X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/cdf8c678988cdad7d10cc52b5ece22196c2491f9..7665e5a03530bcc37ff25e0e838f8aa1b7fb69b5:/lib/SearchDescription.php
diff --git a/lib/SearchDescription.php b/lib/SearchDescription.php
index eba5f6a9..bf32df96 100644
--- a/lib/SearchDescription.php
+++ b/lib/SearchDescription.php
@@ -4,6 +4,7 @@ namespace Nominatim;
require_once(CONST_BasePath.'/lib/SpecialSearchOperator.php');
require_once(CONST_BasePath.'/lib/SearchContext.php');
+require_once(CONST_BasePath.'/lib/Result.php');
/**
* Description of a single interpretation of a search query.
@@ -16,6 +17,8 @@ class SearchDescription
private $sCountryCode = '';
/// List of word ids making up the name of the object.
private $aName = array();
+ /// True if the name is rare enough to force index use on name.
+ private $bRareName = false;
/// List of word ids making up the address of the object.
private $aAddress = array();
/// Subset of word ids of full words making up the address.
@@ -42,7 +45,6 @@ class SearchDescription
/// Index of phrase currently processed.
private $iNamePhrase = -1;
-
/**
* Create an empty search description.
*
@@ -57,7 +59,7 @@ class SearchDescription
/**
* Get current search rank.
*
- * The higher the search rank the lower the likelyhood that the
+ * The higher the search rank the lower the likelihood that the
* search is a correct interpretation of the search query.
*
* @return integer Search rank.
@@ -67,19 +69,6 @@ class SearchDescription
return $this->iSearchRank;
}
- /**
- * Increase the search rank.
- *
- * @param integer $iAddRank Number of ranks to increase.
- *
- * @return void
- */
- public function addToRank($iAddRank)
- {
- $this->iSearchRank += $iAddRank;
- return $this->iSearchRank;
- }
-
/**
* Make this search a POI search.
*
@@ -106,8 +95,8 @@ class SearchDescription
*/
public function looksLikeFullAddress()
{
- return sizeof($this->aName)
- && (sizeof($this->aAddress || $this->sCountryCode))
+ return (!empty($this->aName))
+ && (!empty($this->aAddress) || $this->sCountryCode)
&& preg_match('/[0-9]+/', $this->sHouseNumber);
}
@@ -159,7 +148,7 @@ class SearchDescription
*/
public function isValidSearch()
{
- if (!sizeof($this->aName)) {
+ if (empty($this->aName)) {
if ($this->sHouseNumber) {
return false;
}
@@ -177,55 +166,43 @@ class SearchDescription
/**
* Derive new searches by adding a full term to the existing search.
*
- * @param mixed[] $aSearchTerm Description of the token.
- * @param bool $bHasPartial True if there are also tokens of partial terms
- * with the same name.
- * @param string $sPhraseType Type of phrase the token is contained in.
- * @param bool $bFirstToken True if the token is at the beginning of the
- * query.
- * @param bool $bFirstPhrase True if the token is in the first phrase of
- * the query.
- * @param bool $bLastToken True if the token is at the end of the query.
- * @param integer $iGlobalRank Changable ranking of all searches in the
- * batch.
+ * @param object $oSearchTerm Description of the token.
+ * @param bool $bHasPartial True if there are also tokens of partial terms
+ * with the same name.
+ * @param string $sPhraseType Type of phrase the token is contained in.
+ * @param bool $bFirstToken True if the token is at the beginning of the
+ * query.
+ * @param bool $bFirstPhrase True if the token is in the first phrase of
+ * the query.
+ * @param bool $bLastToken True if the token is at the end of the query.
*
* @return SearchDescription[] List of derived search descriptions.
*/
- public function extendWithFullTerm($aSearchTerm, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken, &$iGlobalRank)
+ public function extendWithFullTerm($oSearchTerm, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken)
{
$aNewSearches = array();
if (($sPhraseType == '' || $sPhraseType == 'country')
- && !empty($aSearchTerm['country_code'])
- && $aSearchTerm['country_code'] != '0'
+ && is_a($oSearchTerm, '\Nominatim\Token\Country')
) {
if (!$this->sCountryCode) {
$oSearch = clone $this;
$oSearch->iSearchRank++;
- $oSearch->sCountryCode = $aSearchTerm['country_code'];
+ $oSearch->sCountryCode = $oSearchTerm->sCountryCode;
// Country is almost always at the end of the string
// - increase score for finding it anywhere else (optimisation)
if (!$bLastToken) {
$oSearch->iSearchRank += 5;
}
$aNewSearches[] = $oSearch;
-
- // If it is at the beginning, we can be almost sure that
- // the terms are in the wrong order. Increase score for all searches.
- if ($bFirstToken) {
- $iGlobalRank++;
- }
}
} elseif (($sPhraseType == '' || $sPhraseType == 'postalcode')
- && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode'
+ && is_a($oSearchTerm, '\Nominatim\Token\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 (!$this->sPostcode
- && $aSearchTerm['word']
- && pg_escape_string($aSearchTerm['word']) == $aSearchTerm['word']
- ) {
+ if (!$this->sPostcode) {
// If we have structured search or this is the first term,
// make the postcode the primary search element.
if ($this->iOperator == Operator::NONE
@@ -236,70 +213,81 @@ class SearchDescription
$oSearch->iOperator = Operator::POSTCODE;
$oSearch->aAddress = array_merge($this->aAddress, $this->aName);
$oSearch->aName =
- array($aSearchTerm['word_id'] => $aSearchTerm['word']);
+ array($oSearchTerm->iId => $oSearchTerm->sPostcode);
$aNewSearches[] = $oSearch;
}
// If we have a structured search or this is not the first term,
// add the postcode as an addendum.
if ($this->iOperator != Operator::POSTCODE
- && ($sPhraseType == 'postalcode' || sizeof($this->aName))
+ && ($sPhraseType == 'postalcode' || !empty($this->aName))
) {
$oSearch = clone $this;
$oSearch->iSearchRank++;
- $oSearch->sPostcode = $aSearchTerm['word'];
+ $oSearch->sPostcode = $oSearchTerm->sPostcode;
$aNewSearches[] = $oSearch;
}
}
} elseif (($sPhraseType == '' || $sPhraseType == 'street')
- && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house'
+ && is_a($oSearchTerm, '\Nominatim\Token\HouseNumber')
) {
if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
$oSearch = clone $this;
$oSearch->iSearchRank++;
- $oSearch->sHouseNumber = trim($aSearchTerm['word_token']);
+ $oSearch->sHouseNumber = $oSearchTerm->sToken;
// sanity check: if the housenumber is not mainly made
// up of numbers, add a penalty
- if (preg_match_all("/[^0-9]/", $oSearch->sHouseNumber, $aMatches) > 2) {
+ if (preg_match_all('/[^0-9]/', $oSearch->sHouseNumber, $aMatches) > 2) {
$oSearch->iSearchRank++;
}
- if (!isset($aSearchTerm['word_id'])) {
+ if (empty($oSearchTerm->iId)) {
$oSearch->iSearchRank++;
}
// also must not appear in the middle of the address
- if (sizeof($this->aAddress) || sizeof($this->aAddressNonSearch)) {
+ if (!empty($this->aAddress)
+ || (!empty($this->aAddressNonSearch))
+ || $this->sPostcode
+ ) {
$oSearch->iSearchRank++;
}
$aNewSearches[] = $oSearch;
}
- } elseif ($sPhraseType == '' && $aSearchTerm['class']) {
+ } elseif ($sPhraseType == ''
+ && is_a($oSearchTerm, '\Nominatim\Token\SpecialTerm')
+ ) {
if ($this->iOperator == Operator::NONE) {
$oSearch = clone $this;
$oSearch->iSearchRank++;
- $iOp = Operator::NEAR; // near == in for the moment
- if ($aSearchTerm['operator'] == '') {
- if (sizeof($this->aName)) {
+ $iOp = $oSearchTerm->iOperator;
+ if ($iOp == Operator::NONE) {
+ if (!empty($this->aName) || $this->oContext->isBoundedSearch()) {
$iOp = Operator::NAME;
+ } else {
+ $iOp = Operator::NEAR;
}
$oSearch->iSearchRank += 2;
}
- $oSearch->setPoiSearch($iOp, $aSearchTerm['class'], $aSearchTerm['type']);
+ $oSearch->setPoiSearch(
+ $iOp,
+ $oSearchTerm->sClass,
+ $oSearchTerm->sType
+ );
$aNewSearches[] = $oSearch;
}
- } elseif (isset($aSearchTerm['word_id'])
- && $aSearchTerm['word_id']
- && $sPhraseType != 'country'
+ } elseif ($sPhraseType != 'country'
+ && is_a($oSearchTerm, '\Nominatim\Token\Word')
) {
- $iWordID = $aSearchTerm['word_id'];
- if (sizeof($this->aName)) {
- if (($sPhraseType == '' || !$bFirstPhrase)
- && $sPhraseType != 'country'
- && !$bHasPartial
- ) {
+ $iWordID = $oSearchTerm->iId;
+ // Full words can only be a name if they appear at the beginning
+ // of the phrase. In structured search the name must forcably in
+ // the first phrase. In unstructured search it may be in a later
+ // phrase when the first phrase is a house number.
+ if (!empty($this->aName) || !($bFirstPhrase || $sPhraseType == '')) {
+ if (($sPhraseType == '' || !$bFirstPhrase) && !$bHasPartial) {
$oSearch = clone $this;
- $oSearch->iSearchRank++;
+ $oSearch->iSearchRank += 2;
$oSearch->aAddress[$iWordID] = $iWordID;
$aNewSearches[] = $oSearch;
} else {
@@ -309,6 +297,11 @@ class SearchDescription
$oSearch = clone $this;
$oSearch->iSearchRank++;
$oSearch->aName = array($iWordID => $iWordID);
+ if (CONST_Search_NameOnlySearchFrequencyThreshold) {
+ $oSearch->bRareName =
+ $oSearchTerm->iSearchNameCount
+ < CONST_Search_NameOnlySearchFrequencyThreshold;
+ }
$aNewSearches[] = $oSearch;
}
}
@@ -319,7 +312,8 @@ class SearchDescription
/**
* Derive new searches by adding a partial term to the existing search.
*
- * @param mixed[] $aSearchTerm Description of the token.
+ * @param string $sToken Term for the token.
+ * @param object $oSearchTerm Description of the token.
* @param bool $bStructuredPhrases True if the search is structured.
* @param integer $iPhrase Number of the phrase the token is in.
* @param array[] $aFullTokens List of full term tokens with the
@@ -327,46 +321,44 @@ class SearchDescription
*
* @return SearchDescription[] List of derived search descriptions.
*/
- public function extendWithPartialTerm($aSearchTerm, $bStructuredPhrases, $iPhrase, $aFullTokens)
+ public function extendWithPartialTerm($sToken, $oSearchTerm, $bStructuredPhrases, $iPhrase, $aFullTokens)
{
// Only allow name terms.
- if (!(isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])) {
+ if (!(is_a($oSearchTerm, '\Nominatim\Token\Word'))) {
return array();
}
$aNewSearches = array();
- $iWordID = $aSearchTerm['word_id'];
+ $iWordID = $oSearchTerm->iId;
if ((!$bStructuredPhrases || $iPhrase > 0)
- && sizeof($this->aName)
- && strpos($aSearchTerm['word_token'], ' ') === false
+ && (!empty($this->aName))
+ && strpos($sToken, ' ') === false
) {
- if ($aSearchTerm['search_name_count'] + 1 < CONST_Max_Word_Frequency) {
+ if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
$oSearch = clone $this;
- $oSearch->iSearchRank++;
+ $oSearch->iSearchRank += 2;
$oSearch->aAddress[$iWordID] = $iWordID;
$aNewSearches[] = $oSearch;
} else {
$oSearch = clone $this;
$oSearch->iSearchRank++;
$oSearch->aAddressNonSearch[$iWordID] = $iWordID;
- if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
+ if (preg_match('#^[0-9]+$#', $sToken)) {
$oSearch->iSearchRank += 2;
}
- if (sizeof($aFullTokens)) {
+ if (!empty($aFullTokens)) {
$oSearch->iSearchRank++;
}
$aNewSearches[] = $oSearch;
// revert to the token version?
- foreach ($aFullTokens as $aSearchTermToken) {
- if (empty($aSearchTermToken['country_code'])
- && empty($aSearchTermToken['lat'])
- && empty($aSearchTermToken['class'])
- ) {
+ foreach ($aFullTokens as $oSearchTermToken) {
+ if (is_a($oSearchTermToken, '\Nominatim\Token\Word')) {
$oSearch = clone $this;
$oSearch->iSearchRank++;
- $oSearch->aAddress[$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
+ $oSearch->aAddress[$oSearchTermToken->iId]
+ = $oSearchTermToken->iId;
$aNewSearches[] = $oSearch;
}
}
@@ -374,17 +366,26 @@ class SearchDescription
}
if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
- && (!sizeof($this->aName) || $this->iNamePhrase == $iPhrase)
+ && (empty($this->aName) || $this->iNamePhrase == $iPhrase)
) {
$oSearch = clone $this;
- $oSearch->iSearchRank++;
- if (!sizeof($this->aName)) {
+ $oSearch->iSearchRank += 2;
+ if (empty($this->aName)) {
$oSearch->iSearchRank += 1;
}
- if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
+ if (preg_match('#^[0-9]+$#', $sToken)) {
$oSearch->iSearchRank += 2;
}
- if ($aSearchTerm['search_name_count'] + 1 < CONST_Max_Word_Frequency) {
+ if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
+ if (empty($this->aName)
+ && CONST_Search_NameOnlySearchFrequencyThreshold
+ ) {
+ $oSearch->bRareName =
+ $oSearchTerm->iSearchNameCount
+ < CONST_Search_NameOnlySearchFrequencyThreshold;
+ } else {
+ $oSearch->bRareName = false;
+ }
$oSearch->aName[$iWordID] = $iWordID;
} else {
$oSearch->aNameNonSearch[$iWordID] = $iWordID;
@@ -402,101 +403,91 @@ class SearchDescription
/**
* Query database for places that match this search.
*
- * @param object $oDB Database connection to use.
- * @param mixed[] $aWordFrequencyScores Number of times tokens appears
- * overall in a planet database.
- * @param mixed[] $aExactMatchCache Saves number of exact matches.
- * @param integer $iMinRank Minimum address rank to restrict
- * search to.
- * @param integer $iMaxRank Maximum address rank to restrict
- * search to.
- * @param integer $iLimit Maximum number of results.
+ * @param object $oDB Database connection to use.
+ * @param integer $iMinRank Minimum address rank to restrict search to.
+ * @param integer $iMaxRank Maximum address rank to restrict search to.
+ * @param integer $iLimit Maximum number of results.
*
* @return mixed[] An array with two fields: IDs contains the list of
* matching place IDs and houseNumber the houseNumber
* if appicable or -1 if not.
*/
- public function query(&$oDB, &$aWordFrequencyScores, &$aExactMatchCache, $iMinRank, $iMaxRank, $iLimit)
+ public function query(&$oDB, $iMinRank, $iMaxRank, $iLimit)
{
- $aPlaceIDs = array();
+ $aResults = array();
$iHousenumber = -1;
if ($this->sCountryCode
- && !sizeof($this->aName)
+ && empty($this->aName)
&& !$this->iOperator
&& !$this->sClass
&& !$this->oContext->hasNearPoint()
) {
// Just looking for a country - look it up
if (4 >= $iMinRank && 4 <= $iMaxRank) {
- $aPlaceIDs = $this->queryCountry($oDB);
+ $aResults = $this->queryCountry($oDB);
}
- } elseif (!sizeof($this->aName) && !sizeof($this->aAddress)) {
+ } elseif (empty($this->aName) && empty($this->aAddress)) {
// Neither name nor address? Then we must be
// looking for a POI in a geographic area.
if ($this->oContext->isBoundedSearch()) {
- $aPlaceIDs = $this->queryNearbyPoi($oDB, $iLimit);
+ $aResults = $this->queryNearbyPoi($oDB, $iLimit);
}
} elseif ($this->iOperator == Operator::POSTCODE) {
// looking for postcode
- $aPlaceIDs = $this->queryPostcode($oDB, $iLimit);
+ $aResults = $this->queryPostcode($oDB, $iLimit);
} else {
// Ordinary search:
// First search for places according to name and address.
- $aNamedPlaceIDs = $this->queryNamedPlace(
+ $aResults = $this->queryNamedPlace(
$oDB,
- $aWordFrequencyScores,
$iMinRank,
$iMaxRank,
$iLimit
);
- if (sizeof($aNamedPlaceIDs)) {
- foreach ($aNamedPlaceIDs as $aRow) {
- $aPlaceIDs[] = $aRow['place_id'];
- $aExactMatchCache[$aRow['place_id']] = $aRow['exactmatch'];
+ //now search for housenumber, if housenumber provided
+ if ($this->sHouseNumber && !empty($aResults)) {
+ // Downgrade the rank of the street results, they are missing
+ // the housenumber.
+ foreach ($aResults as $oRes) {
+ $oRes->iResultRank++;
}
- }
- //now search for housenumber, if housenumber provided
- if ($this->sHouseNumber && sizeof($aPlaceIDs)) {
- $aResult = $this->queryHouseNumber($oDB, $aPlaceIDs, $iLimit);
-
- if (sizeof($aResult)) {
- $iHousenumber = $aResult['iHouseNumber'];
- $aPlaceIDs = $aResult['aPlaceIDs'];
- } elseif (!$this->looksLikeFullAddress()) {
- $aPlaceIDs = array();
+ $aHnResults = $this->queryHouseNumber($oDB, $aResults);
+
+ if (!empty($aHnResults)) {
+ foreach ($aHnResults as $oRes) {
+ $aResults[$oRes->iId] = $oRes;
+ }
}
}
// finally get POIs if requested
- if ($this->sClass && sizeof($aPlaceIDs)) {
- $aPlaceIDs = $this->queryPoiByOperator($oDB, $aPlaceIDs, $iLimit);
+ if ($this->sClass && !empty($aResults)) {
+ $aResults = $this->queryPoiByOperator($oDB, $aResults, $iLimit);
}
}
- if (CONST_Debug) {
- echo "
Place IDs: ";
- var_Dump($aPlaceIDs);
- }
+ Debug::printDebugTable('Place IDs', $aResults);
- if (sizeof($aPlaceIDs) && $this->sPostcode) {
- $sSQL = 'SELECT place_id FROM placex';
- $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')';
- $sSQL .= " AND postcode = '".$this->sPostcode."'";
- if (CONST_Debug) var_dump($sSQL);
- $aFilteredPlaceIDs = chksql($oDB->getCol($sSQL));
- if ($aFilteredPlaceIDs) {
- $aPlaceIDs = $aFilteredPlaceIDs;
- if (CONST_Debug) {
- echo "
Place IDs after postcode filtering: ";
- var_Dump($aPlaceIDs);
+ if (!empty($aResults) && $this->sPostcode) {
+ $sPlaceIds = Result::joinIdsByTable($aResults, Result::TABLE_PLACEX);
+ if ($sPlaceIds) {
+ $sSQL = 'SELECT place_id FROM placex';
+ $sSQL .= ' WHERE place_id in ('.$sPlaceIds.')';
+ $sSQL .= " AND postcode != '".$this->sPostcode."'";
+ Debug::printSQL($sSQL);
+ $aFilteredPlaceIDs = chksql($oDB->getCol($sSQL));
+ if ($aFilteredPlaceIDs) {
+ foreach ($aFilteredPlaceIDs as $iPlaceId) {
+ $aResults[$iPlaceId]->iResultRank++;
+ }
}
}
}
- return array('IDs' => $aPlaceIDs, 'houseNumber' => $iHousenumber);
+ return $aResults;
}
@@ -508,11 +499,16 @@ class SearchDescription
if ($this->oContext->bViewboxBounded) {
$sSQL .= ' AND ST_Intersects('.$this->oContext->sqlViewboxSmall.', geometry)';
}
- $sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1";
+ $sSQL .= ' ORDER BY st_area(geometry) DESC LIMIT 1';
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
+
+ $aResults = array();
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId);
+ }
- return chksql($oDB->getCol($sSQL));
+ return $aResults;
}
private function queryNearbyPoi(&$oDB, $iLimit)
@@ -521,6 +517,7 @@ class SearchDescription
return array();
}
+ $aDBResults = array();
$sPoiTable = $this->poiTable();
$sSQL = 'SELECT count(*) FROM pg_tables WHERE tablename = \''.$sPoiTable."'";
@@ -545,8 +542,8 @@ class SearchDescription
$sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('ct.centroid').' ASC';
}
$sSQL .= " limit $iLimit";
- if (CONST_Debug) var_dump($sSQL);
- return chksql($oDB->getCol($sSQL));
+ Debug::printSQL($sSQL);
+ $aDBResults = chksql($oDB->getCol($sSQL));
}
if ($this->oContext->hasNearPoint()) {
@@ -557,20 +554,25 @@ class SearchDescription
if ($this->oContext->sqlCountryList) {
$sSQL .= ' AND country_code in '.$this->oContext->sqlCountryList;
}
- $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('centroid')." ASC";
+ $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('centroid').' ASC';
$sSQL .= " LIMIT $iLimit";
- if (CONST_Debug) var_dump($sSQL);
- return chksql($oDB->getCol($sSQL));
+ Debug::printSQL($sSQL);
+ $aDBResults = chksql($oDB->getCol($sSQL));
}
- return array();
+ $aResults = array();
+ foreach ($aDBResults as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId);
+ }
+
+ return $aResults;
}
private function queryPostcode(&$oDB, $iLimit)
{
$sSQL = 'SELECT p.place_id FROM location_postcode p ';
- if (sizeof($this->aAddress)) {
+ if (!empty($this->aAddress)) {
$sSQL .= ', search_name s ';
$sSQL .= 'WHERE s.place_id = p.parent_place_id ';
$sSQL .= 'AND array_cat(s.nameaddress_vector, s.name_vector)';
@@ -584,17 +586,26 @@ class SearchDescription
$sSQL .= $this->oContext->excludeSQL(' AND p.place_id');
$sSQL .= " LIMIT $iLimit";
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
+
+ $aResults = array();
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_POSTCODE);
+ }
- return chksql($oDB->getCol($sSQL));
+ return $aResults;
}
- private function queryNamedPlace(&$oDB, $aWordFrequencyScores, $iMinAddressRank, $iMaxAddressRank, $iLimit)
+ private function queryNamedPlace(&$oDB, $iMinAddressRank, $iMaxAddressRank, $iLimit)
{
$aTerms = array();
$aOrder = array();
- if ($this->sHouseNumber && sizeof($this->aAddress)) {
+ // Sort by existence of the requested house number but only if not
+ // too many results are expected for the street, i.e. if the result
+ // will be narrowed down by an address. Remeber that with ordering
+ // every single result has to be checked.
+ if ($this->sHouseNumber && (!empty($this->aAddress) || $this->sPostcode)) {
$sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
$aOrder[] = ' (';
$aOrder[0] .= 'EXISTS(';
@@ -620,16 +631,12 @@ class SearchDescription
$aOrder[0] .= ') DESC';
}
- if (sizeof($this->aName)) {
+ if (!empty($this->aName)) {
$aTerms[] = 'name_vector @> '.getArraySQL($this->aName);
}
- if (sizeof($this->aAddress)) {
+ if (!empty($this->aAddress)) {
// For infrequent name terms disable index usage for address
- if (CONST_Search_NameOnlySearchFrequencyThreshold
- && sizeof($this->aName) == 1
- && $aWordFrequencyScores[$this->aName[reset($this->aName)]]
- < CONST_Search_NameOnlySearchFrequencyThreshold
- ) {
+ if ($this->bRareName) {
$aTerms[] = 'array_cat(nameaddress_vector,ARRAY[]::integer[]) @> '.getArraySQL($this->aAddress);
} else {
$aTerms[] = 'nameaddress_vector @> '.getArraySQL($this->aAddress);
@@ -642,13 +649,13 @@ class SearchDescription
}
if ($this->sHouseNumber) {
- $aTerms[] = "address_rank between 16 and 27";
+ $aTerms[] = 'address_rank between 16 and 27';
} elseif (!$this->sClass || $this->iOperator == Operator::NAME) {
if ($iMinAddressRank > 0) {
- $aTerms[] = "address_rank >= ".$iMinAddressRank;
+ $aTerms[] = 'address_rank >= '.$iMinAddressRank;
}
if ($iMaxAddressRank < 30) {
- $aTerms[] = "address_rank <= ".$iMaxAddressRank;
+ $aTerms[] = 'address_rank <= '.$iMaxAddressRank;
}
}
@@ -656,7 +663,7 @@ class SearchDescription
$aTerms[] = $this->oContext->withinSQL('centroid');
$aOrder[] = $this->oContext->distanceSQL('centroid');
} elseif ($this->sPostcode) {
- if (!sizeof($this->aAddress)) {
+ if (empty($this->aAddress)) {
$aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."' AND ST_DWithin(search_name.centroid, p.geometry, 0.1))";
} else {
$aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."')";
@@ -679,12 +686,12 @@ class SearchDescription
if ($this->sHouseNumber) {
$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)';
+ $sImportanceSQL = '(CASE WHEN importance = 0 OR importance IS NULL THEN 0.75001-(search_rank::float/40) ELSE importance END)';
}
$sImportanceSQL .= $this->oContext->viewboxImportanceSQL('centroid');
$aOrder[] = "$sImportanceSQL DESC";
- if (sizeof($this->aFullNameAddress)) {
+ if (!empty($this->aFullNameAddress)) {
$sExactMatchSQL = ' ( ';
$sExactMatchSQL .= ' SELECT count(*) FROM ( ';
$sExactMatchSQL .= ' SELECT unnest('.getArraySQL($this->aFullNameAddress).')';
@@ -698,49 +705,60 @@ class SearchDescription
}
if ($this->sHouseNumber || $this->sClass) {
- $iLimit = 20;
+ $iLimit = 40;
}
- if (sizeof($aTerms)) {
+ $aResults = array();
+
+ if (!empty($aTerms)) {
$sSQL = 'SELECT place_id,'.$sExactMatchSQL;
$sSQL .= ' FROM search_name';
$sSQL .= ' WHERE '.join(' and ', $aTerms);
$sSQL .= ' ORDER BY '.join(', ', $aOrder);
$sSQL .= ' LIMIT '.$iLimit;
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
- return chksql(
+ $aDBResults = chksql(
$oDB->getAll($sSQL),
- "Could not get places for search terms."
+ 'Could not get places for search terms.'
);
+
+ foreach ($aDBResults as $aResult) {
+ $oResult = new Result($aResult['place_id']);
+ $oResult->iExactMatches = $aResult['exactmatch'];
+ $aResults[$aResult['place_id']] = $oResult;
+ }
}
- return array();
+ return $aResults;
}
- private function queryHouseNumber(&$oDB, $aRoadPlaceIDs, $iLimit)
+ private function queryHouseNumber(&$oDB, $aRoadPlaceIDs)
{
- $sPlaceIDs = join(',', $aRoadPlaceIDs);
+ $aResults = array();
+ $sPlaceIDs = Result::joinIdsByTable($aRoadPlaceIDs, Result::TABLE_PLACEX);
+
+ if (!$sPlaceIDs) {
+ return $aResults;
+ }
$sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
$sSQL = 'SELECT place_id FROM placex ';
$sSQL .= 'WHERE parent_place_id in ('.$sPlaceIDs.')';
$sSQL .= " AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."'";
$sSQL .= $this->oContext->excludeSQL(' AND place_id');
- $sSQL .= " LIMIT $iLimit";
-
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = chksql($oDB->getCol($sSQL));
+ Debug::printSQL($sSQL);
- if (sizeof($aPlaceIDs)) {
- return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => -1);
+ // XXX should inherit the exactMatches from its parent
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId);
}
$bIsIntHouseNumber= (bool) preg_match('/[0-9]+/', $this->sHouseNumber);
$iHousenumber = intval($this->sHouseNumber);
- if ($bIsIntHouseNumber) {
+ if ($bIsIntHouseNumber && empty($aResults)) {
// if nothing found, search in the interpolation line table
$sSQL = 'SELECT distinct place_id FROM location_property_osmline';
$sSQL .= ' WHERE startnumber is not NULL';
@@ -754,40 +772,36 @@ class SearchDescription
$sSQL .= "interpolationtype='odd'";
}
$sSQL .= " or interpolationtype='all') and ";
- $sSQL .= $iHousenumber.">=startnumber and ";
- $sSQL .= $iHousenumber."<=endnumber";
+ $sSQL .= $iHousenumber.'>=startnumber and ';
+ $sSQL .= $iHousenumber.'<=endnumber';
$sSQL .= $this->oContext->excludeSQL(' AND place_id');
- $sSQL .= " limit $iLimit";
-
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = chksql($oDB->getCol($sSQL, 0));
+ Debug::printSQL($sSQL);
- if (sizeof($aPlaceIDs)) {
- return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => $iHousenumber);
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $oResult = new Result($iPlaceId, Result::TABLE_OSMLINE);
+ $oResult->iHouseNumber = $iHousenumber;
+ $aResults[$iPlaceId] = $oResult;
}
}
// If nothing found try the aux fallback table
- if (CONST_Use_Aux_Location_data) {
+ if (CONST_Use_Aux_Location_data && empty($aResults)) {
$sSQL = 'SELECT place_id FROM location_property_aux';
$sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.')';
$sSQL .= " AND housenumber = '".$this->sHouseNumber."'";
$sSQL .= $this->oContext->excludeSQL(' AND place_id');
- $sSQL .= " limit $iLimit";
-
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = chksql($oDB->getCol($sSQL));
+ Debug::printSQL($sSQL);
- if (sizeof($aPlaceIDs)) {
- return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => -1);
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_AUX);
}
}
// If nothing found then search in Tiger data (location_property_tiger)
- if (CONST_Use_US_Tiger_Data && $bIsIntHouseNumber) {
- $sSQL = 'SELECT distinct place_id FROM location_property_tiger';
+ if (CONST_Use_US_Tiger_Data && $bIsIntHouseNumber && empty($aResults)) {
+ $sSQL = 'SELECT place_id FROM location_property_tiger';
$sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.') and (';
if ($iHousenumber % 2 == 0) {
$sSQL .= "interpolationtype='even'";
@@ -795,28 +809,31 @@ class SearchDescription
$sSQL .= "interpolationtype='odd'";
}
$sSQL .= " or interpolationtype='all') and ";
- $sSQL .= $iHousenumber.">=startnumber and ";
- $sSQL .= $iHousenumber."<=endnumber";
+ $sSQL .= $iHousenumber.'>=startnumber and ';
+ $sSQL .= $iHousenumber.'<=endnumber';
$sSQL .= $this->oContext->excludeSQL(' AND place_id');
- $sSQL .= " limit $iLimit";
-
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = chksql($oDB->getCol($sSQL, 0));
+ Debug::printSQL($sSQL);
- if (sizeof($aPlaceIDs)) {
- return array('aPlaceIDs' => $aPlaceIDs, 'iHouseNumber' => $iHousenumber);
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $oResult = new Result($iPlaceId, Result::TABLE_TIGER);
+ $oResult->iHouseNumber = $iHousenumber;
+ $aResults[$iPlaceId] = $oResult;
}
}
- return array();
+ return $aResults;
}
private function queryPoiByOperator(&$oDB, $aParentIDs, $iLimit)
{
- $sPlaceIDs = join(',', $aParentIDs);
- $aClassPlaceIDs = array();
+ $aResults = array();
+ $sPlaceIDs = Result::joinIdsByTable($aParentIDs, Result::TABLE_PLACEX);
+
+ if (!$sPlaceIDs) {
+ return $aResults;
+ }
if ($this->iOperator == Operator::TYPE || $this->iOperator == Operator::NAME) {
// If they were searching for a named class (i.e. 'Kings Head pub')
@@ -825,14 +842,16 @@ class SearchDescription
$sSQL .= " WHERE place_id in ($sPlaceIDs)";
$sSQL .= " AND class='".$this->sClass."' ";
$sSQL .= " AND type='".$this->sType."'";
- $sSQL .= " AND linked_place_id is null";
+ $sSQL .= ' AND linked_place_id is null';
$sSQL .= $this->oContext->excludeSQL(' AND place_id');
- $sSQL .= " ORDER BY rank_search ASC ";
+ $sSQL .= ' ORDER BY rank_search ASC ';
$sSQL .= " LIMIT $iLimit";
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
- $aClassPlaceIDs = chksql($oDB->getCol($sSQL));
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId);
+ }
}
// NEAR and IN are handled the same
@@ -842,7 +861,7 @@ class SearchDescription
$bCacheTable = (bool) chksql($oDB->getOne($sSQL));
$sSQL = "SELECT min(rank_search) FROM placex WHERE place_id in ($sPlaceIDs)";
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
$iMaxRank = (int)chksql($oDB->getOne($sSQL));
// For state / country level searches the normal radius search doesn't work very well
@@ -853,9 +872,9 @@ class SearchDescription
$sSQL .= " WHERE place_id in ($sPlaceIDs)";
$sSQL .= " AND rank_search < $iMaxRank + 5";
$sSQL .= " AND ST_GeometryType(geometry) in ('ST_Polygon','ST_MultiPolygon')";
- $sSQL .= " ORDER BY rank_search ASC ";
- $sSQL .= " LIMIT 1";
- if (CONST_Debug) var_dump($sSQL);
+ $sSQL .= ' ORDER BY rank_search ASC ';
+ $sSQL .= ' LIMIT 1';
+ Debug::printSQL($sSQL);
$sPlaceGeom = chksql($oDB->getOne($sSQL));
}
@@ -865,7 +884,7 @@ class SearchDescription
$iMaxRank += 5;
$sSQL = 'SELECT place_id FROM placex';
$sSQL .= " WHERE place_id in ($sPlaceIDs) and rank_search < $iMaxRank";
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
$aPlaceIDs = chksql($oDB->getCol($sSQL));
$sPlaceIDs = join(',', $aPlaceIDs);
}
@@ -880,7 +899,7 @@ class SearchDescription
if ($this->oContext->hasNearPoint()) {
$sOrderBySQL = $this->oContext->distanceSQL('l.centroid');
} elseif ($sPlaceIDs) {
- $sOrderBySQL = "ST_Distance(l.centroid, f.geometry)";
+ $sOrderBySQL = 'ST_Distance(l.centroid, f.geometry)';
} elseif ($sPlaceGeom) {
$sOrderBySQL = "ST_Distance(st_centroid('".$sPlaceGeom."'), l.centroid)";
}
@@ -896,7 +915,7 @@ class SearchDescription
$sSQL .= ' from '.$sClassTable.' as l';
if ($sPlaceIDs) {
- $sSQL .= ",placex as f WHERE ";
+ $sSQL .= ',placex as f WHERE ';
$sSQL .= "f.place_id in ($sPlaceIDs) ";
$sSQL .= " AND ST_DWithin(l.centroid, f.centroid, $fRange)";
} elseif ($sPlaceGeom) {
@@ -910,9 +929,11 @@ class SearchDescription
}
$sSQL .= " limit $iLimit";
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
- $aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($oDB->getCol($sSQL)));
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId);
+ }
} else {
if ($this->oContext->hasNearPoint()) {
$fRange = $this->oContext->nearRadius();
@@ -922,7 +943,7 @@ class SearchDescription
if ($this->oContext->hasNearPoint()) {
$sOrderBySQL = $this->oContext->distanceSQL('l.geometry');
} else {
- $sOrderBySQL = "ST_Distance(l.geometry, f.geometry)";
+ $sOrderBySQL = 'ST_Distance(l.geometry, f.geometry)';
}
$sSQL = 'SELECT distinct l.place_id';
@@ -936,18 +957,20 @@ class SearchDescription
$sSQL .= " AND l.type='".$this->sType."'";
$sSQL .= $this->oContext->excludeSQL(' AND l.place_id');
if ($sOrderBySQL) {
- $sSQL .= "ORDER BY orderterm ASC";
+ $sSQL .= 'ORDER BY orderterm ASC';
}
$sSQL .= " limit $iLimit";
- if (CONST_Debug) var_dump($sSQL);
+ Debug::printSQL($sSQL);
- $aClassPlaceIDs = array_merge($aClassPlaceIDs, chksql($oDB->getCol($sSQL)));
+ foreach (chksql($oDB->getCol($sSQL)) as $iPlaceId) {
+ $aResults[$iPlaceId] = new Result($iPlaceId);
+ }
}
}
}
- return $aClassPlaceIDs;
+ return $aResults;
}
private function poiTable()
@@ -983,25 +1006,43 @@ class SearchDescription
//////////// Debugging functions
+ public function debugInfo()
+ {
+ return array(
+ 'Search rank' => $this->iSearchRank,
+ 'Country code' => $this->sCountryCode,
+ 'Name terms' => $this->aName,
+ 'Name terms (stop words)' => $this->aNameNonSearch,
+ 'Address terms' => $this->aAddress,
+ 'Address terms (stop words)' => $this->aAddressNonSearch,
+ 'Address terms (full words)' => $this->aFullNameAddress,
+ 'Special search' => $this->iOperator,
+ 'Class' => $this->sClass,
+ 'Type' => $this->sType,
+ 'House number' => $this->sHouseNumber,
+ 'Postcode' => $this->sPostcode
+ );
+ }
+
public function dumpAsHtmlTableRow(&$aWordIDs)
{
$kf = function ($k) use (&$aWordIDs) {
return $aWordIDs[$k];
};
- echo "";
+ echo '
';
echo "| $this->iSearchRank | ";
- echo "".join(', ', array_map($kf, $this->aName))." | ";
- echo "".join(', ', array_map($kf, $this->aNameNonSearch))." | ";
- echo "".join(', ', array_map($kf, $this->aAddress))." | ";
- echo "".join(', ', array_map($kf, $this->aAddressNonSearch))." | ";
- echo "".$this->sCountryCode." | ";
- echo "".Operator::toString($this->iOperator)." | ";
- echo "".$this->sClass." | ";
- echo "".$this->sType." | ";
- echo "".$this->sPostcode." | ";
- echo "".$this->sHouseNumber." | ";
-
- echo "
";
+ echo ''.join(', ', array_map($kf, $this->aName)).' | ';
+ echo ''.join(', ', array_map($kf, $this->aNameNonSearch)).' | ';
+ echo ''.join(', ', array_map($kf, $this->aAddress)).' | ';
+ echo ''.join(', ', array_map($kf, $this->aAddressNonSearch)).' | ';
+ echo ''.$this->sCountryCode.' | ';
+ echo ''.Operator::toString($this->iOperator).' | ';
+ echo ''.$this->sClass.' | ';
+ echo ''.$this->sType.' | ';
+ echo ''.$this->sPostcode.' | ';
+ echo ''.$this->sHouseNumber.' | ';
+
+ echo '';
}
}