]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib/Geocode.php
penalize search order where a country comes first
[nominatim.git] / lib / Geocode.php
index b89337041efe78a3d8183ad1f7118d184fe23d04..bced2a36a555c2166190f9d0c4d0f5caedd93fd3 100644 (file)
@@ -63,7 +63,7 @@ class Geocode
     private function normTerm($sTerm)
     {
         if ($this->oNormalizer === null) {
-            return null;
+            return $sTerm;
         }
 
         return $this->oNormalizer->transliterate($sTerm);
@@ -304,7 +304,7 @@ class Geocode
         $aViewbox = $oParams->getStringList('viewboxlbrt');
         if ($aViewbox) {
             if (count($aViewbox) != 4) {
-                userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
+                userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
             }
             $this->setViewbox($aViewbox);
         } else {
@@ -313,12 +313,7 @@ class Geocode
                 if (count($aViewbox) != 4) {
                     userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
                 }
-                $this->setViewBox(array(
-                                   $aViewbox[0],
-                                   $aViewbox[3],
-                                   $aViewbox[2],
-                                   $aViewbox[1]
-                                  ));
+                $this->setViewBox($aViewbox);
             } else {
                 $aRoute = $oParams->getStringList('route');
                 $fRouteWidth = $oParams->getFloat('routewidth');
@@ -372,7 +367,7 @@ class Geocode
         $this->aAddressRankList = array();
 
         $this->aStructuredQuery = array();
-        $this->sAllowedTypesSQLList = '';
+        $this->sAllowedTypesSQLList = False;
 
         $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
         $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
@@ -385,7 +380,7 @@ class Geocode
         if (sizeof($this->aStructuredQuery) > 0) {
             $this->sQuery = join(', ', $this->aStructuredQuery);
             if ($this->iMaxAddressRank < 30) {
-                $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+                $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
             }
         }
     }
@@ -714,6 +709,8 @@ class Geocode
 
              Score how good the search is so they can be ordered
          */
+        $iGlobalRank = 0;
+
         foreach ($aPhrases as $iPhrase => $aPhrase) {
             $aNewPhraseSearches = array();
             if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
@@ -737,6 +734,8 @@ class Geocode
 
                         // If the token is valid
                         if (isset($aValidTokens[' '.$sToken])) {
+                            // TODO variable should go into aCurrentSearch
+                            $bHavePostcode = false;
                             foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
                                 $aSearch = $aCurrentSearch;
                                 $aSearch['iSearchRank']++;
@@ -748,19 +747,25 @@ class Geocode
                                             $aSearch['iSearchRank'] += 5;
                                         }
                                         if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+                                        // If it is at the beginning, we can be almost sure that this is the wrong order
+                                        // Increase score for all searches.
+                                        if ($iToken == 0 && $iPhrase == 0) {
+                                            $iGlobalRank++;
+                                        }
                                     }
-                                } elseif ($sPhraseType == 'postalcode' || ($aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode')) {
+                                } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') {
                                     // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
-                                    if ($aSearch['sPostcode'] === '' && $aSearch['sHouseNumber'] === '' &&
-                                        isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) {
+                                    if ($aSearch['sPostcode'] === '' &&
+                                        isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) {
                                         // If we have structured search or this is the first term,
                                         // make the postcode the primary search element.
-                                        if ($aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
+                                        if (!$bHavePostcode && $aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
                                             $aNewSearch = $aSearch;
                                             $aNewSearch['sOperator'] = 'postcode';
                                             $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
-                                            $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word'];
+                                            $aNewSearch['aName'] = array($aSearchTerm['word_id'] => $aSearchTerm['word']);
                                             if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;
+                                            $bHavePostcode = true;
                                         }
 
                                         // If we have a structured search or this is not the first term,
@@ -790,7 +795,7 @@ class Geocode
                                     // require a normalized exact match of the term
                                     // if we have the normalizer version of the query
                                     // available
-                                    if ($aSearch['sClass'] === ''
+                                    if ($aSearch['sOperator'] === ''
                                         && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) {
                                         $aSearch['sClass'] = $aSearchTerm['class'];
                                         $aSearch['sType'] = $aSearchTerm['type'];
@@ -915,6 +920,20 @@ class Geocode
 
             //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
         }
+
+        // Revisit searches, giving penalty to unlikely combinations
+        $aGroupedSearches = array();
+        foreach ($aSearches as $aSearch) {
+            if (!$aSearch['aName']) {
+                if ($aSearch['sHouseNumber']) {
+                    continue;
+                }
+            }
+            $aSearch['iSearchRank'] += $iGlobalRank;
+            $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+        }
+        ksort($aGroupedSearches);
+
         return $aGroupedSearches;
     }
 
@@ -1021,15 +1040,24 @@ class Geocode
 
             // Any 'special' terms in the search?
             $bSpecialTerms = false;
-            preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
-            $aSpecialTerms = array();
+            preg_match_all('/\\[([\\w_]*)=([\\w_]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
             foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                 $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
-                $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2];
+                if (!$bSpecialTerms) {
+                    $aNewSearches = array();
+                    foreach ($aSearches as $aSearch) {
+                        $aNewSearch = $aSearch;
+                        $aNewSearch['sClass'] = $aSpecialTerm[1];
+                        $aNewSearch['sType'] = $aSpecialTerm[2];
+                        $aNewSearches[] = $aNewSearch;
+                    }
+
+                    $aSearches = $aNewSearches;
+                    $bSpecialTerms = true;
+                }
             }
 
             preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
-            $aSpecialTerms = array();
             if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) {
                 $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
                 unset($this->aStructuredQuery['amenity']);
@@ -1037,32 +1065,28 @@ class Geocode
 
             foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                 $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
-                $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".$aSpecialTerm[1]."') AS string"));
+                if ($bSpecialTerms) {
+                    continue;
+                }
+
+                $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".pg_escape_string($aSpecialTerm[1])."') AS string"));
                 $sSQL = 'SELECT * ';
                 $sSQL .= 'FROM ( ';
                 $sSQL .= '   SELECT word_id, word_token, word, class, type, country_code, operator';
                 $sSQL .= '   FROM word ';
                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
                 $sSQL .= ') AS x ';
-                $sSQL .= ' WHERE (class is not null AND class not in (\'place\')) ';
-                $sSQL .= ' OR country_code is not null';
+                $sSQL .= ' WHERE (class is not null AND class not in (\'place\'))';
                 if (CONST_Debug) var_Dump($sSQL);
                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
                 $aNewSearches = array();
                 foreach ($aSearches as $aSearch) {
                     foreach ($aSearchWords as $aSearchTerm) {
                         $aNewSearch = $aSearch;
-                        if ($aSearchTerm['country_code']) {
-                            $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
-                            $aNewSearches[] = $aNewSearch;
-                            $bSpecialTerms = true;
-                        }
-                        if ($aSearchTerm['class']) {
-                            $aNewSearch['sClass'] = $aSearchTerm['class'];
-                            $aNewSearch['sType'] = $aSearchTerm['type'];
-                            $aNewSearches[] = $aNewSearch;
-                            $bSpecialTerms = true;
-                        }
+                        $aNewSearch['sClass'] = $aSearchTerm['class'];
+                        $aNewSearch['sType'] = $aSearchTerm['type'];
+                        $aNewSearches[] = $aNewSearch;
+                        $bSpecialTerms = true;
                     }
                 }
                 $aSearches = $aNewSearches;
@@ -1111,14 +1135,10 @@ class Geocode
                 if (CONST_Debug) var_Dump($sSQL);
 
                 $aValidTokens = array();
-                if (sizeof($aTokens)) {
-                    $aDatabaseWords = chksql(
-                        $this->oDB->getAll($sSQL),
-                        "Could not get word tokens."
-                    );
-                } else {
-                    $aDatabaseWords = array();
-                }
+                $aDatabaseWords = chksql(
+                    $this->oDB->getAll($sSQL),
+                    "Could not get word tokens."
+                );
                 $aPossibleMainWordIDs = array();
                 $aWordFrequencyScores = array();
                 foreach ($aDatabaseWords as $aToken) {
@@ -1269,8 +1289,8 @@ class Geocode
                     }
 
                     // No location term?
-                    if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['oNear']) {
-                        if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) {
+                    if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress'])) {
+                        if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber'] && !$aSearch['oNear']) {
                             // Just looking for a country by code - look it up
                             if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) {
                                 $sSQL = "SELECT place_id FROM placex WHERE country_code='".$aSearch['sCountryCode']."' AND rank_search = 4";
@@ -1290,39 +1310,32 @@ class Geocode
                             if (chksql($this->oDB->getOne($sSQL))) {
                                 $sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
                                 if ($sCountryCodesSQL) $sSQL .= " JOIN placex USING (place_id)";
-                                $sSQL .= " WHERE st_contains($this->sViewboxSmallSQL, ct.centroid)";
+                                if ($aSearch['oNear']) {
+                                    $sSQL .= " WHERE ".$aSearch['oNear']->withinSQL('ct.centroid');
+                                } else {
+                                    $sSQL .= " WHERE st_contains($this->sViewboxSmallSQL, ct.centroid)";
+                                }
                                 if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
                                 if (sizeof($this->aExcludePlaceIDs)) {
                                     $sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                 }
-                                if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
+                                if ($this->sViewboxCentreSQL) {
+                                    $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
+                                } elseif ($aSearch['oNear']) {
+                                    $sSQL .= " ORDER BY ".$aSearch['oNear']->distanceSQL('ct.centroid').' ASC';
+                                }
                                 $sSQL .= " limit $this->iLimit";
                                 if (CONST_Debug) var_dump($sSQL);
                                 $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
-
-                                // If excluded place IDs are given, it is fair to assume that
-                                // there have been results in the small box, so no further
-                                // expansion in that case.
-                                // Also don't expand if bounded results were requested.
-                                if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs) && !$this->bBoundedSearch) {
-                                    $sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
-                                    if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
-                                    $sSQL .= " WHERE ST_Contains($this->sViewboxLargeSQL, ct.centroid)";
-                                    if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
-                                    if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
-                                    $sSQL .= " LIMIT $this->iLimit";
-                                    if (CONST_Debug) var_dump($sSQL);
-                                    $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
-                                }
-                            } else {
+                            } else if ($aSearch['oNear']) {
                                 $sSQL = "SELECT place_id ";
                                 $sSQL .= "FROM placex ";
                                 $sSQL .= "WHERE class='".$aSearch['sClass']."' ";
                                 $sSQL .= "  AND type='".$aSearch['sType']."'";
-                                $sSQL .= "  AND ST_Contains($this->sViewboxSmallSQL, geometry) ";
+                                $sSQL .= "  AND ".$aSearch['oNear']->withinSQL('geometry');
                                 $sSQL .= "  AND linked_place_id is null";
                                 if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
-                                if ($this->sViewboxCentreSQL)   $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, centroid) ASC";
+                                $sSQL .= " ORDER BY ".$aSearch['oNear']->distanceSQL('centroid')." ASC";
                                 $sSQL .= " LIMIT $this->iLimit";
                                 if (CONST_Debug) var_dump($sSQL);
                                 $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
@@ -1403,7 +1416,7 @@ class Geocode
                         if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                         if ($aSearch['sHouseNumber']) {
                             $aTerms[] = "address_rank between 16 and 27";
-                        } else {
+                        } elseif (!$aSearch['sClass'] || $aSearch['sOperator'] == 'name') {
                             if ($this->iMinAddressRank > 0) {
                                 $aTerms[] = "address_rank >= ".$this->iMinAddressRank;
                             }
@@ -1570,7 +1583,8 @@ class Geocode
                             }
 
                             // Fallback to the road (if no housenumber was found)
-                            if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) {
+                            if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])
+                                && ($aSearch['aAddress'] || $aSearch['sCountryCode'])) {
                                 $aPlaceIDs = $aRoadPlaceIDs;
                                 //set to -1, if no housenumbers were found
                                 $searchedHousenumber = -1;
@@ -1646,7 +1660,7 @@ class Geocode
                                         } elseif ($sPlaceIDs) {
                                             $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
                                         } elseif ($sPlaceGeom) {
-                                            $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
+                                            $sOrderBySQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
                                         }
 
                                         $sSQL = "select distinct i.place_id".($sOrderBySQL?', i.order_term':'')." from (";
@@ -1682,7 +1696,7 @@ class Geocode
                                             $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
                                         }
 
-                                        $sSQL = "SELECT distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'');
+                                        $sSQL = "SELECT distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'');
                                         $sSQL .= " FROM placex as l, placex as f ";
                                         $sSQL .= " WHERE f.place_id in ($sPlaceIDs) ";
                                         $sSQL .= "  AND ST_DWithin(l.geometry, f.centroid, $fRange) ";
@@ -1692,7 +1706,7 @@ class Geocode
                                             $sSQL .= " AND l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
                                         }
                                         if ($sCountryCodesSQL) $sSQL .= " AND l.country_code in ($sCountryCodesSQL)";
-                                        if ($sOrderBy) $sSQL .= "ORDER BY ".$OrderBysSQL." ASC";
+                                        if ($sOrderBySQL) $sSQL .= "ORDER BY ".$sOrderBySQL." ASC";
                                         if ($this->iOffset) $sSQL .= " OFFSET $this->iOffset";
                                         $sSQL .= " limit $this->iLimit";
                                         if (CONST_Debug) var_dump($sSQL);
@@ -1735,9 +1749,11 @@ class Geocode
                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
                     // reduces the number of place ids, like a filter
                     // rank_address is 30 for interpolated housenumbers
+                    $sWherePlaceId = 'WHERE place_id in (';
+                    $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
+
                     $sSQL = "SELECT place_id ";
-                    $sSQL .= "FROM placex ";
-                    $sSQL .= "WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
+                    $sSQL .= "FROM placex ".$sWherePlaceId;
                     $sSQL .= "  AND (";
                     $sSQL .= "         placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                     if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
@@ -1746,20 +1762,23 @@ class Geocode
                     if ($this->aAddressRankList) {
                         $sSQL .= "     OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
                     }
-                    if (CONST_Use_US_Tiger_Data) {
-                        $sSQL .= "  ) ";
+                    $sSQL .= "  ) UNION ";
+                    $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
+                    $sSQL .= "  AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
+                    if ($this->aAddressRankList) {
+                        $sSQL .= "     OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
+                    }
+                    $sSQL .= ") ";
+                    if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
+                        $sSQL .= "UNION ";
+                        $sSQL .= "  SELECT place_id ";
+                        $sSQL .= "  FROM location_property_tiger ".$sWherePlaceId;
+                    }
+                    if ($this->iMaxAddressRank == 30) {
                         $sSQL .= "UNION ";
                         $sSQL .= "  SELECT place_id ";
-                        $sSQL .= "  FROM location_property_tiger ";
-                        $sSQL .= "  WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
-                        $sSQL .= "    AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
-                        if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',', $this->aAddressRankList).")";
+                        $sSQL .= "  FROM location_property_osmline ".$sWherePlaceId;
                     }
-                    $sSQL .= ") UNION ";
-                    $sSQL .= "  SELECT place_id ";
-                    $sSQL .= "  FROM location_property_osmline ";
-                    $sSQL .= "  WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).")";
-                    $sSQL .= "    AND startnumber is not NULL AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank)";
                     if (CONST_Debug) var_dump($sSQL);
                     $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
                     $tempIDs = array();