X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/1187929a02ee62ae154f0faed14a5ad853f0674f..0fac696f86fc88be31c583f951abce62230c0945:/utils/setup.php
diff --git a/utils/setup.php b/utils/setup.php
index e23a2211..c45cc857 100755
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -22,6 +22,7 @@
array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'),
array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'),
+ array('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'),
array('create-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
array('create-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'),
@@ -88,6 +89,8 @@
$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+ $fPostgisVersion = (float) CONST_Postgis_Version;
+
if ($aCMDResult['create-db'] || $aCMDResult['all'])
{
echo "Create DB\n";
@@ -109,7 +112,7 @@
$oDB =& getDB();
$sVersionString = $oDB->getOne('select version()');
- preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[.]([0-9]+) #', $sVersionString, $aMatches);
+ preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[^0-9]#', $sVersionString, $aMatches);
if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2])
{
echo "ERROR: PostgreSQL version is not correct. Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
@@ -125,7 +128,6 @@
pgsqlRunScript('CREATE EXTENSION hstore');
}
- $fPostgisVersion = (float) CONST_Postgis_Version;
if ($fPostgisVersion < 2.0) {
pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
@@ -147,7 +149,15 @@
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/gb_postcode_table.sql');
+ if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz'))
+ {
+ pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
+ }
+ else
+ {
+ echo "WARNING: external UK postcode table not found.\n";
+ }
pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
@@ -156,6 +166,13 @@
{
pgsqlRunScript('update country_name set partition = 0');
}
+
+ // the following will be needed by create_functions later but
+ // is only defined in the subsequently called create_tables.
+ // Create dummies here that will be overwritten by the proper
+ // versions in create-tables.
+ pgsqlRunScript('CREATE TABLE place_boundingbox ()');
+ pgsqlRunScript('create type wikipedia_article_match as ()');
}
if ($aCMDResult['import-data'] || $aCMDResult['all'])
@@ -174,9 +191,16 @@
{
$osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
}
- $osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';
+ if (CONST_Tablespace_Osm2pgsql_Data)
+ $osm2pgsql .= ' --tablespace-slim-data '.CONST_Tablespace_Osm2pgsql_Data;
+ if (CONST_Tablespace_Osm2pgsql_Index)
+ $osm2pgsql .= ' --tablespace-slim-index '.CONST_Tablespace_Osm2pgsql_Index;
+ if (CONST_Tablespace_Place_Data)
+ $osm2pgsql .= ' --tablespace-main-data '.CONST_Tablespace_Place_Data;
+ if (CONST_Tablespace_Place_Index)
+ $osm2pgsql .= ' --tablespace-main-index '.CONST_Tablespace_Place_Index;
$osm2pgsql .= ' -lsc -O gazetteer --hstore';
- $osm2pgsql .= ' -C 18000';
+ $osm2pgsql .= ' -C 25000';
$osm2pgsql .= ' -P '.$aDSNInfo['port'];
$osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
passthruCheckReturn($osm2pgsql);
@@ -200,6 +224,15 @@
if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
pgsqlRunScript($sTemplate);
+
+ if ($fPostgisVersion < 2.0) {
+ echo "Helper functions for postgis < 2.0\n";
+ $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql');
+ } else {
+ echo "Helper functions for postgis >= 2.0\n";
+ $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_20_aux.sql');
+ }
+ pgsqlRunScript($sTemplate);
}
if ($aCMDResult['create-minimal-tables'])
@@ -234,13 +267,30 @@
if ($aCMDResult['create-tables'] || $aCMDResult['all'])
{
- echo "Tables\n";
$bDidSomething = true;
- pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql');
+
+ echo "Tables\n";
+ $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql');
+ $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:address-data}',
+ CONST_Tablespace_Address_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:address-index}',
+ CONST_Tablespace_Address_Index, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:search-data}',
+ CONST_Tablespace_Search_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:search-index}',
+ CONST_Tablespace_Search_Index, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-data}',
+ CONST_Tablespace_Aux_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-index}',
+ CONST_Tablespace_Aux_Index, $sTemplate);
+ pgsqlRunScript($sTemplate, false);
// re-run the functions
+ echo "Functions\n";
$sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
- $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
+ $sTemplate = str_replace('{modulepath}',
+ CONST_BasePath.'/module', $sTemplate);
pgsqlRunScript($sTemplate);
}
@@ -258,6 +308,18 @@
if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
+ $sTemplate = replace_tablespace('{ts:address-data}',
+ CONST_Tablespace_Address_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:address-index}',
+ CONST_Tablespace_Address_Index, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:search-data}',
+ CONST_Tablespace_Search_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:search-index}',
+ CONST_Tablespace_Search_Index, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-data}',
+ CONST_Tablespace_Aux_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-index}',
+ CONST_Tablespace_Aux_Index, $sTemplate);
preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
foreach($aMatches as $aMatch)
{
@@ -438,7 +500,13 @@
{
$bDidSomething = true;
- pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+ $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
+ $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-data}',
+ CONST_Tablespace_Aux_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-index}',
+ CONST_Tablespace_Aux_Index, $sTemplate);
+ pgsqlRunScript($sTemplate, false);
$aDBInstances = array();
for($i = 0; $i < $iInstances; $i++)
@@ -490,7 +558,13 @@
}
echo "Creating indexes\n";
- pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
+ $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql');
+ $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-data}',
+ CONST_Tablespace_Aux_Data, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-index}',
+ CONST_Tablespace_Aux_Index, $sTemplate);
+ pgsqlRunScript($sTemplate, false);
}
if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
@@ -549,49 +623,59 @@
// Search for the correct state file - uses file timestamps so need to sort by date descending
$sRepURL = CONST_Replication_Url."/";
- $sRep = file_get_contents($sRepURL."?C=M;O=D");
+ $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
// download.geofabrik.de: 000/| 26-Feb-2013 11:53 |
- // planet.openstreetmap.org: 273/ 22-Mar-2013 07:41 -
- preg_match_all('#([0-9]{3}/).*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER);
- $aPrevRepMatch = false;
- foreach($aRepMatches as $aRepMatch)
+ // planet.openstreetmap.org: 273/ 2013-03-11 07:41 -
+ preg_match_all('#([0-9]{3}/)\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
+ if ($aRepMatches)
{
- if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
- $aPrevRepMatch = $aRepMatch;
- }
- if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
+ $aPrevRepMatch = false;
+ foreach($aRepMatches as $aRepMatch)
+ {
+ if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
+ $aPrevRepMatch = $aRepMatch;
+ }
+ if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
- $sRepURL .= $aRepMatch[1];
- $sRep = file_get_contents($sRepURL."?C=M;O=D");
- preg_match_all('#([0-9]{3}/).*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER);
- $aPrevRepMatch = false;
- foreach($aRepMatches as $aRepMatch)
- {
- if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
- $aPrevRepMatch = $aRepMatch;
- }
- if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
+ $sRepURL .= $aRepMatch[1];
+ $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
+ preg_match_all('#([0-9]{3}/)\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
+ $aPrevRepMatch = false;
+ foreach($aRepMatches as $aRepMatch)
+ {
+ if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
+ $aPrevRepMatch = $aRepMatch;
+ }
+ if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
- $sRepURL .= $aRepMatch[1];
- $sRep = file_get_contents($sRepURL."?C=M;O=D");
- preg_match_all('#([0-9]{3}).state.txt.*(([0-9]{2})-([A-z]{3})-([0-9]{4}) ([0-9]{2}):([0-9]{2}))#', $sRep, $aRepMatches, PREG_SET_ORDER);
- $aPrevRepMatch = false;
- foreach($aRepMatches as $aRepMatch)
+ $sRepURL .= $aRepMatch[1];
+ $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
+ preg_match_all('#([0-9]{3}).state.txt\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
+ $aPrevRepMatch = false;
+ foreach($aRepMatches as $aRepMatch)
+ {
+ if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
+ $aPrevRepMatch = $aRepMatch;
+ }
+ if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
+
+ $sRepURL .= $aRepMatch[1].'.state.txt';
+ echo "Getting state file: $sRepURL\n";
+ $sStateFile = file_get_contents($sRepURL);
+ if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
+ file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
+ echo "Updating DB status\n";
+ pg_query($oDB->connection, 'TRUNCATE import_status');
+ $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
+ pg_query($oDB->connection, $sSQL);
+ }
+ else
{
- if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
- $aPrevRepMatch = $aRepMatch;
+ if (!$aCMDResult['all'])
+ {
+ fail("Cannot read state file directory.");
+ }
}
- if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
-
- $sRepURL .= $aRepMatch[1].'.state.txt';
- echo "Getting state file: $sRepURL\n";
- $sStateFile = file_get_contents($sRepURL);
- if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
- file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
- echo "Updating DB status\n";
- pg_query($oDB->connection, 'TRUNCATE import_status');
- $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
- pg_query($oDB->connection, $sSQL);
}
}
@@ -622,6 +706,12 @@
if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
+ $sTemplate = replace_tablespace('{ts:address-index}',
+ CONST_Tablespace_Address_Index, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:search-index}',
+ CONST_Tablespace_Search_Index, $sTemplate);
+ $sTemplate = replace_tablespace('{ts:aux-index}',
+ CONST_Tablespace_Aux_Index, $sTemplate);
preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
foreach($aMatches as $aMatch)
{
@@ -670,6 +760,10 @@
{
showUsage($aCMDOptions, true);
}
+ else
+ {
+ echo "Setup finished.\n";
+ }
function pgsqlRunScriptFile($sFilename)
{
@@ -678,10 +772,29 @@
// Convert database DSN to psql parameters
$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
- $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -f '.$sFilename;
+ $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+
+ $ahGzipPipes = null;
+ if (preg_match('/\\.gz$/', $sFilename))
+ {
+ $aDescriptors = array(
+ 0 => array('pipe', 'r'),
+ 1 => array('pipe', 'w'),
+ 2 => array('file', '/dev/null', 'a')
+ );
+ $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes);
+ if (!is_resource($hGzipProcess)) fail('unable to start zcat');
+ $aReadPipe = $ahGzipPipes[1];
+ fclose($ahGzipPipes[0]);
+ }
+ else
+ {
+ $sCMD .= ' -f '.$sFilename;
+ $aReadPipe = array('pipe', 'r');
+ }
$aDescriptors = array(
- 0 => array('pipe', 'r'),
+ 0 => $aReadPipe,
1 => array('pipe', 'w'),
2 => array('file', '/dev/null', 'a')
);
@@ -689,7 +802,6 @@
$hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
if (!is_resource($hProcess)) fail('unable to start pgsql');
- fclose($ahPipes[0]);
// TODO: error checking
while(!feof($ahPipes[1]))
@@ -698,15 +810,28 @@
}
fclose($ahPipes[1]);
- proc_close($hProcess);
+ $iReturn = proc_close($hProcess);
+ if ($iReturn > 0)
+ {
+ fail("pgsql returned with error code ($iReturn)");
+ }
+ if ($ahGzipPipes)
+ {
+ fclose($ahGzipPipes[1]);
+ proc_close($hGzipProcess);
+ }
+
}
- function pgsqlRunScript($sScript)
+ function pgsqlRunScript($sScript, $bfatal = true)
{
+ global $aCMDResult;
// Convert database DSN to psql parameters
$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
$sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+ if ($bfatal && !$aCMDResult['ignore-errors'])
+ $sCMD .= ' -v ON_ERROR_STOP=1';
$aDescriptors = array(
0 => array('pipe', 'r'),
1 => STDOUT,
@@ -719,10 +844,15 @@
while(strlen($sScript))
{
$written = fwrite($ahPipes[0], $sScript);
+ if ($written <= 0) break;
$sScript = substr($sScript, $written);
}
fclose($ahPipes[0]);
- proc_close($hProcess);
+ $iReturn = proc_close($hProcess);
+ if ($bfatal && $iReturn > 0)
+ {
+ fail("pgsql returned with error code ($iReturn)");
+ }
}
function pgsqlRunRestoreData($sDumpFile)
@@ -750,7 +880,7 @@
}
fclose($ahPipes[1]);
- proc_close($hProcess);
+ $iReturn = proc_close($hProcess);
}
function pgsqlRunDropAndRestore($sDumpFile)
@@ -778,7 +908,7 @@
}
fclose($ahPipes[1]);
- proc_close($hProcess);
+ $iReturn = proc_close($hProcess);
}
function passthruCheckReturn($cmd)
@@ -787,3 +917,15 @@
passthru($cmd, $result);
if ($result != 0) fail('Error executing external command: '.$cmd);
}
+
+ function replace_tablespace($sTemplate, $sTablespace, $sSql)
+ {
+ if ($sTablespace)
+ $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"',
+ $sSql);
+ else
+ $sSql = str_replace($sTemplate, '', $sSql);
+
+ return $sSql;
+ }
+