move word recheck into token collection

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 9115be0b3e333c3c6f08fdbbcac97b254fe3b72d..f7f97593a8a3b01b2e999e97aafe622c93569063 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -669,7 +669,7 @@ class Geocode
         return $aSearchResults;
     }
 
-    public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured, $sNormQuery)
+    public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured)
     {
         /*
              Calculate all searches using aValidTokens i.e.
@@ -707,17 +707,8 @@ class Geocode
                         // If the token is valid
                         if (isset($aValidTokens[' '.$sToken])) {
                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
-                                // Recheck if the original word shows up in the query.
-                                $bWordInQuery = false;
-                                if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
-                                    $bWordInQuery = strpos(
-                                        $sNormQuery,
-                                        $this->normTerm($aSearchTerm['word'])
-                                    ) !== false;
-                                }
                                 $aNewSearches = $oCurrentSearch->extendWithFullTerm(
                                     $aSearchTerm,
-                                    $bWordInQuery,
                                     isset($aValidTokens[$sToken])
                                       && strpos($sToken, ' ') === false,
                                     $sPhraseType,
@@ -802,7 +793,7 @@ class Geocode
         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
         $aGroupedSearches = array();
         foreach ($aSearches as $oSearch) {
-            if (!$oSearch->isValidSearch($this->aCountryCodes)) {
+            if (!$oSearch->isValidSearch()) {
                 continue;
             }
 
@@ -991,13 +982,22 @@ class Geocode
                 );
                 $aWordFrequencyScores = array();
                 foreach ($aDatabaseWords as $aToken) {
-                    // Very special case - require 2 letter country param to match the country code found
-                    if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
-                        && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']
+                    // Filter country tokens that do not match restricted countries.
+                    if ($this->aCountryCodes
+                        && $aToken['country_code']
+                        && !in_array($aToken['country_code'], $this->aCountryCodes)
                     ) {
                         continue;
                     }
 
+                    // Special terms need to appear in their normalized form.
+                    if ($aToken['word'] && $aToken['class']) {
+                        $sNormWord = $this->normTerm($aToken['word']);
+                        if (strpos($sNormQuery, $sNormWord) === false) {
+                            continue;
+                        }
+                    }
+
                     if (isset($aValidTokens[$aToken['word_token']])) {
                         $aValidTokens[$aToken['word_token']][] = $aToken;
                     } else {
@@ -1034,7 +1034,7 @@ class Geocode
                 // Any words that have failed completely?
                 // TODO: suggestions
 
-                $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases, $sNormQuery);
+                $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
 
                 if ($this->bReverseInPlan) {
                     // Reverse phrase array and also reverse the order of the wordsets in
@@ -1045,7 +1045,7 @@ class Geocode
                     if (sizeof($aPhrases) > 1) {
                         $aPhrases[sizeof($aPhrases)-1]->invertWordSets();
                     }
-                    $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false, $sNormQuery);
+                    $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
 
                     foreach ($aGroupedSearches as $aSearches) {
                         foreach ($aSearches as $aSearch) {