X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/6985abc3fd16dd5a8f8de91ec1d798238ea10709..c2629bc08d081262a7fd1cbd9d52acb5f44bd8ec:/utils/setup.php

diff --git a/utils/setup.php b/utils/setup.php
index eeba0a28..86630196 100755
--- a/utils/setup.php
+++ b/utils/setup.php
@@ -65,11 +65,11 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
 $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
 if ($iInstances < 1) {
     $iInstances = 1;
-    echo "WARNING: resetting threads to $iInstances\n";
+    warn("resetting threads to $iInstances");
 }
 if ($iInstances > getProcessorCount()) {
     $iInstances = getProcessorCount();
-    echo "WARNING: resetting threads to $iInstances\n";
+    warn("resetting threads to $iInstances");
 }
 
 // Assume we can steal all the cache memory in the box (unless told otherwise)
@@ -83,7 +83,7 @@ $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
 if ($aCMDResult['create-db'] || $aCMDResult['all']) {
-    echo "Create DB\n";
+    info("Create DB");
     $bDidSomething = true;
     $oDB = DB::connect(CONST_Database_DSN, false);
     if (!PEAR::isError($oDB)) {
@@ -93,11 +93,9 @@ if ($aCMDResult['create-db'] || $aCMDResult['all']) {
 }
 
 if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
-    echo "Setup DB\n";
+    info("Setup DB");
     $bDidSomething = true;
 
-    // TODO: path detection, detection memory, etc.
-    //
     $oDB =& getDB();
 
     $fPostgresVersion = getPostgresVersion($oDB);
@@ -117,7 +115,7 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
 
     if ($iNumFunc == 0) {
         pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable");
-        echo "WARNING: Postgresql is too old. extratags and namedetails API not available.";
+        warn('Postgresql is too old. extratags and namedetails API not available.');
     }
 
     $fPostgisVersion = getPostgisVersion($oDB);
@@ -132,6 +130,26 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
         pgsqlRunScript('ALTER FUNCTION ST_Distance_Spheroid(geometry, geometry, spheroid) RENAME TO ST_DistanceSpheroid');
     }
 
+    $i = chksql($oDB->getOne("select count(*) from pg_user where usename = '".CONST_Database_Web_User."'"));
+    if ($i == 0) {
+        echo "\nERROR: Web user '".CONST_Database_Web_User."' does not exist. Create it with:\n";
+        echo "\n          createuser ".CONST_Database_Web_User."\n\n";
+        exit(1);
+    }
+
+    // Try accessing the C module, so we know early if something is wrong
+    // and can simply error out.
+    $sSQL = "CREATE FUNCTION nominatim_test_import_func(text) RETURNS text AS '";
+    $sSQL .= CONST_InstallPath."/module/nominatim.so', 'transliteration' LANGUAGE c IMMUTABLE STRICT";
+    $sSQL .= ';DROP FUNCTION nominatim_test_import_func(text);';
+    $oResult = $oDB->query($sSQL);
+
+    if (PEAR::isError($oResult)) {
+        echo "\nERROR: Failed to load nominatim module. Reason:\n";
+        echo $oResult->userinfo."\n\n";
+        exit(1);
+    }
+
     if (!file_exists(CONST_ExtraDataPath.'/country_osm_grid.sql.gz')) {
         echo "Error: you need to download the country_osm_grid first:";
         echo "\n    wget -O ".CONST_ExtraDataPath."/country_osm_grid.sql.gz http://www.nominatim.org/data/country_grid.sql.gz\n";
@@ -145,7 +163,7 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
     if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz')) {
         pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
     } else {
-        echo "WARNING: external UK postcode table not found.\n";
+        warn('external UK postcode table not found.');
     }
     if (CONST_Use_Extra_US_Postcodes) {
         pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
@@ -164,7 +182,7 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
 }
 
 if ($aCMDResult['import-data'] || $aCMDResult['all']) {
-    echo "Import\n";
+    info('Import data');
     $bDidSomething = true;
 
     $osm2pgsql = CONST_Osm2pgsql_Binary;
@@ -198,16 +216,18 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
 }
 
 if ($aCMDResult['create-functions'] || $aCMDResult['all']) {
-    echo "Functions\n";
+    info('Create Functions');
     $bDidSomething = true;
-    if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
+    if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) {
+        fail("nominatim module not built");
+    }
     create_sql_functions($aCMDResult);
 }
 
 if ($aCMDResult['create-tables'] || $aCMDResult['all']) {
+    info('Create Tables');
     $bDidSomething = true;
 
-    echo "Tables\n";
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql');
     $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
     $sTemplate = replace_tablespace(
@@ -243,12 +263,12 @@ if ($aCMDResult['create-tables'] || $aCMDResult['all']) {
     pgsqlRunScript($sTemplate, false);
 
     // re-run the functions
-    echo "Functions\n";
+    info('Recreate Functions');
     create_sql_functions($aCMDResult);
 }
 
 if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) {
-    echo "Partition Tables\n";
+    info('Create Partition Tables');
     $bDidSomething = true;
 
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
@@ -288,7 +308,7 @@ if ($aCMDResult['create-partition-tables'] || $aCMDResult['all']) {
 
 
 if ($aCMDResult['create-partition-functions'] || $aCMDResult['all']) {
-    echo "Partition Functions\n";
+    info('Create Partition Functions');
     $bDidSomething = true;
 
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
@@ -301,24 +321,22 @@ if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all']) {
     $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...";
+        info('Importing wikipedia articles');
         pgsqlRunDropAndRestore($sWikiArticlesFile);
-        echo "...done\n";
     } else {
-        echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
+        warn('wikipedia article dump file not found - places will have default importance');
     }
     if (file_exists($sWikiRedirectsFile)) {
-        echo "Importing wikipedia redirects...";
+        info('Importing wikipedia redirects');
         pgsqlRunDropAndRestore($sWikiRedirectsFile);
-        echo "...done\n";
     } else {
-        echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
+        warn('wikipedia redirect dump file not found - some place importance values may be missing');
     }
 }
 
 
 if ($aCMDResult['load-data'] || $aCMDResult['all']) {
-    echo "Drop old Data\n";
+    info('Drop old Data');
     $bDidSomething = true;
 
     $oDB =& getDB();
@@ -361,11 +379,11 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
 
     // pre-create the word list
     if (!$aCMDResult['disable-token-precalc']) {
-        echo "Loading word list\n";
+        info('Loading word list');
         pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
     }
 
-    echo "Load Data\n";
+    info('Load Data');
     $sColumns = 'osm_type, osm_id, class, type, name, admin_level, address, extratags, geometry';
 
     $aDBInstances = array();
@@ -402,17 +420,22 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
         echo '.';
     }
     echo "\n";
-    echo "Reanalysing database...\n";
+    info('Reanalysing database');
     pgsqlRunScript('ANALYSE');
 
     $sDatabaseDate = getDatabaseDate($oDB);
     pg_query($oDB->connection, 'TRUNCATE import_status');
-    $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')";
-    pg_query($oDB->connection, $sSQL);
-    echo "Latest data imported from $sDatabaseDate.\n";
+    if ($sDatabaseDate === false) {
+        warn('could not determine database date.');
+    } else {
+        $sSQL = "INSERT INTO import_status (lastimportdate) VALUES('".$sDatabaseDate."')";
+        pg_query($oDB->connection, $sSQL);
+        echo "Latest data imported from $sDatabaseDate.\n";
+    }
 }
 
 if ($aCMDResult['import-tiger-data']) {
+    info('Import Tiger data');
     $bDidSomething = true;
 
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
@@ -470,7 +493,7 @@ if ($aCMDResult['import-tiger-data']) {
         echo "\n";
     }
 
-    echo "Creating indexes\n";
+    info('Creating indexes on Tiger data');
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql');
     $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
     $sTemplate = replace_tablespace(
@@ -487,25 +510,63 @@ if ($aCMDResult['import-tiger-data']) {
 }
 
 if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) {
+    info('Calculate Postcodes');
     $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,address,country_code,geometry) ";
-    $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',";
-    $sSQL .= "hstore('postcode', pc),country_code,";
-    $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select country_code,";
-    $sSQL .= "address->'postcode' as pc,";
-    $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
-    $sSQL .= "from placex where address ? 'postcode' group by country_code,pc) as x ";
-    $sSQL .= "where ST_Point(x,y) is not null";
-    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+    if (!pg_query($oDB->connection, 'TRUNCATE location_postcode')) {
+        fail(pg_last_error($oDB->connection));
+    }
+
+    $sSQL  = "INSERT INTO location_postcode";
+    $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) ";
+    $sSQL .= "SELECT nextval('seq_place'), 1, country_code,";
+    $sSQL .= "       upper(trim (both ' ' from address->'postcode')) as pc,";
+    $sSQL .= "       ST_Centroid(ST_Collect(ST_Centroid(geometry)))";
+    $sSQL .= "  FROM placex";
+    $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'";
+    $sSQL .= "       AND geometry IS NOT null";
+    $sSQL .= " GROUP BY country_code, pc";
+
+    if (!pg_query($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,address,country_code,geometry) ";
-        $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',";
-        $sSQL .= "hstore('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));
+        // only add postcodes that are not yet available in OSM
+        $sSQL  = "INSERT INTO location_postcode";
+        $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) ";
+        $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,";
+        $sSQL .= "       ST_SetSRID(ST_Point(x,y),4326)";
+        $sSQL .= "  FROM us_postcode WHERE postcode NOT IN";
+        $sSQL .= "        (SELECT postcode FROM location_postcode";
+        $sSQL .= "          WHERE country_code = 'us')";
+
+    } else {
+        $sSQL .= "TRUNCATE TABLE us_postcode";
+    }
+    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+
+    // add missing postcodes for GB (if available)
+    $sSQL  = "INSERT INTO location_postcode";
+    $sSQL .= " (place_id, indexed_status, country_code, postcode, geometry) ";
+    $sSQL .= "SELECT nextval('seq_place'), 1, 'gb', postcode, geometry";
+    $sSQL .= "  FROM gb_postcode WHERE postcode NOT IN";
+    $sSQL .= "           (SELECT postcode FROM location_postcode";
+    $sSQL .= "             WHERE country_code = 'gb')";
+    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
+
+    if (!$aCMDResult['all']) {
+        $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'";
+        $sSQL .= "and word NOT IN (SELECT postcode FROM location_postcode)";
+        if (!pg_query($oDB->connection, $sSQL)) {
+            fail(pg_last_error($oDB->connection));
+        }
+    }
+    $sSQL = "SELECT count(getorcreate_postcode_id(v)) FROM ";
+    $sSQL .= "(SELECT distinct(postcode) as v FROM location_postcode) p";
+
+    if (!pg_query($oDB->connection, $sSQL)) {
+        fail(pg_last_error($oDB->connection));
     }
 }
 
@@ -518,15 +579,23 @@ if ($aCMDResult['index'] || $aCMDResult['all']) {
     $bDidSomething = true;
     $sOutputFile = '';
     $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
+    info('Index ranks 0 - 4');
     passthruCheckReturn($sBaseCmd.' -R 4');
     if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+    info('Index ranks 5 - 25');
     passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
     if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
+    info('Index ranks 26 - 30');
     passthruCheckReturn($sBaseCmd.' -r 26');
+
+    info('Index postcodes');
+    $oDB =& getDB();
+    $sSQL = 'UPDATE location_postcode SET indexed_status = 0';
+    if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 }
 
 if ($aCMDResult['create-search-indices'] || $aCMDResult['all']) {
-    echo "Search indices\n";
+    info('Create Search indices');
     $bDidSomething = true;
 
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
@@ -550,7 +619,7 @@ if ($aCMDResult['create-search-indices'] || $aCMDResult['all']) {
 }
 
 if ($aCMDResult['create-country-names'] || $aCMDResult['all']) {
-    echo 'Creating search index for default country names';
+    info('Create search index for default country names');
     $bDidSomething = true;
 
     pgsqlRunScript("select getorcreate_country(make_standard_name('uk'), 'gb')");
@@ -576,6 +645,7 @@ if ($aCMDResult['create-country-names'] || $aCMDResult['all']) {
 }
 
 if ($aCMDResult['drop']) {
+    info('Drop tables only required for updates');
     // The implementation is potentially a bit dangerous because it uses
     // a positive selection of tables to keep, and deletes everything else.
     // Including any tables that the unsuspecting user might have manually
@@ -632,18 +702,25 @@ if ($aCMDResult['drop']) {
 if (!$bDidSomething) {
     showUsage($aCMDOptions, true);
 } else {
-    echo "Setup finished.\n";
+    echo "Summary of warnings:\n\n";
+    repeatWarnings();
+    echo "\n";
+    info('Setup finished.');
 }
 
 
 function pgsqlRunScriptFile($sFilename)
 {
+    global $aCMDResult;
     if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 
     // Convert database DSN to psql parameters
     $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
     if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
     $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+    if (!$aCMDResult['verbose']) {
+        $sCMD .= ' -q';
+    }
 
     $ahGzipPipes = null;
     if (preg_match('/\\.gz$/', $sFilename)) {
@@ -690,31 +767,12 @@ function pgsqlRunScriptFile($sFilename)
 function pgsqlRunScript($sScript, $bfatal = true)
 {
     global $aCMDResult;
-    // Convert database DSN to psql parameters
-    $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
-    if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
-    $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
-    if ($bfatal && !$aCMDResult['ignore-errors'])
-        $sCMD .= ' -v ON_ERROR_STOP=1';
-    $aDescriptors = array(
-                     0 => array('pipe', 'r'),
-                     1 => STDOUT,
-                     2 => STDERR
-                    );
-    $ahPipes = null;
-    $hProcess = @proc_open($sCMD, $aDescriptors, $ahPipes);
-    if (!is_resource($hProcess)) fail('unable to start pgsql');
-
-    while (strlen($sScript)) {
-        $written = fwrite($ahPipes[0], $sScript);
-        if ($written <= 0) break;
-        $sScript = substr($sScript, $written);
-    }
-    fclose($ahPipes[0]);
-    $iReturn = proc_close($hProcess);
-    if ($bfatal && $iReturn > 0) {
-        fail("pgsql returned with error code ($iReturn)");
-    }
+    runSQLScript(
+        $sScript,
+        $bfatal,
+        $aCMDResult['verbose'],
+        $aCMDResult['ignore-errors']
+    );
 }
 
 function pgsqlRunPartitionScript($sTemplate)