]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
more tolerant regexp for parsing replication state directories
[nominatim.git] / utils / setup.php
index 99d9e52628527643a332bb42dcfb1d3d9e980bfc..65c3d88aaf619df098a6025ec9467e8c4adcdffb 100755 (executable)
@@ -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";
                $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";
                        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');
                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');
                {
                        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'])
                $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);
+               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);
        }
 
 
                        // 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:    <a href="000/">000/</a></td><td align="right">26-Feb-2013 11:53  </td>
-                       // planet.openstreetmap.org: <a href="273/">273/</a>                    22-Mar-2013 07:41    -
-                       preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>.*(([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: <a href="273/">273/</a>                    2013-03-11 07:41    -
+                       preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
+                       var_dump($aRepMatches);
+                       if ($aRepMatches)
                        {
-                               if (strtotime($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('#<a href="[0-9]{3}/">([0-9]{3}/)</a>.*(([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;
+                               $aPrevRepMatch = false;
+                               foreach($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('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
+                               var_dump($aRepMatches);
+                               $aPrevRepMatch = false;
+                               foreach($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('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
+                               var_dump($aRepMatches);
+                               $aPrevRepMatch = false;
+                               foreach($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, $sSQL);
                        }
-                       if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
-
-                       $sRepURL .= $aRepMatch[1];
-                       $sRep = file_get_contents($sRepURL."?C=M;O=D");
-                       preg_match_all('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>.*(([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 $aRepMatch)
+                       else
                        {
-                               if (strtotime($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);
                }
        }
 
        {
                showUsage($aCMDOptions, true);
        }
+       else
+       {
+               echo "Setup finished.\n";
+       }
 
        function pgsqlRunScriptFile($sFilename)
        {
                // 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')
                );
                $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
                if (!is_resource($hProcess)) fail('unable to start pgsql');
 
-               fclose($ahPipes[0]);
 
                // TODO: error checking
                while(!feof($ahPipes[1]))
                }
                fclose($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)
        {
+               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 '.$aDSNInfo['database'];
+               if (!$aCMDResult['ignore-errors'])
+                       $sCMD .= ' -v ON_ERROR_STOP=1';
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 
                while(strlen($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 ($iReturn > 0)
+               {
+                       fail("pgsql returned with error code ($iReturn)");
+               }
        }
 
        function pgsqlRunRestoreData($sDumpFile)
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
        }
 
        function pgsqlRunDropAndRestore($sDumpFile)
                }
                fclose($ahPipes[1]);
 
-               proc_close($hProcess);
+               $iReturn = proc_close($hProcess);
        }
 
        function passthruCheckReturn($cmd)