]> 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
* use osm2pgsql from $PATH if none present locally
[nominatim.git] / utils / setup.php
diff --git a/utils/setup.php b/utils/setup.php
index 4f3886f00bb6da28f7ac2311fbca9b17aab1053e..c7667b15902e3b1ea8e9223bef8be6759fdd57a0 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -10,22 +10,56 @@
                array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
                array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
 
                array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
                array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
 
-               array('all', '', 0, 1, 1, 1, 'realpath', 'Do the complete process'),
+               array('osm-file', '', 0, 1, 1, 1, 'realpath', 'File to import'),
+               array('threads', '', 0, 1, 1, 1, 'int', 'Number of threads (where possible)'),
+
+               array('all', '', 0, 1, 0, 0, 'bool', 'Do the complete process'),
 
                array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
                array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
 
                array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
                array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
-               array('import-data', '', 0, 1, 1, 1, 'realpath', 'Import a osm file'),
+               array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
                array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
                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('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
                array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
                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('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
+               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-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
        $bDidSomething = false;
 
        );
        getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
 
        $bDidSomething = false;
 
-       if ($aCMDResult['create-db'] || isset($aCMDResult['all']))
+       // This is a pretty hard core defult - the number of processors in the box - 1
+       $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
+       if ($iInstances < 1)
+       {
+               $iInstances = 1;
+               echo "WARNING: resetting threads to $iInstances\n";
+       }
+       if ($iInstances > getProcessorCount())
+       {
+               $iInstances = getProcessorCount();
+               echo "WARNING: resetting threads to $iInstances\n";
+       }
+       if (isset($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', $iTime).'T22:00:00Z';
+               }
+       }
+
+       if ($aCMDResult['create-db'] || $aCMDResult['all'])
        {
        {
+               echo "Create DB\n";
                $bDidSomething = true;
                $oDB =& DB::connect(CONST_Database_DSN, false);
                if (!PEAR::isError($oDB))
                $bDidSomething = true;
                $oDB =& DB::connect(CONST_Database_DSN, false);
                if (!PEAR::isError($oDB))
@@ -35,8 +69,9 @@
                passthru('createdb nominatim');
        }
 
                passthru('createdb nominatim');
        }
 
-       if ($aCMDResult['create-db'] || isset($aCMDResult['all']))
+       if ($aCMDResult['create-db'] || $aCMDResult['all'])
        {
        {
+               echo "Create DB (2)\n";
                $bDidSomething = true;
                // TODO: path detection, detection memory, etc.
 
                $bDidSomething = true;
                // TODO: path detection, detection memory, etc.
 
@@ -47,31 +82,43 @@
                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_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_naturaleathdata.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/us_statecounty.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
                pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode.sql');
                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/worldboundaries.sql');
        }
 
                pgsqlRunScriptFile(CONST_BasePath.'/data/worldboundaries.sql');
        }
 
-       if (isset($aCMDResult['all']) && !isset($aCMDResult['import-data'])) $aCMDResult['import-data'] = $aCMDResult['all'];
-       if (isset($aCMDResult['import-data']) && $aCMDResult['import-data'])
+       if ($aCMDResult['import-data'] || $aCMDResult['all'])
        {
        {
+               echo "Import\n";
                $bDidSomething = true;
                $bDidSomething = true;
-               passthru(CONST_BasePath.'/osm2pgsql/osm2pgsql -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['import-data']);
+
+        $osm2pgsql = CONST_BasePath.'/osm2pgsql/osm2pgsql';
+        if (!file_exists($osm2pgsql)) $osm2pgsql = trim(`which osm2pgsql`);
+        if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql");
+        passthru($osm2pgsql.' -lsc -O gazetteer -C 10000 --hstore -d nominatim '.$aCMDResult['osm-file']);
+
+               $oDB =& getDB();
+               $x = $oDB->getRow('select * from place limit 1');
+               if (!$x || PEAR::isError($x)) fail('No Data');
        }
 
        }
 
-       if ($aCMDResult['create-functions'] || isset($aCMDResult['all']))
+       if ($aCMDResult['create-functions'] || $aCMDResult['all'])
        {
        {
+               echo "Functions\n";
                $bDidSomething = true;
                $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', $sTemplate);
                pgsqlRunScript($sTemplate);
        }
 
                $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
                $sTemplate = str_replace('{modulepath}',CONST_BasePath.'/module', $sTemplate);
                pgsqlRunScript($sTemplate);
        }
 
-       if ($aCMDResult['create-tables'] || isset($aCMDResult['all']))
+       if ($aCMDResult['create-tables'] || $aCMDResult['all'])
        {
        {
+               echo "Tables\n";
                $bDidSomething = true;
                pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql');
 
                $bDidSomething = true;
                pgsqlRunScriptFile(CONST_BasePath.'/sql/tables.sql');
 
@@ -81,8 +128,9 @@
                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
-       if ($aCMDResult['create-partitions'] || isset($aCMDResult['all']))
+       if ($aCMDResult['create-partitions'] || $aCMDResult['all'])
        {
        {
+               echo "Partitions\n";
                $bDidSomething = true;
                $oDB =& getDB();
                $sSQL = 'select partition from country_name order by country_code';
                $bDidSomething = true;
                $oDB =& getDB();
                $sSQL = 'select partition from country_name order by country_code';
@@ -104,11 +152,13 @@
                        }
                        $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
                }
                        }
                        $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
                }
+
                pgsqlRunScript($sTemplate);
        }
 
                pgsqlRunScript($sTemplate);
        }
 
-       if ($aCMDResult['load-data'] || isset($aCMDResult['all']))
+       if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
        {
+               echo "Load Data\n";
                $bDidSomething = true;
 
                $oDB =& getDB();
                $bDidSomething = true;
 
                $oDB =& getDB();
@@ -131,7 +181,6 @@
                if (!pg_query($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 '.';
 
-               $iInstances = 16;
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
@@ -139,7 +188,36 @@
                        $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
                        $sSQL .= 'housenumber, street, isin, postcode, country_code, extratags, ';
                        $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i;
                        $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
                        $sSQL .= 'housenumber, street, isin, postcode, country_code, extratags, ';
                        $sSQL .= 'geometry) select * from place where osm_id % '.$iInstances.' = '.$i;
-var_dump($sSQL);
+                       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['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, 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));
                }
                $bAnyBusy = true;
                        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
                }
                $bAnyBusy = true;
@@ -156,6 +234,102 @@ var_dump($sSQL);
                echo "\n";
        }
 
                echo "\n";
        }
 
+       if ($aCMDResult['import-tiger-data'])
+       {
+               $bDidSomething = true;
+
+               $aDBInstances = array();
+               for($i = 0; $i < $iInstances; $i++)
+               {
+                       $aDBInstances[$i] =& getDB(true);
+               }
+
+               foreach(glob(CONST_BasePath.'/data/tiger2009/*.sql') as $sFile)
+               {
+                       echo $sFile.': ';
+                       $hFile = fopen($sFile, "r");
+                       $sSQL = fgets($hFile, 100000);
+                       $iLines = 0;
+
+                       while(true)
+                       {
+                               for($i = 0; $i < $iInstances; $i++)
+                               {
+                                       if (!pg_connection_busy($aDBInstances[$i]->connection))
+                                       {
+                                               while(pg_get_result($aDBInstances[$i]->connection));
+                                               $sSQL = fgets($hFile, 100000);
+                                               if (!$sSQL) break 2;
+                                               if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+                                               $iLines++;
+                                               if ($iLines == 1000)
+                                               {
+                                                       echo ".";
+                                                       $iLines = 0;
+                                               }
+                                       }
+                               }
+                               usleep(10);
+                       }
+
+                       fclose($hFile);
+       
+                       $bAnyBusy = true;
+                       while($bAnyBusy)
+                       {
+                               $bAnyBusy = false;
+                               for($i = 0; $i < $iInstances; $i++)
+                               {
+                                       if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+                               }
+                               usleep(10);
+                       }
+                       echo "\n";
+               }
+       }
+
+       if ($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 .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
+               $sSQL .= "from placex where postcode is not null group by country_code,postcode) as x";
+               if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+
+               $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,country_code,geometry) ";
+               $sSQL .= "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']))
+       {
+               $bDidSomething = true;
+
+               if (!file_exists(CONST_BasePath.'/osmosis-0.38/bin/osmosis')) fail("please download osmosis");
+               if (file_exists(CONST_BasePath.'/settings/configuration.txt')) echo "settings/configuration.txt already exists\n";
+               else passthru(CONST_BasePath.'/osmosis-0.38/bin/osmosis --read-replication-interval-init '.CONST_BasePath.'/settings');
+
+               $sDate = $aCMDResult['osmosis-init-date'];
+               $sURL = 'http://toolserver.org/~mazder/replicate-sequences/?'.$sDate;
+               echo "Getting state file: $sURL\n";
+               $sStateFile = file_get_contents($sURL);
+               if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
+               file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
+       }
+
+       if ($aCMDResult['index'] || $aCMDResult['all'])
+       {
+               $bDidSomething = true;
+               $sOutputFile = '';
+               if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
+               passthru(CONST_BasePath.'/nominatim/nominatim -i -d nominatim -t '.$iInstances.$sOutputFile);
+       }
+
        if (!$bDidSomething)
        {
                showUsage($aCMDOptions, true);
        if (!$bDidSomething)
        {
                showUsage($aCMDOptions, true);
@@ -194,8 +368,8 @@ var_dump($sSQL);
        {
                // Convert database DSN to psql paramaters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
        {
                // Convert database DSN to psql paramaters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
-               $sCMD = 'psql '.$aDSNInfo['database'];
-
+               if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
+               $sCMD = 'psql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database'];
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => array('pipe', 'w'),
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => array('pipe', 'w'),
Open Source search based on OpenStreetMap data