X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/86c085813044e52a4897d6a85e4ae7700bd40f91..77abe882ab160669d5206f94426353883750857f:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 1138824c..bf0782f1 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'); @@ -398,16 +399,20 @@ class Geocode $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, "; - $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 "; - $sSQL .= " ) AS addressimportance, "; + if ($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) "; @@ -457,16 +462,20 @@ class Geocode 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, "; - $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"; - $sSQL .= " ) AS addressimportance, "; + if ($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) "; @@ -505,16 +514,20 @@ class Geocode $sSQL .= " avg(st_x(centroid)) AS lon, "; $sSQL .= " avg(st_y(centroid)) AS lat,"; $sSQL .= " -1.15".$sImportanceSQL." 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"; - $sSQL .= " ) AS addressimportance, "; + if ($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 @@ -555,17 +568,21 @@ class Geocode $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 - $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"; - $sSQL .= " ) AS addressimportance,"; + if ($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 "; @@ -614,16 +631,20 @@ class Geocode $sSQL .= " avg(ST_X(centroid)) AS lon, "; $sSQL .= " avg(ST_Y(centroid)) AS lat, "; $sSQL .= " -1.10".$sImportanceSQL." 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"; - $sSQL .= " ) AS addressimportance, "; + if ($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) "; @@ -648,7 +669,7 @@ class Geocode return $aSearchResults; } - public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery) + public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured, $sNormQuery) { /* Calculate all searches using aValidTokens i.e. @@ -663,15 +684,11 @@ class Geocode */ $iGlobalRank = 0; - foreach ($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; @@ -726,9 +743,8 @@ class Geocode foreach ($aValidTokens[$sToken] as $aSearchTerm) { $aNewSearches = $oCurrentSearch->extendWithPartialTerm( $aSearchTerm, - $bStructuredPhrases, + $bIsStructured, $iPhrase, - $aWordFrequencyScores, isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array() ); @@ -846,18 +862,20 @@ class Geocode $this->aRouteWidth, $this->bBoundedSearch ); - } else if ($this->aViewBox) { + } 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) ); - $sCountryCodesSQL = false; - if ($this->aCountryCodes) { - $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes)); - } $sQuery = $this->sQuery; if (!preg_match('//u', $sQuery)) { @@ -933,10 +951,10 @@ class Geocode // Split query into phrases // Commas are used to reduce the search space by indicating where phrases split if ($this->aStructuredQuery) { - $aPhrases = $this->aStructuredQuery; + $aInPhrases = $this->aStructuredQuery; $bStructuredPhrases = true; } else { - $aPhrases = explode(',', $sQuery); + $aInPhrases = explode(',', $sQuery); $bStructuredPhrases = false; } @@ -945,25 +963,19 @@ 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"), + $aPhrases = array(); + foreach ($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'; @@ -1024,23 +1036,18 @@ class Geocode // 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); + $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases, $sNormQuery); if ($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, $sNormQuery); foreach ($aGroupedSearches as $aSearches) { foreach ($aSearches as $aSearch) { @@ -1082,117 +1089,38 @@ class Geocode if (CONST_Debug) _debugDumpGroupedSearches($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); - } - } elseif (!$oSearch->isNamedSearch()) { - // looking for a POI in a geographic area - if (!$oCtx->isBoundedSearch()) { - continue; - } - - $aPlaceIDs = $oSearch->queryNearbyPoi( - $this->oDB, - $sCountryCodesSQL, - $this->aExcludePlaceIDs ? join(',', $this->aExcludePlaceIDs) : '', - $this->iLimit - ); - } elseif ($oSearch->isOperator(Operator::POSTCODE)) { - $aPlaceIDs = $oSearch->queryPostcode( - $this->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) : '', - $this->iLimit - ); - - if (sizeof($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 = $aResult['aPlaceIDs']; - } elseif (!$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 @@ -1235,14 +1163,13 @@ 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)) { + if (sizeof($aResultPlaceIDs)) { $aSearchResults = $this->getDetails($aResultPlaceIDs, $oCtx); } } else { @@ -1348,36 +1275,43 @@ class Geocode } } - // 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; + $aResult['name'] = $aResult['langaddress']; + + if ($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'] = 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; + } } - } - $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['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; + // 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 { + $aResult['foundorder'] += 0.01; + } } if (CONST_Debug) var_dump($aResult); $aSearchResults[$iResNum] = $aResult;