Changeset 5703 for platform/tdiary/lib

Timestamp:

01/28/08 20:10:16 (10 months ago)

Author:

drry

Message:

platform/tdiary/lib/bayes.rb: use URI.regexp() for URL detection. fixed some regexps.

Files:

• platform/tdiary/lib/bayes.rb (modified) (4 diffs)

platform/tdiary/lib/bayes.rb

@@ -3 +3 @@
 
 require "pstore"
+require "uri"
 
 module Bayes
@@ -84 +85 @@
 
                                 h = host
-                                while /^(.*?)[\.\-_](.*)$/=~h
+                                while /^(.*?)[._-](.*)$/=~h
                                         h = $2
                                         push($1, prefix)
@@ -94 +95 @@
 
                 def add_url(url, prefix=nil)
-                        if %r[^(?:https?|ftp)://(.*?)(?::\d+)?/(.*?)\/?(\?.*)?$] =~ url
-                                host = $1
-                                path = $2
+                        if URI.regexp(%w[http https ftp]) === url
+                                url  = URI.parse url
+                                host = url.host                       # $4
+                                path = url.path.gsub(%r{^/+|/+$}, '') # $7
 
                                 add_host(host, prefix)
@@ -104 +106 @@
 
                                         p = path
-                                        re = %r[^(.*)[/\-\._](.*?)$]
+                                        re = %r[^(.*)[/._-](.*?)$]
                                         while re=~p
                                                 p = $1