]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
projects / nominatim.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
update Tiger import script to use 2014 data
[nominatim.git] / utils / setup.php
diff --git a/utils/setup.php b/utils/setup.php
index 42a132bf5c92aab4e93a067c6e8d925ed4a34c57..3c038e66d66f2f2e3f1ac2849a9ea32e108088f1 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -33,7 +33,6 @@
                array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
                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('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
                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('create-roads', '', 0, 1, 0, 0, 'bool', ''),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
                array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
                array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
                array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
                array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
@@ -191,6 +190,14 @@
                {
                        $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
                }
                {
                        $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
                }
+               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';
                $osm2pgsql .= ' -C '.$iCacheMemory;
                $osm2pgsql .= ' -P '.$aDSNInfo['port'];
                $osm2pgsql .= ' -lsc -O gazetteer --hstore';
                $osm2pgsql .= ' -C '.$iCacheMemory;
                $osm2pgsql .= ' -P '.$aDSNInfo['port'];
@@ -216,6 +223,7 @@
                if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
                if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
                pgsqlRunScript($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";
                        $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql');
                if ($fPostgisVersion < 2.0) {
                        echo "Helper functions for postgis < 2.0\n";
                        $sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql');
@@ -258,13 +266,30 @@
 
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
 
        if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
-               echo "Tables\n";
                $bDidSomething = true;
                $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
 
                // re-run the functions
+               echo "Functions\n";
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
                $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);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
@@ -282,6 +307,18 @@
                if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
                if (!$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)
                {
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
@@ -429,40 +466,17 @@
                pgsqlRunScript('ANALYSE');
        }
 
                pgsqlRunScript('ANALYSE');
        }
 
-       if ($aCMDResult['create-roads'])
-       {
-               $bDidSomething = true;
-
-               $oDB =& getDB();
-               $aDBInstances = array();
-               for($i = 0; $i < $iInstances; $i++)
-               {
-                       $aDBInstances[$i] =& getDB(true);
-                       if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
-                       $sSQL = 'select count(*) from (select insertLocationRoad(partition, place_id, calculated_country_code, geometry) from ';
-                       $sSQL .= 'placex where osm_id % '.$iInstances.' = '.$i.' and rank_search between 26 and 27 and class = \'highway\') as x ';
-                       if ($aCMDResult['verbose']) echo "$sSQL\n";
-                       if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
-               }
-               $bAnyBusy = true;
-               while($bAnyBusy)
-               {
-                       $bAnyBusy = false;
-                       for($i = 0; $i < $iInstances; $i++)
-                       {
-                               if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
-                       }
-                       sleep(1);
-                       echo '.';
-               }
-               echo "\n";
-       }
-
        if ($aCMDResult['import-tiger-data'])
        {
                $bDidSomething = true;
 
        if ($aCMDResult['import-tiger-data'])
        {
                $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++)
 
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
@@ -470,7 +484,7 @@
                        $aDBInstances[$i] =& getDB(true);
                }
 
                        $aDBInstances[$i] =& getDB(true);
                }
 
-               foreach(glob(CONST_BasePath.'/data/tiger2011/*.sql') as $sFile)
+               foreach(glob(CONST_BasePath.'/data/tiger2014/*.sql') as $sFile)
                {
                        echo $sFile.': ';
                        $hFile = fopen($sFile, "r");
                {
                        echo $sFile.': ';
                        $hFile = fopen($sFile, "r");
@@ -514,7 +528,13 @@
                }
 
                echo "Creating indexes\n";
                }
 
                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'])
        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
@@ -563,7 +583,7 @@
                        }
 
                        // Find the last node in the DB
                        }
 
                        // Find the last node in the DB
-                       $iLastOSMID = $oDB->getOne("select max(id) from planet_osm_nodes");
+                       $iLastOSMID = $oDB->getOne("select max(osm_id) from place where osm_type = 'N'");
 
                        // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed)
                        $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1";
 
                        // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed)
                        $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1";
@@ -573,49 +593,59 @@
 
                        // Search for the correct state file - uses file timestamps so need to sort by date descending
                        $sRepURL = CONST_Replication_Url."/";
 
                        // 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>
                        // 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, $aRepMatches, PREG_SET_ORDER);
+                       if ($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('#<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 = $aRepMatch;
-                       }
-                       if ($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);
+                               $aPrevRepMatch = false;
+                               foreach($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('#<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)
+                               $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);
+                               $aPrevRepMatch = false;
+                               foreach($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);
                }
        }
 
                }
        }
 
@@ -646,6 +676,12 @@
                if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
                if (!$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)
                {
                preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
                foreach($aMatches as $aMatch)
                {
@@ -757,14 +793,14 @@
 
        }
 
 
        }
 
-       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'];
        {
                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'])
+               if ($bfatal && !$aCMDResult['ignore-errors'])
                        $sCMD .= ' -v ON_ERROR_STOP=1';
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        $sCMD .= ' -v ON_ERROR_STOP=1';
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
@@ -783,7 +819,7 @@
                }
                fclose($ahPipes[0]);
                $iReturn = proc_close($hProcess);
                }
                fclose($ahPipes[0]);
                $iReturn = proc_close($hProcess);
-               if ($iReturn > 0)
+               if ($bfatal && $iReturn > 0)
                {
                        fail("pgsql returned with error code ($iReturn)");
                }
                {
                        fail("pgsql returned with error code ($iReturn)");
                }
@@ -851,3 +887,15 @@
                passthru($cmd, $result);
                if ($result != 0) fail('Error executing external command: '.$cmd);
        }
                passthru($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;
+       }
+
Open Source search based on OpenStreetMap data