X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/42f6820a7fe67c5daa4dd3645163ac3611ee8e17..25db5f271addc4ff5aa2d3b32f1582c48dca3405:/lib/Geocode.php?ds=sidebyside diff --git a/lib/Geocode.php b/lib/Geocode.php index 545d3b71..532a7720 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -20,6 +20,7 @@ protected $iLimit = 20; protected $iFinalLimit = 10; protected $iOffset = 0; + protected $bFallback = false; protected $aCountryCodes = false; protected $aNearPoint = false; @@ -34,6 +35,7 @@ protected $iMinAddressRank = 0; protected $iMaxAddressRank = 30; protected $aAddressRankList = array(); + protected $exactMatchCache = array(); protected $sAllowedTypesSQLList = false; @@ -119,6 +121,11 @@ $this->iOffset = $iOffset; } + function setFallback($bFallback = true) + { + $this->bFallback = (bool)$bFallback; + } + function setExcludedPlaceIDs($a) { // TODO: force to int @@ -215,6 +222,11 @@ { $this->sQuery = false; + // Reset + $this->iMinAddressRank = 0; + $this->iMaxAddressRank = 30; + $this->aAddressRankList = array(); + $this->aStructuredQuery = array(); $this->sAllowedTypesSQLList = ''; @@ -234,7 +246,29 @@ $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } } + } + + function fallbackStructuredQuery() + { + if (!$this->aStructuredQuery) return false; + + $aParams = $this->aStructuredQuery; + + if (sizeof($aParams) == 1) return false; + + $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state'); + + foreach($aOrderToFallback as $sType) + { + if (isset($aParams[$sType])) + { + unset($aParams[$sType]); + $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']); + return true; + } + } + return false; } function getDetails($aPlaceIDs) @@ -250,13 +284,13 @@ if ($this->sViewboxSmallSQL) $sImportanceSQL .= " case when ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; if ($this->sViewboxLargeSQL) $sImportanceSQL .= " case when ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; - $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,calculated_country_code as country_code,"; + $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id, calculated_country_code as country_code,"; $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,"; $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; $sSQL .= $sImportanceSQL."coalesce(importance,0.75-(rank_search::float/40)) as importance, "; - $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(placex.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; $sSQL .= "(extratags->'place') as extra_place "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank "; @@ -275,13 +309,13 @@ if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) { $sSQL .= " union "; - $sSQL .= "select 'T' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,"; + $sSQL .= "select 'T' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id,'us' as country_code,"; $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "null as placename,"; $sSQL .= "null as ref,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; $sSQL .= $sImportanceSQL."-1.15 as importance, "; - $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; $sSQL .= "null as extra_place "; $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) "; $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank "; @@ -289,13 +323,13 @@ if (!$this->bDeDupe) $sSQL .= ",place_id "; /* $sSQL .= " union "; - $sSQL .= "select 'L' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,"; + $sSQL .= "select 'L' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id,'us' as country_code,"; $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "null as placename,"; $sSQL .= "null as ref,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; $sSQL .= $sImportanceSQL."-1.10 as importance, "; - $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; $sSQL .= "null as extra_place "; $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) "; $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank "; @@ -317,6 +351,36 @@ return $aSearchResults; } + /* Perform the actual query lookup. + + Returns an ordered list of results, each with the following fields: + osm_type: type of corresponding OSM object + N - node + W - way + R - relation + P - postcode (internally computed) + osm_id: id of corresponding OSM object + class: general object class (corresponds to tag key of primary OSM tag) + type: subclass of object (corresponds to tag value of primary OSM tag) + admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level + rank_search: rank in search hierarchy + (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level) + rank_address: rank in address hierarchy (determines orer in address) + place_id: internal key (may differ between different instances) + country_code: ISO country code + langaddress: localized full address + placename: localized name of object + ref: content of ref tag (if available) + lon: longitude + lat: latitude + importance: importance of place based on Wikipedia link count + addressimportance: cumulated importance of address elements + extra_place: type of place (for admin boundaries, if there is a place tag) + aBoundingBox: bounding Box + label: short description of the object class/type (English only) + name: full name (currently the same as langaddress) + foundorder: secondary ordering for places with same importance + */ function lookup() { if (!$this->sQuery && !$this->aStructuredQuery) return false; @@ -330,7 +394,8 @@ } // Hack to make it handle "new york, ny" (and variants) correctly - $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $this->sQuery); + //$sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $this->sQuery); + $sQuery = $this->sQuery; // Conflicts between US state abreviations and various words for 'the' in different languages if (isset($this->aLangPrefOrder['name:en'])) @@ -361,11 +426,12 @@ if ($this->aRoutePoints) { $sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; - $bFirst = false; - foreach($this->aRouteaPoints as $aPoint) + $bFirst = true; + foreach($this->aRoutePoints as $aPoint) { if (!$bFirst) $sViewboxCentreSQL .= ","; $sViewboxCentreSQL .= $aPoint[1].' '.$aPoint[0]; + $bFirst = false; } $sViewboxCentreSQL .= ")'::geometry,4326)"; @@ -1083,7 +1149,7 @@ // TODO: filter out the pointless search terms (2 letter name tokens and less) // they might be right - but they are just too darned expensive to run if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]"; - if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]"; + //if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]"; if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) { // For infrequent name terms disable index usage for address @@ -1091,12 +1157,13 @@ sizeof($aSearch['aName']) == 1 && $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold) { - $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]"; + //$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]"; + $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]"; } else { $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]"; - if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]"; + //if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]"; } } if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; @@ -1124,12 +1191,16 @@ $aOrder[] = "$sImportanceSQL DESC"; if (sizeof($aSearch['aFullNameAddress'])) { - $aOrder[] = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) DESC'; + $sExactMatchSQL = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) as exactmatch'; + $aOrder[] = 'exactmatch DESC'; + } else { + $sExactMatchSQL = '0::int as exactmatch'; } if (sizeof($aTerms)) { - $sSQL = "select place_id"; + $sSQL = "select place_id, "; + $sSQL .= $sExactMatchSQL; $sSQL .= " from search_name"; $sSQL .= " where ".join(' and ',$aTerms); $sSQL .= " order by ".join(', ',$aOrder); @@ -1157,6 +1228,7 @@ //if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1; //else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2; $aPlaceIDs[] = $aViewBoxRow['place_id']; + $this->exactMatchCache[$aViewBoxRow['place_id']] = $aViewBoxRow['exactmatch']; } } //var_Dump($aPlaceIDs); @@ -1385,6 +1457,14 @@ // No results? Done if (!sizeof($aSearchResults)) { + if ($this->bFallback) + { + if ($this->fallbackStructuredQuery()) + { + return $this->lookup(); + } + } + return array(); } @@ -1435,10 +1515,12 @@ { preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); } + /* elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch)) { preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); } + */ elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch)) { $fRadius = 0.01; @@ -1556,7 +1638,24 @@ $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right $aResult['name'] = $aResult['langaddress']; - $aResult['foundorder'] = -$aResult['addressimportance']; + // secondary ordering (for results with same importance (the smaller the better): + // - approximate importance of address parts + $aResult['foundorder'] = -$aResult['addressimportance']/10; + // - number of exact matches from the query + if (isset($this->exactMatchCache[$aResult['place_id']])) + $aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']]; + else if (isset($this->exactMatchCache[$aResult['parent_place_id']])) + $aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']]; + // - importance of the class/type + if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) + && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']) + { + $aResult['foundorder'] = $aResult['foundorder'] + 0.000001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance']; + } + else + { + $aResult['foundorder'] = $aResult['foundorder'] + 0.001; + } $aSearchResults[$iResNum] = $aResult; } uasort($aSearchResults, 'byImportance');