[sup-talk] Xapian: Term too long

[Tero Tilus <tero@tilus.net>]

sup-sync blows up like this

/home/terotil/src/sup/lib/sup/xapian_index.rb:446:in `replace_document': InvalidArgumentError: Term too long (> 245): Lfwd: =?iso-8859-1?q?tekij=e4n_oikeudet=5d?= (ArgumentError)
x-enigmail-version: 0.92.0.0
content-type: multipart/mixed;
 boundary="------------010606010007070802040301"
x-virus-scanned: amavisd-new at cc.jyu.fi
x-spam-status: no, hits=-2.373 required=5 tests=[awl=0.226, bayes_00=-2.599
        from /home/terotil/src/sup/lib/sup/xapian_index.rb:446:in `sync_message'
        from /usr/lib/ruby/1.8/monitor.rb:242:in `synchronize'
        from /home/terotil/src/sup/lib/sup/xapian_index.rb:363:in `synchronize'
        from /home/terotil/src/sup/lib/sup/xapian_index.rb:440:in `sync_message'
        from /home/terotil/src/sup/lib/sup/xapian_index.rb:92:in `add_message'
        from /home/terotil/src/sup/bin/sup-sync:211
	...

Relevant part of the problematic mail looks like this

User-Agent: Debian Thunderbird 1.0.6 (X11/20050802)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: mutikainen@iki.fi
Subject: [Fwd: =?ISO-8859-1?Q?tekij=E4n_oikeudet=5D?=
X-Enigmail-Version: 0.92.0.0
Content-Type: multipart/mixed;
 boundary="------------010606010007070802040301"
X-Virus-Scanned: amavisd-new at cc.jyu.fi
X-Spam-Status: No, hits=-2.373 required=5 tests=[AWL=0.226, BAYES_00=-2.599]
X-Spam-Level: 
X-Sorted: Whitelist
Content-Length: 11892

This is how I solved it for me, for now

diff --git a/lib/sup/xapian_index.rb b/lib/sup/xapian_index.rb
index ad45b0e..d3b3e25 100644
--- a/lib/sup/xapian_index.rb
+++ b/lib/sup/xapian_index.rb
@@ -443,7 +443,11 @@ EOS
         warn "docid underflow, dropping #{m.id.inspect}"
         return
       end
-      @xapian.replace_document docid, doc
+      begin
+        @xapian.replace_document docid, doc
+      rescue StandardError => err
+        warn "Failed to add message #{m.id.inspect} to Xapian index: #{err}"
+      end
     end
 
     m.labels.each { |l| LabelManager << l }

Looks like lib/sup/xapian_index.rb tries to override
Xapian::Document#add_term with a version which is wired to ditch too
long terms.  Only that you can't override methods just by including a
module.  Methods of the including class override methods in included
module.

terotil@sotka:~$ irb
> class Foo; def bar; :bar; end; end
=> nil
> module Baz; def bar; :baz; end; end
=> nil
> class Foo; include Baz; end
=> Foo
> Foo.new.bar
=> :bar
> Foo.ancestors
=> [Foo, Baz, Object, Kernel]  # Foo before Baz, methods in Foo take priority

It is still Foo#bar being called, not Baz#bar.  You need to open up
Xapian::Document and then do alias method chaining to override
methods.  Or you could do tricks like
http://coderrr.wordpress.com/2008/10/29/secure-alias-method-chaining/

-- 
Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/
_______________________________________________
sup-talk mailing list
sup-talk@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-talk