Changeset 2222:30fe556de226 for inc/core/class.dc.store.php

Timestamp:

10/03/13 00:57:41 (12 years ago)

Author:

Denis Jean-Chirstian <contact@…>

Branch:

dcRepo

Message:

Enhance repository search engine (and some fix)

File:

• inc/core/class.dc.store.php (modified) (4 diffs)

inc/core/class.dc.store.php

@@ -15 +15 @@
 @ingroup DC_CORE
 @brief Repository modules XML feed reader
+@since 2.6
 
 Provides an object to parse XML feed of modules from repository.
@@ -24 +25 @@
      /** @var  object    dcModules instance */
      public $modules;
+
+     /** @var  array     Modules fields to search on and their weighting */
+     public static $weighting = array('id' => 10, 'name' => 8, 'author' => 6, 'tags' => 4, 'desc' => 2);
 
      /** @var  string    User agent used to query repository */
@@ -126 +130 @@
 
           # Split query into small clean words
-          $patterns = explode(' ', $pattern);
-          array_walk($patterns, array('dcStore','sanitize'));
+          if (!($patterns = self::patternize($pattern))) {
+               return $result;
+          }
 
           # For each modules
           foreach ($this->data['new'] as $id => $module) {
-
-               # Split modules infos into small clean word
-               $subjects = explode(' ', $id.' '.$module['name'].' '.$module['desc']);
-               array_walk($subjects, array('dcStore','sanitize'));
-
-               # Check contents
-               if (!($nb = preg_match_all('/('.implode('|', $patterns).')/', implode(' ', $subjects), $_))) {
-                    continue;
-               }
-
-               # Add module to result
-               if (!isset($sorter[$id])) {
-                    $sorter[$id] = 0;
-                    $result[$id] = $module;
-               }
-
-               # Increment matches count
-               $sorter[$id] += $nb;
-               $result[$id]['accuracy'] = $sorter[$id];
+               $module['id'] = $id;
+
+               # Loop through required module fields
+               foreach(self::$weighting as $field => $weight) {
+
+                    # Skip fields which not exsist on module
+                    if (empty($module[$field])) {
+                         continue;
+                    }
+
+                    # Split field value into small clean word
+                    if (!($subjects = self::patternize($module[$field]))) {
+                         continue;
+                    }
+
+                    # Check contents
+                    if (!($nb = preg_match_all('/('.implode('|', $patterns).')/', implode(' ', $subjects), $_))) {
+                         continue;
+                    }
+
+                    # Add module to result
+                    if (!isset($sorter[$id])) {
+                         $sorter[$id] = 0;
+                         $result[$id] = $module;
+                    }
+
+                    # Increment score by matches count * field weight
+                    $sorter[$id] += $nb * $weight;
+                    $result[$id]['score'] = $sorter[$id];
+               }
           }
           # Sort response by matches count
@@ -217 +233 @@
 
      /**
-      * Sanitize string.
+      * Split and clean pattern.
       *
       * @param string    $str      String to sanitize
-      * @param null $_        Unused    param
-      */
-     public static function sanitize(&$str, $_)
-     {
-          $str = strtolower(preg_replace('/[^A-Za-z0-9]/', '', $str));
+      * @return     array     Array of cleaned pieces of string or false if none
+      */
+     public static function patternize($str)
+     {
+          $arr = array();
+
+          foreach(explode(' ', $str) as $_) {
+               $_ = strtolower(preg_replace('/[^A-Za-z0-9]/', '', $_));
+               if (strlen($_) > 2) {
+                    $arr[] = $_;
+               }
+          }
+
+          return empty($arr) ? false : $arr;
      }