X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/85530b3c52c2dc20cacb665c695402d2aa503954..a8a700e16d286971c43c013b322a1fd4a339de4d:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 51b79735..8276fe98 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -3,6 +3,7 @@ namespace Nominatim; 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'); @@ -11,21 +12,13 @@ class Geocode { protected $oDB; + protected $oPlaceLookup; + protected $aLangPrefOrder = array(); protected $bIncludeAddressDetails = false; - protected $bIncludeExtraTags = false; - protected $bIncludeNameDetails = false; - - protected $bIncludePolygonAsPoints = false; - protected $bIncludePolygonAsText = false; - protected $bIncludePolygonAsGeoJSON = false; - protected $bIncludePolygonAsKML = false; - protected $bIncludePolygonAsSVG = false; - protected $fPolygonSimplificationThreshold = 0.0; protected $aExcludePlaceIDs = array(); - protected $bDeDupe = true; protected $bReverseInPlan = true; protected $iLimit = 20; @@ -44,7 +37,6 @@ class Geocode protected $iMinAddressRank = 0; protected $iMaxAddressRank = 30; protected $aAddressRankList = array(); - protected $exactMatchCache = array(); protected $sAllowedTypesSQLList = false; @@ -57,6 +49,7 @@ class Geocode public function __construct(&$oDB) { $this->oDB =& $oDB; + $this->oPlaceLookup = new PlaceLookup($this->oDB); $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules); } @@ -87,69 +80,26 @@ 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) $iLimit = 50; @@ -185,20 +135,39 @@ class Geocode public function setViewbox($aViewbox) { - $this->aViewBox = array_map('floatval', $aViewbox); + $aBox = 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])); + $this->aViewBox[0] = max(-180.0, min($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], $aBox[3])); - if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001 - || abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001 + if ($this->aViewBox[2] - $this->aViewBox[0] < 0.000000001 + || $this->aViewBox[3] - $this->aViewBox[1] < 0.000000001 ) { userError("Bad parameter 'viewbox'. Not a box."); } } + private function viewboxImportanceFactor($fX, $fY) + { + $fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2; + $fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2; + + $fXDist = abs($fX - ($this->aViewBox[0] + $this->aViewBox[2])/2); + $fYDist = abs($fY - ($this->aViewBox[1] + $this->aViewBox[3])/2); + + if ($fXDist <= $fWidth && $fYDist <= $fHeight) { + return 1; + } + + if ($fXDist <= $fWidth * 3 && $fYDist <= 3 * $fHeight) { + return 0.5; + } + + return 0.25; + } + public function setQuery($sQueryString) { $this->sQuery = $sQueryString; @@ -211,17 +180,12 @@ class Geocode } - public function loadParamArray($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); @@ -280,6 +244,10 @@ class Geocode } } } + + $this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType); + $this->oPlaceLookup->setIncludeAddressDetails(false); + $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon')); } public function setQueryFromParams($oParams) @@ -364,311 +332,7 @@ class Geocode return false; } - public function getDetails($aPlaceIDs, $oCtx) - { - //$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 = $oCtx->viewboxImportanceSQL('ST_Collect(centroid)'); - $sImportanceSQLGeom = $oCtx->viewboxImportanceSQL('geometry'); - - $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 .= " COALESCE(importance,0.75-(rank_search::float/40)) $sImportanceSQL AS importance, "; - if ($oCtx->hasNearPoint()) { - $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,"; - } else { - $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 .= " (0.75-(rank_search::float/40)) $sImportanceSQLGeom AS importance, "; - if ($oCtx->hasNearPoint()) { - $sSQL .= $oCtx->distanceSQL('geometry')." AS addressimportance,"; - } else { - $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,"; - $sSQL .= " -1.15".$sImportanceSQL." AS importance, "; - if ($oCtx->hasNearPoint()) { - $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,"; - } else { - $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, "; - $sSQL .= " -0.1".$sImportanceSQL." AS importance, "; // slightly smaller than the importance for normal houses with rank 30, which is 0 - if ($oCtx->hasNearPoint()) { - $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,"; - } else { - $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, "; - $sSQL .= " -1.10".$sImportanceSQL." AS importance, "; - if ($oCtx->hasNearPoint()) { - $sSQL .= $oCtx->distanceSQL('ST_Collect(centroid)')." AS addressimportance,"; - } else { - $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, $aValidTokens, $bIsStructured) { /* Calculate all searches using aValidTokens i.e. @@ -681,17 +345,11 @@ 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 = ''; - } + $sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : ''; - foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) { + foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) { // Too many permutations - too expensive if ($iWordSet > 120) break; @@ -710,25 +368,15 @@ class Geocode // If the token is valid if (isset($aValidTokens[' '.$sToken])) { foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) { - // Recheck if the original word shows up in the query. - $bWordInQuery = false; - if (isset($aSearchTerm['word']) && $aSearchTerm['word']) { - $bWordInQuery = strpos( - $sNormQuery, - $this->normTerm($aSearchTerm['word']) - ) !== false; - } $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 + && $iPhrase + 1 == sizeof($aPhrases) ); foreach ($aNewSearches as $oSearch) { @@ -746,9 +394,8 @@ class Geocode foreach ($aValidTokens[$sToken] as $aSearchTerm) { $aNewSearches = $oCurrentSearch->extendWithPartialTerm( $aSearchTerm, - $bStructuredPhrases, + $bIsStructured, $iPhrase, - $aWordFrequencyScores, isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array() ); @@ -806,11 +453,11 @@ class Geocode // Revisit searches, drop bad searches and give penalty to unlikely combinations. $aGroupedSearches = array(); foreach ($aSearches as $oSearch) { - if (!$oSearch->isValidSearch($this->aCountryCodes)) { + if (!$oSearch->isValidSearch()) { continue; } - $iRank = $oSearch->addToRank($iGlobalRank); + $iRank = $oSearch->getRank(); if (!isset($aGroupedSearches[$iRank])) { $aGroupedSearches[$iRank] = array(); } @@ -878,12 +525,12 @@ class Geocode $sNormQuery = $this->normTerm($this->sQuery); $sLanguagePrefArraySQL = getArraySQL( - array_map("getDBQuoted", $this->aLangPrefOrder) + array_map('getDBQuoted', $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 @@ -896,7 +543,7 @@ class Geocode // Do we have anything that looks like a lat/lon pair? $sQuery = $oCtx->setNearPointFromQuery($sQuery); - $aSearchResults = array(); + $aResults = array(); if ($sQuery || $this->aStructuredQuery) { // Start with a single blank search $aSearches = array(new SearchDescription($oCtx)); @@ -930,7 +577,7 @@ class Geocode $sSpecialTerm = pg_escape_string($sSpecialTerm); $sToken = chksql( $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"), - "Cannot decode query. Wrong encoding?" + 'Cannot decode query. Wrong encoding?' ); $sSQL = 'SELECT class, type FROM word '; $sSQL .= ' WHERE word_token in (\' '.$sToken.'\')'; @@ -955,10 +602,10 @@ 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; } @@ -967,54 +614,55 @@ class Geocode // Get all 'sets' of words // Generate a complete list of all $aTokens = array(); - foreach ($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 = chksql( + $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($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); - 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 '; - $sSQL .= ' WHERE word_token in ('.join(',', array_map("getDBQuoted", $aTokens)).')'; + $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')'; if (CONST_Debug) var_Dump($sSQL); $aValidTokens = array(); $aDatabaseWords = chksql( $this->oDB->getAll($sSQL), - "Could not get word tokens." + 'Could not get word tokens.' ); - $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']) != $aToken['country_code'] + // Filter country tokens that do not match restricted countries. + if ($this->aCountryCodes + && $aToken['country_code'] + && !in_array($aToken['country_code'], $this->aCountryCodes) ) { continue; } + // Special terms need to appear in their normalized form. + if ($aToken['word'] && $aToken['class']) { + $sNormWord = $this->normTerm($aToken['word']); + if (strpos($sNormQuery, $sNormWord) === false) { + continue; + } + } + if (isset($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); @@ -1046,19 +694,18 @@ class Geocode // Any words that have failed completely? // TODO: suggestions - $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery); + $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases); if ($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); + $aPhrases[0]->invertWordSets(); if (sizeof($aPhrases) > 1) { - $aFinalPhrase = end($aPhrases); - $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0); + $aPhrases[sizeof($aPhrases)-1]->invertWordSets(); } - $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery); + $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false); foreach ($aGroupedSearches as $aSearches) { foreach ($aSearches as $aSearch) { @@ -1101,8 +748,6 @@ class Geocode if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens); // Start the search process - // array with: placeid => -1 | tiger-housenumber - $aResultPlaceIDs = array(); $iGroupLoop = 0; $iQueryLoop = 0; foreach ($aGroupedSearches as $iGroupedRank => $aSearches) { @@ -1115,93 +760,90 @@ class Geocode _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens); } - $aRes = $oSearch->query( + $aResults += $oSearch->query( $this->oDB, $aWordFrequencyScores, - $this->exactMatchCache, $this->iMinAddressRank, $this->iMaxAddressRank, $this->iLimit ); - foreach ($aRes['IDs'] as $iPlaceID) { - // array for placeID => -1 | Tiger housenumber - $aResultPlaceIDs[$iPlaceID] = $aRes['houseNumber']; - } if ($iQueryLoop > 20) break; } - if (sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) { + if (sizeof($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 ', $aFilterSql); + if (CONST_Debug) var_dump($sSQL); + $aFilteredIDs = chksql($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 = $tempIDs; } - if (sizeof($aResultPlaceIDs)) break; + if (sizeof($aResults)) break; if ($iGroupLoop > 4) break; if ($iQueryLoop > 30) break; } - - // Did we find anything? - if (sizeof($aResultPlaceIDs)) { - $aSearchResults = $this->getDetails($aResultPlaceIDs, $oCtx); - } } else { // Just interpret as a reverse geocode $oReverse = new ReverseGeocode($this->oDB); $oReverse->setZoom(18); - $aLookup = $oReverse->lookupPoint($oCtx->sqlNear, false); + $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false); - if (CONST_Debug) var_dump("Reverse search", $aLookup); + if (CONST_Debug) var_dump('Reverse search', $aLookup); - if ($aLookup['place_id']) { - $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1), $oCtx); - $aResultPlaceIDs[$aLookup['place_id']] = -1; - } else { - $aSearchResults = array(); + if ($oLookup) { + $aResults = array($oLookup->iId => $oLookup); } } // No results? Done - if (!sizeof($aSearchResults)) { + if (!sizeof($aResults)) { if ($this->bFallback) { if ($this->fallbackStructuredQuery()) { return $this->lookup(); @@ -1211,6 +853,17 @@ class Geocode return array(); } + if ($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 = getClassTypesWithImportance(); $aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery); foreach ($aRecheckWords as $i => $sWord) { @@ -1222,23 +875,15 @@ class Geocode var_dump($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); - - foreach ($aSearchResults as $iResNum => $aResult) { + foreach ($aSearchResults as $iIdx => $aResult) { // Default $fDiameter = getResultDiameter($aResult); - $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'; $aResult['type'] = 'city'; @@ -1264,43 +909,30 @@ class Geocode // if tag '&addressdetails=1' is set in query if ($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']]); + $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResults[$aResult['place_id']]->iHouseNumber); if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) { $aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']); } } - if ($this->bIncludeExtraTags) { - if ($aResult['extra']) { - $aResult['sExtraTags'] = json_decode($aResult['extra']); - } else { - $aResult['sExtraTags'] = (object) array(); - } - } - - if ($this->bIncludeNameDetails) { - if ($aResult['names']) { - $aResult['sNameDetails'] = json_decode($aResult['names']); - } else { - $aResult['sNameDetails'] = (object) array(); - } - } - $aResult['name'] = $aResult['langaddress']; - if ($oCtx->hasNearPoint()) - { + if ($oCtx->hasNearPoint()) { $aResult['importance'] = 0.001; $aResult['foundorder'] = $aResult['addressimportance']; } else { // Adjust importance for the number of exact string matches in the result + $aResult['importance'] *= $this->viewboxImportanceFactor( + $aResult['lon'], + $aResult['lat'] + ); $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)) $iCountWords += 0.1; + if (preg_match('/(^|,)\s*'.preg_quote($sWord, '/').'\s*(,|$)/', $sAddress)) $iCountWords += 0.1; } } @@ -1310,11 +942,7 @@ class Geocode // - 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']]; - } + $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches; // - importance of the class/type if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) && $aClassType[$aResult['class'].':'.$aResult['type']]['importance'] @@ -1325,7 +953,7 @@ class Geocode } } if (CONST_Debug) var_dump($aResult); - $aSearchResults[$iResNum] = $aResult; + $aSearchResults[$iIdx] = $aResult; } uasort($aSearchResults, 'byImportance'); @@ -1334,8 +962,10 @@ class Geocode $aToFilter = $aSearchResults; $aSearchResults = array(); + if (CONST_Debug) var_dump($aToFilter); + $bFirst = true; - foreach ($aToFilter as $iResNum => $aResult) { + foreach ($aToFilter as $aResult) { $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id']; if ($bFirst) { $fLat = $aResult['lat']; @@ -1343,7 +973,7 @@ class Geocode if (isset($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'].$aResult['osm_id']] = true; @@ -1355,6 +985,7 @@ class Geocode if (sizeof($aSearchResults) >= $this->iFinalLimit) break; } + if (CONST_Debug) var_dump($aSearchResults); return $aSearchResults; } // end lookup() } // end class