X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/f8d55b5448b6be721a6ea05567930fe244ead7d1..0067555c38cece9cf11a0074ab59456c22c8c97d:/lib/Geocode.php
diff --git a/lib/Geocode.php b/lib/Geocode.php
index d96aaac2..88a969a5 100644
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -709,6 +709,8 @@ class Geocode
Score how good the search is so they can be ordered
*/
+ $iGlobalRank = 0;
+
foreach ($aPhrases as $iPhrase => $aPhrase) {
$aNewPhraseSearches = array();
if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
@@ -745,14 +747,19 @@ 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) {
+ $iGlobalRank++;
+ }
}
} elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') {
// We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
- if (!$bHavePostcode && $aSearch['sPostcode'] === '' && $aSearch['sHouseNumber'] === '' &&
+ if ($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 ($aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
+ if (!$bHavePostcode && $aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
$aNewSearch = $aSearch;
$aNewSearch['sOperator'] = 'postcode';
$aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
@@ -774,8 +781,8 @@ class Geocode
// 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[] = $aSearch;
/*
// Fall back to not searching for this item (better than nothing)
@@ -852,7 +859,8 @@ class Geocode
}
}
- if (!sizeof($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'])) $aSearch['iSearchRank'] += 1;
@@ -914,7 +922,7 @@ class Geocode
//if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
}
- // Revisit searches, giving penalty to unlikely combinations
+ // Revisit searches, drop bad searches and give penalty to unlikely combinations.
$aGroupedSearches = array();
foreach ($aSearches as $aSearch) {
if (!$aSearch['aName']) {
@@ -922,6 +930,12 @@ class Geocode
continue;
}
}
+ if ($this->aCountryCodes && $aSearch['sCountryCode']
+ && !in_array($aSearch['sCountryCode'], $this->aCountryCodes)) {
+ continue;
+ }
+
+ $aSearch['iSearchRank'] += $iGlobalRank;
$aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
}
ksort($aGroupedSearches);
@@ -1009,76 +1023,59 @@ class Geocode
$aSearchResults = array();
if ($sQuery || $this->aStructuredQuery) {
- // 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' => '',
- 'sPostcode' => '',
- 'oNear' => $oNearPoint
- )
- );
-
- // Any 'special' terms in the search?
- $bSpecialTerms = false;
- preg_match_all('/\\[([\\w_]*)=([\\w_]*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
- foreach ($aSpecialTermsRaw as $aSpecialTerm) {
- $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
- if (!$bSpecialTerms) {
- $aNewSearches = array();
- foreach ($aSearches as $aSearch) {
- $aNewSearch = $aSearch;
- $aNewSearch['sClass'] = $aSpecialTerm[1];
- $aNewSearch['sType'] = $aSpecialTerm[2];
- $aNewSearches[] = $aNewSearch;
- }
+ // Start with a single blank search
+ $aSearches = array(new SearchDescription());
- $aSearches = $aNewSearches;
- $bSpecialTerms = true;
- }
+ if ($oNearPoint) {
+ $aSearches[0]->setNear($oNearPoint);
}
- preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
- if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) {
- $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
- unset($this->aStructuredQuery['amenity']);
+ if ($sQuery) {
+ $sQuery = $aSearches[0]->extractKeyValuePairs($sQuery);
}
- foreach ($aSpecialTermsRaw as $aSpecialTerm) {
- $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
- if ($bSpecialTerms) {
- continue;
+ $sSpecialTerm = '';
+ if ($sQuery) {
+ preg_match_all(
+ '/\\[([\\w ]*)\\]/u',
+ $sQuery,
+ $aSpecialTermsRaw,
+ PREG_SET_ORDER
+ );
+ foreach ($aSpecialTermsRaw as $aSpecialTerm) {
+ $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
+ if (!$sSpecialTerm) {
+ $sSpecialTerm = $aSpecialTerm[1];
+ }
}
+ }
+ if (!$sSpecialTerm && $this->aStructuredQuery
+ && isset($this->aStructuredQuery['amenity'])) {
+ $sSpecialTerm = $this->aStructuredQuery['amenity'];
+ unset($this->aStructuredQuery['amenity']);
+ }
- $sToken = chksql($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';
- $sSQL .= ' FROM word ';
+ if ($sSpecialTerm && !$aSearches[0]->hasOperator()) {
+ $sSpecialTerm = pg_escape_string($sSpecialTerm);
+ $sToken = chksql(
+ $this->oDB->getOne("SELECT make_standard_name('$sSpecialTerm')"),
+ "Cannot decode query. Wrong encoding?"
+ );
+ $sSQL = 'SELECT class, type FROM word ';
$sSQL .= ' WHERE word_token in (\' '.$sToken.'\')';
- $sSQL .= ') AS x ';
- $sSQL .= ' WHERE (class is not null AND class not in (\'place\'))';
+ $sSQL .= ' AND class is not null AND class not in (\'place\')';
if (CONST_Debug) var_Dump($sSQL);
$aSearchWords = chksql($this->oDB->getAll($sSQL));
$aNewSearches = array();
- foreach ($aSearches as $aSearch) {
+ foreach ($aSearches as $oSearch) {
foreach ($aSearchWords as $aSearchTerm) {
- $aNewSearch = $aSearch;
- $aNewSearch['sClass'] = $aSearchTerm['class'];
- $aNewSearch['sType'] = $aSearchTerm['type'];
- $aNewSearches[] = $aNewSearch;
- $bSpecialTerms = true;
+ $oNewSearch = clone $oSearch;
+ $oNewSearch->setPoiSearch(
+ Operator::TYPE,
+ $aSearchTerm['class'],
+ $aSearchTerm['type'],
+ );
+ $aNewSearches[] = $oNewSearch;
}
}
$aSearches = $aNewSearches;
@@ -1198,10 +1195,10 @@ class Geocode
foreach ($aGroupedSearches as $aSearches) {
foreach ($aSearches as $aSearch) {
- if ($aSearch['iSearchRank'] < $this->iMaxRank) {
- if (!isset($aReverseGroupedSearches[$aSearch['iSearchRank']])) $aReverseGroupedSearches[$aSearch['iSearchRank']] = array();
- $aReverseGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+ if (!isset($aReverseGroupedSearches[$aSearch->getRank()])) {
+ $aReverseGroupedSearches[$aSearch->getRank()] = array();
}
+ $aReverseGroupedSearches[$aSearch->getRank()][] = $aSearch;
}
}
@@ -1212,38 +1209,9 @@ class Geocode
// 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'] < $this->iMaxRank) {
- if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
- $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
- }
- }
- ksort($aGroupedSearches);
- }
-
- if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) {
- $aCopyGroupedSearches = $aGroupedSearches;
- foreach ($aCopyGroupedSearches as $iGroup => $aSearches) {
- foreach ($aSearches as $iSearch => $aSearch) {
- $aReductionsList = array($aSearch['aAddress']);
- $iSearchRank = $aSearch['iSearchRank'];
- while (sizeof($aReductionsList) > 0) {
- $iSearchRank += 5;
- if ($iSearchRank > iMaxRank) break 3;
- $aNewReductionsList = array();
- foreach ($aReductionsList as $aReductionsWordList) {
- for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++) {
- $aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1));
- $aReverseSearch = $aSearch;
- $aSearch['aAddress'] = $aReductionsWordListResult;
- $aSearch['iSearchRank'] = $iSearchRank;
- $aGroupedSearches[$iSearchRank][] = $aReverseSearch;
- if (sizeof($aReductionsWordListResult) > 0) {
- $aNewReductionsList[] = $aReductionsWordListResult;
- }
- }
- }
- $aReductionsList = $aNewReductionsList;
- }
+ if ($aSearch->getRank() < $this->iMaxRank) {
+ if (!isset($aGroupedSearches[$aSearch->getRank()])) $aGroupedSearches[$aSearch->getRank()] = array();
+ $aGroupedSearches[$aSearch->getRank()][] = $aSearch;
}
}
ksort($aGroupedSearches);
@@ -1276,10 +1244,6 @@ 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($aSearch['aAddress'])) {
if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber'] && !$aSearch['oNear']) {