Merge remote-tracking branch 'upstream/master'

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

diff --git a/lib-php/Geocode.php b/lib-php/Geocode.php
index 14a3315e87931c3e11d304e6f109d29e546c8840..85d871f0d039f1b37072901f1ae09ae0650d9c64 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;
 
@@ -7,6 +15,7 @@ require_once(CONST_LibDir.'/Phrase.php');
 require_once(CONST_LibDir.'/ReverseGeocode.php');
 require_once(CONST_LibDir.'/SearchDescription.php');
 require_once(CONST_LibDir.'/SearchContext.php');
+require_once(CONST_LibDir.'/SearchPosition.php');
 require_once(CONST_LibDir.'/TokenList.php');
 require_once(CONST_TokenizerDir.'/tokenizer.php');
 
@@ -345,50 +354,26 @@ class Geocode
          */
         foreach ($aPhrases as $iPhrase => $oPhrase) {
             $aNewPhraseSearches = array();
-            $sPhraseType = $oPhrase->getPhraseType();
+            $oPosition = new SearchPosition(
+                $oPhrase->getPhraseType(),
+                $iPhrase,
+                count($aPhrases)
+            );
 
             foreach ($oPhrase->getWordSets() as $aWordset) {
                 $aWordsetSearches = $aSearches;
 
                 // Add all words from this wordset
                 foreach ($aWordset as $iToken => $sToken) {
-                    //echo "<br><b>$sToken</b>";
                     $aNewWordsetSearches = array();
+                    $oPosition->setTokenPosition($iToken, count($aWordset));
 
                     foreach ($aWordsetSearches as $oCurrentSearch) {
-                        //echo "<i>";
-                        //var_dump($oCurrentSearch);
-                        //echo "</i>";
-
-                        // Tokens with full name matches.
-                        foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) {
-                            $aNewSearches = $oCurrentSearch->extendWithFullTerm(
-                                $oSearchTerm,
-                                $sPhraseType,
-                                $iToken == 0 && $iPhrase == 0,
-                                $iPhrase == 0,
-                                $iToken + 1 == count($aWordset)
-                                  && $iPhrase + 1 == count($aPhrases)
-                            );
-
-                            foreach ($aNewSearches as $oSearch) {
-                                if ($oSearch->getRank() < $this->iMaxRank) {
-                                    $aNewWordsetSearches[] = $oSearch;
-                                }
-                            }
-                        }
-                        // Look for partial matches.
-                        // Note that there is no point in adding country terms here
-                        // because country is omitted in the address.
-                        if ($sPhraseType != 'country') {
-                            // Allow searching for a word - but at extra cost
-                            foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
-                                $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
-                                    $sToken,
-                                    $oSearchTerm,
-                                    (bool) $sPhraseType,
-                                    $iPhrase,
-                                    $oValidTokens->get(' '.$sToken)
+                        foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
+                            if ($oSearchTerm->isExtendable($oCurrentSearch, $oPosition)) {
+                                $aNewSearches = $oSearchTerm->extendSearch(
+                                    $oCurrentSearch,
+                                    $oPosition
                                 );
 
                                 foreach ($aNewSearches as $oSearch) {
@@ -403,7 +388,6 @@ class Geocode
                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
                 }
-                //var_Dump('<hr>',count($aWordsetSearches)); exit;
 
                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
@@ -522,7 +506,6 @@ class Geocode
         if ($this->aCountryCodes) {
             $oCtx->setCountryList($this->aCountryCodes);
         }
-        $this->oTokenizer->setCountryRestriction($this->aCountryCodes);
 
         Debug::newSection('Query Preprocessing');
 
@@ -531,13 +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);
 
@@ -579,15 +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;
@@ -641,16 +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 {
@@ -720,7 +695,7 @@ class Geocode
                         }
                     }
 
-                    if ($iQueryLoop > 20) {
+                    if ($iQueryLoop > 30) {
                         break;
                     }
                 }
@@ -797,7 +772,7 @@ class Geocode
                     $aResults = $tempIDs;
                 }
 
-                if (!empty($aResults) || $iGroupLoop > 4 || $iQueryLoop > 30) {
+                if (!empty($aResults) || $iGroupLoop > 6 || $iQueryLoop > 40) {
                     break;
                 }
             }
@@ -868,7 +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']