X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/27e6a24718e7d778b6b69df78cf5558aa9917f56..f1c4615c32532cdb03845c017f660fa5c5a06331:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index 0e1d7c1f..7de9c8e2 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -15,23 +15,27 @@ protected $aExcludePlaceIDs = array(); protected $bDeDupe = true; - protected $bReverseInPlan = false; + protected $bReverseInPlan = true; protected $iLimit = 20; protected $iFinalLimit = 10; protected $iOffset = 0; + protected $bFallback = false; protected $aCountryCodes = false; protected $aNearPoint = false; protected $bBoundedSearch = false; protected $aViewBox = false; + protected $sViewboxSmallSQL = false; + protected $sViewboxLargeSQL = false; protected $aRoutePoints = false; protected $iMaxRank = 20; protected $iMinAddressRank = 0; protected $iMaxAddressRank = 30; protected $aAddressRankList = array(); + protected $exactMatchCache = array(); protected $sAllowedTypesSQLList = false; @@ -43,6 +47,11 @@ $this->oDB =& $oDB; } + function setReverseInPlan($bReverse) + { + $this->bReverseInPlan = $bReverse; + } + function setLanguagePreference($aLangPref) { $this->aLangPrefOrder = $aLangPref; @@ -112,6 +121,11 @@ $this->iOffset = $iOffset; } + function setFallback($bFallback = true) + { + $this->bFallback = (bool)$bFallback; + } + function setExcludedPlaceIDs($a) { // TODO: force to int @@ -190,20 +204,131 @@ return $this->sQuery; } + + function loadParamArray($aParams) + { + if (isset($aParams['addressdetails'])) $this->bIncludeAddressDetails = (bool)$aParams['addressdetails']; + if (isset($aParams['bounded'])) $this->bBoundedSearch = (bool)$aParams['bounded']; + if (isset($aParams['dedupe'])) $this->bDeDupe = (bool)$aParams['dedupe']; + + if (isset($aParams['limit'])) $this->setLimit((int)$aParams['limit']); + if (isset($aParams['offset'])) $this->iOffset = (int)$aParams['offset']; + + if (isset($aParams['fallback'])) $this->bFallback = (bool)$aParams['fallback']; + + // List of excluded Place IDs - used for more acurate pageing + if (isset($aParams['exclude_place_ids']) && $aParams['exclude_place_ids']) + { + foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID) + { + $iExcludedPlaceID = (int)$iExcludedPlaceID; + if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID; + } + $this->aExcludePlaceIDs = $aExcludePlaceIDs; + } + + // Only certain ranks of feature + if (isset($aParams['featureType'])) $this->setFeatureType($aParams['featureType']); + if (isset($aParams['featuretype'])) $this->setFeatureType($aParams['featuretype']); + + // Country code list + if (isset($aParams['countrycodes'])) + { + $aCountryCodes = array(); + foreach(explode(',',$aParams['countrycodes']) as $sCountryCode) + { + if (preg_match('/^[a-zA-Z][a-zA-Z]$/', $sCountryCode)) + { + $aCountryCodes[] = strtolower($sCountryCode); + } + } + $this->aCountryCodes = $aCountryCodes; + } + + if (isset($aParams['viewboxlbrt']) && $aParams['viewboxlbrt']) + { + $aCoOrdinatesLBRT = explode(',',$aParams['viewboxlbrt']); + $this->setViewBox($aCoOrdinatesLBRT[0], $aCoOrdinatesLBRT[1], $aCoOrdinatesLBRT[2], $aCoOrdinatesLBRT[3]); + } + else if (isset($aParams['viewbox']) && $aParams['viewbox']) + { + $aCoOrdinatesLTRB = explode(',',$aParams['viewbox']); + $this->setViewBox($aCoOrdinatesLTRB[0], $aCoOrdinatesLTRB[3], $aCoOrdinatesLTRB[2], $aCoOrdinatesLTRB[1]); + } + + if (isset($aParams['route']) && $aParams['route'] && isset($aParams['routewidth']) && $aParams['routewidth']) + { + $aPoints = explode(',',$aParams['route']); + if (sizeof($aPoints) % 2 != 0) + { + userError("Uneven number of points"); + exit; + } + $fPrevCoord = false; + $aRoute = array(); + foreach($aPoints as $i => $fPoint) + { + if ($i%2) + { + $aRoute[] = array((float)$fPoint, $fPrevCoord); + } + else + { + $fPrevCoord = (float)$fPoint; + } + } + $this->aRoutePoints = $aRoute; + } + } + + function setQueryFromParams($aParams) + { + // Search query + $sQuery = (isset($aParams['q'])?trim($aParams['q']):''); + if (!$sQuery) + { + $this->setStructuredQuery(@$aParams['amenity'], @$aParams['street'], @$aParams['city'], @$aParams['county'], @$aParams['state'], @$aParams['country'], @$aParams['postalcode']); + $this->setReverseInPlan(false); + } + else + { + $this->setQuery($sQuery); + } + } + + 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, $aItemListValues); + return true; + } + function setStructuredQuery($sAmentiy = 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(); $this->sAllowedTypesSQLList = ''; - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sAmentiy, 'amenity', 26, 30, false); - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sStreet, 'street', 26, 30, false); - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sCity, 'city', 14, 24, false); - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sCounty, 'county', 9, 13, false); - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sState, 'state', 8, 8, false); - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sCountry, 'country', 4, 4, false); - loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sPostalCode, 'postalcode' , 5, 11, array(5, 11)); + $this->loadStructuredAddressElement($sAmentiy, '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) { @@ -213,7 +338,29 @@ $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } } + } + + function fallbackStructuredQuery() + { + if (!$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; } function getDetails($aPlaceIDs) @@ -225,18 +372,22 @@ // Get the details for display (is this a redundant extra step?) $sPlaceIDs = join(',',$aPlaceIDs); - $sSQL = "select osm_type,osm_id,class,type,admin_level,rank_search,rank_address,min(place_id) as place_id,calculated_country_code as country_code,"; + $sImportanceSQL = ''; + if ($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_name_by_language(name, $sLanguagePrefArraySQL) as placename,"; $sSQL .= "get_name_by_language(name, ARRAY['ref']) as ref,"; $sSQL .= "avg(ST_X(centroid)) as lon,avg(ST_Y(centroid)) as lat, "; - $sSQL .= "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(placex.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $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 .= "(extratags->'place') as extra_place "; $sSQL .= "from placex where place_id in ($sPlaceIDs) "; $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank "; if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'"; - if ($aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$aAddressRankList).")"; + 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 "; @@ -250,35 +401,37 @@ if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank) { $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,'us' as country_code,"; + $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 .= "-0.15 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.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $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) $sSQL .= ",place_id"; + if (!$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,'us' as country_code,"; + $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 .= "-0.10 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.place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, "; + $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"; + $sSQL .= " order by importance desc"; if (CONST_Debug) { echo "
"; var_dump($sSQL); } $aSearchResults = $this->oDB->getAll($sSQL); @@ -290,6 +443,36 @@ return $aSearchResults; } + /* Perform the actual query lookup. + + Returns an ordered list of results, each with the following fields: + osm_type: type of corresponding OSM object + N - node + W - way + R - relation + P - postcode (internally computed) + osm_id: id of corresponding OSM object + class: general object class (corresponds to tag key of primary OSM tag) + type: subclass of object (corresponds to tag value of primary OSM tag) + admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level + rank_search: rank in search hierarchy + (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level) + rank_address: rank in address hierarchy (determines orer in address) + place_id: internal key (may differ between different instances) + country_code: ISO country code + langaddress: localized full address + placename: localized name of object + ref: content of ref tag (if available) + lon: longitude + lat: latitude + importance: importance of place based on Wikipedia link count + addressimportance: cumulated importance of address elements + extra_place: type of place (for admin boundaries, if there is a place tag) + aBoundingBox: bounding Box + label: short description of the object class/type (English only) + name: full name (currently the same as langaddress) + foundorder: secondary ordering for places with same importance + */ function lookup() { if (!$this->sQuery && !$this->aStructuredQuery) return false; @@ -302,19 +485,18 @@ $sCountryCodesSQL = join(',', array_map('addQuotes', $this->aCountryCodes)); } - // Hack to make it handle "new york, ny" (and variants) correctly - $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $this->sQuery); + $sQuery = $this->sQuery; // Conflicts between US state abreviations and various words for 'the' in different languages if (isset($this->aLangPrefOrder['name:en'])) { - $sQuery = preg_replace('/,\s*il\s*(,|$)/',', illinois\1', $sQuery); - $sQuery = preg_replace('/,\s*al\s*(,|$)/',', alabama\1', $sQuery); - $sQuery = preg_replace('/,\s*la\s*(,|$)/',', louisiana\1', $sQuery); + $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/','\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 = $sViewboxSmallSQL = $sViewboxLargeSQL = false; + $sViewboxCentreSQL = false; $bBoundingBoxSearch = false; if ($this->aViewBox) { @@ -325,8 +507,8 @@ $aBigViewBox[1] = $this->aViewBox[1] + $fWidth; $aBigViewBox[3] = $this->aViewBox[3] - $fWidth; - $sViewboxSmallSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$this->aViewBox[0].",".(float)$this->aViewBox[1]."),ST_Point(".(float)$this->aViewBox[2].",".(float)$this->aViewBox[3].")),4326)"; - $sViewboxLargeSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$aBigViewBox[0].",".(float)$aBigViewBox[1]."),ST_Point(".(float)$aBigViewBox[2].",".(float)$aBigViewBox[3].")),4326)"; + $this->sViewboxSmallSQL = "ST_SetSRID(ST_MakeBox2D(ST_Point(".(float)$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 = $this->bBoundedSearch; } @@ -334,62 +516,37 @@ if ($this->aRoutePoints) { $sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; - $bFirst = false; - foreach($this->aRouteaPoints as $aPoint) + $bFirst = true; + foreach($this->aRoutePoints as $aPoint) { if (!$bFirst) $sViewboxCentreSQL .= ","; - $sViewboxCentreSQL .= $aPoint[1].' '.$aPoint[0]; + $sViewboxCentreSQL .= $aPoint[0].' '.$aPoint[1]; + $bFirst = false; } $sViewboxCentreSQL .= ")'::geometry,4326)"; $sSQL = "select st_buffer(".$sViewboxCentreSQL.",".(float)($_GET['routewidth']/69).")"; - $sViewboxSmallSQL = $this->oDB->getOne($sSQL); - if (PEAR::isError($sViewboxSmallSQL)) + $this->sViewboxSmallSQL = $this->oDB->getOne($sSQL); + if (PEAR::isError($this->sViewboxSmallSQL)) { - failInternalError("Could not get small viewbox.", $sSQL, $sViewboxSmallSQL); + failInternalError("Could not get small viewbox.", $sSQL, $this->sViewboxSmallSQL); } - $sViewboxSmallSQL = "'".$sViewboxSmallSQL."'::geometry"; + $this->sViewboxSmallSQL = "'".$this->sViewboxSmallSQL."'::geometry"; $sSQL = "select st_buffer(".$sViewboxCentreSQL.",".(float)($_GET['routewidth']/30).")"; - $sViewboxLargeSQL = $this->oDB->getOne($sSQL); - if (PEAR::isError($sViewboxLargeSQL)) + $this->sViewboxLargeSQL = $this->oDB->getOne($sSQL); + if (PEAR::isError($this->sViewboxLargeSQL)) { - failInternalError("Could not get large viewbox.", $sSQL, $sViewboxLargeSQL); + failInternalError("Could not get large viewbox.", $sSQL, $this->sViewboxLargeSQL); } - $sViewboxLargeSQL = "'".$sViewboxLargeSQL."'::geometry"; + $this->sViewboxLargeSQL = "'".$this->sViewboxLargeSQL."'::geometry"; $bBoundingBoxSearch = $this->bBoundedSearch; } // 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']; } $aSearchResults = array(); @@ -552,7 +709,7 @@ { if (substr($aData[1],-2,1) != ' ') { - $aData[0] = substr($aData[0],0,strlen($aData[1]-1)).' '.substr($aData[0],strlen($aData[1]-1)); + $aData[0] = substr($aData[0],0,strlen($aData[1])-1).' '.substr($aData[0],strlen($aData[1])-1); $aData[1] = substr($aData[1],0,-1).' '.substr($aData[1],-1,1); } $aGBPostcodeLocation = gbPostcodeCalculate($aData[0], $aData[1], $aData[2], $this->oDB); @@ -617,8 +774,11 @@ if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase]; else $sPhraseType = ''; - foreach($aPhrases[$iPhrase]['wordsets'] as $aWordset) + foreach($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset) { + // Too many permutations - too expensive + if ($iWordSet > 120) break; + $aWordsetSearches = $aSearches; // Add all words from this wordset @@ -706,6 +866,9 @@ if ($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) @@ -724,6 +887,7 @@ $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') @@ -770,7 +934,7 @@ { if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) { - if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strlen($sToken) >= 4) + if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false) { $aSearch = $aCurrentSearch; $aSearch['iSearchRank'] += 1; @@ -779,8 +943,11 @@ $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } - elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version? + elseif (isset($aValidTokens[' '.$sToken]) && strlen($sToken) >= 4) // revert to the token version? { + $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + $aSearch['iSearchRank'] += 1; + if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; foreach($aValidTokens[' '.$sToken] as $aSearchTermToken) { if (empty($aSearchTermToken['country_code']) @@ -797,6 +964,7 @@ else { $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } } @@ -804,7 +972,8 @@ if (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase) { $aSearch = $aCurrentSearch; - $aSearch['iSearchRank'] += 2; + $aSearch['iSearchRank'] += 1; + if (!sizeof($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']; @@ -989,10 +1158,16 @@ { $sSQL = "select place_id from placex where calculated_country_code='".$aSearch['sCountryCode']."' and rank_search = 4"; if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; + if ($bBoundingBoxSearch) + $sSQL .= " and st_overlaps($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 { @@ -1003,7 +1178,7 @@ { $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct"; if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; - $sSQL .= " where st_contains($sViewboxSmallSQL, ct.centroid)"; + $sSQL .= " where st_contains($this->sViewboxSmallSQL, ct.centroid)"; if ($sCountryCodesSQL) $sSQL .= " and calculated_country_code in ($sCountryCodesSQL)"; if (sizeof($this->aExcludePlaceIDs)) { @@ -1017,11 +1192,12 @@ // If excluded place IDs are given, it is fair to assume that // there have been results in the small box, so no further // expansion in that case. - if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs)) + // Also don't expand if bounded results were requested. + if (!sizeof($aPlaceIDs) && !sizeof($this->aExcludePlaceIDs) && !$this->bBoundedSearch) { $sSQL = "select place_id from place_classtype_".$aSearch['sClass']."_".$aSearch['sType']." ct"; if ($sCountryCodesSQL) $sSQL .= " join placex using (place_id)"; - $sSQL .= " where st_contains($sViewboxLargeSQL, ct.centroid)"; + $sSQL .= " where st_contains($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"; @@ -1032,7 +1208,7 @@ else { $sSQL = "select place_id from placex where class='".$aSearch['sClass']."' and type='".$aSearch['sType']."'"; - $sSQL .= " and st_contains($sViewboxSmallSQL, geometry) and linked_place_id is null"; + $sSQL .= " and st_contains($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 $this->iLimit"; @@ -1052,7 +1228,7 @@ // TODO: filter out the pointless search terms (2 letter name tokens and less) // they might be right - but they are just too darned expensive to run if (sizeof($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['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]"; if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress']) { // For infrequent name terms disable index usage for address @@ -1060,16 +1236,31 @@ sizeof($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'],$aSearch['aAddressNonSearch']),",")."]"; + //$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]"; + $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]"; } else { $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]"; - if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]"; + //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']) $aTerms[] = "address_rank between 16 and 27"; + if ($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'].")"; @@ -1084,21 +1275,33 @@ $aTerms[] = "country_code in ($sCountryCodesSQL)"; } - if ($bBoundingBoxSearch) $aTerms[] = "centroid && $sViewboxSmallSQL"; + if ($bBoundingBoxSearch) $aTerms[] = "centroid && $this->sViewboxSmallSQL"; if ($sNearPointSQL) $aOrder[] = "ST_Distance($sNearPointSQL, centroid) asc"; - $sImportanceSQL = '(case when importance = 0 OR importance IS NULL then 0.75-(search_rank::float/40) else importance end)'; - if ($sViewboxSmallSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxSmallSQL, centroid) THEN 1 ELSE 0.5 END"; - if ($sViewboxLargeSQL) $sImportanceSQL .= " * case when ST_Contains($sViewboxLargeSQL, centroid) THEN 1 ELSE 0.5 END"; + 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'])) { - $aOrder[] = '(select count(*) from (select unnest(ARRAY['.join($aSearch['aFullNameAddress'],",").']) INTERSECT select unnest(nameaddress_vector))s) DESC'; + $sExactMatchSQL = '(select count(*) from (select unnest(ARRAY['.join($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 = "select place_id, "; + $sSQL .= $sExactMatchSQL; $sSQL .= " from search_name"; $sSQL .= " where ".join(' and ',$aTerms); $sSQL .= " order by ".join(', ',$aOrder); @@ -1126,6 +1329,7 @@ //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); @@ -1137,8 +1341,8 @@ $sPlaceIDs = join(',',$aPlaceIDs); // Now they are indexed look for a house attached to a street we found - $sHouseNumberRegex = '\\\\m'.str_replace(' ','[-,/ ]',$aSearch['sHouseNumber']).'\\\\M'; - $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and housenumber ~* E'".$sHouseNumberRegex."'"; + $sHouseNumberRegex = '\\\\m'.$aSearch['sHouseNumber'].'\\\\M'; + $sSQL = "select place_id from placex where parent_place_id in (".$sPlaceIDs.") and transliteration(housenumber) ~* E'".$sHouseNumberRegex."'"; if (sizeof($this->aExcludePlaceIDs)) { $sSQL .= " and place_id not in (".join(',',$this->aExcludePlaceIDs).")"; @@ -1148,6 +1352,7 @@ $aPlaceIDs = $this->oDB->getCol($sSQL); // If not 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($aSearch['sHouseNumber'])."'"; @@ -1159,6 +1364,7 @@ if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $this->oDB->getCol($sSQL); } + */ if (!sizeof($aPlaceIDs)) { @@ -1261,7 +1467,7 @@ } if ($sCountryCodesSQL) $sSQL .= " and lp.calculated_country_code in ($sCountryCodesSQL)"; if ($sOrderBySQL) $sSQL .= "order by ".$sOrderBySQL." asc"; - if ($iOffset) $sSQL .= " offset $iOffset"; + if ($this->iOffset) $sSQL .= " offset $this->iOffset"; $sSQL .= " limit $this->iLimit"; if (CONST_Debug) var_dump($sSQL); $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL)); @@ -1283,7 +1489,7 @@ } if ($sCountryCodesSQL) $sSQL .= " and l.calculated_country_code in ($sCountryCodesSQL)"; if ($sOrderBy) $sSQL .= "order by ".$OrderBysSQL." asc"; - if ($iOffset) $sSQL .= " offset $iOffset"; + if ($this->iOffset) $sSQL .= " offset $this->iOffset"; $sSQL .= " limit $this->iLimit"; if (CONST_Debug) var_dump($sSQL); $aClassPlaceIDs = array_merge($aClassPlaceIDs, $this->oDB->getCol($sSQL)); @@ -1352,11 +1558,19 @@ // No results? Done if (!sizeof($aSearchResults)) { + if ($this->bFallback) + { + if ($this->fallbackStructuredQuery()) + { + return $this->lookup(); + } + } + return array(); } $aClassType = getClassTypesWithImportance(); - $aRecheckWords = preg_split('/\b/u',$sQuery); + $aRecheckWords = preg_split('/\b[\s,\\-]*/u',$sQuery); foreach($aRecheckWords as $i => $sWord) { if (!$sWord) unset($aRecheckWords[$i]); @@ -1402,10 +1616,12 @@ { 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; @@ -1481,7 +1697,7 @@ $aResult['aPolyPoints'][] = array($aPoint[1], $aPoint[2]); } } - $aResult['aBoundingBox'] = array($aPointPolygon['minlat'],$aPointPolygon['maxlat'],$aPointPolygon['minlon'],$aPointPolygon['maxlon']); + $aResult['aBoundingBox'] = array((string)$aPointPolygon['minlat'],(string)$aPointPolygon['maxlat'],(string)$aPointPolygon['minlon'],(string)$aPointPolygon['maxlon']); } // Is there an icon set for this type of result? @@ -1517,13 +1733,34 @@ $sAddress = $aResult['langaddress']; foreach($aRecheckWords as $i => $sWord) { - if (stripos($sAddress, $sWord)!==false) $iCountWords++; + if (stripos($sAddress, $sWord)!==false) + { + $iCountWords++; + if (preg_match("/(^|,)\s*$sWord\s*(,|$)/", $sAddress)) $iCountWords += 0.1; + } } $aResult['importance'] = $aResult['importance'] + ($iCountWords*0.1); // 0.1 is a completely arbitrary number but something in the range 0.1 to 0.5 would seem right $aResult['name'] = $aResult['langaddress']; - $aResult['foundorder'] = -$aResult['addressimportance']; + // secondary ordering (for results with same importance (the smaller the better): + // - approximate importance of address parts + $aResult['foundorder'] = -$aResult['addressimportance']/10; + // - number of exact matches from the query + if (isset($this->exactMatchCache[$aResult['place_id']])) + $aResult['foundorder'] -= $this->exactMatchCache[$aResult['place_id']]; + else if (isset($this->exactMatchCache[$aResult['parent_place_id']])) + $aResult['foundorder'] -= $this->exactMatchCache[$aResult['parent_place_id']]; + // - importance of the class/type + 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']; + } + else + { + $aResult['foundorder'] = $aResult['foundorder'] + 0.001; + } $aSearchResults[$iResNum] = $aResult; } uasort($aSearchResults, 'byImportance'); @@ -1564,17 +1801,3 @@ } // end class - -/* - if (isset($_GET['route']) && $_GET['route'] && isset($_GET['routewidth']) && $_GET['routewidth']) - { - $aPoints = explode(',',$_GET['route']); - if (sizeof($aPoints) % 2 != 0) - { - userError("Uneven number of points"); - exit; - } - $sViewboxCentreSQL = "ST_SetSRID('LINESTRING("; - $fPrevCoord = false; - } -*/