Merge remote-tracking branch 'upstream/master'

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index 90474fcf2767f224b0c98dcc1d628b857dd48211..e588bb084a5ff285e60e62ee38be4978eada8843 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -22,10 +22,12 @@
                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('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'),
                array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),
                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'),
                array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),

+               array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"),

                array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
                array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
                array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
                array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
                array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
                array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),


@@ -124,7 +126,17 @@
                        pgsqlRunScript('CREATE EXTENSION hstore');
                }
 
                        pgsqlRunScript('CREATE EXTENSION hstore');
                }
 

-               pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
+               $fPostgisVersion = (float) CONST_Postgis_Version;
+               if ($fPostgisVersion < 2.0) {
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
+                       pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
+               } else {
+                       pgsqlRunScript('CREATE EXTENSION postgis');
+               }
+               if ($fPostgisVersion < 2.1) {
+                       // Function was renamed in 2.1 and throws an annoying deprecation warning
+                       pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
+               }

                $sVersionString = $oDB->getOne('select postgis_full_version()');
                preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches);
                if (CONST_Postgis_Version != $aMatches[1].'.'.$aMatches[2])
                $sVersionString = $oDB->getOne('select postgis_full_version()');
                preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches);
                if (CONST_Postgis_Version != $aMatches[1].'.'.$aMatches[2])


@@ -133,14 +145,33 @@
                        exit;
                }
 
                        exit;
                }
 

-               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/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');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');

+
+               if ($aCMDResult['no-partitions'])
+               {
+                       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'])
        }
 
        if ($aCMDResult['import-data'] || $aCMDResult['all'])


@@ -161,7 +192,7 @@
                }
                $osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';
                $osm2pgsql .= ' -lsc -O gazetteer --hstore';
                }
                $osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';
                $osm2pgsql .= ' -lsc -O gazetteer --hstore';

-               $osm2pgsql .= ' -C 16000';
+               $osm2pgsql .= ' -C 25000';

                $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);
                $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);


@@ -183,6 +214,7 @@
                $sTemplate = str_replace('{modulepath}', CONST_BasePath.'/module', $sTemplate);
                if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
                if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
                $sTemplate = str_replace('{modulepath}', CONST_BasePath.'/module', $sTemplate);
                if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
                if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);

+               if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);

                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 


@@ -233,13 +265,13 @@
                echo "Partition Tables\n";
                $bDidSomething = true;
                $oDB =& getDB();
                echo "Partition Tables\n";
                $bDidSomething = true;
                $oDB =& getDB();

-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';

                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }

-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;

 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);


@@ -262,13 +294,13 @@
                echo "Partition Functions\n";
                $bDidSomething = true;
                $oDB =& getDB();
                echo "Partition Functions\n";
                $bDidSomething = true;
                $oDB =& getDB();

-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';

                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }

-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;

 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);


@@ -338,13 +370,13 @@
                if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
                echo '.';
 
                if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
                echo '.';
 

-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';

                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }

-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;

                foreach($aPartitions as $sPartition)
                {
                        if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
                foreach($aPartitions as $sPartition)
                {
                        if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));


@@ -597,13 +629,13 @@
                echo "Search indices\n";
                $bDidSomething = true;
                $oDB =& getDB();
                echo "Search indices\n";
                $bDidSomething = true;
                $oDB =& getDB();

-               $sSQL = 'select partition from country_name order by country_code';
+               $sSQL = 'select distinct partition from country_name';

                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }
                $aPartitions = $oDB->getCol($sSQL);
                if (PEAR::isError($aPartitions))
                {
                        fail($aPartitions->getMessage());
                }

-               $aPartitions[] = 0;
+               if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;

 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);


@@ -662,10 +694,29 @@
                // Convert database DSN to psql parameters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
                // 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(
 
                $aDescriptors = array(

-                       0 => array('pipe', 'r'),
+                       0 => $aReadPipe,

                        1 => array('pipe', 'w'),
                        2 => array('file', '/dev/null', 'a')
                );
                        1 => array('pipe', 'w'),
                        2 => array('file', '/dev/null', 'a')
                );


@@ -673,7 +724,6 @@
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
                if (!is_resource($hProcess)) fail('unable to start pgsql');
 
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
                if (!is_resource($hProcess)) fail('unable to start pgsql');
 

-               fclose($ahPipes[0]);

 
                // TODO: error checking
                while(!feof($ahPipes[1]))
 
                // TODO: error checking
                while(!feof($ahPipes[1]))


@@ -682,15 +732,28 @@
                }
                fclose($ahPipes[1]);
 
                }
                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)
        {

+               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'];
                // 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 (!$aCMDResult['ignore-errors'])
+                       $sCMD .= ' -v ON_ERROR_STOP=1';

                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 


@@ -703,10 +766,15 @@
                while(strlen($sScript))
                {
                        $written = fwrite($ahPipes[0], $sScript);
                while(strlen($sScript))
                {
                        $written = fwrite($ahPipes[0], $sScript);

+                       if ($written <= 0) break;

                        $sScript = substr($sScript, $written);
                }
                fclose($ahPipes[0]);
                        $sScript = substr($sScript, $written);
                }
                fclose($ahPipes[0]);

-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
+               if ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }

        }
 
        function pgsqlRunRestoreData($sDumpFile)
        }
 
        function pgsqlRunRestoreData($sDumpFile)


@@ -734,7 +802,7 @@
                }
                fclose($ahPipes[1]);
 
                }
                fclose($ahPipes[1]);
 

-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);

        }
 
        function pgsqlRunDropAndRestore($sDumpFile)
        }
 
        function pgsqlRunDropAndRestore($sDumpFile)


@@ -762,7 +830,7 @@
                }
                fclose($ahPipes[1]);
 
                }
                fclose($ahPipes[1]);
 

-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);

        }
 
        function passthruCheckReturn($cmd)
        }
 
        function passthruCheckReturn($cmd)