<?php
+ @define('CONST_ConnectionBucket_PageType', 'Search');
+
require_once(dirname(dirname(__FILE__)).'/lib/init-website.php');
require_once(CONST_BasePath.'/lib/log.php');
ini_set('memory_limit', '200M');
+
$oDB =& getDB();
// Display defaults
$sSuggestion = $sSuggestionURL = false;
$bDeDupe = isset($_GET['dedupe'])?(bool)$_GET['dedupe']:true;
$bReverseInPlan = false;
- $iLimit = isset($_GET['limit'])?(int)$_GET['limit']:10;
+ $iFinalLimit = isset($_GET['limit'])?(int)$_GET['limit']:10;
$iOffset = isset($_GET['offset'])?(int)$_GET['offset']:0;
$iMaxRank = 20;
- if ($iLimit > 100) $iLimit = 100;
+ if ($iFinalLimit > 50) $iFinalLimit = 50;
+ $iLimit = $iFinalLimit + min($iFinalLimit, 10);
$iMinAddressRank = 0;
$iMaxAddressRank = 30;
+ $sAllowedTypesSQLList = false;
// Format for output
if (isset($_GET['format']) && ($_GET['format'] == 'html' || $_GET['format'] == 'xml' || $_GET['format'] == 'json' || $_GET['format'] == 'jsonv2'))
}
// Show / use polygons
- $bShowPolygons = isset($_GET['polygon']) && $_GET['polygon'];
+ $bShowPolygons = (boolean)isset($_GET['polygon']) && $_GET['polygon'];
+ if ($sOutputFormat == 'html') {
+ $bAsText = $bShowPolygons;
+ $bShowPolygons = false;
+ $bAsGeoJSON = false;
+ $bAsKML = false;
+ $bAsSVG = false;
+ } else {
+ $bAsGeoJSON = (boolean)isset($_GET['polygon_geojson']) && $_GET['polygon_geojson'];
+ $bAsKML = (boolean)isset($_GET['polygon_kml']) && $_GET['polygon_kml'];
+ $bAsSVG = (boolean)isset($_GET['polygon_svg']) && $_GET['polygon_svg'];
+ $bAsText = (boolean)isset($_GET['polygon_text']) && $_GET['polygon_text'];
+ if ((($bShowPolygons?1:0)
+ + ($bAsGeoJSON?1:0)
+ + ($bAsKML?1:0)
+ + ($bAsSVG?1:0)
+ + ($bAsText?1:0)
+ ) > CONST_PolygonOutput_MaximumTypes) {
+ if (CONST_PolygonOutput_MaximumTypes) {
+ userError("Select only ".CONST_PolygonOutput_MaximumTypes." polgyon output option");
+ } else {
+ userError("Polygon output is disabled");
+ }
+ exit;
+ }
+ }
// Show address breakdown
$bShowAddressDetails = isset($_GET['addressdetails']) && $_GET['addressdetails'];
$aPhrases = array_reverse($aPhrases);
$sQuery = join(', ',$aPhrases);
}
+
+ function structuredAddressElement(&$aStructuredQuery, &$iMinAddressRank, &$iMaxAddressRank, $aParams, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank)
+ {
+ if (!isset($_GET[$sKey])) return false;
+ $sValue = trim($_GET[$sKey]);
+ if (!$sValue) return false;
+ $aStructuredQuery[$sKey] = $sValue;
+ if ($iMinAddressRank == 0 && $iMaxAddressRank == 30) {
+ $iMinAddressRank = $iNewMinAddressRank;
+ $iMaxAddressRank = $iNewMaxAddressRank;
+ }
+ return true;
+ }
+
+ // Structured query?
+ $aStructuredOptions = array(
+ array('amenity', 26, 30),
+ array('street', 26, 30),
+ array('city', 14, 24),
+ array('county', 9, 13),
+ array('state', 8, 8),
+ array('country', 4, 4),
+ array('postalcode', 16, 25),
+ );
+ $aStructuredQuery = array();
+ $sAllowedTypesSQLList = '';
+ foreach($aStructuredOptions as $aStructuredOption)
+ {
+ loadStructuredAddressElement($aStructuredQuery, $iMinAddressRank, $iMaxAddressRank, $_GET, $aStructuredOption[0], $aStructuredOption[1], $aStructuredOption[2]);
+ }
+ if (sizeof($aStructuredQuery) > 0) {
+ $sQuery = join(', ', $aStructuredQuery);
+ if ($iMaxAddressRank < 30)
+ {
+ $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ }
+ }
+
if ($sQuery)
{
$hLog = logStart($oDB, 'search', $sQuery, $aLangPrefOrder);
$aPoints = explode(',',$_GET['route']);
if (sizeof($aPoints) % 2 != 0)
{
- echo "Uneven number of points";
+ userError("Uneven number of points");
exit;
}
$sViewboxCentreSQL = "ST_SetSRID('LINESTRING(";
failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL);
}
$sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry";
+ $bBoundingBoxSearch = true;
}
// Do we have anything that looks like a lat/lon pair?
}
}
- if ($sQuery)
+ if ($sQuery || $aStructuredQuery)
{
// 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'=>'')
);
preg_match_all('/\\[([a-zA-Z]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
$aSpecialTerms = array();
+ if (isset($aStructuredQuery['amenity']) && $aStructuredQuery['amenity'])
+ {
+ $aSpecialTermsRaw[] = array('['.$aStructuredQuery['amenity'].']', $aStructuredQuery['amenity']);
+ unset($aStructuredQuery['amenity']);
+ }
foreach($aSpecialTermsRaw as $aSpecialTerm)
{
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
{
foreach($aSearchWords as $aSearchTerm)
{
- $aNewSearch = $aSearch;
+ $aNewSearch = $aSearch;
if ($aSearchTerm['country_code'])
{
$aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
// Split query into phrases
// Commas are used to reduce the search space by indicating where phrases split
- $aPhrases = explode(',',$sQuery);
+ if (sizeof($aStructuredQuery) > 0)
+ {
+ $aPhrases = $aStructuredQuery;
+ $bStructuredPhrases = true;
+ }
+ else
+ {
+ $aPhrases = explode(',',$sQuery);
+ $bStructuredPhrases = false;
+ }
+
// Convert each phrase to standard form
// Create a list of standard words
$aPhrase = $oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string");
if (PEAR::isError($aPhrase))
{
- echo "Illegal query string (not an UTF-8 string): ".$sPhrase;
+ userError("Illegal query string (not an UTF-8 string): ".$sPhrase);
if (CONST_Debug) var_dump($aPhrase);
exit;
}
}
// reindex phrases - we make assumptions later on
+ $aPhraseTypes = array_keys($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, 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;
+ 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, $aValidTokens);
Calculate all searches using aValidTokens i.e.
'Wodsworth Road, Sheffield' =>
-
+
Phrase Wordset
0 0 (wodsworth road)
0 1 (wodsworth)(road)
1 0 (sheffield)
-
+
Score how good the search is so they can be ordered
*/
foreach($aPhrases as $iPhrase => $sPhrase)
{
$aNewPhraseSearches = array();
+ if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
+ else $sPhraseType = '';
foreach($aPhrases[$iPhrase]['wordsets'] as $aWordset)
{
{
//echo "<br><b>$sToken</b>";
$aNewWordsetSearches = array();
-
+
foreach($aWordsetSearches as $aCurrentSearch)
{
//echo "<i>";
{
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank']++;
- if ($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[] = $aSearch;
}
}
if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
- elseif ($aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house')
+ elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house')
{
if ($aSearch['sHouseNumber'] === '')
{
*/
}
}
- elseif ($aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null)
+ elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null)
{
if ($aSearch['sClass'] === '')
{
{
if (sizeof($aSearch['aName']))
{
- if (!isset($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']) && $aSearchTerm['word_id'])
{
-//var_Dump('<hr>',$aSearch['aName']);
-
- if (sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
+ if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
+ {
+ $aSearch = $aCurrentSearch;
+ $aSearch['iSearchRank'] += 1;
+ if ($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'];
- $aSearch['iNamePhrase'] = $iPhrase;
- if ($aSearch['iSearchRank'] < $iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
+ if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
+ {
+ $aSearch = $aCurrentSearch;
+ $aSearch['iSearchRank'] += 2;
+ if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
+ if ($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[] = $aSearch;
+ }
}
}
}
}
}
+ if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0)
+ {
+ $aCopyGroupedSearches = $aGroupedSearches;
+ foreach($aCopyGroupedSearches as $iGroup => $aSearches)
+ {
+ foreach($aSearches as $iSearch => $aSearch)
+ {
+ $aReductionsList = array($aSearch['aAddress']);
+ $iSearchRank = $aSearch['iSearchRank'];
+ while(sizeof($aReductionsList) > 0)
+ {
+ $iSearchRank += 5;
+ if ($iSearchRank > iMaxRank) break 3;
+ $aNewReductionsList = array();
+ foreach($aReductionsList as $aReductionsWordList)
+ {
+ for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++)
+ {
+ $aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1));
+ $aReverseSearch = $aSearch;
+ $aSearch['aAddress'] = $aReductionsWordListResult;
+ $aSearch['iSearchRank'] = $iSearchRank;
+ $aGroupedSearches[$iSearchRank][] = $aReverseSearch;
+ if (sizeof($aReductionsWordListResult) > 0)
+ {
+ $aNewReductionsList[] = $aReductionsWordListResult;
+ }
+ }
+ }
+ $aReductionsList = $aNewReductionsList;
+ }
+ }
+ }
+ ksort($aGroupedSearches);
+ }
+
// Filter out duplicate searches
$aSearchHash = array();
foreach($aGroupedSearches as $iGroup => $aSearches)
{
$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 ($aSearch['sCountryCode'] && !$aSearch['sClass'])
+ if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber'])
{
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);
else
{
$sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
- $sSQL .= " and st_contains($sViewboxSmallSQL, geometry)";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ $sSQL .= " and st_contains($sViewboxSmallSQL, geometry) and linked_place_id is null";
+ 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'] && $aSearch['fLat'])
if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL";
if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
- $sImportanceSQL = 'case when importance = 0 OR importance IS NULL then 0.92-(search_rank::float/33) 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))
{
{
$sPlaceIDs = join(',',$aPlaceIDs);
+ $aClassPlaceIDs = array();
+
if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'name')
{
// 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']."'";
- if ($sCountryCodesSQL) $sSQL .= " and country_code in ($sCountryCodesSQL)";
+ $sSQL .= " and linked_place_id is null";
+ if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
$sSQL .= " order by rank_search asc limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $oDB->getCol($sSQL);
+ $aClassPlaceIDs = $oDB->getCol($sSQL);
}
if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') // & in
if ($iMaxRank < 9 && $bCacheTable)
{
// Try and get a polygon to search in instead
- $sSQL = "select geometry from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank + 5 and st_geometrytype(geometry) in ('ST_Polygon','ST_MultiPolygon') order by rank_search asc limit 1";
- if (CONST_Debug) var_dump($sSQL);
- $sPlaceGeom = $oDB->getOne($sSQL);
+ $sSQL = "select geometry from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank + 5 and st_geometrytype(geometry) in ('ST_Polygon','ST_MultiPolygon') order by rank_search asc limit 1";
+ if (CONST_Debug) var_dump($sSQL);
+ $sPlaceGeom = $oDB->getOne($sSQL);
}
if ($sPlaceGeom)
else
{
$iMaxRank += 5;
- $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $oDB->getCol($sSQL);
- $sPlaceIDs = join(',',$aPlaceIDs);
+ $sSQL = "select place_id from placex where place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
+ if (CONST_Debug) var_dump($sSQL);
+ $aPlaceIDs = $oDB->getCol($sSQL);
+ $sPlaceIDs = join(',',$aPlaceIDs);
}
if ($sPlaceIDs || $sPlaceGeom)
if ($sNearPointSQL) $sOrderBySQL = "ST_Distance($sNearPointSQL, l.centroid)";
else if ($sPlaceIDs) $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
else if ($sPlaceGeom) $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
-
- $sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
+
+ $sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
if ($sPlaceIDs)
{
$sSQL .= ",placex as f where ";
- $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, st_centroid(f.geometry), $fRange) ";
+ $sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, f.centroid, $fRange) ";
}
if ($sPlaceGeom)
{
{
$sSQL .= " and l.place_id not in (".join(',',$aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
- if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc";
+ if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
+ if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
if ($iOffset) $sSQL .= " offset $iOffset";
$sSQL .= " limit $iLimit";
if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $oDB->getCol($sSQL);
+ $aClassPlaceIDs = array_merge($aClassPlaceIDs, $oDB->getCol($sSQL));
}
else
{
else $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
$sSQL = "select distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'')." from placex as l,placex as f where ";
- $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, st_centroid(f.geometry), $fRange) ";
+ $sSQL .= "f.place_id in ( $sPlaceIDs) and ST_DWithin(l.geometry, f.centroid, $fRange) ";
$sSQL .= "and l.class='".$aSearch['sClass']."' and l.type='".$aSearch['sType']."' ";
if (sizeof($aExcludePlaceIDs))
{
$sSQL .= " and l.place_id not in (".join(',',$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 $iLimit";
if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $oDB->getCol($sSQL);
+ $aClassPlaceIDs = array_merge($aClassPlaceIDs, $oDB->getCol($sSQL));
}
}
}
+
+ $aPlaceIDs = $aClassPlaceIDs;
+
}
}
failInternalError("Could not get place IDs from tokens." ,$sSQL, $aPlaceIDs);
}
- if (CONST_Debug) var_Dump($aPlaceIDs);
+ if (CONST_Debug) { echo "<br><b>Place IDs:</b> "; var_Dump($aPlaceIDs); }
foreach($aPlaceIDs as $iPlaceID)
{
}
if ($iQueryLoop > 20) break;
}
+
//exit;
- if (sizeof($aResultPlaceIDs)) break;
+ if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break;
if ($iGroupLoop > 4) break;
if ($iQueryLoop > 30) break;
}
-//exit;
+
// Did we find anything?
- if (sizeof($aResultPlaceIDs))
+ if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs))
{
//var_Dump($aResultPlaceIDs);exit;
// Get the details for display (is this a redundant extra step?)
$sOrderSQL .= 'when min(place_id) = '.$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 .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, ";
+ $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
// $sSQL .= $sOrderSQL." as porder, ";
- $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) 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,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_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";
$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 .= "avg(ST_X(ST_Centroid(geometry))) as lon,avg(ST_Y(ST_Centroid(geometry))) as lat, ";
+ $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
// $sSQL .= $sOrderSQL." as porder, ";
- $sSQL .= "coalesce(importance,0.9-(rank_search::float/30)) as importance ";
+ $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;
//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]);
$sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,";
$sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,";
$sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),2)) as maxlat,";
- $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon,";
- $sSQL .= "ST_AsText(geometry) as outlinestring from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\'';
+ $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon";
+ if ($bAsGeoJSON) $sSQL .= ",ST_AsGeoJSON(geometry) as asgeojson";
+ if ($bAsKML) $sSQL .= ",ST_AsKML(geometry) as askml";
+ if ($bAsSVG) $sSQL .= ",ST_AsSVG(geometry) as assvg";
+ if ($bAsText || $bShowPolygons) $sSQL .= ",ST_AsText(geometry) as astext";
+ $sSQL .= " from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\'';
$aPointPolygon = $oDB->getRow($sSQL);
if (PEAR::IsError($aPointPolygon))
{
}
if ($aPointPolygon['place_id'])
{
+ if ($bAsGeoJSON) $aResult['asgeojson'] = $aPointPolygon['asgeojson'];
+ if ($bAsKML) $aResult['askml'] = $aPointPolygon['askml'];
+ if ($bAsSVG) $aResult['assvg'] = $aPointPolygon['assvg'];
+ if ($bAsText) $aResult['astext'] = $aPointPolygon['astext'];
+
if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) {
$aResult['lat'] = $aPointPolygon['centrelat'];
$aResult['lon'] = $aPointPolygon['centrelon'];
}
- // Translate geometary string to point array
- if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['outlinestring'],$aMatch))
- {
- preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
- }
- elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['outlinestring'],$aMatch))
+ if ($bShowPolygons)
{
- $fRadius = 0.01;
- $iSteps = ($fRadius * 40000)^2;
- $fStepSize = (2*pi())/$iSteps;
- $aPolyPoints = array();
- for($f = 0; $f < 2*pi(); $f += $fStepSize)
+ // Translate geometary string to point array
+ if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
{
- $aPolyPoints[] = array('',$aMatch[1]+($fRadius*sin($f)),$aMatch[2]+($fRadius*cos($f)));
+ preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
+ }
+ elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
+ {
+ preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
+ }
+ elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch))
+ {
+ $fRadius = 0.01;
+ $iSteps = ($fRadius * 40000)^2;
+ $fStepSize = (2*pi())/$iSteps;
+ $aPolyPoints = array();
+ for($f = 0; $f < 2*pi(); $f += $fStepSize)
+ {
+ $aPolyPoints[] = array('',$aMatch[1]+($fRadius*sin($f)),$aMatch[2]+($fRadius*cos($f)));
+ }
+ $aPointPolygon['minlat'] = $aPointPolygon['minlat'] - $fRadius;
+ $aPointPolygon['maxlat'] = $aPointPolygon['maxlat'] + $fRadius;
+ $aPointPolygon['minlon'] = $aPointPolygon['minlon'] - $fRadius;
+ $aPointPolygon['maxlon'] = $aPointPolygon['maxlon'] + $fRadius;
}
- $aPointPolygon['minlat'] = $aPointPolygon['minlat'] - $fRadius;
- $aPointPolygon['maxlat'] = $aPointPolygon['maxlat'] + $fRadius;
- $aPointPolygon['minlon'] = $aPointPolygon['minlon'] - $fRadius;
- $aPointPolygon['maxlon'] = $aPointPolygon['maxlon'] + $fRadius;
}
// Output data suitable for display (points and a bounding box)
{
if (stripos($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');
-//var_dump($aSearchResults);exit;
-
$aOSMIDDone = array();
$aClassTypeNameDone = array();
$aToFilter = $aSearchResults;
$bFirst = false;
}
if (!$bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
- && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']])))
+ && !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name']])))
{
$aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
- $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['name']] = true;
+ $aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name']] = true;
$aSearchResults[] = $aResult;
}
// Absolute limit on number of results
- if (sizeof($aSearchResults) >= $iLimit) break;
+ if (sizeof($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($_GET['nearlon']))
{
logEnd($oDB, $hLog, sizeof($aToFilter));
}
$sMoreURL = CONST_Website_BaseURL.'search?format='.urlencode($sOutputFormat).'&exclude_place_ids='.join(',',$aExcludePlaceIDs);
- $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"];
+ if (isset($_SERVER["HTTP_ACCEPT_LANGUAGE"])) $sMoreURL .= '&accept-language='.$_SERVER["HTTP_ACCEPT_LANGUAGE"];
if ($bShowPolygons) $sMoreURL .= '&polygon=1';
if ($bShowAddressDetails) $sMoreURL .= '&addressdetails=1';
if (isset($_GET['viewbox']) && $_GET['viewbox']) $sMoreURL .= '&viewbox='.urlencode($_GET['viewbox']);
projects / nominatim.git / blobdiff