Change to regular regex group

[nominatim.git] / lib-php / admin / warm.php

diff --git a/lib-php/admin/warm.php b/lib-php/admin/warm.php
index d7950af9b1912b91d70c2f29b691c06a328fa15c..5cbae89846f52bc9034f9c36d0ed352260b98fd3 100644 (file)
--- a/lib-php/admin/warm.php
+++ b/lib-php/admin/warm.php
@@ -1,9 +1,16 @@
 <?php
 <?php

+/**
+ * SPDX-License-Identifier: GPL-2.0-only
+ *
+ * This file is part of Nominatim. (https://nominatim.org)
+ *
+ * Copyright (C) 2022 by the Nominatim developer community.
+ * For a full list of authors see the git log.
+ */

 @define('CONST_LibDir', dirname(dirname(__FILE__)));
 
 require_once(CONST_LibDir.'/init-cmd.php');
 require_once(CONST_LibDir.'/log.php');
 @define('CONST_LibDir', dirname(dirname(__FILE__)));
 
 require_once(CONST_LibDir.'/init-cmd.php');
 require_once(CONST_LibDir.'/log.php');

-require_once(CONST_LibDir.'/Geocode.php');

 require_once(CONST_LibDir.'/PlaceLookup.php');
 require_once(CONST_LibDir.'/ReverseGeocode.php');
 
 require_once(CONST_LibDir.'/PlaceLookup.php');
 require_once(CONST_LibDir.'/ReverseGeocode.php');
 


@@ -26,16 +33,16 @@ loadSettings($aCMDResult['project-dir'] ?? getcwd());
 @define('CONST_Default_Language', getSetting('DEFAULT_LANGUAGE', false));
 @define('CONST_Log_DB', getSettingBool('LOG_DB'));
 @define('CONST_Log_File', getSetting('LOG_FILE', false));
 @define('CONST_Default_Language', getSetting('DEFAULT_LANGUAGE', false));
 @define('CONST_Log_DB', getSettingBool('LOG_DB'));
 @define('CONST_Log_File', getSetting('LOG_FILE', false));

-@define('CONST_Max_Word_Frequency', getSetting('MAX_WORD_FREQUENCY'));

 @define('CONST_NoAccessControl', getSettingBool('CORS_NOACCESSCONTROL'));
 @define('CONST_Places_Max_ID_count', getSetting('LOOKUP_MAX_COUNT'));
 @define('CONST_PolygonOutput_MaximumTypes', getSetting('POLYGON_OUTPUT_MAX_TYPES'));
 @define('CONST_Search_BatchMode', getSettingBool('SEARCH_BATCH_MODE'));
 @define('CONST_Search_NameOnlySearchFrequencyThreshold', getSetting('SEARCH_NAME_ONLY_THRESHOLD'));
 @define('CONST_NoAccessControl', getSettingBool('CORS_NOACCESSCONTROL'));
 @define('CONST_Places_Max_ID_count', getSetting('LOOKUP_MAX_COUNT'));
 @define('CONST_PolygonOutput_MaximumTypes', getSetting('POLYGON_OUTPUT_MAX_TYPES'));
 @define('CONST_Search_BatchMode', getSettingBool('SEARCH_BATCH_MODE'));
 @define('CONST_Search_NameOnlySearchFrequencyThreshold', getSetting('SEARCH_NAME_ONLY_THRESHOLD'));

-@define('CONST_Term_Normalization_Rules', getSetting('TERM_NORMALIZATION'));

 @define('CONST_Use_US_Tiger_Data', getSettingBool('USE_US_TIGER_DATA'));
 @define('CONST_MapIcon_URL', getSetting('MAPICON_URL', false));
 @define('CONST_Use_US_Tiger_Data', getSettingBool('USE_US_TIGER_DATA'));
 @define('CONST_MapIcon_URL', getSetting('MAPICON_URL', false));

+@define('CONST_TokenizerDir', CONST_InstallDir.'/tokenizer');

 
 

+require_once(CONST_LibDir.'/Geocode.php');

 
 $oDB = new Nominatim\DB();
 $oDB->connect();
 
 $oDB = new Nominatim\DB();
 $oDB->connect();


@@ -63,11 +70,15 @@ if (!$aResult['search-only']) {
     $oPlaceLookup->setLanguagePreference(array('en'));
 
     echo 'Warm reverse: ';
     $oPlaceLookup->setLanguagePreference(array('en'));
 
     echo 'Warm reverse: ';

-    if ($bVerbose) echo "\n";
+    if ($bVerbose) {
+        echo "\n";
+    }

     for ($i = 0; $i < 1000; $i++) {
         $fLat = rand(-9000, 9000) / 100;
         $fLon = rand(-18000, 18000) / 100;
     for ($i = 0; $i < 1000; $i++) {
         $fLat = rand(-9000, 9000) / 100;
         $fLon = rand(-18000, 18000) / 100;

-        if ($bVerbose) echo "$fLat, $fLon = ";
+        if ($bVerbose) {
+            echo "$fLat, $fLon = ";
+        }

 
         $oLookup = $oReverseGeocode->lookup($fLat, $fLon);
         $aSearchResults = $oLookup ? $oPlaceLookup->lookup(array($oLookup->iId => $oLookup)) : null;
 
         $oLookup = $oReverseGeocode->lookup($fLat, $fLon);
         $aSearchResults = $oLookup ? $oPlaceLookup->lookup(array($oLookup->iId => $oLookup)) : null;


@@ -80,10 +91,19 @@ if (!$aResult['reverse-only']) {
     $oGeocode = new Nominatim\Geocode($oDB);
 
     echo 'Warm search: ';
     $oGeocode = new Nominatim\Geocode($oDB);
 
     echo 'Warm search: ';

-    if ($bVerbose) echo "\n";
+    if ($bVerbose) {
+        echo "\n";
+    }
+
+    $oTokenizer = new \Nominatim\Tokenizer($oDB);
+
+    $aWords = $oTokenizer->mostFrequentWords(1000);
+

     $sSQL = 'SELECT word FROM word WHERE word is not null ORDER BY search_name_count DESC LIMIT 1000';
     $sSQL = 'SELECT word FROM word WHERE word is not null ORDER BY search_name_count DESC LIMIT 1000';

-    foreach ($oDB->getCol($sSQL) as $sWord) {
-        if ($bVerbose) echo "$sWord = ";
+    foreach ($aWords as $sWord) {
+        if ($bVerbose) {
+            echo "$sWord = ";
+        }

 
         $oGeocode->setLanguagePreference(array('en'));
         $oGeocode->setQuery($sWord);
 
         $oGeocode->setLanguagePreference(array('en'));
         $oGeocode->setQuery($sWord);