array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
- array('all', '', 0, 1, 1, 1, 'realpath', 'Do the complete process'),
+ array('osm-file', '', 0, 1, 1, 1, 'realpath', 'File to import'),
+ array('threads', '', 0, 1, 1, 1, 'int', 'Number of threads (where possible)'),
+
+ array('all', '', 0, 1, 0, 0, 'bool', 'Do the complete process'),
array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
- array('import-data', '', 0, 1, 1, 1, 'realpath', 'Import a osm file'),
+ array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
array('create-partitions', '', 0, 1, 0, 0, 'bool', 'Create required partition tables and triggers'),
array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
+ array('import-tiger-data', '', 0, 1, 0, 0, 'bool', 'Import tiger data (not included in \'all\')'),
+ array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
+ array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
);
getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
$bDidSomething = false;
- if ($aCMDResult['create-db'] || isset($aCMDResult['all']))
+ // This is a pretty hard core defult - the number of processors in the box - 1
+ $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
+ if ($iInstances < 1)
+ {
+ $iInstances = 1;
+ echo "WARNING: resetting threads to $iInstances\n";
+ }
+ if ($iInstances > getProcessorCount())
+ {
+ $iInstances = getProcessorCount();
+ echo "WARNING: resetting threads to $iInstances\n";
+ }
+
+ if ($aCMDResult['create-db'] || $aCMDResult['all'])
{
$bDidSomething = true;
$oDB =& DB::connect(CONST_Database_DSN, false);
passthru('createdb nominatim');
}
- if ($aCMDResult['create-db'] || isset($aCMDResult['all']))
+ if ($aCMDResult['create-db'] || $aCMDResult['all'])
{
$bDidSomething = true;
// TODO: path detection, detection memory, etc.
pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
+ pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
+ pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/worldboundaries.sql');
}
- if (isset($aCMDResult['all']) && !isset($aCMDResult['import-data'])) $aCMDResult['import-data'] = $aCMDResult['all'];
- if (isset($aCMDResult['import-data']) && $aCMDResult['import-data'])
+ if ($aCMDResult['import-data'] || $aCMDResult['all'])
{
$bDidSomething = true;
- passthru(CONST_BasePath.'/osm2pgsql/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['import-data']);
+ passthru(CONST_BasePath.'/osm2pgsql/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['osm-file']);
+
+ $oDB =& getDB();
+ $x = $oDB->getRow('select * from place limit 1');
+ if (!$x || PEAR::isError($x)) fail('No Data');
}
- if ($aCMDResult['create-functions'] || isset($aCMDResult['all']))
+ if ($aCMDResult['create-functions'] || $aCMDResult['all'])
{
$bDidSomething = true;
+ if (!file_exists(CONST_BasePath.'/module/nominatim.so')) fail("nominatim module not built");
$sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
$sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
pgsqlRunScript($sTemplate);
}
- if ($aCMDResult['create-tables'] || isset($aCMDResult['all']))
+ if ($aCMDResult['create-tables'] || $aCMDResult['all'])
{
$bDidSomething = true;
pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql');
pgsqlRunScript($sTemplate);
}
- if ($aCMDResult['create-partitions'] || isset($aCMDResult['all']))
+ if ($aCMDResult['create-partitions'] || $aCMDResult['all'])
{
$bDidSomething = true;
$oDB =& getDB();
- $sSQL = 'select distinct country_code from country_name order by country_code';
+ $sSQL = 'select partition from country_name order by country_code';
$aPartitions = $oDB->getCol($sSQL);
if (PEAR::isError($aPartitions))
{
fail($aPartitions->getMessage());
}
- $aPartitions[] = 'none';
+ $aPartitions[] = 0;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/partitions.src.sql');
preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
}
$sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
}
+
pgsqlRunScript($sTemplate);
}
- if ($aCMDResult['load-data'] || isset($aCMDResult['all']))
+ if ($aCMDResult['load-data'] || $aCMDResult['all'])
{
$bDidSomething = true;
$oDB =& getDB();
+ if (!pg_query($oDB->connection, 'TRUNCATE word')) fail(pg_last_error($oDB->connection));
+ echo '.';
if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection));
echo '.';
if (!pg_query($oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($oDB->connection));
echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($oDB->connection));
+ echo '.';
if (!pg_query($oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($oDB->connection));
echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE search_name')) fail(pg_last_error($oDB->connection));
+ echo '.';
+ if (!pg_query($oDB->connection, 'TRUNCATE search_name_blank')) fail(pg_last_error($oDB->connection));
+ echo '.';
if (!pg_query($oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($oDB->connection));
echo '.';
if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
echo '.';
- $iInstances = 16;
$aDBInstances = array();
for($i = 0; $i < $iInstances; $i++)
{
$sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
$sSQL .= 'housenumber, street, isin, postcode, country_code, extratags, ';
$sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i;
+ if ($aCMDResult['verbose']) echo "$sSQL\n";
if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
}
$bAnyBusy = true;
echo "\n";
}
+ if ($aCMDResult['import-tiger-data'])
+ {
+ $bDidSomething = true;
+
+ $aDBInstances = array();
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ $aDBInstances[$i] =& getDB(true);
+ }
+
+ foreach(glob(CONST_BasePath.'/data/tiger2009/*.sql') as $sFile)
+ {
+ echo $sFile.': ';
+ if ((int)basename($sFile) <= 53033) continue;
+ $hFile = fopen($sFile, "r");
+ $sSQL = fgets($hFile, 100000);
+ $iLines = 0;
+
+ while(true)
+ {
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ if (!pg_connection_busy($aDBInstances[$i]->connection))
+ {
+ while(pg_get_result($aDBInstances[$i]->connection));
+ $sSQL = fgets($hFile, 100000);
+ if (!$sSQL) break 2;
+ if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+ $iLines++;
+ if ($iLines == 1000)
+ {
+ echo ".";
+ $iLines = 0;
+ }
+ }
+ }
+ usleep(10);
+ }
+
+ fclose($hFile);
+
+ $bAnyBusy = true;
+ while($bAnyBusy)
+ {
+ $bAnyBusy = false;
+ for($i = 0; $i < $iInstances; $i++)
+ {
+ if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+ }
+ usleep(10);
+ }
+ echo "\n";
+ }
+ }
+
+ if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
+ {
+ $oDB =& getDB();
+ if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection));
+ $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
+ $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,country_code,";
+ $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select country_code,postcode,";
+ $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
+ $sSQL .= "from place where postcode is not null group by country_code,postcode) as x";
+ if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+
+ $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
+ $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
+ $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
+ if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+ }
+
+ if ($aCMDResult['index'] || $aCMDResult['all'])
+ {
+ $bDidSomething = true;
+ passthru(CONST_BasePath.'/nominatim/nominatim -i -d nominatim -t '.$iInstances);
+ }
+
if (!$bDidSomething)
{
showUsage($aCMDOptions, true);
projects / nominatim.git / blobdiff