Merge pull request #2784 from lonvia/doscs-customizing-icu-tokenizer

[nominatim.git] / lib-php / Geocode.php

diff --git a/lib-php/Geocode.php b/lib-php/Geocode.php
index c2b4f4e4c0704e035f37518b784c251bc62a8c02..9e3b0b92fbc47638dd4ff4227a45b9617a9c2a6c 100644 (file)
--- a/lib-php/Geocode.php
+++ b/lib-php/Geocode.php
@@ -1,4 +1,12 @@
 <?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.
+ */
 
 namespace Nominatim;
 
@@ -182,7 +190,7 @@ class Geocode
 
         $this->bFallback = $oParams->getBool('fallback', $this->bFallback);
 
-        // List of excluded Place IDs - used for more acurate pageing
+        // List of excluded Place IDs - used for more accurate pageing
         $sExcluded = $oParams->getStringList('exclude_place_ids');
         if ($sExcluded) {
             foreach ($sExcluded as $iExcludedPlaceID) {
@@ -362,15 +370,16 @@ class Geocode
 
                     foreach ($aWordsetSearches as $oCurrentSearch) {
                         foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
-                            $aNewSearches = $oCurrentSearch->extendWithSearchTerm(
-                                $sToken,
-                                $oSearchTerm,
-                                $oPosition
-                            );
-
-                            foreach ($aNewSearches as $oSearch) {
-                                if ($oSearch->getRank() < $this->iMaxRank) {
-                                    $aNewWordsetSearches[] = $oSearch;
+                            if ($oSearchTerm->isExtendable($oCurrentSearch, $oPosition)) {
+                                $aNewSearches = $oSearchTerm->extendSearch(
+                                    $oCurrentSearch,
+                                    $oPosition
+                                );
+
+                                foreach ($aNewSearches as $oSearch) {
+                                    if ($oSearch->getRank() < $this->iMaxRank) {
+                                        $aNewWordsetSearches[] = $oSearch;
+                                    }
                                 }
                             }
                         }
@@ -497,7 +506,6 @@ class Geocode
         if ($this->aCountryCodes) {
             $oCtx->setCountryList($this->aCountryCodes);
         }
-        $this->oTokenizer->setCountryRestriction($this->aCountryCodes);
 
         Debug::newSection('Query Preprocessing');
 
@@ -506,13 +514,6 @@ class Geocode
             userError('Query string is not UTF-8 encoded.');
         }
 
-        // Conflicts between US state abreviations and various words for 'the' in different languages
-        if (isset($this->aLangPrefOrder['name:en'])) {
-            $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
-            $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
-            $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);
-        }
-
         // Do we have anything that looks like a lat/lon pair?
         $sQuery = $oCtx->setNearPointFromQuery($sQuery);
 
@@ -554,15 +555,15 @@ class Geocode
 
                 if (!empty($aTokens)) {
                     $aNewSearches = array();
+                    $oPosition = new SearchPosition('', 0, 1);
+                    $oPosition->setTokenPosition(0, 1);
+
                     foreach ($aSearches as $oSearch) {
                         foreach ($aTokens as $oToken) {
-                            $oNewSearch = clone $oSearch;
-                            $oNewSearch->setPoiSearch(
-                                $oToken->iOperator,
-                                $oToken->sClass,
-                                $oToken->sType
+                            $aNewSearches = array_merge(
+                                $aNewSearches,
+                                $oToken->extendSearch($oSearch, $oPosition)
                             );
-                            $aNewSearches[] = $oNewSearch;
                         }
                     }
                     $aSearches = $aNewSearches;
@@ -616,16 +617,15 @@ class Geocode
                     }
                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);
 
-                    foreach ($aGroupedSearches as $aSearches) {
+                    foreach ($aReverseGroupedSearches as $aSearches) {
                         foreach ($aSearches as $aSearch) {
-                            if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
-                                $aReverseGroupedSearches[$aSearch->getRank()] = array();
+                            if (!isset($aGroupedSearches[$aSearch->getRank()])) {
+                                $aGroupedSearches[$aSearch->getRank()] = array();
                             }
-                            $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
+                            $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
                         }
                     }
 
-                    $aGroupedSearches = $aReverseGroupedSearches;
                     ksort($aGroupedSearches);
                 }
             } else {
@@ -843,7 +843,9 @@ class Geocode
                 $aResult['importance'] = 0.001;
                 $aResult['foundorder'] = $aResult['addressimportance'];
             } else {
-                $aResult['importance'] = max(0.001, $aResult['importance']);
+                if ($aResult['importance'] == 0) {
+                    $aResult['importance'] = 0.0001;
+                }
                 $aResult['importance'] *= $this->viewboxImportanceFactor(
                     $aResult['lon'],
                     $aResult['lat']