X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/c371bbcfe45904ab56ece73f709b75e5ef6502fd..33f7bba69858405517c68a4aaf609ce69e00e277:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 545d3b71..016fc101 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 @@ -197,6 +204,98 @@ return $this->sQuery; } + + function loadParamArray($aParams) + { + if (isset($aParams['addressdetails'])) $this->bIncludeAddressDetails = (bool)$aParams['addressdetails']; + if (isset($aParams['bounded'])) $this->bBoundedSearch = (bool)$aParams['bounded']; + if (isset($aParams['dedupe'])) $this->bDeDupe = (bool)$aParams['dedupe']; + + if (isset($aParams['limit'])) $this->setLimit((int)$aParams['limit']); + if (isset($aParams['offset'])) $this->iOffset = (int)$aParams['offset']; + + if (isset($aParams['fallback'])) $this->bFallback = (bool)$aParams['fallback']; + + // List of excluded Place IDs - used for more acurate pageing + if (isset($aParams['exclude_place_ids']) && $aParams['exclude_place_ids']) + { + foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID) + { + $iExcludedPlaceID = (int)$iExcludedPlaceID; + if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID; + } + $this->aExcludePlaceIDs = $aExcludePlaceIDs; + } + + // Only certain ranks of feature + if (isset($aParams['featureType'])) $this->setFeatureType($aParams['featureType']); + if (isset($aParams['featuretype'])) $this->setFeatureType($aParams['featuretype']); + + // Country code list + if (isset($aParams['countrycodes'])) + { + $aCountryCodes = array(); + foreach(explode(',',$aParams['countrycodes']) as $sCountryCode) + { + if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) + { + $aCountryCodes[] = strtolower($sCountryCode); + } + } + $this->aCountryCodes = $aCountryCodes; + } + + if (isset($aParams['viewboxlbrt']) && $aParams['viewboxlbrt']) + { + $aCoOrdinatesLBRT = explode(',',$aParams['viewboxlbrt']); + $this->setViewBox($aCoOrdinatesLBRT[0], $aCoOrdinatesLBRT[1], $aCoOrdinatesLBRT[2], $aCoOrdinatesLBRT[3]); + } + else if (isset($aParams['viewbox']) && $aParams['viewbox']) + { + $aCoOrdinatesLTRB = explode(',',$aParams['viewbox']); + $this->setViewBox($aCoOrdinatesLTRB[0], $aCoOrdinatesLTRB[3], $aCoOrdinatesLTRB[2], $aCoOrdinatesLTRB[1]); + } + + if (isset($aParams['route']) && $aParams['route'] && isset($aParams['routewidth']) && $aParams['routewidth']) + { + $aPoints = explode(',',$aParams['route']); + if (sizeof($aPoints) % 2 != 0) + { + userError("Uneven number of points"); + exit; + } + $fPrevCoord = false; + $aRoute = array(); + foreach($aPoints as $i => $fPoint) + { + if ($i%2) + { + $aRoute[] = array((float)$fPoint, $fPrevCoord); + } + else + { + $fPrevCoord = (float)$fPoint; + } + } + $this->aRoutePoints = $aRoute; + } + } + + function setQueryFromParams($aParams) + { + // Search query + $sQuery = (isset($aParams['q'])?trim($aParams['q']):''); + if (!$sQuery) + { + $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']); + $this->setReverseInPlan(false); + } + else + { + $this->setQuery($sQuery); + } + } + function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues) { $sValue = trim($sValue); @@ -215,6 +314,11 @@ { $this->sQuery = false; + // Reset + $this->iMinAddressRank = 0; + $this->iMaxAddressRank = 30; + $this->aAddressRankList = array(); + $this->aStructuredQuery = array(); $this->sAllowedTypesSQLList = ''; @@ -234,7 +338,29 @@ $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } } + } + + function fallbackStructuredQuery() + { + if (!$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($aPlaceIDs) @@ -250,13 +376,13 @@ if ($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 +401,13 @@ 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 $this->iMaxAddressRank "; @@ -289,13 +415,13 @@ if (!$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 +443,36 @@ 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; @@ -330,14 +486,15 @@ } // 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'])) { - $sQuery = preg_replace('/,\s*il\s*(,|$)/',', illinois\1', $sQuery); - $sQuery = preg_replace('/,\s*al\s*(,|$)/',', alabama\1', $sQuery); - $sQuery = preg_replace('/,\s*la\s*(,|$)/',', louisiana\1', $sQuery); + $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/','\1illinois\2', $sQuery); + $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/','\1alabama\2', $sQuery); + $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/','\1louisiana\2', $sQuery); } // View Box SQL @@ -361,11 +518,12 @@ if ($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; } $sViewboxCentreSQL .= ")'::geometry,4326)"; @@ -1083,7 +1241,7 @@ // TODO: filter out the pointless search terms (2 letter name tokens and less) // they might be right - but they are just too darned expensive to run if (sizeof($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'] != $aSearch['aAddress']) { // For infrequent name terms disable index usage for address @@ -1091,12 +1249,13 @@ sizeof($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'])."'"; @@ -1118,18 +1277,30 @@ if ($bBoundingBoxSearch) $aTerms[] = "centroid && $this->sViewboxSmallSQL"; if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)'; + if ($aSearch['sHouseNumber']) + { + $sImportanceSQL = '- abs(26 - address_rank) + 3'; + } + else + { + $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)'; + } if ($this->sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($this->sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END"; if ($this->sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($this->sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END"; + $aOrder[] = "$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 +1328,7 @@ //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 +1557,14 @@ // No results? Done if (!sizeof($aSearchResults)) { + if ($this->bFallback) + { + if ($this->fallbackStructuredQuery()) + { + return $this->lookup(); + } + } + return array(); } @@ -1435,10 +1615,12 @@ { preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$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; @@ -1514,7 +1696,7 @@ $aResult['aPolyPoints'][] = array($aPoint[1], $aPoint[2]); } } - $aResult['aBoundingBox'] = array($aPointPolygon['minlat'],$aPointPolygon['maxlat'],$aPointPolygon['minlon'],$aPointPolygon['maxlon']); + $aResult['aBoundingBox'] = array((string)$aPointPolygon['minlat'],(string)$aPointPolygon['maxlat'],(string)$aPointPolygon['minlon'],(string)$aPointPolygon['maxlon']); } // Is there an icon set for this type of result? @@ -1556,7 +1738,24 @@ $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');