<?php
+ require_once(CONST_BasePath.'/lib/PlaceLookup.php');
+
class Geocode
{
protected $oDB;
protected $aLangPrefOrder = array();
protected $bIncludeAddressDetails = false;
+ protected $bIncludeExtraTags = false;
+ protected $bIncludeNameDetails = false;
protected $bIncludePolygonAsPoints = false;
protected $bIncludePolygonAsText = false;
protected $bIncludePolygonAsGeoJSON = false;
protected $bIncludePolygonAsKML = false;
protected $bIncludePolygonAsSVG = false;
+ protected $fPolygonSimplificationThreshold = 0.0;
protected $aExcludePlaceIDs = array();
protected $bDeDupe = true;
return $this->bIncludeAddressDetails;
}
+ function getIncludeExtraTags()
+ {
+ return $this->bIncludeExtraTags;
+ }
+
+ function getIncludeNameDetails()
+ {
+ return $this->bIncludeNameDetails;
+ }
+
function setIncludePolygonAsPoints($b = true)
{
$this->bIncludePolygonAsPoints = $b;
$this->bIncludePolygonAsSVG = $b;
}
+ function setPolygonSimplificationThreshold($f)
+ {
+ $this->fPolygonSimplificationThreshold = $f;
+ }
+
function setDeDupe($bDeDupe = true)
{
$this->bDeDupe = (bool)$bDeDupe;
function loadParamArray($aParams)
{
if (isset($aParams['addressdetails'])) $this->bIncludeAddressDetails = (bool)$aParams['addressdetails'];
+ 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'];
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
$this->loadStructuredAddressElement($sPostalCode, 'postalcode' , 5, 11, array(5, 11));
$this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
- if (sizeof($this->aStructuredQuery) > 0)
+ if (sizeof($this->aStructuredQuery) > 0)
{
$this->sQuery = join(', ', $this->aStructuredQuery);
if ($this->iMaxAddressRank < 30)
function getDetails($aPlaceIDs)
{
+ //$aPlaceIDs is an array with key: placeID and value: tiger-housenumber, if found, else -1
if (sizeof($aPlaceIDs) == 0) return array();
$sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted",$this->aLangPrefOrder))."]";
// Get the details for display (is this a redundant extra step?)
- $sPlaceIDs = join(',',$aPlaceIDs);
+ $sPlaceIDs = join(',', array_keys($aPlaceIDs));
$sImportanceSQL = '';
if ($this->sViewboxSmallSQL) $sImportanceSQL .= " case when ST_Contains($this->sViewboxSmallSQL, ST_Collect(centroid)) 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 * ";
$sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id, min(parent_place_id) as parent_place_id, calculated_country_code as country_code,";
- $sSQL .= "get_address_by_language(place_id, $sLanguagePrefArraySQL) as langaddress,";
+ $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,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, ";
$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)
{
+ //query also location_property_tiger and location_property_aux
+ //Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
+ //only Tiger housenumbers 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 .= ", ";
+ }
+
+ $sSQL .= "union ";
+ $sSQL .= "select 'T' 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, 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, 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(blub.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 (select place_id";
+ //interpolate the Tiger housenumbers here
+ $sSQL .= ", ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) as centroid, parent_place_id, housenumber_for_place ";
+ $sSQL .= "from (location_property_tiger ";
+ $sSQL .= " join (values ".$sHousenumbers.") as housenumbers(place_id, housenumber_for_place) using(place_id)) ";
+ $sSQL .= " where housenumber_for_place>=0 and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here
+ $sSQL .= " group by place_id, housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
+ if (!$this->bDeDupe) $sSQL .= ", place_id ";
+
$sSQL .= " union ";
- $sSQL .= "select 'T' 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,";
- $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, ";
- $sSQL .= "null as extra_place ";
- $sSQL .= "from location_property_tiger where place_id in ($sPlaceIDs) ";
- $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
- $sSQL .= "group by place_id";
- if (!$this->bDeDupe) $sSQL .= ",place_id ";
- $sSQL .= " 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,";
- $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, ";
+ $sSQL .= "select 'L' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, 0 as rank_search, 0 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, -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, 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, $sLanguagePrefArraySQL) ";
+ if (!$this->bDeDupe) $sSQL .= ", place_id";
+ $sSQL .= ", get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
}
$sSQL .= " order by importance desc";
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
+ 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))
{
}
// Do we have anything that looks like a lat/lon pair?
- if (preg_match('/\\b([NS])[ ]+([0-9]+[0-9.]*)[ ]+([0-9.]+)?[, ]+([EW])[ ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?\\b/', $sQuery, $aData))
- {
- $fQueryLat = ($aData[1]=='N'?1:-1) * ($aData[2] + $aData[3]/60);
- $fQueryLon = ($aData[4]=='E'?1:-1) * ($aData[5] + $aData[6]/60);
- if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
- {
- $this->setNearPoint(array($fQueryLat, $fQueryLon));
- $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
- }
- }
- elseif (preg_match('/\\b([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([NS])[, ]+([0-9]+)[ ]+([0-9]+[0-9.]*)?[ ]+([EW])\\b/', $sQuery, $aData))
- {
- $fQueryLat = ($aData[3]=='N'?1:-1) * ($aData[1] + $aData[2]/60);
- $fQueryLon = ($aData[6]=='E'?1:-1) * ($aData[4] + $aData[5]/60);
- if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
- {
- $this->setNearPoint(array($fQueryLat, $fQueryLon));
- $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
- }
- }
- elseif (preg_match('/(\\[|^|\\b)(-?[0-9]+[0-9]*\\.[0-9]+)[, ]+(-?[0-9]+[0-9]*\\.[0-9]+)(\\]|$|\\b)/', $sQuery, $aData))
- {
- $fQueryLat = $aData[2];
- $fQueryLon = $aData[3];
- if ($fQueryLat <= 90.1 && $fQueryLat >= -90.1 && $fQueryLon <= 180.1 && $fQueryLon >= -180.1)
- {
- $this->setNearPoint(array($fQueryLat, $fQueryLon));
- $sQuery = trim(str_replace($aData[0], ' ', $sQuery));
- }
+ if ( $aLooksLike = looksLikeLatLonPair($sQuery) )
+ {
+ $this->setNearPoint(array($aLooksLike['lat'], $aLooksLike['lon']));
+ $sQuery = $aLooksLike['query'];
}
$aSearchResults = array();
{
// 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?
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 $aSpecialTerm)
{
foreach($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']) != $aToken['country_code'])
{
continue;
}
{
if (substr($aData[1],-2,1) != ' ')
{
- $aData[0] = substr($aData[0],0,strlen($aData[1]-1)).' '.substr($aData[0],strlen($aData[1]-1));
+ $aData[0] = substr($aData[0],0,strlen($aData[1])-1).' '.substr($aData[0],strlen($aData[1])-1);
$aData[1] = substr($aData[1],0,-1).' '.substr($aData[1],-1,1);
}
$aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $this->oDB);
// TODO: suggestions
// Start the search process
+ // array with: placeid => -1 | tiger-housenumber
$aResultPlaceIDs = array();
- /*
- Calculate all searches using aValidTokens i.e.
- 'Wodsworth Road, Sheffield' =>
-
- Phrase Wordset
- 0 0 (wodsworth road)
- 0 1 (wodsworth)(road)
- 1 0 (sheffield)
+ $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases);
- Score how good the search is so they can be ordered
- */
- foreach($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 $aSearch)
{
- //echo "<br><b>$sToken</b>";
- $aNewWordsetSearches = array();
-
- foreach($aWordsetSearches as $aCurrentSearch)
+ if ($aSearch['iSearchRank'] < $this->iMaxRank)
{
- //echo "<i>";
- //var_dump($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 reverse order is enabled, it may appear at the beginning as well.
- if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases)) &&
- (!$this->bReverseInPlan || $iToken > 0 || $iPhrase > 0))
- {
- $aSearch['iSearchRank'] += 5;
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- elseif (isset($aSearchTerm['lat']) && $aSearchTerm['lat'] !== '' && $aSearchTerm['lat'] !== null)
- {
- if ($aSearch['fLat'] === '')
- {
- $aSearch['fLat'] = $aSearchTerm['lat'];
- $aSearch['fLon'] = $aSearchTerm['lon'];
- $aSearch['fRadius'] = $aSearchTerm['radius'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- elseif ($sPhraseType == 'postalcode')
- {
- // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
- if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
- {
- // If we already have a name try putting the postcode first
- if (sizeof($aSearch['aName']))
- {
- $aNewSearch = $aSearch;
- $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
- $aNewSearch['aName'] = array();
- $aNewSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;
- }
-
- if (sizeof($aSearch['aName']))
- {
- if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
- {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- }
- else
- {
- $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1000; // skip;
- }
- }
- else
- {
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- //$aSearch['iNamePhrase'] = $iPhrase;
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
-
- }
- elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house')
- {
- if ($aSearch['sHouseNumber'] === '')
- {
- $aSearch['sHouseNumber'] = $sToken;
- // sanity check: if the housenumber is not mainly made
- // up of numbers, add a penalty
- if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- /*
- // Fall back to not searching for this item (better than nothing)
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- */
- }
- }
- elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null)
- {
- if ($aSearch['sClass'] === '')
- {
- $aSearch['sOperator'] = $aSearchTerm['operator'];
- $aSearch['sClass'] = $aSearchTerm['class'];
- $aSearch['sType'] = $aSearchTerm['type'];
- if (sizeof($aSearch['aName'])) $aSearch['sOperator'] = 'name';
- else $aSearch['sOperator'] = 'near'; // near = in for the moment
- if (strlen($aSearchTerm['operator']) == 0) $aSearch['iSearchRank'] += 1;
-
- // Do we have a shortcut id?
- if ($aSearch['sOperator'] == 'name')
- {
- $sSQL = "select get_tagpair('".$aSearch['sClass']."', '".$aSearch['sType']."')";
- if ($iAmenityID = $this->oDB->getOne($sSQL))
- {
- $aValidTokens[$aSearch['sClass'].':'.$aSearch['sType']] = array('word_id' => $iAmenityID);
- $aSearch['aName'][$iAmenityID] = $iAmenityID;
- $aSearch['sClass'] = '';
- $aSearch['sType'] = '';
- }
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
- {
- if (sizeof($aSearch['aName']))
- {
- if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false))
- {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- }
- else
- {
- $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1000; // skip;
- }
- }
- else
- {
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- //$aSearch['iNamePhrase'] = $iPhrase;
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- }
- if (isset($aValidTokens[$sToken]))
- {
- // Allow searching for a word - but at extra cost
- foreach($aValidTokens[$sToken] as $aSearchTerm)
- {
- if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])
- {
- if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4)
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
- {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version?
- {
- foreach($aValidTokens[' '.$sToken] as $aSearchTermToken)
- {
- if (empty($aSearchTermToken['country_code'])
- && empty($aSearchTermToken['lat'])
- && empty($aSearchTermToken['class']))
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- }
- else
- {
- $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
-
- if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)
- {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 2;
- if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
- if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency)
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- else
- $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iNamePhrase'] = $iPhrase;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- }
- }
- else
- {
- // Allow skipping a word - but at EXTREAM cost
- //$aSearch = $aCurrentSearch;
- //$aSearch['iSearchRank']+=100;
- //$aNewWordsetSearches[] = $aSearch;
- }
+ if (!isset($aReverseGroupedSearches[$aSearch['iSearchRank']])) $aReverseGroupedSearches[$aSearch['iSearchRank']] = array();
+ $aReverseGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
}
- // Sort and cut
- usort($aNewWordsetSearches, 'bySearchRank');
- $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
- }
- //var_Dump('<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;
- }
- }
+ $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, $aValidTokens);
-
}
-
}
else
{
if (CONST_Debug) var_Dump($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 => $aSearches)
foreach($aSearches as $aSearch)
{
$iQueryLoop++;
+ $searchedHousenumber = -1;
if (CONST_Debug) { echo "<hr><b>Search Loop, group $iGroupLoop, loop $iQueryLoop</b>"; }
if (CONST_Debug) _debugDumpGroupedSearches(array($iGroupedRank => array($aSearch)), $aValidTokens);
{
$sSQL = "select place_id from placex where calculated_country_code='".$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);
$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'],",")."]";
$sSQL .= " where ".join(' and ',$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'])
$sSQL .= " limit 1";
else
//var_Dump($aPlaceIDs);
//exit;
+ //now search for housenumber, if housenumber provided
if ($aSearch['sHouseNumber'] && sizeof($aPlaceIDs))
{
$aRoadPlaceIDs = $aPlaceIDs;
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $this->oDB->getCol($sSQL);
- // If not try the aux fallback table
+ // 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($aSearch['sHouseNumber'])."'";
if (CONST_Debug) var_dump($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)
+ $searchedHousenumber = intval($aSearch['sHouseNumber']);
if (!sizeof($aPlaceIDs))
{
- $sSQL = "select place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
+ //new query for lines, not housenumbers anymore
+ if($searchedHousenumber%2 == 0){
+ //if housenumber is even, look for housenumber in streets with interpolationtype even or all
+ $sSQL = "select distinct place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='even' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber";
+ }else{
+ //look for housenumber in streets with interpolationtype odd or all
+ $sSQL = "select distinct place_id from location_property_tiger where parent_place_id in (".$sPlaceIDs.") and (interpolationtype='odd' or interpolationtype='all') and ".$searchedHousenumber.">=startnumber and ".$searchedHousenumber."<=endnumber";
+ }
+
if (sizeof($this->aExcludePlaceIDs))
{
- $sSQL .= " and place_id not in (".join(',',$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);
+ //get place IDs
+ $aPlaceIDs = $this->oDB->getCol($sSQL, 0);
}
- // Fallback to the road
+ // Fallback to the road (if no housenumber was found)
if (!sizeof($aPlaceIDs) && preg_match('/[0-9]+/', $aSearch['sHouseNumber']))
{
$aPlaceIDs = $aRoadPlaceIDs;
+ //set to -1, if no housenumbers were found
+ $searchedHousenumber = -1;
}
-
+ //else: housenumber was found, remains saved in searchedHousenumber
}
+
if ($aSearch['sClass'] && sizeof($aPlaceIDs))
{
- $sPlaceIDs = join(',',$aPlaceIDs);
+ $sPlaceIDs = join(',', $aPlaceIDs);
$aClassPlaceIDs = array();
if (!$aSearch['sOperator'] || $aSearch['sOperator'] == 'name')
foreach($aPlaceIDs as $iPlaceID)
{
- $aResultPlaceIDs[$iPlaceID] = $iPlaceID;
+ // array for placeID => -1 | Tiger housenumber
+ $aResultPlaceIDs[$iPlaceID] = $searchedHousenumber;
}
if ($iQueryLoop > 20) break;
}
if (isset($aResultPlaceIDs) && sizeof($aResultPlaceIDs) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30))
{
// Need to verify passes rank limits before dropping out of the loop (yuk!)
- $sSQL = "select place_id from placex where place_id in (".join(',',$aResultPlaceIDs).") ";
+ // reduces the number of place ids, like a filter
+ $sSQL = "select place_id from placex where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";
$sSQL .= "and (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).")";
- $sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',',$aResultPlaceIDs).") ";
+ $sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";
$sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',',$this->aAddressRankList).")";
$sSQL .= ")";
if (CONST_Debug) var_dump($sSQL);
- $aResultPlaceIDs = $this->oDB->getCol($sSQL);
+ $aFilteredPlaceIDs = $this->oDB->getCol($sSQL);
+ $tempIDs = array();
+ foreach($aFilteredPlaceIDs as $placeID)
+ {
+ $tempIDs[$placeID] = $aResultPlaceIDs[$placeID]; //assign housenumber to placeID
+ }
+ $aResultPlaceIDs = $tempIDs;
}
//exit;
$aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery);
foreach($aRecheckWords as $i => $sWord)
{
- if (!$sWord) unset($aRecheckWords[$i]);
+ if (!preg_match('/\pL/', $sWord)) unset($aRecheckWords[$i]);
}
+ if (CONST_Debug) { echo '<i>Recheck words:<\i>'; var_dump($aRecheckWords); }
+
foreach($aSearchResults as $iResNum => $aResult)
{
- if (CONST_Search_AreaPolygons)
+ // Default
+ $fDiameter = getResultDiameter($aResult);
+
+ $oPlaceLookup = new PlaceLookup($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);
+
+ $aOutlineResult = $oPlaceLookup->getOutlines($aResult['place_id'], $aResult['lon'], $aResult['lat'], $fDiameter/2);
+ if ($aOutlineResult)
{
- // 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, $aPointPolygon);
- }
-
- if ($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'] + $fRadius;
- }
- }
-
- // Output data suitable for display (points and a bounding box)
- if ($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']);
- }
+ $aResult = array_merge($aResult, $aOutlineResult);
}
-
+
if ($aResult['extra_place'] == 'city')
{
$aResult['class'] = 'place';
$aResult['rank_search'] = 16;
}
- if (!isset($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'])
{
$aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label'];
}
-
+ // if tag '&addressdetails=1' is set in query
if ($this->bIncludeAddressDetails)
{
- $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code']);
+ // getAddressDetails() is defined in lib.php and uses the SQL function get_addressdata in functions.sql
+ $aResult['address'] = getAddressDetails($this->oDB, $sLanguagePrefArraySQL, $aResult['place_id'], $aResult['country_code'], $aResultPlaceIDs[$aResult['place_id']]);
if ($aResult['extra_place'] == 'city' && !isset($aResult['address']['city']))
{
$aResult['address'] = array_merge(array('city' => array_shift(array_values($aResult['address']))), $aResult['address']);
}
}
+ if ($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']);
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)) $iCountWords += 0.1;
}
}
if (isset($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');
$bFirst = true;
foreach($aToFilter as $iResNum => $aResult)
{
- if ($aResult['type'] == 'adminitrative') $aResult['type'] = 'administrative';
$this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
if ($bFirst)
{
projects / nominatim.git / blobdiff