Merge pull request #2204 from darkshredder/tiger-data

Ported tiger-data-import to Python and Added Tarball Support

nominatim/cli.py

@@ -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:

nominatim/clicmd/transition.py

@@ -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)

nominatim/tools/tiger_data.py

@@ -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')