X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/79b30a8c0f695746b3ed79e4fb484accd5bf9bec..fa7730d4af7c330d5443c6ae32f5ab73c48bf7a3:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 7de9c8e2..48055ec6 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -1,4 +1,6 @@ bIncludeAddressDetails; } + function getIncludeExtraTags() + { + return $this->bIncludeExtraTags; + } + + function getIncludeNameDetails() + { + return $this->bIncludeNameDetails; + } + function setIncludePolygonAsPoints($b = true) { $this->bIncludePolygonAsPoints = $b; @@ -102,6 +117,11 @@ $this->bIncludePolygonAsSVG = $b; } + function setPolygonSimplificationThreshold($f) + { + $this->fPolygonSimplificationThreshold = $f; + } + function setDeDupe($bDeDupe = true) { $this->bDeDupe = (bool)$bDeDupe; @@ -208,6 +228,11 @@ function loadParamArray($aParams) { if (isset($aParams['addressdetails'])) $this->bIncludeAddressDetails = (bool)$aParams['addressdetails']; + if ((float) CONST_Postgresql_Version > 9.2) + { + if (isset($aParams['extratags'])) $this->bIncludeExtraTags = (bool)$aParams['extratags']; + if (isset($aParams['namedetails'])) $this->bIncludeNameDetails = (bool)$aParams['namedetails']; + } if (isset($aParams['bounded'])) $this->bBoundedSearch = (bool)$aParams['bounded']; if (isset($aParams['dedupe'])) $this->bDeDupe = (bool)$aParams['dedupe']; @@ -222,9 +247,12 @@ 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 @@ -380,6 +408,8 @@ $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "get_name_by_language(name, $sLanguagePrefArraySQL) as placename,"; $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,"; + if ($this->bIncludeExtraTags) $sSQL .= "hstore_to_json(extratags)::text as extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "hstore_to_json(name)::text as names,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; $sSQL .= $sImportanceSQL."coalesce(importance,0.75-(rank_search::float/40)) as importance, "; $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; @@ -396,6 +426,8 @@ $sSQL .= ",langaddress "; $sSQL .= ",placename "; $sSQL .= ",ref "; + if ($this->bIncludeExtraTags) $sSQL .= ",extratags"; + if ($this->bIncludeNameDetails) $sSQL .= ",name"; $sSQL .= ",extratags->'place' "; if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) @@ -405,6 +437,8 @@ $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "null as placename,"; $sSQL .= "null as ref,"; + if ($this->bIncludeExtraTags) $sSQL .= "null as extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "null as names,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; $sSQL .= $sImportanceSQL."-1.15 as importance, "; $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_tiger.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; @@ -419,6 +453,8 @@ $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,"; $sSQL .= "null as placename,"; $sSQL .= "null as ref,"; + if ($this->bIncludeExtraTags) $sSQL .= "null as extra,"; + if ($this->bIncludeNameDetails) $sSQL .= "null as names,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; $sSQL .= $sImportanceSQL."-1.10 as importance, "; $sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; @@ -443,42 +479,318 @@ 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: - osm_type: type of corresponding OSM object + osm_type: type of corresponding OSM object N - node W - way R - relation P - postcode (internally computed) - osm_id: id of corresponding OSM object - class: general object class (corresponds to tag key of primary OSM tag) - type: subclass of object (corresponds to tag value of primary OSM tag) - admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level - rank_search: rank in search hierarchy + osm_id: id of corresponding OSM object + class: general object class (corresponds to tag key of primary OSM tag) + type: subclass of object (corresponds to tag value of primary OSM tag) + admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level + rank_search: rank in search hierarchy (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level) - rank_address: rank in address hierarchy (determines orer in address) - place_id: internal key (may differ between different instances) - country_code: ISO country code - langaddress: localized full address - placename: localized name of object - ref: content of ref tag (if available) - lon: longitude - lat: latitude - importance: importance of place based on Wikipedia link count - addressimportance: cumulated importance of address elements - extra_place: type of place (for admin boundaries, if there is a place tag) - aBoundingBox: bounding Box - label: short description of the object class/type (English only) - name: full name (currently the same as langaddress) - foundorder: secondary ordering for places with same importance + rank_address: rank in address hierarchy (determines orer in address) + place_id: internal key (may differ between different instances) + country_code: ISO country code + langaddress: localized full address + placename: localized name of object + ref: content of ref tag (if available) + lon: longitude + lat: latitude + importance: importance of place based on Wikipedia link count + addressimportance: cumulated importance of address elements + extra_place: type of place (for admin boundaries, if there is a place tag) + aBoundingBox: bounding Box + label: short description of the object class/type (English only) + name: full name (currently the same as langaddress) + foundorder: secondary ordering for places with same importance */ function lookup() { if (!$this->sQuery && !$this->aStructuredQuery) return false; $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$this->aLangPrefOrder))."]"; - $sCountryCodesSQL = false; if ($this->aCountryCodes && sizeof($this->aCountryCodes)) { @@ -554,9 +866,23 @@ { // 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'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'') + array('iSearchRank' => 0, + 'iNamePhrase' => -1, + 'sCountryCode' => false, + 'aName' => array(), + 'aAddress' => array(), + 'aFullNameAddress' => array(), + 'aNameNonSearch' => array(), + 'aAddressNonSearch' => array(), + 'sOperator' => '', + 'aFeatureName' => array(), + 'sClass' => '', + 'sType' => '', + 'sHouseNumber' => '', + 'fLat' => '', + 'fLon' => '', + 'fRadius' => '' + ) ); // Do we have a radius search? @@ -581,10 +907,10 @@ 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 $aSpecialTerm) { @@ -682,8 +1008,8 @@ foreach($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; } @@ -757,287 +1083,38 @@ // 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, $aValidTokens); - } - } else { @@ -1056,30 +1133,7 @@ if (CONST_Debug) var_Dump($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) @@ -1158,8 +1212,8 @@ { $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4"; if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; - if ($bBoundingBoxSearch) - $sSQL .= " and st_overlaps($this->sViewboxSmallSQL, geometry)"; + 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); @@ -1225,6 +1279,12 @@ $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'],",")."]"; @@ -1306,7 +1366,7 @@ $sSQL .= " where ".join(' and ',$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 @@ -1573,84 +1633,26 @@ $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$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) - { - // 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,"; - $sSQL .= "ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),1)) as minlon,ST_X(ST_PointN(ST_ExteriorRing(Box2D(geometry)),3)) as maxlon"; - if ($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\''; - $aPointPolygon = $this->oDB->getRow($sSQL); - if (PEAR::IsError($aPointPolygon)) - { - failInternalError("Could not get outline.", $sSQL, $aPointPolygon); - } + // Default + $fDiameter = getResultDiameter($aResult); - if ($aPointPolygon['place_id']) - { - if ($this->bIncludePolygonAsGeoJSON) $aResult['asgeojson'] = $aPointPolygon['asgeojson']; - if ($this->bIncludePolygonAsKML) $aResult['askml'] = $aPointPolygon['askml']; - if ($this->bIncludePolygonAsSVG) $aResult['assvg'] = $aPointPolygon['assvg']; - if ($this->bIncludePolygonAsText) $aResult['astext'] = $aPointPolygon['astext']; - - if ($aPointPolygon['centrelon'] !== null && $aPointPolygon['centrelat'] !== null ) - { - $aResult['lat'] = $aPointPolygon['centrelat']; - $aResult['lon'] = $aPointPolygon['centrelon']; - } - - if ($this->bIncludePolygonAsPoints) - { - // Translate geometary 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); - } - /* - elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch)) - { - preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER); - } - */ - elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch)) - { - $fRadius = 0.01; - $iSteps = ($fRadius * 40000)^2; - $fStepSize = (2*pi())/$iSteps; - $aPolyPoints = array(); - for($f = 0; $f < 2*pi(); $f += $fStepSize) - { - $aPolyPoints[] = array('',$aMatch[1]+($fRadius*sin($f)),$aMatch[2]+($fRadius*cos($f))); - } - $aPointPolygon['minlat'] = $aPointPolygon['minlat'] - $fRadius; - $aPointPolygon['maxlat'] = $aPointPolygon['maxlat'] + $fRadius; - $aPointPolygon['minlon'] = $aPointPolygon['minlon'] - $fRadius; - $aPointPolygon['maxlon'] = $aPointPolygon['maxlon'] + $fRadius; - } - } + $oPlaceLookup = new PlaceLookup($this->oDB); + $oPlaceLookup->setIncludePolygonAsPoints($this->bIncludePolygonAsPoints); + $oPlaceLookup->setIncludePolygonAsText($this->bIncludePolygonAsText); + $oPlaceLookup->setIncludePolygonAsGeoJSON($this->bIncludePolygonAsGeoJSON); + $oPlaceLookup->setIncludePolygonAsKML($this->bIncludePolygonAsKML); + $oPlaceLookup->setIncludePolygonAsSVG($this->bIncludePolygonAsSVG); + $oPlaceLookup->setPolygonSimplificationThreshold($this->fPolygonSimplificationThreshold); - // Output data suitable for display (points and a bounding box) - if ($this->bIncludePolygonAsPoints && isset($aPolyPoints)) - { - $aResult['aPolyPoints'] = array(); - foreach($aPolyPoints as $aPoint) - { - $aResult['aPolyPoints'][] = array($aPoint[1], $aPoint[2]); - } - } - $aResult['aBoundingBox'] = array($aPointPolygon['minlat'],$aPointPolygon['maxlat'],$aPointPolygon['minlon'],$aPointPolygon['maxlon']); - } - } + $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2); + $aResult = array_merge($aResult, $aOutlineResult); if ($aResult['extra_place'] == 'city') { @@ -1659,47 +1661,6 @@ $aResult['rank_search'] = 16; } - if (!isset($aResult['aBoundingBox'])) - { - // Default - $fDiameter = 0.0001; - - if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defdiameter']) - && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defdiameter']) - { - $fDiameter = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['defzoom']; - } - elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']) - && $aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']) - { - $fDiameter = $aClassType[$aResult['class'].':'.$aResult['type']]['defdiameter']; - } - $fRadius = $fDiameter / 2; - - $iSteps = max(8,min(100,$fRadius * 3.14 * 100000)); - $fStepSize = (2*pi())/$iSteps; - $aPolyPoints = array(); - for($f = 0; $f < 2*pi(); $f += $fStepSize) - { - $aPolyPoints[] = array('',$aResult['lon']+($fRadius*sin($f)),$aResult['lat']+($fRadius*cos($f))); - } - $aPointPolygon['minlat'] = $aResult['lat'] - $fRadius; - $aPointPolygon['maxlat'] = $aResult['lat'] + $fRadius; - $aPointPolygon['minlon'] = $aResult['lon'] - $fRadius; - $aPointPolygon['maxlon'] = $aResult['lon'] + $fRadius; - - // Output data suitable for display (points and a bounding box) - if ($this->bIncludePolygonAsPoints) - { - $aResult['aPolyPoints'] = array(); - foreach($aPolyPoints as $aPoint) - { - $aResult['aPolyPoints'][] = array($aPoint[1], $aPoint[2]); - } - } - $aResult['aBoundingBox'] = array((string)$aPointPolygon['minlat'],(string)$aPointPolygon['maxlat'],(string)$aPointPolygon['minlon'],(string)$aPointPolygon['maxlon']); - } - // Is there an icon set for this type of result? if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['icon']) && $aClassType[$aResult['class'].':'.$aResult['type']]['icon']) @@ -1727,6 +1688,30 @@ } } + if ($this->bIncludeExtraTags) + { + if ($aResult['extra']) + { + $aResult['sExtraTags'] = json_decode($aResult['extra']); + } + else + { + $aResult['sExtraTags'] = (object) array(); + } + } + + if ($this->bIncludeNameDetails) + { + if ($aResult['names']) + { + $aResult['sNameDetails'] = json_decode($aResult['names']); + } + else + { + $aResult['sNameDetails'] = (object) array(); + } + } + // Adjust importance for the number of exact string matches in the result $aResult['importance'] = max(0.001,$aResult['importance']); $iCountWords = 0; @@ -1736,7 +1721,7 @@ if (stripos($sAddress, $sWord)!==false) { $iCountWords++; - if (preg_match("/(^|,)\s*$sWord\s*(,|$)/", $sAddress)) $iCountWords += 0.1; + if (preg_match("/(^|,)\s*".preg_quote($sWord, '/')."\s*(,|$)/", $sAddress)) $iCountWords += 0.1; } } @@ -1755,12 +1740,13 @@ if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['importance']) && $aClassType[$aResult['class'].':'.$aResult['type']]['importance']) { - $aResult['foundorder'] = $aResult['foundorder'] + 0.000001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance']; + $aResult['foundorder'] += 0.0001 * $aClassType[$aResult['class'].':'.$aResult['type']]['importance']; } else { - $aResult['foundorder'] = $aResult['foundorder'] + 0.001; + $aResult['foundorder'] += 0.01; } + if (CONST_Debug) { var_dump($aResult); } $aSearchResults[$iResNum] = $aResult; } uasort($aSearchResults, 'byImportance'); @@ -1773,7 +1759,6 @@ $bFirst = true; foreach($aToFilter as $iResNum => $aResult) { - if ($aResult['type'] == 'adminitrative') $aResult['type'] = 'administrative'; $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id']; if ($bFirst) {