]> git.openstreetmap.org Git - nominatim.git/commitdiff
Merge remote-tracking branch 'upstream/master'
authorSarah Hoffmann <lonvia@denofr.de>
Fri, 12 Mar 2021 14:48:26 +0000 (15:48 +0100)
committerSarah Hoffmann <lonvia@denofr.de>
Fri, 12 Mar 2021 14:48:26 +0000 (15:48 +0100)
15 files changed:
docs/admin/Migration.md
lib-php/Geocode.php
lib-php/SearchDescription.php
lib-php/admin/setup.php
lib-php/setup/SetupClass.php
lib-sql/tiger_import_finish.sql
nominatim/cli.py
nominatim/clicmd/admin.py
nominatim/clicmd/transition.py
nominatim/tools/migration.py [new file with mode: 0644]
nominatim/tools/tiger_data.py [new file with mode: 0644]
test/python/conftest.py
test/python/test_cli.py
test/python/test_db_sql_preprocessor.py
test/python/test_tools_tiger_data.py [new file with mode: 0644]

index dc94310bb5d127e1aad7de64f6d8d986ef4a0ca4..52970a33d65c078189aabf4a0537e1e15abe3385 100644 (file)
@@ -1,22 +1,21 @@
 # Database Migrations
 
-This page describes database migrations necessary to update existing databases
-to newer versions of Nominatim.
+Since version 3.7.0 Nominatim offers automatic migrations. Please follow
+the following steps:
 
-SQL statements should be executed from the PostgreSQL commandline. Execute
-`psql nominatim` to enter command line mode.
+* stop any updates that are potentially running
+* update Nominatim to the nwer version
+* goto your project directory and run `nominatim admin --migrate`
+* (optionally) restart updates
 
-## 3.6.0 -> master
-
-### Status table contains now time zone information
+Below you find additional migrations and hints about other structural and
+breaking changes.
 
-The `import_status` table has been changed to include timezone information
-with the time stamp. You need to alter an existing table before running
-any replication functions with:
+!!! note
+    If you are migrating from a version <3.6, then you still have to follow
+    the manual migration steps up to 3.6.
 
-```sql
-ALTER TABLE import_status ALTER COLUMN lastimportdate TYPE timestamp with time zone;
-```
+## 3.6.0 -> master
 
 ### New location for data files
 
index b475add22e8eefaf3cd85d0cbc85c46e7c66bc98..f638af9a9a300a927f066a59690c23b6ee58e966 100644 (file)
@@ -779,7 +779,8 @@ class Geocode
                     $aSplitResults = Result::splitResults($aResults);
                     Debug::printVar('Split results', $aSplitResults);
                     if ($iGroupLoop <= 4
-                        && reset($aSplitResults['head'])->iResultRank > 0) {
+                        && reset($aSplitResults['head'])->iResultRank > 0
+                        && $iGroupedRank !== array_key_last($aGroupedSearches)) {
                         // Haven't found an exact match for the query yet.
                         // Therefore add result from the next group level.
                         $aNextResults = $aSplitResults['head'];
index ea9d7f58feb5dd06b453033f39ac1ec446f13e31..228b099917d6377f102a2987bf9e6f76af68058b 100644 (file)
@@ -179,6 +179,7 @@ class SearchDescription
                 // - increase score for finding it anywhere else (optimisation)
                 if (!$bLastToken) {
                     $oSearch->iSearchRank += 5;
+                    $oSearch->iNamePhrase = -1;
                 }
                 $aNewSearches[] = $oSearch;
             }
@@ -205,6 +206,7 @@ class SearchDescription
                 ) {
                     $oSearch = clone $this;
                     $oSearch->iSearchRank++;
+                    $oSearch->iNamePhrase = -1;
                     if (strlen($oSearchTerm->sPostcode) < 4) {
                         $oSearch->iSearchRank += 4 - strlen($oSearchTerm->sPostcode);
                     }
@@ -218,7 +220,11 @@ class SearchDescription
             if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
+                $oSearch->iNamePhrase = -1;
                 $oSearch->sHouseNumber = $oSearchTerm->sToken;
+                if ($this->iOperator != Operator::NONE) {
+                    $oSearch->iSearchRank++;
+                }
                 // sanity check: if the housenumber is not mainly made
                 // up of numbers, add a penalty
                 if (preg_match('/\\d/', $oSearch->sHouseNumber) === 0
@@ -255,7 +261,8 @@ class SearchDescription
         ) {
             if ($this->iOperator == Operator::NONE) {
                 $oSearch = clone $this;
-                $oSearch->iSearchRank++;
+                $oSearch->iSearchRank += 2;
+                $oSearch->iNamePhrase = -1;
 
                 $iOp = $oSearchTerm->iOperator;
                 if ($iOp == Operator::NONE) {
@@ -265,6 +272,11 @@ class SearchDescription
                         $iOp = Operator::NEAR;
                     }
                     $oSearch->iSearchRank += 2;
+                } elseif (!$bFirstToken && !$bLastToken) {
+                    $oSearch->iSearchRank += 2;
+                }
+                if ($this->sHouseNumber) {
+                    $oSearch->iSearchRank++;
                 }
 
                 $oSearch->setPoiSearch(
@@ -285,11 +297,12 @@ class SearchDescription
             if (!empty($this->aName) || !($bFirstPhrase || $sPhraseType == '')) {
                 if (($sPhraseType == '' || !$bFirstPhrase) && !$bHasPartial) {
                     $oSearch = clone $this;
+                    $oSearch->iNamePhrase = -1;
                     $oSearch->iSearchRank += 3 * $oSearchTerm->iTermCount;
                     $oSearch->aAddress[$iWordID] = $iWordID;
                     $aNewSearches[] = $oSearch;
                 }
-            } else {
+            } elseif (empty($this->aNameNonSearch)) {
                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
                 $oSearch->aName = array($iWordID => $iWordID);
@@ -329,51 +342,34 @@ class SearchDescription
 
         if ((!$bStructuredPhrases || $iPhrase > 0)
             && (!empty($this->aName))
-            && strpos($sToken, ' ') === false
         ) {
+            $oSearch = clone $this;
+            $oSearch->iSearchRank++;
+            if (preg_match('#^[0-9 ]+$#', $sToken)) {
+                $oSearch->iSearchRank++;
+            }
             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
-                $oSearch = clone $this;
-                $oSearch->iSearchRank += $oSearchTerm->iTermCount + 1;
-                if (empty($this->aName)) {
-                    $oSearch->iSearchRank++;
-                }
-                if (preg_match('#^[0-9]+$#', $sToken)) {
-                    $oSearch->iSearchRank++;
-                }
                 $oSearch->aAddress[$iWordID] = $iWordID;
-                $aNewSearches[] = $oSearch;
             } else {
-                $oSearch = clone $this;
-                $oSearch->iSearchRank += $oSearchTerm->iTermCount + 1;
                 $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
                 if (!empty($aFullTokens)) {
                     $oSearch->iSearchRank++;
                 }
-                $aNewSearches[] = $oSearch;
-
-                // revert to the token version?
-                foreach ($aFullTokens as $oSearchTermToken) {
-                    if (is_a($oSearchTermToken, '\Nominatim\Token\Word')) {
-                        $oSearch = clone $this;
-                        $oSearch->iSearchRank += 3;
-                        $oSearch->aAddress[$oSearchTermToken->iId]
-                            = $oSearchTermToken->iId;
-                        $aNewSearches[] = $oSearch;
-                    }
-                }
             }
+            $aNewSearches[] = $oSearch;
         }
 
         if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
-            && (empty($this->aName) || $this->iNamePhrase == $iPhrase)
+            && ((empty($this->aName) && empty($this->aNameNonSearch)) || $this->iNamePhrase == $iPhrase)
+            && strpos($sToken, ' ') === false
         ) {
             $oSearch = clone $this;
-            $oSearch->iSearchRank += 2;
-            if (empty($this->aName)) {
-                $oSearch->iSearchRank += 1;
+            $oSearch->iSearchRank++;
+            if (empty($this->aName) && empty($this->aNameNonSearch)) {
+                $oSearch->iSearchRank++;
             }
-            if (preg_match('#^[0-9]+$#', $sToken)) {
-                $oSearch->iSearchRank += 2;
+            if (preg_match('#^[0-9 ]+$#', $sToken)) {
+                $oSearch->iSearchRank++;
             }
             if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
                 if (empty($this->aName)
@@ -387,6 +383,9 @@ class SearchDescription
                 }
                 $oSearch->aName[$iWordID] = $iWordID;
             } else {
+                if (!empty($aFullTokens)) {
+                    $oSearch->iSearchRank++;
+                }
                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;
             }
             $oSearch->iNamePhrase = $iPhrase;
index f81c0ca89e915925b1b1e0da30c9647771fec5e8..0b0084957d6ecef3081409ce3a6e21da7cb6cfb9 100644 (file)
@@ -161,7 +161,7 @@ if ($aCMDResult['load-data'] || $aCMDResult['all']) {
 if ($aCMDResult['import-tiger-data']) {
     $bDidSomething = true;
     $sTigerPath = getSetting('TIGER_DATA_PATH', CONST_InstallDir.'/tiger');
-    $oSetup->importTigerData($sTigerPath);
+    run((clone($oNominatimCmd))->addParams('transition', '--tiger-data', $sTigerPath));
 }
 
 if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all']) {
index 4b6439a95c9142b0ce0724627cfb243cc3e8e210..cf2ac6da29ff0749c58f9cb92fa31725ca0e9e83 100755 (executable)
@@ -67,76 +67,6 @@ class SetupFunctions
         }
     }
 
-    public function importTigerData($sTigerPath)
-    {
-        info('Import Tiger data');
-
-        $aFilenames = glob($sTigerPath.'/*.sql');
-        info('Found '.count($aFilenames).' SQL files in path '.$sTigerPath);
-        if (empty($aFilenames)) {
-            warn('Tiger data import selected but no files found in path '.$sTigerPath);
-            return;
-        }
-        $sTemplate = file_get_contents(CONST_SqlDir.'/tiger_import_start.sql');
-        $sTemplate = $this->replaceSqlPatterns($sTemplate);
-
-        $this->pgsqlRunScript($sTemplate, false);
-
-        $aDBInstances = array();
-        for ($i = 0; $i < $this->iInstances; $i++) {
-            // https://secure.php.net/manual/en/function.pg-connect.php
-            $DSN = getSetting('DATABASE_DSN');
-            $DSN = preg_replace('/^pgsql:/', '', $DSN);
-            $DSN = preg_replace('/;/', ' ', $DSN);
-            $aDBInstances[$i] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW | PGSQL_CONNECT_ASYNC);
-            pg_ping($aDBInstances[$i]);
-        }
-
-        foreach ($aFilenames as $sFile) {
-            echo $sFile.': ';
-            $hFile = fopen($sFile, 'r');
-            $sSQL = fgets($hFile, 100000);
-            $iLines = 0;
-            while (true) {
-                for ($i = 0; $i < $this->iInstances; $i++) {
-                    if (!pg_connection_busy($aDBInstances[$i])) {
-                        while (pg_get_result($aDBInstances[$i]));
-                        $sSQL = fgets($hFile, 100000);
-                        if (!$sSQL) break 2;
-                        if (!pg_send_query($aDBInstances[$i], $sSQL)) fail(pg_last_error($aDBInstances[$i]));
-                        $iLines++;
-                        if ($iLines == 1000) {
-                            echo '.';
-                            $iLines = 0;
-                        }
-                    }
-                }
-                usleep(10);
-            }
-            fclose($hFile);
-
-            $bAnyBusy = true;
-            while ($bAnyBusy) {
-                $bAnyBusy = false;
-                for ($i = 0; $i < $this->iInstances; $i++) {
-                    if (pg_connection_busy($aDBInstances[$i])) $bAnyBusy = true;
-                }
-                usleep(10);
-            }
-            echo "\n";
-        }
-
-        for ($i = 0; $i < $this->iInstances; $i++) {
-            pg_close($aDBInstances[$i]);
-        }
-
-        info('Creating indexes on Tiger data');
-        $sTemplate = file_get_contents(CONST_SqlDir.'/tiger_import_finish.sql');
-        $sTemplate = $this->replaceSqlPatterns($sTemplate);
-
-        $this->pgsqlRunScript($sTemplate, false);
-    }
-
     public function calculatePostcodes($bCMDResultAll)
     {
         info('Calculate Postcodes');
index 374c00b39bcaa5bfa37aa1a66b703c89e61583f1..39ab1ae36f6394e7fdc5ee8d93a39ad172bcf6d3 100644 (file)
@@ -1,13 +1,15 @@
 --index only on parent_place_id
-CREATE INDEX idx_location_property_tiger_parent_place_id_imp ON location_property_tiger_import (parent_place_id) {ts:aux-index};
-CREATE UNIQUE INDEX idx_location_property_tiger_place_id_imp ON location_property_tiger_import (place_id) {ts:aux-index};
+CREATE INDEX {{sql.if_index_not_exists}} idx_location_property_tiger_place_id_imp
+  ON location_property_tiger_import (parent_place_id) {{db.tablespace.aux_index}};
+CREATE UNIQUE INDEX {{sql.if_index_not_exists}} idx_location_property_tiger_place_id_imp
+  ON location_property_tiger_import (place_id) {{db.tablespace.aux_index}};
 
-GRANT SELECT ON location_property_tiger_import TO "{www-user}";
+GRANT SELECT ON location_property_tiger_import TO "{{config.DATABASE_WEBUSER}}";
 
 DROP TABLE IF EXISTS location_property_tiger;
 ALTER TABLE location_property_tiger_import RENAME TO location_property_tiger;
 
-ALTER INDEX idx_location_property_tiger_parent_place_id_imp RENAME TO idx_location_property_tiger_housenumber_parent_place_id;
-ALTER INDEX idx_location_property_tiger_place_id_imp RENAME TO idx_location_property_tiger_place_id;
+ALTER INDEX IF EXISTS idx_location_property_tiger_parent_place_id_imp RENAME TO idx_location_property_tiger_housenumber_parent_place_id;
+ALTER INDEX IF EXISTS idx_location_property_tiger_place_id_imp RENAME TO idx_location_property_tiger_place_id;
 
 DROP FUNCTION tiger_line_import (linegeo geometry, in_startnumber integer, in_endnumber integer, interpolationtype text, in_street text, in_isin text, in_postcode text);
index 7459711f9b43dd1c8e370c209d14e6c38989fa6d..b3d9eee6d735742ef61941379a00ca5477930d68 100644 (file)
@@ -13,6 +13,7 @@ from .tools.exec_utils import run_legacy_script, run_php_server
 from .errors import UsageError
 from . import clicmd
 from .clicmd.args import NominatimArgs
+from .tools import tiger_data
 
 LOG = logging.getLogger()
 
@@ -166,8 +167,11 @@ class UpdateAddData:
     @staticmethod
     def run(args):
         if args.tiger_data:
-            os.environ['NOMINATIM_TIGER_DATA_PATH'] = args.tiger_data
-            return run_legacy_script('setup.php', '--import-tiger-data', nominatim_env=args)
+            return tiger_data.add_tiger_data(args.config.get_libpq_dsn(),
+                                             args.tiger_data,
+                                             args.threads or 1,
+                                             args.config,
+                                             args.sqllib_dir)
 
         params = ['update.php']
         if args.file:
index fd9382ebb4ed5704f1d538c4c9527ba673d103d8..03d7ca8a73e8c55a63f9571dca302f2a4bceff0c 100644 (file)
@@ -20,13 +20,16 @@ class AdminFuncs:
 
     @staticmethod
     def add_args(parser):
-        group = parser.add_argument_group('Admin task arguments')
-        group.add_argument('--warm', action='store_true',
-                           help='Warm database caches for search and reverse queries.')
-        group.add_argument('--check-database', action='store_true',
-                           help='Check that the database is complete and operational.')
-        group.add_argument('--analyse-indexing', action='store_true',
-                           help='Print performance analysis of the indexing process.')
+        group = parser.add_argument_group('Admin tasks')
+        objs = group.add_mutually_exclusive_group(required=True)
+        objs.add_argument('--warm', action='store_true',
+                          help='Warm database caches for search and reverse queries.')
+        objs.add_argument('--check-database', action='store_true',
+                          help='Check that the database is complete and operational.')
+        objs.add_argument('--migrate', action='store_true',
+                          help='Migrate the database to a new software version.')
+        objs.add_argument('--analyse-indexing', action='store_true',
+                          help='Print performance analysis of the indexing process.')
         group = parser.add_argument_group('Arguments for cache warming')
         group.add_argument('--search-only', action='store_const', dest='target',
                            const='search',
@@ -44,7 +47,7 @@ class AdminFuncs:
     @staticmethod
     def run(args):
         if args.warm:
-            AdminFuncs._warm(args)
+            return AdminFuncs._warm(args)
 
         if args.check_database:
             LOG.warning('Checking database')
@@ -56,8 +59,14 @@ class AdminFuncs:
             from ..tools import admin
             with connect(args.config.get_libpq_dsn()) as conn:
                 admin.analyse_indexing(conn, osm_id=args.osm_id, place_id=args.place_id)
+            return 0
 
-        return 0
+        if args.migrate:
+            LOG.warning('Checking for necessary database migrations')
+            from ..tools import migration
+            return migration.migrate(args.config, args)
+
+        return 1
 
 
     @staticmethod
index b8db1a38ee745fd2062f2dd63aa49d21707e3710..0a89cb037260dfa7eeb9344d9da68c215a1b2dfa 100644 (file)
@@ -58,10 +58,12 @@ class AdminTransition:
                            help="Ignore certain erros on import.")
         group.add_argument('--reverse-only', action='store_true',
                            help='Do not create search tables and indexes')
+        group.add_argument('--tiger-data', metavar='FILE',
+                           help='File to import')
 
     @staticmethod
     def run(args):
-        from ..tools import database_import
+        from ..tools import database_import, tiger_data
         from ..tools import refresh
 
         if args.create_db:
@@ -127,3 +129,11 @@ class AdminTransition:
             LOG.warning('Create Search indices')
             with connect(args.config.get_libpq_dsn()) as conn:
                 database_import.create_search_indices(conn, args.config, args.sqllib_dir, args.drop)
+
+        if args.tiger_data:
+            LOG.warning('Tiger data')
+            tiger_data.add_tiger_data(args.config.get_libpq_dsn(),
+                                      args.tiger_data,
+                                      args.threads or 1,
+                                      args.config,
+                                      args.sqllib_dir)
diff --git a/nominatim/tools/migration.py b/nominatim/tools/migration.py
new file mode 100644 (file)
index 0000000..756a1d3
--- /dev/null
@@ -0,0 +1,132 @@
+"""
+Functions for database migration to newer software versions.
+"""
+import logging
+
+from ..db import properties
+from ..db.connection import connect
+from ..version import NOMINATIM_VERSION
+from . import refresh, database_import
+from ..errors import UsageError
+
+LOG = logging.getLogger()
+
+_MIGRATION_FUNCTIONS = []
+
+def migrate(config, paths):
+    """ Check for the current database version and execute migrations,
+        if necesssary.
+    """
+    with connect(config.get_libpq_dsn()) as conn:
+        if conn.table_exists('nominatim_properties'):
+            db_version_str = properties.get_property(conn, 'database_version')
+        else:
+            db_version_str = None
+
+        if db_version_str is not None:
+            parts = db_version_str.split('.')
+            db_version = tuple([int(x) for x in parts[:2] + parts[2].split('-')])
+
+            if db_version == NOMINATIM_VERSION:
+                LOG.warning("Database already at latest version (%s)", db_version_str)
+                return 0
+
+            LOG.info("Detected database version: %s", db_version_str)
+        else:
+            db_version = _guess_version(conn)
+
+
+        has_run_migration = False
+        for version, func in _MIGRATION_FUNCTIONS:
+            if db_version <= version:
+                LOG.warning("Runnning: %s (%s)", func.__doc__.split('\n', 1)[0],
+                            '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(version))
+                kwargs = dict(conn=conn, config=config, paths=paths)
+                func(**kwargs)
+                has_run_migration = True
+
+        if has_run_migration:
+            LOG.warning('Updating SQL functions.')
+            refresh.create_functions(conn, config, paths.sqllib_dir)
+
+        properties.set_property(conn, 'database_version',
+                                '{0[0]}.{0[1]}.{0[2]}-{0[3]}'.format(NOMINATIM_VERSION))
+
+        conn.commit()
+
+    return 0
+
+
+def _guess_version(conn):
+    """ Guess a database version when there is no property table yet.
+        Only migrations for 3.6 and later are supported, so bail out
+        when the version seems older.
+    """
+    with conn.cursor() as cur:
+        # In version 3.6, the country_name table was updated. Check for that.
+        cnt = cur.scalar("""SELECT count(*) FROM
+                            (SELECT svals(name) FROM  country_name
+                             WHERE country_code = 'gb')x;
+                         """)
+        if cnt < 100:
+            LOG.fatal('It looks like your database was imported with a version '
+                      'prior to 3.6.0. Automatic migration not possible.')
+            raise UsageError('Migration not possible.')
+
+    return (3, 5, 0, 99)
+
+
+
+def _migration(major, minor, patch=0, dbpatch=0):
+    """ Decorator for a single migration step. The parameters describe the
+        version after which the migration is applicable, i.e before changing
+        from the given version to the next, the migration is required.
+
+        All migrations are run in the order in which they are defined in this
+        file. Do not run global SQL scripts for migrations as you cannot be sure
+        that these scripts do the same in later versions.
+
+        Functions will always be reimported in full at the end of the migration
+        process, so the migration functions may leave a temporary state behind
+        there.
+    """
+    def decorator(func):
+        _MIGRATION_FUNCTIONS.append(((major, minor, patch, dbpatch), func))
+
+    return decorator
+
+
+@_migration(3, 5, 0, 99)
+def import_status_timestamp_change(conn, **_):
+    """ Add timezone to timestamp in status table.
+
+        The import_status table has been changed to include timezone information
+        with the time stamp.
+    """
+    with conn.cursor() as cur:
+        cur.execute("""ALTER TABLE import_status ALTER COLUMN lastimportdate
+                       TYPE timestamp with time zone;""")
+
+
+@_migration(3, 5, 0, 99)
+def install_database_module_in_project_directory(conn, config, paths, **_):
+    """ Install database module in project directory.
+
+        The database module needs to be present in the project directory
+        since those were introduced.
+    """
+    database_import.install_module(paths.module_dir, paths.project_dir,
+                                   config.DATABASE_MODULE_PATH, conn=conn)
+
+
+@_migration(3, 5, 0, 99)
+def add_nominatim_property_table(conn, config, **_):
+    """ Add nominatim_property table.
+    """
+    if not conn.table_exists('nominatim_properties'):
+        with conn.cursor() as cur:
+            cur.execute("""CREATE TABLE nominatim_properties (
+                               property TEXT,
+                               value TEXT);
+                           GRANT SELECT ON TABLE nominatim_properties TO "{}";
+                        """.format(config.DATABASE_WEBUSER))
diff --git a/nominatim/tools/tiger_data.py b/nominatim/tools/tiger_data.py
new file mode 100644 (file)
index 0000000..9b960e2
--- /dev/null
@@ -0,0 +1,112 @@
+"""
+Functions for importing tiger data and handling tarbar and directory files
+"""
+import logging
+import os
+import tarfile
+import selectors
+
+from ..db.connection import connect
+from ..db.async_connection import DBConnection
+from ..db.sql_preprocessor import SQLPreprocessor
+
+
+LOG = logging.getLogger()
+
+
+def handle_tarfile_or_directory(data_dir):
+    """ Handles tarfile or directory for importing tiger data
+    """
+
+    tar = None
+    if data_dir.endswith('.tar.gz'):
+        tar = tarfile.open(data_dir)
+        sql_files = [i for i in tar.getmembers() if i.name.endswith('.sql')]
+        LOG.warning("Found %d SQL files in tarfile with path %s", len(sql_files), data_dir)
+        if not sql_files:
+            LOG.warning("Tiger data import selected but no files in tarfile's path %s", data_dir)
+            return None, None
+    else:
+        files = os.listdir(data_dir)
+        sql_files = [os.path.join(data_dir, i) for i in files if i.endswith('.sql')]
+        LOG.warning("Found %d SQL files in path %s", len(sql_files), data_dir)
+        if not sql_files:
+            LOG.warning("Tiger data import selected but no files found in path %s", data_dir)
+            return None, None
+
+    return sql_files, tar
+
+
+def handle_threaded_sql_statements(sel, file):
+    """ Handles sql statement with multiplexing
+    """
+
+    lines = 0
+    end_of_file = False
+    # Using pool of database connections to execute sql statements
+    while not end_of_file:
+        for key, _ in sel.select(1):
+            conn = key.data
+            try:
+                if conn.is_done():
+                    sql_query = file.readline()
+                    lines += 1
+                    if not sql_query:
+                        end_of_file = True
+                        break
+                    conn.perform(sql_query)
+                    if lines == 1000:
+                        print('. ', end='', flush=True)
+                        lines = 0
+            except Exception as exc: # pylint: disable=broad-except
+                LOG.info('Wrong SQL statement: %s', exc)
+
+
+def add_tiger_data(dsn, data_dir, threads, config, sqllib_dir):
+    """ Import tiger data from directory or tar file
+    """
+
+    sql_files, tar = handle_tarfile_or_directory(data_dir)
+
+    if not sql_files:
+        return
+
+    with connect(dsn) as conn:
+        sql = SQLPreprocessor(conn, config, sqllib_dir)
+        sql.run_sql_file(conn, 'tiger_import_start.sql')
+
+    # Reading sql_files and then for each file line handling
+    # sql_query in <threads - 1> chunks.
+    sel = selectors.DefaultSelector()
+    place_threads = max(1, threads - 1)
+
+    # Creates a pool of database connections
+    for _ in range(place_threads):
+        conn = DBConnection(dsn)
+        conn.connect()
+        sel.register(conn, selectors.EVENT_WRITE, conn)
+
+    for sql_file in sql_files:
+        if not tar:
+            file = open(sql_file)
+        else:
+            file = tar.extractfile(sql_file)
+
+        handle_threaded_sql_statements(sel, file)
+
+    # Unregistering pool of database connections
+    while place_threads > 0:
+        for key, _ in sel.select(1):
+            conn = key.data
+            sel.unregister(conn)
+            conn.wait()
+            conn.close()
+            place_threads -= 1
+
+    if tar:
+        tar.close()
+    print('\n')
+    LOG.warning("Creating indexes on Tiger data")
+    with connect(dsn) as conn:
+        sql = SQLPreprocessor(conn, config, sqllib_dir)
+        sql.run_sql_file(conn, 'tiger_import_finish.sql')
index d16dceffccb1008a1fe6082ab0a5e9d04cd2a439..4b7cccc39f09992b73c48d348ebd150a4942d8b9 100644 (file)
@@ -13,6 +13,7 @@ sys.path.insert(0, str(SRC_DIR.resolve()))
 
 from nominatim.config import Configuration
 from nominatim.db import connection
+from nominatim.db.sql_preprocessor import SQLPreprocessor
 
 class _TestingCursor(psycopg2.extras.DictCursor):
     """ Extension to the DictCursor class that provides execution
@@ -266,3 +267,9 @@ def osm2pgsql_options(temp_db):
                 flatnode_file='',
                 tablespaces=dict(slim_data='', slim_index='',
                                  main_data='', main_index=''))
+
+@pytest.fixture
+def sql_preprocessor(temp_db_conn, tmp_path, def_config, monkeypatch, table_factory):
+    monkeypatch.setenv('NOMINATIM_DATABASE_MODULE_PATH', '.')
+    table_factory('country_name', 'partition INT', (0, 1, 2))
+    return SQLPreprocessor(temp_db_conn, def_config, tmp_path)
index 418f4bcfdeb7b1f4ded6a1b8c474e6c59c45d4d5..1f9cd06d4be7e0401e6162ac2cc60da2bea49149 100644 (file)
@@ -65,7 +65,6 @@ def test_cli_help(capsys):
 
 @pytest.mark.parametrize("command,script", [
                          (('special-phrases',), 'specialphrases'),
-                         (('add-data', '--tiger-data', 'tiger'), 'setup'),
                          (('add-data', '--file', 'foo.osm'), 'update'),
                          (('export',), 'export')
                          ])
index 3c10000f9169c883f56547665fef1e171f27ba90..08a195bd2d794d9c6f9f01cd18116df64d466b42 100644 (file)
@@ -5,8 +5,6 @@ from pathlib import Path
 
 import pytest
 
-from nominatim.db.sql_preprocessor import SQLPreprocessor
-
 @pytest.fixture
 def sql_factory(tmp_path):
     def _mk_sql(sql_body):
@@ -21,13 +19,6 @@ def sql_factory(tmp_path):
 
     return _mk_sql
 
-
-@pytest.fixture
-def sql_preprocessor(temp_db_conn, tmp_path, def_config, monkeypatch, table_factory):
-    monkeypatch.setenv('NOMINATIM_DATABASE_MODULE_PATH', '.')
-    table_factory('country_name', 'partition INT', (0, 1, 2))
-    return SQLPreprocessor(temp_db_conn, def_config, tmp_path)
-
 @pytest.mark.parametrize("expr,ret", [
     ("'a'", 'a'),
     ("'{{db.partitions|join}}'", '012'),
diff --git a/test/python/test_tools_tiger_data.py b/test/python/test_tools_tiger_data.py
new file mode 100644 (file)
index 0000000..5029a13
--- /dev/null
@@ -0,0 +1,37 @@
+"""
+Test for tiger data function
+"""
+from pathlib import Path
+
+import pytest
+import tarfile
+
+from nominatim.tools import tiger_data, database_import
+
+
+@pytest.mark.parametrize("threads", (1, 5))
+def test_add_tiger_data(dsn, src_dir, def_config, tmp_path, sql_preprocessor,
+                        temp_db_cursor, threads, temp_db):
+    temp_db_cursor.execute('CREATE EXTENSION hstore')
+    temp_db_cursor.execute('CREATE EXTENSION postgis')
+    temp_db_cursor.execute('CREATE TABLE place (id INT)')
+    sqlfile = tmp_path / '1010.sql'
+    sqlfile.write_text("""INSERT INTO place values (1)""")
+    tiger_data.add_tiger_data(dsn, str(tmp_path), threads, def_config, src_dir / 'lib-sql')
+
+    assert temp_db_cursor.table_rows('place') == 1
+
+@pytest.mark.parametrize("threads", (1, 5))
+def test_add_tiger_data_tarfile(dsn, src_dir, def_config, tmp_path,
+                        temp_db_cursor, threads, temp_db, sql_preprocessor):
+    temp_db_cursor.execute('CREATE EXTENSION hstore')
+    temp_db_cursor.execute('CREATE EXTENSION postgis')
+    temp_db_cursor.execute('CREATE TABLE place (id INT)')
+    sqlfile = tmp_path / '1010.sql'
+    sqlfile.write_text("""INSERT INTO place values (1)""")
+    tar = tarfile.open("sample.tar.gz", "w:gz")
+    tar.add(sqlfile)
+    tar.close()
+    tiger_data.add_tiger_data(dsn, str(src_dir / 'sample.tar.gz'), threads, def_config, src_dir / 'lib-sql')
+    
+    assert temp_db_cursor.table_rows('place') == 1
\ No newline at end of file