if (isset($aLangPrefOrder['name:de'])) $bReverseInPlan = true;
if (isset($aLangPrefOrder['name:ru'])) $bReverseInPlan = true;
if (isset($aLangPrefOrder['name:ja'])) $bReverseInPlan = true;
+ if (isset($aLangPrefOrder['name:pl'])) $bReverseInPlan = true;
$sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$aLangPrefOrder))."]";
{
// Start with a blank search
$aSearches = array(
- array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(),
+ array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 'aFullNameAddress'=>array(),
+ 'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(),
'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'')
);
// Check which tokens we have, get the ID numbers
$sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator, search_name_count';
$sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')';
- $sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency;
+// $sSQL .= ' and search_name_count < '.CONST_Max_Word_Frequency;
// $sSQL .= ' group by word_token, word, class, type, location, country_code';
if (CONST_Debug) var_Dump($sSQL);
failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords);
}
$aPossibleMainWordIDs = array();
+ $aWordFrequencyScores = array();
foreach($aDatabaseWords as $aToken)
{
+ // Very special case - require 2 letter country param to match the country code found
+ if ($bStructuredPhrases && $aToken['country_code'] && !empty($aStructuredQuery['country'])
+ && strlen($aStructuredQuery['country']) == 2 && strtolower($aStructuredQuery['country']) != $aToken['country_code'])
+ {
+ continue;
+ }
+
if (isset($aValidTokens[$aToken['word_token']]))
{
$aValidTokens[$aToken['word_token']][] = $aToken;
{
$aValidTokens[$aToken['word_token']] = array($aToken);
}
- if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1 + $aToken['search_name_count'];
+ if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
+ $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
}
if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
{
if (sizeof($aSearch['aName']))
{
- if (($sPhraseType != 'street' && $sPhraseType != 'country') && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
+ if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
{
$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
}
else
{
+ $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
$aSearch['iSearchRank'] += 1000; // skip;
}
}
{
if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
{
- if (($sPhraseType != 'street') && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
+ if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
{
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1;
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
+ {
+ $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version?
+ {
+ foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
+ {
+ if (empty($aSearchTermToken['country_code'])
+ && empty($aSearchTermToken['lat'])
+ && empty($aSearchTermToken['class']))
+ {
+ $aSearch = $aCurrentSearch;
+ $aSearch['iSearchRank'] += 1;
+ $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
+ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ }
+ }
+ else
+ {
+ $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
}
if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 2;
if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
+ $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ else
+ $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
$aSearch['iNamePhrase'] = $iPhrase;
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
// TODO: filter out the pointless search terms (2 letter name tokens and less)
// they might be right - but they are just too darned expensive to run
if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";
+ if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";
if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress'])
{
// For infrequent name terms disable index usage for address
if (CONST_Search_NameOnlySearchFrequencyThreshold &&
sizeof($aSearch['aName']) == 1 &&
- $aPossibleMainWordIDs[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
+ $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
{
- $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]";
+ $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
}
else
{
$aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
+ if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";
}
}
if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
- $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end';
-
+ $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)';
if ($sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END";
if ($sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END";
$aOrder[] = "$sImportanceSQL DESC";
+ if (sizeof($aSearch['aFullNameAddress']))
+ $aOrder[] = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) DESC';
+
if (sizeof($aTerms))
{
else
$sSQL .= " limit ".$iLimit;
- if (CONST_Debug) { var_dump($sSQL); }
+ if (CONST_Debug) var_dump($sSQL);
+ $iStartTime = time();
$aViewBoxPlaceIDs = $oDB->getAll($sSQL);
if (PEAR::IsError($aViewBoxPlaceIDs))
{
failInternalError("Could not get places for search terms.", $sSQL, $aViewBoxPlaceIDs);
}
+ if (time() - $iStartTime > 60) {
+ file_put_contents(CONST_BasePath.'/log/long_queries.log', date('Y-m-d H:i:s', $iStartTime).' '.$sSQL."\n", FILE_APPEND);
+ }
+
//var_dump($aViewBoxPlaceIDs);
// Did we have an viewbox matches?
$aPlaceIDs = array();
$sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
$sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
// $sSQL .= $sOrderSQL." as porder, ";
- $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance ";
+ $sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance, ";
+ $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(placex.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";
$sSQL .= "from placex where place_id in ($sPlaceIDs) ";
$sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
if ($sAllowedTypesSQLList) $sSQL .= "and placex.class in $sAllowedTypesSQLList ";
$sSQL .= "and linked_place_id is null ";
$sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
if (!$bDeDupe) $sSQL .= ",place_id";
- $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
- $sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
- $sSQL .= ",get_name_by_language(name, ARRAY['ref']) ";
+ $sSQL .= ",langaddress ";
+ $sSQL .= ",placename ";
+ $sSQL .= ",ref ";
$sSQL .= " union ";
$sSQL .= "select 'T' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id,'us' as country_code,";
$sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
$sSQL .= "null as ref,";
$sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
// $sSQL .= $sOrderSQL." as porder, ";
- $sSQL .= "-0.15 as importance ";
+ $sSQL .= "-0.15 as importance, ";
+ $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";
$sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
$sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
$sSQL .= "group by place_id";
$sSQL .= "null as ref,";
$sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
// $sSQL .= $sOrderSQL." as porder, ";
- $sSQL .= "-0.10 as importance ";
+ $sSQL .= "-0.10 as importance, ";
+ $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";
$sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
$sSQL .= "and 30 between $iMinAddressRank and $iMaxAddressRank ";
$sSQL .= "group by place_id";
//var_Dump($aSearchResults);
//exit;
$aClassType = getClassTypesWithImportance();
- $aRecheckWords = preg_split('/\b/',$sQuery);
+ $aRecheckWords = preg_split('/\b/u',$sQuery);
foreach($aRecheckWords as $i => $sWord)
{
if (!$sWord) unset($aRecheckWords[$i]);
{
if (stripos($sAddress, $sWord)!==false) $iCountWords++;
}
+
$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
//if (CONST_Debug) var_dump($aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']);
}
*/
$aResult['name'] = $aResult['langaddress'];
- $aResult['foundorder'] = $iResNum;
+ $aResult['foundorder'] = -$aResult['addressimportance'];
$aSearchResults[$iResNum] = $aResult;
}
uasort($aSearchResults, 'byImportance');
if (sizeof($aSearchResults) >= $iFinalLimit) break;
}
- $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '1 day'::interval,'YYYY/MM/DD') from import_status limit 1");
+ $sDataDate = $oDB->getOne("select TO_CHAR(lastimportdate - '2 minutes'::interval,'YYYY/MM/DD HH24:MI')||' GMT' from import_status limit 1");
if (isset($_GET['nearlat']) && isset($_GET['nearlon']))
{
projects / nominatim.git / blobdiff