Nominatim/lib/Geocode.php

1869 lines
98 KiB
PHP
Raw Normal View History

<?php
namespace Nominatim;
2017-03-15 22:43:08 +03:00
require_once(CONST_BasePath.'/lib/NearPoint.php');
2016-09-04 04:19:48 +03:00
require_once(CONST_BasePath.'/lib/PlaceLookup.php');
require_once(CONST_BasePath.'/lib/ReverseGeocode.php');
class Geocode
{
protected $oDB;
protected $aLangPrefOrder = array();
protected $bIncludeAddressDetails = false;
protected $bIncludeExtraTags = false;
protected $bIncludeNameDetails = false;
protected $bIncludePolygonAsPoints = false;
protected $bIncludePolygonAsText = false;
protected $bIncludePolygonAsGeoJSON = false;
protected $bIncludePolygonAsKML = false;
protected $bIncludePolygonAsSVG = false;
protected $fPolygonSimplificationThreshold = 0.0;
protected $aExcludePlaceIDs = array();
protected $bDeDupe = true;
protected $bReverseInPlan = false;
protected $iLimit = 20;
protected $iFinalLimit = 10;
protected $iOffset = 0;
protected $bFallback = false;
protected $aCountryCodes = false;
protected $bBoundedSearch = false;
protected $aViewBox = false;
protected $sViewboxCentreSQL = false;
protected $sViewboxSmallSQL = false;
protected $sViewboxLargeSQL = false;
protected $iMaxRank = 20;
protected $iMinAddressRank = 0;
protected $iMaxAddressRank = 30;
protected $aAddressRankList = array();
protected $exactMatchCache = array();
protected $sAllowedTypesSQLList = false;
protected $sQuery = false;
protected $aStructuredQuery = false;
public function __construct(&$oDB)
2016-09-04 04:19:48 +03:00
{
$this->oDB =& $oDB;
}
public function setReverseInPlan($bReverse)
2016-09-04 04:19:48 +03:00
{
$this->bReverseInPlan = $bReverse;
}
public function setLanguagePreference($aLangPref)
2016-09-04 04:19:48 +03:00
{
$this->aLangPrefOrder = $aLangPref;
}
public function getIncludeAddressDetails()
2016-09-04 04:19:48 +03:00
{
return $this->bIncludeAddressDetails;
}
public function getIncludeExtraTags()
2016-09-04 04:19:48 +03:00
{
return $this->bIncludeExtraTags;
}
public function getIncludeNameDetails()
2016-09-04 04:19:48 +03:00
{
return $this->bIncludeNameDetails;
}
public function setIncludePolygonAsPoints($b = true)
2016-09-04 04:19:48 +03:00
{
$this->bIncludePolygonAsPoints = $b;
}
public function setIncludePolygonAsText($b = true)
2016-09-04 04:19:48 +03:00
{
$this->bIncludePolygonAsText = $b;
}
public function setIncludePolygonAsGeoJSON($b = true)
2016-09-04 04:19:48 +03:00
{
$this->bIncludePolygonAsGeoJSON = $b;
}
public function setIncludePolygonAsKML($b = true)
2016-09-04 04:19:48 +03:00
{
$this->bIncludePolygonAsKML = $b;
}
public function setIncludePolygonAsSVG($b = true)
2016-09-04 04:19:48 +03:00
{
$this->bIncludePolygonAsSVG = $b;
}
public function setPolygonSimplificationThreshold($f)
2016-09-04 04:19:48 +03:00
{
$this->fPolygonSimplificationThreshold = $f;
}
public function setLimit($iLimit = 10)
2016-09-04 04:19:48 +03:00
{
if ($iLimit > 50) $iLimit = 50;
if ($iLimit < 1) $iLimit = 1;
$this->iFinalLimit = $iLimit;
$this->iLimit = $iLimit + min($iLimit, 10);
}
public function getExcludedPlaceIDs()
2016-09-04 04:19:48 +03:00
{
return $this->aExcludePlaceIDs;
}
public function getCountryCodes()
{
return $this->aCountryCodes;
}
public function getViewBoxString()
2016-09-04 04:19:48 +03:00
{
if (!$this->aViewBox) return null;
return $this->aViewBox[0].','.$this->aViewBox[3].','.$this->aViewBox[2].','.$this->aViewBox[1];
}
public function setFeatureType($sFeatureType)
2016-09-04 04:19:48 +03:00
{
switch ($sFeatureType) {
case 'country':
$this->setRankRange(4, 4);
break;
case 'state':
$this->setRankRange(8, 8);
break;
case 'city':
$this->setRankRange(14, 16);
break;
case 'settlement':
$this->setRankRange(8, 20);
break;
2016-09-04 04:19:48 +03:00
}
}
public function setRankRange($iMin, $iMax)
2016-09-04 04:19:48 +03:00
{
$this->iMinAddressRank = $iMin;
$this->iMaxAddressRank = $iMax;
}
public function setRoute($aRoutePoints, $fRouteWidth)
2016-09-04 04:19:48 +03:00
{
$this->aViewBox = false;
$this->sViewboxCentreSQL = "ST_SetSRID('LINESTRING(";
$sSep = '';
2016-10-09 22:41:23 +03:00
foreach ($aRoutePoints as $aPoint) {
2016-09-04 04:19:48 +03:00
$fPoint = (float)$aPoint;
$this->sViewboxCentreSQL .= $sSep.$fPoint;
$sSep = ($sSep == ' ') ? ',' : ' ';
}
$this->sViewboxCentreSQL .= ")'::geometry,4326)";
$this->sViewboxSmallSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL;
2016-09-04 04:19:48 +03:00
$this->sViewboxSmallSQL .= ','.($fRouteWidth/69).')';
$this->sViewboxLargeSQL = 'ST_BUFFER('.$this->sViewboxCentreSQL;
2016-09-04 04:19:48 +03:00
$this->sViewboxLargeSQL .= ','.($fRouteWidth/30).')';
}
public function setViewbox($aViewbox)
2016-09-04 04:19:48 +03:00
{
$this->aViewBox = array_map('floatval', $aViewbox);
$this->aViewBox[0] = max(-180.0, min(180, $this->aViewBox[0]));
$this->aViewBox[1] = max(-90.0, min(90, $this->aViewBox[1]));
$this->aViewBox[2] = max(-180.0, min(180, $this->aViewBox[2]));
$this->aViewBox[3] = max(-90.0, min(90, $this->aViewBox[3]));
if (abs($this->aViewBox[0] - $this->aViewBox[2]) < 0.000000001
|| abs($this->aViewBox[1] - $this->aViewBox[3]) < 0.000000001
) {
userError("Bad parameter 'viewbox'. Not a box.");
}
2016-09-04 04:19:48 +03:00
$fHeight = $this->aViewBox[0] - $this->aViewBox[2];
$fWidth = $this->aViewBox[1] - $this->aViewBox[3];
$aBigViewBox[0] = $this->aViewBox[0] + $fHeight;
$aBigViewBox[2] = $this->aViewBox[2] - $fHeight;
$aBigViewBox[1] = $this->aViewBox[1] + $fWidth;
$aBigViewBox[3] = $this->aViewBox[3] - $fWidth;
$this->sViewboxCentreSQL = false;
$this->sViewboxSmallSQL = sprintf(
'ST_SetSRID(ST_MakeBox2D(ST_Point(%F,%F),ST_Point(%F,%F)),4326)',
$this->aViewBox[0],
$this->aViewBox[1],
$this->aViewBox[2],
$this->aViewBox[3]
);
$this->sViewboxLargeSQL = sprintf(
'ST_SetSRID(ST_MakeBox2D(ST_Point(%F,%F),ST_Point(%F,%F)),4326)',
$aBigViewBox[0],
$aBigViewBox[1],
$aBigViewBox[2],
$aBigViewBox[3]
);
2016-09-04 04:19:48 +03:00
}
public function setQuery($sQueryString)
2016-09-04 04:19:48 +03:00
{
$this->sQuery = $sQueryString;
$this->aStructuredQuery = false;
}
public function getQueryString()
2016-09-04 04:19:48 +03:00
{
return $this->sQuery;
}
public function loadParamArray($oParams)
2016-09-04 04:19:48 +03:00
{
$this->bIncludeAddressDetails
= $oParams->getBool('addressdetails', $this->bIncludeAddressDetails);
$this->bIncludeExtraTags
= $oParams->getBool('extratags', $this->bIncludeExtraTags);
$this->bIncludeNameDetails
= $oParams->getBool('namedetails', $this->bIncludeNameDetails);
2016-09-04 04:19:48 +03:00
$this->bBoundedSearch = $oParams->getBool('bounded', $this->bBoundedSearch);
$this->bDeDupe = $oParams->getBool('dedupe', $this->bDeDupe);
$this->setLimit($oParams->getInt('limit', $this->iFinalLimit));
$this->iOffset = $oParams->getInt('offset', $this->iOffset);
$this->bFallback = $oParams->getBool('fallback', $this->bFallback);
// List of excluded Place IDs - used for more acurate pageing
$sExcluded = $oParams->getStringList('exclude_place_ids');
if ($sExcluded) {
foreach ($sExcluded as $iExcludedPlaceID) {
2016-09-04 04:19:48 +03:00
$iExcludedPlaceID = (int)$iExcludedPlaceID;
if ($iExcludedPlaceID)
$aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
}
if (isset($aExcludePlaceIDs))
$this->aExcludePlaceIDs = $aExcludePlaceIDs;
}
// Only certain ranks of feature
$sFeatureType = $oParams->getString('featureType');
if (!$sFeatureType) $sFeatureType = $oParams->getString('featuretype');
if ($sFeatureType) $this->setFeatureType($sFeatureType);
// Country code list
$sCountries = $oParams->getStringList('countrycodes');
if ($sCountries) {
foreach ($sCountries as $sCountryCode) {
if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) {
2016-09-04 04:19:48 +03:00
$aCountries[] = strtolower($sCountryCode);
}
}
if (isset($aCountries))
2016-09-04 04:19:48 +03:00
$this->aCountryCodes = $aCountries;
}
$aViewbox = $oParams->getStringList('viewboxlbrt');
if ($aViewbox) {
if (count($aViewbox) != 4) {
userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
}
2016-09-04 04:19:48 +03:00
$this->setViewbox($aViewbox);
} else {
2016-09-04 04:19:48 +03:00
$aViewbox = $oParams->getStringList('viewbox');
if ($aViewbox) {
if (count($aViewbox) != 4) {
userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
}
$this->setViewBox(array(
$aViewbox[0],
$aViewbox[3],
$aViewbox[2],
$aViewbox[1]
));
} else {
2016-09-04 04:19:48 +03:00
$aRoute = $oParams->getStringList('route');
$fRouteWidth = $oParams->getFloat('routewidth');
if ($aRoute && $fRouteWidth) {
2016-09-04 04:19:48 +03:00
$this->setRoute($aRoute, $fRouteWidth);
}
}
}
}
public function setQueryFromParams($oParams)
2016-09-04 04:19:48 +03:00
{
// Search query
$sQuery = $oParams->getString('q');
if (!$sQuery) {
$this->setStructuredQuery(
$oParams->getString('amenity'),
$oParams->getString('street'),
$oParams->getString('city'),
$oParams->getString('county'),
$oParams->getString('state'),
$oParams->getString('country'),
$oParams->getString('postalcode')
);
2016-09-04 04:19:48 +03:00
$this->setReverseInPlan(false);
} else {
2016-09-04 04:19:48 +03:00
$this->setQuery($sQuery);
}
}
public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
2016-09-04 04:19:48 +03:00
{
$sValue = trim($sValue);
if (!$sValue) return false;
$this->aStructuredQuery[$sKey] = $sValue;
if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
2016-09-04 04:19:48 +03:00
$this->iMinAddressRank = $iNewMinAddressRank;
$this->iMaxAddressRank = $iNewMaxAddressRank;
}
if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
return true;
}
public function setStructuredQuery($sAmentiy = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
2016-09-04 04:19:48 +03:00
{
$this->sQuery = false;
// Reset
$this->iMinAddressRank = 0;
$this->iMaxAddressRank = 30;
$this->aAddressRankList = array();
$this->aStructuredQuery = array();
$this->sAllowedTypesSQLList = '';
$this->loadStructuredAddressElement($sAmentiy, 'amenity', 26, 30, false);
$this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
$this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
$this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
$this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
$this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
2016-09-04 04:19:48 +03:00
$this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
if (sizeof($this->aStructuredQuery) > 0) {
2016-09-04 04:19:48 +03:00
$this->sQuery = join(', ', $this->aStructuredQuery);
if ($this->iMaxAddressRank < 30) {
2016-09-04 04:19:48 +03:00
$sAllowedTypesSQLList = '(\'place\',\'boundary\')';
}
}
}
public function fallbackStructuredQuery()
2016-09-04 04:19:48 +03:00
{
if (!$this->aStructuredQuery) return false;
$aParams = $this->aStructuredQuery;
if (sizeof($aParams) == 1) return false;
$aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
foreach ($aOrderToFallback as $sType) {
if (isset($aParams[$sType])) {
2016-09-04 04:19:48 +03:00
unset($aParams[$sType]);
$this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
return true;
}
}
return false;
}
public function getDetails($aPlaceIDs)
2016-09-04 04:19:48 +03:00
{
//$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1
if (sizeof($aPlaceIDs) == 0) return array();
2016-09-04 04:19:48 +03:00
$sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]";
2016-09-04 04:19:48 +03:00
// Get the details for display (is this a redundant extra step?)
$sPlaceIDs = join(',', array_keys($aPlaceIDs));
$sImportanceSQL = '';
if ($this->sViewboxSmallSQL) $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
if ($this->sViewboxLargeSQL) $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
$sSQL = "SELECT ";
$sSQL .= " osm_type,";
$sSQL .= " osm_id,";
$sSQL .= " class,";
$sSQL .= " type,";
$sSQL .= " admin_level,";
$sSQL .= " rank_search,";
$sSQL .= " rank_address,";
$sSQL .= " min(place_id) AS place_id, ";
$sSQL .= " min(parent_place_id) AS parent_place_id, ";
$sSQL .= " calculated_country_code AS country_code, ";
$sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
$sSQL .= " get_name_by_language(name, $sLanguagePrefArraySQL) AS placename,";
$sSQL .= " get_name_by_language(name, ARRAY['ref']) AS ref,";
if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text AS extra,";
if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text AS names,";
$sSQL .= " avg(ST_X(centroid)) AS lon, ";
$sSQL .= " avg(ST_Y(centroid)) AS lat, ";
$sSQL .= " ".$sImportanceSQL."COALESCE(importance,0.75-(rank_search::float/40)) AS importance, ";
$sSQL .= " ( ";
$sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
$sSQL .= " FROM ";
$sSQL .= " place_addressline s, ";
$sSQL .= " placex p";
$sSQL .= " WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
$sSQL .= " AND p.place_id = s.address_place_id ";
$sSQL .= " AND s.isaddress ";
$sSQL .= " AND p.importance is not null ";
$sSQL .= " ) AS addressimportance, ";
$sSQL .= " (extratags->'place') AS extra_place ";
$sSQL .= " FROM placex";
$sSQL .= " WHERE place_id in ($sPlaceIDs) ";
$sSQL .= " AND (";
$sSQL .= " placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
$sSQL .= " OR (extratags->'place') = 'city'";
}
if ($this->aAddressRankList) {
$sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
}
$sSQL .= " ) ";
if ($this->sAllowedTypesSQLList) {
$sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
}
$sSQL .= " AND linked_place_id is null ";
$sSQL .= " GROUP BY ";
$sSQL .= " osm_type, ";
$sSQL .= " osm_id, ";
$sSQL .= " class, ";
$sSQL .= " type, ";
$sSQL .= " admin_level, ";
$sSQL .= " rank_search, ";
$sSQL .= " rank_address, ";
$sSQL .= " calculated_country_code, ";
$sSQL .= " importance, ";
if (!$this->bDeDupe) $sSQL .= "place_id,";
$sSQL .= " langaddress, ";
$sSQL .= " placename, ";
$sSQL .= " ref, ";
if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
if ($this->bIncludeNameDetails) $sSQL .= "name, ";
$sSQL .= " extratags->'place' ";
2016-09-04 04:19:48 +03:00
if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
// only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
2016-09-04 04:19:48 +03:00
// with start- and endnumber, the common osm housenumbers are usually saved as points
$sHousenumbers = "";
$i = 0;
$length = count($aPlaceIDs);
foreach ($aPlaceIDs as $placeID => $housenumber) {
2016-09-04 04:19:48 +03:00
$i++;
$sHousenumbers .= "(".$placeID.", ".$housenumber.")";
if ($i<$length) $sHousenumbers .= ", ";
2016-09-04 04:19:48 +03:00
}
if (CONST_Use_US_Tiger_Data) {
// Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
2016-09-04 04:19:48 +03:00
$sSQL .= " union";
$sSQL .= " SELECT ";
$sSQL .= " 'T' AS osm_type, ";
$sSQL .= " (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
$sSQL .= " 'place' AS class, ";
$sSQL .= " 'house' AS type, ";
$sSQL .= " null AS admin_level, ";
$sSQL .= " 30 AS rank_search, ";
$sSQL .= " 30 AS rank_address, ";
$sSQL .= " min(place_id) AS place_id, ";
$sSQL .= " min(parent_place_id) AS parent_place_id, ";
$sSQL .= " 'us' AS country_code, ";
$sSQL .= " get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress,";
$sSQL .= " null AS placename, ";
$sSQL .= " null AS ref, ";
if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
$sSQL .= " avg(st_x(centroid)) AS lon, ";
$sSQL .= " avg(st_y(centroid)) AS lat,";
$sSQL .= " ".$sImportanceSQL."-1.15 AS importance, ";
$sSQL .= " (";
$sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
$sSQL .= " FROM ";
$sSQL .= " place_addressline s, ";
$sSQL .= " placex p";
$sSQL .= " WHERE s.place_id = min(blub.parent_place_id)";
$sSQL .= " AND p.place_id = s.address_place_id ";
$sSQL .= " AND s.isaddress";
$sSQL .= " AND p.importance is not null";
$sSQL .= " ) AS addressimportance, ";
$sSQL .= " null AS extra_place ";
$sSQL .= " FROM (";
$sSQL .= " SELECT place_id, "; // interpolate the Tiger housenumbers here
$sSQL .= " ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
$sSQL .= " parent_place_id, ";
$sSQL .= " housenumber_for_place";
$sSQL .= " FROM (";
$sSQL .= " location_property_tiger ";
$sSQL .= " JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
$sSQL .= " WHERE ";
$sSQL .= " housenumber_for_place>=0";
$sSQL .= " AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
$sSQL .= " ) AS blub"; //postgres wants an alias here
$sSQL .= " GROUP BY";
$sSQL .= " place_id, ";
$sSQL .= " housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
2016-09-04 04:19:48 +03:00
if (!$this->bDeDupe) $sSQL .= ", place_id ";
}
// osmline
// interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
$sSQL .= " UNION ";
$sSQL .= "SELECT ";
$sSQL .= " 'W' AS osm_type, ";
$sSQL .= " osm_id, ";
$sSQL .= " 'place' AS class, ";
$sSQL .= " 'house' AS type, ";
$sSQL .= " null AS admin_level, ";
$sSQL .= " 30 AS rank_search, ";
$sSQL .= " 30 AS rank_address, ";
$sSQL .= " min(place_id) as place_id, ";
$sSQL .= " min(parent_place_id) AS parent_place_id, ";
$sSQL .= " calculated_country_code AS country_code, ";
$sSQL .= " get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
$sSQL .= " null AS placename, ";
$sSQL .= " null AS ref, ";
if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
$sSQL .= " AVG(st_x(centroid)) AS lon, ";
$sSQL .= " AVG(st_y(centroid)) AS lat, ";
$sSQL .= " ".$sImportanceSQL."-0.1 AS importance, "; // slightly smaller than the importance for normal houses with rank 30, which is 0
$sSQL .= " (";
$sSQL .= " SELECT ";
$sSQL .= " MAX(p.importance*(p.rank_address+2)) ";
$sSQL .= " FROM";
$sSQL .= " place_addressline s, ";
$sSQL .= " placex p";
$sSQL .= " WHERE s.place_id = min(blub.parent_place_id) ";
$sSQL .= " AND p.place_id = s.address_place_id ";
$sSQL .= " AND s.isaddress ";
$sSQL .= " AND p.importance is not null";
$sSQL .= " ) AS addressimportance,";
$sSQL .= " null AS extra_place ";
$sSQL .= " FROM (";
$sSQL .= " SELECT ";
$sSQL .= " osm_id, ";
$sSQL .= " place_id, ";
$sSQL .= " calculated_country_code, ";
$sSQL .= " CASE "; // interpolate the housenumbers here
$sSQL .= " WHEN startnumber != endnumber ";
$sSQL .= " THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
$sSQL .= " ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
$sSQL .= " END as centroid, ";
$sSQL .= " parent_place_id, ";
$sSQL .= " housenumber_for_place ";
$sSQL .= " FROM (";
$sSQL .= " location_property_osmline ";
$sSQL .= " JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
$sSQL .= " ) ";
$sSQL .= " WHERE housenumber_for_place>=0 ";
$sSQL .= " AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
$sSQL .= " ) as blub"; //postgres wants an alias here
$sSQL .= " GROUP BY ";
$sSQL .= " osm_id, ";
$sSQL .= " place_id, ";
$sSQL .= " housenumber_for_place, ";
$sSQL .= " calculated_country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
2016-09-04 04:19:48 +03:00
if (!$this->bDeDupe) $sSQL .= ", place_id ";
if (CONST_Use_Aux_Location_data) {
$sSQL .= " UNION ";
$sSQL .= " SELECT ";
$sSQL .= " 'L' AS osm_type, ";
$sSQL .= " place_id AS osm_id, ";
$sSQL .= " 'place' AS class,";
$sSQL .= " 'house' AS type, ";
$sSQL .= " null AS admin_level, ";
$sSQL .= " 0 AS rank_search,";
$sSQL .= " 0 AS rank_address, ";
$sSQL .= " min(place_id) AS place_id,";
$sSQL .= " min(parent_place_id) AS parent_place_id, ";
$sSQL .= " 'us' AS country_code, ";
$sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress, ";
$sSQL .= " null AS placename, ";
$sSQL .= " null AS ref, ";
if ($this->bIncludeExtraTags) $sSQL .= "null AS extra, ";
if ($this->bIncludeNameDetails) $sSQL .= "null AS names, ";
$sSQL .= " avg(ST_X(centroid)) AS lon, ";
$sSQL .= " avg(ST_Y(centroid)) AS lat, ";
$sSQL .= " ".$sImportanceSQL."-1.10 AS importance, ";
$sSQL .= " ( ";
$sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
$sSQL .= " FROM ";
$sSQL .= " place_addressline s, ";
$sSQL .= " placex p";
$sSQL .= " WHERE s.place_id = min(location_property_aux.parent_place_id)";
$sSQL .= " AND p.place_id = s.address_place_id ";
$sSQL .= " AND s.isaddress";
$sSQL .= " AND p.importance is not null";
$sSQL .= " ) AS addressimportance, ";
$sSQL .= " null AS extra_place ";
$sSQL .= " FROM location_property_aux ";
$sSQL .= " WHERE place_id in ($sPlaceIDs) ";
$sSQL .= " AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
$sSQL .= " GROUP BY ";
$sSQL .= " place_id, ";
if (!$this->bDeDupe) $sSQL .= "place_id, ";
$sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
2016-09-04 04:19:48 +03:00
}
}
$sSQL .= " order by importance desc";
if (CONST_Debug) {
echo "<hr>";
var_dump($sSQL);
}
$aSearchResults = chksql(
$this->oDB->getAll($sSQL),
"Could not get details for place."
);
2016-09-04 04:19:48 +03:00
return $aSearchResults;
}
public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases)
2016-09-04 04:19:48 +03:00
{
/*
Calculate all searches using aValidTokens i.e.
'Wodsworth Road, Sheffield' =>
Phrase Wordset
0 0 (wodsworth road)
0 1 (wodsworth)(road)
1 0 (sheffield)
Score how good the search is so they can be ordered
*/
foreach ($aPhrases as $iPhrase => $sPhrase) {
2016-09-04 04:19:48 +03:00
$aNewPhraseSearches = array();
if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
else $sPhraseType = '';
foreach ($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset) {
2016-09-04 04:19:48 +03:00
// Too many permutations - too expensive
if ($iWordSet > 120) break;
$aWordsetSearches = $aSearches;
// Add all words from this wordset
foreach ($aWordset as $iToken => $sToken) {
2016-09-04 04:19:48 +03:00
//echo "<br><b>$sToken</b>";
$aNewWordsetSearches = array();
foreach ($aWordsetSearches as $aCurrentSearch) {
2016-09-04 04:19:48 +03:00
//echo "<i>";
//var_dump($aCurrentSearch);
//echo "</i>";
// If the token is valid
if (isset($aValidTokens[' '.$sToken])) {
foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
2016-09-04 04:19:48 +03:00
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank']++;
if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0') {
if ($aSearch['sCountryCode'] === false) {
2016-09-04 04:19:48 +03:00
$aSearch['sCountryCode'] = strtolower($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))) {
2016-09-04 04:19:48 +03:00
$aSearch['iSearchRank'] += 5;
}
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
} elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) {
2017-03-17 00:04:30 +03:00
if ($aSearch['oNear'] === false) {
$aSearch['oNear'] = new NearPoint(
$aSearchTerm['lat'],
$aSearchTerm['lon'],
$aSearchTerm['radius']);
2016-09-04 04:19:48 +03:00
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
} elseif ($sPhraseType == 'postalcode') {
2016-09-04 04:19:48 +03:00
// We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
2016-09-04 04:19:48 +03:00
// If we already have a name try putting the postcode first
if (sizeof($aSearch['aName'])) {
2016-09-04 04:19:48 +03:00
$aNewSearch = $aSearch;
$aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
$aNewSearch['aName'] = array();
$aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;
}
if (sizeof($aSearch['aName'])) {
if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) {
2016-09-04 04:19:48 +03:00
$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
} else {
2016-09-04 04:19:48 +03:00
$aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
$aSearch['iSearchRank'] += 1000; // skip;
}
} else {
2016-09-04 04:19:48 +03:00
$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
//$aSearch['iNamePhrase'] = $iPhrase;
}
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
} elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') {
if ($aSearch['sHouseNumber'] === '') {
2016-09-04 04:19:48 +03:00
$aSearch['sHouseNumber'] = $sToken;
// sanity check: if the housenumber is not mainly made
// up of numbers, add a penalty
if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;
// also housenumbers should appear in the first or second phrase
if ($iPhrase > 1) $aSearch['iSearchRank'] += 1;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
/*
// Fall back to not searching for this item (better than nothing)
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
*/
}
} elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) {
if ($aSearch['sClass'] === '') {
2016-09-04 04:19:48 +03:00
$aSearch['sOperator'] = $aSearchTerm['operator'];
$aSearch['sClass'] = $aSearchTerm['class'];
$aSearch['sType'] = $aSearchTerm['type'];
if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name';
else $aSearch['sOperator'] = 'near'; // near = in for the moment
if (strlen($aSearchTerm['operator']) == 0) $aSearch['iSearchRank'] += 1;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
} elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
if (sizeof($aSearch['aName'])) {
if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) {
2016-09-04 04:19:48 +03:00
$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
} else {
2016-09-04 04:19:48 +03:00
$aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
$aSearch['iSearchRank'] += 1000; // skip;
}
} else {
2016-09-04 04:19:48 +03:00
$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
//$aSearch['iNamePhrase'] = $iPhrase;
}
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
}
// Look for partial matches.
// Note that there is no point in adding country terms here
// because country are omitted in the address.
if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
2016-09-04 04:19:48 +03:00
// Allow searching for a word - but at extra cost
foreach ($aValidTokens[$sToken] as $aSearchTerm) {
if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false) {
2016-09-04 04:19:48 +03:00
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1;
if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) {
2016-09-04 04:19:48 +03:00
$aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
} elseif (isset($aValidTokens[' '.$sToken])) { // revert to the token version?
2016-09-04 04:19:48 +03:00
$aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
$aSearch['iSearchRank'] += 1;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
foreach ($aValidTokens[' '.$sToken] as $aSearchTermToken) {
2016-09-04 04:19:48 +03:00
if (empty($aSearchTermToken['country_code'])
&& empty($aSearchTermToken['lat'])
&& empty($aSearchTermToken['class'])
) {
2016-09-04 04:19:48 +03:00
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1;
$aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
} else {
2016-09-04 04:19:48 +03:00
$aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) {
2016-09-04 04:19:48 +03:00
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1;
if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;
if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) {
2016-09-04 04:19:48 +03:00
$aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
} else {
2016-09-04 04:19:48 +03:00
$aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
}
2016-09-04 04:19:48 +03:00
$aSearch['iNamePhrase'] = $iPhrase;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
}
} else {
2016-09-04 04:19:48 +03:00
// Allow skipping a word - but at EXTREAM cost
//$aSearch = $aCurrentSearch;
//$aSearch['iSearchRank']+=100;
//$aNewWordsetSearches[] = $aSearch;
}
}
// Sort and cut
usort($aNewWordsetSearches, 'bySearchRank');
$aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
}
//var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
$aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
usort($aNewPhraseSearches, 'bySearchRank');
$aSearchHash = array();
foreach ($aNewPhraseSearches as $iSearch => $aSearch) {
2016-09-04 04:19:48 +03:00
$sHash = serialize($aSearch);
if (isset($aSearchHash[$sHash])) unset($aNewPhraseSearches[$iSearch]);
else $aSearchHash[$sHash] = 1;
}
$aNewPhraseSearches = array_slice($aNewPhraseSearches, 0, 50);
}
// Re-group the searches by their score, junk anything over 20 as just not worth trying
$aGroupedSearches = array();
foreach ($aNewPhraseSearches as $aSearch) {
if ($aSearch['iSearchRank'] < $this->iMaxRank) {
2016-09-04 04:19:48 +03:00
if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
$aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
}
}
ksort($aGroupedSearches);
$iSearchCount = 0;
$aSearches = array();
foreach ($aGroupedSearches as $iScore => $aNewSearches) {
2016-09-04 04:19:48 +03:00
$iSearchCount += sizeof($aNewSearches);
$aSearches = array_merge($aSearches, $aNewSearches);
if ($iSearchCount > 50) break;
}
//if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
}
return $aGroupedSearches;
}
/* Perform the actual query lookup.
Returns an ordered list of results, each with the following fields:
osm_type: type of corresponding OSM object
N - node
W - way
R - relation
P - postcode (internally computed)
osm_id: id of corresponding OSM object
class: general object class (corresponds to tag key of primary OSM tag)
type: subclass of object (corresponds to tag value of primary OSM tag)
admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
rank_search: rank in search hierarchy
(see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
rank_address: rank in address hierarchy (determines orer in address)
place_id: internal key (may differ between different instances)
country_code: ISO country code
langaddress: localized full address
placename: localized name of object
ref: content of ref tag (if available)
lon: longitude
lat: latitude
importance: importance of place based on Wikipedia link count
addressimportance: cumulated importance of address elements
extra_place: type of place (for admin boundaries, if there is a place tag)
aBoundingBox: bounding Box
label: short description of the object class/type (English only)
name: full name (currently the same as langaddress)
foundorder: secondary ordering for places with same importance
*/
public function lookup()
2016-09-04 04:19:48 +03:00
{
if (!$this->sQuery && !$this->aStructuredQuery) return false;
$sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]";
2016-09-04 04:19:48 +03:00
$sCountryCodesSQL = false;
if ($this->aCountryCodes) {
2016-09-04 04:19:48 +03:00
$sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes));
}
$sQuery = $this->sQuery;
2016-10-12 23:25:04 +03:00
if (!preg_match('//u', $sQuery)) {
userError("Query string is not UTF-8 encoded.");
}
2016-09-04 04:19:48 +03:00
// Conflicts between US state abreviations and various words for 'the' in different languages
if (isset($this->aLangPrefOrder['name:en'])) {
$sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
$sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
$sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
2016-09-04 04:19:48 +03:00
}
$bBoundingBoxSearch = $this->bBoundedSearch && $this->sViewboxSmallSQL;
if ($this->sViewboxCentreSQL) {
2016-09-04 04:19:48 +03:00
// For complex viewboxes (routes) precompute the bounding geometry
$sGeom = chksql(
$this->oDB->getOne("select ".$this->sViewboxSmallSQL),
"Could not get small viewbox"
);
2016-09-04 04:19:48 +03:00
$this->sViewboxSmallSQL = "'".$sGeom."'::geometry";
$sGeom = chksql(
$this->oDB->getOne("select ".$this->sViewboxLargeSQL),
"Could not get large viewbox"
);
2016-09-04 04:19:48 +03:00
$this->sViewboxLargeSQL = "'".$sGeom."'::geometry";
}
// Do we have anything that looks like a lat/lon pair?
2017-03-15 22:43:08 +03:00
$oNearPoint = false;
if ($aLooksLike = NearPoint::extractFromQuery($sQuery)) {
$oNearPoint = $aLooksLike['pt'];
2016-09-04 04:19:48 +03:00
$sQuery = $aLooksLike['query'];
}
$aSearchResults = array();
if ($sQuery || $this->aStructuredQuery) {
2016-09-04 04:19:48 +03:00
// Start with a blank search
$aSearches = array(
array(
'iSearchRank' => 0,
'iNamePhrase' => -1,
'sCountryCode' => false,
'aName' => array(),
'aAddress' => array(),
'aFullNameAddress' => array(),
'aNameNonSearch' => array(),
'aAddressNonSearch' => array(),
'sOperator' => '',
'aFeatureName' => array(),
'sClass' => '',
'sType' => '',
'sHouseNumber' => '',
2017-03-17 00:04:30 +03:00
'oNear' => $oNearPoint
)
);
2016-09-04 04:19:48 +03:00
// Any 'special' terms in the search?
$bSpecialTerms = false;
preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
$aSpecialTerms = array();
foreach ($aSpecialTermsRaw as $aSpecialTerm) {
2016-09-04 04:19:48 +03:00
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
$aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2];
}
preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
$aSpecialTerms = array();
if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) {
2016-09-04 04:19:48 +03:00
$aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
unset($this->aStructuredQuery['amenity']);
}
foreach ($aSpecialTermsRaw as $aSpecialTerm) {
2016-09-04 04:19:48 +03:00
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
$sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".$aSpecialTerm[1]."') AS string"));
$sSQL = 'SELECT * ';
$sSQL .= 'FROM ( ';
$sSQL .= ' SELECT word_id, word_token, word, class, type, country_code, operator';
$sSQL .= ' FROM word ';
$sSQL .= ' WHERE word_token in (\' '.$sToken.'\')';
$sSQL .= ') AS x ';
$sSQL .= ' WHERE (class is not null AND class not in (\'place\')) ';
$sSQL .= ' OR country_code is not null';
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_Dump($sSQL);
$aSearchWords = chksql($this->oDB->getAll($sSQL));
$aNewSearches = array();
foreach ($aSearches as $aSearch) {
foreach ($aSearchWords as $aSearchTerm) {
2016-09-04 04:19:48 +03:00
$aNewSearch = $aSearch;
if ($aSearchTerm['country_code']) {
2016-09-04 04:19:48 +03:00
$aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
$aNewSearches[] = $aNewSearch;
$bSpecialTerms = true;
}
if ($aSearchTerm['class']) {
2016-09-04 04:19:48 +03:00
$aNewSearch['sClass'] = $aSearchTerm['class'];
$aNewSearch['sType'] = $aSearchTerm['type'];
$aNewSearches[] = $aNewSearch;
$bSpecialTerms = true;
}
}
}
$aSearches = $aNewSearches;
}
// Split query into phrases
// Commas are used to reduce the search space by indicating where phrases split
if ($this->aStructuredQuery) {
2016-09-04 04:19:48 +03:00
$aPhrases = $this->aStructuredQuery;
$bStructuredPhrases = true;
} else {
$aPhrases = explode(',', $sQuery);
2016-09-04 04:19:48 +03:00
$bStructuredPhrases = false;
}
// Convert each phrase to standard form
// Create a list of standard words
// Get all 'sets' of words
// Generate a complete list of all
$aTokens = array();
foreach ($aPhrases as $iPhrase => $sPhrase) {
$aPhrase = chksql(
$this->oDB->getRow("SELECT make_standard_name('".pg_escape_string($sPhrase)."') as string"),
2016-10-12 21:21:12 +03:00
"Cannot normalize query string (is it a UTF-8 string?)"
);
if (trim($aPhrase['string'])) {
2016-09-04 04:19:48 +03:00
$aPhrases[$iPhrase] = $aPhrase;
$aPhrases[$iPhrase]['words'] = explode(' ', $aPhrases[$iPhrase]['string']);
2016-09-04 04:19:48 +03:00
$aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0);
$aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets']));
} else {
2016-09-04 04:19:48 +03:00
unset($aPhrases[$iPhrase]);
}
}
// Reindex phrases - we make assumptions later on that they are numerically keyed in order
$aPhraseTypes = array_keys($aPhrases);
$aPhrases = array_values($aPhrases);
if (sizeof($aTokens)) {
2016-09-04 04:19:48 +03:00
// 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", $aTokens)).')';
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_Dump($sSQL);
$aValidTokens = array();
if (sizeof($aTokens)) {
$aDatabaseWords = chksql(
$this->oDB->getAll($sSQL),
"Could not get word tokens."
);
} else {
2016-09-04 04:19:48 +03:00
$aDatabaseWords = array();
}
$aPossibleMainWordIDs = array();
$aWordFrequencyScores = array();
foreach ($aDatabaseWords as $aToken) {
2016-09-04 04:19:48 +03:00
// Very special case - require 2 letter country param to match the country code found
if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
&& strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']
) {
2016-09-04 04:19:48 +03:00
continue;
}
if (isset($aValidTokens[$aToken['word_token']])) {
2016-09-04 04:19:48 +03:00
$aValidTokens[$aToken['word_token']][] = $aToken;
} else {
2016-09-04 04:19:48 +03:00
$aValidTokens[$aToken['word_token']] = array($aToken);
}
if (!$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, $aValidTokens);
// Try and calculate GB postcodes we might be missing
foreach ($aTokens as $sToken) {
2016-09-04 04:19:48 +03:00
// Source of gb postcodes is now definitive - always use
if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData)) {
if (substr($aData[1], -2, 1) != ' ') {
$aData[0] = substr($aData[0], 0, strlen($aData[1])-1).' '.substr($aData[0], strlen($aData[1])-1);
$aData[1] = substr($aData[1], 0, -1).' '.substr($aData[1], -1, 1);
2016-09-04 04:19:48 +03:00
}
$aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $this->oDB);
if ($aGBPostcodeLocation) {
2016-09-04 04:19:48 +03:00
$aValidTokens[$sToken] = $aGBPostcodeLocation;
}
} elseif (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
// US ZIP+4 codes - if there is no token,
// merge in the 5-digit ZIP code
if (isset($aValidTokens[$aData[1]])) {
foreach ($aValidTokens[$aData[1]] as $aToken) {
if (!$aToken['class']) {
if (isset($aValidTokens[$sToken])) {
2016-09-04 04:19:48 +03:00
$aValidTokens[$sToken][] = $aToken;
} else {
2016-09-04 04:19:48 +03:00
$aValidTokens[$sToken] = array($aToken);
}
}
}
}
}
}
foreach ($aTokens as $sToken) {
2016-09-04 04:19:48 +03:00
// Unknown single word token with a number - assume it is a house number
if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken, ' ') === false && preg_match('/[0-9]/', $sToken)) {
$aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house'));
2016-09-04 04:19:48 +03:00
}
}
// Any words that have failed completely?
// TODO: suggestions
// Start the search process
// array with: placeid => -1 | tiger-housenumber
$aResultPlaceIDs = array();
$aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
if ($this->bReverseInPlan) {
2016-09-04 04:19:48 +03:00
// Reverse phrase array and also reverse the order of the wordsets in
// the first and final phrase. Don't bother about phrases in the middle
// because order in the address doesn't matter.
$aPhrases = array_reverse($aPhrases);
$aPhrases[0]['wordsets'] = getInverseWordSets($aPhrases[0]['words'], 0);
if (sizeof($aPhrases) > 1) {
2016-09-04 04:19:48 +03:00
$aFinalPhrase = end($aPhrases);
$aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
}
$aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
foreach ($aGroupedSearches as $aSearches) {
foreach ($aSearches as $aSearch) {
if ($aSearch['iSearchRank'] < $this->iMaxRank) {
2016-09-04 04:19:48 +03:00
if (!isset($aReverseGroupedSearches[$aSearch['iSearchRank']])) $aReverseGroupedSearches[$aSearch['iSearchRank']] = array();
$aReverseGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
}
}
}
$aGroupedSearches = $aReverseGroupedSearches;
ksort($aGroupedSearches);
}
} else {
2016-09-04 04:19:48 +03:00
// Re-group the searches by their score, junk anything over 20 as just not worth trying
$aGroupedSearches = array();
foreach ($aSearches as $aSearch) {
if ($aSearch['iSearchRank'] < $this->iMaxRank) {
2016-09-04 04:19:48 +03:00
if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
$aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
}
}
ksort($aGroupedSearches);
}
if (CONST_Debug) var_Dump($aGroupedSearches);
if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) {
2016-09-04 04:19:48 +03:00
$aCopyGroupedSearches = $aGroupedSearches;
foreach ($aCopyGroupedSearches as $iGroup => $aSearches) {
foreach ($aSearches as $iSearch => $aSearch) {
2016-09-04 04:19:48 +03:00
$aReductionsList = array($aSearch['aAddress']);
$iSearchRank = $aSearch['iSearchRank'];
while (sizeof($aReductionsList) > 0) {
2016-09-04 04:19:48 +03:00
$iSearchRank += 5;
if ($iSearchRank > iMaxRank) break 3;
$aNewReductionsList = array();
foreach ($aReductionsList as $aReductionsWordList) {
for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++) {
2016-09-04 04:19:48 +03:00
$aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1));
$aReverseSearch = $aSearch;
$aSearch['aAddress'] = $aReductionsWordListResult;
$aSearch['iSearchRank'] = $iSearchRank;
$aGroupedSearches[$iSearchRank][] = $aReverseSearch;
if (sizeof($aReductionsWordListResult) > 0) {
2016-09-04 04:19:48 +03:00
$aNewReductionsList[] = $aReductionsWordListResult;
}
}
}
$aReductionsList = $aNewReductionsList;
}
}
}
ksort($aGroupedSearches);
}
// Filter out duplicate searches
$aSearchHash = array();
foreach ($aGroupedSearches as $iGroup => $aSearches) {
foreach ($aSearches as $iSearch => $aSearch) {
2016-09-04 04:19:48 +03:00
$sHash = serialize($aSearch);
if (isset($aSearchHash[$sHash])) {
2016-09-04 04:19:48 +03:00
unset($aGroupedSearches[$iGroup][$iSearch]);
if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
} else {
2016-09-04 04:19:48 +03:00
$aSearchHash[$sHash] = 1;
}
}
}
if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
$iGroupLoop = 0;
$iQueryLoop = 0;
foreach ($aGroupedSearches as $iGroupedRank => $aSearches) {
2016-09-04 04:19:48 +03:00
$iGroupLoop++;
foreach ($aSearches as $aSearch) {
2016-09-04 04:19:48 +03:00
$iQueryLoop++;
$searchedHousenumber = -1;
if (CONST_Debug) echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
// No location term?
2017-03-17 00:04:30 +03:00
if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['oNear']) {
if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) {
2016-09-04 04:19:48 +03:00
// Just looking for a country by code - look it up
if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank) {
$sSQL = "SELECT place_id FROM placex WHERE calculated_country_code='".$aSearch['sCountryCode']."' AND rank_search = 4";
if ($sCountryCodesSQL) $sSQL .= " AND calculated_country_code in ($sCountryCodesSQL)";
2016-09-04 04:19:48 +03:00
if ($bBoundingBoxSearch)
$sSQL .= " AND _st_intersects($this->sViewboxSmallSQL, geometry)";
$sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
} else {
2016-09-04 04:19:48 +03:00
$aPlaceIDs = array();
}
} else {
2017-03-17 00:04:30 +03:00
if (!$bBoundingBoxSearch && !$aSearch['oNear']) continue;
2016-09-04 04:19:48 +03:00
if (!$aSearch['sClass']) continue;
$sSQL = "SELECT COUNT(*) FROM pg_tables WHERE tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
if (chksql($this->oDB->getOne($sSQL))) {
$sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
if ($sCountryCodesSQL) $sSQL .= " JOIN placex USING (place_id)";
$sSQL .= " WHERE st_contains($this->sViewboxSmallSQL, ct.centroid)";
if ($sCountryCodesSQL) $sSQL .= " AND calculated_country_code in ($sCountryCodesSQL)";
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
2016-09-04 04:19:48 +03:00
$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
// If excluded place IDs are given, it is fair to assume that
// there have been results in the small box, so no further
// expansion in that case.
// Also don't expand if bounded results were requested.
if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs) && !$this->bBoundedSearch) {
$sSQL = "SELECT place_id FROM place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
2016-09-04 04:19:48 +03:00
if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
$sSQL .= " WHERE ST_Contains($this->sViewboxLargeSQL, ct.centroid)";
if ($sCountryCodesSQL) $sSQL .= " AND calculated_country_code in ($sCountryCodesSQL)";
if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
$sSQL .= " LIMIT $this->iLimit";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
}
} else {
$sSQL = "SELECT place_id ";
$sSQL .= "FROM placex ";
$sSQL .= "WHERE class='".$aSearch['sClass']."' ";
$sSQL .= " AND type='".$aSearch['sType']."'";
$sSQL .= " AND ST_Contains($this->sViewboxSmallSQL, geometry) ";
$sSQL .= " AND linked_place_id is null";
if ($sCountryCodesSQL) $sSQL .= " AND calculated_country_code in ($sCountryCodesSQL)";
if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, centroid) ASC";
$sSQL .= " LIMIT $this->iLimit";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
}
}
2017-03-17 00:04:30 +03:00
} elseif ($aSearch['oNear'] && !sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['sClass']) {
// If a coordinate is given, the search must either
// be for a name or a special search. Ignore everythin else.
2016-09-04 04:19:48 +03:00
$aPlaceIDs = array();
} else {
2016-09-04 04:19:48 +03:00
$aPlaceIDs = array();
// First we need a position, either aName or fLat or both
$aTerms = array();
$aOrder = array();
if ($aSearch['sHouseNumber'] && sizeof($aSearch['aAddress'])) {
2016-09-04 04:19:48 +03:00
$sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M';
$aOrder[] = "";
$aOrder[0] = " (";
$aOrder[0] .= " EXISTS(";
$aOrder[0] .= " SELECT place_id ";
$aOrder[0] .= " FROM placex ";
$aOrder[0] .= " WHERE parent_place_id = search_name.place_id";
$aOrder[0] .= " AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."' ";
$aOrder[0] .= " LIMIT 1";
$aOrder[0] .= " ) ";
2016-09-04 04:19:48 +03:00
// also housenumbers from interpolation lines table are needed
$aOrder[0] .= " OR EXISTS(";
$aOrder[0] .= " SELECT place_id ";
$aOrder[0] .= " FROM location_property_osmline ";
$aOrder[0] .= " WHERE parent_place_id = search_name.place_id";
$aOrder[0] .= " AND startnumber is not NULL";
$aOrder[0] .= " AND ".intval($aSearch['sHouseNumber']).">=startnumber ";
$aOrder[0] .= " AND ".intval($aSearch['sHouseNumber'])."<=endnumber ";
$aOrder[0] .= " LIMIT 1";
$aOrder[0] .= " )";
$aOrder[0] .= " )";
$aOrder[0] .= " DESC";
2016-09-04 04:19:48 +03:00
}
// TODO: filter out the pointless search terms (2 letter name tokens and less)
// they might be right - but they are just too darned expensive to run
if (sizeof($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'] != $aSearch['aAddress']) {
2016-09-04 04:19:48 +03:00
// For infrequent name terms disable index usage for address
if (CONST_Search_NameOnlySearchFrequencyThreshold
&& sizeof($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'], ",")."]";
}
2016-09-04 04:19:48 +03:00
}
}
if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
if ($aSearch['sHouseNumber']) {
2016-09-04 04:19:48 +03:00
$aTerms[] = "address_rank between 16 and 27";
} else {
if ($this->iMinAddressRank > 0) {
2016-09-04 04:19:48 +03:00
$aTerms[] = "address_rank >= ".$this->iMinAddressRank;
}
if ($this->iMaxAddressRank < 30) {
2016-09-04 04:19:48 +03:00
$aTerms[] = "address_rank <= ".$this->iMaxAddressRank;
}
}
2017-03-17 00:04:30 +03:00
if ($aSearch['oNear']) {
$aTerms[] = $aSearch['oNear']->withinSQL('centroid');
2017-03-17 00:04:30 +03:00
$aOrder[] = $aSearch['oNear']->distanceSQL('centroid');
2016-09-04 04:19:48 +03:00
}
if (sizeof($this->aExcludePlaceIDs)) {
$aTerms[] = "place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
if ($sCountryCodesSQL) {
2016-09-04 04:19:48 +03:00
$aTerms[] = "country_code in ($sCountryCodesSQL)";
}
if ($bBoundingBoxSearch) $aTerms[] = "centroid && $this->sViewboxSmallSQL";
2017-03-15 22:43:08 +03:00
if ($oNearPoint) {
$aOrder[] = $oNearPoint->distanceSQL('centroid');
}
2016-09-04 04:19:48 +03:00
if ($aSearch['sHouseNumber']) {
2016-09-04 04:19:48 +03:00
$sImportanceSQL = '- abs(26 - address_rank) + 3';
} else {
$sImportanceSQL = '(CASE WHEN importance = 0 OR importance IS NULL THEN 0.75-(search_rank::float/40) ELSE importance END)';
2016-09-04 04:19:48 +03:00
}
if ($this->sViewboxSmallSQL) $sImportanceSQL .= " * CASE WHEN ST_Contains($this->sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END";
if ($this->sViewboxLargeSQL) $sImportanceSQL .= " * CASE WHEN ST_Contains($this->sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END";
2016-09-04 04:19:48 +03:00
$aOrder[] = "$sImportanceSQL DESC";
if (sizeof($aSearch['aFullNameAddress'])) {
$sExactMatchSQL = ' ( ';
$sExactMatchSQL .= ' SELECT count(*) FROM ( ';
$sExactMatchSQL .= ' SELECT unnest(ARRAY['.join($aSearch['aFullNameAddress'], ",").']) ';
$sExactMatchSQL .= ' INTERSECT ';
$sExactMatchSQL .= ' SELECT unnest(nameaddress_vector)';
$sExactMatchSQL .= ' ) s';
$sExactMatchSQL .= ') as exactmatch';
2016-09-04 04:19:48 +03:00
$aOrder[] = 'exactmatch DESC';
} else {
$sExactMatchSQL = '0::int as exactmatch';
}
if (sizeof($aTerms)) {
$sSQL = "SELECT place_id, ";
2016-09-04 04:19:48 +03:00
$sSQL .= $sExactMatchSQL;
$sSQL .= " FROM search_name";
$sSQL .= " WHERE ".join(' and ', $aTerms);
$sSQL .= " ORDER BY ".join(', ', $aOrder);
if ($aSearch['sHouseNumber'] || $aSearch['sClass']) {
$sSQL .= " LIMIT 20";
} elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass']) {
$sSQL .= " LIMIT 1";
} else {
$sSQL .= " LIMIT ".$this->iLimit;
}
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aViewBoxPlaceIDs = chksql(
$this->oDB->getAll($sSQL),
"Could not get places for search terms."
);
2016-09-04 04:19:48 +03:00
//var_dump($aViewBoxPlaceIDs);
// Did we have an viewbox matches?
$aPlaceIDs = array();
$bViewBoxMatch = false;
foreach ($aViewBoxPlaceIDs as $aViewBoxRow) {
2016-09-04 04:19:48 +03:00
//if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break;
//if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break;
//if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1;
//else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2;
$aPlaceIDs[] = $aViewBoxRow['place_id'];
$this->exactMatchCache[$aViewBoxRow['place_id']] = $aViewBoxRow['exactmatch'];
}
}
//var_Dump($aPlaceIDs);
//exit;
//now search for housenumber, if housenumber provided
if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs)) {
2016-09-04 04:19:48 +03:00
$searchedHousenumber = intval($aSearch['sHouseNumber']);
$aRoadPlaceIDs = $aPlaceIDs;
$sPlaceIDs = join(',', $aPlaceIDs);
2016-09-04 04:19:48 +03:00
// Now they are indexed, look for a house attached to a street we found
$sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M';
$sSQL = "SELECT place_id FROM placex ";
$sSQL .= "WHERE parent_place_id in (".$sPlaceIDs.") and transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
$sSQL .= " LIMIT $this->iLimit";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
2016-10-13 00:30:24 +03:00
2016-09-04 04:19:48 +03:00
// if nothing found, search in the interpolation line table
if (!sizeof($aPlaceIDs)) {
2016-09-04 04:19:48 +03:00
// do we need to use transliteration and the regex for housenumbers???
//new query for lines, not housenumbers anymore
$sSQL = "SELECT distinct place_id FROM location_property_osmline";
$sSQL .= " WHERE startnumber is not NULL and parent_place_id in (".$sPlaceIDs.") and (";
if ($searchedHousenumber%2 == 0) {
2016-09-04 04:19:48 +03:00
//if housenumber is even, look for housenumber in streets with interpolationtype even or all
2016-10-13 00:30:24 +03:00
$sSQL .= "interpolationtype='even'";
} else {
2016-09-04 04:19:48 +03:00
//look for housenumber in streets with interpolationtype odd or all
2016-10-13 00:30:24 +03:00
$sSQL .= "interpolationtype='odd'";
2016-09-04 04:19:48 +03:00
}
2016-10-13 00:30:24 +03:00
$sSQL .= " or interpolationtype='all') and ";
$sSQL .= $searchedHousenumber.">=startnumber and ";
$sSQL .= $searchedHousenumber."<=endnumber";
2016-09-04 04:19:48 +03:00
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
//$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
//get place IDs
$aPlaceIDs = chksql($this->oDB->getCol($sSQL, 0));
}
2016-10-13 00:30:24 +03:00
2016-09-04 04:19:48 +03:00
// If nothing found try the aux fallback table
if (CONST_Use_Aux_Location_data && !sizeof($aPlaceIDs)) {
$sSQL = "SELECT place_id FROM location_property_aux ";
$sSQL .= " WHERE parent_place_id in (".$sPlaceIDs.") ";
$sSQL .= " AND housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND parent_place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
//$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
}
//if nothing was found in placex or location_property_aux, then search in Tiger data for this housenumber(location_property_tiger)
if (CONST_Use_US_Tiger_Data && !sizeof($aPlaceIDs)) {
$sSQL = "SELECT distinct place_id FROM location_property_tiger";
$sSQL .= " WHERE parent_place_id in (".$sPlaceIDs.") and (";
if ($searchedHousenumber%2 == 0) {
2016-10-13 00:30:24 +03:00
$sSQL .= "interpolationtype='even'";
} else {
2016-10-13 00:30:24 +03:00
$sSQL .= "interpolationtype='odd'";
2016-09-04 04:19:48 +03:00
}
2016-10-13 00:30:24 +03:00
$sSQL .= " or interpolationtype='all') and ";
$sSQL .= $searchedHousenumber.">=startnumber and ";
$sSQL .= $searchedHousenumber."<=endnumber";
2016-09-04 04:19:48 +03:00
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
//$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
//get place IDs
$aPlaceIDs = chksql($this->oDB->getCol($sSQL, 0));
}
// Fallback to the road (if no housenumber was found)
if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) {
2016-09-04 04:19:48 +03:00
$aPlaceIDs = $aRoadPlaceIDs;
//set to -1, if no housenumbers were found
$searchedHousenumber = -1;
}
//else: housenumber was found, remains saved in searchedHousenumber
}
if ($aSearch['sClass'] && sizeof($aPlaceIDs)) {
2016-09-04 04:19:48 +03:00
$sPlaceIDs = join(',', $aPlaceIDs);
$aClassPlaceIDs = array();
if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'name') {
2016-09-04 04:19:48 +03:00
// If they were searching for a named class (i.e. 'Kings Head pub') then we might have an extra match
$sSQL = "SELECT place_id ";
$sSQL .= " FROM placex ";
$sSQL .= " WHERE place_id in ($sPlaceIDs) ";
$sSQL .= " AND class='".$aSearch['sClass']."' ";
$sSQL .= " AND type='".$aSearch['sType']."'";
$sSQL .= " AND linked_place_id is null";
if ($sCountryCodesSQL) $sSQL .= " AND calculated_country_code in ($sCountryCodesSQL)";
$sSQL .= " ORDER BY rank_search ASC ";
$sSQL .= " LIMIT $this->iLimit";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aClassPlaceIDs = chksql($this->oDB->getCol($sSQL));
}
if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') { // & in
$sSQL = "SELECT count(*) FROM pg_tables ";
$sSQL .= "WHERE tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
2016-09-04 04:19:48 +03:00
$bCacheTable = chksql($this->oDB->getOne($sSQL));
$sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$this->iMaxRank = ((int)chksql($this->oDB->getOne($sSQL)));
// For state / country level searches the normal radius search doesn't work very well
$sPlaceGeom = false;
if ($this->iMaxRank < 9 && $bCacheTable) {
2016-09-04 04:19:48 +03:00
// Try and get a polygon to search in instead
$sSQL = "SELECT geometry ";
$sSQL .= " FROM placex";
$sSQL .= " WHERE place_id in ($sPlaceIDs)";
$sSQL .= " AND rank_search < $this->iMaxRank + 5";
$sSQL .= " AND ST_Geometrytype(geometry) in ('ST_Polygon','ST_MultiPolygon')";
$sSQL .= " ORDER BY rank_search ASC ";
$sSQL .= " LIMIT 1";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$sPlaceGeom = chksql($this->oDB->getOne($sSQL));
}
if ($sPlaceGeom) {
2016-09-04 04:19:48 +03:00
$sPlaceIDs = false;
} else {
2016-09-04 04:19:48 +03:00
$this->iMaxRank += 5;
$sSQL = "SELECT place_id FROM placex WHERE place_id in ($sPlaceIDs) and rank_search < $this->iMaxRank";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
$sPlaceIDs = join(',', $aPlaceIDs);
2016-09-04 04:19:48 +03:00
}
if ($sPlaceIDs || $sPlaceGeom) {
2016-09-04 04:19:48 +03:00
$fRange = 0.01;
if ($bCacheTable) {
2016-09-04 04:19:48 +03:00
// More efficient - can make the range bigger
$fRange = 0.05;
$sOrderBySQL = '';
2017-03-15 22:43:08 +03:00
if ($oNearPoint) {
$sOrderBySQL = $oNearPoint->distanceSQL('l.centroid');
} elseif ($sPlaceIDs) {
$sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
} elseif ($sPlaceGeom) {
$sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
}
2016-09-04 04:19:48 +03:00
$sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
if ($sPlaceIDs) {
2016-09-04 04:19:48 +03:00
$sSQL .= ",placex as f where ";
$sSQL .= "f.place_id in ($sPlaceIDs) and ST_DWithin(l.centroid, f.centroid, $fRange) ";
}
if ($sPlaceGeom) {
2016-09-04 04:19:48 +03:00
$sSQL .= " where ";
$sSQL .= "ST_Contains('".$sPlaceGeom."', l.centroid) ";
}
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " and l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
if ($this->iOffset) $sSQL .= " offset $this->iOffset";
$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
$aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($this->oDB->getCol($sSQL)));
} else {
2017-03-17 00:04:30 +03:00
if ($aSearch['oNear']) {
$fRange = $aSearch['oNear']->radius();
}
2016-09-04 04:19:48 +03:00
$sOrderBySQL = '';
2017-03-15 22:43:08 +03:00
if ($oNearPoint) {
$sOrderBySQL = $oNearPoint->distanceSQL('l.geometry');
} else {
$sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
}
2016-09-04 04:19:48 +03:00
$sSQL = "SELECT distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'');
$sSQL .= " FROM placex as l, placex as f ";
$sSQL .= " WHERE f.place_id in ($sPlaceIDs) ";
$sSQL .= " AND ST_DWithin(l.geometry, f.centroid, $fRange) ";
$sSQL .= " AND l.class='".$aSearch['sClass']."' ";
$sSQL .= " AND l.type='".$aSearch['sType']."' ";
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
2016-09-04 04:19:48 +03:00
}
if ($sCountryCodesSQL) $sSQL .= " AND l.calculated_country_code in ($sCountryCodesSQL)";
if ($sOrderBy) $sSQL .= "ORDER BY ".$OrderBysSQL." ASC";
if ($this->iOffset) $sSQL .= " OFFSET $this->iOffset";
2016-09-04 04:19:48 +03:00
$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
$aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($this->oDB->getCol($sSQL)));
}
}
}
$aPlaceIDs = $aClassPlaceIDs;
}
}
if (CONST_Debug) {
echo "<br><b>Place IDs:</b> ";
var_Dump($aPlaceIDs);
}
2016-09-04 04:19:48 +03:00
foreach ($aPlaceIDs as $iPlaceID) {
2016-09-04 04:19:48 +03:00
// array for placeID => -1 | Tiger housenumber
$aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
}
if ($iQueryLoop > 20) break;
}
if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
2016-09-04 04:19:48 +03:00
// Need to verify passes rank limits before dropping out of the loop (yuk!)
// reduces the number of place ids, like a filter
// rank_address is 30 for interpolated housenumbers
$sSQL = "SELECT place_id ";
$sSQL .= "FROM placex ";
$sSQL .= "WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
$sSQL .= " AND (";
$sSQL .= " placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
$sSQL .= " OR (extratags->'place') = 'city'";
}
if ($this->aAddressRankList) {
$sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
}
if (CONST_Use_US_Tiger_Data) {
$sSQL .= " ) ";
$sSQL .= "UNION ";
$sSQL .= " SELECT place_id ";
$sSQL .= " FROM location_property_tiger ";
$sSQL .= " WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
$sSQL .= " AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',', $this->aAddressRankList).")";
2016-09-04 04:19:48 +03:00
}
$sSQL .= ") UNION ";
$sSQL .= " SELECT place_id ";
$sSQL .= " FROM location_property_osmline ";
$sSQL .= " WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).")";
$sSQL .= " AND startnumber is not NULL AND (30 between $this->iMinAddressRank and $this->iMaxAddressRank)";
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump($sSQL);
$aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
$tempIDs = array();
foreach ($aFilteredPlaceIDs as $placeID) {
2016-09-04 04:19:48 +03:00
$tempIDs[$placeID] = $aResultPlaceIDs[$placeID]; //assign housenumber to placeID
}
$aResultPlaceIDs = $tempIDs;
}
//exit;
if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) break;
if ($iGroupLoop > 4) break;
if ($iQueryLoop > 30) break;
}
// Did we find anything?
if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs)) {
2016-09-04 04:19:48 +03:00
$aSearchResults = $this->getDetails($aResultPlaceIDs);
}
} else {
2016-09-04 04:19:48 +03:00
// Just interpret as a reverse geocode
2016-09-30 23:42:12 +03:00
$oReverse = new ReverseGeocode($this->oDB);
2016-09-04 04:19:48 +03:00
$oReverse->setZoom(18);
$aLookup = $oReverse->lookup(
2017-03-15 22:43:08 +03:00
$oNearPoint->lat(),
$oNearPoint->lon(),
false
);
2016-09-04 04:19:48 +03:00
if (CONST_Debug) var_dump("Reverse search", $aLookup);
if ($aLookup['place_id']) {
2016-09-04 04:19:48 +03:00
$aSearchResults = $this->getDetails(array($aLookup['place_id'] => -1));
$aResultPlaceIDs[$aLookup['place_id']] = -1;
} else {
2016-09-04 04:19:48 +03:00
$aSearchResults = array();
}
2016-09-04 04:19:48 +03:00
}
// No results? Done
if (!sizeof($aSearchResults)) {
if ($this->bFallback) {
if ($this->fallbackStructuredQuery()) {
2016-09-04 04:19:48 +03:00
return $this->lookup();
}
}
return array();
}
$aClassType = getClassTypesWithImportance();
$aRecheckWords = preg_split('/\b[\s,\\-]*/u', $sQuery);
foreach ($aRecheckWords as $i => $sWord) {
if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
2016-09-04 04:19:48 +03:00
}
if (CONST_Debug) {
echo '<i>Recheck words:<\i>';
var_dump($aRecheckWords);
}
2016-09-04 04:19:48 +03:00
$oPlaceLookup = new PlaceLookup($this->oDB);
$oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints);
$oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText);
$oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON);
$oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML);
$oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG);
$oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold);
foreach ($aSearchResults as $iResNum => $aResult) {
2016-09-04 04:19:48 +03:00
// Default
$fDiameter = getResultDiameter($aResult);
$aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
if ($aOutlineResult) {
2016-09-04 04:19:48 +03:00
$aResult = array_merge($aResult, $aOutlineResult);
}
if ($aResult['extra_place'] == 'city') {
2016-09-04 04:19:48 +03:00
$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']
) {
2016-09-04 04:19:48 +03:00
$aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png';
}
if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label'])
&& $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']
) {
2016-09-04 04:19:48 +03:00
$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']
) {
2016-09-04 04:19:48 +03:00
$aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
}
// if tag '&addressdetails=1' is set in query
if ($this->bIncludeAddressDetails) {
2016-09-04 04:19:48 +03:00
// 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'], $aResultPlaceIDs[$aResult['place_id']]);
if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city'])) {
$aResult['address'] = array_merge(array('city' => array_values($aResult['address'])[0]), $aResult['address']);
2016-09-04 04:19:48 +03:00
}
}
if ($this->bIncludeExtraTags) {
if ($aResult['extra']) {
2016-09-04 04:19:48 +03:00
$aResult['sExtraTags'] = json_decode($aResult['extra']);
} else {
2016-09-04 04:19:48 +03:00
$aResult['sExtraTags'] = (object) array();
}
}
if ($this->bIncludeNameDetails) {
if ($aResult['names']) {
2016-09-04 04:19:48 +03:00
$aResult['sNameDetails'] = json_decode($aResult['names']);
} else {
2016-09-04 04:19:48 +03:00
$aResult['sNameDetails'] = (object) array();
}
}
// Adjust importance for the number of exact string matches in the result
$aResult['importance'] = max(0.001, $aResult['importance']);
2016-09-04 04:19:48 +03:00
$iCountWords = 0;
$sAddress = $aResult['langaddress'];
foreach ($aRecheckWords as $i => $sWord) {
if (stripos($sAddress, $sWord)!==false) {
2016-09-04 04:19:48 +03:00
$iCountWords++;
if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
}
}
$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
$aResult['name'] = $aResult['langaddress'];
// secondary ordering (for results with same importance (the smaller the better):
// - approximate importance of address parts
2016-09-04 04:19:48 +03:00
$aResult['foundorder'] = -$aResult['addressimportance']/10;
// - number of exact matches from the query
if (isset($this->exactMatchCache[$aResult['place_id']])) {
2016-09-04 04:19:48 +03:00
$aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']];
} elseif (isset($this->exactMatchCache[$aResult['parent_place_id']])) {
2016-09-04 04:19:48 +03:00
$aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']];
}
// - importance of the class/type
2016-09-04 04:19:48 +03:00
if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance'])
&& $aClassType[$aResult['class'].':'.$aResult['type']]['importance']
) {
2016-09-04 04:19:48 +03:00
$aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance'];
} else {
2016-09-04 04:19:48 +03:00
$aResult['foundorder'] += 0.01;
}
if (CONST_Debug) var_dump($aResult);
2016-09-04 04:19:48 +03:00
$aSearchResults[$iResNum] = $aResult;
}
uasort($aSearchResults, 'byImportance');
$aOSMIDDone = array();
$aClassTypeNameDone = array();
$aToFilter = $aSearchResults;
$aSearchResults = array();
$bFirst = true;
foreach ($aToFilter as $iResNum => $aResult) {
2016-09-04 04:19:48 +03:00
$this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
if ($bFirst) {
2016-09-04 04:19:48 +03:00
$fLat = $aResult['lat'];
$fLon = $aResult['lon'];
if (isset($aResult['zoom'])) $iZoom = $aResult['zoom'];
$bFirst = false;
}
if (!$this->bDeDupe || (!isset($aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']])
&& !isset($aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']]))
) {
2016-09-04 04:19:48 +03:00
$aOSMIDDone[$aResult['osm_type'].$aResult['osm_id']] = true;
$aClassTypeNameDone[$aResult['osm_type'].$aResult['class'].$aResult['type'].$aResult['name'].$aResult['admin_level']] = true;
$aSearchResults[] = $aResult;
}
// Absolute limit on number of results
if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
}
return $aSearchResults;
} // end lookup()
} // end class