Merge pull request #724 from lonvia/update-country-list

[nominatim.git] / utils / setup.php

diff --git a/utils/setup.php b/utils/setup.php
index fc5b8e7470274294987da4cc20b22aae43878589..b0a3bf600bbfcdccfeeec4a03c45d35e5cb1fca2 100755 (executable)
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -38,6 +38,7 @@ $aCMDOptions
    array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
    array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
    array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
    array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
    array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
    array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),

+   array('create-country-names', '', 0, 1, 0, 0, 'bool', 'Create default list of searchable country names'),

    array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'),
   );
 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
    array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'),
   );
 getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);


@@ -94,8 +95,9 @@ if ($aCMDResult['create-db'] || $aCMDResult['all']) {
 if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
     echo "Setup DB\n";
     $bDidSomething = true;
 if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
     echo "Setup DB\n";
     $bDidSomething = true;

-    // TODO: path detection, detection memory, etc.

 
 

+    // TODO: path detection, detection memory, etc.
+    //

     $oDB =& getDB();
 
     $fPostgresVersion = getPostgresVersion($oDB);
     $oDB =& getDB();
 
     $fPostgresVersion = getPostgresVersion($oDB);


@@ -122,13 +124,23 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
     echo 'Postgis version found: '.$fPostgisVersion."\n";
 
     if ($fPostgisVersion < 2.1) {
     echo 'Postgis version found: '.$fPostgisVersion."\n";
 
     if ($fPostgisVersion < 2.1) {

-        // Function was renamed in 2.1 and throws an annoying deprecation warning
+        // Functions were renamed in 2.1 and throw an annoying deprecation warning

         pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
         pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');

+        pgsqlRunScript('ALTER FUNCTION ST_Line_Locate_Point(geometry, geometry) RENAME TO ST_LineLocatePoint');
+    }
+    if ($fPostgisVersion < 2.2) {
+        pgsqlRunScript('ALTER FUNCTION ST_Distance_Spheroid(geometry, geometry, spheroid) RENAME TO ST_DistanceSpheroid');
+    }
+
+    if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) {
+        echo "Error: you need to download the country_osm_grid first:";
+        echo "\n    wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz http://www.nominatim.org/data/country_grid.sql.gz\n";
+        exit(1);

     }
 
     pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
     pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
     }
 
     pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
     pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');

-    pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
+    pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql.gz');

     pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
     if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) {
         pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
     pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
     if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) {
         pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');


@@ -157,7 +169,8 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
 
     $osm2pgsql = CONST_Osm2pgsql_Binary;
     if (!file_exists($osm2pgsql)) {
 
     $osm2pgsql = CONST_Osm2pgsql_Binary;
     if (!file_exists($osm2pgsql)) {

-        echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
+        echo "Check CONST_Osm2pgsql_Binary in your local settings file.\n";
+        echo "Normally you should not need to set this manually.\n";

         fail("osm2pgsql not found in '$osm2pgsql'");
     }
 
         fail("osm2pgsql not found in '$osm2pgsql'");
     }
 


@@ -179,7 +192,7 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
     passthruCheckReturn($osm2pgsql);
 
     $oDB =& getDB();
     passthruCheckReturn($osm2pgsql);
 
     $oDB =& getDB();

-    if (!chksql($oDB->getRow('select * from place limit 1'))) {
+    if (!$aCMDResult['ignore-errors'] && !chksql($oDB->getRow('select * from place limit 1'))) {

         fail('No Data');
     }
 }
         fail('No Data');
     }
 }


@@ -285,8 +298,8 @@ if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) {
 
 if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all']) {
     $bDidSomething = true;
 
 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';
+    $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin';
+    $sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin';

     if (file_exists($sWikiArticlesFile)) {
         echo "Importing wikipedia articles...";
         pgsqlRunDropAndRestore($sWikiArticlesFile);
     if (file_exists($sWikiArticlesFile)) {
         echo "Importing wikipedia articles...";
         pgsqlRunDropAndRestore($sWikiArticlesFile);


@@ -339,7 +352,11 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
     }
 
     // used by getorcreate_word_id to ignore frequent partial words
     }
 
     // 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));
+    $sSQL = 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS ';
+    $sSQL .= '$$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE';
+    if (!pg_query($oDB->connection, $sSQL)) {
+        fail(pg_last_error($oDB->connection));
+    }

     echo ".\n";
 
     // pre-create the word list
     echo ".\n";
 
     // pre-create the word list


@@ -349,24 +366,31 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
     }
 
     echo "Load Data\n";
     }
 
     echo "Load Data\n";

+    $sColumns = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry';
+

     $aDBInstances = array();
     $iLoadThreads = max(1, $iInstances - 1);
     for ($i = 0; $i < $iLoadThreads; $i++) {
         $aDBInstances[$i] =& getDB(true);
     $aDBInstances = array();
     $iLoadThreads = max(1, $iInstances - 1);
     for ($i = 0; $i < $iLoadThreads; $i++) {
         $aDBInstances[$i] =& getDB(true);

-        $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
-        $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, ';
-        $sSQL .= 'geometry) select * from place where osm_id % '.$iLoadThreads.' = '.$i;
-        $sSQL .= " and not (class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString')";
+        $sSQL = "INSERT INTO placex ($sColumns) SELECT $sColumns FROM place WHERE osm_id % $iLoadThreads = $i";
+        $sSQL .= " and not (class='place' and type='houses' and osm_type='W'";
+        $sSQL .= "          and ST_GeometryType(geometry) = 'ST_LineString')";
+        $sSQL .= " and ST_IsValid(geometry)";

         if ($aCMDResult['verbose']) echo "$sSQL\n";
         if ($aCMDResult['verbose']) echo "$sSQL\n";

-        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+        if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) {
+            fail(pg_last_error($aDBInstances[$i]->connection));
+        }

     }
     // last thread for interpolation lines
     $aDBInstances[$iLoadThreads] =& getDB(true);
     }
     // last thread for interpolation lines
     $aDBInstances[$iLoadThreads] =& getDB(true);

-    $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, ';
-    $sSQL .= 'geometry) from place where ';
+    $sSQL = 'insert into location_property_osmline';
+    $sSQL .= ' (osm_id, address, linegeo)';
+    $sSQL .= ' SELECT osm_id, address, geometry from place where ';

     $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'";
     if ($aCMDResult['verbose']) echo "$sSQL\n";
     $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'";
     if ($aCMDResult['verbose']) echo "$sSQL\n";

-    if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+    if (!pg_send_query($aDBInstances[$iLoadThreads]->connection, $sSQL)) {
+        fail(pg_last_error($aDBInstances[$iLoadThreads]->connection));
+    }

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


@@ -460,15 +484,16 @@ 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));
     $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,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 = "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 .= "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 calculated_country_code,postcode) as x";
+    $sSQL .= "from placex where postcode is not null group by country_code,postcode) as x ";
+    $sSQL .= "where ST_Point(x,y) is not null";

     if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
     if (CONST_Use_Extra_US_Postcodes) {
     if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 
     if (CONST_Use_Extra_US_Postcodes) {

-        $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
+        $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));
         $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));


@@ -589,6 +614,32 @@ if ($aCMDResult['create-search-indices'] || $aCMDResult['all']) {
     pgsqlRunScript($sTemplate);
 }
 
     pgsqlRunScript($sTemplate);
 }
 

+if ($aCMDResult['create-country-names'] || $aCMDResult['all']) {
+    echo 'Creating search index for default country names';
+    $bDidSomething = true;
+
+    pgsqlRunScript("select getorcreate_country(make_standard_name('uk'), 'gb')");
+    pgsqlRunScript("select getorcreate_country(make_standard_name('united states'), 'us')");
+    pgsqlRunScript("select count(*) from (select getorcreate_country(make_standard_name(country_code), country_code) from country_name where country_code is not null) as x");
+    pgsqlRunScript("select count(*) from (select getorcreate_country(make_standard_name(name->'name'), country_code) from country_name where name ? 'name') as x");
+
+    $sSQL = 'select count(*) from (select getorcreate_country(make_standard_name(v), country_code) from (select country_code, skeys(name) as k, svals(name) as v from country_name) x where k ';
+    if (CONST_Languages) {
+        $sSQL .= 'in ';
+        $sDelim = '(';
+        foreach (explode(',', CONST_Languages) as $sLang) {
+            $sSQL .= $sDelim."'name:$sLang'";
+            $sDelim = ',';
+        }
+        $sSQL .= ')';
+    } else {
+        // all include all simple name tags
+        $sSQL .= "like 'name:%'";
+    }
+    $sSQL .= ') v';
+    pgsqlRunScript($sSQL);
+}
+

 if ($aCMDResult['drop']) {
     // The implementation is potentially a bit dangerous because it uses
     // a positive selection of tables to keep, and deletes everything else.
 if ($aCMDResult['drop']) {
     // The implementation is potentially a bit dangerous because it uses
     // a positive selection of tables to keep, and deletes everything else.


@@ -649,6 +700,7 @@ if (!$bDidSomething) {
     echo "Setup finished.\n";
 }
 
     echo "Setup finished.\n";
 }
 

+

 function pgsqlRunScriptFile($sFilename)
 {
     if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 function pgsqlRunScriptFile($sFilename)
 {
     if (!file_exists($sFilename)) fail('unable to find '.$sFilename);


@@ -698,7 +750,6 @@ function pgsqlRunScriptFile($sFilename)
         fclose($ahGzipPipes[1]);
         proc_close($hGzipProcess);
     }
         fclose($ahGzipPipes[1]);
         proc_close($hGzipProcess);
     }

-

 }
 
 function pgsqlRunScript($sScript, $bfatal = true)
 }
 
 function pgsqlRunScript($sScript, $bfatal = true)


@@ -712,7 +763,7 @@ function pgsqlRunScript($sScript, $bfatal = true)
         $sCMD .= ' -v ON_ERROR_STOP=1';
     $aDescriptors = array(
                      0 => array('pipe', 'r'),
         $sCMD .= ' -v ON_ERROR_STOP=1';
     $aDescriptors = array(
                      0 => array('pipe', 'r'),

-                     1 => STDOUT, 
+                     1 => STDOUT,

                      2 => STDERR
                     );
     $ahPipes = null;
                      2 => STDERR
                     );
     $ahPipes = null;


@@ -844,6 +895,4 @@ function create_sql_functions($aCMDResult)
         $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
     }
     pgsqlRunScript($sTemplate);
         $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
     }
     pgsqlRunScript($sTemplate);

-

 }
 }

-