Merge remote-tracking branch 'upstream/master'

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

diff --git a/nominatim/tools/tiger_data.py b/nominatim/tools/tiger_data.py
index 26f6beb24ca13891e13e1407b3a848fd54568bbc..ff498f777e527a38adbc38a48e500f5b514bc684 100644 (file)
--- a/nominatim/tools/tiger_data.py
+++ b/nominatim/tools/tiger_data.py
@@ -1,98 +1,108 @@
 """
 """

-Functions for setting up and importing a new Nominatim database.
+Functions for importing tiger data and handling tarbar and directory files

 """
 """

+import csv
+import io

 import logging
 import os
 import tarfile
 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
+import psycopg2.extras

 
 

-# pylint: disable=R0912
-# pylint: disable=R0914,R0915,W0702
+from nominatim.db.connection import connect
+from nominatim.db.async_connection import WorkerPool
+from nominatim.db.sql_preprocessor import SQLPreprocessor
+from nominatim.errors import UsageError

 
 LOG = logging.getLogger()
 
 
 
 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'):
     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:
+        try:
+            tar = tarfile.open(data_dir)
+        except tarfile.ReadError as err:
+            LOG.fatal("Cannot open '%s'. Is this a tar file?", data_dir)
+            raise UsageError("Cannot open Tiger data file.") from err
+
+        csv_files = [i for i in tar.getmembers() if i.name.endswith('.csv')]
+        LOG.warning("Found %d CSV files in tarfile with path %s", len(csv_files), data_dir)
+        if not csv_files:

             LOG.warning("Tiger data import selected but no files in tarfile's path %s", data_dir)
             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)
     else:
         files = os.listdir(data_dir)

-        sql_files = [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:
+        csv_files = [os.path.join(data_dir, i) for i in files if i.endswith('.csv')]
+        LOG.warning("Found %d CSV files in path %s", len(csv_files), data_dir)
+        if not csv_files:

             LOG.warning("Tiger data import selected but no files found in path %s", data_dir)
             LOG.warning("Tiger data import selected but no files found in path %s", data_dir)

-            return
+            return None, None
+
+    return csv_files, tar
+
+
+def handle_threaded_sql_statements(pool, fd, analyzer):
+    """ Handles sql statement with multiplexing
+    """
+    lines = 0
+    # Using pool of database connections to execute sql statements
+
+    sql = "SELECT tiger_line_import(%s, %s, %s, %s, %s, %s)"
+
+    for row in csv.DictReader(fd, delimiter=';'):
+        try:
+            address = dict(street=row['street'], postcode=row['postcode'])
+            args = ('SRID=4326;' + row['geometry'],
+                    int(row['from']), int(row['to']), row['interpolation'],
+                    psycopg2.extras.Json(analyzer.process_place(dict(address=address))),
+                    analyzer.normalize_postcode(row['postcode']))
+        except ValueError:
+            continue
+        pool.next_free_worker().perform(sql, args=args)
+
+        lines += 1
+        if lines == 1000:
+            print('.', end='', flush=True)
+            lines = 0
+
+
+def add_tiger_data(data_dir, config, threads, tokenizer):
+    """ Import tiger data from directory or tar file `data dir`.
+    """
+    dsn = config.get_libpq_dsn()
+    files, tar = handle_tarfile_or_directory(data_dir)
+
+    if not files:
+        return

 
     with connect(dsn) as conn:
 
     with connect(dsn) as conn:

-        sql = SQLPreprocessor(conn, config, sqllib_dir)
+        sql = SQLPreprocessor(conn, config)

         sql.run_sql_file(conn, 'tiger_import_start.sql')
 
         sql.run_sql_file(conn, 'tiger_import_start.sql')
 

-    # Reading sql_files and then for each file line handling
+    # Reading files and then for each file line handling

     # sql_query in <threads - 1> chunks.
     # sql_query in <threads - 1> chunks.

-    sel = selectors.DefaultSelector()

     place_threads = max(1, threads - 1)
     place_threads = max(1, threads - 1)

-    for sql_file in sql_files:
-        if not is_tarfile:
-            file_path = os.path.join(data_dir, sql_file)
-            file = open(file_path)
-        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:
+
+    with WorkerPool(dsn, place_threads, ignore_sql_errors=True) as pool:
+        with tokenizer.name_analyzer() as analyzer:
+            for fname in files:
+                if not tar:
+                    fd = open(fname)
+                else:
+                    fd = io.TextIOWrapper(tar.extractfile(fname))
+
+                handle_threaded_sql_statements(pool, fd, analyzer)
+
+                fd.close()
+
+    if tar:

         tar.close()
     print('\n')
     LOG.warning("Creating indexes on Tiger data")
     with connect(dsn) as conn:
         tar.close()
     print('\n')
     LOG.warning("Creating indexes on Tiger data")
     with connect(dsn) as conn:

-        sql = SQLPreprocessor(conn, config, sqllib_dir)
+        sql = SQLPreprocessor(conn, config)

         sql.run_sql_file(conn, 'tiger_import_finish.sql')
         sql.run_sql_file(conn, 'tiger_import_finish.sql')

-    
\ No newline at end of file