hans/Nominatim
1
0
Fork 1
mirror of https://github.com/osm-search/Nominatim.git synced 2026-03-11 05:14:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

performance: for low frequency name terms disable use of postgresql address index. Theshold is probably about 1000 but optimal value will probably depend on hardware.

Browse Source
This commit is contained in:
Brian Quinion
2012-12-20 17:47:57 +00:00
parent ae7f676b73
commit d1a224bad4
2 changed files with 20 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
settings/settings.php
View File

@@ -60,6 +60,7 @@
@define('CONST_Suggestions_Enabled', false); @define('CONST_Suggestions_Enabled', false);
@define('CONST_Search_TryDroppedAddressTerms', false); @define('CONST_Search_TryDroppedAddressTerms', false);
@define('CONST_Search_NameOnlySearchFrequencyThreshold', false);
// Set to zero to disable polygon output // Set to zero to disable polygon output
@define('CONST_PolygonOutput_MaximumTypes', 1); @define('CONST_PolygonOutput_MaximumTypes', 1);

25
website/search.php
View File

@@ -324,7 +324,7 @@
{ {
foreach($aSearchWords as $aSearchTerm) foreach($aSearchWords as $aSearchTerm)
{ {
$aNewSearch = $aSearch; $aNewSearch = $aSearch;
if ($aSearchTerm['country_code']) if ($aSearchTerm['country_code'])
{ {
$aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']); $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
@@ -392,7 +392,7 @@
{ {
// Check which tokens we have, get the ID numbers // 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)).')'; $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$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'; // $sSQL .= ' group by word_token, word, class, type, location, country_code';
@@ -419,7 +419,7 @@
{ {
$aValidTokens[$aToken['word_token']] = array($aToken); $aValidTokens[$aToken['word_token']] = array($aToken);
} }
if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1; if ($aToken['word_token'][0]==' ' && !$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1 + $aToken['search_name_count'];
} }
if (CONST_Debug) var_Dump($aPhrases, $aValidTokens); if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
@@ -527,7 +527,7 @@
{ {
$aSearch = $aCurrentSearch; $aSearch = $aCurrentSearch;
$aSearch['iSearchRank']++; $aSearch['iSearchRank']++;
if (($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) if ($aSearch['sCountryCode'] === false)
{ {
@@ -860,7 +860,7 @@
if (CONST_Debug) var_dump('<hr>',$aSearch); if (CONST_Debug) var_dump('<hr>',$aSearch);
if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens); if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
$aPlaceIDs = array(); $aPlaceIDs = array();
// First we need a position, either aName or fLat or both // First we need a position, either aName or fLat or both
$aTerms = array(); $aTerms = array();
$aOrder = array(); $aOrder = array();
@@ -868,7 +868,20 @@
// TODO: filter out the pointless search terms (2 letter name tokens and less) // TODO: filter out the pointless search terms (2 letter name tokens and less)
// they might be right - but they are just too darned expensive to run // they might be right - but they are just too darned expensive to run
if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]"; if (sizeof($aSearch['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['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)
{
$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]";
}
else
{
$aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
}
}
if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)"; if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank in (26,27)";
if ($aSearch['fLon'] && $aSearch['fLat']) if ($aSearch['fLon'] && $aSearch['fLat'])