Changeset 6031 for platform/tdiary/lib

Timestamp:

02/02/08 15:11:57 (10 months ago)

Author:

hsbt

Message:

platform/tdiary/tdiary/lib/bayes.rb: backport from official repos.

Files:

• platform/tdiary/lib/bayes.rb (modified) (6 diffs)

platform/tdiary/lib/bayes.rb

@@ -3 +3 @@
 
 require "pstore"
-require "uri"
 
 module Bayes
@@ -17 +16 @@
                         KCODE = "e"
                         KATAKANA = "\xa5\xa2-\xa5\xf3"
+      BAR = "\xa1\xbc"
                         KANJI = "\xb0\xa1-\xfc\xfe"
-                        RE_KATAKANA = /[#{KATAKANA}]{2,}/eo
+                        RE_KATAKANA = /[#{KATAKANA}#{BAR}]{2,}/eo
                         RE_KANJI = /[#{KANJI}]{2,}/eo
 
@@ -33 +33 @@
                         end
                         KATAKANA = utf_range(0x30a0, 0x30ff)
+      BAR = c2u(0x30fc)
                         KANJI = utf_range(0x4e00, 0x9faf)
-                        RE_KATAKANA = /[#{KATAKANA}]{2,}/uo
+                        RE_KATAKANA = /[#{KATAKANA}#{BAR}]{2,}/uo
                         RE_KANJI = /[#{KANJI}]{2,}/uo
 
@@ -85 +86 @@
 
                                 h = host
-                                while /^(.*?)[._-](.*)$/=~h
+                                while /^(.*?)[\.\-_](.*)$/=~h
                                         h = $2
                                         push($1, prefix)
@@ -95 +96 @@
 
                 def add_url(url, prefix=nil)
-                        if URI.regexp(%w[http https ftp]) === url
-                                url  = URI.parse url
-                                host = url.host                       # $4
-                                path = url.path.gsub(%r{^/+|/+$}, '') # $7
+                        if %r[^(?:https?|ftp)://(.*?)(?::\d+)?/(.*?)\/?(\?.*)?$] =~ url
+                                host = $1
+                                path = $2
 
                                 add_host(host, prefix)
@@ -106 +106 @@
 
                                         p = path
-                                        re = %r[^(.*)[/._-](.*?)$]
+                                        re = %r[^(.*)[/\-\._](.*?)$]
                                         while re=~p
                                                 p = $1