Merge pull request #2250 from lonvia/save-transliterated-housenumbers

[nominatim.git] / nominatim / tools / tiger_data.py

diff --git a/nominatim/tools/tiger_data.py b/nominatim/tools/tiger_data.py
index 26f6beb24ca13891e13e1407b3a848fd54568bbc..9b960e2d54ee82df6f05c921ccb44062f33d9ec6 100644 (file)
--- a/nominatim/tools/tiger_data.py
+++ b/nominatim/tools/tiger_data.py
@@ -1,5 +1,5 @@
 """
-Functions for setting up and importing a new Nominatim database.
+Functions for importing tiger data and handling tarbar and directory files
 """
 import logging
 import os
@@ -10,32 +10,66 @@ from ..db.connection import connect
 from ..db.async_connection import DBConnection
 from ..db.sql_preprocessor import SQLPreprocessor
 
-# pylint: disable=R0912
-# pylint: disable=R0914,R0915,W0702
 
 LOG = logging.getLogger()
 
 
-def add_tiger_data(dsn, data_dir, threads, config, sqllib_dir):
-    """ Import tiger data from directory or tar file
+def handle_tarfile_or_directory(data_dir):
+    """ Handles tarfile or directory for importing tiger data
     """
-    # Handling directory or tarball file.
-    is_tarfile = False
+
+    tar = None
     if data_dir.endswith('.tar.gz'):
-        is_tarfile = True
         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
+            return None, None
     else:
         files = os.listdir(data_dir)
-        sql_files = [i for i in files if i.endswith('.sql')]
+        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
+            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)
@@ -45,54 +79,34 @@ def add_tiger_data(dsn, data_dir, threads, config, sqllib_dir):
     # 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 is_tarfile:
-            file_path = os.path.join(data_dir, sql_file)
-            file = open(file_path)
+        if not tar:
+            file = open(sql_file)
         else:
             file = tar.extractfile(sql_file)
-        lines = 0
-        end_of_file = False
-        total_used_threads = place_threads
-        while True :
-            if end_of_file:
-                break
-            for imod in range(place_threads):
-                conn = DBConnection(dsn)
-                conn.connect()
-
-                sql_query = file.readline()
-                lines += 1
-
-                if not sql_query:
-                    end_of_file = True
-                    total_used_threads = imod
-                    break
-
-                conn.perform(sql_query)
-                sel.register(conn, selectors.EVENT_READ, conn)
-
-                if lines == 1000:
-                    print('. ', end='', flush=True)
-                    lines = 0
-
-            todo = min(place_threads, total_used_threads)
-            while todo > 0:
-                for key, _ in sel.select(1):
-                    try:
-                        conn = key.data
-                        sel.unregister(conn)
-                        conn.wait()
-                        conn.close()
-                        todo -= 1
-                    except:
-                        todo -= 1
-
-    if is_tarfile:
+
+        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')
-    
\ No newline at end of file