Work on setup/update scripts, unit tests, and documentation to enable Postgres server to be optionally configured on a remote host

data/words.sql

@@ -23,6 +23,7 @@ CREATE TABLE word_frequencies (
     count bigint
 );
 
+TRUNCATE TABLE word_frequencies;
 
 --
 -- Data for Name: word_frequencies; Type: TABLE DATA; Schema: public; Owner: -

docs/admin/Faq.md

@@ -112,7 +112,7 @@ to get the full error message.
 On CentOS v7 the PostgreSQL server is started with `systemd`.
 Check if `/usr/lib/systemd/system/httpd.service` contains a line `PrivateTmp=true`.
 If so then Apache cannot see the `/tmp/.s.PGSQL.5432` file. It's a good security feature,
-so use the [preferred solution](../appendix/Install-on-Centos-7/#adding-selinux-security-settings).
+so use the [[#PostgreSQL_UNIX_Socket_Location_on_CentOS|preferred solution]]
 
 However, you can solve this the quick and dirty way by commenting out that line and then run
 

lib/cmd.php

@@ -158,6 +158,16 @@ function runSQLScript($sScript, $bfatal = true, $bVerbose = false, $bIgnoreError
     $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
     if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
     $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $sCMD .= ' -h ' . $aDSNInfo['hostspec'];
+    }
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $sCMD .= ' -U ' . $aDSNInfo['username'];
+    }
+    $procenv = null;
+    if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
+        $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
+    }
     if (!$bVerbose) {
         $sCMD .= ' -q';
     }
@@ -170,7 +180,7 @@ function runSQLScript($sScript, $bfatal = true, $bVerbose = false, $bIgnoreError
                      2 => STDERR
                     );
     $ahPipes = null;
-    $hProcess = @proc_open($sCMD, $aDescriptors, $ahPipes);
+    $hProcess = @proc_open($sCMD, $aDescriptors, $ahPipes, null, $procenv);
     if (!is_resource($hProcess)) {
         fail('unable to start pgsql');
     }

nominatim/index.c

@@ -262,10 +262,13 @@ struct index_thread_data * thread_data, const char *structuredoutputfile)
 
 void nominatim_index(int rank_min, int rank_max, int num_threads, const char *conninfo, const char *structuredoutputfile)
 {
-    struct index_thread_data * thread_data;
+    struct index_thread_data *thread_data;
 
     PGconn *conn;
-    PGresult * res;
+    PGresult *res;
+    int num_rows = 0, status_code = 0;
+    int db_has_locale = 0;
+    char *result_string = NULL;
 
     int rank;
 
@@ -283,6 +286,23 @@ void nominatim_index(int rank_min, int rank_max, int num_threads, const char *co
         exit(EXIT_FAILURE);
     }
 
+    res = PQexec(conn, "SHOW lc_messages");
+    status_code = PQresultStatus(res);
+    if (status_code != PGRES_TUPLES_OK && status_code != PGRES_SINGLE_TUPLE) {
+        fprintf(stderr, "Failed determining database locale: %s\n", PQerrorMessage(conn));
+        exit(EXIT_FAILURE);
+    }
+    num_rows = PQntuples(res);
+    if (num_rows > 0)
+    {
+        result_string = PQgetvalue(res, 0, 0);
+        if (result_string && (strlen(result_string) > 0) && (strcasecmp(result_string, "C") != 0))
+        {
+            // non-default locale if the result exists, is non-empty, and is not "C"
+            db_has_locale = 1;
+        }
+    }
+
     pg_prepare_params[0] = PG_OID_INT4;
     res = PQprepare(conn, "index_sectors",
                     "select geometry_sector,count(*) from placex where rank_search = $1 and indexed_status > 0 group by geometry_sector order by geometry_sector",
@@ -392,19 +412,20 @@ void nominatim_index(int rank_min, int rank_max, int num_threads, const char *co
         }
         PQclear(res);
 
-        // Make sure the error message is not localized as we parse it later.
+        if (db_has_locale)
-        res = PQexec(thread_data[i].conn, "SET lc_messages TO 'C'");
-        if (PQresultStatus(res) != PGRES_COMMAND_OK)
         {
-            fprintf(stderr, "Failed to set langauge: %s\n", PQerrorMessage(thread_data[i].conn));
+            // Make sure the error message is not localized as we parse it later.
-            exit(EXIT_FAILURE);
+            res = PQexec(thread_data[i].conn, "SET lc_messages TO 'C'");
+            if (PQresultStatus(res) != PGRES_COMMAND_OK)
+            {
+                fprintf(stderr, "Failed to set langauge: %s\n", PQerrorMessage(thread_data[i].conn));
+                exit(EXIT_FAILURE);
+            }
+            PQclear(res);
         }
-        PQclear(res);
-
         nominatim_exportCreatePreparedQueries(thread_data[i].conn);
     }
 
-
     fprintf(stderr, "Starting indexing rank (%i to %i) using %i threads\n", rank_min, rank_max, num_threads);
 
     for (rank = rank_min; rank <= rank_max; rank++)

sql/partition-tables.src.sql

@@ -48,6 +48,7 @@ CREATE INDEX idx_search_name_-partition-_place_id ON search_name_-partition- USI
 CREATE INDEX idx_search_name_-partition-_centroid ON search_name_-partition- USING GIST (centroid) {ts:address-index};
 CREATE INDEX idx_search_name_-partition-_name_vector ON search_name_-partition- USING GIN (name_vector) WITH (fastupdate = off) {ts:address-index};
 
+DROP TABLE IF EXISTS location_road_-partition-;
 CREATE TABLE location_road_-partition- (
   place_id BIGINT,
   partition SMALLINT,

test/README.md

@@ -73,6 +73,11 @@ The tests can be configured with a set of environment variables:
                    the test databases (db tests)
  * `TEST_DB` - name of test database (db tests)
  * `ABI_TEST_DB` - name of the database containing the API test data (api tests)
+ * `DB_HOST` - (optional) hostname of database host
+ * `DB_USER` - (optional) username of database login
+ * `DB_PASS` - (optional) password for database login
+ * `SERVER_MODULE_PATH` - (optional) path on the Postgres server to Nominatim
+ *                        module shared library file
  * `TEST_SETTINGS_TEMPLATE` - file to write temporary Nominatim settings to
  * `REMOVE_TEMPLATE` - if true, the template database will not be reused during
                        the next run. Reusing the base templates speeds up tests

test/bdd/environment.py

@@ -14,10 +14,14 @@ userconfig = {
     'BUILDDIR' : os.path.join(os.path.split(__file__)[0], "../../build"),
     'REMOVE_TEMPLATE' : False,
     'KEEP_TEST_DB' : False,
+    'DB_HOST' : None,
+    'DB_USER' : None,
+    'DB_PASS' : None,
     'TEMPLATE_DB' : 'test_template_nominatim',
     'TEST_DB' : 'test_nominatim',
     'API_TEST_DB' : 'test_api_nominatim',
     'TEST_SETTINGS_FILE' : '/tmp/nominatim_settings.php',
+    'SERVER_MODULE_PATH' : None,
     'PHPCOV' : False, # set to output directory to enable code coverage
 }
 
@@ -30,9 +34,13 @@ class NominatimEnvironment(object):
     def __init__(self, config):
         self.build_dir = os.path.abspath(config['BUILDDIR'])
         self.src_dir = os.path.abspath(os.path.join(os.path.split(__file__)[0], "../.."))
+        self.db_host = config['DB_HOST']
+        self.db_user = config['DB_USER']
+        self.db_pass = config['DB_PASS']
         self.template_db = config['TEMPLATE_DB']
         self.test_db = config['TEST_DB']
         self.api_test_db = config['API_TEST_DB']
+        self.server_module_path = config['SERVER_MODULE_PATH']
         self.local_settings_file = config['TEST_SETTINGS_FILE']
         self.reuse_template = not config['REMOVE_TEMPLATE']
         self.keep_scenario_db = config['KEEP_TEST_DB']
@@ -42,6 +50,17 @@ class NominatimEnvironment(object):
 
         self.template_db_done = False
 
+    def connect_database(self, dbname):
+        dbargs = {'database': dbname}
+        if self.db_host:
+            dbargs['host'] = self.db_host
+        if self.db_user:
+            dbargs['user'] = self.db_user
+        if self.db_pass:
+            dbargs['password'] = self.db_pass
+        conn = psycopg2.connect(**dbargs)
+        return conn
+
     def next_code_coverage_file(self):
         fn = os.path.join(self.code_coverage_path, "%06d.cov" % self.code_coverage_id)
         self.code_coverage_id += 1
@@ -50,7 +69,11 @@ class NominatimEnvironment(object):
 
     def write_nominatim_config(self, dbname):
         f = open(self.local_settings_file, 'w')
-        f.write("<?php\n  @define('CONST_Database_DSN', 'pgsql://@/%s');\n" % dbname)
+        f.write("<?php\n  @define('CONST_Database_DSN', 'pgsql://%s:%s@%s/%s');\n" %
+                (self.db_user if self.db_user else '',
+                 self.db_pass if self.db_pass else '',
+                 self.db_host if self.db_host else '',
+                 dbname))
         f.write("@define('CONST_Osm2pgsql_Flatnode_File', null);\n")
         f.close()
 
@@ -61,7 +84,7 @@ class NominatimEnvironment(object):
             pass # ignore missing file
 
     def db_drop_database(self, name):
-        conn = psycopg2.connect(database='postgres')
+        conn = self.connect_database('postgres')
         conn.set_isolation_level(0)
         cur = conn.cursor()
         cur.execute('DROP DATABASE IF EXISTS %s' % (name, ))
@@ -75,7 +98,7 @@ class NominatimEnvironment(object):
 
         if self.reuse_template:
             # check that the template is there
-            conn = psycopg2.connect(database='postgres')
+            conn = self.connect_database('postgres')
             cur = conn.cursor()
             cur.execute('select count(*) from pg_database where datname = %s',
                         (self.template_db,))
@@ -91,7 +114,7 @@ class NominatimEnvironment(object):
             self.write_nominatim_config(self.template_db)
             self.run_setup_script('create-db', 'setup-db')
             # remove external data to speed up indexing for tests
-            conn = psycopg2.connect(database=self.template_db)
+            conn = self.connect_database(self.template_db)
             cur = conn.cursor()
             cur.execute("""select tablename from pg_tables
                            where tablename in ('gb_postcode', 'us_postcode')""")
@@ -128,13 +151,13 @@ class NominatimEnvironment(object):
     def setup_db(self, context):
         self.setup_template_db()
         self.write_nominatim_config(self.test_db)
-        conn = psycopg2.connect(database=self.template_db)
+        conn = self.connect_database(self.template_db)
         conn.set_isolation_level(0)
         cur = conn.cursor()
         cur.execute('DROP DATABASE IF EXISTS %s' % (self.test_db, ))
         cur.execute('CREATE DATABASE %s TEMPLATE = %s' % (self.test_db, self.template_db))
         conn.close()
-        context.db = psycopg2.connect(database=self.test_db)
+        context.db = self.connect_database(self.test_db)
         if python_version[0] < 3:
             psycopg2.extras.register_hstore(context.db, globally=False, unicode=True)
         else:
@@ -148,6 +171,9 @@ class NominatimEnvironment(object):
             self.db_drop_database(self.test_db)
 
     def run_setup_script(self, *args, **kwargs):
+        if self.server_module_path:
+            kwargs = dict(kwargs)
+            kwargs['module_path'] = self.server_module_path
         self.run_nominatim_script('setup', *args, **kwargs)
 
     def run_update_script(self, *args, **kwargs):
@@ -271,4 +297,3 @@ def before_scenario(context, scenario):
 def after_scenario(context, scenario):
     if 'DB' in context.tags:
         context.nominatim.teardown_db(context)
-

utils/setup.php

@@ -5,6 +5,8 @@ require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
 require_once(CONST_BasePath.'/lib/init-cmd.php');
 ini_set('memory_limit', '800M');
 
+# (long-opt, short-opt, min-occurs, max-occurs, num-arguments, num-arguments, type, help)
+
 $aCMDOptions
 = array(
    'Create and setup nominatim search system',
@@ -14,6 +16,7 @@ $aCMDOptions
 
    array('osm-file', '', 0, 1, 1, 1, 'realpath', 'File to import'),
    array('threads', '', 0, 1, 1, 1, 'int', 'Number of threads (where possible)'),
+   array('module-path', '', 0, 1, 1, 1, 'string', 'Directory on Postgres server containing Nominatim module'),
 
    array('all', '', 0, 1, 0, 0, 'bool', 'Do the complete process'),
 
@@ -60,7 +63,6 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
     }
 }
 
-
 // by default, use all but one processor, but never more than 15.
 $iInstances = isset($aCMDResult['threads'])
               ? $aCMDResult['threads']
@@ -70,10 +72,6 @@ if ($iInstances < 1) {
     $iInstances = 1;
     warn("resetting threads to $iInstances");
 }
-if ($iInstances > getProcessorCount()) {
-    $iInstances = getProcessorCount();
-    warn("resetting threads to $iInstances");
-}
 
 // Assume we can steal all the cache memory in the box (unless told otherwise)
 if (isset($aCMDResult['osm2pgsql-cache'])) {
@@ -82,6 +80,12 @@ if (isset($aCMDResult['osm2pgsql-cache'])) {
     $iCacheMemory = getCacheMemoryMB();
 }
 
+$modulePath = CONST_InstallPath . '/module';
+if (isset($aCMDResult['module-path'])) {
+    $modulePath = $aCMDResult['module-path'];
+    echo 'module path: ' . $modulePath . '\n';
+}
+
 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 
@@ -92,7 +96,22 @@ if ($aCMDResult['create-db'] || $aCMDResult['all']) {
     if (!PEAR::isError($oDB)) {
         fail('database already exists ('.CONST_Database_DSN.')');
     }
-    passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']);
+
+    $createdbCmd = 'createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database'];
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $createdbCmd .= ' -U ' . $aDSNInfo['username'];
+    }
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $createdbCmd .= ' -h ' . $aDSNInfo['hostspec'];
+    }
+
+    $procenv = null;
+    if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
+        $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
+    }
+
+    $result = runWithEnv($createdbCmd, $procenv);
+    if ($result != 0) fail('Error executing external command: '.$createdbCmd);
 }
 
 if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
@@ -143,7 +162,7 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
     // 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 .= $modulePath."/nominatim.so', 'transliteration' LANGUAGE c IMMUTABLE STRICT";
     $sSQL .= ';DROP FUNCTION nominatim_test_import_func(text);';
     $oResult = $oDB->query($sSQL);
 
@@ -180,8 +199,8 @@ if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
     // is only defined in the subsequently called create_tables.
     // Create dummies here that will be overwritten by the proper
     // versions in create-tables.
-    pgsqlRunScript('CREATE TABLE place_boundingbox ()');
+    pgsqlRunScript('CREATE TABLE IF NOT EXISTS place_boundingbox ()');
-    pgsqlRunScript('create type wikipedia_article_match as ()');
+    pgsqlRunScript('CREATE TYPE wikipedia_article_match AS ()', false);
 }
 
 if ($aCMDResult['import-data'] || $aCMDResult['all']) {
@@ -209,8 +228,20 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
     $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1';
     $osm2pgsql .= ' -C '.$iCacheMemory;
     $osm2pgsql .= ' -P '.$aDSNInfo['port'];
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $osm2pgsql .= ' -U ' . $aDSNInfo['username'];
+    }
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $osm2pgsql .= ' -H ' . $aDSNInfo['hostspec'];
+    }
+
+    $procenv = null;
+    if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
+        $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
+    }
+
     $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
-    passthruCheckReturn($osm2pgsql);
+    runWithEnv($osm2pgsql, $procenv);
 
     $oDB =& getDB();
     if (!$aCMDResult['ignore-errors'] && !chksql($oDB->getRow('select * from place limit 1'))) {
@@ -221,9 +252,6 @@ if ($aCMDResult['import-data'] || $aCMDResult['all']) {
 if ($aCMDResult['create-functions'] || $aCMDResult['all']) {
     info('Create Functions');
     $bDidSomething = true;
-    if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) {
-        fail('nominatim module not built');
-    }
     create_sql_functions($aCMDResult);
 }
 
@@ -413,14 +441,23 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
         fail(pg_last_error($aDBInstances[$iLoadThreads]->connection));
     }
 
-    $bAnyBusy = true;
+    $failed = false;
-    while ($bAnyBusy) {
+    for ($i = 0; $i <= $iLoadThreads; $i++) {
-        $bAnyBusy = false;
+        while (($pgresult = pg_get_result($aDBInstances[$i]->connection)) !== false) {
-        for ($i = 0; $i <= $iLoadThreads; $i++) {
+            $resultStatus = pg_result_status($pgresult);
-            if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
+            // PGSQL_EMPTY_QUERY, PGSQL_COMMAND_OK, PGSQL_TUPLES_OK,
+            // PGSQL_COPY_OUT, PGSQL_COPY_IN, PGSQL_BAD_RESPONSE,
+            // PGSQL_NONFATAL_ERROR and PGSQL_FATAL_ERROR
+            echo 'Query result ' . $i . ' is: ' . $resultStatus . '\n';
+            if ($resultStatus != PGSQL_COMMAND_OK && $resultStatus != PGSQL_TUPLES_OK) {
+                $resultError = pg_result_error($pgresult);
+                echo '-- error text ' . $i . ': ' . $resultError . '\n';
+                $failed = true;
+            }
         }
-        sleep(1);
+    }
-        echo '.';
+    if ($failed) {
+        fail('SQL errors loading placex and/or location_property_osmline tables');
     }
     echo "\n";
     info('Reanalysing database');
@@ -579,14 +616,34 @@ if ($aCMDResult['index'] || $aCMDResult['all']) {
     $bDidSomething = true;
     $sOutputFile = '';
     $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $sBaseCmd .= ' -H ' . $aDSNInfo['hostspec'];
+    }
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $sBaseCmd .= ' -U ' . $aDSNInfo['username'];
+    }
+    $procenv = null;
+    if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
+        $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
+    }
+
     info('Index ranks 0 - 4');
-    passthruCheckReturn($sBaseCmd.' -R 4');
+    $status = runWithEnv($sBaseCmd.' -R 4', $procenv);
+    if ($status != 0) {
+        fail('error status ' . $status . ' running nominatim!');
+    }
     if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
     info('Index ranks 5 - 25');
-    passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
+    $status = runWithEnv($sBaseCmd.' -r 5 -R 25', $procenv);
+    if ($status != 0) {
+        fail('error status ' . $status . ' running nominatim!');
+    }
     if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
     info('Index ranks 26 - 30');
-    passthruCheckReturn($sBaseCmd.' -r 26');
+    $status = runWithEnv($sBaseCmd.' -r 26', $procenv);
+    if ($status != 0) {
+        fail('error status ' . $status . ' running nominatim!');
+    }
 
     info('Index postcodes');
     $oDB =& getDB();
@@ -722,6 +779,16 @@ function pgsqlRunScriptFile($sFilename)
     if (!$aCMDResult['verbose']) {
         $sCMD .= ' -q';
     }
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $sCMD .= ' -h ' . $aDSNInfo['hostspec'];
+    }
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $sCMD .= ' -U ' . $aDSNInfo['username'];
+    }
+    $procenv = null;
+    if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
+        $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
+    }
 
     $ahGzipPipes = null;
     if (preg_match('/\\.gz$/', $sFilename)) {
@@ -745,10 +812,9 @@ function pgsqlRunScriptFile($sFilename)
                      2 => array('file', '/dev/null', 'a')
                     );
     $ahPipes = null;
-    $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+    $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes, null, $procenv);
     if (!is_resource($hProcess)) fail('unable to start pgsql');
 
-
     // TODO: error checking
     while (!feof($ahPipes[1])) {
         echo fread($ahPipes[1], 4096);
@@ -830,32 +896,41 @@ function pgsqlRunDropAndRestore($sDumpFile)
     $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
     if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
     $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc --clean '.$sDumpFile;
-
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
-    $aDescriptors = array(
+        $sCMD .= ' -h ' . $aDSNInfo['hostspec'];
-                     0 => array('pipe', 'r'),
+    }
-                     1 => array('pipe', 'w'),
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
-                     2 => array('file', '/dev/null', 'a')
+        $sCMD .= ' -U ' . $aDSNInfo['username'];
-                    );
+    }
-    $ahPipes = null;
+    $procenv = null;
-    $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
+    if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
-    if (!is_resource($hProcess)) fail('unable to start pg_restore');
+        $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
-
-    fclose($ahPipes[0]);
-
-    // TODO: error checking
-    while (!feof($ahPipes[1])) {
-        echo fread($ahPipes[1], 4096);
     }
-    fclose($ahPipes[1]);
 
-    $iReturn = proc_close($hProcess);
+    $iReturn = runWithEnv($sCMD, $procenv);
 }
 
 function passthruCheckReturn($cmd)
 {
     $result = -1;
     passthru($cmd, $result);
-    if ($result != 0) fail('Error executing external command: '.$cmd);
+}
+
+function runWithEnv($cmd, $env)
+{
+    $fds = array(0 => array('pipe', 'r'),
+                 1 => STDOUT,
+                 2 => STDERR);
+    $pipes = null;
+    $proc = @proc_open($cmd, $fds, $pipes, null, $env);
+    if (!is_resource($proc)) {
+        fail('unable to run command:' . $cmd);
+    }
+
+    fclose($pipes[0]); // no stdin
+
+    $stat = proc_close($proc);
+    return $stat;
 }
 
 function replace_tablespace($sTemplate, $sTablespace, $sSql)
@@ -871,8 +946,9 @@ function replace_tablespace($sTemplate, $sTablespace, $sSql)
 
 function create_sql_functions($aCMDResult)
 {
+    global $modulePath;
     $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
-    $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/module', $sTemplate);
+    $sTemplate = str_replace('{modulepath}', $modulePath, $sTemplate);
     if ($aCMDResult['enable-diff-updates']) {
         $sTemplate = str_replace('RETURN NEW; -- %DIFFUPDATES%', '--', $sTemplate);
     }

utils/update.php

@@ -5,6 +5,8 @@ require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
 require_once(CONST_BasePath.'/lib/init-cmd.php');
 ini_set('memory_limit', '800M');
 
+# (long-opt, short-opt, min-occurs, max-occurs, num-arguments, num-arguments, type, help)
+
 $aCMDOptions
 = array(
    'Import / update / index osm data',
@@ -14,6 +16,7 @@ $aCMDOptions
 
    array('init-updates', '', 0, 1, 0, 0, 'bool', 'Set up database for updating'),
    array('check-for-updates', '', 0, 1, 0, 0, 'bool', 'Check if new updates are available'),
+   array('update-functions', '', 0, 1, 0, 0, 'bool', 'Update trigger functions to support differential updates'),
    array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import updates once'),
    array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import updates forever'),
    array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
@@ -56,6 +59,17 @@ if ($iCacheMemory + 500 > getTotalMemoryMB()) {
     echo "WARNING: resetting cache memory to $iCacheMemory\n";
 }
 $sOsm2pgsqlCmd = CONST_Osm2pgsql_Binary.' -klas --number-processes 1 -C '.$iCacheMemory.' -O gazetteer -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'];
+if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+    $sOsm2pgsqlCmd .= ' -U ' . $aDSNInfo['username'];
+}
+if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+    $sOsm2pgsqlCmd .= ' -H ' . $aDSNInfo['hostspec'];
+}
+$procenv = null;
+if (isset($aDSNInfo['password']) && $aDSNInfo['password']) {
+    $procenv = array_merge(array('PGPASSWORD' => $aDSNInfo['password']), $_ENV);
+}
+
 if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
     $sOsm2pgsqlCmd .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
 }
@@ -84,11 +98,13 @@ if ($aResult['init-updates']) {
         echo "and have set up CONST_Pyosmium_Binary to point to pyosmium-get-changes.\n";
         fail('pyosmium-get-changes not found or not usable');
     }
-    $sSetup = CONST_InstallPath.'/utils/setup.php';
+    if ($aResult['update-functions']) {
-    $iRet = -1;
+        $sSetup = CONST_InstallPath.'/utils/setup.php';
-    passthru($sSetup.' --create-functions --enable-diff-updates', $iRet);
+        $iRet = -1;
-    if ($iRet != 0) {
+        passthru($argv[0].' '.$sSetup.' --create-functions --enable-diff-updates', $iRet);
-        fail('Error running setup script');
+        if ($iRet != 0) {
+            fail('Error running setup script');
+        }
     }
 
     $sDatabaseDate = getDatabaseDate($oDB);
@@ -106,8 +122,8 @@ if ($aResult['init-updates']) {
     }
 
     pg_query($oDB->connection, 'TRUNCATE import_status');
-    $sSQL = "INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES('";
+    $sSQL = 'INSERT INTO import_status (lastimportdate, sequence_id, indexed) VALUES(';
-    $sSQL .= $sDatabaseDate."',".$aOutput[0].', true)';
+    $sSQL .= "'".$sDatabaseDate."',".$aOutput[0].', true)';
     if (!pg_query($oDB->connection, $sSQL)) {
         fail('Could not enter sequence into database.');
     }
@@ -137,7 +153,7 @@ if (isset($aResult['import-diff']) || isset($aResult['import-file'])) {
     // Import the file
     $sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
     echo $sCMD."\n";
-    exec($sCMD, $sJunk, $iErrorLevel);
+    $iErrorLevel = runWithEnv($sCMD, $procenv);
 
     if ($iErrorLevel) {
         fail("Error from osm2pgsql, $iErrorLevel\n");
@@ -189,7 +205,7 @@ if ($bHaveDiff) {
     // import generated change file
     $sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
     echo $sCMD."\n";
-    exec($sCMD, $sJunk, $iErrorLevel);
+    $iErrorLevel = runWithEnv($sCMD, $procenv);
     if ($iErrorLevel) {
         fail("osm2pgsql exited with error level $iErrorLevel\n");
     }
@@ -273,7 +289,15 @@ if ($aResult['recompute-word-counts']) {
 }
 
 if ($aResult['index']) {
-    passthru(CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank']);
+    $cmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'].' -r '.$aResult['index-rank'];
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $cmd .= ' -H ' . $aDSNInfo['hostspec'];
+    }
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $cmd .= ' -U ' . $aDSNInfo['username'];
+    }
+
+    runWithEnv($cmd, $procenv);
 }
 
 if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
@@ -287,6 +311,12 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
     $sCMDDownload = CONST_Pyosmium_Binary.' --server '.CONST_Replication_Url.' -o '.$sImportFile.' -s '.CONST_Replication_Max_Diff_size;
     $sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
     $sCMDIndex = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
+    if (isset($aDSNInfo['hostspec']) && $aDSNInfo['hostspec']) {
+        $sCMDIndex .= ' -H ' . $aDSNInfo['hostspec'];
+    }
+    if (isset($aDSNInfo['username']) && $aDSNInfo['username']) {
+        $sCMDIndex .= ' -U ' . $aDSNInfo['username'];
+    }
 
     while (true) {
         $fStartTime = time();
@@ -354,7 +384,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
             $fCMDStartTime = time();
             echo $sCMDImport."\n";
             unset($sJunk);
-            exec($sCMDImport, $sJunk, $iErrorLevel);
+            $iErrorLevel = runWithEnv($sCMDImport, $procenv);
             if ($iErrorLevel) {
                 echo "Error executing osm2pgsql: $iErrorLevel\n";
                 exit($iErrorLevel);
@@ -383,7 +413,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
             $fCMDStartTime = time();
 
             echo "$sThisIndexCmd\n";
-            exec($sThisIndexCmd, $sJunk, $iErrorLevel);
+            $iErrorLevel = runWithEnv($sThisIndexCmd, $procenv);
             if ($iErrorLevel) {
                 echo "Error: $iErrorLevel\n";
                 exit($iErrorLevel);
@@ -398,7 +428,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
             $oDB->query($sSQL);
             echo date('Y-m-d H:i:s')." Completed index step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60, 2)." minutes\n";
 
-            $sSQL = 'update import_status set indexed = true';
+            $sSQL = 'UPDATE import_status SET indexed = true';
             $oDB->query($sSQL);
         }
 
@@ -407,3 +437,20 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
         if (!$aResult['import-osmosis-all']) exit(0);
     }
 }
+
+function runWithEnv($cmd, $env)
+{
+    $fds = array(0 => array('pipe', 'r'),
+                 1 => STDOUT,
+                 2 => STDERR);
+    $pipes = null;
+    $proc = @proc_open($cmd, $fds, $pipes, null, $env);
+    if (!is_resource($proc)) {
+        fail('unable to run command:' . $cmd);
+    }
+
+    fclose($pipes[0]); // no stdin
+
+    $stat = proc_close($proc);
+    return $stat;
+}