X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/6bf90eb22f4aa567a7bc11e1f1ce0986d8ce98c9..16268f92ccd11f0c11fe07b2a001b197af3a5b92:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 273b28cb..cce38de9 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -2,6 +2,7 @@ namespace Nominatim; +require_once(CONST_BasePath.'/lib/NearPoint.php'); require_once(CONST_BasePath.'/lib/PlaceLookup.php'); require_once(CONST_BasePath.'/lib/ReverseGeocode.php'); @@ -32,7 +33,6 @@ class Geocode protected $bFallback = false; protected $aCountryCodes = false; - protected $aNearPoint = false; protected $bBoundedSearch = false; protected $aViewBox = false; @@ -51,10 +51,22 @@ class Geocode protected $sQuery = false; protected $aStructuredQuery = false; + protected $oNormalizer = null; + public function __construct(&$oDB) { $this->oDB =& $oDB; + $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules); + } + + private function normTerm($sTerm) + { + if ($this->oNormalizer === null) { + return $sTerm; + } + + return $this->oNormalizer->transliterate($sTerm); } public function setReverseInPlan($bReverse) @@ -67,19 +79,45 @@ class Geocode $this->aLangPrefOrder = $aLangPref; } - public function getIncludeAddressDetails() + public function getMoreUrlParams() { - return $this->bIncludeAddressDetails; - } + if ($this->aStructuredQuery) { + $aParams = $this->aStructuredQuery; + } else { + $aParams = array('q' => $this->sQuery); + } - public function getIncludeExtraTags() - { - return $this->bIncludeExtraTags; - } + if ($this->aExcludePlaceIDs) { + $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs); + } - public function getIncludeNameDetails() - { - return $this->bIncludeNameDetails; + if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1'; + if ($this->bIncludeExtraTags) $aParams['extratags'] = '1'; + if ($this->bIncludeNameDetails) $aParams['namedetails'] = '1'; + + if ($this->bIncludePolygonAsPoints) $aParams['polygon'] = '1'; + if ($this->bIncludePolygonAsText) $aParams['polygon_text'] = '1'; + if ($this->bIncludePolygonAsGeoJSON) $aParams['polygon_geojson'] = '1'; + if ($this->bIncludePolygonAsKML) $aParams['polygon_kml'] = '1'; + if ($this->bIncludePolygonAsSVG) $aParams['polygon_svg'] = '1'; + + if ($this->fPolygonSimplificationThreshold > 0.0) { + $aParams['polygon_threshold'] = $this->fPolygonSimplificationThreshold; + } + + if ($this->bBoundedSearch) $aParams['bounded'] = '1'; + if (!$this->bDeDupe) $aParams['dedupe'] = '0'; + + if ($this->aCountryCodes) { + $aParams['countrycodes'] = implode(',', $this->aCountryCodes); + } + + if ($this->aViewBox) { + $aParams['viewbox'] = $this->aViewBox[0].','.$this->aViewBox[3] + .','.$this->aViewBox[2].','.$this->aViewBox[1]; + } + + return $aParams; } public function setIncludePolygonAsPoints($b = true) @@ -121,17 +159,6 @@ class Geocode $this->iLimit = $iLimit + min($iLimit, 10); } - public function getExcludedPlaceIDs() - { - return $this->aExcludePlaceIDs; - } - - public function getViewBoxString() - { - if (!$this->aViewBox) return null; - return $this->aViewBox[0].','.$this->aViewBox[3].','.$this->aViewBox[2].','.$this->aViewBox[1]; - } - public function setFeatureType($sFeatureType) { switch ($sFeatureType) { @@ -162,17 +189,17 @@ class Geocode $this->sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; $sSep = ''; - foreach ($this->aRoutePoints as $aPoint) { + foreach ($aRoutePoints as $aPoint) { $fPoint = (float)$aPoint; $this->sViewboxCentreSQL .= $sSep.$fPoint; $sSep = ($sSep == ' ') ? ',' : ' '; } $this->sViewboxCentreSQL .= ")'::geometry,4326)"; - $this->sViewboxSmallSQL = 'st_buffer('.$this->sViewboxCentreSQL; + $this->sViewboxSmallSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL; $this->sViewboxSmallSQL .= ','.($fRouteWidth/69).')'; - $this->sViewboxLargeSQL = 'st_buffer('.$this->sViewboxCentreSQL; + $this->sViewboxLargeSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL; $this->sViewboxLargeSQL .= ','.($fRouteWidth/30).')'; } @@ -180,6 +207,17 @@ class Geocode { $this->aViewBox = array_map('floatval', $aViewbox); + $this->aViewBox[0] = max(-180.0, min(180, $this->aViewBox[0])); + $this->aViewBox[1] = max(-90.0, min(90, $this->aViewBox[1])); + $this->aViewBox[2] = max(-180.0, min(180, $this->aViewBox[2])); + $this->aViewBox[3] = max(-90.0, min(90, $this->aViewBox[3])); + + if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001 + || abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001 + ) { + userError("Bad parameter 'viewbox'. Not a box."); + } + $fHeight = $this->aViewBox[0] - $this->aViewBox[2]; $fWidth = $this->aViewBox[1] - $this->aViewBox[3]; $aBigViewBox[0] = $this->aViewBox[0] + $fHeight; @@ -188,13 +226,20 @@ class Geocode $aBigViewBox[3] = $this->aViewBox[3] - $fWidth; $this->sViewboxCentreSQL = false; - $this->sViewboxSmallSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".$this->aViewBox[0].",".$this->aViewBox[1]."),ST_Point(".$this->aViewBox[2].",".$this->aViewBox[3].")),4326)"; - $this->sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".$aBigViewBox[0].",".$aBigViewBox[1]."),ST_Point(".$aBigViewBox[2].",".$aBigViewBox[3].")),4326)"; - } - - public function setNearPoint($aNearPoint, $fRadiusDeg = 0.1) - { - $this->aNearPoint = array((float)$aNearPoint[0], (float)$aNearPoint[1], (float)$fRadiusDeg); + $this->sViewboxSmallSQL = sprintf( + 'ST_SetSRID(ST_MakeBox2D(ST_Point(%F,%F),ST_Point(%F,%F)),4326)', + $this->aViewBox[0], + $this->aViewBox[1], + $this->aViewBox[2], + $this->aViewBox[3] + ); + $this->sViewboxLargeSQL = sprintf( + 'ST_SetSRID(ST_MakeBox2D(ST_Point(%F,%F),ST_Point(%F,%F)),4326)', + $aBigViewBox[0], + $aBigViewBox[1], + $aBigViewBox[2], + $aBigViewBox[3] + ); } public function setQuery($sQueryString) @@ -252,22 +297,23 @@ class Geocode $aCountries[] = strtolower($sCountryCode); } } - if (isset($aCountryCodes)) + if (isset($aCountries)) $this->aCountryCodes = $aCountries; } $aViewbox = $oParams->getStringList('viewboxlbrt'); if ($aViewbox) { + if (count($aViewbox) != 4) { + userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates."); + } $this->setViewbox($aViewbox); } else { $aViewbox = $oParams->getStringList('viewbox'); if ($aViewbox) { - $this->setViewBox(array( - $aViewbox[0], - $aViewbox[3], - $aViewbox[2], - $aViewbox[1] - )); + if (count($aViewbox) != 4) { + userError("Bad parmater 'viewbox'. Expected 4 coordinates."); + } + $this->setViewBox($aViewbox); } else { $aRoute = $oParams->getStringList('route'); $fRouteWidth = $oParams->getFloat('routewidth'); @@ -311,7 +357,7 @@ class Geocode return true; } - public function setStructuredQuery($sAmentiy = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false) + public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false) { $this->sQuery = false; @@ -321,9 +367,9 @@ class Geocode $this->aAddressRankList = array(); $this->aStructuredQuery = array(); - $this->sAllowedTypesSQLList = ''; + $this->sAllowedTypesSQLList = False; - $this->loadStructuredAddressElement($sAmentiy, 'amenity', 26, 30, false); + $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false); $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false); $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false); $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false); @@ -334,7 +380,7 @@ class Geocode if (sizeof($this->aStructuredQuery) > 0) { $this->sQuery = join(', ', $this->aStructuredQuery); if ($this->iMaxAddressRank < 30) { - $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; + $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } } } @@ -365,40 +411,116 @@ class Geocode //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1 if (sizeof($aPlaceIDs) == 0) return array(); - $sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]"; + $sLanguagePrefArraySQL = getArraySQL( + array_map("getDBQuoted", + $this->aLangPrefOrder) + ); // Get the details for display (is this a redundant extra step?) $sPlaceIDs = join(',', array_keys($aPlaceIDs)); $sImportanceSQL = ''; - if ($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, min(parent_place_id) as parent_place_id, calculated_country_code as country_code,"; - $sSQL .= "get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) as langaddress,"; - $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,"; - $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,"; - if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text as extra,"; - if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text as names,"; - $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(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 "; - if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'"; - if ($this->aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")"; - $sSQL .= ") "; - if ($this->sAllowedTypesSQLList) $sSQL .= "and placex.class in $this->sAllowedTypesSQLList "; - $sSQL .= "and linked_place_id is null "; - $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance"; - if (!$this->bDeDupe) $sSQL .= ",place_id"; - $sSQL .= ",langaddress "; - $sSQL .= ",placename "; - $sSQL .= ",ref "; - if ($this->bIncludeExtraTags) $sSQL .= ",extratags"; - if ($this->bIncludeNameDetails) $sSQL .= ",name"; - $sSQL .= ",extratags->'place' "; + $sImportanceSQLGeom = ''; + if ($this->sViewboxSmallSQL) { + $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * "; + $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, geometry) 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 * "; + $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, geometry) THEN 1 ELSE 0.75 END * "; + } + + $sSQL = "SELECT "; + $sSQL .= " osm_type,"; + $sSQL .= " osm_id,"; + $sSQL .= " class,"; + $sSQL .= " type,"; + $sSQL .= " admin_level,"; + $sSQL .= " rank_search,"; + $sSQL .= " rank_address,"; + $sSQL .= " min(place_id) AS place_id, "; + $sSQL .= " min(parent_place_id) AS parent_place_id, "; + $sSQL .= " country_code, "; + $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,"; + $sSQL .= " get_name_by_language(name, $sLanguagePrefArraySQL) AS placename,"; + $sSQL .= " get_name_by_language(name, ARRAY['ref']) AS ref,"; + if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text AS extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text AS names,"; + $sSQL .= " avg(ST_X(centroid)) AS lon, "; + $sSQL .= " avg(ST_Y(centroid)) AS lat, "; + $sSQL .= " ".$sImportanceSQL."COALESCE(importance,0.75-(rank_search::float/40)) AS importance, "; + $sSQL .= " ( "; + $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; + $sSQL .= " FROM "; + $sSQL .= " place_addressline s, "; + $sSQL .= " placex p"; + $sSQL .= " WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)"; + $sSQL .= " AND p.place_id = s.address_place_id "; + $sSQL .= " AND s.isaddress "; + $sSQL .= " AND p.importance is not null "; + $sSQL .= " ) AS addressimportance, "; + $sSQL .= " (extratags->'place') AS extra_place "; + $sSQL .= " FROM placex"; + $sSQL .= " WHERE place_id in ($sPlaceIDs) "; + $sSQL .= " AND ("; + $sSQL .= " placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank "; + if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) { + $sSQL .= " OR (extratags->'place') = 'city'"; + } + if ($this->aAddressRankList) { + $sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")"; + } + $sSQL .= " ) "; + if ($this->sAllowedTypesSQLList) { + $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList "; + } + $sSQL .= " AND linked_place_id is null "; + $sSQL .= " GROUP BY "; + $sSQL .= " osm_type, "; + $sSQL .= " osm_id, "; + $sSQL .= " class, "; + $sSQL .= " type, "; + $sSQL .= " admin_level, "; + $sSQL .= " rank_search, "; + $sSQL .= " rank_address, "; + $sSQL .= " country_code, "; + $sSQL .= " importance, "; + if (!$this->bDeDupe) $sSQL .= "place_id,"; + $sSQL .= " langaddress, "; + $sSQL .= " placename, "; + $sSQL .= " ref, "; + if ($this->bIncludeExtraTags) $sSQL .= "extratags, "; + if ($this->bIncludeNameDetails) $sSQL .= "name, "; + $sSQL .= " extratags->'place' "; + + // postcode table + $sSQL .= "UNION "; + $sSQL .= "SELECT"; + $sSQL .= " 'P' as osm_type,"; + $sSQL .= " (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,"; + $sSQL .= " 'place' as class, 'postcode' as type,"; + $sSQL .= " null as admin_level, rank_search, rank_address,"; + $sSQL .= " place_id, parent_place_id, country_code,"; + $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,"; + $sSQL .= " postcode as placename,"; + $sSQL .= " postcode as ref,"; + if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "null AS names,"; + $sSQL .= " ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,"; + $sSQL .= $sImportanceSQLGeom."(0.75-(rank_search::float/40)) AS importance, "; + $sSQL .= " ("; + $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; + $sSQL .= " FROM "; + $sSQL .= " place_addressline s, "; + $sSQL .= " placex p"; + $sSQL .= " WHERE s.place_id = lp.parent_place_id"; + $sSQL .= " AND p.place_id = s.address_place_id "; + $sSQL .= " AND s.isaddress"; + $sSQL .= " AND p.importance is not null"; + $sSQL .= " ) AS addressimportance, "; + $sSQL .= " null AS extra_place "; + $sSQL .= "FROM location_postcode lp"; + $sSQL .= " WHERE place_id in ($sPlaceIDs) "; if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) { // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines @@ -411,70 +533,156 @@ class Geocode $sHousenumbers .= "(".$placeID.", ".$housenumber.")"; if ($i<$length) $sHousenumbers .= ", "; } + if (CONST_Use_US_Tiger_Data) { // Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join) $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, min(parent_place_id) as parent_place_id, 'us' as country_code"; - $sSQL .= ", get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) as langaddress "; - $sSQL .= ", null as placename"; - $sSQL .= ", null as ref"; - if ($this->bIncludeExtraTags) $sSQL .= ", null as extra"; - if ($this->bIncludeNameDetails) $sSQL .= ", null as names"; - $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(blub.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 (select place_id"; - // interpolate the Tiger housenumbers here - $sSQL .= ", ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) as centroid, parent_place_id, housenumber_for_place"; - $sSQL .= " from (location_property_tiger "; - $sSQL .= " join (values ".$sHousenumbers.") as housenumbers(place_id, housenumber_for_place) using(place_id)) "; - $sSQL .= " where housenumber_for_place>=0 and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here - $sSQL .= " group by place_id, housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique + $sSQL .= " SELECT "; + $sSQL .= " 'T' AS osm_type, "; + $sSQL .= " (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, "; + $sSQL .= " 'place' AS class, "; + $sSQL .= " 'house' AS type, "; + $sSQL .= " null AS admin_level, "; + $sSQL .= " 30 AS rank_search, "; + $sSQL .= " 30 AS rank_address, "; + $sSQL .= " min(place_id) AS place_id, "; + $sSQL .= " min(parent_place_id) AS parent_place_id, "; + $sSQL .= " 'us' AS country_code, "; + $sSQL .= " get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,"; + $sSQL .= " null AS placename, "; + $sSQL .= " null AS ref, "; + if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "null AS names,"; + $sSQL .= " avg(st_x(centroid)) AS lon, "; + $sSQL .= " avg(st_y(centroid)) AS lat,"; + $sSQL .= " ".$sImportanceSQL."-1.15 AS importance, "; + $sSQL .= " ("; + $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; + $sSQL .= " FROM "; + $sSQL .= " place_addressline s, "; + $sSQL .= " placex p"; + $sSQL .= " WHERE s.place_id = min(blub.parent_place_id)"; + $sSQL .= " AND p.place_id = s.address_place_id "; + $sSQL .= " AND s.isaddress"; + $sSQL .= " AND p.importance is not null"; + $sSQL .= " ) AS addressimportance, "; + $sSQL .= " null AS extra_place "; + $sSQL .= " FROM ("; + $sSQL .= " SELECT place_id, "; // interpolate the Tiger housenumbers here + $sSQL .= " ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, "; + $sSQL .= " parent_place_id, "; + $sSQL .= " housenumber_for_place"; + $sSQL .= " FROM ("; + $sSQL .= " location_property_tiger "; + $sSQL .= " JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) "; + $sSQL .= " WHERE "; + $sSQL .= " housenumber_for_place>=0"; + $sSQL .= " AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank"; + $sSQL .= " ) AS blub"; //postgres wants an alias here + $sSQL .= " GROUP BY"; + $sSQL .= " place_id, "; + $sSQL .= " housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique if (!$this->bDeDupe) $sSQL .= ", place_id "; } // osmline // interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join) - $sSQL .= " union "; - $sSQL .= "select 'W' 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, calculated_country_code as country_code, "; - $sSQL .= "get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) as langaddress, "; - $sSQL .= "null as placename, "; - $sSQL .= "null as ref, "; - if ($this->bIncludeExtraTags) $sSQL .= "null as extra, "; - if ($this->bIncludeNameDetails) $sSQL .= "null as names, "; - $sSQL .= " avg(st_x(centroid)) as lon, avg(st_y(centroid)) as lat,"; - $sSQL .= $sImportanceSQL."-0.1 as importance, "; // slightly smaller than the importance for normal houses with rank 30, which is 0 - $sSQL .= " (select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p"; - $sSQL .= " where s.place_id = min(blub.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 (select place_id, calculated_country_code "; - // interpolate the housenumbers here - $sSQL .= ", CASE WHEN startnumber != endnumber THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) "; - $sSQL .= " ELSE ST_LineInterpolatePoint(linegeo, 0.5) END as centroid"; - $sSQL .= ", parent_place_id, housenumber_for_place "; - $sSQL .= " from (location_property_osmline "; - $sSQL .= " join (values ".$sHousenumbers.") as housenumbers(place_id, housenumber_for_place) using(place_id)) "; - $sSQL .= " where housenumber_for_place>=0 and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here - $sSQL .= " group by place_id, housenumber_for_place, calculated_country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique + $sSQL .= " UNION "; + $sSQL .= "SELECT "; + $sSQL .= " 'W' AS osm_type, "; + $sSQL .= " osm_id, "; + $sSQL .= " 'place' AS class, "; + $sSQL .= " 'house' AS type, "; + $sSQL .= " null AS admin_level, "; + $sSQL .= " 30 AS rank_search, "; + $sSQL .= " 30 AS rank_address, "; + $sSQL .= " min(place_id) as place_id, "; + $sSQL .= " min(parent_place_id) AS parent_place_id, "; + $sSQL .= " country_code, "; + $sSQL .= " get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, "; + $sSQL .= " null AS placename, "; + $sSQL .= " null AS ref, "; + if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, "; + if ($this->bIncludeNameDetails) $sSQL .= "null AS names, "; + $sSQL .= " AVG(st_x(centroid)) AS lon, "; + $sSQL .= " AVG(st_y(centroid)) AS lat, "; + $sSQL .= " ".$sImportanceSQL."-0.1 AS importance, "; // slightly smaller than the importance for normal houses with rank 30, which is 0 + $sSQL .= " ("; + $sSQL .= " SELECT "; + $sSQL .= " MAX(p.importance*(p.rank_address+2)) "; + $sSQL .= " FROM"; + $sSQL .= " place_addressline s, "; + $sSQL .= " placex p"; + $sSQL .= " WHERE s.place_id = min(blub.parent_place_id) "; + $sSQL .= " AND p.place_id = s.address_place_id "; + $sSQL .= " AND s.isaddress "; + $sSQL .= " AND p.importance is not null"; + $sSQL .= " ) AS addressimportance,"; + $sSQL .= " null AS extra_place "; + $sSQL .= " FROM ("; + $sSQL .= " SELECT "; + $sSQL .= " osm_id, "; + $sSQL .= " place_id, "; + $sSQL .= " country_code, "; + $sSQL .= " CASE "; // interpolate the housenumbers here + $sSQL .= " WHEN startnumber != endnumber "; + $sSQL .= " THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) "; + $sSQL .= " ELSE ST_LineInterpolatePoint(linegeo, 0.5) "; + $sSQL .= " END as centroid, "; + $sSQL .= " parent_place_id, "; + $sSQL .= " housenumber_for_place "; + $sSQL .= " FROM ("; + $sSQL .= " location_property_osmline "; + $sSQL .= " JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)"; + $sSQL .= " ) "; + $sSQL .= " WHERE housenumber_for_place>=0 "; + $sSQL .= " AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank"; + $sSQL .= " ) as blub"; //postgres wants an alias here + $sSQL .= " GROUP BY "; + $sSQL .= " osm_id, "; + $sSQL .= " place_id, "; + $sSQL .= " housenumber_for_place, "; + $sSQL .= " country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique if (!$this->bDeDupe) $sSQL .= ", place_id "; if (CONST_Use_Aux_Location_data) { - $sSQL .= " union "; - $sSQL .= "select 'L' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, 0 as rank_search, 0 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, -1, $sLanguagePrefArraySQL) as langaddress, "; - $sSQL .= "null as placename, "; - $sSQL .= "null as ref, "; - if ($this->bIncludeExtraTags) $sSQL .= "null as extra, "; - if ($this->bIncludeNameDetails) $sSQL .= "null as names, "; - $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.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 "; - $sSQL .= "group by place_id"; - if (!$this->bDeDupe) $sSQL .= ", place_id"; - $sSQL .= ", get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) "; + $sSQL .= " UNION "; + $sSQL .= " SELECT "; + $sSQL .= " 'L' AS osm_type, "; + $sSQL .= " place_id AS osm_id, "; + $sSQL .= " 'place' AS class,"; + $sSQL .= " 'house' AS type, "; + $sSQL .= " null AS admin_level, "; + $sSQL .= " 0 AS rank_search,"; + $sSQL .= " 0 AS rank_address, "; + $sSQL .= " min(place_id) AS place_id,"; + $sSQL .= " min(parent_place_id) AS parent_place_id, "; + $sSQL .= " 'us' AS country_code, "; + $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, "; + $sSQL .= " null AS placename, "; + $sSQL .= " null AS ref, "; + if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, "; + if ($this->bIncludeNameDetails) $sSQL .= "null AS names, "; + $sSQL .= " avg(ST_X(centroid)) AS lon, "; + $sSQL .= " avg(ST_Y(centroid)) AS lat, "; + $sSQL .= " ".$sImportanceSQL."-1.10 AS importance, "; + $sSQL .= " ( "; + $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; + $sSQL .= " FROM "; + $sSQL .= " place_addressline s, "; + $sSQL .= " placex p"; + $sSQL .= " WHERE s.place_id = min(location_property_aux.parent_place_id)"; + $sSQL .= " AND p.place_id = s.address_place_id "; + $sSQL .= " AND s.isaddress"; + $sSQL .= " AND p.importance is not null"; + $sSQL .= " ) AS addressimportance, "; + $sSQL .= " null AS extra_place "; + $sSQL .= " FROM location_property_aux "; + $sSQL .= " WHERE place_id in ($sPlaceIDs) "; + $sSQL .= " AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank "; + $sSQL .= " GROUP BY "; + $sSQL .= " place_id, "; + if (!$this->bDeDupe) $sSQL .= "place_id, "; + $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) "; } } @@ -491,7 +699,7 @@ class Geocode return $aSearchResults; } - public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases) + public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery) { /* Calculate all searches using aValidTokens i.e. @@ -504,12 +712,17 @@ class Geocode Score how good the search is so they can be ordered */ - foreach ($aPhrases as $iPhrase => $sPhrase) { + $iGlobalRank = 0; + + foreach ($aPhrases as $iPhrase => $aPhrase) { $aNewPhraseSearches = array(); - if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase]; - else $sPhraseType = ''; + if ($bStructuredPhrases) { + $sPhraseType = $aPhraseTypes[$iPhrase]; + } else { + $sPhraseType = ''; + } - foreach ($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset) { + foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) { // Too many permutations - too expensive if ($iWordSet > 120) break; @@ -520,155 +733,60 @@ class Geocode //echo "
$sToken"; $aNewWordsetSearches = array(); - foreach ($aWordsetSearches as $aCurrentSearch) { + foreach ($aWordsetSearches as $oCurrentSearch) { //echo ""; - //var_dump($aCurrentSearch); + //var_dump($oCurrentSearch); //echo ""; // If the token is valid if (isset($aValidTokens[' '.$sToken])) { + // Recheck if the original word shows up in the query. + $bWordInQuery = false; + if (isset($aSearchTerm['word']) && $aSearchTerm['word']) { + $bWordInQuery = $this->normTerm($aSearchTerm['word'])) !== false; + } foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank']++; - if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0') { - if ($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 (optimisation) - if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases))) { - $aSearch['iSearchRank'] += 5; - } - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) { - if ($aSearch['fLat'] === '') { - $aSearch['fLat'] = $aSearchTerm['lat']; - $aSearch['fLon'] = $aSearchTerm['lon']; - $aSearch['fRadius'] = $aSearchTerm['radius']; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } elseif ($sPhraseType == 'postalcode') { - // 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 (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { - // If we already have a name try putting the postcode first - if (sizeof($aSearch['aName'])) { - $aNewSearch = $aSearch; - $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']); - $aNewSearch['aName'] = array(); - $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch; - } - - if (sizeof($aSearch['aName'])) { - if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) { - $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - } else { - $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - $aSearch['iSearchRank'] += 1000; // skip; - } - } else { - $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - //$aSearch['iNamePhrase'] = $iPhrase; - } - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') { - if ($aSearch['sHouseNumber'] === '') { - $aSearch['sHouseNumber'] = $sToken; - // sanity check: if the housenumber is not mainly made - // up of numbers, add a penalty - if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++; - // also housenumbers should appear in the first or second phrase - if ($iPhrase > 1) $aSearch['iSearchRank'] += 1; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - /* - // Fall back to not searching for this item (better than nothing) - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 1; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - */ + $aNewSearches = $oCurrentSearch->extendWithFullTerm( + $aSearchTerm, + $bWordInQuery, + isset($aValidTokens[$sToken]) + && strpos($sToken, ' ') === false, + $sPhraseType, + $iToken == 0 && $iPhrase == 0, + $iPhrase == 0, + $iToken + 1 == sizeof($aWordset) + && $iPhrase + 1 == sizeof($aPhrases), + $iGlobalRank + ); + + foreach ($aNewSearches as $oSearch) { + if ($oSearch->getRank() < $this->iMaxRank) { + $aNewWordsetSearches[] = $oSearch; } - } elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) { - if ($aSearch['sClass'] === '') { - $aSearch['sOperator'] = $aSearchTerm['operator']; - $aSearch['sClass'] = $aSearchTerm['class']; - $aSearch['sType'] = $aSearchTerm['type']; - if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name'; - else $aSearch['sOperator'] = 'near'; // near = in for the moment - if (strlen($aSearchTerm['operator']) == 0) $aSearch['iSearchRank'] += 1; - - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { - if (sizeof($aSearch['aName'])) { - if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) { - $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - } else { - $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - $aSearch['iSearchRank'] += 1000; // skip; - } - } else { - $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - //$aSearch['iNamePhrase'] = $iPhrase; - } - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } } } // Look for partial matches. // Note that there is no point in adding country terms here - // because country are omitted in the address. + // because country is omitted in the address. if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') { // Allow searching for a word - but at extra cost foreach ($aValidTokens[$sToken] as $aSearchTerm) { - if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { - if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false) { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 1; - if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) { - $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } elseif (isset($aValidTokens[' '.$sToken])) { // revert to the token version? - $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - $aSearch['iSearchRank'] += 1; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - foreach ($aValidTokens[' '.$sToken] as $aSearchTermToken) { - if (empty($aSearchTermToken['country_code']) - && empty($aSearchTermToken['lat']) - && empty($aSearchTermToken['class']) - ) { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 1; - $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id']; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } - } else { - $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } - - if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 1; - if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1; - if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; - if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) { - $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - } else { - $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; - } - $aSearch['iNamePhrase'] = $iPhrase; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; + $aNewSearches = $oCurrentSearch->extendWithPartialTerm( + $aSearchTerm, + $bStructuredPhrases, + $iPhrase, + $aWordFrequencyScores, + isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array() + ); + + foreach ($aNewSearches as $oSearch) { + if ($oSearch->getRank() < $this->iMaxRank) { + $aNewWordsetSearches[] = $oSearch; } } + } - } else { - // Allow skipping a word - but at EXTREAM cost - //$aSearch = $aCurrentSearch; - //$aSearch['iSearchRank']+=100; - //$aNewWordsetSearches[] = $aSearch; } } // Sort and cut @@ -693,9 +811,12 @@ class Geocode // Re-group the searches by their score, junk anything over 20 as just not worth trying $aGroupedSearches = array(); foreach ($aNewPhraseSearches as $aSearch) { - if ($aSearch['iSearchRank'] < $this->iMaxRank) { - if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array(); - $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; + $iRank = $aSearch->getRank(); + if ($iRank < $this->iMaxRank) { + if (!isset($aGroupedSearches[$iRank])) { + $aGroupedSearches[$iRank] = array(); + } + $aGroupedSearches[$iRank][] = $aSearch; } } ksort($aGroupedSearches); @@ -710,6 +831,22 @@ class Geocode //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); } + + // Revisit searches, drop bad searches and give penalty to unlikely combinations. + $aGroupedSearches = array(); + foreach ($aSearches as $oSearch) { + if (!$oSearch->isValidSearch()) { + continue; + } + + $iRank = $oSearch->addToRank($iGlobalRank); + if (!isset($aGroupedSearches[$iRank]) { + $aGroupedSearches[$iRank] = array(); + } + $aGroupedSearches[$iRank][] = $oSearch; + } + ksort($aGroupedSearches); + return $aGroupedSearches; } @@ -747,15 +884,22 @@ class Geocode public function lookup() { - if (!$this->sQuery && !$this->aStructuredQuery) return false; + if (!$this->sQuery && !$this->aStructuredQuery) return array(); - $sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]"; + $sNormQuery = $this->normTerm($this->sQuery); + $sLanguagePrefArraySQL = getArraySQL( + array_map("getDBQuoted", + $this->aLangPrefOrder) + ); $sCountryCodesSQL = false; if ($this->aCountryCodes) { $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes)); } $sQuery = $this->sQuery; + if (!preg_match('//u', $sQuery)) { + userError("Query string is not UTF-8 encoded."); + } // Conflicts between US state abreviations and various words for 'the' in different languages if (isset($this->aLangPrefOrder['name:en'])) { @@ -781,82 +925,67 @@ class Geocode } // Do we have anything that looks like a lat/lon pair? - if ($aLooksLike = looksLikeLatLonPair($sQuery)) { - $this->setNearPoint(array($aLooksLike['lat'], $aLooksLike['lon'])); + $oNearPoint = false; + if ($aLooksLike = NearPoint::extractFromQuery($sQuery)) { + $oNearPoint = $aLooksLike['pt']; $sQuery = $aLooksLike['query']; } $aSearchResults = array(); if ($sQuery || $this->aStructuredQuery) { - // Start with a blank search - $aSearches = array( - array( - 'iSearchRank' => 0, - 'iNamePhrase' => -1, - 'sCountryCode' => false, - 'aName' => array(), - 'aAddress' => array(), - 'aFullNameAddress' => array(), - 'aNameNonSearch' => array(), - 'aAddressNonSearch' => array(), - 'sOperator' => '', - 'aFeatureName' => array(), - 'sClass' => '', - 'sType' => '', - 'sHouseNumber' => '', - 'fLat' => '', - 'fLon' => '', - 'fRadius' => '' - ) - ); - - // Do we have a radius search? - $sNearPointSQL = false; - if ($this->aNearPoint) { - $sNearPointSQL = "ST_SetSRID(ST_Point(".(float)$this->aNearPoint[1].",".(float)$this->aNearPoint[0]."),4326)"; - $aSearches[0]['fLat'] = (float)$this->aNearPoint[0]; - $aSearches[0]['fLon'] = (float)$this->aNearPoint[1]; - $aSearches[0]['fRadius'] = (float)$this->aNearPoint[2]; + // Start with a single blank search + $aSearches = array(new SearchDescription()); + + if ($oNearPoint) { + $aSearches[0]->setNear($oNearPoint); } - // Any 'special' terms in the search? - $bSpecialTerms = false; - preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); - $aSpecialTerms = array(); - foreach ($aSpecialTermsRaw as $aSpecialTerm) { - $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); - $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2]; + if ($sQuery) { + $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery); } - 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']); + $sSpecialTerm = ''; + if ($sQuery) { + preg_match_all( + '/\\[([\\w ]*)\\]/u', + $sQuery, + $aSpecialTermsRaw, + PREG_SET_ORDER + ); + foreach ($aSpecialTermsRaw as $aSpecialTerm) { + $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); + if (!$sSpecialTerm) { + $sSpecialTerm = $aSpecialTerm[1]; + } + } + } + if (!$sSpecialTerm && $this->aStructuredQuery + && isset($this->aStructuredQuery['amenity'])) { + $sSpecialTerm = $this->aStructuredQuery['amenity']; unset($this->aStructuredQuery['amenity']); } - foreach ($aSpecialTermsRaw as $aSpecialTerm) { - $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); - $sToken = chksql($this->oDB->getOne("select make_standard_name('".$aSpecialTerm[1]."') as string")); - $sSQL = 'select * from (select word_id,word_token, word, class, type, country_code, operator'; - $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\')) or country_code is not null'; + if ($sSpecialTerm && !$aSearches[0]->hasOperator()) { + $sSpecialTerm = pg_escape_string($sSpecialTerm); + $sToken = chksql( + $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"), + "Cannot decode query. Wrong encoding?" + ); + $sSQL = 'SELECT class, type FROM word '; + $sSQL .= ' WHERE word_token in (\' '.$sToken.'\')'; + $sSQL .= ' AND class is not null AND class not in (\'place\')'; if (CONST_Debug) var_Dump($sSQL); $aSearchWords = chksql($this->oDB->getAll($sSQL)); $aNewSearches = array(); - foreach ($aSearches as $aSearch) { + foreach ($aSearches as $oSearch) { foreach ($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; - } + $oNewSearch = clone $oSearch; + $oNewSearch->setPoiSearch( + Operator::TYPE, + $aSearchTerm['class'], + $aSearchTerm['type'], + ); + $aNewSearches[] = $oNewSearch; } } $aSearches = $aNewSearches; @@ -879,8 +1008,8 @@ class Geocode $aTokens = array(); foreach ($aPhrases as $iPhrase => $sPhrase) { $aPhrase = chksql( - $this->oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string"), - "Cannot nomralize query string (is it an UTF-8 string?)" + $this->oDB->getRow("SELECT make_standard_name('".pg_escape_string($sPhrase)."') as string"), + "Cannot normalize query string (is it a UTF-8 string?)" ); if (trim($aPhrase['string'])) { $aPhrases[$iPhrase] = $aPhrase; @@ -898,20 +1027,17 @@ class Geocode if (sizeof($aTokens)) { // Check which tokens we have, get the ID numbers - $sSQL = 'select word_id,word_token, word, class, type, country_code, operator, search_name_count'; - $sSQL .= ' from word where word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')'; + $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count'; + $sSQL .= ' FROM word '; + $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')'; if (CONST_Debug) var_Dump($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 $aToken) { @@ -932,21 +1058,9 @@ class Geocode } if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); - // Try and calculate GB postcodes we might be missing + // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code foreach ($aTokens as $sToken) { - // Source of gb postcodes is now definitive - always use - if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) { - if (substr($aData[1], -2, 1) != ' ') { - $aData[0] = substr($aData[0], 0, strlen($aData[1])-1).' '.substr($aData[0], strlen($aData[1])-1); - $aData[1] = substr($aData[1], 0, -1).' '.substr($aData[1], -1, 1); - } - $aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $this->oDB); - if ($aGBPostcodeLocation) { - $aValidTokens[$sToken] = $aGBPostcodeLocation; - } - } elseif (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) { - // US ZIP+4 codes - if there is no token, - // merge in the 5-digit ZIP code + if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) { if (isset($aValidTokens[$aData[1]])) { foreach ($aValidTokens[$aData[1]] as $aToken) { if (!$aToken['class']) { @@ -975,7 +1089,7 @@ class Geocode // array with: placeid => -1 | tiger-housenumber $aResultPlaceIDs = array(); - $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases); + $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery); if ($this->bReverseInPlan) { // Reverse phrase array and also reverse the order of the wordsets in @@ -987,14 +1101,14 @@ class Geocode $aFinalPhrase = end($aPhrases); $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0); } - $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false); + $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery); foreach ($aGroupedSearches as $aSearches) { foreach ($aSearches as $aSearch) { - if ($aSearch['iSearchRank'] < $this->iMaxRank) { - if (!isset($aReverseGroupedSearches[$aSearch['iSearchRank']])) $aReverseGroupedSearches[$aSearch['iSearchRank']] = array(); - $aReverseGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; + if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) { + $aReverseGroupedSearches[$aSearch->getRank()] = array(); } + $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch; } } @@ -1005,39 +1119,9 @@ class Geocode // Re-group the searches by their score, junk anything over 20 as just not worth trying $aGroupedSearches = array(); foreach ($aSearches as $aSearch) { - if ($aSearch['iSearchRank'] < $this->iMaxRank) { - if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array(); - $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; - } - } - ksort($aGroupedSearches); - } - - if (CONST_Debug) var_Dump($aGroupedSearches); - if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) { - $aCopyGroupedSearches = $aGroupedSearches; - foreach ($aCopyGroupedSearches as $iGroup => $aSearches) { - foreach ($aSearches as $iSearch => $aSearch) { - $aReductionsList = array($aSearch['aAddress']); - $iSearchRank = $aSearch['iSearchRank']; - while (sizeof($aReductionsList) > 0) { - $iSearchRank += 5; - if ($iSearchRank > iMaxRank) break 3; - $aNewReductionsList = array(); - foreach ($aReductionsList as $aReductionsWordList) { - for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++) { - $aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1)); - $aReverseSearch = $aSearch; - $aSearch['aAddress'] = $aReductionsWordListResult; - $aSearch['iSearchRank'] = $iSearchRank; - $aGroupedSearches[$iSearchRank][] = $aReverseSearch; - if (sizeof($aReductionsWordListResult) > 0) { - $aNewReductionsList[] = $aReductionsWordListResult; - } - } - } - $aReductionsList = $aNewReductionsList; - } + if ($aSearch->getRank() < $this->iMaxRank) { + if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array(); + $aGroupedSearches[$aSearch->getRank()][] = $aSearch; } } ksort($aGroupedSearches); @@ -1063,359 +1147,89 @@ class Geocode $iQueryLoop = 0; foreach ($aGroupedSearches as $iGroupedRank => $aSearches) { $iGroupLoop++; - foreach ($aSearches as $aSearch) { + foreach ($aSearches as $oSearch) { $iQueryLoop++; $searchedHousenumber = -1; if (CONST_Debug) echo "
Search Loop, group $iGroupLoop, loop $iQueryLoop"; - if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); - - // No location term? - if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon']) { - if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) { - // Just looking for a country by code - look it up - if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) { - $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4"; - if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; - if ($bBoundingBoxSearch) - $sSQL .= " and _st_intersects($this->sViewboxSmallSQL, geometry)"; - $sSQL .= " order by st_area(geometry) desc limit 1"; - if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = chksql($this->oDB->getCol($sSQL)); - } else { - $aPlaceIDs = array(); - } - } else { - if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue; - if (!$aSearch['sClass']) continue; - - $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; - if (chksql($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 ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($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"; - $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) $sSQL .= " and calculated_country_code in ($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 { - $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; - $sSQL .= " and st_contains($this->sViewboxSmallSQL, geometry) and linked_place_id is null"; - if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; - if ($this->sViewboxCentreSQL) $sSQL .= " order by st_distance($this->sViewboxCentreSQL, centroid) asc"; - $sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = chksql($this->oDB->getCol($sSQL)); - } - } - } elseif ($aSearch['fLon'] && !sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['sClass']) { - // If a coordinate is given, the search must either - // be for a name or a special search. Ignore everythin else. - $aPlaceIDs = array(); - } else { - $aPlaceIDs = array(); - - // First we need a position, either aName or fLat or both - $aTerms = array(); - $aOrder = array(); - - if ($aSearch['sHouseNumber'] && sizeof($aSearch['aAddress'])) { - $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M'; - $aOrder[] = ""; - $aOrder[0] = " (exists(select place_id from placex where parent_place_id = search_name.place_id"; - $aOrder[0] .= " and transliteration(housenumber) ~* E'".$sHouseNumberRegex."' limit 1) "; - // also housenumbers from interpolation lines table are needed - $aOrder[0] .= " or exists(select place_id from location_property_osmline where parent_place_id = search_name.place_id"; - $aOrder[0] .= " and ".intval($aSearch['sHouseNumber']).">=startnumber and ".intval($aSearch['sHouseNumber'])."<=endnumber limit 1))"; - $aOrder[0] .= " desc"; - } - - // TODO: filter out the pointless search terms (2 letter name tokens and less) - // they might be right - but they are just too darned expensive to run - if (sizeof($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['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) { - // For infrequent name terms disable index usage for address - if (CONST_Search_NameOnlySearchFrequencyThreshold - && sizeof($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']), ",")."]"; - } else { - $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'], ",")."]"; - 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'])."'"; - if ($aSearch['sHouseNumber']) { - $aTerms[] = "address_rank between 16 and 27"; - } else { - if ($this->iMinAddressRank > 0) { - $aTerms[] = "address_rank >= ".$this->iMinAddressRank; - } - if ($this->iMaxAddressRank < 30) { - $aTerms[] = "address_rank <= ".$this->iMaxAddressRank; - } - } - if ($aSearch['fLon'] && $aSearch['fLat']) { - $aTerms[] = "ST_DWithin(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326), ".$aSearch['fRadius'].")"; - $aOrder[] = "ST_Distance(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326)) ASC"; - } - if (sizeof($this->aExcludePlaceIDs)) { - $aTerms[] = "place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - if ($sCountryCodesSQL) { - $aTerms[] = "country_code in ($sCountryCodesSQL)"; - } - - if ($bBoundingBoxSearch) $aTerms[] = "centroid && $this->sViewboxSmallSQL"; - if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - - if ($aSearch['sHouseNumber']) { - $sImportanceSQL = '- abs(26 - address_rank) + 3'; - } else { - $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)'; + if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens); + + $aPlaceIDs = array(); + if ($oSearch->isCountrySearch()) { + // Just looking for a country - look it up + if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) { + $aPlaceIDs = $oSearch->queryCountry( + $this->oDB, + $bBoundingBoxSearch ? $this->sViewboxSmallSQL : '' + ); } - if ($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'])) { - $sExactMatchSQL = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'], ",").']) INTERSECT select unnest(nameaddress_vector))s) as exactmatch'; - $aOrder[] = 'exactmatch DESC'; - } else { - $sExactMatchSQL = '0::int as exactmatch'; + } elseif (!$oSearch->isNamedSearch()) { + // looking for a POI in a geographic area + if (!$bBoundingBoxSearch && !$oSearch->isNearSearch()) { + continue; } - if (sizeof($aTerms)) { - $sSQL = "select place_id, "; - $sSQL .= $sExactMatchSQL; - $sSQL .= " from search_name"; - $sSQL .= " where ".join(' and ', $aTerms); - $sSQL .= " order by ".join(', ', $aOrder); - if ($aSearch['sHouseNumber'] || $aSearch['sClass']) { - $sSQL .= " limit 20"; - } elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass']) { - $sSQL .= " limit 1"; - } else { - $sSQL .= " limit ".$this->iLimit; - } - - if (CONST_Debug) var_dump($sSQL); - $aViewBoxPlaceIDs = chksql( - $this->oDB->getAll($sSQL), - "Could not get places for search terms." - ); - //var_dump($aViewBoxPlaceIDs); - // Did we have an viewbox matches? - $aPlaceIDs = array(); - $bViewBoxMatch = false; - foreach ($aViewBoxPlaceIDs as $aViewBoxRow) { - //if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break; - //if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break; - //if ($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']; + $aPlaceIDs = $oSearch->queryNearbyPoi( + $this->oDB, + $sCountryCodesSQL, + $bBoundingBoxSearch ? $this->sViewboxSmallSQL : '', + $sViewboxCentreSQL, + $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '', + $this->iLimit + ); + } elseif ($oSearch->isOperator(Operator::POSTCODE)) { + $aPlaceIDs = $oSearch->queryPostcode( + $oDB, + $sCountryCodesSQL, + $this->iLimit + ); + } else { + // Ordinary search: + // First search for places according to name and address. + $aNamedPlaceIDs = $oSearch->queryNamedPlace( + $this->oDB, + $aWordFrequencyScores, + $sCountryCodesSQL, + $this->iMinAddressRank, + $this->iMaxAddressRank, + $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '', + $bBoundingBoxSearch ? $this->sViewboxSmallSQL : '', + $bBoundingBoxSearch ? $this->sViewboxLargeSQL : '', + $this->iLimit + ); + + if (sizeof($aNamedPlaceIDs)) { + foreach ($aNamedPlaceIDs as $aRow) { + $aPlaceIDs[] = $aRow['place_id']; + $this->exactMatchCache[$aRow['place_id']] = $aRow['exactmatch']; } } - //var_Dump($aPlaceIDs); - //exit; //now search for housenumber, if housenumber provided - if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs)) { - $searchedHousenumber = intval($aSearch['sHouseNumber']); - $aRoadPlaceIDs = $aPlaceIDs; - $sPlaceIDs = join(',', $aPlaceIDs); - - // Now they are indexed, look for a house attached to a street we found - $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M'; - $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and transliteration(housenumber) ~* E'".$sHouseNumberRegex."'"; - if (sizeof($this->aExcludePlaceIDs)) { - $sSQL .= " and place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - $sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = chksql($this->oDB->getCol($sSQL)); - - // if nothing found, search in the interpolation line table - if (!sizeof($aPlaceIDs)) { - // do we need to use transliteration and the regex for housenumbers??? - //new query for lines, not housenumbers anymore - if ($searchedHousenumber%2 == 0) { - //if housenumber is even, look for housenumber in streets with interpolationtype even or all - $sSQL = "select distinct place_id from location_property_osmline where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='even' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber"; - } else { - //look for housenumber in streets with interpolationtype odd or all - $sSQL = "select distinct place_id from location_property_osmline where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='odd' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber"; - } - - if (sizeof($this->aExcludePlaceIDs)) { - $sSQL .= " and place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - //$sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - //get place IDs - $aPlaceIDs = chksql($this->oDB->getCol($sSQL, 0)); - } - - // If nothing found try the aux fallback table - if (CONST_Use_Aux_Location_data && !sizeof($aPlaceIDs)) { - $sSQL = "select place_id from location_property_aux where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'"; - if (sizeof($this->aExcludePlaceIDs)) { - $sSQL .= " and parent_place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - //$sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = chksql($this->oDB->getCol($sSQL)); - } - - //if nothing was found in placex or location_property_aux, then search in Tiger data for this housenumber(location_property_tiger) - if (CONST_Use_US_Tiger_Data && !sizeof($aPlaceIDs)) { - //new query for lines, not housenumbers anymore - if ($searchedHousenumber%2 == 0) { - //if housenumber is even, look for housenumber in streets with interpolationtype even or all - $sSQL = "select distinct place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='even' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber"; - } else { - //look for housenumber in streets with interpolationtype odd or all - $sSQL = "select distinct place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='odd' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber"; - } - - if (sizeof($this->aExcludePlaceIDs)) { - $sSQL .= " and place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - //$sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - //get place IDs - $aPlaceIDs = chksql($this->oDB->getCol($sSQL, 0)); - } + if ($oSearch->hasHouseNumber() && sizeof($aPlaceIDs)) { + $aResult = $oSearch->queryHouseNumber( + $this->oDB, + $aPlaceIDs, + $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '' + $this->iLimit + ); - // Fallback to the road (if no housenumber was found) - if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) { - $aPlaceIDs = $aRoadPlaceIDs; - //set to -1, if no housenumbers were found - $searchedHousenumber = -1; + if (sizeof($aResult)) { + $searchedHousenumber = $aResult['iHouseNumber']; + $aPlaceIDs = $aResults['aPlaceIDs']; + } elseif (!$oSearch->looksLikeFullAddress()) { + $aPlaceIDs = array(); } - //else: housenumber was found, remains saved in searchedHousenumber } - - if ($aSearch['sClass'] && sizeof($aPlaceIDs)) { - $sPlaceIDs = join(',', $aPlaceIDs); - $aClassPlaceIDs = array(); - - if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'name') { - // If they were searching for a named class (i.e. 'Kings Head pub') then we might have an extra match - $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; - $sSQL .= " and linked_place_id is null"; - if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; - $sSQL .= " order by rank_search asc limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - $aClassPlaceIDs = chksql($this->oDB->getCol($sSQL)); - } - - if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') { // & in - $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'"; - $bCacheTable = chksql($this->oDB->getOne($sSQL)); - - $sSQL = "select min(rank_search) from placex where place_id in ($sPlaceIDs)"; - - if (CONST_Debug) var_dump($sSQL); - $this->iMaxRank = ((int)chksql($this->oDB->getOne($sSQL))); - - // For state / country level searches the normal radius search doesn't work very well - $sPlaceGeom = false; - if ($this->iMaxRank < 9 && $bCacheTable) { - // Try and get a polygon to search in instead - $sSQL = "select geometry from placex where place_id in ($sPlaceIDs) and rank_search < $this->iMaxRank + 5 and st_geometrytype(geometry) in ('ST_Polygon','ST_MultiPolygon') order by rank_search asc limit 1"; - if (CONST_Debug) var_dump($sSQL); - $sPlaceGeom = chksql($this->oDB->getOne($sSQL)); - } - - if ($sPlaceGeom) { - $sPlaceIDs = false; - } else { - $this->iMaxRank += 5; - $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and rank_search < $this->iMaxRank"; - if (CONST_Debug) var_dump($sSQL); - $aPlaceIDs = chksql($this->oDB->getCol($sSQL)); - $sPlaceIDs = join(',', $aPlaceIDs); - } - - if ($sPlaceIDs || $sPlaceGeom) { - $fRange = 0.01; - if ($bCacheTable) { - // More efficient - can make the range bigger - $fRange = 0.05; - - $sOrderBySQL = ''; - if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.centroid)"; - elseif ($sPlaceIDs) $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)"; - elseif ($sPlaceGeom) $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)"; - - $sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l"; - if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)"; - if ($sPlaceIDs) { - $sSQL .= ",placex as f where "; - $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, f.centroid, $fRange) "; - } - if ($sPlaceGeom) { - $sSQL .= " where "; - $sSQL .= "ST_Contains('".$sPlaceGeom."', l.centroid) "; - } - if (sizeof($this->aExcludePlaceIDs)) { - $sSQL .= " and l.place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)"; - if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc"; - if ($this->iOffset) $sSQL .= " offset $this->iOffset"; - $sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - $aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($this->oDB->getCol($sSQL))); - } else { - 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)"; - - $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where "; - $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, f.centroid, $fRange) "; - $sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' "; - if (sizeof($this->aExcludePlaceIDs)) { - $sSQL .= " and l.place_id not in (".join(',', $this->aExcludePlaceIDs).")"; - } - if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)"; - if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc"; - if ($this->iOffset) $sSQL .= " offset $this->iOffset"; - $sSQL .= " limit $this->iLimit"; - if (CONST_Debug) var_dump($sSQL); - $aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($this->oDB->getCol($sSQL))); - } - } - } - $aPlaceIDs = $aClassPlaceIDs; + // finally get POIs if requested + if ($oSearch->isPoiSearch() && sizeof($aPlaceIDs)) { + $aPlaceIDs = $oSearch->queryPoiByOperator( + $this->oDB, + $aPlaceIDs, + $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '' + $this->iLimit + ); } } @@ -1424,6 +1238,21 @@ class Geocode var_Dump($aPlaceIDs); } + if (sizeof($aPlaceIDs) && $oSearch->getPostcode()) { + $sSQL = 'SELECT place_id FROM placex'; + $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')'; + $sSQL .= " AND postcode = '".$oSearch->getPostcode()."'"; + if (CONST_Debug) var_dump($sSQL); + $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL)); + if ($aFilteredPlaceIDs) { + $aPlaceIDs = $aFilteredPlaceIDs; + if (CONST_Debug) { + echo "
Place IDs after postcode filtering: "; + var_Dump($aPlaceIDs); + } + } + } + foreach ($aPlaceIDs as $iPlaceID) { // array for placeID => -1 | Tiger housenumber $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber; @@ -1435,17 +1264,36 @@ 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 - $sSQL = "select place_id from placex where place_id in (".join(',', array_keys($aResultPlaceIDs)).") "; - $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank "; - if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'"; - if ($this->aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")"; - if (CONST_Use_US_Tiger_Data) { - $sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',', array_keys($aResultPlaceIDs)).") "; - $sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank "; - if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',', $this->aAddressRankList).")"; + $sWherePlaceId = 'WHERE place_id in ('; + $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') '; + + $sSQL = "SELECT place_id "; + $sSQL .= "FROM placex ".$sWherePlaceId; + $sSQL .= " AND ("; + $sSQL .= " placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank "; + if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) { + $sSQL .= " OR (extratags->'place') = 'city'"; + } + if ($this->aAddressRankList) { + $sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")"; + } + $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 "; + $sSQL .= " FROM location_property_osmline ".$sWherePlaceId; } - $sSQL .= ") UNION select place_id from location_property_osmline where place_id in (".join(',', array_keys($aResultPlaceIDs)).")"; - $sSQL .= " and (30 between $this->iMinAddressRank and $this->iMaxAddressRank)"; if (CONST_Debug) var_dump($sSQL); $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL)); $tempIDs = array(); @@ -1471,8 +1319,8 @@ class Geocode $oReverse->setZoom(18); $aLookup = $oReverse->lookup( - (float)$this->aNearPoint[0], - (float)$this->aNearPoint[1], + $oNearPoint->lat(), + $oNearPoint->lon(), false ); @@ -1480,6 +1328,7 @@ class Geocode if ($aLookup['place_id']) { $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1)); + $aResultPlaceIDs[$aLookup['place_id']] = -1; } else { $aSearchResults = array(); } @@ -1499,7 +1348,7 @@ class Geocode $aClassType = getClassTypesWithImportance(); $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery); foreach ($aRecheckWords as $i => $sWord) { - if (!preg_match('/\pL/', $sWord)) unset($aRecheckWords[$i]); + if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]); } if (CONST_Debug) { @@ -1551,7 +1400,7 @@ class Geocode // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]); if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) { - $aResult['address'] = array_merge(array('city' => array_shift(array_values($aResult['address']))), $aResult['address']); + $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']); } }