Changeset 15861

Timestamp:

07/15/08 23:55:39 (5 years ago)

Author:

sasezaki

Message:

lang/php/Scraper: abstactのgetValuesにて処理するように

Location:

lang/php/Scraper/library/Diggin

Files:

• Scraper.php (modified) (1 diff)
• Scraper/Strategy/Abstract.php (modified) (2 diffs)
• Scraper/Strategy/Selector.php (modified) (5 diffs)
• Scraper/Strategy/Xpath.php (modified) (4 diffs)

lang/php/Scraper/library/Diggin/Scraper.php

@@ -306 +306 @@
         $context = new Diggin_Scraper_Context($this->getStrategy($resource));
         foreach (self::$_processes as $process) {
-            $values = self::$_strategy->getValue($context, $process);
-
-            if ($process->filters) {
-                require_once 'Diggin/Scraper/Filter.php';
-                $values = Diggin_Scraper_Filter::run($values, $process->filters);
-            }
-  
+            $values = self::$_strategy->getValues($context, $process);
+
             $this->results[$process->name] = $values;
         }

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

@@ -39 +39 @@
     }
     
-    /**
-     * 
-     */
-    public function getData()
-    {
-        //if !is_readble($this->getBody)...
-        
-        return $this->readData($this->getResponse());
-    }
-    
     public function scrapedData($process)
     {   
@@ -61 +51 @@
     protected abstract function scrape($response, $process);
     
-    protected abstract function getValue($context, $process);
+    protected abstract function getValue($values, $process);
+    
+    protected abstract static function extract($values, $process);
+
+    public function getValues($context, $process)
+    {
+        if (!isset($process->type)) {
+            return $context->scrape($process);
+        }
+        
+        if ($context instanceof Diggin_Scraper_Context) {
+            $values = $context->scrape($process);
+        } else {
+            $values = $this->extract($context, $process);
+        }
+                
+        if ($process->type instanceof scraper) {
+            foreach ($values as $count => $val) {
+                foreach ($process->type->processes as $proc) {
+                    $returns[$count][$proc->name] = $this->getValues($val, $proc);
+                }
+                
+                if (($process->arrayflag === false) && $count === 0) break;
+            }
+           return $returns;
+        }
+        
+        $values = $this->getValue($values, $process);
+        
+        if ($process->arrayflag === false && strtoupper($process->type) === 'RAW') {
+            $values = array_shift($values);
+        } elseif ($process->arrayflag === false) {
+            $values = (string) array_shift($values);
+        }
+        
+        if ($process->filters) {
+            require_once 'Diggin/Scraper/Filter.php';
+            $values = Diggin_Scraper_Filter::run($values, $process->filters);
+        }
+        
+        return $values;
+    }
 }

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

@@ -16 +16 @@
 
 require_once 'Diggin/Scraper/Strategy/Abstract.php';
-
+require_once dirname(__FILE__).'/Selector/sfDomCssSelector.class.php';
 class Diggin_Scraper_Strategy_Selector extends Diggin_Scraper_Strategy_Abstract 
 {    
@@ -71 +71 @@
         $simplexml = $this->getAdapter()->readData($respose);
 
+        return self::extract($simplexml, $process);
+    }
+    
+    public static function extract($simplexml, $process)
+    {
         $dom = dom_import_simplexml($simplexml);
-
-        require_once dirname(__FILE__).'/Selector/sfDomCssSelector.class.php';
+        
         $selector = new sfDomCssSelector($dom);
         
@@ -83 +87 @@
         return $results;
     }
+    
 
     /**
@@ -91 +96 @@
      * @return mixed
      */
-    public function getValue($context, $process)
-    {
-        
-        if (!isset($process->type)) {
-            return $context->scrape($process);
-        }
-        
-        if ($context instanceof Diggin_Scraper_Context) {
-            $values = $context->scrape($process);
-        } else {
-            $dom = dom_import_simplexml($context); 
-            $selector = new sfDomCssSelector($dom);
-            
-            $values = array();       
-            foreach ($selector->getElements($process->expression) as $result) {
-                $values[] = simplexml_import_dom($result); 
-            }
-        }
-
-        if ($process->type instanceof scraper) {
-            foreach ($values as $count => $val) {
-                foreach ($process->type->processes as $proc) {
-                    $returns[$count][$proc->name] = $this->getValue($val, $proc);
-                }
-            }
-            
-            if($process->arrayflag === false) {
-                $returns = array_shift($returns);
-            }
-            
-           return $returns;
-        }
-        
+    public function getValue($values, $process)
+    {        
         //type
         if (strtoupper(($process->type)) === 'RAW'){
@@ -161 +135 @@
         }
         
-        if ($process->arrayflag === false && strtoupper($process->type) === 'RAW') {
-            $strings = array_shift($strings);
-        } elseif ($process->arrayflag === false) {
-            $strings = (string) array_shift($strings);
-        }
-
         return $strings;
     }

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

@@ -72 +72 @@
     {
         $simplexml = $this->getAdapter()->readData($respose);
- 
-        $results = array();       
-        foreach ($simplexml->xpath($process->expression) as $result) {
+        
+        return self::extract($simplexml, $process);
+    }
+    
+    public static function extract($values, $process)
+    {
+        $results = array();
+        foreach ($values->xpath($process->expression) as $result) {
             $results[] = $result; 
         }
@@ -80 +85 @@
         return $results;
     }
-    
+
     /**
      * get value with DSL
@@ -88 +93 @@
      * @return mixed
      */
-    public function getValue($context, $process)
+    public function getValue($values, $process)
     {
-        if (!isset($process->type)) {
-            return $context->scrape($process);
-        }
-        
-        if ($context instanceof Diggin_Scraper_Context) {
-            $values = $context->scrape($process);
-        } else {
-            
-            $values = array();
-            foreach ($context->xpath($process->expression) as $result) {
-                $values[] = $result; 
-            }
-        }
-                
-        if ($process->type instanceof scraper) {
-            foreach ($values as $count => $val) {
-                foreach ($process->type->processes as $proc) {
-                    $returns[$count][$proc->name] = $this->getValue($val, $proc);
-                }
-            }
-            
-            if($process->arrayflag === false) {
-                $returns = array_shift($returns);
-            }
-           return $returns;
-        }
-
         //type
         if (strtoupper(($process->type)) === 'RAW') {
             $strings = $values;
         } elseif (strtoupper(($process->type)) === 'TEXT') {
-            
             $strings = array();
             foreach ($values as $value) {
@@ -153 +130 @@
             throw new Diggin_Scraper_Strategy_Exception("can not understand type :".$process->type);
         }
-
-        if ($process->arrayflag === false && strtoupper($process->type) === 'RAW') {
-            $strings = array_shift($strings);
-        } elseif ($process->arrayflag === false) {
-            $strings = (string) array_shift($strings);
-        }
         
         return $strings;
     }
+
 }