]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/SearchDescription.php
update osm2pgsql (disable install rule)
[nominatim.git] / lib / SearchDescription.php
index 079cb8a6e65d1c72d8dd8693e40d740569206d45..2053082f6bbff4a13bd6a5ba6c08bc89366e0c38 100644 (file)
@@ -2,9 +2,9 @@
 
 namespace Nominatim;
 
-require_once(CONST_BasePath.'/lib/SpecialSearchOperator.php');
-require_once(CONST_BasePath.'/lib/SearchContext.php');
-require_once(CONST_BasePath.'/lib/Result.php');
+require_once(CONST_LibDir.'/SpecialSearchOperator.php');
+require_once(CONST_LibDir.'/SearchContext.php');
+require_once(CONST_LibDir.'/Result.php');
 
 /**
  * Description of a single interpretation of a search query.
@@ -21,8 +21,6 @@ class SearchDescription
     private $bRareName = false;
     /// List of word ids making up the address of the object.
     private $aAddress = array();
-    /// Subset of word ids of full words making up the address.
-    private $aFullNameAddress = array();
     /// List of word ids that appear in the name but should be ignored.
     private $aNameNonSearch = array();
     /// List of word ids that appear in the address but should be ignored.
@@ -88,18 +86,6 @@ class SearchDescription
         $this->sType = $sType;
     }
 
-    /**
-     * Check if this might be a full address search.
-     *
-     * @return bool True if the search contains name, address and housenumber.
-     */
-    public function looksLikeFullAddress()
-    {
-        return (!empty($this->aName))
-               && (!empty($this->aAddress) || $this->sCountryCode)
-               && preg_match('/[0-9]+/', $this->sHouseNumber);
-    }
-
     /**
      * Check if any operator is set.
      *
@@ -199,15 +185,10 @@ class SearchDescription
         } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode')
                   && is_a($oSearchTerm, '\Nominatim\Token\Postcode')
         ) {
-            // We need to try the case where the postal code is the primary element
-            // (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode)
-            // so try both.
             if (!$this->sPostcode) {
                 // If we have structured search or this is the first term,
                 // make the postcode the primary search element.
-                if ($this->iOperator == Operator::NONE
-                    && ($sPhraseType == 'postalcode' || $bFirstToken)
-                ) {
+                if ($this->iOperator == Operator::NONE && $bFirstToken) {
                     $oSearch = clone $this;
                     $oSearch->iSearchRank++;
                     $oSearch->iOperator = Operator::POSTCODE;
@@ -224,6 +205,9 @@ class SearchDescription
                 ) {
                     $oSearch = clone $this;
                     $oSearch->iSearchRank++;
+                    if (strlen($oSearchTerm->sPostcode) < 4) {
+                        $oSearch->iSearchRank += 4 - strlen($oSearchTerm->sPostcode);
+                    }
                     $oSearch->sPostcode = $oSearchTerm->sPostcode;
                     $aNewSearches[] = $oSearch;
                 }
@@ -237,7 +221,8 @@ class SearchDescription
                 $oSearch->sHouseNumber = $oSearchTerm->sToken;
                 // sanity check: if the housenumber is not mainly made
                 // up of numbers, add a penalty
-                if (preg_match_all('/[^0-9]/', $oSearch->sHouseNumber, $aMatches) > 2) {
+                if (preg_match('/\\d/', $oSearch->sHouseNumber) === 0
+                    || preg_match_all('/[^0-9]/', $oSearch->sHouseNumber, $aMatches) > 2) {
                     $oSearch->iSearchRank++;
                 }
                 if (empty($oSearchTerm->iId)) {
@@ -251,6 +236,19 @@ class SearchDescription
                     $oSearch->iSearchRank++;
                 }
                 $aNewSearches[] = $oSearch;
+                // Housenumbers may appear in the name when the place has its own
+                // address terms.
+                if ($oSearchTerm->iId !== null
+                    && ($this->iNamePhrase >= 0 || empty($this->aName))
+                    && empty($this->aAddress)
+                   ) {
+                    $oSearch = clone $this;
+                    $oSearch->iSearchRank++;
+                    $oSearch->aAddress = $this->aName;
+                    $oSearch->bRareName = false;
+                    $oSearch->aName = array($oSearchTerm->iId => $oSearchTerm->iId);
+                    $aNewSearches[] = $oSearch;
+                }
             }
         } elseif ($sPhraseType == ''
                   && is_a($oSearchTerm, '\Nominatim\Token\SpecialTerm')
@@ -287,11 +285,9 @@ class SearchDescription
             if (!empty($this->aName) || !($bFirstPhrase || $sPhraseType == '')) {
                 if (($sPhraseType == '' || !$bFirstPhrase) && !$bHasPartial) {
                     $oSearch = clone $this;
-                    $oSearch->iSearchRank++;
+                    $oSearch->iSearchRank += 3 * $oSearchTerm->iTermCount;
                     $oSearch->aAddress[$iWordID] = $iWordID;
                     $aNewSearches[] = $oSearch;
-                } else {
-                    $this->aFullNameAddress[$iWordID] = $iWordID;
                 }
             } else {
                 $oSearch = clone $this;
@@ -337,16 +333,19 @@ class SearchDescription
         ) {
             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
                 $oSearch = clone $this;
-                $oSearch->iSearchRank += 2;
+                $oSearch->iSearchRank += $oSearchTerm->iTermCount + 1;
+                if (empty($this->aName)) {
+                    $oSearch->iSearchRank++;
+                }
+                if (preg_match('#^[0-9]+$#', $sToken)) {
+                    $oSearch->iSearchRank++;
+                }
                 $oSearch->aAddress[$iWordID] = $iWordID;
                 $aNewSearches[] = $oSearch;
             } else {
                 $oSearch = clone $this;
-                $oSearch->iSearchRank++;
+                $oSearch->iSearchRank += $oSearchTerm->iTermCount + 1;
                 $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
-                if (preg_match('#^[0-9]+$#', $sToken)) {
-                    $oSearch->iSearchRank += 2;
-                }
                 if (!empty($aFullTokens)) {
                     $oSearch->iSearchRank++;
                 }
@@ -356,7 +355,7 @@ class SearchDescription
                 foreach ($aFullTokens as $oSearchTermToken) {
                     if (is_a($oSearchTermToken, '\Nominatim\Token\Word')) {
                         $oSearch = clone $this;
-                        $oSearch->iSearchRank++;
+                        $oSearch->iSearchRank += 3;
                         $oSearch->aAddress[$oSearchTermToken->iId]
                             = $oSearchTermToken->iId;
                         $aNewSearches[] = $oSearch;
@@ -403,7 +402,7 @@ class SearchDescription
     /**
      * Query database for places that match this search.
      *
-     * @param object  $oDB      Database connection to use.
+     * @param object  $oDB      Nominatim::DB instance to use.
      * @param integer $iMinRank Minimum address rank to restrict search to.
      * @param integer $iMaxRank Maximum address rank to restrict search to.
      * @param integer $iLimit   Maximum number of results.
@@ -446,13 +445,20 @@ class SearchDescription
                 $iLimit
             );
 
-            //now search for housenumber, if housenumber provided
-            if ($this->sHouseNumber && !empty($aResults)) {
-                $aNamedPlaceIDs = $aResults;
-                $aResults = $this->queryHouseNumber($oDB, $aNamedPlaceIDs);
+            // Now search for housenumber, if housenumber provided. Can be zero.
+            if (($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);
 
-                if (empty($aResults) && $this->looksLikeFullAddress()) {
-                    $aResults = $aNamedPlaceIDs;
+                if (!empty($aHnResults)) {
+                    foreach ($aHnResults as $oRes) {
+                        $aResults[$oRes->iId] = $oRes;
+                    }
                 }
             }
 
@@ -469,16 +475,13 @@ class SearchDescription
             if ($sPlaceIds) {
                 $sSQL = 'SELECT place_id FROM placex';
                 $sSQL .= ' WHERE place_id in ('.$sPlaceIds.')';
-                $sSQL .= " AND postcode = '".$this->sPostcode."'";
+                $sSQL .= " AND postcode != '".$this->sPostcode."'";
                 Debug::printSQL($sSQL);
-                $aFilteredPlaceIDs = chksql($oDB->getCol($sSQL));
+                $aFilteredPlaceIDs = $oDB->getCol($sSQL);
                 if ($aFilteredPlaceIDs) {
-                    $aNewResults = array();
                     foreach ($aFilteredPlaceIDs as $iPlaceId) {
-                        $aNewResults[$iPlaceId] = $aResults[$iPlaceId];
+                        $aResults[$iPlaceId]->iResultRank++;
                     }
-                    $aResults = $aNewResults;
-                    Debug::printVar('Place IDs after postcode filtering', $aResults);
                 }
             }
         }
@@ -499,8 +502,10 @@ class SearchDescription
 
         Debug::printSQL($sSQL);
 
+        $iPlaceId = $oDB->getOne($sSQL);
+
         $aResults = array();
-        foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+        if ($iPlaceId) {
             $aResults[$iPlaceId] = new Result($iPlaceId);
         }
 
@@ -516,8 +521,7 @@ class SearchDescription
         $aDBResults = array();
         $sPoiTable = $this->poiTable();
 
-        $sSQL = 'SELECT count(*) FROM pg_tables WHERE tablename = \''.$sPoiTable."'";
-        if (chksql($oDB->getOne($sSQL))) {
+        if ($oDB->tableExists($sPoiTable)) {
             $sSQL = 'SELECT place_id FROM '.$sPoiTable.' ct';
             if ($this->oContext->sqlCountryList) {
                 $sSQL .= ' JOIN placex USING (place_id)';
@@ -537,14 +541,14 @@ class SearchDescription
             } elseif ($this->oContext->hasNearPoint()) {
                 $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('ct.centroid').' ASC';
             }
-            $sSQL .= " limit $iLimit";
+            $sSQL .= " LIMIT $iLimit";
             Debug::printSQL($sSQL);
-            $aDBResults = chksql($oDB->getCol($sSQL));
+            $aDBResults = $oDB->getCol($sSQL);
         }
 
         if ($this->oContext->hasNearPoint()) {
             $sSQL = 'SELECT place_id FROM placex WHERE ';
-            $sSQL .= 'class=\''.$this->sClass."' and type='".$this->sType."'";
+            $sSQL .= 'class = :class and type = :type';
             $sSQL .= ' AND '.$this->oContext->withinSQL('geometry');
             $sSQL .= ' AND linked_place_id is null';
             if ($this->oContext->sqlCountryList) {
@@ -553,7 +557,10 @@ class SearchDescription
             $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('centroid').' ASC';
             $sSQL .= " LIMIT $iLimit";
             Debug::printSQL($sSQL);
-            $aDBResults = chksql($oDB->getCol($sSQL));
+            $aDBResults = $oDB->getCol(
+                $sSQL,
+                array(':class' => $this->sClass, ':type' => $this->sType)
+            );
         }
 
         $aResults = array();
@@ -572,20 +579,23 @@ class SearchDescription
             $sSQL .= ', search_name s ';
             $sSQL .= 'WHERE s.place_id = p.parent_place_id ';
             $sSQL .= 'AND array_cat(s.nameaddress_vector, s.name_vector)';
-            $sSQL .= '      @> '.getArraySQL($this->aAddress).' AND ';
+            $sSQL .= '      @> '.$oDB->getArraySQL($this->aAddress).' AND ';
         } else {
             $sSQL .= 'WHERE ';
         }
 
         $sSQL .= "p.postcode = '".reset($this->aName)."'";
         $sSQL .= $this->countryCodeSQL(' AND p.country_code');
+        if ($this->oContext->bViewboxBounded) {
+            $sSQL .= ' AND ST_Intersects('.$this->oContext->sqlViewboxSmall.', geometry)';
+        }
         $sSQL .= $this->oContext->excludeSQL(' AND p.place_id');
         $sSQL .= " LIMIT $iLimit";
 
         Debug::printSQL($sSQL);
 
         $aResults = array();
-        foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+        foreach ($oDB->getCol($sSQL) as $iPlaceId) {
             $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_POSTCODE);
         }
 
@@ -628,14 +638,14 @@ class SearchDescription
         }
 
         if (!empty($this->aName)) {
-            $aTerms[] = 'name_vector @> '.getArraySQL($this->aName);
+            $aTerms[] = 'name_vector @> '.$oDB->getArraySQL($this->aName);
         }
         if (!empty($this->aAddress)) {
             // For infrequent name terms disable index usage for address
             if ($this->bRareName) {
-                $aTerms[] = 'array_cat(nameaddress_vector,ARRAY[]::integer[]) @> '.getArraySQL($this->aAddress);
+                $aTerms[] = 'array_cat(nameaddress_vector,ARRAY[]::integer[]) @> '.$oDB->getArraySQL($this->aAddress);
             } else {
-                $aTerms[] = 'nameaddress_vector @> '.getArraySQL($this->aAddress);
+                $aTerms[] = 'nameaddress_vector @> '.$oDB->getArraySQL($this->aAddress);
             }
         }
 
@@ -645,13 +655,10 @@ class SearchDescription
         }
 
         if ($this->sHouseNumber) {
-            $aTerms[] = 'address_rank between 16 and 27';
+            $aTerms[] = 'address_rank between 16 and 30';
         } elseif (!$this->sClass || $this->iOperator == Operator::NAME) {
             if ($iMinAddressRank > 0) {
-                $aTerms[] = 'address_rank >= '.$iMinAddressRank;
-            }
-            if ($iMaxAddressRank < 30) {
-                $aTerms[] = 'address_rank <= '.$iMaxAddressRank;
+                $aTerms[] = "((address_rank between $iMinAddressRank and $iMaxAddressRank) or (search_rank between $iMinAddressRank and $iMaxAddressRank))";
             }
         }
 
@@ -687,10 +694,11 @@ class SearchDescription
         $sImportanceSQL .= $this->oContext->viewboxImportanceSQL('centroid');
         $aOrder[] = "$sImportanceSQL DESC";
 
-        if (!empty($this->aFullNameAddress)) {
+        $aFullNameAddress = $this->oContext->getFullNameTerms();
+        if (!empty($aFullNameAddress)) {
             $sExactMatchSQL = ' ( ';
             $sExactMatchSQL .= ' SELECT count(*) FROM ( ';
-            $sExactMatchSQL .= '  SELECT unnest('.getArraySQL($this->aFullNameAddress).')';
+            $sExactMatchSQL .= '  SELECT unnest('.$oDB->getArraySQL($aFullNameAddress).')';
             $sExactMatchSQL .= '    INTERSECT ';
             $sExactMatchSQL .= '  SELECT unnest(nameaddress_vector)';
             $sExactMatchSQL .= ' ) s';
@@ -701,7 +709,7 @@ class SearchDescription
         }
 
         if ($this->sHouseNumber || $this->sClass) {
-            $iLimit = 20;
+            $iLimit = 40;
         }
 
         $aResults = array();
@@ -715,10 +723,7 @@ class SearchDescription
 
             Debug::printSQL($sSQL);
 
-            $aDBResults = chksql(
-                $oDB->getAll($sSQL),
-                'Could not get places for search terms.'
-            );
+            $aDBResults = $oDB->getAll($sSQL, null, 'Could not get places for search terms.');
 
             foreach ($aDBResults as $aResult) {
                 $oResult = new Result($aResult['place_id']);
@@ -748,7 +753,7 @@ class SearchDescription
         Debug::printSQL($sSQL);
 
         // XXX should inherit the exactMatches from its parent
-        foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+        foreach ($oDB->getCol($sSQL) as $iPlaceId) {
             $aResults[$iPlaceId] = new Result($iPlaceId);
         }
 
@@ -774,7 +779,7 @@ class SearchDescription
 
             Debug::printSQL($sSQL);
 
-            foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+            foreach ($oDB->getCol($sSQL) as $iPlaceId) {
                 $oResult = new Result($iPlaceId, Result::TABLE_OSMLINE);
                 $oResult->iHouseNumber = $iHousenumber;
                 $aResults[$iPlaceId] = $oResult;
@@ -790,7 +795,7 @@ class SearchDescription
 
             Debug::printSQL($sSQL);
 
-            foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+            foreach ($oDB->getCol($sSQL) as $iPlaceId) {
                 $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_AUX);
             }
         }
@@ -811,7 +816,7 @@ class SearchDescription
 
             Debug::printSQL($sSQL);
 
-            foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+            foreach ($oDB->getCol($sSQL) as $iPlaceId) {
                 $oResult = new Result($iPlaceId, Result::TABLE_TIGER);
                 $oResult->iHouseNumber = $iHousenumber;
                 $aResults[$iPlaceId] = $oResult;
@@ -845,7 +850,7 @@ class SearchDescription
 
             Debug::printSQL($sSQL);
 
-            foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+            foreach ($oDB->getCol($sSQL) as $iPlaceId) {
                 $aResults[$iPlaceId] = new Result($iPlaceId);
             }
         }
@@ -853,12 +858,11 @@ class SearchDescription
         // NEAR and IN are handled the same
         if ($this->iOperator == Operator::TYPE || $this->iOperator == Operator::NEAR) {
             $sClassTable = $this->poiTable();
-            $sSQL = "SELECT count(*) FROM pg_tables WHERE tablename = '$sClassTable'";
-            $bCacheTable = (bool) chksql($oDB->getOne($sSQL));
+            $bCacheTable = $oDB->tableExists($sClassTable);
 
             $sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
             Debug::printSQL($sSQL);
-            $iMaxRank = (int)chksql($oDB->getOne($sSQL));
+            $iMaxRank = (int) $oDB->getOne($sSQL);
 
             // For state / country level searches the normal radius search doesn't work very well
             $sPlaceGeom = false;
@@ -871,7 +875,7 @@ class SearchDescription
                 $sSQL .= ' ORDER BY rank_search ASC ';
                 $sSQL .= ' LIMIT 1';
                 Debug::printSQL($sSQL);
-                $sPlaceGeom = chksql($oDB->getOne($sSQL));
+                $sPlaceGeom = $oDB->getOne($sSQL);
             }
 
             if ($sPlaceGeom) {
@@ -881,7 +885,7 @@ class SearchDescription
                 $sSQL = 'SELECT place_id FROM placex';
                 $sSQL .= " WHERE place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
                 Debug::printSQL($sSQL);
-                $aPlaceIDs = chksql($oDB->getCol($sSQL));
+                $aPlaceIDs = $oDB->getCol($sSQL);
                 $sPlaceIDs = join(',', $aPlaceIDs);
             }
 
@@ -927,7 +931,7 @@ class SearchDescription
 
                     Debug::printSQL($sSQL);
 
-                    foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+                    foreach ($oDB->getCol($sSQL) as $iPlaceId) {
                         $aResults[$iPlaceId] = new Result($iPlaceId);
                     }
                 } else {
@@ -959,7 +963,7 @@ class SearchDescription
 
                     Debug::printSQL($sSQL);
 
-                    foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+                    foreach ($oDB->getCol($sSQL) as $iPlaceId) {
                         $aResults[$iPlaceId] = new Result($iPlaceId);
                     }
                 }
@@ -1011,7 +1015,7 @@ class SearchDescription
                 'Name terms (stop words)' => $this->aNameNonSearch,
                 'Address terms' => $this->aAddress,
                 'Address terms (stop words)' => $this->aAddressNonSearch,
-                'Address terms (full words)' => $this->aFullNameAddress,
+                'Address terms (full words)' => $this->aFullNameAddress ?? '',
                 'Special search' => $this->iOperator,
                 'Class' => $this->sClass,
                 'Type' => $this->sType,
@@ -1023,7 +1027,7 @@ class SearchDescription
     public function dumpAsHtmlTableRow(&$aWordIDs)
     {
         $kf = function ($k) use (&$aWordIDs) {
-            return $aWordIDs[$k];
+            return $aWordIDs[$k] ?? '['.$k.']';
         };
 
         echo '<tr>';