Changeset 12925

Timestamp:

10/21/07 18:26:35 (1 year ago)

Author:

rambo

Message:

indexing/search multilang, first tests

Files:

• branches/MidCOM_2_8/midcom.core/midcom/services/indexer/backend/solr.php (modified) (9 diffs)
• branches/MidCOM_2_8/midcom.core/midcom/services/indexer/document.php (modified) (5 diffs)
• branches/MidCOM_2_8/midcom.core/midcom/services/indexer/document/midcom.php (modified) (2 diffs)
• branches/MidCOM_2_8/midcom.helper.search/viewer.php (modified) (2 diffs)

branches/MidCOM_2_8/midcom.core/midcom/services/indexer/backend/solr.php

@@ -108 +108 @@
-    function query ($query, $filter)
-    {
+        debug_push_class(__CLASS__, __FUNCTION__);
-        if ($filter !== null) 
-        {
@@ -119 +120 @@
-            }
-        }
+        /* In fact this is probably best left for midcom.helper.search to decide
+        if ($GLOBALS['midcom_config']['i18n_multilang_strict'])
+        {
+            $query .= ' AND (__LANG:"' . $_MIDCOM->i18n->get_current_language() . '" OR __LANG:"")';
+        }
+        */
-
-        $url = "http://" . $GLOBALS['midcom_config']['indexer_xmltcp_host'] . 
@@ -136 +143 @@
-            $msg = (is_object($err)) ? $err->getMessage() : "";
-            debug_add("Failed to execute Request {$url}:{$this->code} {$msg}", MIDCOM_LOG_WARN); 
+            debug_pop();
-            return false;
-        }
-
-
+
+
+
-        $xquery = new DomXPath($response);
-        $result = array();
@@ -148 +157 @@
-        }
-
- {
-    
+
+{
-            $doc = new midcom_services_indexer_document();
-            foreach ($res->childNodes as $str) {
@@ -163 +172 @@
-                
-            }
-
-
+
+
+
+
-        return $result;
-    }
@@ -283 +294 @@
-     * posts the xml to the suggested url using HTTP_Request.
-     * */
-
-
+
+
+
-        $options = array();
-        $options['method'] = HTTP_REQUEST_METHOD_POST ;
@@ -295 +307 @@
-        $this->request->addHeader('Accept-Charset', 'UTF-8');
-        $this->request->addHeader('Content-type', 'text/xml; charset=utf-8');
+        debug_add("POSTing XML to {$url}\n===\n{$xml}\n===\n");
-        $err = $this->request->sendRequest(true);
-
@@ -301 +314 @@
-        if ($this->code != 200 || PEAR :: isError($err)) {
-            debug_add("Failed to execute Request {$url}:{$this->code} {$err}", MIDCOM_LOG_WARN); 
-
+
+
-            return false;
-        }
@@ -312 +326 @@
-            debug_add("Failed to execute Request {$url}: {$err->getMessage()}", MIDCOM_LOG_WARN); 
-            debug_add("Request content: \n$xml", MIDCOM_LOG_INFO); 
+            debug_pop();
-            return false;
-        }
+        debug_add('POST ok');
+        debug_pop();
-        return true;
-

branches/MidCOM_2_8/midcom.core/midcom/services/indexer/document.php

@@ -88 +88 @@
-
-    /**
+     * Two letter language code of the document contnet
+     *
+     * This field is optional.
+     *
+     * @var string
+     */
+    var $lang = '';
+
+    /**
-     * The GUID of the topic the document is assigned to. May be empty for
-     * non-midgard resources.
@@ -444 +453 @@
-
-        // __RI does not need to be populated, this is done by backends.
+        $this->add_unindexed('__LANG', $this->lang);
-        $this->add_text('__TOPIC_GUID', $this->topic_guid);
-        $this->add_text('__COMPONENT', $this->component);
@@ -493 +503 @@
-    {
-        $this->RI = $this->get_field('__RI');
+        $this->lang = $this->get_field('__LANG');
-        $this->topic_guid = $this->get_field('__TOPIC_GUID');
-        $this->component = $this->get_field('__COMPONENT');
@@ -517 +528 @@
-        $this->type = $this->get_field('__TYPE');
-        $this->security = $this->get_field('__SECURITY');
+
+        // Strip language code from end of RI if it looks like "<GUID>_<LANG>" (because *many* places suppose it's plain GUID)
+        $this->RI = preg_replace('/^([0-9a-f]{32,80})_[a-z]{2}$/', '\\1', $this->RI);
-    }
-
@@ -568 +582 @@
-        debug_add("Editor: " . (! is_object($this->editor) ? 'none' : $this->editor->name), $loglevel);
-        debug_add("RI: {$this->RI}", $loglevel);
+        debug_add("Language: {$this->lang}", $loglevel);
-        debug_add("Score: {$this->score}", $loglevel);
-        debug_add("Source: {$this->source}", $loglevel);

branches/MidCOM_2_8/midcom.core/midcom/services/indexer/document/midcom.php

@@ -71 +71 @@
-        
-        $this->source = $this->_metadata->object->guid();
-
+
+
-        $this->document_url = $_MIDCOM->permalinks->create_permalink($this->source);
-        
@@ -77 +78 @@
-        $this->_process_topic();
-    }
-
+
+
+
+
+
+
+
+
+
-    
-    /**

branches/MidCOM_2_8/midcom.helper.search/viewer.php

@@ -97 +97 @@
-        
-        $data['type'] = $_REQUEST['type'];
+        $data['query'] = trim($_REQUEST['query']);
+        if (   $GLOBALS['midcom_config']['indexer_backend'] != 'solr'
+            && count(explode(' ', $data['query'])) == 1
+            && strpos($data['query'], '*') === false)
+        {
+            $data['query'] .= '*';
+        }
+
-        
-        switch ($data['type'])
-        {
-            case 'basic':
-                $data['query'] = trim($_REQUEST['query']);
-                
-                                if (   count(explode(' ', $data['query'])) == 1
-                                    && !strstr($data['query'], '*'))
-                                {
-                                    // Single search term, append *
-                    if ($GLOBALS['midcom_config']['indexer_backend'] != 'solr') 
-                    {
-                                        $data['query'] .= '*';
-                    }
-                                }
-                                
-
+
+
+
+
+
+
-                break;
-            
-            case 'advanced':
-
-
-
-
-
-
-
-
-
+
-                                $data['request_topic'] = trim($_REQUEST['topic']);
-                $data['component'] = trim($_REQUEST['component']);
@@ -163 +157 @@
-                    }
-                    $final_query .= "__COMPONENT:{$data['component']}";
+                }
+
+                if ($GLOBALS['midcom_config']['i18n_multilang_strict'])
+                {
+                    $final_query .= ' AND (__LANG:"' . $_MIDCOM->i18n->get_current_language() . '" OR __LANG:"")';
-                }
-                debug_add("Final query: {$final_query}");