X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/d72c8633531c859ebf20ed366f1d6976853ffe0d..c11d1d78e90c1c7338ad9e67ae2c457773452aa5:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index f3cc10da..7a9f5ad4 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -2,29 +2,22 @@ namespace Nominatim; -require_once(CONST_BasePath.'/lib/NearPoint.php'); require_once(CONST_BasePath.'/lib/PlaceLookup.php'); +require_once(CONST_BasePath.'/lib/Phrase.php'); require_once(CONST_BasePath.'/lib/ReverseGeocode.php'); +require_once(CONST_BasePath.'/lib/SearchDescription.php'); +require_once(CONST_BasePath.'/lib/SearchContext.php'); +require_once(CONST_BasePath.'/lib/TokenList.php'); class Geocode { protected $oDB; - protected $aLangPrefOrder = array(); - - protected $bIncludeAddressDetails = false; - protected $bIncludeExtraTags = false; - protected $bIncludeNameDetails = false; + protected $oPlaceLookup; - protected $bIncludePolygonAsPoints = false; - protected $bIncludePolygonAsText = false; - protected $bIncludePolygonAsGeoJSON = false; - protected $bIncludePolygonAsKML = false; - protected $bIncludePolygonAsSVG = false; - protected $fPolygonSimplificationThreshold = 0.0; + protected $aLangPrefOrder = array(); protected $aExcludePlaceIDs = array(); - protected $bDeDupe = true; protected $bReverseInPlan = false; protected $iLimit = 20; @@ -36,15 +29,13 @@ class Geocode protected $bBoundedSearch = false; protected $aViewBox = false; - protected $sViewboxCentreSQL = false; - protected $sViewboxSmallSQL = false; - protected $sViewboxLargeSQL = false; + protected $aRoutePoints = false; + protected $aRouteWidth = false; protected $iMaxRank = 20; protected $iMinAddressRank = 0; protected $iMaxAddressRank = 30; protected $aAddressRankList = array(); - protected $exactMatchCache = array(); protected $sAllowedTypesSQLList = false; @@ -57,6 +48,7 @@ class Geocode public function __construct(&$oDB) { $this->oDB =& $oDB; + $this->oPlaceLookup = new PlaceLookup($this->oDB); $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules); } @@ -87,69 +79,25 @@ class Geocode $aParams = array('q' => $this->sQuery); } + $aParams = array_merge($aParams, $this->oPlaceLookup->getMoreUrlParams()); + if ($this->aExcludePlaceIDs) { $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs); } - if ($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]; + $aParams['viewbox'] = join(',', $this->aViewBox); } return $aParams; } - public function setIncludePolygonAsPoints($b = true) - { - $this->bIncludePolygonAsPoints = $b; - } - - public function setIncludePolygonAsText($b = true) - { - $this->bIncludePolygonAsText = $b; - } - - public function setIncludePolygonAsGeoJSON($b = true) - { - $this->bIncludePolygonAsGeoJSON = $b; - } - - public function setIncludePolygonAsKML($b = true) - { - $this->bIncludePolygonAsKML = $b; - } - - public function setIncludePolygonAsSVG($b = true) - { - $this->bIncludePolygonAsSVG = $b; - } - - public function setPolygonSimplificationThreshold($f) - { - $this->fPolygonSimplificationThreshold = $f; - } - public function setLimit($iLimit = 10) { if ($iLimit > 50) $iLimit = 50; @@ -183,63 +131,43 @@ class Geocode $this->iMaxAddressRank = $iMax; } - public function setRoute($aRoutePoints, $fRouteWidth) + public function setViewbox($aViewbox) { - $this->aViewBox = false; - - $this->sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; - $sSep = ''; - foreach ($aRoutePoints as $aPoint) { - $fPoint = (float)$aPoint; - $this->sViewboxCentreSQL .= $sSep.$fPoint; - $sSep = ($sSep == ' ') ? ',' : ' '; - } - $this->sViewboxCentreSQL .= ")'::geometry,4326)"; + $aBox = array_map('floatval', $aViewbox); - $this->sViewboxSmallSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL; - $this->sViewboxSmallSQL .= ','.($fRouteWidth/69).')'; + $this->aViewBox[0] = max(-180.0, min($aBox[0], $aBox[2])); + $this->aViewBox[1] = max(-90.0, min($aBox[1], $aBox[3])); + $this->aViewBox[2] = min(180.0, max($aBox[0], $aBox[2])); + $this->aViewBox[3] = min(90.0, max($aBox[1], $aBox[3])); - $this->sViewboxLargeSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL; - $this->sViewboxLargeSQL .= ','.($fRouteWidth/30).')'; + if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001 + || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001 + ) { + userError("Bad parameter 'viewbox'. Not a box."); + } } - public function setViewbox($aViewbox) + private function viewboxImportanceFactor($fX, $fY) { - $this->aViewBox = array_map('floatval', $aViewbox); + if (!$this->aViewBox) { + return 1; + } - $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])); + $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2; + $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2; - if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001 - || abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001 - ) { - userError("Bad parameter 'viewbox'. Not a box."); + $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2); + $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2); + + if ($fXDist <= $fWidth && $fYDist <= $fHeight) { + return 1; } - $fHeight = $this->aViewBox[0] - $this->aViewBox[2]; - $fWidth = $this->aViewBox[1] - $this->aViewBox[3]; - $aBigViewBox[0] = $this->aViewBox[0] + $fHeight; - $aBigViewBox[2] = $this->aViewBox[2] - $fHeight; - $aBigViewBox[1] = $this->aViewBox[1] + $fWidth; - $aBigViewBox[3] = $this->aViewBox[3] - $fWidth; - - $this->sViewboxCentreSQL = false; - $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], - $aBigViewBox[3] - ); + if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) { + return 0.5; + } + + return 0.25; } public function setQuery($sQueryString) @@ -254,17 +182,9 @@ class Geocode } - public function loadParamArray($oParams) + public function loadParamArray($oParams, $sForceGeometryType = null) { - $this->bIncludeAddressDetails - = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails); - $this->bIncludeExtraTags - = $oParams->getBool('extratags', $this->bIncludeExtraTags); - $this->bIncludeNameDetails - = $oParams->getBool('namedetails', $this->bIncludeNameDetails); - $this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch); - $this->bDeDupe = $oParams->getBool('dedupe', $this->bDeDupe); $this->setLimit($oParams->getInt('limit', $this->iFinalLimit)); $this->iOffset = $oParams->getInt('offset', $this->iOffset); @@ -304,24 +224,29 @@ class Geocode $aViewbox = $oParams->getStringList('viewboxlbrt'); if ($aViewbox) { if (count($aViewbox) != 4) { - userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates."); + userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates."); } $this->setViewbox($aViewbox); } else { $aViewbox = $oParams->getStringList('viewbox'); if ($aViewbox) { if (count($aViewbox) != 4) { - userError("Bad parmater 'viewbox'. Expected 4 coordinates."); + userError("Bad parameter 'viewbox'. Expected 4 coordinates."); } $this->setViewBox($aViewbox); } else { $aRoute = $oParams->getStringList('route'); $fRouteWidth = $oParams->getFloat('routewidth'); if ($aRoute && $fRouteWidth) { - $this->setRoute($aRoute, $fRouteWidth); + $this->aRoutePoints = $aRoute; + $this->aRouteWidth = $fRouteWidth; } } } + + $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType); + $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon')); + $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false)); } public function setQueryFromParams($oParams) @@ -367,7 +292,7 @@ class Geocode $this->aAddressRankList = array(); $this->aStructuredQuery = array(); - $this->sAllowedTypesSQLList = False; + $this->sAllowedTypesSQLList = false; $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false); $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false); @@ -377,7 +302,7 @@ class Geocode $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11)); $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false); - if (sizeof($this->aStructuredQuery) > 0) { + if (!empty($this->aStructuredQuery)) { $this->sQuery = join(', ', $this->aStructuredQuery); if ($this->iMaxAddressRank < 30) { $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')'; @@ -391,7 +316,7 @@ class Geocode $aParams = $this->aStructuredQuery; - if (sizeof($aParams) == 1) return false; + if (count($aParams) == 1) return false; $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state'); @@ -406,303 +331,10 @@ class Geocode return false; } - public function getDetails($aPlaceIDs) - { - //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1 - if (sizeof($aPlaceIDs) == 0) return array(); - - $sLanguagePrefArraySQL = getArraySQL( - array_map("getDBQuoted", - $this->aLangPrefOrder) - ); - - // Get the details for display (is this a redundant extra step?) - $sPlaceIDs = join(',', array_keys($aPlaceIDs)); - - $sImportanceSQL = ''; - $sImportanceSQLGeom = ''; - if ($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 - // with start- and endnumber, the common osm housenumbers are usually saved as points - $sHousenumbers = ""; - $i = 0; - $length = count($aPlaceIDs); - foreach ($aPlaceIDs as $placeID => $housenumber) { - $i++; - $sHousenumbers .= "(".$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 "; - $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 "; - $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 "; - $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) "; - } - } - - $sSQL .= " order by importance desc"; - if (CONST_Debug) { - echo "
"; - var_dump($sSQL); - } - $aSearchResults = chksql( - $this->oDB->getAll($sSQL), - "Could not get details for place." - ); - - return $aSearchResults; - } - - public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery) + public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured) { /* - Calculate all searches using aValidTokens i.e. + Calculate all searches using oValidTokens i.e. 'Wodsworth Road, Sheffield' => Phrase Wordset @@ -712,17 +344,11 @@ class Geocode Score how good the search is so they can be ordered */ - $iGlobalRank = 0; - - foreach ($aPhrases as $iPhrase => $aPhrase) { + foreach ($aPhrases as $iPhrase => $oPhrase) { $aNewPhraseSearches = array(); - if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase]; - else $sPhraseType = ''; - - foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) { - // Too many permutations - too expensive - if ($iWordSet > 120) break; + $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : ''; + foreach ($oPhrase->getWordSets() as $aWordset) { $aWordsetSearches = $aSearches; // Add all words from this wordset @@ -730,169 +356,60 @@ 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])) { - // TODO variable should go into aCurrentSearch - $bHavePostcode = false; - foreach ($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[] = $aSearch; - // If it is at the beginning, we can be almost sure that this is the wrong order - // Increase score for all searches. - if ($iToken == 0 && $iPhrase == 0) { - $iGlobalRank++; - } - } - } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') { - // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both - if ($aSearch['sPostcode'] === '' && - isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) { - // If we have structured search or this is the first term, - // make the postcode the primary search element. - if (!$bHavePostcode && $aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) { - $aNewSearch = $aSearch; - $aNewSearch['sOperator'] = 'postcode'; - $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']); - $aNewSearch['aName'] = array($aSearchTerm['word_id'] => $aSearchTerm['word']); - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch; - $bHavePostcode = true; - } - - // If we have a structured search or this is not the first term, - // add the postcode as an addendum. - if ($aSearch['sOperator'] !== 'postcode' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']))) { - $aSearch['sPostcode'] = $aSearchTerm['word']; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } - } elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') { - if ($aSearch['sHouseNumber'] === '' && $aSearch['sOperator'] !== 'postcode') { - $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) $aSearch['iSearchRank']++; - // also must not appear in the middle of the address - if ($aSearch['aAddress'] || $aSearch['aAddressNonSearch']) $aSearch['iSearchRank'] += 1; - if ($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[] = $aSearch; - */ - } - } elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) { - // require a normalized exact match of the term - // if we have the normalizer version of the query - // available - if ($aSearch['sOperator'] === '' - && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) { - $aSearch['sClass'] = $aSearchTerm['class']; - $aSearch['sType'] = $aSearchTerm['type']; - if ($aSearchTerm['operator'] == '') { - $aSearch['sOperator'] = sizeof($aSearch['aName']) ? 'name' : 'near'; - $aSearch['iSearchRank'] += 2; - } else { - $aSearch['sOperator'] = 'near'; // near = in for the moment - } - - if ($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[] = $aSearch; + // Tokens with full name matches. + foreach ($oValidTokens->get(' '.$sToken) as $oSearchTerm) { + $aNewSearches = $oCurrentSearch->extendWithFullTerm( + $oSearchTerm, + $oValidTokens->contains($sToken) + && strpos($sToken, ' ') === false, + $sPhraseType, + $iToken == 0 && $iPhrase == 0, + $iPhrase == 0, + $iToken + 1 == count($aWordset) + && $iPhrase + 1 == count($aPhrases) + ); + + foreach ($aNewSearches as $oSearch) { + if ($oSearch->getRank() < $this->iMaxRank) { + $aNewWordsetSearches[] = $oSearch; } } } // Look for partial matches. // Note that there is no point in adding country terms here - // because country are omitted in the address. - if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') { + // because country is omitted in the address. + if ($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[] = $aSearch; - } - } - - if ((!$aCurrentSearch['sPostcode'] && !$aCurrentSearch['aAddress'] && !$aCurrentSearch['aAddressNonSearch']) - && (!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[] = $aSearch; + foreach ($oValidTokens->get($sToken) as $oSearchTerm) { + $aNewSearches = $oCurrentSearch->extendWithPartialTerm( + $sToken, + $oSearchTerm, + $bIsStructured, + $iPhrase, + $oValidTokens->get(' '.$sToken) + ); + + 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[] = $aSearch; } } // Sort and cut - usort($aNewWordsetSearches, 'bySearchRank'); + usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank')); $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50); } - //var_Dump('
',sizeof($aWordsetSearches)); exit; + //var_Dump('
',count($aWordsetSearches)); exit; $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches); - usort($aNewPhraseSearches, 'bySearchRank'); + usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank')); $aSearchHash = array(); foreach ($aNewPhraseSearches as $iSearch => $aSearch) { @@ -907,9 +424,12 @@ 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); @@ -917,29 +437,24 @@ class Geocode $iSearchCount = 0; $aSearches = array(); foreach ($aGroupedSearches as $iScore => $aNewSearches) { - $iSearchCount += sizeof($aNewSearches); + $iSearchCount += count($aNewSearches); $aSearches = array_merge($aSearches, $aNewSearches); if ($iSearchCount > 50) break; } - - //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); } // Revisit searches, drop bad searches and give penalty to unlikely combinations. $aGroupedSearches = array(); - foreach ($aSearches as $aSearch) { - if (!$aSearch['aName']) { - if ($aSearch['sHouseNumber']) { - continue; - } - } - if ($this->aCountryCodes && $aSearch['sCountryCode'] - && !in_array($aSearch['sCountryCode'], $this->aCountryCodes)) { + foreach ($aSearches as $oSearch) { + if (!$oSearch->isValidSearch()) { continue; } - $aSearch['iSearchRank'] += $iGlobalRank; - $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch; + $iRank = $oSearch->getRank(); + if (!isset($aGroupedSearches[$iRank])) { + $aGroupedSearches[$iRank] = array(); + } + $aGroupedSearches[$iRank][] = $oSearch; } ksort($aGroupedSearches); @@ -957,9 +472,9 @@ class Geocode 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 + admin_level: see https://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) + (see also https://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 @@ -980,61 +495,57 @@ class Geocode public function lookup() { + Debug::newFunction('Geocode::lookup'); if (!$this->sQuery && !$this->aStructuredQuery) return array(); - $sNormQuery = $this->normTerm($this->sQuery); - $sLanguagePrefArraySQL = getArraySQL( - array_map("getDBQuoted", - $this->aLangPrefOrder) - ); - $sCountryCodesSQL = false; + Debug::printDebugArray('Geocode', $this); + + $oCtx = new SearchContext(); + + if ($this->aRoutePoints) { + $oCtx->setViewboxFromRoute( + $this->oDB, + $this->aRoutePoints, + $this->aRouteWidth, + $this->bBoundedSearch + ); + } elseif ($this->aViewBox) { + $oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch); + } + if ($this->aExcludePlaceIDs) { + $oCtx->setExcludeList($this->aExcludePlaceIDs); + } if ($this->aCountryCodes) { - $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes)); + $oCtx->setCountryList($this->aCountryCodes); } + Debug::newSection('Query Preprocessing'); + + $sNormQuery = $this->normTerm($this->sQuery); + Debug::printVar('Normalized query', $sNormQuery); + + $sLanguagePrefArraySQL = $this->oDB->getArraySQL( + $this->oDB->getDBQuotedList($this->aLangPrefOrder) + ); + $sQuery = $this->sQuery; if (!preg_match('//u', $sQuery)) { - userError("Query string is not UTF-8 encoded."); + userError('Query string is not UTF-8 encoded.'); } // Conflicts between US state abreviations and various words for 'the' in different languages if (isset($this->aLangPrefOrder['name:en'])) { - $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery); - $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery); - $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery); - } - - $bBoundingBoxSearch = $this->bBoundedSearch && $this->sViewboxSmallSQL; - if ($this->sViewboxCentreSQL) { - // For complex viewboxes (routes) precompute the bounding geometry - $sGeom = chksql( - $this->oDB->getOne("select ".$this->sViewboxSmallSQL), - "Could not get small viewbox" - ); - $this->sViewboxSmallSQL = "'".$sGeom."'::geometry"; - - $sGeom = chksql( - $this->oDB->getOne("select ".$this->sViewboxLargeSQL), - "Could not get large viewbox" - ); - $this->sViewboxLargeSQL = "'".$sGeom."'::geometry"; + $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery); + $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery); + $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery); } // Do we have anything that looks like a lat/lon pair? - $oNearPoint = false; - if ($aLooksLike = NearPoint::extractFromQuery($sQuery)) { - $oNearPoint = $aLooksLike['pt']; - $sQuery = $aLooksLike['query']; - } + $sQuery = $oCtx->setNearPointFromQuery($sQuery); - $aSearchResults = array(); if ($sQuery || $this->aStructuredQuery) { // Start with a single blank search - $aSearches = array(new SearchDescription()); - - if ($oNearPoint) { - $aSearches[0]->setNear($oNearPoint); - } + $aSearches = array(new SearchDescription($oCtx)); if ($sQuery) { $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery); @@ -1048,6 +559,10 @@ class Geocode $aSpecialTermsRaw, PREG_SET_ORDER ); + if (!empty($aSpecialTermsRaw)) { + Debug::printVar('Special terms', $aSpecialTermsRaw); + } + foreach ($aSpecialTermsRaw as $aSpecialTerm) { $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery); if (!$sSpecialTerm) { @@ -1063,15 +578,17 @@ class Geocode 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?" + $sToken = $this->oDB->getOne( + 'SELECT make_standard_name(:term)', + array(':term' => $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($sSQL)); + + Debug::printSQL($sSQL); + $aSearchWords = $this->oDB->getAll($sSQL); $aNewSearches = array(); foreach ($aSearches as $oSearch) { foreach ($aSearchWords as $aSearchTerm) { @@ -1079,7 +596,7 @@ class Geocode $oNewSearch->setPoiSearch( Operator::TYPE, $aSearchTerm['class'], - $aSearchTerm['type'], + $aSearchTerm['type'] ); $aNewSearches[] = $oNewSearch; } @@ -1090,114 +607,100 @@ class Geocode // Split query into phrases // Commas are used to reduce the search space by indicating where phrases split if ($this->aStructuredQuery) { - $aPhrases = $this->aStructuredQuery; + $aInPhrases = $this->aStructuredQuery; $bStructuredPhrases = true; } else { - $aPhrases = explode(',', $sQuery); + $aInPhrases = explode(',', $sQuery); $bStructuredPhrases = false; } + Debug::printDebugArray('Search context', $oCtx); + Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]); + Debug::printVar('Final query phrases', $aInPhrases); + // Convert each phrase to standard form // Create a list of standard words // Get all 'sets' of words // Generate a complete list of all + Debug::newSection('Tokenization'); $aTokens = array(); - foreach ($aPhrases as $iPhrase => $sPhrase) { - $aPhrase = chksql( - $this->oDB->getRow("SELECT make_standard_name('".pg_escape_string($sPhrase)."') as string"), - "Cannot normalize query string (is it a UTF-8 string?)" + $aPhrases = array(); + foreach ($aInPhrases as $iPhrase => $sPhrase) { + $sPhrase = $this->oDB->getOne( + 'SELECT make_standard_name(:phrase)', + array(':phrase' => $sPhrase), + 'Cannot normalize query string (is it a UTF-8 string?)' ); - if (trim($aPhrase['string'])) { - $aPhrases[$iPhrase] = $aPhrase; - $aPhrases[$iPhrase]['words'] = explode(' ', $aPhrases[$iPhrase]['string']); - $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0); - $aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets'])); - } else { - unset($aPhrases[$iPhrase]); + if (trim($sPhrase)) { + $oPhrase = new Phrase($sPhrase, is_string($iPhrase) ? $iPhrase : ''); + $oPhrase->addTokens($aTokens); + $aPhrases[] = $oPhrase; } } - // Reindex phrases - we make assumptions later on that they are numerically keyed in order - $aPhraseTypes = array_keys($aPhrases); - $aPhrases = array_values($aPhrases); + Debug::printVar('Tokens', $aTokens); - if (sizeof($aTokens)) { - // Check which tokens we have, get the ID numbers + $oValidTokens = new TokenList(); + + if (!empty($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)).')'; + $sSQL .= ' WHERE word_token in ('.join(',', $this->oDB->getDBQuotedList($aTokens)).')'; - if (CONST_Debug) var_Dump($sSQL); + Debug::printSQL($sSQL); - $aValidTokens = array(); - $aDatabaseWords = chksql( - $this->oDB->getAll($sSQL), - "Could not get word tokens." + $oValidTokens->addTokensFromDB( + $this->oDB, + $aTokens, + $this->aCountryCodes, + $sNormQuery, + $this->oNormalizer ); - $aPossibleMainWordIDs = array(); - $aWordFrequencyScores = array(); - foreach ($aDatabaseWords as $aToken) { - // Very special case - require 2 letter country param to match the country code found - if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country']) - && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code'] - ) { - continue; - } - if (isset($aValidTokens[$aToken['word_token']])) { - $aValidTokens[$aToken['word_token']][] = $aToken; - } else { - $aValidTokens[$aToken['word_token']] = array($aToken); - } - if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1; - $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1; - } - if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); - - // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code + // Try more interpretations for Tokens that could not be matched. foreach ($aTokens as $sToken) { - if (!isset($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']) { - if (isset($aValidTokens[$sToken])) { - $aValidTokens[$sToken][] = $aToken; - } else { - $aValidTokens[$sToken] = array($aToken); - } - } - } + if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) { + if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) { + // US ZIP+4 codes - merge in the 5-digit ZIP code + $oValidTokens->addToken( + $sToken, + new Token\Postcode(null, $aData[1], 'us') + ); + } elseif (preg_match('/^ [0-9]+$/', $sToken)) { + // Unknown single word token with a number. + // Assume it is a house number. + $oValidTokens->addToken( + $sToken, + new Token\HouseNumber(null, trim($sToken)) + ); } } } - foreach ($aTokens as $sToken) { - // Unknown single word token with a number - assume it is a house number - if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/[0-9]/', $sToken)) { - $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house')); - } - } - // Any words that have failed completely? // TODO: suggestions - // Start the search process - // array with: placeid => -1 | tiger-housenumber - $aResultPlaceIDs = array(); + Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo()); - $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery); + foreach ($aPhrases as $oPhrase) { + $oPhrase->computeWordSets($oValidTokens); + } + Debug::printDebugTable('Phrases', $aPhrases); + + Debug::newSection('Search candidates'); + + $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bStructuredPhrases); if ($this->bReverseInPlan) { // Reverse phrase array and also reverse the order of the wordsets in // the first and final phrase. Don't bother about phrases in the middle // because order in the address doesn't matter. $aPhrases = array_reverse($aPhrases); - $aPhrases[0]['wordsets'] = getInverseWordSets($aPhrases[0]['words'], 0); - if (sizeof($aPhrases) > 1) { - $aFinalPhrase = end($aPhrases); - $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0); + $aPhrases[0]->invertWordSets(); + if (count($aPhrases) > 1) { + $aPhrases[count($aPhrases)-1]->invertWordSets(); } - $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery); + $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false); foreach ($aGroupedSearches as $aSearches) { foreach ($aSearches as $aSearch) { @@ -1230,208 +733,143 @@ class Geocode $sHash = serialize($aSearch); if (isset($aSearchHash[$sHash])) { unset($aGroupedSearches[$iGroup][$iSearch]); - if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]); + if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]); } else { $aSearchHash[$sHash] = 1; } } } - if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); + Debug::printGroupedSearch( + $aGroupedSearches, + $oValidTokens->debugTokenByWordIdList() + ); + // Start the search process $iGroupLoop = 0; $iQueryLoop = 0; + $aNextResults = array(); foreach ($aGroupedSearches as $iGroupedRank => $aSearches) { $iGroupLoop++; + $aResults = $aNextResults; foreach ($aSearches as $oSearch) { $iQueryLoop++; - $searchedHousenumber = -1; - - if (CONST_Debug) echo "
Search Loop, group $iGroupLoop, loop $iQueryLoop"; - if (CONST_Debug) _debugDumpGroupedSearches(array($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 ? $this->sViewboxSmallSQL : '' - ); - } - } elseif (!$oSearch->isNamedSearch()) { - // looking for a POI in a geographic area - if (!$bBoundingBoxSearch && !$oSearch->isNearSearch()) { - continue; - } - $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']] = $aRow['exactmatch']; - } - } - - //now search for housenumber, if housenumber provided - if ($oSearch->hasHouseNumber() && sizeof($aPlaceIDs)) { - $aResult = $oSearch->queryHouseNumber( - $this->oDB, - $aPlaceIDs, - $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '' - $this->iLimit - ); - - if (sizeof($aResult)) { - $searchedHousenumber = $aResult['iHouseNumber']; - $aPlaceIDs = $aResults['aPlaceIDs']; - } elseif (!$oSearch->looksLikeFullAddress()) { - $aPlaceIDs = array(); - } - } - - // finally get POIs if requested - if ($oSearch->isPoiSearch() && sizeof($aPlaceIDs)) { - $aPlaceIDs = $oSearch->queryPoiByOperator( - $this->oDB, - $aPlaceIDs, - $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '' - $this->iLimit - ); + Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop"); + Debug::printGroupedSearch( + array($iGroupedRank => array($oSearch)), + $oValidTokens->debugTokenByWordIdList() + ); + + $aNewResults = $oSearch->query( + $this->oDB, + $this->iMinAddressRank, + $this->iMaxAddressRank, + $this->iLimit + ); + + // The same result may appear in different rounds, only + // use the one with minimal rank. + foreach ($aNewResults as $iPlace => $oRes) { + if (!isset($aResults[$iPlace]) + || $aResults[$iPlace]->iResultRank > $oRes->iResultRank) { + $aResults[$iPlace] = $oRes; } } - if (CONST_Debug) { - echo "
Place IDs: "; - var_Dump($aPlaceIDs); - } + if ($iQueryLoop > 20) break; + } - if (sizeof($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); - } + if (!empty($aResults)) { + $aSplitResults = Result::splitResults($aResults); + Debug::printVar('Split results', $aSplitResults); + if ($iGroupLoop <= 4 && empty($aSplitResults['tail']) + && reset($aSplitResults['head'])->iResultRank > 0) { + // Haven't found an exact match for the query yet. + // Therefore add result from the next group level. + $aNextResults = $aSplitResults['head']; + foreach ($aNextResults as $oRes) { + $oRes->iResultRank--; } + $aResults = array(); + } else { + $aResults = $aSplitResults['head']; } - - foreach ($aPlaceIDs as $iPlaceID) { - // array for placeID => -1 | Tiger housenumber - $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber; - } - if ($iQueryLoop > 20) break; } - if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) { + if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) { // Need to verify passes rank limits before dropping out of the loop (yuk!) // reduces the number of place ids, like a filter // rank_address is 30 for interpolated housenumbers - $sWherePlaceId = 'WHERE place_id in ('; - $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') '; - - $sSQL = "SELECT place_id "; - $sSQL .= "FROM placex ".$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).")"; + $aFilterSql = array(); + $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX); + if ($sPlaceIds) { + $sSQL = 'SELECT place_id 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 .= ')'; + $aFilterSql[] = $sSQL; } - $sSQL .= ") "; - if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) { - $sSQL .= "UNION "; - $sSQL .= " SELECT place_id "; - $sSQL .= " FROM location_property_tiger ".$sWherePlaceId; + $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_POSTCODE); + if ($sPlaceIds) { + $sSQL = ' SELECT place_id FROM location_postcode lp '; + $sSQL .= 'WHERE place_id in ('.$sPlaceIds.') '; + $sSQL .= " AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank "; + if ($this->aAddressRankList) { + $sSQL .= ' OR lp.rank_address in ('.join(',', $this->aAddressRankList).')'; + } + $sSQL .= ') '; + $aFilterSql[] = $sSQL; } - if ($this->iMaxAddressRank == 30) { - $sSQL .= "UNION "; - $sSQL .= " SELECT place_id "; - $sSQL .= " FROM location_property_osmline ".$sWherePlaceId; + + $aFilteredIDs = array(); + if ($aFilterSql) { + $sSQL = join(' UNION ', $aFilterSql); + Debug::printSQL($sSQL); + $aFilteredIDs = $this->oDB->getCol($sSQL); } - if (CONST_Debug) var_dump($sSQL); - $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL)); + $tempIDs = array(); - foreach ($aFilteredPlaceIDs as $placeID) { - $tempIDs[$placeID] = $aResultPlaceIDs[$placeID]; //assign housenumber to placeID + foreach ($aResults as $oResult) { + if (($this->iMaxAddressRank == 30 && + ($oResult->iTable == Result::TABLE_OSMLINE + || $oResult->iTable == Result::TABLE_AUX + || $oResult->iTable == Result::TABLE_TIGER)) + || in_array($oResult->iId, $aFilteredIDs) + ) { + $tempIDs[$oResult->iId] = $oResult; + } } - $aResultPlaceIDs = $tempIDs; + $aResults = $tempIDs; } - //exit; - if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break; + if (!empty($aResults)) break; if ($iGroupLoop > 4) break; if ($iQueryLoop > 30) break; } - - // Did we find anything? - if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) { - $aSearchResults = $this->getDetails($aResultPlaceIDs); - } } else { // Just interpret as a reverse geocode $oReverse = new ReverseGeocode($this->oDB); $oReverse->setZoom(18); - $aLookup = $oReverse->lookup( - $oNearPoint->lat(), - $oNearPoint->lon(), - false - ); + $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false); - if (CONST_Debug) var_dump("Reverse search", $aLookup); + Debug::printVar('Reverse search', $oLookup); - if ($aLookup['place_id']) { - $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1)); - $aResultPlaceIDs[$aLookup['place_id']] = -1; - } else { - $aSearchResults = array(); + if ($oLookup) { + $aResults = array($oLookup->iId => $oLookup); } } // No results? Done - if (!sizeof($aSearchResults)) { + if (empty($aResults)) { if ($this->bFallback) { if ($this->fallbackStructuredQuery()) { return $this->lookup(); @@ -1441,34 +879,34 @@ class Geocode return array(); } - $aClassType = getClassTypesWithImportance(); + if ($this->aAddressRankList) { + $this->oPlaceLookup->setAddressRankList($this->aAddressRankList); + } + $this->oPlaceLookup->setAllowedTypesSQLList($this->sAllowedTypesSQLList); + $this->oPlaceLookup->setLanguagePreference($this->aLangPrefOrder); + if ($oCtx->hasNearPoint()) { + $this->oPlaceLookup->setAnchorSql($oCtx->sqlNear); + } + + $aSearchResults = $this->oPlaceLookup->lookup($aResults); + + $aClassType = ClassTypes\getListWithImportance(); $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery); foreach ($aRecheckWords as $i => $sWord) { if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]); } - if (CONST_Debug) { - echo 'Recheck words:<\i>'; - var_dump($aRecheckWords); - } - - $oPlaceLookup = new PlaceLookup($this->oDB); - $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints); - $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText); - $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON); - $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML); - $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG); - $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold); + Debug::printVar('Recheck words', $aRecheckWords); - foreach ($aSearchResults as $iResNum => $aResult) { + foreach ($aSearchResults as $iIdx => $aResult) { // Default - $fDiameter = getResultDiameter($aResult); + $fDiameter = ClassTypes\getProperty($aResult, 'defdiameter', 0.0001); - $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2); + $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2); if ($aOutlineResult) { $aResult = array_merge($aResult, $aOutlineResult); } - + if ($aResult['extra_place'] == 'city') { $aResult['class'] = 'place'; $aResult['type'] = 'city'; @@ -1476,81 +914,59 @@ class Geocode } // Is there an icon set for this type of result? - if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon']) - && $aClassType[$aResult['class'].':'.$aResult['type']]['icon'] - ) { - $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png'; - } + $aClassInfo = ClassTypes\getInfo($aResult); - if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']) - && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'] - ) { - $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']; - } elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label']) - && $aClassType[$aResult['class'].':'.$aResult['type']]['label'] - ) { - $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label']; - } - // if tag '&addressdetails=1' is set in query - if ($this->bIncludeAddressDetails) { - // 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'])) { - $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']); + if ($aClassInfo) { + if (isset($aClassInfo['icon'])) { + $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassInfo['icon'].'.p.20.png'; } - } - if ($this->bIncludeExtraTags) { - if ($aResult['extra']) { - $aResult['sExtraTags'] = json_decode($aResult['extra']); - } else { - $aResult['sExtraTags'] = (object) array(); + if (isset($aClassInfo['label'])) { + $aResult['label'] = $aClassInfo['label']; } } - if ($this->bIncludeNameDetails) { - if ($aResult['names']) { - $aResult['sNameDetails'] = json_decode($aResult['names']); - } else { - $aResult['sNameDetails'] = (object) array(); - } - } + $aResult['name'] = $aResult['langaddress']; - // Adjust importance for the number of exact string matches in the result - $aResult['importance'] = max(0.001, $aResult['importance']); - $iCountWords = 0; - $sAddress = $aResult['langaddress']; - foreach ($aRecheckWords as $i => $sWord) { - if (stripos($sAddress, $sWord)!==false) { - $iCountWords++; - if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1; + if ($oCtx->hasNearPoint()) { + $aResult['importance'] = 0.001; + $aResult['foundorder'] = $aResult['addressimportance']; + } else { + $aResult['importance'] = max(0.001, $aResult['importance']); + $aResult['importance'] *= $this->viewboxImportanceFactor( + $aResult['lon'], + $aResult['lat'] + ); + // Adjust importance for the number of exact string matches in the result + $iCountWords = 0; + $sAddress = $aResult['langaddress']; + foreach ($aRecheckWords as $i => $sWord) { + if (stripos($sAddress, $sWord)!==false) { + $iCountWords++; + if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1; + } } - } - - $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']; - // 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']]; - } elseif (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'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance']; - } else { - $aResult['foundorder'] += 0.01; + $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 + + // 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 + $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches; + // - importance of the class/type + if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) + && $aClassType[$aResult['class'].':'.$aResult['type']]['importance'] + ) { + $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance']; + } else { + $aResult['foundorder'] += 0.01; + } } - if (CONST_Debug) var_dump($aResult); - $aSearchResults[$iResNum] = $aResult; + $aSearchResults[$iIdx] = $aResult; } uasort($aSearchResults, 'byImportance'); + Debug::printVar('Pre-filter results', $aSearchResults); $aOSMIDDone = array(); $aClassTypeNameDone = array(); @@ -1558,7 +974,7 @@ class Geocode $aSearchResults = array(); $bFirst = true; - foreach ($aToFilter as $iResNum => $aResult) { + foreach ($aToFilter as $aResult) { $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id']; if ($bFirst) { $fLat = $aResult['lat']; @@ -1566,7 +982,7 @@ class Geocode if (isset($aResult['zoom'])) $iZoom = $aResult['zoom']; $bFirst = false; } - if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']]) + if (!$this->oPlaceLookup->doDeDupe() || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']]) && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']])) ) { $aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true; @@ -1575,9 +991,32 @@ class Geocode } // Absolute limit on number of results - if (sizeof($aSearchResults) >= $this->iFinalLimit) break; + if (count($aSearchResults) >= $this->iFinalLimit) break; } + Debug::printVar('Post-filter results', $aSearchResults); return $aSearchResults; } // end lookup() + + public function debugInfo() + { + return array( + 'Query' => $this->sQuery, + 'Structured query' => $this->aStructuredQuery, + 'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder), + 'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs), + 'Try reversed query'=> $this->bReverseInPlan, + 'Limit (for searches)' => $this->iLimit, + 'Limit (for results)'=> $this->iFinalLimit, + 'Country codes' => Debug::fmtArrayVals($this->aCountryCodes), + 'Bounded search' => $this->bBoundedSearch, + 'Viewbox' => Debug::fmtArrayVals($this->aViewBox), + 'Route points' => Debug::fmtArrayVals($this->aRoutePoints), + 'Route width' => $this->aRouteWidth, + 'Max rank' => $this->iMaxRank, + 'Min address rank' => $this->iMinAddressRank, + 'Max address rank' => $this->iMaxAddressRank, + 'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList) + ); + } } // end class