Changeset 20204 for lang/php

Timestamp:

09/29/08 20:15:18 (2 months ago)

Author:

sasezaki

Message:

Flexible.php追加

Location:

lang/php/Scraper/library/Diggin

Files:

• Scraper/Adapter/Htmlscraping.php (modified) (6 diffs)
• Scraper/Strategy/Flexible.php (added)
• Scraper/Strategy/Selector.php (modified) (1 diff)
• Scraper/Strategy/Xpath.php (modified) (2 diffs)
• Uri/Http.php (modified) (1 diff)

lang/php/Scraper/library/Diggin/Scraper/Adapter/Htmlscraping.php

@@ -72 +72 @@
         
         if ($bases = $xml_object->xpath('//base[@href]')) {
-            $bases[0]['href'] = $this->getAbsoluteUrl((string) $bases[0]['href'], $this->config['url']);
+            require_once 'Diggin/Uri/Http.php';
+            $bases[0]['href'] = Diggin_Uri_Http::getAbsoluteUrl((string) $bases[0]['href'], $this->config['url']);
         } else {
             if (!$xml_object->head) {
@@ -203 +204 @@
         $responseBody = preg_replace('/<\?xml\b[^>]*?\?>/si', '', $responseBody);
         if (preg_match('/^\s*$/s', $responseBody)) {
-            throw new Exception('The entity body became empty after preprocessing.');
+            require_once 'Diggin/Scraper/Adapter/Exception.php';
+            throw new Diggin_Scraper_Adapter_Exception('The entity body became empty after preprocessing.');
         }
         /*
@@ -230 +232 @@
                 @mb_detect_order('ASCII, JIS, UTF-8, EUC-JP, SJIS');
                 if (false === $encoding = @mb_preferred_mime_name(@mb_detect_encoding($responseBody))) {
-                    throw new Exception('Failed detecting character encoding.');
+                    require_once 'Diggin/Scraper/Adapter/Exception.php';
+                    throw new Diggin_Scraper_Adapter_Exception('Failed detecting character encoding.');
                 }
             }
@@ -240 +243 @@
             foreach ($this->backup as $key => $value) {
                 if (false === $this->backup[$key] = @iconv($encoding, 'UTF-8', $value)) {
-                    throw new Exception('Failed converting character encoding.');
+                    require_once 'Diggin/Scraper/Adapter/Exception.php';
+                    throw new Diggin_Scraper_Adapter_Exception('Failed converting character encoding.');
                 }
             }
@@ -306 +310 @@
 
     /**
-     * @param  string  $name
-     * @param  string  $string
      * @return mixed
      */
@@ -354 +356 @@
         return $this;
     }
-
-    /**
-     * @param  string  $url
-     * @param  string  $base_url
-     * @return string
-     */
-    public static function getAbsoluteUrl($url, $base_url)
-    {
-        if (preg_match('/^[\w\+\-\.]+:/', $url) or false === $bases = @parse_url($base_url)) {
-            return $url;
-        } elseif (0 === strpos($url, '/')) {
-            return "$bases[scheme]://$bases[host]".(isset($bases['port'])? ":$bases[port]": '').$url;
-        } else {
-            if (!isset($bases['path'])) {
-                $bases['path'] = '/';
-            }
-            require_once 'Net/URL2.php';
-            return "$bases[scheme]://$bases[host]".(isset($bases['port'])? ":$bases[port]": '').
-                Net_URL2::resolvePath(substr($bases['path'], 0, strrpos($bases['path'], '/') +1).$url);
-        }
-    }
 }
 

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

@@ -122 +122 @@
         } elseif (strpos($process->type, '@') === 0) {
             $strings = array();
+            require_once 'Diggin/Uri/Http.php';
             foreach ($values as $value) {
-                //@todo
-                if (($process->type == '@href' OR $process->type == '@src')
-                    && method_exists($this->getAdapter(), 'getAbsoluteUrl')) {
-                    array_push($strings, $this->getAdapter()->getAbsoluteUrl((string)$value[substr($process->type, 1)], self::$_adapterconfig['url']));
-                    //require_once 'Diggin/Uri/Http.php';
-                    //array_push($strings, Diggin_Uri_Http::getAbsoluteUrl((string)$value[substr($process->type, 1)], self::$_adapterconfig['url']));
+                if (($process->type == '@href' OR $process->type == '@src')) {
+                    array_push($strings, Diggin_Uri_Http::getAbsoluteUrl((string)$value[substr($process->type, 1)], self::$_adapterconfig['url']));
                 } else {
                     array_push($strings, (string) $value[substr($process->type, 1)]);

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

@@ -123 +123 @@
         } elseif (strpos($process->type, '@') === 0) {
             $strings = array();
-            require_once 'Diggin/Scraper/Adapter/Htmlscraping.php';
+            require_once 'Diggin/Uri/Http.php';
             foreach ($values as $value) {
                 if (($process->type == '@href' OR $process->type == '@src')) {
-                    array_push($strings, Diggin_Scraper_Adapter_Htmlscraping::getAbsoluteUrl((string)$value[substr($process->type, 1)], self::$_adapterconfig['url']));
-
-                    //require_once 'Diggin/Uri/Http.php';
-                    //array_push($strings, Diggin_Uri_Http::getAbsoluteUrl((string)$value[substr($process->type, 1)], self::$_adapterconfig['url']));
+                    array_push($strings, Diggin_Uri_Http::getAbsoluteUrl((string)$value[substr($process->type, 1)], self::$_adapterconfig['url']));
                 } else {
                     array_push($strings, (string) $value[substr($process->type, 1)]);
@@ -142 +139 @@
         return $strings;
     }
-
 }
-

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

@@ -28 +28 @@
     public static function getAbsoluteUrl($url, $base_url)
     {
-        $parse = parse_url($url);
-        if (isset($parse["host"])) {
-            $build = $url;
+        //using pecl_http
+        if (extension_loaded('http')) {
+            static $base_url;
+            if (array_key_exists('host', parse_url($url))) {
+                return $url;
+            } else {
+                if (strpos(pathinfo(parse_url($base_url, PHP_URL_PATH), PATHINFO_DIRNAME), '/') === false) {
+                    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,), 
+                    HTTP_URL_JOIN_PATH | HTTP_URL_STRIP_QUERY | HTTP_URL_STRIP_FRAGMENT);
+                }
+            }
+        //Net_URL2 ver 0.2.0
         } else {
-            $uridir = pathinfo(parse_url($base_url, PHP_URL_PATH), PATHINFO_DIRNAME);
-            $slash = strpos($uridir, '/');
-            if ($slash === false) {
-                $build = http_build_url($base_url, array("path" => $url,),
-                HTTP_URL_STRIP_QUERY | HTTP_URL_STRIP_FRAGMENT);
-            } else {            
-                $build = http_build_url($base_url, array("path" => $url,), 
-                HTTP_URL_JOIN_PATH | HTTP_URL_STRIP_QUERY | HTTP_URL_STRIP_FRAGMENT);
-            }
+            if (!class_exists('Net_URL2')) require_once 'Net/URL2.php';
+            $neturl2 = new Net_URL2($base_url);
+            return $neturl2->resolve($url)->getUrl();
         }
-        
-        return $build;
     }
 }
+