X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/a92907186b2acc6e2074c0db270de49e9f266a9d..095dbe671880a9457960295425300e054feb6a1a:/lib/Geocode.php?ds=inline diff --git a/lib/Geocode.php b/lib/Geocode.php index f87748c3..8616be13 100644 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@ -15,7 +15,7 @@ protected $aExcludePlaceIDs = array(); protected $bDeDupe = true; - protected $bReverseInPlan = false; + protected $bReverseInPlan = true; protected $iLimit = 20; protected $iFinalLimit = 10; @@ -27,6 +27,8 @@ protected $bBoundedSearch = false; protected $aViewBox = false; + protected $sViewboxSmallSQL = false; + protected $sViewboxLargeSQL = false; protected $aRoutePoints = false; protected $iMaxRank = 20; @@ -370,12 +372,16 @@ // Get the details for display (is this a redundant extra step?) $sPlaceIDs = join(',',$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_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 .= $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) "; @@ -400,20 +406,21 @@ $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 .= $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, 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 .= $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) "; @@ -421,9 +428,10 @@ $sSQL .= "group by place_id"; if (!$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); @@ -477,8 +485,7 @@ $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'])) @@ -489,7 +496,7 @@ } // View Box SQL - $sViewboxCentreSQL = $sViewboxSmallSQL = $sViewboxLargeSQL = false; + $sViewboxCentreSQL = false; $bBoundingBoxSearch = false; if ($this->aViewBox) { @@ -500,8 +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; } @@ -519,20 +526,20 @@ $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 = $this->bBoundedSearch; } @@ -885,6 +892,9 @@ if ($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) @@ -903,6 +913,7 @@ $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') @@ -949,7 +960,7 @@ { 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; @@ -958,8 +969,11 @@ $aSearch['aAddress'][$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']) @@ -976,6 +990,7 @@ else { $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id']; + if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2; if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch; } } @@ -983,7 +998,8 @@ if (!sizeof($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']] = $aSearchTerm['word_id']; @@ -1172,6 +1188,10 @@ if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $this->oDB->getCol($sSQL); } + else + { + $aPlaceIDs = array(); + } } else { @@ -1182,7 +1202,7 @@ { $sSQL = "select place_id from place_classtype_".$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($this->aExcludePlaceIDs)) { @@ -1196,11 +1216,12 @@ // If excluded place IDs are given, it is fair to assume that // there have been results in the small box, so no further // expansion in that case. - if (!sizeof($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"; @@ -1211,7 +1232,7 @@ 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 $this->iLimit"; @@ -1231,7 +1252,7 @@ // TODO: filter out the pointless search terms (2 letter name tokens and less) // they might be right - but they are just too darned expensive to run if (sizeof($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'] != $aSearch['aAddress']) { // For infrequent name terms disable index usage for address @@ -1239,12 +1260,13 @@ sizeof($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'])."'"; @@ -1263,12 +1285,20 @@ $aTerms[] = "country_code in ($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'])) { @@ -1332,6 +1362,7 @@ $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($aSearch['sHouseNumber'])."'"; @@ -1343,6 +1374,7 @@ if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $this->oDB->getCol($sSQL); } + */ if (!sizeof($aPlaceIDs)) { @@ -1445,7 +1477,7 @@ } if ($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)); @@ -1467,7 +1499,7 @@ } 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)); @@ -1548,7 +1580,7 @@ } $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]); @@ -1594,10 +1626,12 @@ { 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; @@ -1709,7 +1743,11 @@ $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 @@ -1773,17 +1811,3 @@ } // 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; - } -*/