ignore frequent partial search terms

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index 07faa2df2a92e4f878be9eba89f6a82d193fdff4..bfc7c7205e47866b00511e763c5f5490ddeab6a6 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -43,6 +43,26 @@
 
        $bDidSomething = false;
 
+       // 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)
@@ -89,9 +109,9 @@
                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.
 
@@ -129,7 +149,7 @@
                $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');
@@ -251,7 +271,7 @@
 
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
-               echo "Load Data\n";
+               echo "Drop old Data\n";
                $bDidSomething = true;
 
                $oDB =& getDB();
@@ -287,17 +307,18 @@
                        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'])
                {
-                       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 '.';
+                       echo "Loading word list\n";
+                       pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
                }
 
+               echo "Load Data\n";
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
@@ -357,6 +378,8 @@
        {
                $bDidSomething = true;
 
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
@@ -405,6 +428,9 @@
                        }
                        echo "\n";
                }
+
+               echo "Creating indexes\n";
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');
        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
@@ -432,7 +458,12 @@
 
                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');
+               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');
+               }
 
                $sDate = $aCMDResult['osmosis-init-date'];
                $aDate = date_parse_from_format("Y-m-d\TH-i", $sDate);
@@ -457,11 +488,11 @@
                $sOutputFile = '';
                if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
                $sBaseCmd = CONST_BasePath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -t '.$iInstances.$sOutputFile;
-               passthru($sBaseCmd.' -R 4');
+               passthruCheckReturn($sBaseCmd.' -R 4');
                if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
-               passthru($sBaseCmd.' -r 5 -R 25');
+               passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
                if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
-               passthru($sBaseCmd.' -r 26');
+               passthruCheckReturn($sBaseCmd.' -r 26');
        }
 
        if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
@@ -625,3 +656,10 @@
 
                proc_close($hProcess);
        }
+
+       function passthruCheckReturn($cmd)
+       {
+               $result = -1;
+               passthru($cmd, $result);
+               if ($result != 0) fail('Error executing external command: '.$cmd);
+       }