]> git.openstreetmap.org Git - nominatim.git/blobdiff - website/search.php
Merge branch 'patch-1' of https://github.com/Firefishy/Nominatim
[nominatim.git] / website / search.php
index 2765ab454e6716b2ea66b3c65c441226993991f8..5e2ff371efdaa9ec7b0307b39809e0b4112f8970 100755 (executable)
@@ -36,8 +36,8 @@
        // Show address breakdown
        $bShowAddressDetails = isset($_GET['addressdetails']) && $_GET['addressdetails'];
 
-       // Prefered language    
-       $aLangPrefOrder = getPrefferedLangauges();
+       // Preferred language   
+       $aLangPrefOrder = getPreferredLanguages();
        if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true;
        if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true;
        if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true;
@@ -53,7 +53,7 @@
                }
        }
 
-  // Only certain ranks of feature
+       // Only certain ranks of feature
        if (isset($_GET['featureType']) && !isset($_GET['featuretype'])) $_GET['featuretype'] = $_GET['featureType'];
 
        if (isset($_GET['featuretype']))
                
        // Search query
        $sQuery = (isset($_GET['q'])?trim($_GET['q']):'');
-       if (!$sQuery && $_SERVER['PATH_INFO'] && $_SERVER['PATH_INFO'][0] == '/')
+       if (!$sQuery && isset($_SERVER['PATH_INFO']) && $_SERVER['PATH_INFO'][0] == '/')
        {
                $sQuery = substr($_SERVER['PATH_INFO'], 1);
 
-               // reverse order of '/' seperated string
+               // reverse order of '/' separated string
                $aPhrases = explode('/', $sQuery);              
                $aPhrases = array_reverse($aPhrases); 
                $sQuery = join(', ',$aPhrases);
        }
-
        if ($sQuery)
        {
                $hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder);
                 $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $sQuery);
                if (isset($aLangPrefOrder['name:en']))          
                {
-                       $sQuery = preg_replace('/\\bil\\b/','illinois', $sQuery);
-                       $sQuery = preg_replace('/\\bal\\b/','alabama', $sQuery);
-                       $sQuery = preg_replace('/\\bla\\b/','louisiana', $sQuery);
+                       $sQuery = preg_replace('/,\s*il\s*(,|$)/',', illinois\1', $sQuery);
+                       $sQuery = preg_replace('/,\s*al\s*(,|$)/',', alabama\1', $sQuery);
+                       $sQuery = preg_replace('/,\s*la\s*(,|$)/',', louisiana\1', $sQuery);
                }
 
                // If we have a view box create the SQL
                        $sViewboxSmallSQL = $oDB->getOne($sSQL);
                        if (PEAR::isError($sViewboxSmallSQL))
                        {
-                               var_dump($sViewboxSmallSQL);
-                               exit;
+                               failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL);
                        }
                        $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry";
 
                        $sViewboxLargeSQL = $oDB->getOne($sSQL);
                        if (PEAR::isError($sViewboxLargeSQL))
                        {
-                               var_dump($sViewboxLargeSQL);
-                               exit;
+                               failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL);
                        }
                        $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry";
                }
                // Do we have anything that looks like a lat/lon pair?
                if (preg_match('/\\b([NS])[ ]+([0-9]+[0-9.]*)[ ]+([0-9.]+)?[, ]+([EW])[ ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?\\b/', $sQuery, $aData))
                {
-                       $_GET['nearlat'] = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60);
-                       $_GET['nearlon'] = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60);
-                       $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60);
+                       $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60);
+                       if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
+                       {
+                               $_GET['nearlat'] = $fQueryLat;
+                               $_GET['nearlon'] = $fQueryLon;
+                               $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       }
                }
                elseif (preg_match('/\\b([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([NS])[, ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([EW])\\b/', $sQuery, $aData))
                {
-                       $_GET['nearlat'] = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60);
-                       $_GET['nearlon'] = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
-                       $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       $fQueryLat = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60);
+                       $fQueryLon = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
+                       if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
+                       {
+                               $_GET['nearlat'] = $fQueryLat;
+                               $_GET['nearlon'] = $fQueryLon;
+                               $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       }
                }
-               elseif (preg_match('/(\\[|\\b)(-?[0-9]+[0-9.]*)[, ]+(-?[0-9]+[0-9.]*)(\\]|\\b)/', $sQuery, $aData))
+               elseif (preg_match('/(\\[|^|\\b)(-?[0-9]+[0-9.]*)[, ]+(-?[0-9]+[0-9.]*)(\\]|$|\\b)/', $sQuery, $aData))
                {
-                       $_GET['nearlat'] = $aData[2];
-                       $_GET['nearlon'] = $aData[3];
-                       $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       $fQueryLat = $aData[2];
+                       $fQueryLon = $aData[3];
+                       if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
+                       {
+                               $_GET['nearlat'] = $fQueryLat;
+                               $_GET['nearlon'] = $fQueryLon;
+                               $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
+                       }
                }
 
                if ($sQuery)
                                $sToken = $oDB->getOne("select make_standard_name('".$aSpecialTerm[1]."') as string");
                                $sSQL = 'select * from (select word_id,word_token, word, class, type, location, country_code, operator';
                                $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\',\'highway\')) or country_code is not null';
+                               if (CONST_Debug) var_Dump($sSQL);
                                $aSearchWords = $oDB->getAll($sSQL);
                                $aNewSearches = array();
                                foreach($aSearches as $aSearch)
                                $aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string");
                                if (PEAR::isError($aPhrase))
                                {
-                                       var_dump($aPhrase);
+                                       echo "Illegal query string (not an UTF-8 string): ".$sPhrase;
+                                       if (CONST_Debug) var_dump($aPhrase);
                                        exit;
                                }
                                if (trim($aPhrase['string']))
                                $aDatabaseWords = array();
                        if (PEAR::IsError($aDatabaseWords))
                        {
-                               var_dump($sSQL, $aDatabaseWords);
-                               exit;
+                               failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords);
                        }
                        $aPossibleMainWordIDs = array();
                        foreach($aDatabaseWords as $aToken)
                                }
                        }
                        if ($bSuggestion) $sSuggestion = join(', ',$aSuggestion);
-/*
+
                        // Try and calculate GB postcodes we might be missing
                        foreach($aTokens as $sToken)
                        {
-                               if (!isset($aValidTokens[$sToken]) && !isset($aValidTokens[' '.$sToken]) && preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData))
+                               // Source of gb postcodes is now definitive - always use
+                               if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData))
                                {
                                        if (substr($aData[1],-2,1) != ' ')
                                        {
                                        }
                                }
                        }
-*/
 
                        foreach($aTokens as $sToken)
                        {
                                // Unknown single word token with a number - assume it is a house number
                                if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken,' ') === false && preg_match('/[0-9]/', $sToken))
                                {
-                                       $aValidTokens[' '.$sToken] = array('class'=>'place','type'=>'house');
+                                       $aValidTokens[' '.$sToken] = array(array('class'=>'place','type'=>'house'));
                                }
                        }
 
                                
                                Score how good the search is so they can be ordered
                        */
-
                                foreach($aPhrases as $iPhrase => $sPhrase)
                                {
                                        $aNewPhraseSearches = array();
                                                                                        if ($aSearch['sCountryCode'] === false)
                                                                                        {
                                                                                                $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
-                                                                                               // Country is almost always at the end of the string - increase score for finding it anywhere else (opimisation)
+                                                                                               // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
                                                                                                if ($iWordset+1 != sizeof($aPhrases[$iPhrase]['wordsets']) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5;
                                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
                                                                                }
-                                                                               elseif ($aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
+                                                                               elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
                                                                                {
                                                                                        if ($aSearch['fLat'] === '')
                                                                                        {
                                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                        }
                                                                                }
-                                                                               else
+                                                                               elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
                                                                                {
                                                                                        if (sizeof($aSearch['aName']))
                                                                                        {
                                                                        // Allow searching for a word - but at extra cost
                                                                        foreach($aValidTokens[$sToken] as $aSearchTerm)
                                                                        {
+                                                                               if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
+                                                                               {
 //var_Dump('<hr>',$aSearch['aName']);
 
                                                                                if (sizeof($aCurrentSearch['aName'])  && strlen($sToken) >= 4)
                                                                                        $aSearch['iNamePhrase'] = $iPhrase;
                                                                                if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
                                                                                }
+                                                                               }
                                                                        }
                                                                }
                                                                else
                                                        // First we need a position, either aName or fLat or both
                                                        $aTerms = array();
                                                        $aOrder = array();
+
+                                                       // TODO: filter out the pointless search terms (2 letter name tokens and less)
+                                                       // they might be right - but they are just too darned expensive to run
                                                        if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";
                                                        if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
                                                        if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                                                                $aViewBoxPlaceIDs = $oDB->getAll($sSQL);
                                                                if (PEAR::IsError($aViewBoxPlaceIDs))
                                                                {
-                                                                       var_dump($sSQL, $aViewBoxPlaceIDs);                                     
-                                                                       exit;
+                                                                       failInternalError("Could not get places for search terms.", $sSQL, $aViewBoxPlaceIDs);
                                                                }
 //var_dump($aViewBoxPlaceIDs);
                                                                // Did we have an viewbox matches?
                                                                $sPlaceIDs = join(',',$aPlaceIDs);
        
                                                                // Now they are indexed look for a house attached to a street we found
-                                                               $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-, ]',$aSearch['sHouseNumber']).'\\\\M';                                         
+                                                               $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M';                                                
                                                                $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'";
                                                                if (sizeof($aExcludePlaceIDs))
                                                                {
                                                                                $fRange = 0.05;
 
                                                                                $sOrderBySQL = '';
-                                                                               if ($sNearPointSQL) $sOrderBySQL .= "ST_Distance($sNearPointSQL, l.centroid)";
-                                                                               else if ($sPlaceIDs) $sOrderBySQL .= "ST_Distance(l.centroid, f.geometry)";
-                                                                               else if ($sPlaceGeom) $sOrderBysSQL .= "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
+                                                                               if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.centroid)";
+                                                                               else if ($sPlaceIDs) $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
+                                                                               else if ($sPlaceGeom) $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
                                                                                
                                                                                $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
                                                                                if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
                                                                                if (isset($aSearch['fRadius']) && $aSearch['fRadius']) $fRange = $aSearch['fRadius'];
 
                                                                                $sOrderBySQL = '';
-                                                                               if ($sNearPointSQL) $sOrderBySQL .= "ST_Distance($sNearPointSQL, l.geometry)";
-                                                                               else $sOrderBySQL .= "ST_Distance(l.geometry, f.geometry)";
+                                                                               if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.geometry)";
+                                                                               else $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
 
                                                                                $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where ";
                                                                                $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) ";
 
                                                if (PEAR::IsError($aPlaceIDs))
                                                {
-                                                       var_dump($sSQL, $aPlaceIDs);                                    
-                                                       exit;
+                                                       failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs);
                                                }
 
                                                if (CONST_Debug) var_Dump($aPlaceIDs);
                                        $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance ";
                                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
                                        $sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
+                                       $sSQL .= "and linked_place_id is null ";
                                        $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance";
                                        if (!$bDeDupe) $sSQL .= ",place_id";
                                        $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
 
                                        if (PEAR::IsError($aSearchResults))
                                        {
-                                               var_dump($sSQL, $aSearchResults);                                       
-                                               exit;
+                                               failInternalError("Could not get details for place.", $sSQL, $aSearchResults);
                                        }
                                }
                        } // end if ($sQuery)
 
                                        if (PEAR::IsError($aSearchResults))
                                        {
-                                               var_dump($sSQL, $aSearchResults);                                       
-                                               exit;
+                        failInternalError("Could not get details for place (near).", $sSQL, $aSearchResults);
                                        }
                                }
                        }
 //var_Dump($aSearchResults);
 //exit;
        $aClassType = getClassTypesWithImportance();
+       $aRecheckWords = preg_split('/\b/',$sQuery);
+       foreach($aRecheckWords as $i => $sWord)
+       {
+               if (!$sWord) unset($aRecheckWords[$i]);
+       }
        foreach($aSearchResults as $iResNum => $aResult)
        {
-               if (CONST_Search_AreaPolygons || true)
+               if (CONST_Search_AreaPolygons)
                {
                        // Get the bounding box and outline polygon
                        $sSQL = "select place_id,numfeatures,area,outline,";
-                       $sSQL .= "ST_Y(ST_PointN(ExteriorRing(ST_Box2D(outline)),4)) as minlat,ST_Y(ST_PointN(ExteriorRing(ST_Box2D(outline)),2)) as maxlat,";
-                       $sSQL .= "ST_X(ST_PointN(ExteriorRing(ST_Box2D(outline)),1)) as minlon,ST_X(ST_PointN(ExteriorRing(ST_Box2D(outline)),3)) as maxlon,";
+                       $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(outline)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(outline)),2)) as maxlat,";
+                       $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(outline)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(outline)),3)) as maxlon,";
                        $sSQL .= "ST_AsText(outline) as outlinestring from get_place_boundingbox_quick(".$aResult['place_id'].")";
 
                        $sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,";
-                       $sSQL .= "ST_Y(ST_PointN(ExteriorRing(ST_Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ExteriorRing(ST_Box2D(geometry)),2)) as maxlat,";
-                       $sSQL .= "ST_X(ST_PointN(ExteriorRing(ST_Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ExteriorRing(ST_Box2D(geometry)),3)) as maxlon,";
-                       $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(ST_Box2D(geometry)) = \'ST_Polygon\'';
+                       $sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,";
+                       $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),2)) as maxlat,";
+                       $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon,";
+                       $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\'';
                        $aPointPolygon = $oDB->getRow($sSQL);
                        if (PEAR::IsError($aPointPolygon))
                        {
-                               var_dump($sSQL, $aPointPolygon);
-                               exit;
+                               failInternalError("Could not get outline.", $sSQL, $aPointPolygon);
                        }
                        if ($aPointPolygon['place_id'])
                        {
+                               if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) {
+                                       $aResult['lat'] = $aPointPolygon['centrelat'];
+                                       $aResult['lon'] = $aPointPolygon['centrelon'];
+                               }
                                // Translate geometary string to point array
                                if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['outlinestring'],$aMatch))
                                {
                                }
 
                                // Output data suitable for display (points and a bounding box)
-                               if ($bShowPolygons)
+                               if ($bShowPolygons && isset($aPolyPoints))
                                {
                                        $aResult['aPolyPoints'] = array();
                                        foreach($aPolyPoints as $aPoint)
 //exit;
                }
 
+               // Adjust importance for the number of exact string matches in the result
+               $aResult['importance'] = max(0.001,$aResult['importance']);
+               $iCountWords = 0;
+               $sAddress = $aResult['langaddress'];
+               foreach($aRecheckWords as $i => $sWord)
+               {
+                       if (stripos($sAddress, $sWord)!==false) $iCountWords++;
+               }
+               $aResult['importance'] = $aResult['importance'] + $iCountWords;
+
 //if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']);
 /*
                if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['importance']) 
                $aResult['foundorder'] = $iResNum;
                $aSearchResults[$iResNum] = $aResult;
        }
-       
        uasort($aSearchResults, 'byImportance');
 
 //var_dump($aSearchResults);exit;
                        $bFirst = false;
                }
                if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
-                       && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']])))
+                       && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']])))
                {
                        $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
-                       $aClassTypeNameDone[$aResult['osm_type'].$aResult['osm_class'].$aResult['name']] = true;
+                       $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']] = true;
                        $aSearchResults[] = $aResult;
                }