X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/4d6da459da6237a1fe383c6edc7e1166e56fc8e5..27753fb8dbce6bb595f95bad637078ab2106d794:/lib/Geocode.php
diff --git a/lib/Geocode.php b/lib/Geocode.php
index bb463445..5f4a0174 100644
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -222,9 +222,12 @@
foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID)
{
$iExcludedPlaceID = (int)$iExcludedPlaceID;
- if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
+ if ($iExcludedPlaceID)
+ $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID;
}
- $this->aExcludePlaceIDs = $aExcludePlaceIDs;
+
+ if (isset($aExcludePlaceIDs))
+ $this->aExcludePlaceIDs = $aExcludePlaceIDs;
}
// Only certain ranks of feature
@@ -441,6 +444,283 @@
return $aSearchResults;
}
+ function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases)
+ {
+ /*
+ Calculate all searches using aValidTokens i.e.
+ 'Wodsworth Road, Sheffield' =>
+
+ Phrase Wordset
+ 0 0 (wodsworth road)
+ 0 1 (wodsworth)(road)
+ 1 0 (sheffield)
+
+ Score how good the search is so they can be ordered
+ */
+ foreach($aPhrases as $iPhrase => $sPhrase)
+ {
+ $aNewPhraseSearches = array();
+ if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
+ else $sPhraseType = '';
+
+ foreach($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset)
+ {
+ // Too many permutations - too expensive
+ if ($iWordSet > 120) break;
+
+ $aWordsetSearches = $aSearches;
+
+ // Add all words from this wordset
+ foreach($aWordset as $iToken => $sToken)
+ {
+ //echo "
$sToken";
+ $aNewWordsetSearches = array();
+
+ foreach($aWordsetSearches as $aCurrentSearch)
+ {
+ //echo "";
+ //var_dump($aCurrentSearch);
+ //echo "";
+
+ // If the token is valid
+ if (isset($aValidTokens[' '.$sToken]))
+ {
+ foreach($aValidTokens[' '.$sToken] as $aSearchTerm)
+ {
+ $aSearch = $aCurrentSearch;
+ $aSearch['iSearchRank']++;
+ if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0')
+ {
+ if ($aSearch['sCountryCode'] === false)
+ {
+ $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
+ // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
+ if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)))
+ {
+ $aSearch['iSearchRank'] += 5;
+ }
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ }
+ elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
+ {
+ if ($aSearch['fLat'] === '')
+ {
+ $aSearch['fLat'] = $aSearchTerm['lat'];
+ $aSearch['fLon'] = $aSearchTerm['lon'];
+ $aSearch['fRadius'] = $aSearchTerm['radius'];
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ }
+ elseif ($sPhraseType == 'postalcode')
+ {
+ // 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']))
+ {
+ $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))
+ {
+ $aSearch['aAddress'][$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'] = $iPhrase;
+ }
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+
+ }
+ elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house')
+ {
+ if ($aSearch['sHouseNumber'] === '')
+ {
+ $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'] === '')
+ {
+ $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))
+ {
+ $aSearch['aAddress'][$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'] = $iPhrase;
+ }
+ if ($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]) && $sPhraseType != 'country')
+ {
+ // Allow searching for a word - but at extra cost
+ foreach($aValidTokens[$sToken] as $aSearchTerm)
+ {
+ if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
+ {
+ if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false)
+ {
+ $aSearch = $aCurrentSearch;
+ $aSearch['iSearchRank'] += 1;
+ if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
+ {
+ $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version?
+ {
+ $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ $aSearch['iSearchRank'] += 1;
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
+ {
+ if (empty($aSearchTermToken['country_code'])
+ && empty($aSearchTermToken['lat'])
+ && empty($aSearchTermToken['class']))
+ {
+ $aSearch = $aCurrentSearch;
+ $aSearch['iSearchRank'] += 1;
+ $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ }
+ }
+ else
+ {
+ $aSearch['aAddressNonSearch'][$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)
+ {
+ $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)
+ $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ else
+ $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
+ $aSearch['iNamePhrase'] = $iPhrase;
+ if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ }
+ }
+ }
+ }
+ else
+ {
+ // Allow skipping a word - but at EXTREAM cost
+ //$aSearch = $aCurrentSearch;
+ //$aSearch['iSearchRank']+=100;
+ //$aNewWordsetSearches[] = $aSearch;
+ }
+ }
+ // Sort and cut
+ usort($aNewWordsetSearches, 'bySearchRank');
+ $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
+ }
+ //var_Dump('
',sizeof($aWordsetSearches)); exit;
+
+ $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
+ usort($aNewPhraseSearches, 'bySearchRank');
+
+ $aSearchHash = array();
+ foreach($aNewPhraseSearches as $iSearch => $aSearch)
+ {
+ $sHash = serialize($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'] < $this->iMaxRank)
+ {
+ if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
+ $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+ }
+ }
+ ksort($aGroupedSearches);
+
+ $iSearchCount = 0;
+ $aSearches = array();
+ foreach($aGroupedSearches as $iScore => $aNewSearches)
+ {
+ $iSearchCount += sizeof($aNewSearches);
+ $aSearches = array_merge($aSearches, $aNewSearches);
+ if ($iSearchCount > 50) break;
+ }
+
+ //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
+
+ }
+ return $aGroupedSearches;
+
+ }
+
/* Perform the actual query lookup.
Returns an ordered list of results, each with the following fields:
@@ -476,7 +756,6 @@
if (!$this->sQuery && !$this->aStructuredQuery) return false;
$sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$this->aLangPrefOrder))."]";
-
$sCountryCodesSQL = false;
if ($this->aCountryCodes && sizeof($this->aCountryCodes))
{
@@ -579,10 +858,10 @@
preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
$aSpecialTerms = array();
- if (isset($aStructuredQuery['amenity']) && $aStructuredQuery['amenity'])
+ if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity'])
{
- $aSpecialTermsRaw[] = array('['.$aStructuredQuery['amenity'].']', $aStructuredQuery['amenity']);
- unset($aStructuredQuery['amenity']);
+ $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
+ unset($this->aStructuredQuery['amenity']);
}
foreach($aSpecialTermsRaw as $aSpecialTerm)
{
@@ -680,8 +959,8 @@
foreach($aDatabaseWords as $aToken)
{
// Very special case - require 2 letter country param to match the country code found
- if ($bStructuredPhrases && $aToken['country_code'] && !empty($aStructuredQuery['country'])
- && strlen($aStructuredQuery['country']) == 2 && strtolower($aStructuredQuery['country']) != $aToken['country_code'])
+ if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
+ && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code'])
{
continue;
}
@@ -755,287 +1034,38 @@
// Start the search process
$aResultPlaceIDs = array();
- /*
- Calculate all searches using aValidTokens i.e.
- 'Wodsworth Road, Sheffield' =>
+ $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
- Phrase Wordset
- 0 0 (wodsworth road)
- 0 1 (wodsworth)(road)
- 1 0 (sheffield)
-
- Score how good the search is so they can be ordered
- */
- foreach($aPhrases as $iPhrase => $sPhrase)
+ if ($this->bReverseInPlan)
{
- $aNewPhraseSearches = array();
- if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
- else $sPhraseType = '';
-
- foreach($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset)
+ // 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($aPhrases) > 1)
{
- // Too many permutations - too expensive
- if ($iWordSet > 120) break;
-
- $aWordsetSearches = $aSearches;
+ $aFinalPhrase = end($aPhrases);
+ $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
+ }
+ $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
- // Add all words from this wordset
- foreach($aWordset as $iToken => $sToken)
+ foreach($aGroupedSearches as $aSearches)
+ {
+ foreach($aSearches as $aSearch)
{
- //echo "
$sToken";
- $aNewWordsetSearches = array();
-
- foreach($aWordsetSearches as $aCurrentSearch)
+ if ($aSearch['iSearchRank'] < $this->iMaxRank)
{
- //echo "";
- //var_dump($aCurrentSearch);
- //echo "";
-
- // If the token is valid
- if (isset($aValidTokens[' '.$sToken]))
- {
- foreach($aValidTokens[' '.$sToken] as $aSearchTerm)
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank']++;
- if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0')
- {
- if ($aSearch['sCountryCode'] === false)
- {
- $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
- // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
- // If reverse order is enabled, it may appear at the beginning as well.
- if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) &&
- (!$this->bReverseInPlan || $iToken > 0 || $iPhrase > 0))
- {
- $aSearch['iSearchRank'] += 5;
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
- {
- if ($aSearch['fLat'] === '')
- {
- $aSearch['fLat'] = $aSearchTerm['lat'];
- $aSearch['fLon'] = $aSearchTerm['lon'];
- $aSearch['fRadius'] = $aSearchTerm['radius'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- elseif ($sPhraseType == 'postalcode')
- {
- // 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']))
- {
- $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[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
- {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- }
- else
- {
- $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1000; // skip;
- }
- }
- else
- {
- $aSearch['aName'][$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'] === '')
- {
- $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) $aSearch['iSearchRank']++;
- if ($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'] === '')
- {
- $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) $aSearch['iSearchRank'] += 1;
-
- // Do we have a shortcut id?
- if ($aSearch['sOperator'] == 'name')
- {
- $sSQL = "select get_tagpair('".$aSearch['sClass']."', '".$aSearch['sType']."')";
- if ($iAmenityID = $this->oDB->getOne($sSQL))
- {
- $aValidTokens[$aSearch['sClass'].':'.$aSearch['sType']] = array('word_id' => $iAmenityID);
- $aSearch['aName'][$iAmenityID] = $iAmenityID;
- $aSearch['sClass'] = '';
- $aSearch['sType'] = '';
- }
- }
- if ($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[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
- {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- }
- else
- {
- $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1000; // skip;
- }
- }
- else
- {
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- //$aSearch['iNamePhrase'] = $iPhrase;
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- }
- if (isset($aValidTokens[$sToken]))
- {
- // Allow searching for a word - but at extra cost
- foreach($aValidTokens[$sToken] as $aSearchTerm)
- {
- if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
- {
- if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false)
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
- {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- elseif (isset($aValidTokens[' '.$sToken]) && strlen($sToken) >= 4) // revert to the token version?
- {
- $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
- {
- if (empty($aSearchTermToken['country_code'])
- && empty($aSearchTermToken['lat'])
- && empty($aSearchTermToken['class']))
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- }
- else
- {
- $aSearch['aAddressNonSearch'][$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)
- {
- $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)
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- else
- $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iNamePhrase'] = $iPhrase;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- }
- }
- else
- {
- // Allow skipping a word - but at EXTREAM cost
- //$aSearch = $aCurrentSearch;
- //$aSearch['iSearchRank']+=100;
- //$aNewWordsetSearches[] = $aSearch;
- }
+ if (!isset($aReverseGroupedSearches[$aSearch['iSearchRank']])) $aReverseGroupedSearches[$aSearch['iSearchRank']] = array();
+ $aReverseGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
}
- // Sort and cut
- usort($aNewWordsetSearches, 'bySearchRank');
- $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
- }
- //var_Dump('
',sizeof($aWordsetSearches)); exit;
-
- $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
- usort($aNewPhraseSearches, 'bySearchRank');
- $aSearchHash = array();
- foreach($aNewPhraseSearches as $iSearch => $aSearch)
- {
- $sHash = serialize($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'] < $this->iMaxRank)
- {
- if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
- $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
- }
- }
+ $aGroupedSearches = $aReverseGroupedSearches;
ksort($aGroupedSearches);
-
- $iSearchCount = 0;
- $aSearches = array();
- foreach($aGroupedSearches as $iScore => $aNewSearches)
- {
- $iSearchCount += sizeof($aNewSearches);
- $aSearches = array_merge($aSearches, $aNewSearches);
- if ($iSearchCount > 50) break;
- }
-
- //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
-
}
-
}
else
{
@@ -1054,30 +1084,7 @@
if (CONST_Debug) var_Dump($aGroupedSearches);
- if ($this->bReverseInPlan)
- {
- $aCopyGroupedSearches = $aGroupedSearches;
- foreach($aCopyGroupedSearches as $iGroup => $aSearches)
- {
- foreach($aSearches as $iSearch => $aSearch)
- {
- if (sizeof($aSearch['aAddress']))
- {
- $iReverseItem = array_pop($aSearch['aAddress']);
- if (isset($aPossibleMainWordIDs[$iReverseItem]))
- {
- $aSearch['aAddress'] = array_merge($aSearch['aAddress'], $aSearch['aName']);
- $aSearch['aName'] = array($iReverseItem);
- $aGroupedSearches[$iGroup][] = $aSearch;
- }
- //$aReverseSearch['aName'][$iReverseItem] = $iReverseItem;
- //$aGroupedSearches[$iGroup][] = $aReverseSearch;
- }
- }
- }
- }
-
- if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0)
+ if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0)
{
$aCopyGroupedSearches = $aGroupedSearches;
foreach($aCopyGroupedSearches as $iGroup => $aSearches)
@@ -1156,6 +1163,8 @@
{
$sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4";
if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
+ if ($bBoundingBoxSearch)
+ $sSQL .= " and _st_intersects($this->sViewboxSmallSQL, geometry)";
$sSQL .= " order by st_area(geometry) desc limit 1";
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $this->oDB->getCol($sSQL);
@@ -1221,6 +1230,12 @@
$aTerms = array();
$aOrder = array();
+ if ($aSearch['sHouseNumber'] && sizeof($aSearch['aAddress']))
+ {
+ $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M';
+ $aOrder[] = "exists(select place_id from placex where parent_place_id = search_name.place_id and transliteration(housenumber) ~* E'".$sHouseNumberRegex."' limit 1) desc";
+ }
+
// 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'],",")."]";
@@ -1301,7 +1316,7 @@
$sSQL .= " where ".join(' and ',$aTerms);
$sSQL .= " order by ".join(', ',$aOrder);
if ($aSearch['sHouseNumber'] || $aSearch['sClass'])
- $sSQL .= " limit 50";
+ $sSQL .= " limit 20";
elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass'])
$sSQL .= " limit 1";
else
@@ -1566,9 +1581,11 @@
$aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery);
foreach($aRecheckWords as $i => $sWord)
{
- if (!$sWord) unset($aRecheckWords[$i]);
+ if (!preg_match('/\pL/', $sWord)) unset($aRecheckWords[$i]);
}
+ if (CONST_Debug) { echo 'Recheck words:<\i>'; var_dump($aRecheckWords); }
+
foreach($aSearchResults as $iResNum => $aResult)
{
if (CONST_Search_AreaPolygons)
@@ -1576,13 +1593,13 @@
// Get the bounding box and outline polygon
$sSQL = "select place_id,0 as numfeatures,st_area(geometry) as area,";
$sSQL .= "ST_Y(centroid) as centrelat,ST_X(centroid) as centrelon,";
- $sSQL .= "ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),4)) as minlat,ST_Y(ST_PointN(ST_ExteriorRing(Box2D(geometry)),2)) as maxlat,";
- $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon";
+ $sSQL .= "ST_YMin(geometry) as minlat,ST_YMax(geometry) as maxlat,";
+ $sSQL .= "ST_XMin(geometry) as minlon,ST_XMax(geometry) as maxlon";
if ($this->bIncludePolygonAsGeoJSON) $sSQL .= ",ST_AsGeoJSON(geometry) as asgeojson";
if ($this->bIncludePolygonAsKML) $sSQL .= ",ST_AsKML(geometry) as askml";
if ($this->bIncludePolygonAsSVG) $sSQL .= ",ST_AsSVG(geometry) as assvg";
if ($this->bIncludePolygonAsText || $this->bIncludePolygonAsPoints) $sSQL .= ",ST_AsText(geometry) as astext";
- $sSQL .= " from placex where place_id = ".$aResult['place_id'].' and st_geometrytype(Box2D(geometry)) = \'ST_Polygon\'';
+ $sSQL .= " from placex where place_id = ".$aResult['place_id'];
$aPointPolygon = $this->oDB->getRow($sSQL);
if (PEAR::IsError($aPointPolygon))
{
@@ -1604,7 +1621,7 @@
if ($this->bIncludePolygonAsPoints)
{
- // Translate geometary string to point array
+ // Translate geometry string to point array
if (preg_match('#POLYGON\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
{
preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
@@ -1727,7 +1744,7 @@
if (stripos($sAddress, $sWord)!==false)
{
$iCountWords++;
- if (preg_match("/(^|,)\s*$sWord\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
+ if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1;
}
}
@@ -1764,7 +1781,6 @@
$bFirst = true;
foreach($aToFilter as $iResNum => $aResult)
{
- if ($aResult['type'] == 'adminitrative') $aResult['type'] = 'administrative';
$this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
if ($bFirst)
{