- $sSQL .= " union ";
- $sSQL .= "select 'L' as osm_type,place_id as osm_id,'place' as class,'house' as type,null as admin_level,30 as rank_search,30 as rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id,'us' as country_code,";
- $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, ";
- $sSQL .= "null as extra_place ";
- $sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
- $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
- $sSQL .= "group by place_id";
- if (!$this->bDeDupe) $sSQL .= ",place_id";
- $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) ";
- }
-
- $sSQL .= " order by importance desc";
- if (CONST_Debug) { echo "<hr>"; var_dump($sSQL); }
- $aSearchResults = $this->oDB->getAll($sSQL);
-
- if (PEAR::IsError($aSearchResults))
- {
- failInternalError("Could not get details for place.", $sSQL, $aSearchResults);
- }
-
- 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 => $sToken)
- {
- //echo "<br><b>$sToken</b>";
- $aNewWordsetSearches = array();
-
- foreach($aWordsetSearches as $aCurrentSearch)
- {
- //echo "<i>";
- //var_dump($aCurrentSearch);
- //echo "</i>";
-
- // 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('<hr>',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 $aGroupedSearches;
-
- }
-
- /* Perform the actual query lookup.
-
- Returns an ordered list of results, each with the following fields:
- osm_type: type of corresponding OSM object
- N - node
- W - way
- R - relation
- P - postcode (internally computed)
- osm_id: id of corresponding OSM object
- class: general object class (corresponds to tag key of primary OSM tag)
- type: subclass of object (corresponds to tag value of primary OSM tag)
- admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
- rank_search: rank in search hierarchy
- (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
- rank_address: rank in address hierarchy (determines orer in address)
- place_id: internal key (may differ between different instances)
- country_code: ISO country code
- langaddress: localized full address
- placename: localized name of object
- ref: content of ref tag (if available)
- lon: longitude
- lat: latitude
- importance: importance of place based on Wikipedia link count
- addressimportance: cumulated importance of address elements
- extra_place: type of place (for admin boundaries, if there is a place tag)
- aBoundingBox: bounding Box
- label: short description of the object class/type (English only)
- name: full name (currently the same as langaddress)
- foundorder: secondary ordering for places with same importance
- */
- function lookup()
- {
- if (!$this->sQuery && !$this->aStructuredQuery) return false;
-
- $sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$this->aLangPrefOrder))."]";
- $sCountryCodesSQL = false;
- if ($this->aCountryCodes && sizeof($this->aCountryCodes))
- {
- $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes));
- }
-
- $sQuery = $this->sQuery;
-
- // Conflicts between US state abreviations and various words for 'the' in different languages
- if (isset($this->aLangPrefOrder['name:en']))
- {
- $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/','\1illinois\2', $sQuery);
- $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/','\1alabama\2', $sQuery);
- $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/','\1louisiana\2', $sQuery);
- }
-
- // View Box SQL
- $sViewboxCentreSQL = false;
- $bBoundingBoxSearch = false;
- if ($this->aViewBox)
- {
- $fHeight = $this->aViewBox[0]-$this->aViewBox[2];
- $fWidth = $this->aViewBox[1]-$this->aViewBox[3];
- $aBigViewBox[0] = $this->aViewBox[0] + $fHeight;
- $aBigViewBox[2] = $this->aViewBox[2] - $fHeight;
- $aBigViewBox[1] = $this->aViewBox[1] + $fWidth;
- $aBigViewBox[3] = $this->aViewBox[3] - $fWidth;
-
- $this->sViewboxSmallSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$this->aViewBox[0].",".(float)$this->aViewBox[1]."),ST_Point(".(float)$this->aViewBox[2].",".(float)$this->aViewBox[3].")),4326)";
- $this->sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$aBigViewBox[0].",".(float)$aBigViewBox[1]."),ST_Point(".(float)$aBigViewBox[2].",".(float)$aBigViewBox[3].")),4326)";
- $bBoundingBoxSearch = $this->bBoundedSearch;
- }
-
- // Route SQL
- if ($this->aRoutePoints)
- {
- $sViewboxCentreSQL = "ST_SetSRID('LINESTRING(";
- $bFirst = true;
- foreach($this->aRoutePoints as $aPoint)
- {
- if (!$bFirst) $sViewboxCentreSQL .= ",";
- $sViewboxCentreSQL .= $aPoint[0].' '.$aPoint[1];
- $bFirst = false;
- }
- $sViewboxCentreSQL .= ")'::geometry,4326)";
-
- $sSQL = "select st_buffer(".$sViewboxCentreSQL.",".(float)($_GET['routewidth']/69).")";
- $this->sViewboxSmallSQL = $this->oDB->getOne($sSQL);
- if (PEAR::isError($this->sViewboxSmallSQL))
- {
- failInternalError("Could not get small viewbox.", $sSQL, $this->sViewboxSmallSQL);
- }
- $this->sViewboxSmallSQL = "'".$this->sViewboxSmallSQL."'::geometry";
-
- $sSQL = "select st_buffer(".$sViewboxCentreSQL.",".(float)($_GET['routewidth']/30).")";
- $this->sViewboxLargeSQL = $this->oDB->getOne($sSQL);
- if (PEAR::isError($this->sViewboxLargeSQL))
- {
- failInternalError("Could not get large viewbox.", $sSQL, $this->sViewboxLargeSQL);
- }
- $this->sViewboxLargeSQL = "'".$this->sViewboxLargeSQL."'::geometry";
- $bBoundingBoxSearch = $this->bBoundedSearch;
- }
-
- // Do we have anything that looks like a lat/lon pair?
- if ( $aLooksLike = looksLikeLatLonPair($sQuery) ){
- $this->setNearPoint(array($aLooksLike['lat'], $aLooksLike['lon']));
- $sQuery = $aLooksLike['query'];
- }
-
- $aSearchResults = array();
- if ($sQuery || $this->aStructuredQuery)
- {
- // Start with a blank search
- $aSearches = array(
- array('iSearchRank' => 0,
- 'iNamePhrase' => -1,
- 'sCountryCode' => false,
- 'aName' => array(),
- 'aAddress' => array(),
- 'aFullNameAddress' => array(),
- 'aNameNonSearch' => array(),
- 'aAddressNonSearch' => array(),
- 'sOperator' => '',
- 'aFeatureName' => array(),
- 'sClass' => '',
- 'sType' => '',
- 'sHouseNumber' => '',
- 'fLat' => '',
- 'fLon' => '',
- 'fRadius' => ''
- )
- );
-
- // Do we have a radius search?
- $sNearPointSQL = false;
- if ($this->aNearPoint)
- {
- $sNearPointSQL = "ST_SetSRID(ST_Point(".(float)$this->aNearPoint[1].",".(float)$this->aNearPoint[0]."),4326)";
- $aSearches[0]['fLat'] = (float)$this->aNearPoint[0];
- $aSearches[0]['fLon'] = (float)$this->aNearPoint[1];
- $aSearches[0]['fRadius'] = (float)$this->aNearPoint[2];
- }
-
- // Any 'special' terms in the search?
- $bSpecialTerms = false;
- preg_match_all('/\\[(.*)=(.*)\\]/', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
- $aSpecialTerms = array();
- foreach($aSpecialTermsRaw as $aSpecialTerm)
- {
- $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
- $aSpecialTerms[strtolower($aSpecialTerm[1])] = $aSpecialTerm[2];
- }
-
- preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER);
- $aSpecialTerms = array();
- if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity'])
- {
- $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']);
- unset($this->aStructuredQuery['amenity']);
- }
- foreach($aSpecialTermsRaw as $aSpecialTerm)
- {
- $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
- $sToken = $this->oDB->getOne("select make_standard_name('".$aSpecialTerm[1]."') as string");
- $sSQL = 'select * from (select word_id,word_token, word, class, type, country_code, operator';
- $sSQL .= ' from word where word_token in (\' '.$sToken.'\')) as x where (class is not null and class not in (\'place\')) or country_code is not null';
- if (CONST_Debug) var_Dump($sSQL);
- $aSearchWords = $this->oDB->getAll($sSQL);
- $aNewSearches = array();
- foreach($aSearches as $aSearch)
- {
- foreach($aSearchWords as $aSearchTerm)
- {
- $aNewSearch = $aSearch;
- if ($aSearchTerm['country_code'])
- {
- $aNewSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
- $aNewSearches[] = $aNewSearch;
- $bSpecialTerms = true;
- }
- if ($aSearchTerm['class'])
- {
- $aNewSearch['sClass'] = $aSearchTerm['class'];
- $aNewSearch['sType'] = $aSearchTerm['type'];
- $aNewSearches[] = $aNewSearch;
- $bSpecialTerms = true;
- }
- }
- }
- $aSearches = $aNewSearches;
- }
-
- // Split query into phrases
- // Commas are used to reduce the search space by indicating where phrases split
- if ($this->aStructuredQuery)
- {
- $aPhrases = $this->aStructuredQuery;
- $bStructuredPhrases = true;
- }
- else
- {
- $aPhrases = explode(',',$sQuery);
- $bStructuredPhrases = false;
- }
-
- // Convert each phrase to standard form
- // Create a list of standard words
- // Get all 'sets' of words
- // Generate a complete list of all
- $aTokens = array();
- foreach($aPhrases as $iPhrase => $sPhrase)
- {
- $aPhrase = $this->oDB->getRow("select make_standard_name('".pg_escape_string($sPhrase)."') as string");
- if (PEAR::isError($aPhrase))
- {
- userError("Illegal query string (not an UTF-8 string): ".$sPhrase);
- if (CONST_Debug) var_dump($aPhrase);
- exit;
- }
- if (trim($aPhrase['string']))
- {
- $aPhrases[$iPhrase] = $aPhrase;
- $aPhrases[$iPhrase]['words'] = explode(' ',$aPhrases[$iPhrase]['string']);
- $aPhrases[$iPhrase]['wordsets'] = getWordSets($aPhrases[$iPhrase]['words'], 0);
- $aTokens = array_merge($aTokens, getTokensFromSets($aPhrases[$iPhrase]['wordsets']));
- }
- else
- {
- unset($aPhrases[$iPhrase]);
- }
- }
-
- // Reindex phrases - we make assumptions later on that they are numerically keyed in order
- $aPhraseTypes = array_keys($aPhrases);
- $aPhrases = array_values($aPhrases);
-
- if (sizeof($aTokens))
- {
- // Check which tokens we have, get the ID numbers
- $sSQL = 'select word_id,word_token, word, class, type, country_code, operator, search_name_count';
- $sSQL .= ' from word where word_token in ('.join(',',array_map("getDBQuoted",$aTokens)).')';
-
- if (CONST_Debug) var_Dump($sSQL);
-
- $aValidTokens = array();
- if (sizeof($aTokens)) $aDatabaseWords = $this->oDB->getAll($sSQL);
- else $aDatabaseWords = array();
- if (PEAR::IsError($aDatabaseWords))
- {
- failInternalError("Could not get word tokens.", $sSQL, $aDatabaseWords);
- }
- $aPossibleMainWordIDs = array();
- $aWordFrequencyScores = array();
- foreach($aDatabaseWords as $aToken)
- {
- // Very special case - require 2 letter country param to match the country code found
- if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country'])
- && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code'])
- {
- continue;
- }
-
- if (isset($aValidTokens[$aToken['word_token']]))
- {
- $aValidTokens[$aToken['word_token']][] = $aToken;
- }
- else
- {
- $aValidTokens[$aToken['word_token']] = array($aToken);
- }
- if (!$aToken['class'] && !$aToken['country_code']) $aPossibleMainWordIDs[$aToken['word_id']] = 1;
- $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;
- }
- if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);
-
- // Try and calculate GB postcodes we might be missing
- foreach($aTokens as $sToken)
- {
- // Source of gb postcodes is now definitive - always use
- if (preg_match('/^([A-Z][A-Z]?[0-9][0-9A-Z]? ?[0-9])([A-Z][A-Z])$/', strtoupper(trim($sToken)), $aData))
- {
- if (substr($aData[1],-2,1) != ' ')
- {
- $aData[0] = substr($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], $this->oDB);
- if ($aGBPostcodeLocation)
- {
- $aValidTokens[$sToken] = $aGBPostcodeLocation;
- }
- }
- // US ZIP+4 codes - if there is no token,
- // merge in the 5-digit ZIP code
- else if (!isset($aValidTokens[$sToken]) && preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData))
- {
- if (isset($aValidTokens[$aData[1]]))
- {
- foreach($aValidTokens[$aData[1]] as $aToken)
- {
- if (!$aToken['class'])
- {
- if (isset($aValidTokens[$sToken]))
- {
- $aValidTokens[$sToken][] = $aToken;
- }
- else
- {
- $aValidTokens[$sToken] = array($aToken);
- }
- }
- }
- }
- }
- }
-
- foreach($aTokens as $sToken)
- {
- // Unknown single word token with a number - assume it is a house number
- if (!isset($aValidTokens[' '.$sToken]) && strpos($sToken,' ') === false && preg_match('/[0-9]/', $sToken))
- {
- $aValidTokens[' '.$sToken] = array(array('class'=>'place','type'=>'house'));
- }
- }
-
- // Any words that have failed completely?
- // TODO: suggestions
-
- // Start the search process
- $aResultPlaceIDs = array();
-
- $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
-
- if ($this->bReverseInPlan)
- {
- // 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)
- {
- $aFinalPhrase = end($aPhrases);
- $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0);
- }
- $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false);
-
- foreach($aGroupedSearches as $aSearches)
- {
- foreach($aSearches as $aSearch)
- {
- if ($aSearch['iSearchRank'] < $this->iMaxRank)
- {
- if (!isset($aReverseGroupedSearches[$aSearch['iSearchRank']])) $aReverseGroupedSearches[$aSearch['iSearchRank']] = array();
- $aReverseGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
- }
-
- }
- }
-
- $aGroupedSearches = $aReverseGroupedSearches;
- ksort($aGroupedSearches);
- }
- }
- else
- {
- // Re-group the searches by their score, junk anything over 20 as just not worth trying
- $aGroupedSearches = array();
- foreach($aSearches as $aSearch)
- {
- if ($aSearch['iSearchRank'] < $this->iMaxRank)
- {
- if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
- $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
- }
- }
- ksort($aGroupedSearches);
- }
-
- if (CONST_Debug) var_Dump($aGroupedSearches);
-
- if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0)
- {
- $aCopyGroupedSearches = $aGroupedSearches;
- foreach($aCopyGroupedSearches as $iGroup => $aSearches)
- {
- foreach($aSearches as $iSearch => $aSearch)
- {
- $aReductionsList = array($aSearch['aAddress']);
- $iSearchRank = $aSearch['iSearchRank'];
- while(sizeof($aReductionsList) > 0)
- {
- $iSearchRank += 5;
- if ($iSearchRank > iMaxRank) break 3;
- $aNewReductionsList = array();
- foreach($aReductionsList as $aReductionsWordList)
- {
- for ($iReductionWord = 0; $iReductionWord < sizeof($aReductionsWordList); $iReductionWord++)
- {
- $aReductionsWordListResult = array_merge(array_slice($aReductionsWordList, 0, $iReductionWord), array_slice($aReductionsWordList, $iReductionWord+1));
- $aReverseSearch = $aSearch;
- $aSearch['aAddress'] = $aReductionsWordListResult;
- $aSearch['iSearchRank'] = $iSearchRank;
- $aGroupedSearches[$iSearchRank][] = $aReverseSearch;
- if (sizeof($aReductionsWordListResult) > 0)
- {
- $aNewReductionsList[] = $aReductionsWordListResult;
- }
- }
- }
- $aReductionsList = $aNewReductionsList;
- }
- }
- }
- ksort($aGroupedSearches);
- }
-
- // Filter out duplicate searches
- $aSearchHash = array();
- foreach($aGroupedSearches as $iGroup => $aSearches)
- {
- foreach($aSearches as $iSearch => $aSearch)
- {
- $sHash = serialize($aSearch);
- if (isset($aSearchHash[$sHash]))
- {
- unset($aGroupedSearches[$iGroup][$iSearch]);
- if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
- }
- else
- {
- $aSearchHash[$sHash] = 1;
- }
- }
- }
-
- if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);
-
- $iGroupLoop = 0;
- $iQueryLoop = 0;
- foreach($aGroupedSearches as $iGroupedRank => $aSearches)
- {
- $iGroupLoop++;
- foreach($aSearches as $aSearch)
- {
- $iQueryLoop++;
-
- if (CONST_Debug) { echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>"; }
- if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
-
- // No location term?
- if (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && !$aSearch['fLon'])
- {
- if ($aSearch['sCountryCode'] && !$aSearch['sClass'] && !$aSearch['sHouseNumber'])
- {
- // Just looking for a country by code - look it up
- if (4 >= $this->iMinAddressRank && 4 <= $this->iMaxAddressRank)
- {
- $sSQL = "select place_id from placex where calculated_country_code='".$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
- {
- $aPlaceIDs = array();
- }
- }
- else
- {
- if (!$bBoundingBoxSearch && !$aSearch['fLon']) continue;
- if (!$aSearch['sClass']) continue;
- $sSQL = "select count(*) from pg_tables where tablename = 'place_classtype_".$aSearch['sClass']."_".$aSearch['sType']."'";
- if ($this->oDB->getOne($sSQL))
- {
- $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct";
- if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)";
- $sSQL .= " where st_contains($this->sViewboxSmallSQL, ct.centroid)";
- if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
- if (sizeof($this->aExcludePlaceIDs))
- {
- $sSQL .= " and place_id not in (".join(',',$this->aExcludePlaceIDs).")";
- }
- if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
- $sSQL .= " limit $this->iLimit";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $this->oDB->getCol($sSQL);
-
- // 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.
- // 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)";
- $sSQL .= " where st_contains($this->sViewboxLargeSQL, ct.centroid)";
- if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
- if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, ct.centroid) asc";
- $sSQL .= " limit $this->iLimit";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $this->oDB->getCol($sSQL);
- }
- }
- else
- {
- $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'";
- $sSQL .= " and st_contains($this->sViewboxSmallSQL, geometry) and linked_place_id is null";
- if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)";
- if ($sViewboxCentreSQL) $sSQL .= " order by st_distance($sViewboxCentreSQL, centroid) asc";
- $sSQL .= " limit $this->iLimit";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $this->oDB->getCol($sSQL);
- }
- }
- }
- else
- {
- $aPlaceIDs = array();
-
- // First we need a position, either aName or fLat or both
- $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'],",")."]";
- if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";
- if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress'])
- {
- // For infrequent name terms disable index usage for address
- if (CONST_Search_NameOnlySearchFrequencyThreshold &&
- sizeof($aSearch['aName']) == 1 &&
- $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
- {
- $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
- }
- else
- {
- $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
- if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";
- }
- }
- if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
- 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'].")";
- $aOrder[] = "ST_Distance(centroid, ST_SetSRID(ST_Point(".$aSearch['fLon'].",".$aSearch['fLat']."),4326)) ASC";
- }
- if (sizeof($this->aExcludePlaceIDs))
- {
- $aTerms[] = "place_id not in (".join(',',$this->aExcludePlaceIDs).")";
- }
- if ($sCountryCodesSQL)
- {
- $aTerms[] = "country_code in ($sCountryCodesSQL)";
- }
-
- if ($bBoundingBoxSearch) $aTerms[] = "centroid && $this->sViewboxSmallSQL";
- if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc";
-
- if ($aSearch['sHouseNumber'])
- {
- $sImportanceSQL = '- abs(26 - address_rank) + 3';
- }
- else
- {
- $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)';
- }
- if ($this->sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($this->sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END";
- if ($this->sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($this->sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END";
-
- $aOrder[] = "$sImportanceSQL DESC";
- if (sizeof($aSearch['aFullNameAddress']))
- {
- $sExactMatchSQL = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) as exactmatch';
- $aOrder[] = 'exactmatch DESC';
- } else {
- $sExactMatchSQL = '0::int as exactmatch';
- }
-
- if (sizeof($aTerms))
- {
- $sSQL = "select place_id, ";
- $sSQL .= $sExactMatchSQL;
- $sSQL .= " from search_name";
- $sSQL .= " where ".join(' and ',$aTerms);
- $sSQL .= " order by ".join(', ',$aOrder);
- if ($aSearch['sHouseNumber'] || $aSearch['sClass'])
- $sSQL .= " limit 20";
- elseif (!sizeof($aSearch['aName']) && !sizeof($aSearch['aAddress']) && $aSearch['sClass'])
- $sSQL .= " limit 1";
- else
- $sSQL .= " limit ".$this->iLimit;
-
- if (CONST_Debug) { var_dump($sSQL); }
- $aViewBoxPlaceIDs = $this->oDB->getAll($sSQL);
- if (PEAR::IsError($aViewBoxPlaceIDs))
- {
- failInternalError("Could not get places for search terms.", $sSQL, $aViewBoxPlaceIDs);
- }
- //var_dump($aViewBoxPlaceIDs);
- // Did we have an viewbox matches?
- $aPlaceIDs = array();
- $bViewBoxMatch = false;
- foreach($aViewBoxPlaceIDs as $aViewBoxRow)
- {
- //if ($bViewBoxMatch == 1 && $aViewBoxRow['in_small'] == 'f') break;
- //if ($bViewBoxMatch == 2 && $aViewBoxRow['in_large'] == 'f') break;
- //if ($aViewBoxRow['in_small'] == 't') $bViewBoxMatch = 1;
- //else if ($aViewBoxRow['in_large'] == 't') $bViewBoxMatch = 2;
- $aPlaceIDs[] = $aViewBoxRow['place_id'];
- $this->exactMatchCache[$aViewBoxRow['place_id']] = $aViewBoxRow['exactmatch'];
- }
- }
- //var_Dump($aPlaceIDs);
- //exit;
-
- //now search for housenumber, if housenumber provided
- if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs))
- {
- $aRoadPlaceIDs = $aPlaceIDs;
- $sPlaceIDs = join(',',$aPlaceIDs);
-
- // Now they are indexed look for a house attached to a street we found
- $sHouseNumberRegex = '\\\\m'.$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).")";
- }
- $sSQL .= " limit $this->iLimit";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $this->oDB->getCol($sSQL);
-
- // If nothing found try the aux fallback table
- if (!sizeof($aPlaceIDs))
- {
- $sSQL = "select place_id from location_property_aux where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
- if (sizeof($this->aExcludePlaceIDs))
- {
- $sSQL .= " and place_id not in (".join(',',$this->aExcludePlaceIDs).")";
- }
- //$sSQL .= " limit $this->iLimit";
- if (CONST_Debug) var_dump($sSQL);
- $aPlaceIDs = $this->oDB->getCol($sSQL);
- }
- //if nothing was found in placex or location_property_aux, then search in Tiger data for this housenumber(location_property_tiger_line)
- if (!sizeof($aPlaceIDs))
- {
- //$sSQL = "select place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
- //new query for lines, not housenumbers anymore
- $this->searchedHousenumber = intval($aSearch['sHouseNumber']);
- if($this->searchedHousenumber%2==0){
- //if housenumber is even, look for housenumber in streets with interpolationtype even or all
- $sSQL = "select distinct place_id from location_property_tiger_line where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='even' or interpolationtype='all') and ".$this->searchedHousenumber.">=startnumber and ".$this->searchedHousenumber."<=endnumber";
- }else{
- //look for housenumber in streets with interpolationtype odd or all
- $sSQL = "select distinct place_id from location_property_tiger_line where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='odd' or interpolationtype='all') and ".$this->searchedHousenumber.">=startnumber and ".$this->searchedHousenumber."<=endnumber";
+ }
+ }
+ }
+
+ public function setQueryFromParams($oParams)
+ {
+ // Search query
+ $sQuery = $oParams->getString('q');
+ if (!$sQuery) {
+ $this->setStructuredQuery(
+ $oParams->getString('amenity'),
+ $oParams->getString('street'),
+ $oParams->getString('city'),
+ $oParams->getString('county'),
+ $oParams->getString('state'),
+ $oParams->getString('country'),
+ $oParams->getString('postalcode')
+ );
+ $this->setReverseInPlan(false);
+ } else {
+ $this->setQuery($sQuery);
+ }
+ }
+
+ public function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
+ {
+ $sValue = trim($sValue);
+ if (!$sValue) return false;
+ $this->aStructuredQuery[$sKey] = $sValue;
+ if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30) {
+ $this->iMinAddressRank = $iNewMinAddressRank;
+ $this->iMaxAddressRank = $iNewMaxAddressRank;
+ }
+ if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
+ return true;
+ }
+
+ public function setStructuredQuery($sAmenity = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
+ {
+ $this->sQuery = false;
+
+ // Reset
+ $this->iMinAddressRank = 0;
+ $this->iMaxAddressRank = 30;
+ $this->aAddressRankList = array();
+
+ $this->aStructuredQuery = array();
+ $this->sAllowedTypesSQLList = false;
+
+ $this->loadStructuredAddressElement($sAmenity, 'amenity', 26, 30, false);
+ $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
+ $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
+ $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
+ $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
+ $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
+ $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
+
+ if (sizeof($this->aStructuredQuery) > 0) {
+ $this->sQuery = join(', ', $this->aStructuredQuery);
+ if ($this->iMaxAddressRank < 30) {
+ $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
+ }
+ }
+ }
+
+ public function fallbackStructuredQuery()
+ {
+ if (!$this->aStructuredQuery) return false;
+
+ $aParams = $this->aStructuredQuery;
+
+ if (sizeof($aParams) == 1) return false;
+
+ $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
+
+ foreach ($aOrderToFallback as $sType) {
+ if (isset($aParams[$sType])) {
+ unset($aParams[$sType]);
+ $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']);
+ return true;
+ }
+ }
+
+ return false;
+ }
+
+ public function getDetails($aPlaceIDs)
+ {
+ //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1
+ if (sizeof($aPlaceIDs) == 0) return array();
+
+ $sLanguagePrefArraySQL = getArraySQL(
+ array_map("getDBQuoted", $this->aLangPrefOrder)
+ );
+
+ // Get the details for display (is this a redundant extra step?)
+ $sPlaceIDs = join(',', array_keys($aPlaceIDs));
+
+ $sImportanceSQL = '';
+ $sImportanceSQLGeom = '';
+ if ($this->sViewboxSmallSQL) {
+ $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+ $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxSmallSQL, geometry) THEN 1 ELSE 0.75 END * ";
+ }
+ if ($this->sViewboxLargeSQL) {
+ $sImportanceSQL .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, ST_Collect(centroid)) THEN 1 ELSE 0.75 END * ";
+ $sImportanceSQLGeom .= " CASE WHEN ST_Contains($this->sViewboxLargeSQL, geometry) THEN 1 ELSE 0.75 END * ";
+ }
+
+ $sSQL = "SELECT ";
+ $sSQL .= " osm_type,";
+ $sSQL .= " osm_id,";
+ $sSQL .= " class,";
+ $sSQL .= " type,";
+ $sSQL .= " admin_level,";
+ $sSQL .= " rank_search,";
+ $sSQL .= " rank_address,";
+ $sSQL .= " min(place_id) AS place_id, ";
+ $sSQL .= " min(parent_place_id) AS parent_place_id, ";
+ $sSQL .= " country_code, ";
+ $sSQL .= " get_address_by_language(place_id, -1, $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, ";
+ $sSQL .= " avg(ST_Y(centroid)) AS lat, ";
+ $sSQL .= " ".$sImportanceSQL."COALESCE(importance,0.75-(rank_search::float/40)) AS importance, ";
+ $sSQL .= " ( ";
+ $sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
+ $sSQL .= " FROM ";
+ $sSQL .= " place_addressline s, ";
+ $sSQL .= " placex p";
+ $sSQL .= " WHERE s.place_id = min(CASE WHEN placex.rank_search < 28 THEN placex.place_id ELSE placex.parent_place_id END)";
+ $sSQL .= " AND p.place_id = s.address_place_id ";
+ $sSQL .= " AND s.isaddress ";
+ $sSQL .= " AND p.importance is not null ";
+ $sSQL .= " ) AS addressimportance, ";
+ $sSQL .= " (extratags->'place') AS extra_place ";
+ $sSQL .= " FROM placex";
+ $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
+ $sSQL .= " AND (";
+ $sSQL .= " placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
+ if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) {
+ $sSQL .= " OR (extratags->'place') = 'city'";
+ }
+ if ($this->aAddressRankList) {
+ $sSQL .= " OR placex.rank_address in (".join(',', $this->aAddressRankList).")";
+ }
+ $sSQL .= " ) ";
+ if ($this->sAllowedTypesSQLList) {
+ $sSQL .= "AND placex.class in $this->sAllowedTypesSQLList ";
+ }
+ $sSQL .= " AND linked_place_id is null ";
+ $sSQL .= " GROUP BY ";
+ $sSQL .= " osm_type, ";
+ $sSQL .= " osm_id, ";
+ $sSQL .= " class, ";
+ $sSQL .= " type, ";
+ $sSQL .= " admin_level, ";
+ $sSQL .= " rank_search, ";
+ $sSQL .= " rank_address, ";
+ $sSQL .= " country_code, ";
+ $sSQL .= " importance, ";
+ if (!$this->bDeDupe) $sSQL .= "place_id,";
+ $sSQL .= " langaddress, ";
+ $sSQL .= " placename, ";
+ $sSQL .= " ref, ";
+ if ($this->bIncludeExtraTags) $sSQL .= "extratags, ";
+ if ($this->bIncludeNameDetails) $sSQL .= "name, ";
+ $sSQL .= " extratags->'place' ";
+
+ // postcode table
+ $sSQL .= "UNION ";
+ $sSQL .= "SELECT";
+ $sSQL .= " 'P' as osm_type,";
+ $sSQL .= " (SELECT osm_id from placex p WHERE p.place_id = lp.parent_place_id) as osm_id,";
+ $sSQL .= " 'place' as class, 'postcode' as type,";
+ $sSQL .= " null as admin_level, rank_search, rank_address,";
+ $sSQL .= " place_id, parent_place_id, country_code,";
+ $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) AS langaddress,";
+ $sSQL .= " postcode as placename,";
+ $sSQL .= " postcode as ref,";
+ if ($this->bIncludeExtraTags) $sSQL .= "null AS extra,";
+ if ($this->bIncludeNameDetails) $sSQL .= "null AS names,";
+ $sSQL .= " ST_x(st_centroid(geometry)) AS lon, ST_y(st_centroid(geometry)) AS lat,";
+ $sSQL .= $sImportanceSQLGeom."(0.75-(rank_search::float/40)) AS importance, ";
+ $sSQL .= " (";
+ $sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
+ $sSQL .= " FROM ";
+ $sSQL .= " place_addressline s, ";
+ $sSQL .= " placex p";
+ $sSQL .= " WHERE s.place_id = lp.parent_place_id";
+ $sSQL .= " AND p.place_id = s.address_place_id ";
+ $sSQL .= " AND s.isaddress";
+ $sSQL .= " AND p.importance is not null";
+ $sSQL .= " ) AS addressimportance, ";
+ $sSQL .= " null AS extra_place ";
+ $sSQL .= "FROM location_postcode lp";
+ $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
+
+ if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) {
+ // only Tiger housenumbers and interpolation lines need to be interpolated, because they are saved as lines
+ // with start- and endnumber, the common osm housenumbers are usually saved as points
+ $sHousenumbers = "";
+ $i = 0;
+ $length = count($aPlaceIDs);
+ foreach ($aPlaceIDs as $placeID => $housenumber) {
+ $i++;
+ $sHousenumbers .= "(".$placeID.", ".$housenumber.")";
+ if ($i<$length) $sHousenumbers .= ", ";
+ }
+
+ if (CONST_Use_US_Tiger_Data) {
+ // Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
+ $sSQL .= " union";
+ $sSQL .= " SELECT ";
+ $sSQL .= " 'T' AS osm_type, ";
+ $sSQL .= " (SELECT osm_id from placex p WHERE p.place_id=min(blub.parent_place_id)) as osm_id, ";
+ $sSQL .= " 'place' AS class, ";
+ $sSQL .= " 'house' AS type, ";
+ $sSQL .= " null AS admin_level, ";
+ $sSQL .= " 30 AS rank_search, ";
+ $sSQL .= " 30 AS rank_address, ";
+ $sSQL .= " min(place_id) AS place_id, ";
+ $sSQL .= " min(parent_place_id) AS parent_place_id, ";
+ $sSQL .= " 'us' AS country_code, ";
+ $sSQL .= " get_address_by_language(place_id, housenumber_for_place, $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, ";
+ $sSQL .= " avg(st_y(centroid)) AS lat,";
+ $sSQL .= " ".$sImportanceSQL."-1.15 AS importance, ";
+ $sSQL .= " (";
+ $sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
+ $sSQL .= " FROM ";
+ $sSQL .= " place_addressline s, ";
+ $sSQL .= " placex p";
+ $sSQL .= " WHERE s.place_id = min(blub.parent_place_id)";
+ $sSQL .= " AND p.place_id = s.address_place_id ";
+ $sSQL .= " AND s.isaddress";
+ $sSQL .= " AND p.importance is not null";
+ $sSQL .= " ) AS addressimportance, ";
+ $sSQL .= " null AS extra_place ";
+ $sSQL .= " FROM (";
+ $sSQL .= " SELECT place_id, "; // interpolate the Tiger housenumbers here
+ $sSQL .= " ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) AS centroid, ";
+ $sSQL .= " parent_place_id, ";
+ $sSQL .= " housenumber_for_place";
+ $sSQL .= " FROM (";
+ $sSQL .= " location_property_tiger ";
+ $sSQL .= " JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)) ";
+ $sSQL .= " WHERE ";
+ $sSQL .= " housenumber_for_place>=0";
+ $sSQL .= " AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
+ $sSQL .= " ) AS blub"; //postgres wants an alias here
+ $sSQL .= " GROUP BY";
+ $sSQL .= " place_id, ";
+ $sSQL .= " housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
+ if (!$this->bDeDupe) $sSQL .= ", place_id ";
+ }
+ // osmline
+ // interpolation line search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
+ $sSQL .= " UNION ";
+ $sSQL .= "SELECT ";
+ $sSQL .= " 'W' AS osm_type, ";
+ $sSQL .= " osm_id, ";
+ $sSQL .= " 'place' AS class, ";
+ $sSQL .= " 'house' AS type, ";
+ $sSQL .= " null AS admin_level, ";
+ $sSQL .= " 30 AS rank_search, ";
+ $sSQL .= " 30 AS rank_address, ";
+ $sSQL .= " min(place_id) as place_id, ";
+ $sSQL .= " min(parent_place_id) AS parent_place_id, ";
+ $sSQL .= " country_code, ";
+ $sSQL .= " get_address_by_language(place_id, housenumber_for_place, $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, ";
+ $sSQL .= " AVG(st_y(centroid)) AS lat, ";
+ $sSQL .= " ".$sImportanceSQL."-0.1 AS importance, "; // slightly smaller than the importance for normal houses with rank 30, which is 0
+ $sSQL .= " (";
+ $sSQL .= " SELECT ";
+ $sSQL .= " MAX(p.importance*(p.rank_address+2)) ";
+ $sSQL .= " FROM";
+ $sSQL .= " place_addressline s, ";
+ $sSQL .= " placex p";
+ $sSQL .= " WHERE s.place_id = min(blub.parent_place_id) ";
+ $sSQL .= " AND p.place_id = s.address_place_id ";
+ $sSQL .= " AND s.isaddress ";
+ $sSQL .= " AND p.importance is not null";
+ $sSQL .= " ) AS addressimportance,";
+ $sSQL .= " null AS extra_place ";
+ $sSQL .= " FROM (";
+ $sSQL .= " SELECT ";
+ $sSQL .= " osm_id, ";
+ $sSQL .= " place_id, ";
+ $sSQL .= " country_code, ";
+ $sSQL .= " CASE "; // interpolate the housenumbers here
+ $sSQL .= " WHEN startnumber != endnumber ";
+ $sSQL .= " THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
+ $sSQL .= " ELSE ST_LineInterpolatePoint(linegeo, 0.5) ";
+ $sSQL .= " END as centroid, ";
+ $sSQL .= " parent_place_id, ";
+ $sSQL .= " housenumber_for_place ";
+ $sSQL .= " FROM (";
+ $sSQL .= " location_property_osmline ";
+ $sSQL .= " JOIN (values ".$sHousenumbers.") AS housenumbers(place_id, housenumber_for_place) USING(place_id)";
+ $sSQL .= " ) ";
+ $sSQL .= " WHERE housenumber_for_place>=0 ";
+ $sSQL .= " AND 30 between $this->iMinAddressRank AND $this->iMaxAddressRank";
+ $sSQL .= " ) as blub"; //postgres wants an alias here
+ $sSQL .= " GROUP BY ";
+ $sSQL .= " osm_id, ";
+ $sSQL .= " place_id, ";
+ $sSQL .= " housenumber_for_place, ";
+ $sSQL .= " country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
+ if (!$this->bDeDupe) $sSQL .= ", place_id ";
+
+ if (CONST_Use_Aux_Location_data) {
+ $sSQL .= " UNION ";
+ $sSQL .= " SELECT ";
+ $sSQL .= " 'L' AS osm_type, ";
+ $sSQL .= " place_id AS osm_id, ";
+ $sSQL .= " 'place' AS class,";
+ $sSQL .= " 'house' AS type, ";
+ $sSQL .= " null AS admin_level, ";
+ $sSQL .= " 0 AS rank_search,";
+ $sSQL .= " 0 AS rank_address, ";
+ $sSQL .= " min(place_id) AS place_id,";
+ $sSQL .= " min(parent_place_id) AS parent_place_id, ";
+ $sSQL .= " 'us' AS country_code, ";
+ $sSQL .= " get_address_by_language(place_id, -1, $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, ";
+ $sSQL .= " avg(ST_Y(centroid)) AS lat, ";
+ $sSQL .= " ".$sImportanceSQL."-1.10 AS importance, ";
+ $sSQL .= " ( ";
+ $sSQL .= " SELECT max(p.importance*(p.rank_address+2))";
+ $sSQL .= " FROM ";
+ $sSQL .= " place_addressline s, ";
+ $sSQL .= " placex p";
+ $sSQL .= " WHERE s.place_id = min(location_property_aux.parent_place_id)";
+ $sSQL .= " AND p.place_id = s.address_place_id ";
+ $sSQL .= " AND s.isaddress";
+ $sSQL .= " AND p.importance is not null";
+ $sSQL .= " ) AS addressimportance, ";
+ $sSQL .= " null AS extra_place ";
+ $sSQL .= " FROM location_property_aux ";
+ $sSQL .= " WHERE place_id in ($sPlaceIDs) ";
+ $sSQL .= " AND 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
+ $sSQL .= " GROUP BY ";
+ $sSQL .= " place_id, ";
+ if (!$this->bDeDupe) $sSQL .= "place_id, ";
+ $sSQL .= " get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
+ }
+ }
+
+ $sSQL .= " order by importance desc";
+ if (CONST_Debug) {
+ echo "<hr>";
+ var_dump($sSQL);
+ }
+ $aSearchResults = chksql(
+ $this->oDB->getAll($sSQL),
+ "Could not get details for place."
+ );
+
+ return $aSearchResults;
+ }
+
+ public function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases, $sNormQuery)
+ {
+ /*
+ 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
+ */
+ $iGlobalRank = 0;
+
+ foreach ($aPhrases as $iPhrase => $aPhrase) {
+ $aNewPhraseSearches = array();
+ if ($bStructuredPhrases) {
+ $sPhraseType = $aPhraseTypes[$iPhrase];
+ } else {
+ $sPhraseType = '';
+ }
+
+ foreach ($aPhrase['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 => $sToken) {
+ //echo "<br><b>$sToken</b>";
+ $aNewWordsetSearches = array();
+
+ foreach ($aWordsetSearches as $oCurrentSearch) {
+ //echo "<i>";
+ //var_dump($oCurrentSearch);
+ //echo "</i>";
+
+ // If the token is valid
+ if (isset($aValidTokens[' '.$sToken])) {
+ foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
+ // Recheck if the original word shows up in the query.
+ $bWordInQuery = false;
+ if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
+ $bWordInQuery = strpos(
+ $sNormQuery,
+ $this->normTerm($aSearchTerm['word'])
+ ) !== false;
+ }
+ $aNewSearches = $oCurrentSearch->extendWithFullTerm(
+ $aSearchTerm,
+ $bWordInQuery,
+ isset($aValidTokens[$sToken])
+ && strpos($sToken, ' ') === false,
+ $sPhraseType,
+ $iToken == 0 && $iPhrase == 0,
+ $iPhrase == 0,
+ $iToken + 1 == sizeof($aWordset)
+ && $iPhrase + 1 == sizeof($aPhrases),
+ $iGlobalRank
+ );
+
+ foreach ($aNewSearches as $oSearch) {
+ if ($oSearch->getRank() < $this->iMaxRank) {
+ $aNewWordsetSearches[] = $oSearch;
+ }
+ }
+ }
+ }
+ // Look for partial matches.
+ // Note that there is no point in adding country terms here
+ // because country is omitted in the address.
+ if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
+ // Allow searching for a word - but at extra cost
+ foreach ($aValidTokens[$sToken] as $aSearchTerm) {
+ $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
+ $aSearchTerm,
+ $bStructuredPhrases,
+ $iPhrase,
+ $aWordFrequencyScores,
+ isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
+ );
+
+ foreach ($aNewSearches as $oSearch) {
+ if ($oSearch->getRank() < $this->iMaxRank) {
+ $aNewWordsetSearches[] = $oSearch;
projects / nominatim.git / blobdiff