X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/624e135a83ced039b620ce0077d1f8ddbdabce7a..b25ecf13db2481cc6be7745424ef54ab2de6a4cb:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index d660e8b3..e3fa25b6 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -20,6 +20,7 @@ protected $iLimit = 20; protected $iFinalLimit = 10; protected $iOffset = 0; + protected $bFallback = false; protected $aCountryCodes = false; protected $aNearPoint = false; @@ -32,6 +33,7 @@ protected $iMinAddressRank = 0; protected $iMaxAddressRank = 30; protected $aAddressRankList = array(); + protected $exactMatchCache = array(); protected $sAllowedTypesSQLList = false; @@ -43,6 +45,11 @@ $this->oDB =& $oDB; } + function setReverseInPlan($bReverse) + { + $this->bReverseInPlan = $bReverse; + } + function setLanguagePreference($aLangPref) { $this->aLangPrefOrder = $aLangPref; @@ -112,6 +119,11 @@ $this->iOffset = $iOffset; } + function setFallback($bFallback = true) + { + $this->bFallback = (bool)$bFallback; + } + function setExcludedPlaceIDs($a) { // TODO: force to int @@ -190,20 +202,39 @@ return $this->sQuery; } + function loadStructuredAddressElement($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) { @@ -213,10 +244,32 @@ $sAllowedTypesSQLList = '(\'place\',\'boundary\')'; } } + } + + function fallbackStructuredQuery() + { + if (!$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, $iMinAddressRank = 0, $iMaxAddressRank = 30, $aAddressRankList = false, $sAllowedTypesSQLList = false, $bDeDupe = false) + function getDetails($aPlaceIDs) { if (sizeof($aPlaceIDs) == 0) return array(); @@ -225,56 +278,56 @@ // 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,"; + $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 .= "(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 $iMinAddressRank and $iMaxAddressRank "; - if (14 >= $iMinAddressRank && 14 <= $iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'"; - if ($aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$aAddressRankList).")"; + $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).")"; $sSQL .= ") "; - if ($sAllowedTypesSQLList) $sSQL .= "and placex.class in $sAllowedTypesSQLList "; + if ($this->sAllowedTypesSQLList) $sSQL .= "and placex.class in $this->sAllowedTypesSQLList "; $sSQL .= "and linked_place_id is null "; $sSQL .= "group by osm_type,osm_id,class,type,admin_level,rank_search,rank_address,calculated_country_code,importance"; - if (!$bDeDupe) $sSQL .= ",place_id"; + if (!$this->bDeDupe) $sSQL .= ",place_id"; $sSQL .= ",langaddress "; $sSQL .= ",placename "; $sSQL .= ",ref "; $sSQL .= ",extratags->'place' "; - if (30 >= $iMinAddressRank && 30 <= $iMaxAddressRank) + 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 .= "(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 $iMinAddressRank and $iMaxAddressRank "; + $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank "; $sSQL .= "group by place_id"; - if (!$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 .= "(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 $iMinAddressRank and $iMaxAddressRank "; + $sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank "; $sSQL .= "group by place_id"; - if (!$bDeDupe) $sSQL .= ",place_id"; + if (!$this->bDeDupe) $sSQL .= ",place_id"; $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) "; } @@ -290,6 +343,36 @@ 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 && !$this->aStructuredQuery) return false; @@ -617,8 +700,11 @@ if ($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 @@ -1093,12 +1179,16 @@ $aOrder[] = "$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 +1216,7 @@ //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); @@ -1352,6 +1443,14 @@ // No results? Done if (!sizeof($aSearchResults)) { + if ($this->bFallback) + { + if ($this->fallbackStructuredQuery()) + { + return $this->lookup(); + } + } + return array(); } @@ -1491,7 +1590,12 @@ $aResult['icon'] = CONST_Website_BaseURL.'images/mapicons/'.$aClassType[$aResult['class'].':'.$aResult['type']]['icon'].'.p.20.png'; } - if (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label']) + if (isset($aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']) + && $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']) + { + $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type'].':'.$aResult['admin_level']]['label']; + } + elseif (isset($aClassType[$aResult['class'].':'.$aResult['type']]['label']) && $aClassType[$aResult['class'].':'.$aResult['type']]['label']) { $aResult['label'] = $aClassType[$aResult['class'].':'.$aResult['type']]['label']; @@ -1518,7 +1622,24 @@ $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');