Changeset 20533

Timestamp:

10/02/08 23:45:03 (3 months ago)

Author:

sasezaki

Message:

modify getAbsoulteUrl() etc.

Location:

lang/php/Scraper

Files:

• library/Diggin/Scraper/Strategy/Flexible.php (modified) (1 diff)
• library/Diggin/Uri/Http.php (modified) (3 diffs)
• tests/Diggin/Uri/HttpTest.php (modified) (2 diffs)

lang/php/Scraper/library/Diggin/Scraper/Strategy/Flexible.php

@@ -100 +100 @@
     protected static function _xpathOrCss2Xpath($exp){
         if (preg_match('!^(?:/|id\()!', $exp)) {
-            return $exp;
+            return '.'.$exp;
         } else {
             if ($exp === '.') {

lang/php/Scraper/library/Diggin/Uri/Http.php

@@ -15 +15 @@
  */
 
+/**
+ * Utils For Http
+ */
 class Diggin_Uri_Http
 {
@@ -35 +38 @@
             } else {
                 if (strpos(pathinfo(parse_url($base_url, PHP_URL_PATH), PATHINFO_DIRNAME), '/') === false) {
-                    return http_build_url($base_url, array("path" => $url,),
+                    return http_build_url($base_url, array("path" => $url),
                     HTTP_URL_STRIP_QUERY | HTTP_URL_STRIP_FRAGMENT);
                 } else {            
-                    return http_build_url($base_url, array("path" => $url,), 
+                    return http_build_url($base_url, array("path" => $url), 
                     HTTP_URL_JOIN_PATH | HTTP_URL_STRIP_QUERY | HTTP_URL_STRIP_FRAGMENT);
                 }
@@ -44 +47 @@
         //Net_URL2 ver 0.2.0
         } else {
-            if (!class_exists('Net_URL2')) require_once 'Net/URL2.php';
+                        if (!class_exists('Net_URL2')) require_once 'Net/URL2.php';
+                        static $neturl2;
             $neturl2 = new Net_URL2($base_url);
-            return $neturl2->resolve($url)->getUrl();
-        }
+            return $neturl2->resolve(str_replace(chr(32), '%20', $url))->getUrl();
+        } 
     }
 }

lang/php/Scraper/tests/Diggin/Uri/HttpTest.php

@@ -40 +40 @@
      * check getting absoluteUrl
      *
-     *
-     * import from rhaco's doc-test
-     * @see http://rhaco.googlecode.com/svn/trunk/1_6_1/network/Url.php
      */
     public function testGetAbsoluteUrl()
     {
+        //if 
+        $this->assertEquals('http://yahoo.com/test/', 
+                             $this->object->getAbsoluteUrl('http://yahoo.com/test/', 'http://www.rhaco.org/'));
+        
+        // import from rhaco's doc-test
+        // @see http://rhaco.googlecode.com/svn/trunk/1_6_1/network/Url.php
         $this->assertEquals('http://www.rhaco.org/doc/ja/index.html', 
                              $this->object->getAbsoluteUrl('/doc/ja/index.html', 'http://www.rhaco.org/'));
@@ -70 +73 @@
         $this->assertEquals('http://www.rhaco.org/index.html', 
                              $this->object->getAbsoluteUrl('/index.html', 'http://www.rhaco.org/doc/ja'));
+                             
+                //@see http://d.hatena.ne.jp/kitamomonga/20080410/ruby_mechanize_percent_url_bug
+        $this->assertEquals('http://test.org/doc/ja/index.cgi?param=hoge', 
+                             $this->object->getAbsoluteUrl('?param=hoge', 'http://test.org/doc/ja/index.cgi?test=bar'));
+        $this->assertEquals('http://test.org/index.php?param=hoge', 
+                             $this->object->getAbsoluteUrl('?param=hoge', 'http://test.org/index.php'));                
+        //if space,
+        $this->assertEquals('http://www.rhaco.org/doc/ja/'.rawurlencode('test space.html'),
+                             $this->object->getAbsoluteUrl('test space.html', 'http://www.rhaco.org/doc/ja/'));
+        
     }
 }