -109,7 +109,16 @@ module ::Net
# TODO
# debug "full inflation failed (#{e}), trying to recover as much as possible"
end
+ when /^(?:iso-8859-\d+|windows-\d+|utf-8|utf8)$/i
+ # B0rked servers (Freshmeat being one of them) sometimes return the charset
+ # in the content-encoding; in this case we assume that the document has
+ # a standarc content-encoding
+ old_hsh = self.to_hash
+ self['content-type']= self['content-type']+"; charset="+method.downcase
+ warning "Charset vs content-encoding confusion, trying to recover: from\n#{old_hsh.pretty_inspect}to\n#{self.to_hash.pretty_inspect}"
+ return str
else
+ debug self.to_hash
raise "Unhandled content encoding #{method}"
end
end