Merge pull request #2373 from lonvia/tweak-search-cost

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

diff --git a/lib-php/SearchDescription.php b/lib-php/SearchDescription.php
index 2053082f6bbff4a13bd6a5ba6c08bc89366e0c38..3c572f2fc872e1e41a47580b80d156834870b964 100644 (file)
--- a/lib-php/SearchDescription.php
+++ b/lib-php/SearchDescription.php
@@ -153,8 +153,6 @@ class SearchDescription
      * Derive new searches by adding a full term to the existing search.
      *
      * @param object $oSearchTerm  Description of the token.
      * Derive new searches by adding a full term to the existing search.
      *
      * @param object $oSearchTerm  Description of the token.

-     * @param bool   $bHasPartial  True if there are also tokens of partial terms
-     *                             with the same name.

      * @param string $sPhraseType  Type of phrase the token is contained in.
      * @param bool   $bFirstToken  True if the token is at the beginning of the
      *                             query.
      * @param string $sPhraseType  Type of phrase the token is contained in.
      * @param bool   $bFirstToken  True if the token is at the beginning of the
      *                             query.


@@ -164,7 +162,7 @@ class SearchDescription
      *
      * @return SearchDescription[] List of derived search descriptions.
      */
      *
      * @return SearchDescription[] List of derived search descriptions.
      */

-    public function extendWithFullTerm($oSearchTerm, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken)
+    public function extendWithFullTerm($oSearchTerm, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken)

     {
         $aNewSearches = array();
 
     {
         $aNewSearches = array();
 


@@ -179,6 +177,7 @@ class SearchDescription
                 // - increase score for finding it anywhere else (optimisation)
                 if (!$bLastToken) {
                     $oSearch->iSearchRank += 5;
                 // - increase score for finding it anywhere else (optimisation)
                 if (!$bLastToken) {
                     $oSearch->iSearchRank += 5;

+                    $oSearch->iNamePhrase = -1;

                 }
                 $aNewSearches[] = $oSearch;
             }
                 }
                 $aNewSearches[] = $oSearch;
             }


@@ -205,6 +204,7 @@ class SearchDescription
                 ) {
                     $oSearch = clone $this;
                     $oSearch->iSearchRank++;
                 ) {
                     $oSearch = clone $this;
                     $oSearch->iSearchRank++;

+                    $oSearch->iNamePhrase = -1;

                     if (strlen($oSearchTerm->sPostcode) < 4) {
                         $oSearch->iSearchRank += 4 - strlen($oSearchTerm->sPostcode);
                     }
                     if (strlen($oSearchTerm->sPostcode) < 4) {
                         $oSearch->iSearchRank += 4 - strlen($oSearchTerm->sPostcode);
                     }


@@ -216,26 +216,33 @@ class SearchDescription
                  && is_a($oSearchTerm, '\Nominatim\Token\HouseNumber')
         ) {
             if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
                  && is_a($oSearchTerm, '\Nominatim\Token\HouseNumber')
         ) {
             if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {

-                $oSearch = clone $this;
-                $oSearch->iSearchRank++;
-                $oSearch->sHouseNumber = $oSearchTerm->sToken;

                 // sanity check: if the housenumber is not mainly made
                 // up of numbers, add a penalty
                 // sanity check: if the housenumber is not mainly made
                 // up of numbers, add a penalty

-                if (preg_match('/\\d/', $oSearch->sHouseNumber) === 0
-                    || preg_match_all('/[^0-9]/', $oSearch->sHouseNumber, $aMatches) > 2) {
-                    $oSearch->iSearchRank++;
+                $iSearchCost = 1;
+                if (preg_match('/\\d/', $oSearchTerm->sToken) === 0
+                    || preg_match_all('/[^0-9]/', $oSearchTerm->sToken, $aMatches) > 2) {
+                    $iSearchCost++;
+                }
+                if ($this->iOperator != Operator::NONE) {
+                    $iSearchCost++;

                 }
                 if (empty($oSearchTerm->iId)) {
                 }
                 if (empty($oSearchTerm->iId)) {

-                    $oSearch->iSearchRank++;
+                    $iSearchCost++;

                 }
                 // also must not appear in the middle of the address
                 if (!empty($this->aAddress)
                     || (!empty($this->aAddressNonSearch))
                     || $this->sPostcode
                 ) {
                 }
                 // also must not appear in the middle of the address
                 if (!empty($this->aAddress)
                     || (!empty($this->aAddressNonSearch))
                     || $this->sPostcode
                 ) {

-                    $oSearch->iSearchRank++;
+                    $iSearchCost++;

                 }
                 }

+
+                $oSearch = clone $this;
+                $oSearch->iSearchRank += $iSearchCost;
+                $oSearch->iNamePhrase = -1;
+                $oSearch->sHouseNumber = $oSearchTerm->sToken;

                 $aNewSearches[] = $oSearch;
                 $aNewSearches[] = $oSearch;

+

                 // Housenumbers may appear in the name when the place has its own
                 // address terms.
                 if ($oSearchTerm->iId !== null
                 // Housenumbers may appear in the name when the place has its own
                 // address terms.
                 if ($oSearchTerm->iId !== null


@@ -243,7 +250,7 @@ class SearchDescription
                     && empty($this->aAddress)
                    ) {
                     $oSearch = clone $this;
                     && empty($this->aAddress)
                    ) {
                     $oSearch = clone $this;

-                    $oSearch->iSearchRank++;
+                    $oSearch->iSearchRank += $iSearchCost;

                     $oSearch->aAddress = $this->aName;
                     $oSearch->bRareName = false;
                     $oSearch->aName = array($oSearchTerm->iId => $oSearchTerm->iId);
                     $oSearch->aAddress = $this->aName;
                     $oSearch->bRareName = false;
                     $oSearch->aName = array($oSearchTerm->iId => $oSearchTerm->iId);


@@ -255,7 +262,8 @@ class SearchDescription
         ) {
             if ($this->iOperator == Operator::NONE) {
                 $oSearch = clone $this;
         ) {
             if ($this->iOperator == Operator::NONE) {
                 $oSearch = clone $this;

-                $oSearch->iSearchRank++;
+                $oSearch->iSearchRank += 2;
+                $oSearch->iNamePhrase = -1;

 
                 $iOp = $oSearchTerm->iOperator;
                 if ($iOp == Operator::NONE) {
 
                 $iOp = $oSearchTerm->iOperator;
                 if ($iOp == Operator::NONE) {


@@ -265,6 +273,11 @@ class SearchDescription
                         $iOp = Operator::NEAR;
                     }
                     $oSearch->iSearchRank += 2;
                         $iOp = Operator::NEAR;
                     }
                     $oSearch->iSearchRank += 2;

+                } elseif (!$bFirstToken && !$bLastToken) {
+                    $oSearch->iSearchRank += 2;
+                }
+                if ($this->sHouseNumber) {
+                    $oSearch->iSearchRank++;

                 }
 
                 $oSearch->setPoiSearch(
                 }
 
                 $oSearch->setPoiSearch(


@@ -283,13 +296,14 @@ class SearchDescription
             // the first phrase. In unstructured search it may be in a later
             // phrase when the first phrase is a house number.
             if (!empty($this->aName) || !($bFirstPhrase || $sPhraseType == '')) {
             // the first phrase. In unstructured search it may be in a later
             // phrase when the first phrase is a house number.
             if (!empty($this->aName) || !($bFirstPhrase || $sPhraseType == '')) {

-                if (($sPhraseType == '' || !$bFirstPhrase) && !$bHasPartial) {
+                if (($sPhraseType == '' || !$bFirstPhrase) && $oSearchTerm->iTermCount > 1) {

                     $oSearch = clone $this;
                     $oSearch = clone $this;

-                    $oSearch->iSearchRank += 3 * $oSearchTerm->iTermCount;
+                    $oSearch->iNamePhrase = -1;
+                    $oSearch->iSearchRank += 1;

                     $oSearch->aAddress[$iWordID] = $iWordID;
                     $aNewSearches[] = $oSearch;
                 }
                     $oSearch->aAddress[$iWordID] = $iWordID;
                     $aNewSearches[] = $oSearch;
                 }

-            } else {
+            } elseif (empty($this->aNameNonSearch)) {

                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
                 $oSearch->aName = array($iWordID => $iWordID);
                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
                 $oSearch->aName = array($iWordID => $iWordID);


@@ -320,7 +334,9 @@ class SearchDescription
     public function extendWithPartialTerm($sToken, $oSearchTerm, $bStructuredPhrases, $iPhrase, $aFullTokens)
     {
         // Only allow name terms.
     public function extendWithPartialTerm($sToken, $oSearchTerm, $bStructuredPhrases, $iPhrase, $aFullTokens)
     {
         // Only allow name terms.

-        if (!(is_a($oSearchTerm, '\Nominatim\Token\Word'))) {
+        if (!(is_a($oSearchTerm, '\Nominatim\Token\Word'))
+            || strpos($sToken, ' ') !== false
+        ) {

             return array();
         }
 
             return array();
         }
 


@@ -329,51 +345,33 @@ class SearchDescription
 
         if ((!$bStructuredPhrases || $iPhrase > 0)
             && (!empty($this->aName))
 
         if ((!$bStructuredPhrases || $iPhrase > 0)
             && (!empty($this->aName))

-            && strpos($sToken, ' ') === false

         ) {
         ) {

+            $oSearch = clone $this;
+            $oSearch->iSearchRank++;
+            if (preg_match('#^[0-9 ]+$#', $sToken)) {
+                $oSearch->iSearchRank++;
+            }

             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {

-                $oSearch = clone $this;
-                $oSearch->iSearchRank += $oSearchTerm->iTermCount + 1;
-                if (empty($this->aName)) {
-                    $oSearch->iSearchRank++;
-                }
-                if (preg_match('#^[0-9]+$#', $sToken)) {
-                    $oSearch->iSearchRank++;
-                }

                 $oSearch->aAddress[$iWordID] = $iWordID;
                 $oSearch->aAddress[$iWordID] = $iWordID;

-                $aNewSearches[] = $oSearch;

             } else {
             } else {

-                $oSearch = clone $this;
-                $oSearch->iSearchRank += $oSearchTerm->iTermCount + 1;

                 $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
                 if (!empty($aFullTokens)) {
                     $oSearch->iSearchRank++;
                 }
                 $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
                 if (!empty($aFullTokens)) {
                     $oSearch->iSearchRank++;
                 }

-                $aNewSearches[] = $oSearch;
-
-                // revert to the token version?
-                foreach ($aFullTokens as $oSearchTermToken) {
-                    if (is_a($oSearchTermToken, '\Nominatim\Token\Word')) {
-                        $oSearch = clone $this;
-                        $oSearch->iSearchRank += 3;
-                        $oSearch->aAddress[$oSearchTermToken->iId]
-                            = $oSearchTermToken->iId;
-                        $aNewSearches[] = $oSearch;
-                    }
-                }

             }
             }

+            $aNewSearches[] = $oSearch;

         }
 
         if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
         }
 
         if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)

-            && (empty($this->aName) || $this->iNamePhrase == $iPhrase)
+            && ((empty($this->aName) && empty($this->aNameNonSearch)) || $this->iNamePhrase == $iPhrase)

         ) {
             $oSearch = clone $this;
         ) {
             $oSearch = clone $this;

-            $oSearch->iSearchRank += 2;
-            if (empty($this->aName)) {
-                $oSearch->iSearchRank += 1;
+            $oSearch->iSearchRank++;
+            if (empty($this->aName) && empty($this->aNameNonSearch)) {
+                $oSearch->iSearchRank++;

             }
             }

-            if (preg_match('#^[0-9]+$#', $sToken)) {
-                $oSearch->iSearchRank += 2;
+            if (preg_match('#^[0-9 ]+$#', $sToken)) {
+                $oSearch->iSearchRank++;

             }
             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
                 if (empty($this->aName)
             }
             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
                 if (empty($this->aName)


@@ -387,6 +385,9 @@ class SearchDescription
                 }
                 $oSearch->aName[$iWordID] = $iWordID;
             } else {
                 }
                 $oSearch->aName[$iWordID] = $iWordID;
             } else {

+                if (!empty($aFullTokens)) {
+                    $oSearch->iSearchRank++;
+                }

                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;
             }
             $oSearch->iNamePhrase = $iPhrase;
                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;
             }
             $oSearch->iNamePhrase = $iPhrase;


@@ -447,19 +448,24 @@ class SearchDescription
 
             // Now search for housenumber, if housenumber provided. Can be zero.
             if (($this->sHouseNumber || $this->sHouseNumber === '0') && !empty($aResults)) {
 
             // Now search for housenumber, if housenumber provided. Can be zero.
             if (($this->sHouseNumber || $this->sHouseNumber === '0') && !empty($aResults)) {

-                // Downgrade the rank of the street results, they are missing
-                // the housenumber.
-                foreach ($aResults as $oRes) {
-                    $oRes->iResultRank++;
-                }
-

                 $aHnResults = $this->queryHouseNumber($oDB, $aResults);
 
                 $aHnResults = $this->queryHouseNumber($oDB, $aResults);
 

-                if (!empty($aHnResults)) {
-                    foreach ($aHnResults as $oRes) {
-                        $aResults[$oRes->iId] = $oRes;
+                // Downgrade the rank of the street results, they are missing
+                // the housenumber. Also drop POI places (rank 30) here, they
+                // cannot be a parent place and therefore must not be shown
+                // as a result for a search with a missing housenumber.
+                foreach ($aResults as $oRes) {
+                    if ($oRes->iAddressRank < 28) {
+                        if ($oRes->iAddressRank >= 26) {
+                            $oRes->iResultRank++;
+                        } else {
+                            $oRes->iResultRank += 2;
+                        }
+                        $aHnResults[$oRes->iId] = $oRes;

                     }
                 }
                     }
                 }

+
+                $aResults = $aHnResults;

             }
 
             // finally get POIs if requested
             }
 
             // finally get POIs if requested


@@ -611,14 +617,14 @@ class SearchDescription
         // too many results are expected for the street, i.e. if the result
         // will be narrowed down by an address. Remeber that with ordering
         // every single result has to be checked.
         // too many results are expected for the street, i.e. if the result
         // will be narrowed down by an address. Remeber that with ordering
         // every single result has to be checked.

-        if ($this->sHouseNumber && (!empty($this->aAddress) || $this->sPostcode)) {
+        if ($this->sHouseNumber && ($this->bRareName || !empty($this->aAddress) || $this->sPostcode)) {

             $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
             $aOrder[] = ' (';
             $aOrder[0] .= 'EXISTS(';
             $aOrder[0] .= '  SELECT place_id';
             $aOrder[0] .= '  FROM placex';
             $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
             $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
             $aOrder[] = ' (';
             $aOrder[0] .= 'EXISTS(';
             $aOrder[0] .= '  SELECT place_id';
             $aOrder[0] .= '  FROM placex';
             $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';

-            $aOrder[0] .= "    AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
+            $aOrder[0] .= "    AND housenumber ~* E'".$sHouseNumberRegex."'";

             $aOrder[0] .= '  LIMIT 1';
             $aOrder[0] .= ') ';
             // also housenumbers from interpolation lines table are needed
             $aOrder[0] .= '  LIMIT 1';
             $aOrder[0] .= ') ';
             // also housenumbers from interpolation lines table are needed


@@ -715,7 +721,7 @@ class SearchDescription
         $aResults = array();
 
         if (!empty($aTerms)) {
         $aResults = array();
 
         if (!empty($aTerms)) {

-            $sSQL = 'SELECT place_id,'.$sExactMatchSQL;
+            $sSQL = 'SELECT place_id, address_rank,'.$sExactMatchSQL;

             $sSQL .= ' FROM search_name';
             $sSQL .= ' WHERE '.join(' and ', $aTerms);
             $sSQL .= ' ORDER BY '.join(', ', $aOrder);
             $sSQL .= ' FROM search_name';
             $sSQL .= ' WHERE '.join(' and ', $aTerms);
             $sSQL .= ' ORDER BY '.join(', ', $aOrder);


@@ -728,6 +734,7 @@ class SearchDescription
             foreach ($aDBResults as $aResult) {
                 $oResult = new Result($aResult['place_id']);
                 $oResult->iExactMatches = $aResult['exactmatch'];
             foreach ($aDBResults as $aResult) {
                 $oResult = new Result($aResult['place_id']);
                 $oResult->iExactMatches = $aResult['exactmatch'];

+                $oResult->iAddressRank = $aResult['address_rank'];

                 $aResults[$aResult['place_id']] = $oResult;
             }
         }
                 $aResults[$aResult['place_id']] = $oResult;
             }
         }


@@ -738,16 +745,33 @@ class SearchDescription
     private function queryHouseNumber(&$oDB, $aRoadPlaceIDs)
     {
         $aResults = array();
     private function queryHouseNumber(&$oDB, $aRoadPlaceIDs)
     {
         $aResults = array();

-        $sPlaceIDs = Result::joinIdsByTable($aRoadPlaceIDs, Result::TABLE_PLACEX);
+        $sRoadPlaceIDs = Result::joinIdsByTableMaxRank(
+            $aRoadPlaceIDs,
+            Result::TABLE_PLACEX,
+            27
+        );
+        $sPOIPlaceIDs = Result::joinIdsByTableMinRank(
+            $aRoadPlaceIDs,
+            Result::TABLE_PLACEX,
+            30
+        );

 
 

-        if (!$sPlaceIDs) {
+        $aIDCondition = array();
+        if ($sRoadPlaceIDs) {
+            $aIDCondition[] = 'parent_place_id in ('.$sRoadPlaceIDs.')';
+        }
+        if ($sPOIPlaceIDs) {
+            $aIDCondition[] = 'place_id in ('.$sPOIPlaceIDs.')';
+        }
+
+        if (empty($aIDCondition)) {

             return $aResults;
         }
 
         $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
             return $aResults;
         }
 
         $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';

-        $sSQL = 'SELECT place_id FROM placex ';
-        $sSQL .= 'WHERE parent_place_id in ('.$sPlaceIDs.')';
-        $sSQL .= "  AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
+        $sSQL = 'SELECT place_id FROM placex WHERE';
+        $sSQL .= "  housenumber ~* E'".$sHouseNumberRegex."'";
+        $sSQL .= ' AND ('.join(' OR ', $aIDCondition).')';

         $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 
         Debug::printSQL($sSQL);
         $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 
         Debug::printSQL($sSQL);


@@ -759,11 +783,11 @@ class SearchDescription
 
         $bIsIntHouseNumber= (bool) preg_match('/[0-9]+/', $this->sHouseNumber);
         $iHousenumber = intval($this->sHouseNumber);
 
         $bIsIntHouseNumber= (bool) preg_match('/[0-9]+/', $this->sHouseNumber);
         $iHousenumber = intval($this->sHouseNumber);

-        if ($bIsIntHouseNumber && empty($aResults)) {
+        if ($bIsIntHouseNumber && $sRoadPlaceIDs && empty($aResults)) {

             // if nothing found, search in the interpolation line table
             $sSQL = 'SELECT distinct place_id FROM location_property_osmline';
             $sSQL .= ' WHERE startnumber is not NULL';
             // if nothing found, search in the interpolation line table
             $sSQL = 'SELECT distinct place_id FROM location_property_osmline';
             $sSQL .= ' WHERE startnumber is not NULL';

-            $sSQL .= '  AND parent_place_id in ('.$sPlaceIDs.') AND (';
+            $sSQL .= '  AND parent_place_id in ('.$sRoadPlaceIDs.') AND (';

             if ($iHousenumber % 2 == 0) {
                 // If housenumber is even, look for housenumber in streets
                 // with interpolationtype even or all.
             if ($iHousenumber % 2 == 0) {
                 // If housenumber is even, look for housenumber in streets
                 // with interpolationtype even or all.


@@ -786,24 +810,10 @@ class SearchDescription
             }
         }
 
             }
         }
 

-        // If nothing found try the aux fallback table
-        if (CONST_Use_Aux_Location_data && empty($aResults)) {
-            $sSQL = 'SELECT place_id FROM location_property_aux';
-            $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.')';
-            $sSQL .= " AND housenumber = '".$this->sHouseNumber."'";
-            $sSQL .= $this->oContext->excludeSQL(' AND place_id');
-
-            Debug::printSQL($sSQL);
-
-            foreach ($oDB->getCol($sSQL) as $iPlaceId) {
-                $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_AUX);
-            }
-        }
-

         // If nothing found then search in Tiger data (location_property_tiger)
         // If nothing found then search in Tiger data (location_property_tiger)

-        if (CONST_Use_US_Tiger_Data && $bIsIntHouseNumber && empty($aResults)) {
+        if (CONST_Use_US_Tiger_Data && $sRoadPlaceIDs && $bIsIntHouseNumber && empty($aResults)) {

             $sSQL = 'SELECT place_id FROM location_property_tiger';
             $sSQL = 'SELECT place_id FROM location_property_tiger';

-            $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.') and (';
+            $sSQL .= ' WHERE parent_place_id in ('.$sRoadPlaceIDs.') and (';

             if ($iHousenumber % 2 == 0) {
                 $sSQL .= "interpolationtype='even'";
             } else {
             if ($iHousenumber % 2 == 0) {
                 $sSQL .= "interpolationtype='even'";
             } else {