calculate importance from wikipedia

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index 94888e4430fa28c98dfc1cd4db5c989c7e1d52e0..b9ee9a8996f19fb8cf5374b6b99c3b2057f5df98 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -18,10 +18,12 @@
                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, 0, 0, 'bool', 'Import a osm file'),
                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, 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('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('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
                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('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('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'),


@@ -49,6 +51,15 @@
                $iInstances = getProcessorCount();
                echo "WARNING: resetting threads to $iInstances\n";
        }
                $iInstances = getProcessorCount();
                echo "WARNING: resetting threads to $iInstances\n";
        }

+
+       // Assume we can steal all the cache memory in the box (unless told otherwise)
+       $iCacheMemory = (isset($aCMDResult['osm2pgsql-cache'])?$aCMDResult['osm2pgsql-cache']:getCacheMemoryMB());
+       if ($iCacheMemory > getTotalMemoryMB())
+       {
+               $iCacheMemory = getCacheMemoryMB();
+               echo "WARNING: resetting cache memory to $iCacheMemory\n";
+       }
+

        if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date']))
        {
                $sBaseFile = basename($aCMDResult['osm-file']);
        if (isset($aCMDResult['osm-file']) && !isset($aCMDResult['osmosis-init-date']))
        {
                $sBaseFile = basename($aCMDResult['osm-file']);


@@ -71,7 +82,7 @@
                {
                        fail('database already exists ('.CONST_Database_DSN.')');
                }
                {
                        fail('database already exists ('.CONST_Database_DSN.')');
                }

-               passthru('createdb '.$aDSNInfo['database']);
+               passthru('createdb -E UTF-8 '.$aDSNInfo['database']);

        }
 
        if ($aCMDResult['create-db'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['create-db'] || $aCMDResult['all'])


@@ -82,7 +93,7 @@
 
                $oDB =& getDB();
                passthru('createlang plpgsql '.$aDSNInfo['database']);
 
                $oDB =& getDB();
                passthru('createlang plpgsql '.$aDSNInfo['database']);

-        $pgver = (float) CONST_Postgresql_Version;
+               $pgver = (float) CONST_Postgresql_Version;

                if ($pgver < 9.1) {
                        pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
                } else {
                if ($pgver < 9.1) {
                        pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
                } else {


@@ -107,7 +118,10 @@
 
                $osm2pgsql = CONST_Osm2pgsql_Binary;
                if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql");
 
                $osm2pgsql = CONST_Osm2pgsql_Binary;
                if (!file_exists($osm2pgsql)) fail("please download and build osm2pgsql");

-               passthru($osm2pgsql.' -lsc -O gazetteer -C 12000 --hstore -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file']);
+               $osm2pgsql .= ' -lsc -O gazetteer --hstore';
+               $osm2pgsql .= ' -C '.$iCacheMemory;
+               $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
+               passthru($osm2pgsql);

 
                $oDB =& getDB();
                $x = $oDB->getRow('select * from place limit 1');
 
                $oDB =& getDB();
                $x = $oDB->getRow('select * from place limit 1');


@@ -194,6 +208,34 @@
                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...";
+                       pgsqlRunRestoreData($sWikiArticlesFile);
+                       echo "...done\n";
+               }
+               else
+               {
+                       echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+               }
+               if (file_exists($sWikiRedirectsFile))
+               {
+                       echo "Importing wikipedia redirects...";
+                       pgsqlRunRestoreData($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'] || $aCMDResult['all'])
        {
                echo "Load Data\n";
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
                echo "Load Data\n";


@@ -313,7 +355,7 @@
                        }
 
                        fclose($hFile);
                        }
 
                        fclose($hFile);

-       
+

                        $bAnyBusy = true;
                        while($bAnyBusy)
                        {
                        $bAnyBusy = true;
                        while($bAnyBusy)
                        {


@@ -349,17 +391,27 @@
        if (($aCMDResult['osmosis-init'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date']))
        {
                $bDidSomething = true;
        if (($aCMDResult['osmosis-init'] || $aCMDResult['all']) && isset($aCMDResult['osmosis-init-date']))
        {
                $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'];
 
                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'];

-               $sURL = 'http://toolserver.org/~mazder/replicate-sequences/?'.$sDate;
+               $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 (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
                file_put_contents(CONST_BasePath.'/settings/state.txt', $sStateFile);
                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);

+               echo "Updating DB status\n";
+               pg_query($oDB->connection, 'TRUNCATE import_status');
+               $sSQL = "INSERT INTO import_status VALUES('".$sDate."')";
+               pg_query($oDB->connection, $sSQL);
+

        }
 
        if ($aCMDResult['index'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['index'] || $aCMDResult['all'])


@@ -428,7 +480,8 @@
 
                // Convert database DSN to psql paramaters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 
                // Convert database DSN to psql paramaters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);

-               $sCMD = 'psql -f '.$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'),
 
                $aDescriptors = array(
                        0 => array('pipe', 'r'),


@@ -456,7 +509,7 @@
                // Convert database DSN to psql paramaters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
                // Convert database DSN to psql paramaters
                $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
                if (!isset($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, 
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 


@@ -474,3 +527,31 @@
                fclose($ahPipes[0]);
                proc_close($hProcess);
        }
                fclose($ahPipes[0]);
                proc_close($hProcess);
        }

+
+       function pgsqlRunRestoreData($sDumpFile)
+       {
+               // Convert database DSN to psql paramaters
+               $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
+               if (!isset($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);
+       }