X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/16268f92ccd11f0c11fe07b2a001b197af3a5b92..c8780da19c32c2ce5f6ea02d86a9db2d96e27195:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index cce38de9..e02aae94 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -2,9 +2,10 @@ namespace Nominatim; -require_once(CONST_BasePath.'/lib/NearPoint.php'); require_once(CONST_BasePath.'/lib/PlaceLookup.php'); require_once(CONST_BasePath.'/lib/ReverseGeocode.php'); +require_once(CONST_BasePath.'/lib/SearchDescription.php'); +require_once(CONST_BasePath.'/lib/SearchContext.php'); class Geocode { @@ -36,9 +37,8 @@ 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; @@ -183,26 +183,6 @@ class Geocode $this->iMaxAddressRank = $iMax; } - public function setRoute($aRoutePoints, $fRouteWidth) - { - $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)"; - - $this->sViewboxSmallSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL; - $this->sViewboxSmallSQL .= ','.($fRouteWidth/69).')'; - - $this->sViewboxLargeSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL; - $this->sViewboxLargeSQL .= ','.($fRouteWidth/30).')'; - } - public function setViewbox($aViewbox) { $this->aViewBox = array_map('floatval', $aViewbox); @@ -217,29 +197,6 @@ class Geocode ) { userError("Bad parameter 'viewbox'. Not a box."); } - - $fHeight = $this->aViewBox[0] - $this->aViewBox[2]; - $fWidth = $this->aViewBox[1] - $this->aViewBox[3]; - $aBigViewBox[0] = $this->aViewBox[0] + $fHeight; - $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], - $aBigViewBox[3] - ); } public function setQuery($sQueryString) @@ -318,7 +275,8 @@ class Geocode $aRoute = $oParams->getStringList('route'); $fRouteWidth = $oParams->getFloat('routewidth'); if ($aRoute && $fRouteWidth) { - $this->setRoute($aRoute, $fRouteWidth); + $this->aRoutePoints = $aRoute; + $this->aRouteWidth = $fRouteWidth; } } } @@ -367,7 +325,7 @@ 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); @@ -406,29 +364,20 @@ class Geocode return false; } - public function getDetails($aPlaceIDs) + public function getDetails($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) + 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 * "; - } + $sImportanceSQL = $oCtx->viewboxImportanceSQL('ST_Collect(centroid)'); + $sImportanceSQLGeom = $oCtx->viewboxImportanceSQL('geometry'); $sSQL = "SELECT "; $sSQL .= " osm_type,"; @@ -448,7 +397,7 @@ class Geocode if ($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 .= " COALESCE(importance,0.75-(rank_search::float/40)) $sImportanceSQL AS importance, "; $sSQL .= " ( "; $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; $sSQL .= " FROM "; @@ -507,7 +456,7 @@ class Geocode if ($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 .= " (0.75-(rank_search::float/40)) $sImportanceSQLGeom AS importance, "; $sSQL .= " ("; $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; $sSQL .= " FROM "; @@ -555,7 +504,7 @@ class Geocode if ($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, "; + $sSQL .= " -1.15".$sImportanceSQL." AS importance, "; $sSQL .= " ("; $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; $sSQL .= " FROM "; @@ -605,7 +554,7 @@ class Geocode if ($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 + $sSQL .= " -0.1".$sImportanceSQL." AS importance, "; // slightly smaller than the importance for normal houses with rank 30, which is 0 $sSQL .= " ("; $sSQL .= " SELECT "; $sSQL .= " MAX(p.importance*(p.rank_address+2)) "; @@ -664,7 +613,7 @@ class Geocode if ($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, "; + $sSQL .= " -1.10".$sImportanceSQL." AS importance, "; $sSQL .= " ( "; $sSQL .= " SELECT max(p.importance*(p.rank_address+2))"; $sSQL .= " FROM "; @@ -740,12 +689,15 @@ class Geocode // If the token is valid if (isset($aValidTokens[' '.$sToken])) { - // Recheck if the original word shows up in the query. - $bWordInQuery = false; - if (isset($aSearchTerm['word']) && $aSearchTerm['word']) { - $bWordInQuery = $this->normTerm($aSearchTerm['word'])) !== false; - } foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) { + // 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, @@ -785,18 +737,17 @@ class Geocode $aNewWordsetSearches[] = $oSearch; } } - } } } // Sort and cut - usort($aNewWordsetSearches, 'bySearchRank'); + usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank')); $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50); } //var_Dump('
',sizeof($aWordsetSearches)); exit; $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches); - usort($aNewPhraseSearches, 'bySearchRank'); + usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank')); $aSearchHash = array(); foreach ($aNewPhraseSearches as $iSearch => $aSearch) { @@ -835,12 +786,12 @@ class Geocode // Revisit searches, drop bad searches and give penalty to unlikely combinations. $aGroupedSearches = array(); foreach ($aSearches as $oSearch) { - if (!$oSearch->isValidSearch()) { + if (!$oSearch->isValidSearch($this->aCountryCodes)) { continue; } $iRank = $oSearch->addToRank($iGlobalRank); - if (!isset($aGroupedSearches[$iRank]) { + if (!isset($aGroupedSearches[$iRank])) { $aGroupedSearches[$iRank] = array(); } $aGroupedSearches[$iRank][] = $oSearch; @@ -886,15 +837,29 @@ class Geocode { if (!$this->sQuery && !$this->aStructuredQuery) return array(); + $oCtx = new SearchContext(); + + if ($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) { + $oCtx->setCountryList($this->aCountryCodes); + } + $sNormQuery = $this->normTerm($this->sQuery); $sLanguagePrefArraySQL = getArraySQL( - array_map("getDBQuoted", - $this->aLangPrefOrder) + array_map("getDBQuoted", $this->aLangPrefOrder) ); - $sCountryCodesSQL = false; - if ($this->aCountryCodes) { - $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes)); - } $sQuery = $this->sQuery; if (!preg_match('//u', $sQuery)) { @@ -908,37 +873,13 @@ class Geocode $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\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"; - } - // 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); @@ -983,7 +924,7 @@ class Geocode $oNewSearch->setPoiSearch( Operator::TYPE, $aSearchTerm['class'], - $aSearchTerm['type'], + $aSearchTerm['type'] ); $aNewSearches[] = $oNewSearch; } @@ -1077,18 +1018,14 @@ class Geocode foreach ($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')); + if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/^[0-9]+$/', $sToken)) { + $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken)); } } // Any words that have failed completely? // TODO: suggestions - // Start the search process - // array with: placeid => -1 | tiger-housenumber - $aResultPlaceIDs = array(); - $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery); if ($this->bReverseInPlan) { @@ -1143,124 +1080,38 @@ 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) { $iGroupLoop++; foreach ($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) : '' - $this->iLimit - ); - } - } if (CONST_Debug) { - echo "
Place IDs: "; - var_Dump($aPlaceIDs); + echo "
Search Loop, group $iGroupLoop, loop $iQueryLoop"; + _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens); } - if (sizeof($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); - } - } - } + $aRes = $oSearch->query( + $this->oDB, + $aWordFrequencyScores, + $this->exactMatchCache, + $this->iMinAddressRank, + $this->iMaxAddressRank, + $this->iLimit + ); - foreach ($aPlaceIDs as $iPlaceID) { + foreach ($aRes['IDs'] as $iPlaceID) { // array for placeID => -1 | Tiger housenumber - $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber; + $aResultPlaceIDs[$iPlaceID] = $aRes['houseNumber']; } if ($iQueryLoop > 20) break; } - if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) { + if (sizeof($aResultPlaceIDs) && ($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 @@ -1303,31 +1154,26 @@ class Geocode $aResultPlaceIDs = $tempIDs; } - //exit; - if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break; + if (sizeof($aResultPlaceIDs)) break; if ($iGroupLoop > 4) break; if ($iQueryLoop > 30) break; } // Did we find anything? - if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) { - $aSearchResults = $this->getDetails($aResultPlaceIDs); + 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->lookup( - $oNearPoint->lat(), - $oNearPoint->lon(), - false - ); + $aLookup = $oReverse->lookupPoint($oCtx->sqlNear, false); if (CONST_Debug) var_dump("Reverse search", $aLookup); if ($aLookup['place_id']) { - $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1)); + $aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1), $oCtx); $aResultPlaceIDs[$aLookup['place_id']] = -1; } else { $aSearchResults = array();