From: Sarah Hoffmann Date: Wed, 3 Sep 2014 19:10:14 +0000 (+0200) Subject: Merge remote-tracking branch 'upstream/master' X-Git-Tag: deploy~520 X-Git-Url: https://git.openstreetmap.org./nominatim.git/commitdiff_plain/ededee63989ddcdf048833aef932bf02a08a1ab4?hp=-c Merge remote-tracking branch 'upstream/master' --- ededee63989ddcdf048833aef932bf02a08a1ab4 diff --combined lib/Geocode.php index 5451c382,0b3bd51f..6908ac98 --- a/lib/Geocode.php +++ b/lib/Geocode.php @@@ -15,7 -15,7 +15,7 @@@ protected $aExcludePlaceIDs = array(); protected $bDeDupe = true; - protected $bReverseInPlan = false; + protected $bReverseInPlan = true; protected $iLimit = 20; protected $iFinalLimit = 10; @@@ -413,7 -413,6 +413,7 @@@ $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,"; @@@ -428,7 -427,6 +428,7 @@@ $sSQL .= "group by place_id"; if (!$this->bDeDupe) $sSQL .= ",place_id"; $sSQL .= ",get_address_by_language(place_id, $sLanguagePrefArraySQL) "; + */ } $sSQL .= " order by importance desc"; @@@ -520,7 -518,7 +520,7 @@@ foreach($this->aRoutePoints as $aPoint) { if (!$bFirst) $sViewboxCentreSQL .= ","; - $sViewboxCentreSQL .= $aPoint[1].' '.$aPoint[0]; + $sViewboxCentreSQL .= $aPoint[0].' '.$aPoint[1]; $bFirst = false; } $sViewboxCentreSQL .= ")'::geometry,4326)"; @@@ -960,7 -958,7 +960,7 @@@ { if (isset($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; @@@ -969,11 -967,8 +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']) @@@ -990,7 -985,6 +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[] = $aSearch; } } @@@ -998,8 -992,7 +998,8 @@@ if (!sizeof($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']] = $aSearchTerm['word_id']; @@@ -1252,7 -1245,7 +1252,7 @@@ // TODO: filter out the pointless search terms (2 letter name tokens and less) // they might be right - but they are just too darned expensive to run if (sizeof($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'] != $aSearch['aAddress']) { // For infrequent name terms disable index usage for address @@@ -1260,17 -1253,30 +1260,31 @@@ sizeof($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'].")"; @@@ -1362,7 -1368,6 +1376,7 @@@ $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($aSearch['sHouseNumber'])."'"; @@@ -1374,7 -1379,6 +1388,7 @@@ if (CONST_Debug) var_dump($sSQL); $aPlaceIDs = $this->oDB->getCol($sSQL); } + */ if (!sizeof($aPlaceIDs)) { @@@ -1626,12 -1630,10 +1640,12 @@@ { preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$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; diff --combined sql/functions.sql index 9b083609,a8c5058a..7b06b450 --- a/sql/functions.sql +++ b/sql/functions.sql @@@ -936,11 -936,6 +936,11 @@@ DECLAR BEGIN --DEBUG: RAISE WARNING '% %',NEW.osm_type,NEW.osm_id; + -- remove operator tag for most places, messes too much with search_name indexes + IF NEW.class not in ('amenity', 'shop') THEN + NEW.name := delete(NEW.name, 'operator'); + END IF; + -- just block these IF NEW.class in ('landuse','natural') and NEW.name is null THEN -- RAISE WARNING 'empty landuse %',NEW.osm_id; @@@ -2067,11 -2062,6 +2067,11 @@@ BEGI --DEBUG: RAISE WARNING '%', existingplacex; END IF; + -- remove operator tag for most places, messes too much with search_name indexes + IF NEW.class not in ('amenity', 'shop') THEN + NEW.name := delete(NEW.name, 'operator'); + END IF; + -- Just block these - lots and pointless IF NEW.class in ('landuse','natural') and NEW.name is null THEN -- if the name tag was removed, older versions might still be lurking in the place table @@@ -2262,12 -2252,6 +2262,12 @@@ END IF; + -- refuse to update multiplpoygons with too many objects, too much of a performance hit + IF ST_NumGeometries(NEW.geometry) > 2000 THEN + RAISE WARNING 'Dropping update of % % because of geometry complexity.', NEW.osm_type, NEW.osm_id; + RETURN NULL; + END IF; + IF coalesce(existing.name::text, '') != coalesce(NEW.name::text, '') OR coalesce(existing.extratags::text, '') != coalesce(NEW.extratags::text, '') OR coalesce(existing.housenumber, '') != coalesce(NEW.housenumber, '') @@@ -2352,29 -2336,29 +2352,29 @@@ END $$ LANGUAGE plpgsql IMMUTABLE; - CREATE OR REPLACE FUNCTION get_connected_ways(way_ids INTEGER[]) RETURNS SETOF planet_osm_ways - AS $$ - DECLARE - searchnodes INTEGER[]; - location RECORD; - j INTEGER; - BEGIN - - searchnodes := '{}'; - FOR j IN 1..array_upper(way_ids, 1) LOOP - FOR location IN - select nodes from planet_osm_ways where id = way_ids[j] LIMIT 1 - LOOP - IF not (ARRAY[location.nodes] <@ searchnodes) THEN - searchnodes := searchnodes || location.nodes; - END IF; - END LOOP; - END LOOP; - - RETURN QUERY select * from planet_osm_ways where nodes && searchnodes and NOT ARRAY[id] <@ way_ids; - END; - $$ - LANGUAGE plpgsql IMMUTABLE; + --CREATE OR REPLACE FUNCTION get_connected_ways(way_ids INTEGER[]) RETURNS SETOF planet_osm_ways + -- AS $$ + --DECLARE + -- searchnodes INTEGER[]; + -- location RECORD; + -- j INTEGER; + --BEGIN + -- + -- searchnodes := '{}'; + -- FOR j IN 1..array_upper(way_ids, 1) LOOP + -- FOR location IN + -- select nodes from planet_osm_ways where id = way_ids[j] LIMIT 1 + -- LOOP + -- IF not (ARRAY[location.nodes] <@ searchnodes) THEN + -- searchnodes := searchnodes || location.nodes; + -- END IF; + -- END LOOP; + -- END LOOP; + -- + -- RETURN QUERY select * from planet_osm_ways where nodes && searchnodes and NOT ARRAY[id] <@ way_ids; + --END; + --$$ + --LANGUAGE plpgsql IMMUTABLE; CREATE OR REPLACE FUNCTION get_address_postcode(for_place_id BIGINT) RETURNS TEXT AS $$