introduce name analyzer

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

diff --git a/lib-php/Geocode.php b/lib-php/Geocode.php
index 12f9da37550f856e47a128641a275813d9c4e9de..6cec6a85e088ee40c69917eb152e54bf571fc1a7 100644 (file)
--- a/lib-php/Geocode.php
+++ b/lib-php/Geocode.php
@@ -18,7 +18,6 @@ class Geocode
     protected $aLangPrefOrder = array();
 
     protected $aExcludePlaceIDs = array();
     protected $aLangPrefOrder = array();
 
     protected $aExcludePlaceIDs = array();

-    protected $bReverseInPlan = true;

 
     protected $iLimit = 20;
     protected $iFinalLimit = 10;
 
     protected $iLimit = 20;
     protected $iFinalLimit = 10;


@@ -61,11 +60,6 @@ class Geocode
         return $this->oNormalizer->transliterate($sTerm);
     }
 
         return $this->oNormalizer->transliterate($sTerm);
     }
 

-    public function setReverseInPlan($bReverse)
-    {
-        $this->bReverseInPlan = $bReverse;
-    }
-

     public function setLanguagePreference($aLangPref)
     {
         $this->aLangPrefOrder = $aLangPref;
     public function setLanguagePreference($aLangPref)
     {
         $this->aLangPrefOrder = $aLangPref;


@@ -262,7 +256,6 @@ class Geocode
                 $oParams->getString('country'),
                 $oParams->getString('postalcode')
             );
                 $oParams->getString('country'),
                 $oParams->getString('postalcode')
             );

-            $this->setReverseInPlan(false);

         } else {
             $this->setQuery($sQuery);
         }
         } else {
             $this->setQuery($sQuery);
         }


@@ -330,7 +323,7 @@ class Geocode
         return false;
     }
 
         return false;
     }
 

-    public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
+    public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens)

     {
         /*
              Calculate all searches using oValidTokens i.e.
     {
         /*
              Calculate all searches using oValidTokens i.e.


@@ -345,7 +338,7 @@ class Geocode
          */
         foreach ($aPhrases as $iPhrase => $oPhrase) {
             $aNewPhraseSearches = array();
          */
         foreach ($aPhrases as $iPhrase => $oPhrase) {
             $aNewPhraseSearches = array();

-            $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
+            $sPhraseType = $oPhrase->getPhraseType();

 
             foreach ($oPhrase->getWordSets() as $aWordset) {
                 $aWordsetSearches = $aSearches;
 
             foreach ($oPhrase->getWordSets() as $aWordset) {
                 $aWordsetSearches = $aSearches;


@@ -388,7 +381,7 @@ class Geocode
                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
                                     $sToken,
                                     $oSearchTerm,
                                 $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
                                     $sToken,
                                     $oSearchTerm,

-                                    $bIsStructured,
+                                    (bool) $sPhraseType,

                                     $iPhrase,
                                     $oValidTokens->get(' '.$sToken)
                                 );
                                     $iPhrase,
                                     $oValidTokens->get(' '.$sToken)
                                 );


@@ -607,10 +600,8 @@ class Geocode
             // Commas are used to reduce the search space by indicating where phrases split
             if ($this->aStructuredQuery) {
                 $aInPhrases = $this->aStructuredQuery;
             // Commas are used to reduce the search space by indicating where phrases split
             if ($this->aStructuredQuery) {
                 $aInPhrases = $this->aStructuredQuery;

-                $bStructuredPhrases = true;

             } else {
                 $aInPhrases = explode(',', $sQuery);
             } else {
                 $aInPhrases = explode(',', $sQuery);

-                $bStructuredPhrases = false;

             }
 
             Debug::printDebugArray('Search context', $oCtx);
             }
 
             Debug::printDebugArray('Search context', $oCtx);


@@ -684,9 +675,9 @@ class Geocode
 
                 Debug::newSection('Search candidates');
 
 
                 Debug::newSection('Search candidates');
 

-                $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases);
+                $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);

 
 

-                if ($this->bReverseInPlan) {
+                if (!$this->aStructuredQuery) {

                     // Reverse phrase array and also reverse the order of the wordsets in
                     // the first and final phrase. Don't bother about phrases in the middle
                     // because order in the address doesn't matter.
                     // Reverse phrase array and also reverse the order of the wordsets in
                     // the first and final phrase. Don't bother about phrases in the middle
                     // because order in the address doesn't matter.


@@ -695,7 +686,7 @@ class Geocode
                     if (count($aPhrases) > 1) {
                         $aPhrases[count($aPhrases)-1]->invertWordSets();
                     }
                     if (count($aPhrases) > 1) {
                         $aPhrases[count($aPhrases)-1]->invertWordSets();
                     }

-                    $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
+                    $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens);

 
                     foreach ($aGroupedSearches as $aSearches) {
                         foreach ($aSearches as $aSearch) {
 
                     foreach ($aGroupedSearches as $aSearches) {
                         foreach ($aSearches as $aSearch) {


@@ -778,14 +769,19 @@ class Geocode
                 if (!empty($aResults)) {
                     $aSplitResults = Result::splitResults($aResults);
                     Debug::printVar('Split results', $aSplitResults);
                 if (!empty($aResults)) {
                     $aSplitResults = Result::splitResults($aResults);
                     Debug::printVar('Split results', $aSplitResults);

-                    if ($iGroupLoop <= 4 && empty($aSplitResults['tail'])
-                        && reset($aSplitResults['head'])->iResultRank > 0) {
+                    if ($iGroupLoop <= 4
+                        && reset($aSplitResults['head'])->iResultRank > 0
+                        && $iGroupedRank !== array_key_last($aGroupedSearches)) {

                         // Haven't found an exact match for the query yet.
                         // Therefore add result from the next group level.
                         $aNextResults = $aSplitResults['head'];
                         foreach ($aNextResults as $oRes) {
                             $oRes->iResultRank--;
                         }
                         // Haven't found an exact match for the query yet.
                         // Therefore add result from the next group level.
                         $aNextResults = $aSplitResults['head'];
                         foreach ($aNextResults as $oRes) {
                             $oRes->iResultRank--;
                         }

+                        foreach ($aSplitResults['tail'] as $oRes) {
+                            $oRes->iResultRank--;
+                            $aNextResults[$oRes->iId] = $oRes;
+                        }

                         $aResults = array();
                     } else {
                         $aResults = $aSplitResults['head'];
                         $aResults = array();
                     } else {
                         $aResults = $aSplitResults['head'];


@@ -833,7 +829,6 @@ class Geocode
                     foreach ($aResults as $oResult) {
                         if (($this->iMaxAddressRank == 30 &&
                              ($oResult->iTable == Result::TABLE_OSMLINE
                     foreach ($aResults as $oResult) {
                         if (($this->iMaxAddressRank == 30 &&
                              ($oResult->iTable == Result::TABLE_OSMLINE

-                              || $oResult->iTable == Result::TABLE_AUX

                               || $oResult->iTable == Result::TABLE_TIGER))
                             || in_array($oResult->iId, $aFilteredIDs)
                         ) {
                               || $oResult->iTable == Result::TABLE_TIGER))
                             || in_array($oResult->iId, $aFilteredIDs)
                         ) {


@@ -994,7 +989,6 @@ class Geocode
                 'Structured query' => $this->aStructuredQuery,
                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
                 'Structured query' => $this->aStructuredQuery,
                 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
                 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),

-                'Try reversed query'=> $this->bReverseInPlan,

                 'Limit (for searches)' => $this->iLimit,
                 'Limit (for results)'=> $this->iFinalLimit,
                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
                 'Limit (for searches)' => $this->iLimit,
                 'Limit (for results)'=> $this->iFinalLimit,
                 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),