remove intermediate ANALYSE while indexing

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index db15aa446b640c32f9e481c7d4396fd59aaf80b6..4b60815cd89fad0f41ef15fd7a9a175e677c8e0e 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -23,6 +23,7 @@
                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'),
@@ -210,6 +211,34 @@
                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'] || $aCMDResult['all'])
        {
                echo "Load Data\n";
@@ -235,6 +264,14 @@
                if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
                echo '.';
 
+               // pre-create the word list
+               if (!pg_query($oDB->connection, 'select count(make_keywords(v)) from (select distinct svals(name) as v from place) as w where v is not null;')) fail(pg_last_error($oDB->connection));
+               echo '.';
+               if (!pg_query($oDB->connection, 'select count(make_keywords(v)) from (select distinct postcode as v from place) as w where v is not null;')) fail(pg_last_error($oDB->connection));
+               echo '.';
+               if (!pg_query($oDB->connection, 'select count(getorcreate_housenumber_id(v)) from (select distinct housenumber as v from place where housenumber is not null) as w;')) fail(pg_last_error($oDB->connection));
+               echo '.';
+
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
@@ -329,7 +366,7 @@
                        }
 
                        fclose($hFile);
-       
+
                        $bAnyBusy = true;
                        while($bAnyBusy)
                        {
@@ -394,11 +431,7 @@
                $sOutputFile = '';
                if (isset($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($sBaseCmd.' -r 26');
+               passthru($sBaseCmd);
        }
 
        if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
@@ -454,7 +487,8 @@
 
                // 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'),
@@ -482,7 +516,7 @@
                // 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'];
                $aDescriptors = array(
                        0 => array('pipe', 'r'),
                        1 => STDOUT, 
@@ -500,3 +534,31 @@
                fclose($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);
+       }