Japanese edge cases, meet your match.

README.md

@@ -45,6 +45,8 @@ If you are creating a new twitter-text library in a different programming langua
     * [FIX] Japanese autolink including long vowel mark (chouon)
     * [FIX] Japanese autolink after a full-width exclamation point
     * [FIX] Japanese autolink including ideographic iteration mark
+    * [FIX] Add hashtag extraction with indices test for new language hashtags
+    * [FIX] Add hashtag extraction with indices test for multiple latin hashtags
 
   * v1.4.2 - 2011-07-08 [ Git tag v1.4.2 ]
     * [FIX] Additional Japanese hashtag autolinking tests

autolink.yml

@@ -347,11 +347,14 @@ tests:
       text: "できましたよー！＃日本語ハッシュタグ。"
       expected: "できましたよー！<a href=\"http://twitter.com/search?q=%23日本語ハッシュタグ\" title=\"#日本語ハッシュタグ\" class=\"tweet-url hashtag\">＃日本語ハッシュタグ</a>。"
 
-
     - description: "Autolink a hashtag containing ideographic iteration mark"
       text: "#云々"
       expected: "<a href=\"http://twitter.com/search?q=%23云々\" title=\"#云々\" class=\"tweet-url hashtag\">#云々</a>"
 
+    - description: "Autolink multiple hashtags in multiple languages"
+      text: "Hashtags in #中文, #日本語, #한국말, and #русский! Try it out!"
+      expected: "Hashtags in <a href=\"http://twitter.com/search?q=%23中文\" title=\"#中文\" class=\"tweet-url hashtag\">#中文</a>, <a href=\"http://twitter.com/search?q=%23日本語\" title=\"#日本語\" class=\"tweet-url hashtag\">#日本語</a>, <a href=\"http://twitter.com/search?q=%23한국말\" title=\"#한국말\" class=\"tweet-url hashtag\">#한국말</a>, and <a href=\"http://twitter.com/search?q=%23русский\" title=\"#русский\" class=\"tweet-url hashtag\">#русский</a>! Try it out!"
+
   urls:
     - description: "Autolink URL with pipe character"
       text: "text http://example.com/pipe|character?yes|pipe|character"

extract.yml

@@ -445,9 +445,35 @@ tests:
         - hashtag: "hashtag"
           indices: [7, 15]
 
-    - description: "Extract a hastag in a string of multi-byte characters"
+    - description: "Extract a hashtag in a string of multi-byte characters"
       text: "会議中 #hashtag 会議中"
       expected:
         - hashtag: "hashtag"
           indices: [4, 12]
 
+    - description: "Extract multiple valid hashtags"
+      text: "One #two three #four"
+      expected:
+        - hashtag: "two"
+          indices: [4, 8]
+        - hashtag: "four"
+          indices: [15, 20]
+
+    - description: "Extract a non-latin hashtag"
+      text: "Hashtags in #русский!"
+      expected:
+        - hashtag: "русский"
+          indices: [12, 20]
+
+    - description: "Extract multiple non-latin hashtags"
+      text: "Hashtags in #中文, #日本語, #한국말, and #русский! Try it out!"
+      expected:
+        - hashtag: "中文"
+          indices: [12, 15]
+        - hashtag: "日本語"
+          indices: [17, 21]
+        - hashtag: "한국말"
+          indices: [23, 27]
+        - hashtag: "русский"
+          indices: [33, 41]
+