fix more html syntax errors in ban page

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index ec8eb3489e6daaa15f11ba3d6aa0029284abb114..b0f0c764ca50eabb387de7bbd36b124be4f08e41 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -25,7 +25,7 @@ class Geocode
 
     protected $aExcludePlaceIDs = array();
     protected $bDeDupe = true;
-    protected $bReverseInPlan = false;
+    protected $bReverseInPlan = true;
 
     protected $iLimit = 20;
     protected $iFinalLimit = 10;
@@ -653,7 +653,7 @@ class Geocode
         return $aSearchResults;
     }
 
-    public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases)
+    public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery)
     {
         /*
              Calculate all searches using aValidTokens i.e.
@@ -752,13 +752,19 @@ class Geocode
                                          */
                                     }
                                 } elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) {
-                                    if ($aSearch['sClass'] === '') {
-                                        $aSearch['sOperator'] = $aSearchTerm['operator'];
+                                    // require a normalized exact match of the term
+                                    // if we have the normalizer version of the query
+                                    // available
+                                    if ($aSearch['sClass'] === ''
+                                        && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) {
                                         $aSearch['sClass'] = $aSearchTerm['class'];
                                         $aSearch['sType'] = $aSearchTerm['type'];
-                                        if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name';
-                                        else $aSearch['sOperator'] = 'near'; // near = in for the moment
-                                        if (strlen($aSearchTerm['operator']) == 0) $aSearch['iSearchRank'] += 1;
+                                        if ($aSearchTerm['operator'] == '') {
+                                            $aSearch['sOperator'] = sizeof($aSearch['aName']) ? 'name' :  'near';
+                                            $aSearch['iSearchRank'] += 2;
+                                        } else {
+                                            $aSearch['sOperator'] = 'near'; // near = in for the moment
+                                        }
 
                                         if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                     }
@@ -913,6 +919,13 @@ class Geocode
     {
         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 
+        $oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
+        if ($oNormalizer !== null) {
+            $sNormQuery = $oNormalizer->transliterate($this->sQuery);
+        } else {
+            $sNormQuery = null;
+        }
+
         $sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]";
         $sCountryCodesSQL = false;
         if ($this->aCountryCodes) {
@@ -1139,7 +1152,7 @@ class Geocode
                 // array with: placeid => -1 | tiger-housenumber
                 $aResultPlaceIDs = array();
 
-                $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
+                $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery);
 
                 if ($this->bReverseInPlan) {
                     // Reverse phrase array and also reverse the order of the wordsets in
@@ -1151,7 +1164,7 @@ class Geocode
                         $aFinalPhrase = end($aPhrases);
                         $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
                     }
-                    $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
+                    $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery);
 
                     foreach ($aGroupedSearches as $aSearches) {
                         foreach ($aSearches as $aSearch) {
@@ -1334,19 +1347,20 @@ class Geocode
                         // TODO: filter out the pointless search terms (2 letter name tokens and less)
                         // they might be right - but they are just too darned expensive to run
                         if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'], ",")."]";
-                        if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'], ",")."]";
+                        //if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'], ",")."]";
                         if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) {
                             // For infrequent name terms disable index usage for address
                             if (CONST_Search_NameOnlySearchFrequencyThreshold
                                 && sizeof($aSearch['aName']) == 1
                                 && $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold
                             ) {
-                                $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'], $aSearch['aAddressNonSearch']), ",")."]";
+                                //$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'], $aSearch['aAddressNonSearch']), ",")."]";
+                                $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]";
                             } else {
                                 $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'], ",")."]";
-                                if (sizeof($aSearch['aAddressNonSearch'])) {
+                                /*if (sizeof($aSearch['aAddressNonSearch'])) {
                                     $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'], ",")."]";
-                                }
+                                }*/
                             }
                         }
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
@@ -1542,8 +1556,9 @@ class Geocode
                             }
 
                             if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') { // & in
+                                $sClassTable = 'place_classtype_'.$aSearch['sClass'].'_'.$aSearch['sType'];
                                 $sSQL = "SELECT count(*) FROM pg_tables ";
-                                $sSQL .= "WHERE tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
+                                $sSQL .= "WHERE tablename = '$sClassTable'";
                                 $bCacheTable = chksql($this->oDB->getOne($sSQL));
 
                                 $sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
@@ -1591,7 +1606,8 @@ class Geocode
                                             $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
                                         }
 
-                                        $sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
+                                        $sSQL = "select distinct i.place_id".($sOrderBySQL?', i.order_term':'')." from (";
+                                        $sSQL .= "select l.place_id".($sOrderBySQL?','.$sOrderBySQL.' as order_term':'')." from ".$sClassTable." as l";
                                         if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
                                         if ($sPlaceIDs) {
                                             $sSQL .= ",placex as f where ";
@@ -1605,7 +1621,8 @@ class Geocode
                                             $sSQL .= " and l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                         }
                                         if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
-                                        if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
+                                        $sSQL .= 'limit 300) i ';
+                                        if ($sOrderBySQL) $sSQL .= "order by order_term asc";
                                         if ($this->iOffset) $sSQL .= " offset $this->iOffset";
                                         $sSQL .= " limit $this->iLimit";
                                         if (CONST_Debug) var_dump($sSQL);