X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/70f0d6ba8bc0e8c4115155b27a36e8a9a037d727..1812c89c31b6e163a14144ae05fff34b97634367:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 32b50580..48b3b1ad 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -443,6 +443,295 @@ 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) $aSearch['iSearchRank'] += 1; + + // Do we have a shortcut id? + if ($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[$sToken]) || strpos($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 $aGroupedSearches; + + } + /* Perform the actual query lookup. Returns an ordered list of results, each with the following fields: @@ -520,7 +809,7 @@ foreach($this->aRoutePoints as $aPoint) { if (!$bFirst) $sViewboxCentreSQL .= ","; - $sViewboxCentreSQL .= $aPoint[1].' '.$aPoint[0]; + $sViewboxCentreSQL .= $aPoint[0].' '.$aPoint[1]; $bFirst = false; } $sViewboxCentreSQL .= ")'::geometry,4326)"; @@ -544,35 +833,9 @@ } // Do we have anything that looks like a lat/lon pair? - if (preg_match('/\\b([NS])[ ]+([0-9]+[0-9.]*)[ ]+([0-9.]+)?[, ]+([EW])[ ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?\\b/', $sQuery, $aData)) - { - $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60); - $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60); - if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1) - { - $this->setNearPoint(array($fQueryLat, $fQueryLon)); - $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); - } - } - elseif (preg_match('/\\b([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([NS])[, ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([EW])\\b/', $sQuery, $aData)) - { - $fQueryLat = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60); - $fQueryLon = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60); - if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1) - { - $this->setNearPoint(array($fQueryLat, $fQueryLon)); - $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); - } - } - elseif (preg_match('/(\\[|^|\\b)(-?[0-9]+[0-9]*\\.[0-9]+)[, ]+(-?[0-9]+[0-9]*\\.[0-9]+)(\\]|$|\\b)/', $sQuery, $aData)) - { - $fQueryLat = $aData[2]; - $fQueryLon = $aData[3]; - if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1) - { - $this->setNearPoint(array($fQueryLat, $fQueryLon)); - $sQuery = trim(str_replace($aData[0], ' ', $sQuery)); - } + if ( $aLooksLike = looksLikeLatLonPair($sQuery) ){ + $this->setNearPoint(array($aLooksLike['lat'], $aLooksLike['lon'])); + $sQuery = $aLooksLike['query']; } $aSearchResults = array(); @@ -735,7 +998,7 @@ { if (substr($aData[1],-2,1) != ' ') { - $aData[0] = substr($aData[0],0,strlen($aData[1]-1)).' '.substr($aData[0],strlen($aData[1]-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); @@ -783,282 +1046,38 @@ // 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']) && strlen($sToken) >= 4) - { - $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[' '.$sToken])) // revert to the token version? - { - foreach($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']] = $aSearchTerm['word_id']; - if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; - } - } - - if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) - { - $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 2; - if (preg_match('#^[0-9]+$#', $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, $aValidTokens); - } - } else { @@ -1077,29 +1096,6 @@ if (CONST_Debug) var_Dump($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) { $aCopyGroupedSearches = $aGroupedSearches; @@ -1179,10 +1175,16 @@ { $sSQL = "select place_id from placex where calculated_country_code='".$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); } + else + { + $aPlaceIDs = array(); + } } else { @@ -1207,7 +1209,8 @@ // If excluded place IDs are given, it is fair to assume that // there have been results in the small box, so no further // expansion in that case. - if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs)) + // Also don't expand if bounded results were requested. + if (!sizeof($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)"; @@ -1239,6 +1242,12 @@ $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'],",")."]"; @@ -1260,7 +1269,21 @@ } } if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'"; - if ($aSearch['sHouseNumber']) $aTerms[] = "address_rank between 16 and 27"; + if ($aSearch['sHouseNumber']) + { + $aTerms[] = "address_rank between 16 and 27"; + } + else + { + if ($this->iMinAddressRank > 0) + { + $aTerms[] = "address_rank >= ".$this->iMinAddressRank; + } + if ($this->iMaxAddressRank < 30) + { + $aTerms[] = "address_rank <= ".$this->iMaxAddressRank; + } + } if ($aSearch['fLon'] && $aSearch['fLat']) { $aTerms[] = "ST_DWithin(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326), ".$aSearch['fRadius'].")"; @@ -1306,7 +1329,7 @@ $sSQL .= " where ".join(' and ',$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 @@ -1341,8 +1364,8 @@ $sPlaceIDs = join(',',$aPlaceIDs); // Now they are indexed look for a house attached to a street we found - $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M'; - $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'"; + $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M'; + $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and transliteration(housenumber) ~* E'".$sHouseNumberRegex."'"; if (sizeof($this->aExcludePlaceIDs)) { $sSQL .= " and place_id not in (".join(',',$this->aExcludePlaceIDs).")"; @@ -1467,7 +1490,7 @@ } if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)"; if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc"; - if ($iOffset) $sSQL .= " offset $iOffset"; + if ($this->iOffset) $sSQL .= " offset $this->iOffset"; $sSQL .= " limit $this->iLimit"; if (CONST_Debug) var_dump($sSQL); $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL)); @@ -1489,7 +1512,7 @@ } if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)"; if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc"; - if ($iOffset) $sSQL .= " offset $iOffset"; + if ($this->iOffset) $sSQL .= " offset $this->iOffset"; $sSQL .= " limit $this->iLimit"; if (CONST_Debug) var_dump($sSQL); $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL)); @@ -1570,7 +1593,7 @@ } $aClassType = getClassTypesWithImportance(); - $aRecheckWords = preg_split('/\b/u',$sQuery); + $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery); foreach($aRecheckWords as $i => $sWord) { if (!$sWord) unset($aRecheckWords[$i]); @@ -1583,13 +1606,13 @@ // 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($aPointPolygon)) { @@ -1611,7 +1634,7 @@ if ($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); @@ -1733,7 +1756,11 @@ $sAddress = $aResult['langaddress']; foreach($aRecheckWords as $i => $sWord) { - if (stripos($sAddress, $sWord)!==false) $iCountWords++; + if (stripos($sAddress, $sWord)!==false) + { + $iCountWords++; + if (preg_match("/(^|,)\s*$sWord\s*(,|$)/", $sAddress)) $iCountWords += 0.1; + } } $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right @@ -1769,7 +1796,6 @@ $bFirst = true; foreach($aToFilter as $iResNum => $aResult) { - if ($aResult['type'] == 'adminitrative') $aResult['type'] = 'administrative'; $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id']; if ($bFirst) { @@ -1797,17 +1823,3 @@ } // end class - -/* - if (isset($_GET['route']) && $_GET['route'] && isset($_GET['routewidth']) && $_GET['routewidth']) - { - $aPoints = explode(',',$_GET['route']); - if (sizeof($aPoints) % 2 != 0) - { - userError("Uneven number of points"); - exit; - } - $sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; - $fPrevCoord = false; - } -*/