X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/ecee3828b38dbcf1712c1a1a3ccee1c3bc774f78..92061778f82345dce7138d79519d2ada5733dccd:/lib/Geocode.php?ds=inline
diff --git a/lib/Geocode.php b/lib/Geocode.php
index 3726f337..15915f94 100644
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -25,7 +25,7 @@ class Geocode
protected $aExcludePlaceIDs = array();
protected $bDeDupe = true;
- protected $bReverseInPlan = false;
+ protected $bReverseInPlan = true;
protected $iLimit = 20;
protected $iFinalLimit = 10;
@@ -51,10 +51,22 @@ class Geocode
protected $sQuery = false;
protected $aStructuredQuery = false;
+ protected $oNormalizer = null;
+
public function __construct(&$oDB)
{
$this->oDB =& $oDB;
+ $this->oNormalizer = \Transliterator::createFromRules(CONST_Term_Normalization_Rules);
+ }
+
+ private function normTerm($sTerm)
+ {
+ if ($this->oNormalizer === null) {
+ return $sTerm;
+ }
+
+ return $this->oNormalizer->transliterate($sTerm);
}
public function setReverseInPlan($bReverse)
@@ -67,19 +79,45 @@ class Geocode
$this->aLangPrefOrder = $aLangPref;
}
- public function getIncludeAddressDetails()
+ public function getMoreUrlParams()
{
- return $this->bIncludeAddressDetails;
- }
+ if ($this->aStructuredQuery) {
+ $aParams = $this->aStructuredQuery;
+ } else {
+ $aParams = array('q' => $this->sQuery);
+ }
- public function getIncludeExtraTags()
- {
- return $this->bIncludeExtraTags;
- }
+ if ($this->aExcludePlaceIDs) {
+ $aParams['exclude_place_ids'] = implode(',', $this->aExcludePlaceIDs);
+ }
- public function getIncludeNameDetails()
- {
- return $this->bIncludeNameDetails;
+ if ($this->bIncludeAddressDetails) $aParams['addressdetails'] = '1';
+ if ($this->bIncludeExtraTags) $aParams['extratags'] = '1';
+ if ($this->bIncludeNameDetails) $aParams['namedetails'] = '1';
+
+ if ($this->bIncludePolygonAsPoints) $aParams['polygon'] = '1';
+ if ($this->bIncludePolygonAsText) $aParams['polygon_text'] = '1';
+ if ($this->bIncludePolygonAsGeoJSON) $aParams['polygon_geojson'] = '1';
+ if ($this->bIncludePolygonAsKML) $aParams['polygon_kml'] = '1';
+ if ($this->bIncludePolygonAsSVG) $aParams['polygon_svg'] = '1';
+
+ if ($this->fPolygonSimplificationThreshold > 0.0) {
+ $aParams['polygon_threshold'] = $this->fPolygonSimplificationThreshold;
+ }
+
+ if ($this->bBoundedSearch) $aParams['bounded'] = '1';
+ if (!$this->bDeDupe) $aParams['dedupe'] = '0';
+
+ if ($this->aCountryCodes) {
+ $aParams['countrycodes'] = implode(',', $this->aCountryCodes);
+ }
+
+ if ($this->aViewBox) {
+ $aParams['viewbox'] = $this->aViewBox[0].','.$this->aViewBox[3]
+ .','.$this->aViewBox[2].','.$this->aViewBox[1];
+ }
+
+ return $aParams;
}
public function setIncludePolygonAsPoints($b = true)
@@ -121,23 +159,6 @@ class Geocode
$this->iLimit = $iLimit + min($iLimit, 10);
}
- public function getExcludedPlaceIDs()
- {
- return $this->aExcludePlaceIDs;
- }
-
-
- public function getCountryCodes()
- {
- return $this->aCountryCodes;
- }
-
- public function getViewBoxString()
- {
- if (!$this->aViewBox) return null;
- return $this->aViewBox[0].','.$this->aViewBox[3].','.$this->aViewBox[2].','.$this->aViewBox[1];
- }
-
public function setFeatureType($sFeatureType)
{
switch ($sFeatureType) {
@@ -283,7 +304,7 @@ class Geocode
$aViewbox = $oParams->getStringList('viewboxlbrt');
if ($aViewbox) {
if (count($aViewbox) != 4) {
- userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
+ userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
}
$this->setViewbox($aViewbox);
} else {
@@ -292,12 +313,7 @@ class Geocode
if (count($aViewbox) != 4) {
userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
}
- $this->setViewBox(array(
- $aViewbox[0],
- $aViewbox[3],
- $aViewbox[2],
- $aViewbox[1]
- ));
+ $this->setViewBox($aViewbox);
} else {
$aRoute = $oParams->getStringList('route');
$fRouteWidth = $oParams->getFloat('routewidth');
@@ -341,7 +357,7 @@ class Geocode
return true;
}
- public function setStructuredQuery($sAmentiy = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
+ public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
{
$this->sQuery = false;
@@ -351,9 +367,9 @@ class Geocode
$this->aAddressRankList = array();
$this->aStructuredQuery = array();
- $this->sAllowedTypesSQLList = '';
+ $this->sAllowedTypesSQLList = False;
- $this->loadStructuredAddressElement($sAmentiy, 'amenity', 26, 30, false);
+ $this->loadStructuredAddressElement($sAmenity, '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);
@@ -364,7 +380,7 @@ class Geocode
if (sizeof($this->aStructuredQuery) > 0) {
$this->sQuery = join(', ', $this->aStructuredQuery);
if ($this->iMaxAddressRank < 30) {
- $sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
}
}
}
@@ -401,8 +417,15 @@ class Geocode
$sPlaceIDs = join(',', array_keys($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 * ";
+ $sImportanceSQLGeom = '';
+ if ($this->sViewboxSmallSQL) {
+ $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+ $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, geometry) 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 * ";
+ $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, geometry) THEN 1 ELSE 0.75 END * ";
+ }
$sSQL = "SELECT ";
$sSQL .= " osm_type,";
@@ -414,7 +437,7 @@ class Geocode
$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 .= " 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,";
@@ -457,7 +480,7 @@ class Geocode
$sSQL .= " admin_level, ";
$sSQL .= " rank_search, ";
$sSQL .= " rank_address, ";
- $sSQL .= " calculated_country_code, ";
+ $sSQL .= " country_code, ";
$sSQL .= " importance, ";
if (!$this->bDeDupe) $sSQL .= "place_id,";
$sSQL .= " langaddress, ";
@@ -467,6 +490,35 @@ class Geocode
if ($this->bIncludeNameDetails) $sSQL .= "name, ";
$sSQL .= " extratags->'place' ";
+ // postcode table
+ $sSQL .= "UNION ";
+ $sSQL .= "SELECT";
+ $sSQL .= " 'P' as osm_type,";
+ $sSQL .= " (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
+ $sSQL .= " 'place' as class, 'postcode' as type,";
+ $sSQL .= " null as admin_level, rank_search, rank_address,";
+ $sSQL .= " place_id, parent_place_id, country_code,";
+ $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
+ $sSQL .= " postcode as placename,";
+ $sSQL .= " postcode as ref,";
+ if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
+ if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
+ $sSQL .= " ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
+ $sSQL .= $sImportanceSQLGeom."(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 = lp.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_postcode lp";
+ $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
+
if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
// only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
// with start- and endnumber, the common osm housenumbers are usually saved as points
@@ -542,7 +594,7 @@ class Geocode
$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 .= " country_code, ";
$sSQL .= " get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) AS langaddress, ";
$sSQL .= " null AS placename, ";
$sSQL .= " null AS ref, ";
@@ -567,7 +619,7 @@ class Geocode
$sSQL .= " SELECT ";
$sSQL .= " osm_id, ";
$sSQL .= " place_id, ";
- $sSQL .= " calculated_country_code, ";
+ $sSQL .= " 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) ";
@@ -586,7 +638,7 @@ class Geocode
$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
+ $sSQL .= " country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
if (!$this->bDeDupe) $sSQL .= ", place_id ";
if (CONST_Use_Aux_Location_data) {
@@ -644,7 +696,7 @@ class Geocode
return $aSearchResults;
}
- public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases)
+ public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery)
{
/*
Calculate all searches using aValidTokens i.e.
@@ -657,12 +709,14 @@ class Geocode
Score how good the search is so they can be ordered
*/
- foreach ($aPhrases as $iPhrase => $sPhrase) {
+ $iGlobalRank = 0;
+
+ foreach ($aPhrases as $iPhrase => $aPhrase) {
$aNewPhraseSearches = array();
if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
else $sPhraseType = '';
- foreach ($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset) {
+ foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) {
// Too many permutations - too expensive
if ($iWordSet > 120) break;
@@ -680,6 +734,8 @@ class Geocode
// If the token is valid
if (isset($aValidTokens[' '.$sToken])) {
+ // TODO variable should go into aCurrentSearch
+ $bHavePostcode = false;
foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank']++;
@@ -691,49 +747,42 @@ class Geocode
$aSearch['iSearchRank'] += 5;
}
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ // If it is at the beginning, we can be almost sure that this is the wrong order
+ // Increase score for all searches.
+ if ($iToken == 0 && $iPhrase == 0) {
+ $iGlobalRank++;
+ }
}
- } elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null) {
- if ($aSearch['oNear'] === false) {
- $aSearch['oNear'] = new NearPoint(
- $aSearchTerm['lat'],
- $aSearchTerm['lon'],
- $aSearchTerm['radius']
- );
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- } elseif ($sPhraseType == 'postalcode') {
+ } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') {
// We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
- if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
- // If we already have a name try putting the postcode first
- if (sizeof($aSearch['aName'])) {
+ if ($aSearch['sPostcode'] === '' &&
+ isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) {
+ // If we have structured search or this is the first term,
+ // make the postcode the primary search element.
+ if (!$bHavePostcode && $aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
$aNewSearch = $aSearch;
+ $aNewSearch['sOperator'] = 'postcode';
$aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
- $aNewSearch['aName'] = array();
- $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ $aNewSearch['aName'] = array($aSearchTerm['word_id'] => $aSearchTerm['word']);
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;
+ $bHavePostcode = true;
}
- if (sizeof($aSearch['aName'])) {
- if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- } else {
- $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1000; // skip;
- }
- } else {
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- //$aSearch['iNamePhrase'] = $iPhrase;
+ // If we have a structured search or this is not the first term,
+ // add the postcode as an addendum.
+ if ($aSearch['sOperator'] !== 'postcode' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']))) {
+ $aSearch['sPostcode'] = $aSearchTerm['word'];
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
} elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') {
- if ($aSearch['sHouseNumber'] === '') {
+ if ($aSearch['sHouseNumber'] === '' && $aSearch['sOperator'] !== 'postcode') {
$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) $aSearch['iSearchRank'] += 1;
+ // also must not appear in the middle of the address
+ if ($aSearch['aAddress'] || $aSearch['aAddressNonSearch']) $aSearch['iSearchRank'] += 1;
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
/*
// Fall back to not searching for this item (better than nothing)
@@ -743,13 +792,19 @@ class Geocode
*/
}
} elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) {
- if ($aSearch['sClass'] === '') {
- $aSearch['sOperator'] = $aSearchTerm['operator'];
+ // require a normalized exact match of the term
+ // if we have the normalizer version of the query
+ // available
+ if ($aSearch['sOperator'] === ''
+ && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) {
$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) $aSearch['iSearchRank'] += 1;
+ if ($aSearchTerm['operator'] == '') {
+ $aSearch['sOperator'] = sizeof($aSearch['aName']) ? 'name' : 'near';
+ $aSearch['iSearchRank'] += 2;
+ } else {
+ $aSearch['sOperator'] = 'near'; // near = in for the moment
+ }
if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
@@ -804,7 +859,8 @@ class Geocode
}
}
- if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) {
+ if ((!$aCurrentSearch['sPostcode'] && !$aCurrentSearch['aAddress'] && !$aCurrentSearch['aAddressNonSearch'])
+ && (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)) {
$aSearch = $aCurrentSearch;
$aSearch['iSearchRank'] += 1;
if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;
@@ -865,6 +921,20 @@ class Geocode
//if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
}
+
+ // Revisit searches, giving penalty to unlikely combinations
+ $aGroupedSearches = array();
+ foreach ($aSearches as $aSearch) {
+ if (!$aSearch['aName']) {
+ if ($aSearch['sHouseNumber']) {
+ continue;
+ }
+ }
+ $aSearch['iSearchRank'] += $iGlobalRank;
+ $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+ }
+ ksort($aGroupedSearches);
+
return $aGroupedSearches;
}
@@ -902,8 +972,9 @@ class Geocode
public function lookup()
{
- if (!$this->sQuery && !$this->aStructuredQuery) return false;
+ if (!$this->sQuery && !$this->aStructuredQuery) return array();
+ $sNormQuery = $this->normTerm($this->sQuery);
$sLanguagePrefArraySQL = "ARRAY[".join(',', array_map("getDBQuoted", $this->aLangPrefOrder))."]";
$sCountryCodesSQL = false;
if ($this->aCountryCodes) {
@@ -963,21 +1034,31 @@ class Geocode
'sClass' => '',
'sType' => '',
'sHouseNumber' => '',
+ 'sPostcode' => '',
'oNear' => $oNearPoint
)
);
// Any 'special' terms in the search?
$bSpecialTerms = false;
- preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
- $aSpecialTerms = array();
+ preg_match_all('/\\[([\\w_]*)=([\\w_]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
foreach ($aSpecialTermsRaw as $aSpecialTerm) {
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
- $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2];
+ if (!$bSpecialTerms) {
+ $aNewSearches = array();
+ foreach ($aSearches as $aSearch) {
+ $aNewSearch = $aSearch;
+ $aNewSearch['sClass'] = $aSpecialTerm[1];
+ $aNewSearch['sType'] = $aSpecialTerm[2];
+ $aNewSearches[] = $aNewSearch;
+ }
+
+ $aSearches = $aNewSearches;
+ $bSpecialTerms = true;
+ }
}
preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
- $aSpecialTerms = array();
if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) {
$aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
unset($this->aStructuredQuery['amenity']);
@@ -985,32 +1066,28 @@ class Geocode
foreach ($aSpecialTermsRaw as $aSpecialTerm) {
$sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
- $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".$aSpecialTerm[1]."') AS string"));
+ if ($bSpecialTerms) {
+ continue;
+ }
+
+ $sToken = chksql($this->oDB->getOne("SELECT make_standard_name('".pg_escape_string($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';
+ $sSQL .= ' WHERE (class is not null AND class not in (\'place\'))';
if (CONST_Debug) var_Dump($sSQL);
$aSearchWords = chksql($this->oDB->getAll($sSQL));
$aNewSearches = array();
foreach ($aSearches as $aSearch) {
foreach ($aSearchWords as $aSearchTerm) {
$aNewSearch = $aSearch;
- if ($aSearchTerm['country_code']) {
- $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
- $aNewSearches[] = $aNewSearch;
- $bSpecialTerms = true;
- }
- if ($aSearchTerm['class']) {
- $aNewSearch['sClass'] = $aSearchTerm['class'];
- $aNewSearch['sType'] = $aSearchTerm['type'];
- $aNewSearches[] = $aNewSearch;
- $bSpecialTerms = true;
- }
+ $aNewSearch['sClass'] = $aSearchTerm['class'];
+ $aNewSearch['sType'] = $aSearchTerm['type'];
+ $aNewSearches[] = $aNewSearch;
+ $bSpecialTerms = true;
}
}
$aSearches = $aNewSearches;
@@ -1059,14 +1136,10 @@ class Geocode
if (CONST_Debug) var_Dump($sSQL);
$aValidTokens = array();
- if (sizeof($aTokens)) {
- $aDatabaseWords = chksql(
- $this->oDB->getAll($sSQL),
- "Could not get word tokens."
- );
- } else {
- $aDatabaseWords = array();
- }
+ $aDatabaseWords = chksql(
+ $this->oDB->getAll($sSQL),
+ "Could not get word tokens."
+ );
$aPossibleMainWordIDs = array();
$aWordFrequencyScores = array();
foreach ($aDatabaseWords as $aToken) {
@@ -1087,21 +1160,9 @@ class Geocode
}
if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
- // Try and calculate GB postcodes we might be missing
+ // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
foreach ($aTokens as $sToken) {
- // Source of gb postcodes is now definitive - always use
- if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', 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);
- }
- $aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $this->oDB);
- if ($aGBPostcodeLocation) {
- $aValidTokens[$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[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
if (isset($aValidTokens[$aData[1]])) {
foreach ($aValidTokens[$aData[1]] as $aToken) {
if (!$aToken['class']) {
@@ -1130,7 +1191,7 @@ class Geocode
// array with: placeid => -1 | tiger-housenumber
$aResultPlaceIDs = array();
- $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
+ $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery);
if ($this->bReverseInPlan) {
// Reverse phrase array and also reverse the order of the wordsets in
@@ -1142,7 +1203,7 @@ class Geocode
$aFinalPhrase = end($aPhrases);
$aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
}
- $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
+ $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false, $sNormQuery);
foreach ($aGroupedSearches as $aSearches) {
foreach ($aSearches as $aSearch) {
@@ -1168,7 +1229,6 @@ class Geocode
ksort($aGroupedSearches);
}
- if (CONST_Debug) var_Dump($aGroupedSearches);
if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) {
$aCopyGroupedSearches = $aGroupedSearches;
foreach ($aCopyGroupedSearches as $iGroup => $aSearches) {
@@ -1225,13 +1285,16 @@ class Geocode
if (CONST_Debug) echo "
Search Loop, group $iGroupLoop, loop $iQueryLoop";
if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
+ if ($sCountryCodesSQL && $aSearch['sCountryCode'] && !in_array($aSearch['sCountryCode'], $this->aCountryCodes)) {
+ continue;
+ }
+
// No location term?
- if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['oNear']) {
- if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber']) {
+ if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress'])) {
+ if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber'] && !$aSearch['oNear']) {
// 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)";
+ $sSQL = "SELECT place_id FROM placex WHERE country_code='".$aSearch['sCountryCode']."' AND rank_search = 4";
if ($bBoundingBoxSearch)
$sSQL .= " AND _st_intersects($this->sViewboxSmallSQL, geometry)";
$sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1";
@@ -1248,39 +1311,32 @@ class Geocode
if (chksql($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 ($aSearch['oNear']) {
+ $sSQL .= " WHERE ".$aSearch['oNear']->withinSQL('ct.centroid');
+ } else {
+ $sSQL .= " WHERE st_contains($this->sViewboxSmallSQL, ct.centroid)";
+ }
+ if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND place_id not in (".join(',', $this->aExcludePlaceIDs).")";
}
- if ($this->sViewboxCentreSQL) $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
+ if ($this->sViewboxCentreSQL) {
+ $sSQL .= " ORDER BY ST_Distance($this->sViewboxCentreSQL, ct.centroid) ASC";
+ } elseif ($aSearch['oNear']) {
+ $sSQL .= " ORDER BY ".$aSearch['oNear']->distanceSQL('ct.centroid').' ASC';
+ }
$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']."_".$aSearch['sType']." ct";
- if ($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 $this->iLimit";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
- }
- } else {
+ } else if ($aSearch['oNear']) {
$sSQL = "SELECT place_id ";
$sSQL .= "FROM placex ";
$sSQL .= "WHERE class='".$aSearch['sClass']."' ";
$sSQL .= " AND type='".$aSearch['sType']."'";
- $sSQL .= " AND ST_Contains($this->sViewboxSmallSQL, geometry) ";
+ $sSQL .= " AND ".$aSearch['oNear']->withinSQL('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";
+ if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
+ $sSQL .= " ORDER BY ".$aSearch['oNear']->distanceSQL('centroid')." ASC";
$sSQL .= " LIMIT $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = chksql($this->oDB->getCol($sSQL));
@@ -1290,6 +1346,24 @@ class Geocode
// If a coordinate is given, the search must either
// be for a name or a special search. Ignore everythin else.
$aPlaceIDs = array();
+ } elseif ($aSearch['sOperator'] == 'postcode') {
+ $sSQL = "SELECT p.place_id FROM location_postcode p ";
+ if (sizeof($aSearch['aAddress'])) {
+ $sSQL .= ", search_name s ";
+ $sSQL .= "WHERE s.place_id = p.parent_place_id ";
+ $sSQL .= "AND array_cat(s.nameaddress_vector, s.name_vector) @> ARRAY[".join($aSearch['aAddress'], ",")."] AND ";
+ } else {
+ $sSQL .= " WHERE ";
+ }
+ $sSQL .= "p.postcode = '".pg_escape_string(reset($aSearch['aName']))."'";
+ if ($aSearch['sCountryCode']) {
+ $sSQL .= " AND p.country_code = '".$aSearch['sCountryCode']."'";
+ } elseif ($sCountryCodesSQL) {
+ $sSQL .= " AND p.country_code in ($sCountryCodesSQL)";
+ }
+ $sSQL .= " LIMIT $this->iLimit";
+ if (CONST_Debug) var_dump($sSQL);
+ $aPlaceIDs = chksql($this->oDB->getCol($sSQL));
} else {
$aPlaceIDs = array();
@@ -1343,7 +1417,7 @@ class Geocode
if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
if ($aSearch['sHouseNumber']) {
$aTerms[] = "address_rank between 16 and 27";
- } else {
+ } elseif (!$aSearch['sClass'] || $aSearch['sOperator'] == 'name') {
if ($this->iMinAddressRank > 0) {
$aTerms[] = "address_rank >= ".$this->iMinAddressRank;
}
@@ -1355,6 +1429,12 @@ class Geocode
$aTerms[] = $aSearch['oNear']->withinSQL('centroid');
$aOrder[] = $aSearch['oNear']->distanceSQL('centroid');
+ } elseif ($aSearch['sPostcode']) {
+ if (!sizeof($aSearch['aAddress'])) {
+ $aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$aSearch['sPostcode']."' AND ST_DWithin(search_name.centroid, p.geometry, 0.1))";
+ } else {
+ $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$aSearch['sPostcode']."')";
+ }
}
if (sizeof($this->aExcludePlaceIDs)) {
$aTerms[] = "place_id not in (".join(',', $this->aExcludePlaceIDs).")";
@@ -1504,7 +1584,8 @@ class Geocode
}
// Fallback to the road (if no housenumber was found)
- if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])) {
+ if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber'])
+ && ($aSearch['aAddress'] || $aSearch['sCountryCode'])) {
$aPlaceIDs = $aRoadPlaceIDs;
//set to -1, if no housenumbers were found
$searchedHousenumber = -1;
@@ -1525,7 +1606,7 @@ class Geocode
$sSQL .= " AND class='".$aSearch['sClass']."' ";
$sSQL .= " AND type='".$aSearch['sType']."'";
$sSQL .= " AND linked_place_id is null";
- if ($sCountryCodesSQL) $sSQL .= " AND calculated_country_code in ($sCountryCodesSQL)";
+ if ($sCountryCodesSQL) $sSQL .= " AND country_code in ($sCountryCodesSQL)";
$sSQL .= " ORDER BY rank_search ASC ";
$sSQL .= " LIMIT $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
@@ -1533,8 +1614,9 @@ class Geocode
}
if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'near') { // & in
+ $sClassTable = 'place_classtype_'.$aSearch['sClass'].'_'.$aSearch['sType'];
$sSQL = "SELECT count(*) FROM pg_tables ";
- $sSQL .= "WHERE tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
+ $sSQL .= "WHERE tablename = '$sClassTable'";
$bCacheTable = chksql($this->oDB->getOne($sSQL));
$sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
@@ -1579,10 +1661,11 @@ class Geocode
} elseif ($sPlaceIDs) {
$sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
} elseif ($sPlaceGeom) {
- $sOrderBysSQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
+ $sOrderBySQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
}
- $sSQL = "select distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'')." from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." as l";
+ $sSQL = "select distinct i.place_id".($sOrderBySQL?', i.order_term':'')." from (";
+ $sSQL .= "select l.place_id".($sOrderBySQL?','.$sOrderBySQL.' as order_term':'')." from ".$sClassTable." as l";
if ($sCountryCodesSQL) $sSQL .= " join placex as lp using (place_id)";
if ($sPlaceIDs) {
$sSQL .= ",placex as f where ";
@@ -1595,8 +1678,9 @@ class Geocode
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " and l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)";
- if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc";
+ if ($sCountryCodesSQL) $sSQL .= " and lp.country_code in ($sCountryCodesSQL)";
+ $sSQL .= 'limit 300) i ';
+ if ($sOrderBySQL) $sSQL .= "order by order_term asc";
if ($this->iOffset) $sSQL .= " offset $this->iOffset";
$sSQL .= " limit $this->iLimit";
if (CONST_Debug) var_dump($sSQL);
@@ -1613,7 +1697,7 @@ class Geocode
$sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
}
- $sSQL = "SELECT distinct l.place_id".($sOrderBysSQL?','.$sOrderBysSQL:'');
+ $sSQL = "SELECT distinct l.place_id".($sOrderBySQL?','.$sOrderBySQL:'');
$sSQL .= " FROM placex as l, placex as f ";
$sSQL .= " WHERE f.place_id in ($sPlaceIDs) ";
$sSQL .= " AND ST_DWithin(l.geometry, f.centroid, $fRange) ";
@@ -1622,8 +1706,8 @@ class Geocode
if (sizeof($this->aExcludePlaceIDs)) {
$sSQL .= " AND l.place_id not in (".join(',', $this->aExcludePlaceIDs).")";
}
- if ($sCountryCodesSQL) $sSQL .= " AND l.calculated_country_code in ($sCountryCodesSQL)";
- if ($sOrderBy) $sSQL .= "ORDER BY ".$OrderBysSQL." ASC";
+ if ($sCountryCodesSQL) $sSQL .= " AND l.country_code in ($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);
@@ -1640,6 +1724,21 @@ class Geocode
var_Dump($aPlaceIDs);
}
+ if (sizeof($aPlaceIDs) && $aSearch['sPostcode']) {
+ $sSQL = 'SELECT place_id FROM placex';
+ $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')';
+ $sSQL .= " AND postcode = '".pg_escape_string($aSearch['sPostcode'])."'";
+ if (CONST_Debug) var_dump($sSQL);
+ $aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
+ if ($aFilteredPlaceIDs) {
+ $aPlaceIDs = $aFilteredPlaceIDs;
+ if (CONST_Debug) {
+ echo "
Place IDs after postcode filtering: ";
+ var_Dump($aPlaceIDs);
+ }
+ }
+ }
+
foreach ($aPlaceIDs as $iPlaceID) {
// array for placeID => -1 | Tiger housenumber
$aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
@@ -1651,9 +1750,11 @@ class Geocode
// 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
+ $sWherePlaceId = 'WHERE place_id in (';
+ $sWherePlaceId .= join(',', array_keys($aResultPlaceIDs)).') ';
+
$sSQL = "SELECT place_id ";
- $sSQL .= "FROM placex ";
- $sSQL .= "WHERE place_id in (".join(',', array_keys($aResultPlaceIDs)).") ";
+ $sSQL .= "FROM placex ".$sWherePlaceId;
$sSQL .= " AND (";
$sSQL .= " placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
@@ -1662,20 +1763,23 @@ class Geocode
if ($this->aAddressRankList) {
$sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
}
- if (CONST_Use_US_Tiger_Data) {
- $sSQL .= " ) ";
+ $sSQL .= " ) UNION ";
+ $sSQL .= " SELECT place_id FROM location_postcode lp ".$sWherePlaceId;
+ $sSQL .= " AND (lp.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
+ if ($this->aAddressRankList) {
+ $sSQL .= " OR lp.rank_address in (".join(',', $this->aAddressRankList).")";
+ }
+ $sSQL .= ") ";
+ if (CONST_Use_US_Tiger_Data && $this->iMaxAddressRank == 30) {
+ $sSQL .= "UNION ";
+ $sSQL .= " SELECT place_id ";
+ $sSQL .= " FROM location_property_tiger ".$sWherePlaceId;
+ }
+ if ($this->iMaxAddressRank == 30) {
$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).")";
+ $sSQL .= " FROM location_property_osmline ".$sWherePlaceId;
}
- $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 $this->iMaxAddressRank)";
if (CONST_Debug) var_dump($sSQL);
$aFilteredPlaceIDs = chksql($this->oDB->getCol($sSQL));
$tempIDs = array();