From f6e9d17f9bde89d4ff2528f62f69f1d7944a5025 Mon Sep 17 00:00:00 2001
From: Jacques Distler <distler@golem.ph.utexas.edu>
Date: Thu, 22 Jul 2010 14:08:21 -0500
Subject: [PATCH] Non-Latin WikiWords

Fix bug in WikiWord processing
reproted by Alexander Hambug.

Also fix Cyrillic Capital/Lowercase
regexps.
---
 lib/wiki_content.rb | 2 +-
 lib/wiki_words.rb   | 9 +++++----
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/lib/wiki_content.rb b/lib/wiki_content.rb
index a5dc473d..95e04271 100644
--- a/lib/wiki_content.rb
+++ b/lib/wiki_content.rb
@@ -181,7 +181,7 @@ class WikiContent < ActiveSupport::SafeBuffer
     @options[:engine].apply_to(copy)
 
     copy.inside_chunks(@options[:hide_chunks]) do |id|
-      @chunks_by_id[id.to_i].revert
+      @chunks_by_id[id.to_i].revert if @chunks_by_id[id.to_i]
     end
   end
 
diff --git a/lib/wiki_words.rb b/lib/wiki_words.rb
index bc3e1824..3839c381 100644
--- a/lib/wiki_words.rb
+++ b/lib/wiki_words.rb
@@ -8,16 +8,17 @@ module WikiWords
   I18N_HIGHER_CASE_LETTERS =
     "ÀÁÂÃÄÅĀĄĂÆÇĆČĈĊĎĐÈÉÊËĒĘĚĔĖĜĞĠĢĤĦÌÍÎÏĪĨĬĮİIJĴĶŁĽĹĻĿÑŃŇŅŊÒÓÔÕÖØŌŐŎŒŔŘŖŚŠŞŜȘŤŢŦȚÙÚÛÜŪŮŰŬŨŲŴŶŸȲÝŹŽŻ" + 
     "ΑΒΓΔΕΖΗΘΙΚΛΜΝΞΟΠΡΣΤΥΦΧΨΩ" + 
-    "ΆΈΉΊΌΎΏѠѢѤѦѨѪѬѮѰѲѴѶѸѺѼѾҀҊҌҎҐҒҔҖҘҚҜҞҠҢҤҦҨҪҬҮҰҲҴҶҸҺҼҾӁӃӅӇӉӋӍӏӐӒӔӖӘӚӜӞӠӢӤӦӨӪӬӮӰӲӴӸЖ" +
+    "ЀЁЂЃЄЅІЇЈЉЊЋЌЍЎЏАБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯѠѢѤѦѨѪѬѮѰѲѴѶѸѺѼѾҀҊҌҎҐҒҔҖҘҚҜҞҠҢҤҦҨҪҬҮҰҲҴҶҸҺҼҾӀӁӃӅӇӉӋӍӐӒӔӖӘӚӜӞӠӢӤӦӨӪӬӮӰӲӴӶӸӺӼӾԀԂԄԆԈԊԌԎԐԒԔԖԘԚԜԞԠԢ" +
     "ԱԲԳԴԵԶԷԸԹԺԻԼԽԾԿՀՁՂՃՄՅՆՇՈՉՊՋՌՍՎՏՐՑՒՓՔՕՖ"
 
   I18N_LOWER_CASE_LETTERS =
     "àáâãäåāąăæçćĉċčďđèéêëēęěĕėƒĝğġģĥħìíîïīĩĭįıijĵķĸłľĺļŀñńňņʼnŋòóôõöøōŏőœŕřŗśŝšşșťţŧțùúûüūůűŭũųŵýÿŷžżźÞþßſð" +
     "άέήίΰαβγδεζηθικλμνξοπρςστυφχψωϊϋόύώΐ" +
-    "абвгдежзийклмнопрстуфхцчшщъыьэюяѐёђѓєѕіїјљћќѝўџѡѣѥѧѩѫѭѯѱѳѵѷѹѻѽѿҁҋҌҍҏґғҕҗҙқҝҟҡңҥҧҩҫҭүұҳҵҷҹһҽҿӀӂӄӆӈӊӌӎӑӓӕӗәӛӝӟӡӣӥӧөӫӭӯӱӳӵӹ" +
-    "աբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցւփքօֆև"
+    "абвгдежзийклмнопрстуфхцчшщъыьэюяѐёђѓєѕіїјљњћќѝўџѡѣѥѧѩѫѭѯѱѳѵѷѹѻѽѿҁҋҍҏґғҕҗҙқҝҟҡңҥҧҩҫҭүұҳҵҷҹһҽҿӂӄӆӈӊӌӎӏӑӓӕӗәӛӝӟӡӣӥӧөӫӭӯӱӳӵӷӹӻӽӿԁԃԅԇԉԋԍԏԑԓԕԗԙԛԝԟԡԣ" +
+     "աբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցւփքօֆև"
 
-  WIKI_WORD_PATTERN = '[A-Z' + I18N_HIGHER_CASE_LETTERS + ']+[a-z' + I18N_LOWER_CASE_LETTERS + ']+[A-Z' + I18N_HIGHER_CASE_LETTERS + ']\w+'
+  WIKI_WORD_PATTERN = '[A-Z' + I18N_HIGHER_CASE_LETTERS + ']+[a-z' + I18N_LOWER_CASE_LETTERS + ']+[A-Z' + I18N_HIGHER_CASE_LETTERS + 
+    '][A-Za-z0-9_' + I18N_HIGHER_CASE_LETTERS +  I18N_LOWER_CASE_LETTERS + ']+'
   CAMEL_CASED_WORD_BORDER = /([a-z#{I18N_LOWER_CASE_LETTERS}])([A-Z#{I18N_HIGHER_CASE_LETTERS}])/u
 
   def self.separate(wiki_word)