]> git.openstreetmap.org Git - nominatim.git/blobdiff - utils/setup.php
move deletable and polygons into default website
[nominatim.git] / utils / setup.php
index 6415ea123d843358165029d638a2aadb6eab0a31..3fe5a7abec3594f4e455443a23f3e3e1d87c4358 100755 (executable)
                array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
                array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
                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-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
                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('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-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', 'Calculate postcode centroids'),
                array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
-               array('osmosis-init-date', '', 0, 1, 1, 1, 'string', '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('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
                array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
                array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'),
 
        $bDidSomething = false;
 
-       // This is a pretty hard core defult - the number of processors in the box - 1
+       // Check if osm-file is set and points to a valid file if --all or --import-data is given
+       if ($aCMDResult['import-data'] || $aCMDResult['all'])
+       {
+               if (!isset($aCMDResult['osm-file']))
+               {
+                       fail('missing --osm-file for data import');
+               }
+
+               if (!file_exists($aCMDResult['osm-file']))
+               {
+                       fail('the path supplied to --osm-file does not exist');
+               }
+
+               if (!is_readable($aCMDResult['osm-file']))
+               {
+                       fail('osm-file "'.$aCMDResult['osm-file'].'" not readable');
+               }
+       }
+
+
+       // This is a pretty hard core default - the number of processors in the box - 1
        $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
        if ($iInstances < 1)
        {
                echo "WARNING: resetting cache memory to $iCacheMemory\n";
        }
 
-       if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date']))
-       {
-               $sBaseFile = basename($aCMDResult['osm-file']);
-               if (preg_match('#^planet-([0-9]{2})([0-9]{2})([0-9]{2})[.]#', $sBaseFile, $aMatch))
-               {
-                       $iTime = mktime(0, 0, 0, $aMatch[2], $aMatch[3], '20'.$aMatch[1]);
-                       $iTime -= (60*60*24);
-                       $aCMDResult['osmosis-init-date'] = date('Y-m-d', $iTime).'T22:00:00Z';
-               }
-       }
        $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
        if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
@@ -84,9 +98,9 @@
                passthru('createdb -E UTF-8 '.$aDSNInfo['database']);
        }
 
-       if ($aCMDResult['create-db'] || $aCMDResult['all'])
+       if ($aCMDResult['setup-db'] || $aCMDResult['all'])
        {
-               echo "Create DB (2)\n";
+               echo "Setup DB\n";
                $bDidSomething = true;
                // TODO: path detection, detection memory, etc.
 
                $pgver = (float) CONST_Postgresql_Version;
                if ($pgver < 9.1) {
                        pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
+                       pgsqlRunScriptFile(CONST_BasePath.'/sql/hstore_compatability_9_0.sql');
                } else {
                        pgsqlRunScript('CREATE EXTENSION hstore');
                }
                $bDidSomething = true;
 
                $osm2pgsql = CONST_Osm2pgsql_Binary;
-               if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql");
+               if (!file_exists($osm2pgsql))
+               {
+                       echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+                       fail("osm2pgsql not found in '$osm2pgsql'");
+               }
                $osm2pgsql .= ' -lsc -O gazetteer --hstore';
                $osm2pgsql .= ' -C '.$iCacheMemory;
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
-               passthru($osm2pgsql);
+               passthruCheckReturn($osm2pgsql);
 
                $oDB =& getDB();
                $x = $oDB->getRow('select * from place limit 1');
-               if (!$x || PEAR::isError($x)) fail('No Data');
+               if (PEAR::isError($x)) {
+                       fail($x->getMessage());
+               }
+               if (!$x) fail('No Data');
        }
 
        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);
+               $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:', '', $sTemplate);
                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
+       if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all'])
+       {
+               $bDidSomething = true;
+               $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin';
+               $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin';
+               if (file_exists($sWikiArticlesFile))
+               {
+                       echo "Importing wikipedia articles...";
+                       pgsqlRunDropAndRestore($sWikiArticlesFile);
+                       echo "...done\n";
+               }
+               else
+               {
+                       echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+               }
+               if (file_exists($sWikiRedirectsFile))
+               {
+                       echo "Importing wikipedia redirects...";
+                       pgsqlRunDropAndRestore($sWikiRedirectsFile);
+                       echo "...done\n";
+               }
+               else
+               {
+                       echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
+               }
+       }
+
+
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
-               echo "Load Data\n";
+               echo "Drop old Data\n";
                $bDidSomething = true;
 
                $oDB =& getDB();
                if (!pg_query($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';
+               $aPartitions = $oDB->getCol($sSQL);
+               if (PEAR::isError($aPartitions))
+               {
+                       fail($aPartitions->getMessage());
+               }
+               $aPartitions[] = 0;
+               foreach($aPartitions as $sPartition)
+               {
+                       if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
+                       echo '.';
+               }
+
+               // used by getorcreate_word_id to ignore frequent partial words
+               if (!pg_query($oDB->connection, 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS $$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE')) fail(pg_last_error($oDB->connection));
+               echo ".\n";
+
+               // pre-create the word list
+               if (!$aCMDResult['disable-token-precalc'])
+               {
+                       echo "Loading word list\n";
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
+               }
+
+               echo "Load Data\n";
                $aDBInstances = array();
                for($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, country_code, geometry) from ';
+                       $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));
        {
                $bDidSomething = true;
 
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                        }
 
                        fclose($hFile);
-       
+
                        $bAnyBusy = true;
                        while($bAnyBusy)
                        {
                        }
                        echo "\n";
                }
+
+               echo "Creating indexes\n";
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
                $bDidSomething = true;
                $oDB =& getDB();
                if (!pg_query($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 = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
+               $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,calculated_country_code,";
+               $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select calculated_country_code,postcode,";
                $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
-               $sSQL .= "from placex where postcode is not null group by country_code,postcode) as x";
+               $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
                if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
-               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
+               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_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['osmosis-init'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date']))
+       if ($aCMDResult['osmosis-init'] || $aCMDResult['all'])
        {
                $bDidSomething = true;
                $oDB =& getDB();
 
                if (!file_exists(CONST_Osmosis_Binary)) fail("please download osmosis");
-               if (file_exists(CONST_BasePath.'/settings/configuration.txt')) echo "settings/configuration.txt already exists\n";
-               else passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
-
-               $sDate = $aCMDResult['osmosis-init-date'];
-               $aDate = date_parse_from_format("Y-m-d\TH-i", $sDate);
-               $sURL = 'http://toolserver.org/~mazder/replicate-sequences/?';
-               $sURL .= 'Y='.$aDate['year'].'&m='.$aDate['month'].'&d='.$aDate['day'];
-               $sURL .= '&H='.$aDate['hour'].'&i='.$aDate['minute'].'&s=0';
-               $sURL .= '&stream=minute';
-               echo "Getting state file: $sURL\n";
-               $sStateFile = file_get_contents($sURL);
+               if (file_exists(CONST_BasePath.'/settings/configuration.txt'))
+               {
+                       echo "settings/configuration.txt already exists\n";
+               }
+               else
+               {
+                       passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_BasePath.'/settings');
+                       // server layout changed afer license change, fix path to minutely diffs
+                       passthru("sed -i 's:minute-replicate:replication/minute:' ".CONST_BasePath.'/settings/configuration.txt');
+               }
+
+               // Find the last node in the DB
+               $iLastOSMID = $oDB->getOne("select max(osm_id) as 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;
+               $sLastNodeXML = file_get_contents($sLastNodeURL);
+               preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate);
+               $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60);
+
+
+               // Search for the correct state file - uses file timestamps
+               $sRepURL = 'http://planet.openstreetmap.org/replication/minute/';
+               $sRep = file_get_contents($sRepURL);
+               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);
+               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);
+               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 $aRepMatch)
+               {
+                       if (strtotime($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');
-               $sSQL = "INSERT INTO import_status VALUES('".$sDate."')";
+               $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
                pg_query($oDB->connection, $sSQL);
-
        }
 
        if ($aCMDResult['index'] || $aCMDResult['all'])
                $sOutputFile = '';
                if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
                $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$iInstances.$sOutputFile;
-               passthru($sBaseCmd.' -R 4');
-               pgsqlRunScript('ANALYSE');
-               passthru($sBaseCmd.' -r 5 -R 25');
-               pgsqlRunScript('ANALYSE');
-               passthru($sBaseCmd.' -r 26');
+               passthruCheckReturn($sBaseCmd.' -R 4');
+               if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+               passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
+               if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+               passthruCheckReturn($sBaseCmd.' -r 26');
        }
 
        if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
        {
                $bDidSomething = true;
                $sTargetDir = $aCMDResult['create-website'];
-               if (!is_dir($sTargetDir)) fail('please specify a directory to setup');
+               if (!is_dir($sTargetDir))
+               {
+                       echo "You must create the website directory before calling this function.\n";
+                       fail("Target directory does not exist.");
+               }
+
                @symlink(CONST_BasePath.'/website/details.php', $sTargetDir.'/details.php');
                @symlink(CONST_BasePath.'/website/reverse.php', $sTargetDir.'/reverse.php');
                @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/search.php');
                @symlink(CONST_BasePath.'/website/search.php', $sTargetDir.'/index.php');
+               @symlink(CONST_BasePath.'/website/deletable.php', $sTargetDir.'/deletable.php');
+               @symlink(CONST_BasePath.'/website/polygons.php', $sTargetDir.'/polygons.php');
                @symlink(CONST_BasePath.'/website/images', $sTargetDir.'/images');
                @symlink(CONST_BasePath.'/website/js', $sTargetDir.'/js');
+               @symlink(CONST_BasePath.'/website/css', $sTargetDir.'/css');
                echo "Symlinks created\n";
        }
 
        {
                if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 
-               // Convert database DSN to psql paramaters
+               // Convert database DSN to psql parameters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
-               $sCMD = 'psql -f '.$sFilename.' '.$aDSNInfo['database'];
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -f '.$sFilename;
 
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
 
        function pgsqlRunScript($sScript)
        {
-               // Convert database DSN to psql paramaters
+               // 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'].' '.$aDSNInfo['database'];
+               $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 
                fclose($ahPipes[0]);
                proc_close($hProcess);
        }
+
+       function pgsqlRunRestoreData($sDumpFile)
+       {
+               // Convert database DSN to psql parameters
+               $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc -a '.$sDumpFile;
+
+               $aDescriptors = array(
+                       0 => array('pipe', 'r'),
+                       1 => array('pipe', 'w'),
+                       2 => array('file', '/dev/null', 'a')
+               );
+               $ahPipes = null;
+               $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+               if (!is_resource($hProcess)) fail('unable to start pg_restore');
+
+               fclose($ahPipes[0]);
+
+               // TODO: error checking
+               while(!feof($ahPipes[1]))
+               {
+                       echo fread($ahPipes[1], 4096);
+               }
+               fclose($ahPipes[1]);
+
+               proc_close($hProcess);
+       }
+
+       function pgsqlRunDropAndRestore($sDumpFile)
+       {
+               // Convert database DSN to psql parameters
+               $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc --clean '.$sDumpFile;
+
+               $aDescriptors = array(
+                       0 => array('pipe', 'r'),
+                       1 => array('pipe', 'w'),
+                       2 => array('file', '/dev/null', 'a')
+               );
+               $ahPipes = null;
+               $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+               if (!is_resource($hProcess)) fail('unable to start pg_restore');
+
+               fclose($ahPipes[0]);
+
+               // TODO: error checking
+               while(!feof($ahPipes[1]))
+               {
+                       echo fread($ahPipes[1], 4096);
+               }
+               fclose($ahPipes[1]);
+
+               proc_close($hProcess);
+       }
+
+       function passthruCheckReturn($cmd)
+       {
+               $result = -1;
+               passthru($cmd, $result);
+               if ($result != 0) fail('Error executing external command: '.$cmd);
+       }