use calculated_country_code for postcodes

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index c842ba33e40b927b2ff1ab23fc0770dcb608b236..2542bd9e1289831a20187d70987a214d854ed453 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -43,6 +43,26 @@
 
        $bDidSomething = false;
 
 
        $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)
        // 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']);
        }
 
                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.
 
                $bDidSomething = true;
                // TODO: path detection, detection memory, etc.
 


@@ -126,8 +146,9 @@
                        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 '$osm2pgsql'");
                }
                        echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
                        fail("osm2pgsql not found in '$osm2pgsql'");
                }

+               $osm2pgsql .= ' --tablespace-slim-index ssd --tablespace-main-index ssd --tablespace-main-data ssd --tablespace-slim-data data';

                $osm2pgsql .= ' -lsc -O gazetteer --hstore';
                $osm2pgsql .= ' -lsc -O gazetteer --hstore';

-               $osm2pgsql .= ' -C '.$iCacheMemory;
+               $osm2pgsql .= ' -C 16000';

                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);
 
                $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
                passthruCheckReturn($osm2pgsql);
 


@@ -251,7 +272,7 @@
 
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {
 
        if ($aCMDResult['load-data'] || $aCMDResult['all'])
        {

-               echo "Load Data\n";
+               echo "Drop old Data\n";

                $bDidSomething = true;
 
                $oDB =& getDB();
                $bDidSomething = true;
 
                $oDB =& getDB();


@@ -287,17 +308,18 @@
                        echo '.';
                }
 
                        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 (!$aCMDResult['disable-token-precalc'])
                {
                // pre-create the word list
                if (!$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++)
                {
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {


@@ -320,6 +342,7 @@
                        echo '.';
                }
                echo "\n";
                        echo '.';
                }
                echo "\n";

+               pgsqlRunScript('ALTER TABLE place SET TABLESPACE "data"');

                echo "Reanalysing database...\n";
                pgsqlRunScript('ANALYSE');
        }
                echo "Reanalysing database...\n";
                pgsqlRunScript('ANALYSE');
        }


@@ -334,7 +357,7 @@
                {
                        $aDBInstances[$i] =& getDB(true);
                        if (!pg_query($aDBInstances[$i]->connection, 'set enable_bitmapscan = off')) fail(pg_last_error($oDB->connection));
                {
                        $aDBInstances[$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));
                        $sSQL .= 'placex where osm_id % '.$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));


@@ -357,6 +380,8 @@
        {
                $bDidSomething = true;
 
        {
                $bDidSomething = true;
 

+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_start.sql');
+

                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {
                $aDBInstances = array();
                for($i = 0; $i < $iInstances; $i++)
                {


@@ -405,6 +430,9 @@
                        }
                        echo "\n";
                }
                        }
                        echo "\n";
                }

+
+               echo "Creating indexes\n";
+               pgsqlRunScriptFile(CONST_BasePath.'/sql/tiger_import_finish.sql');

        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
        }
 
        if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])


@@ -412,14 +440,14 @@
                $bDidSomething = true;
                $oDB =& getDB();
                if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection));
                $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 .= "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));
 
                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,calcuclated_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));
                $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));


@@ -432,7 +460,12 @@
 
                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";
 
                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);
 
                $sDate = $aCMDResult['osmosis-init-date'];
                $aDate = date_parse_from_format("Y-m-d\TH-i", $sDate);