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');
protected $aExcludePlaceIDs = array();
protected $bDeDupe = true;
- protected $bReverseInPlan = false;
+ protected $bReverseInPlan = true;
protected $iLimit = 20;
protected $iFinalLimit = 10;
$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) ";
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) ";
$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
$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 ";
$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) ";
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.
*/
$iGlobalRank = 0;
- foreach ($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;
// 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,
$sPhraseType,
foreach ($aValidTokens[$sToken] as $aSearchTerm) {
$aNewSearches = $oCurrentSearch->extendWithPartialTerm(
$aSearchTerm,
- $bStructuredPhrases,
+ $bIsStructured,
$iPhrase,
- $aWordFrequencyScores,
isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
);
// 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;
}
$this->aRouteWidth,
$this->bBoundedSearch
);
- } else if ($this->aViewBox) {
+ } elseif ($this->aViewBox) {
$oCtx->setViewboxFromBox($this->aViewBox, $this->bBoundedSearch);
}
if ($this->aExcludePlaceIDs) {
// 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;
}
// 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';
$this->oDB->getAll($sSQL),
"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);
// 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) {
}
}
- // 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;
projects / nominatim.git / blobdiff