X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/b121da0ca35c2ff08ecd5a02dd4eb5d4a9dcc3b1..131527bdab6738baff679171c421ea7fa863e100:/lib/Geocode.php diff --git a/lib/Geocode.php b/lib/Geocode.php index d96b6cb8..611ca6de 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -1,4 +1,6 @@ bIncludeAddressDetails; } + function getIncludeExtraTags() + { + return $this->bIncludeExtraTags; + } + + function getIncludeNameDetails() + { + return $this->bIncludeNameDetails; + } + function setIncludePolygonAsPoints($b = true) { $this->bIncludePolygonAsPoints = $b; @@ -102,6 +117,11 @@ $this->bIncludePolygonAsSVG = $b; } + function setPolygonSimplificationThreshold($f) + { + $this->fPolygonSimplificationThreshold = $f; + } + function setDeDupe($bDeDupe = true) { $this->bDeDupe = (bool)$bDeDupe; @@ -208,6 +228,9 @@ 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']; @@ -222,9 +245,12 @@ foreach(explode(',',$aParams['exclude_place_ids']) as $iExcludedPlaceID) { $iExcludedPlaceID = (int)$iExcludedPlaceID; - if ($iExcludedPlaceID) $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID; + if ($iExcludedPlaceID) + $aExcludePlaceIDs[$iExcludedPlaceID] = $iExcludedPlaceID; } - $this->aExcludePlaceIDs = $aExcludePlaceIDs; + + if (isset($aExcludePlaceIDs)) + $this->aExcludePlaceIDs = $aExcludePlaceIDs; } // Only certain ranks of feature @@ -330,7 +356,7 @@ $this->loadStructuredAddressElement($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) @@ -365,21 +391,24 @@ 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, "; @@ -396,37 +425,68 @@ $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) { - $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 .= $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 (CONST_Use_US_Tiger_Data) + { + //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 "; + } + + if (CONST_Use_Aux_Location_data) + { + $sSQL .= " union "; + $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, -1, $sLanguagePrefArraySQL) "; + } } $sSQL .= " order by importance desc"; @@ -444,15 +504,15 @@ function getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases) { /* - Calculate all searches using aValidTokens i.e. - 'Wodsworth Road, Sheffield' => + Calculate all searches using aValidTokens i.e. + 'Wodsworth Road, Sheffield' => - Phrase Wordset - 0 0 (wodsworth road) - 0 1 (wodsworth)(road) - 1 0 (sheffield) + Phrase Wordset + 0 0 (wodsworth road) + 0 1 (wodsworth)(road) + 1 0 (sheffield) - Score how good the search is so they can be ordered + Score how good the search is so they can be ordered */ foreach($aPhrases as $iPhrase => $sPhrase) { @@ -526,7 +586,7 @@ if (sizeof($aSearch['aName'])) { - if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false)) + if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) { $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; } @@ -553,6 +613,8 @@ // 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) @@ -573,18 +635,6 @@ 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[] = $aSearch; } } @@ -592,7 +642,7 @@ { if (sizeof($aSearch['aName'])) { - if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strlen($sToken) < 4 || strpos($sToken, ' ') !== false)) + if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) { $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; } @@ -611,7 +661,10 @@ } } } - if (isset($aValidTokens[$sToken])) + // Look for partial matches. + // Note that there is no point in adding country terms here + // because country are omitted in the address. + if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') { // Allow searching for a word - but at extra cost foreach($aValidTokens[$sToken] as $aSearchTerm) @@ -627,7 +680,7 @@ $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } - elseif (isset($aValidTokens[' '.$sToken]) && strlen($sToken) >= 4) // revert to the token version? + elseif (isset($aValidTokens[' '.$sToken])) // revert to the token version? { $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; $aSearch['iSearchRank'] += 1; @@ -728,39 +781,38 @@ /* 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)) { @@ -826,9 +878,10 @@ } // Do we have anything that looks like a lat/lon pair? - if ( $aLooksLike = looksLikeLatLonPair($sQuery) ){ + if ( $aLooksLike = looksLikeLatLonPair($sQuery) ) + { $this->setNearPoint(array($aLooksLike['lat'], $aLooksLike['lon'])); - $sQuery = $aLooksLike['query']; + $sQuery = $aLooksLike['query']; } $aSearchResults = array(); @@ -836,9 +889,23 @@ { // Start with a blank search $aSearches = array( - array('iSearchRank' => 0, 'iNamePhrase' => -1, 'sCountryCode' => false, 'aName'=>array(), 'aAddress'=>array(), 'aFullNameAddress'=>array(), - 'aNameNonSearch'=>array(), 'aAddressNonSearch'=>array(), - 'sOperator'=>'', 'aFeatureName' => array(), 'sClass'=>'', 'sType'=>'', 'sHouseNumber'=>'', 'fLat'=>'', 'fLon'=>'', 'fRadius'=>'') + array('iSearchRank' => 0, + 'iNamePhrase' => -1, + 'sCountryCode' => false, + 'aName' => array(), + 'aAddress' => array(), + 'aFullNameAddress' => array(), + 'aNameNonSearch' => array(), + 'aAddressNonSearch' => array(), + 'sOperator' => '', + 'aFeatureName' => array(), + 'sClass' => '', + 'sType' => '', + 'sHouseNumber' => '', + 'fLat' => '', + 'fLon' => '', + 'fRadius' => '' + ) ); // Do we have a radius search? @@ -863,10 +930,10 @@ preg_match_all('/\\[([\\w ]*)\\]/u', $sQuery, $aSpecialTermsRaw, PREG_SET_ORDER); $aSpecialTerms = array(); - if (isset($aStructuredQuery['amenity']) && $aStructuredQuery['amenity']) + if (isset($this->aStructuredQuery['amenity']) && $this->aStructuredQuery['amenity']) { - $aSpecialTermsRaw[] = array('['.$aStructuredQuery['amenity'].']', $aStructuredQuery['amenity']); - unset($aStructuredQuery['amenity']); + $aSpecialTermsRaw[] = array('['.$this->aStructuredQuery['amenity'].']', $this->aStructuredQuery['amenity']); + unset($this->aStructuredQuery['amenity']); } foreach($aSpecialTermsRaw as $aSpecialTerm) { @@ -964,8 +1031,8 @@ foreach($aDatabaseWords as $aToken) { // Very special case - require 2 letter country param to match the country code found - if ($bStructuredPhrases && $aToken['country_code'] && !empty($aStructuredQuery['country']) - && strlen($aStructuredQuery['country']) == 2 && strtolower($aStructuredQuery['country']) != $aToken['country_code']) + if ($bStructuredPhrases && $aToken['country_code'] && !empty($this->aStructuredQuery['country']) + && strlen($this->aStructuredQuery['country']) == 2 && strtolower($this->aStructuredQuery['country']) != $aToken['country_code']) { continue; } @@ -1037,6 +1104,7 @@ // TODO: suggestions // Start the search process + // array with: placeid => -1 | tiger-housenumber $aResultPlaceIDs = array(); $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhraseTypes, $aPhrases, $aValidTokens, $aWordFrequencyScores, $bStructuredPhrases); @@ -1051,7 +1119,7 @@ if (sizeof($aPhrases) > 1) { $aFinalPhrase = end($aPhrases); - $aFinalPhrase['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0); + $aPhrases[sizeof($aPhrases)-1]['wordsets'] = getInverseWordSets($aFinalPhrase['words'], 0); } $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, null, $aPhrases, $aValidTokens, $aWordFrequencyScores, false); @@ -1089,7 +1157,7 @@ if (CONST_Debug) var_Dump($aGroupedSearches); - if (CONST_Search_TryDroppedAddressTerms && sizeof($aStructuredQuery) > 0) + if (CONST_Search_TryDroppedAddressTerms && sizeof($this->aStructuredQuery) > 0) { $aCopyGroupedSearches = $aGroupedSearches; foreach($aCopyGroupedSearches as $iGroup => $aSearches) @@ -1154,6 +1222,7 @@ foreach($aSearches as $aSearch) { $iQueryLoop++; + $searchedHousenumber = -1; if (CONST_Debug) { echo "