From 969a49426f0c6716076915fc301020170d1631fa Mon Sep 17 00:00:00 2001
From: John Klehm <xixsimplicityxix@gmail.com>
Date: Mon, 7 Apr 2008 15:25:47 -0500
Subject: runIndexer.php: Script that indexes existing database

---
 runIndexer.php |  144 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 144 insertions(+), 0 deletions(-)
 create mode 100644 runIndexer.php

diff --git a/runIndexer.php b/runIndexer.php
new file mode 100644
index 0000000..aefec27
--- /dev/null
+++ b/runIndexer.php
@@ -0,0 +1,144 @@
+<?php
+require_once('./path.php');
+require_once(BASE.'include/incl.php');
+require_once(BASE.'include/util.php');
+require_once(BASE.'include/nice_urls.php');
+$out = '';
+
+function createInsertString($name, $oRow, $rank)
+{
+    $cleanAppName = query_escape_string($oRow->appName);
+    return "('" . md5($name) . "',"     .
+           "$oRow->appId,"              .
+           SEARCH_INDEX_TYPE_APP . ","  .
+           "'$cleanAppName',"           .
+           "$rank),"; 
+}
+
+function hashFrontToBack($name, $nameLen, $oRow)
+{
+    $tempIndexValues = "";
+    for($i = 0; $i < $nameLen;)
+    {
+        #skip to separator
+        while (($name[$i++] != '-') && ($i < $nameLen));
+        if ($i < $nameLen)
+        {
+            $partName = substr($name, $i,$nameLen-$i);
+            $out .= "<p>Part name: $partName</p>\n";
+
+            $tempIndexValues .= createInsertString($partName, $oRow, 0);
+        }
+    }
+
+    return $tempIndexValues;
+}
+
+function hashBackToFront($name, $nameLen, $oRow)
+{
+    $tempIndexValues = "";
+    for($i = $nameLen; $i > 0;)
+    {
+        #skip to separator
+        while (($name[--$i] != '-') && ($i > 0));
+        if ($i > 0)
+        {
+            $partName = substr($name, 0, $i);
+            $out .= "<p>Part name: $partName</p>\n";
+
+            $tempIndexValues .= createInsertString($partName, $oRow, 0);
+        }
+    }
+
+    return $tempIndexValues;
+}
+
+function appFamilyQuery($idStart,$idEnd)
+{
+    return 'SELECT appId, appName, keywords FROM appFamily WHERE state = \'accepted\' AND appId >= ' .
+                  $idStart . ' AND appId < ' . $idEnd;
+}
+
+
+
+#1000 inserts at a time works
+#$vendorQuery  = 'SELECT vendorId, vendorName FROM vendor WHERE vendorId >= ' . 
+#                $queryId . ' AND vendorId < ';
+#$appVersionQuery = 'SELECT versionId, versionName FROM appVersion WHERE versionId >= ' .
+#                   $queryId. ' AND versionId < ';
+#$vendorQuery .= $queryId;
+#$appVersionQuery .= $queryId;
+
+$queryId = 0;
+$maxQueryId = 0;
+if ($oMax = query_fetch_object(query_appdb('SELECT MAX(appId) AS max from appFamily',"Find highest app id")))
+{
+    $maxQueryId = $oMax->max;
+}
+else
+{
+    $out .= "Could not find max id from table.";
+}
+
+while ($result = query_appdb(appFamilyQuery($queryId,($queryId += APPDB_MYSQL_MAX_INSERTS_PER_QUERY)),""))
+{
+    $indexValues = "";
+    while ($oRow = query_fetch_object($result))
+    {
+        $out .= "<p>======================================</p>\n";
+        $out .= "<p>Name: $oRow->appName</p>\n";
+
+        # store the exact match
+        $goodName = nice_url_name($oRow->appName);
+        $out .= "<p>Good Name: $goodName</p>\n";
+
+        // direct hit on name should have highest ranked hash
+        $indexValues .= createInsertString($goodName, $oRow, APPDB_SEARCH_INDEX_MAX_RANK);
+        
+        # chop up the exact matching teram and hash the pieces
+        $gNameLen = strlen($goodName);
+
+        # drop terms from front to back, cool-edit-pro => edit-pro => pro
+        $indexValues .= hashFrontToBack($goodName, $gNameLen, $oRow);
+
+        # drop terms from back to front, cool-edit-pro -> cool-edit => cool
+        $indexValues .= hashBackToFront($goodName, $gNameLen, $oRow);
+
+        # TODO:  hash mixed terms, keywords??
+    }
+
+    if ($indexValues != "")
+    {
+        $indexValLastCharPos = strlen($indexValues) - 1;
+        if ($indexValues[$indexValLastCharPos] == ',')
+        {
+            $indexValues = substr($indexValues, 0, $indexValLastCharPos);
+        }
+
+        $out .= "<p>index values: $indexValues</p>\n";
+        if (APPDB_ENABLE_NICE_URLS_INDEXER)
+        {
+            if(query_appdb("INSERT INTO searchIndex (hash, objectId, type, name, rank)".
+                           " VALUES $indexValues", "Index data for appdb"))
+            {
+                $out .= "<p>Indexer Query Success</p>\n";
+            }
+            else
+            {
+                $out .= "<p>Indexer Query Fail</p>\n";   
+            }
+        }
+        else
+        {
+            $out .= "<p>Indexer off.</p>\n";
+        }
+    }
+    
+    if ($queryId > $maxQueryId)
+    {
+        break;
+    }
+}
+
+nice_urls_display($out, "AppDB: Nice URLs Indexer");
+?>
-- 
1.5.4.5