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 $aLangPrefOrder = array();
- protected $bIncludeAddressDetails = false;
-
protected $aExcludePlaceIDs = array();
protected $bReverseInPlan = false;
$aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
}
- if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
if ($this->bBoundedSearch) $aParams['bounded'] = '1';
if ($this->aCountryCodes) {
private function viewboxImportanceFactor($fX, $fY)
{
+ if (!$this->aViewBox) {
+ return 1;
+ }
+
$fWidth = ($this->aViewBox[2] - $this->aViewBox[0])/2;
$fHeight = ($this->aViewBox[3] - $this->aViewBox[1])/2;
public function loadParamArray($oParams, $sForceGeometryType = null)
{
- $this->bIncludeAddressDetails
- = $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
-
$this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
$this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
}
$this->oPlaceLookup->loadParamArray($oParams, $sForceGeometryType);
- $this->oPlaceLookup->setIncludeAddressDetails(false);
- $this->oPlaceLookup->setIncludePolygonAsPoints($oParams->getBool('polygon'));
+ $this->oPlaceLookup->setIncludeAddressDetails($oParams->getBool('addressdetails', false));
}
public function setQueryFromParams($oParams)
return false;
}
- public function getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bIsStructured)
+ public function getGroupedSearches($aSearches, $aPhrases, $oValidTokens, $bIsStructured)
{
/*
- Calculate all searches using aValidTokens i.e.
+ Calculate all searches using oValidTokens i.e.
'Wodsworth Road, Sheffield' =>
Phrase Wordset
$aNewPhraseSearches = array();
$sPhraseType = $bIsStructured ? $oPhrase->getPhraseType() : '';
- foreach ($oPhrase->getWordSets() as $iWordSet => $aWordset) {
- // Too many permutations - too expensive
- if ($iWordSet > 120) break;
-
+ foreach ($oPhrase->getWordSets() as $aWordset) {
$aWordsetSearches = $aSearches;
// Add all words from this wordset
//var_dump($oCurrentSearch);
//echo "</i>";
- // If the token is valid
- if (isset($aValidTokens[' '.$sToken])) {
- foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
- $aNewSearches = $oCurrentSearch->extendWithFullTerm(
- $aSearchTerm,
- isset($aValidTokens[$sToken])
- && strpos($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[] = $oSearch;
- }
+ // Tokens with full name matches.
+ foreach ($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[] = $oSearch;
}
}
}
// Look for partial matches.
// Note that there is no point in adding country terms here
// because country is omitted in the address.
- if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
+ if ($sPhraseType != 'country') {
// Allow searching for a word - but at extra cost
- foreach ($aValidTokens[$sToken] as $aSearchTerm) {
+ foreach ($oValidTokens->get($sToken) as $oSearchTerm) {
$aNewSearches = $oCurrentSearch->extendWithPartialTerm(
- $aSearchTerm,
+ $sToken,
+ $oSearchTerm,
$bIsStructured,
$iPhrase,
- isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
+ $oValidTokens->get(' '.$sToken)
);
foreach ($aNewSearches as $oSearch) {
$aSearches = array_merge($aSearches, $aNewSearches);
if ($iSearchCount > 50) break;
}
-
- //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
}
// Revisit searches, drop bad searches and give penalty to unlikely combinations.
public function lookup()
{
+ Debug::newFunction('Geocode::lookup');
if (!$this->sQuery && !$this->aStructuredQuery) return array();
+ Debug::printDebugArray('Geocode', $this);
+
$oCtx = new SearchContext();
if ($this->aRoutePoints) {
$oCtx->setCountryList($this->aCountryCodes);
}
+ Debug::newSection('Query Preprocessing');
+
$sNormQuery = $this->normTerm($this->sQuery);
- $sLanguagePrefArraySQL = getArraySQL(
- array_map('getDBQuoted', $this->aLangPrefOrder)
+ Debug::printVar('Normalized query', $sNormQuery);
+
+ $sLanguagePrefArraySQL = $this->oDB->getArraySQL(
+ $this->oDB->getDBQuotedList($this->aLangPrefOrder)
);
$sQuery = $this->sQuery;
// Do we have anything that looks like a lat/lon pair?
$sQuery = $oCtx->setNearPointFromQuery($sQuery);
- $aResults = array();
if ($sQuery || $this->aStructuredQuery) {
// Start with a single blank search
$aSearches = array(new SearchDescription($oCtx));
$aSpecialTermsRaw,
PREG_SET_ORDER
);
+ if (!empty($aSpecialTermsRaw)) {
+ Debug::printVar('Special terms', $aSpecialTermsRaw);
+ }
+
foreach ($aSpecialTermsRaw as $aSpecialTerm) {
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
if (!$sSpecialTerm) {
if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
$sSpecialTerm = pg_escape_string($sSpecialTerm);
- $sToken = chksql(
- $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
+ $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) {
$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();
$aPhrases = array();
foreach ($aInPhrases as $iPhrase => $sPhrase) {
- $sPhrase = chksql(
- $this->oDB->getOne('SELECT make_standard_name('.getDBQuoted($sPhrase).')'),
+ $sPhrase = $this->oDB->getOne(
+ 'SELECT make_standard_name(:phrase)',
+ array(':phrase' => $sPhrase),
'Cannot normalize query string (is it a UTF-8 string?)'
);
if (trim($sPhrase)) {
}
}
- if (!empty($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)).')';
+ Debug::printVar('Tokens', $aTokens);
- if (CONST_Debug) var_Dump($sSQL);
+ $oValidTokens = new TokenList();
- $aValidTokens = array();
- $aDatabaseWords = chksql(
- $this->oDB->getAll($sSQL),
- 'Could not get word tokens.'
+ if (!empty($aTokens)) {
+ $oValidTokens->addTokensFromDB(
+ $this->oDB,
+ $aTokens,
+ $this->aCountryCodes,
+ $sNormQuery,
+ $this->oNormalizer
);
- $aWordFrequencyScores = array();
- foreach ($aDatabaseWords as $aToken) {
- // Filter country tokens that do not match restricted countries.
- if ($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;
- }
- }
+ $oCtx->setFullNameWords($oValidTokens->getFullWordIDs());
- if (isset($aValidTokens[$aToken['word_token']])) {
- $aValidTokens[$aToken['word_token']][] = $aToken;
- } else {
- $aValidTokens[$aToken['word_token']] = array($aToken);
- }
- $aWordFrequencyScores[$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', 'word_token' => ' '.$sToken));
- }
- }
-
// Any words that have failed completely?
// TODO: suggestions
- $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
+ Debug::printGroupTable('Valid Tokens', $oValidTokens->debugInfo());
+
+ foreach ($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
if (count($aPhrases) > 1) {
$aPhrases[count($aPhrases)-1]->invertWordSets();
}
- $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
+ $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $oValidTokens, false);
foreach ($aGroupedSearches as $aSearches) {
foreach ($aSearches as $aSearch) {
}
}
- if (CONST_Debug) _debugDumpGroupedSearches($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++;
- if (CONST_Debug) {
- echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
- _debugDumpGroupedSearches(array($iGroupedRank => array($oSearch)), $aValidTokens);
- }
+ Debug::newSection("Search Loop, group $iGroupLoop, loop $iQueryLoop");
+ Debug::printGroupedSearch(
+ array($iGroupedRank => array($oSearch)),
+ $oValidTokens->debugTokenByWordIdList()
+ );
- $aResults += $oSearch->query(
+ $aNewResults = $oSearch->query(
$this->oDB,
- $aWordFrequencyScores,
$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] = $oRes;
+ }
+ }
+
if ($iQueryLoop > 20) break;
}
+ 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 = $aSplitResults['head'];
+ }
+ }
+
if (!empty($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
$sSQL .= 'WHERE place_id in ('.$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'";
- }
+ $sSQL .= " OR placex.rank_search between $this->iMinAddressRank and $this->iMaxAddressRank ";
if ($this->aAddressRankList) {
$sSQL .= ' OR placex.rank_address in ('.join(',', $this->aAddressRankList).')';
}
$aFilteredIDs = array();
if ($aFilterSql) {
$sSQL = join(' UNION ', $aFilterSql);
- if (CONST_Debug) var_dump($sSQL);
- $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
+ Debug::printSQL($sSQL);
+ $aFilteredIDs = $this->oDB->getCol($sSQL);
}
$tempIDs = array();
$oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
- if (CONST_Debug) var_dump('Reverse search', $aLookup);
+ Debug::printVar('Reverse search', $oLookup);
if ($oLookup) {
$aResults = array($oLookup->iId => $oLookup);
$aSearchResults = $this->oPlaceLookup->lookup($aResults);
- $aClassType = getClassTypesWithImportance();
$aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
foreach ($aRecheckWords as $i => $sWord) {
if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
}
- if (CONST_Debug) {
- echo '<i>Recheck words:<\i>';
- var_dump($aRecheckWords);
- }
+ Debug::printVar('Recheck words', $aRecheckWords);
foreach ($aSearchResults as $iIdx => $aResult) {
- // Default
- $fDiameter = getResultDiameter($aResult);
+ $fRadius = ClassTypes\getDefRadius($aResult);
- $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
+ $aOutlineResult = $this->oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fRadius);
if ($aOutlineResult) {
$aResult = array_merge($aResult, $aOutlineResult);
}
- if ($aResult['extra_place'] == 'city') {
- $aResult['class'] = 'place';
- $aResult['type'] = 'city';
- $aResult['rank_search'] = 16;
- }
-
// Is there an icon set for this type of result?
- if (isset($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';
+ $sIcon = ClassTypes\getIconFile($aResult);
+ if (isset($sIcon)) {
+ $aResult['icon'] = $sIcon;
}
- if (isset($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'], $aResults[$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']);
- }
+ $sLabel = ClassTypes\getLabel($aResult);
+ if (isset($sLabel)) {
+ $aResult['label'] = $sLabel;
}
-
$aResult['name'] = $aResult['langaddress'];
if ($oCtx->hasNearPoint()) {
$aResult['lon'],
$aResult['lat']
);
+
+ // secondary ordering (for results with same importance (the smaller the better):
+ // - approximate importance of address parts
+ if (isset($aResult['addressimportance']) && $aResult['addressimportance']) {
+ $aResult['foundorder'] = -$aResult['addressimportance']/10;
+ } else {
+ $aResult['foundorder'] = -$aResult['importance'];
+ }
+ // - number of exact matches from the query
+ $aResult['foundorder'] -= $aResults[$aResult['place_id']]->iExactMatches;
+ // - importance of the class/type
+ $iClassImportance = ClassTypes\getImportance($aResult);
+ if (isset($iClassImportance)) {
+ $aResult['foundorder'] += 0.0001 * $iClassImportance;
+ } else {
+ $aResult['foundorder'] += 0.01;
+ }
+ // - rank
+ $aResult['foundorder'] -= 0.00001 * (30 - $aResult['rank_search']);
+
// Adjust importance for the number of exact string matches in the result
$iCountWords = 0;
$sAddress = $aResult['langaddress'];
}
}
- $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;
- }
+ // 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);
}
- if (CONST_Debug) var_dump($aResult);
$aSearchResults[$iIdx] = $aResult;
}
uasort($aSearchResults, 'byImportance');
+ Debug::printVar('Pre-filter results', $aSearchResults);
$aOSMIDDone = array();
$aClassTypeNameDone = array();
$aToFilter = $aSearchResults;
$aSearchResults = array();
- if (CONST_Debug) var_dump($aToFilter);
-
$bFirst = true;
foreach ($aToFilter as $aResult) {
$this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
if (count($aSearchResults) >= $this->iFinalLimit) break;
}
- if (CONST_Debug) var_dump($aSearchResults);
+ 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
projects / nominatim.git / blobdiff