Changeset 24890 for lang

Timestamp:

11/26/08 04:35:02 (4 years ago)

Author:

drry

Message:

• User-Agent ヘッダから不正な文字を除去しました。
  
• 正規表現を修正しました。
  
• ほか。
  

Location:

lang/ruby/enokidu_antenna

Files:

• enokidu.rb (modified) (7 diffs)
• enokidu_test.rb (modified) (1 diff)

lang/ruby/enokidu_antenna/enokidu.rb

@@ -32 +32 @@
       @timeout = options[:timeout] || 30
       @debug   = options[:debug] || false
-      @request = Request.new(:timeout => @timeout, :debug   => @debug)
+      @request = Request.new(:timeout => @timeout, :debug  => @debug)
     end
 
     def detective(options = {})
-      page    = options[:page]
+      page = options[:page]
 
       page.body ||= ''
@@ -76 +76 @@
   class Request
     def initialize(options = {})
-      @http_header = {'User-Agent' => Enokidu::UserAgent::NAME, 'Connection' => 'close'}
+      @http_header = {
+        'User-Agent' => "#{Enokidu::UserAgent::NAME}/#{Enokidu::UserAgent::VERSION}".gsub(/:+/, ''),
+        'Connection' => 'close'
+      }
       @timeout = options[:timeout] || 30
       @debug   = options[:debug] || false
@@ -114 +117 @@
 
         # ロケーションが空または不完全な場合
-        if !location.empty? and !location.match(%r[^http://])
+        if !location.empty? and !location.match(%r[^https?://])
           uri = URI.parse(old_uri)
           path, query = location.split('?')
@@ -136 +139 @@
       if page.title.empty?
         debug_print "get page title"
-        page.title = body.scan(/<title>([^<]*)/im).to_s.strip
+        page.title = body.scan(/<title(?:\s[^>]*)?>\s*([^<]*)/im).to_s.rstrip
         page.title = 'no title' if page.title.empty?
       end
@@ -144 +147 @@
       else
         body
-      end.gsub(/<[^>]*>|\s|\n/, '')
+      end.gsub(/<[^>]*>|\s+/, '')
 
       # 更新チェック
@@ -189 +192 @@
       end
 
-      # Etag で更新チェック
+      # ETag で更新チェック
       if !lm and etag and page.etag != etag
         page.etag = etag
@@ -196 +199 @@
       end
 
-      # Last-Modified も Etag も使えない
+      # Last-Modified も ETag も使えない
       raise InvalidHeadResponse unless lm or etag
 

lang/ruby/enokidu_antenna/enokidu_test.rb

@@ -46 +46 @@
     before_modified_at = page.last_modified_at
     page = detective(:page => page)
-    
+
     assert page.updated?
     assert before_modified_at < page.last_modified_at