}
if (sizeof($aStructuredQuery) > 0) {
$sQuery = join(', ', $aStructuredQuery);
- $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ if ($iMaxAddressRank < 30)
+ {
+ $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ }
}
if ($sQuery)
// Start with a blank search
$aSearches = array(
array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(),
+ 'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(),
'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'')
);
{
foreach($aSearchWords as $aSearchTerm)
{
- $aNewSearch = $aSearch;
+ $aNewSearch = $aSearch;
if ($aSearchTerm['country_code'])
{
$aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
{
// Check which tokens we have, get the ID numbers
- $sSQL = 'select word_id,word_token, word, class, type, location, country_code, operator';
+ $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;
+ $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
}
if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
{
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank']++;
- if (($sPhraseType == '' || $sPhraseType == 'country') && $aSearchTerm['country_code'] !== null && $aSearchTerm['country_code'] != '0')
+ if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0')
{
if ($aSearch['sCountryCode'] === false)
{
$aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
// Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
- if ($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) $aSearch['iSearchRank'] += 5;
+ // If reverse order is enabled, it may appear at the beginning as well.
+ if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) &&
+ (!$bReverseInPlan || $iToken > 0 || $iPhrase > 0))
+ {
+ $aSearch['iSearchRank'] += 5;
+ }
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
{
if (sizeof($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']] = $aSearchTerm['word_id'];
}
{
if (isset($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;
}
{
$iQueryLoop++;
+ if (CONST_Debug) { echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>"; }
+ if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
+
+
// Must have a location term
if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon'])
{
{
if (4 >= $iMinAddressRank && 4 <= $iMaxAddressRank)
{
- $sSQL = "select place_id from placex where country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
$sSQL .= " order by st_area(geometry) desc limit 1";
+ if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $oDB->getCol($sSQL);
}
}
{
if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue;
if (!$aSearch['sClass']) continue;
- if (CONST_Debug) var_dump('<hr>',$aSearch);
- if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
$sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
if ($oDB->getOne($sSQL))
{
$sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
$sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
+ if (sizeof($aExcludePlaceIDs))
+ {
+ $sSQL .= " and place_id not in (".join(',',$aExcludePlaceIDs).")";
+ }
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $oDB->getCol($sSQL);
- if (!sizeof($aPlaceIDs))
+ // If excluded place IDs are given, it is fair to assume that
+ // there have been results in the small box, so no further
+ // expansion in that case.
+ if (!sizeof($aPlaceIDs) && !sizeof($aExcludePlaceIDs))
{
$sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
$sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
{
$sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
$sSQL .= " and st_contains($sViewboxSmallSQL, geometry) and linked_place_id is null";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
}
else
{
- if (CONST_Debug) var_dump('<hr>',$aSearch);
- if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
$aPlaceIDs = array();
-
+
// First we need a position, either aName or fLat or both
$aTerms = array();
$aOrder = array();
// TODO: filter out the pointless search terms (2 letter name tokens and less)
// they might be right - but they are just too darned expensive to run
if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";
- if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
+ 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 &&
+ $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
+ {
+ $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($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 ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)";
if ($aSearch['fLon'] && $aSearch['fLat'])
// If they were searching for a named class (i.e. 'Kings Head pub') then we might have an extra match
$sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
$sSQL .= " and linked_place_id is null";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
$sSQL .= " order by rank_search asc limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
$aClassPlaceIDs = $oDB->getCol($sSQL);
{
$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
if ($iOffset) $sSQL .= " offset $iOffset";
$sSQL .= " limit $iLimit";
{
$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " and l.country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)";
if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
if ($iOffset) $sSQL .= " offset $iOffset";
$sSQL .= " limit $iLimit";
failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs);
}
- if (CONST_Debug) var_Dump($aPlaceIDs);
+ if (CONST_Debug) { echo "<br><b>Place IDs:</b> "; var_Dump($aPlaceIDs); }
foreach($aPlaceIDs as $iPlaceID)
{
if ($iGroupLoop > 4) break;
if ($iQueryLoop > 30) break;
}
-//exit;
+
// Did we find anything?
if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs))
{
$sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' ';
}
$sOrderSQL .= ' ELSE 10000000 END';
- $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,country_code,";
+ $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,calculated_country_code as country_code,";
$sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
$sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
$sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
$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,country_code,importance";
+ $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
if (!$bDeDupe) $sSQL .= ",place_id";
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= "order by importance desc";
// $sSQL .= "order by rank_search,rank_address,porder asc";
- if (CONST_Debug) var_dump('<hr>',$sSQL);
+ if (CONST_Debug) { echo "<hr>"; var_dump($sSQL); }
$aSearchResults = $oDB->getAll($sSQL);
//var_dump($sSQL,$aSearchResults);exit;
$sOrderSQL .= 'when min(place_id) = '.$iPlaceID.' then '.$iOrder.' ';
}
$sOrderSQL .= ' ELSE 10000000 END';
- $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,country_code,";
+ $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,calculated_country_code as country_code,";
$sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
$sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,";
$sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,";
$sSQL .= "coalesce(importance,0.75-(rank_search::float/40)) as importance ";
$sSQL .= "from placex where place_id in ($sPlaceIDs) ";
$sSQL .= "and placex.rank_address between $iMinAddressRank and $iMaxAddressRank ";
- $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,country_code,importance";
+ $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance";
if (!$bDeDupe) $sSQL .= ",place_id";
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= ",get_name_by_language(name, $sLanguagePrefArraySQL) ";
$sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
$sSQL .= "order by importance desc";
// $sSQL .= "order by rank_search,rank_address,porder asc";
- if (CONST_Debug) var_dump('<hr>',$sSQL);
+ if (CONST_Debug) { echo "<hr>", var_dump($sSQL); }
$aSearchResults = $oDB->getAll($sSQL);
//var_dump($sSQL,$aSearchResults);exit;
projects / nominatim.git / blobdiff