Speed tweaks for pokedex load in SQLite

[zzz-pokedex.git] / pokedex / db / load.py

diff --git a/pokedex/db/load.py b/pokedex/db/load.py
index d5502ea..e6b3287 100644 (file)
--- a/pokedex/db/load.py
+++ b/pokedex/db/load.py
@@ -2,7 +2,6 @@
 import csv
 import fnmatch
 import os.path
-import pkg_resources
 import sys
 
 from sqlalchemy.orm.attributes import instrumentation_registry
@@ -11,6 +10,7 @@ import sqlalchemy.types
 
 from pokedex.db import metadata
 import pokedex.db.tables as tables
+from pokedex.defaults import get_default_csv_dir
 
 
 def _get_table_names(metadata, patterns):
@@ -95,7 +95,7 @@ def _get_verbose_prints(verbose):
     return print_start, print_status, print_done
 
 
-def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
+def load(session, tables=[], directory=None, drop_tables=False, verbose=False, safe=True):
     """Load data from CSV files into the given database session.
 
     Tables are created automatically.
@@ -115,19 +115,27 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
 
     `verbose`
         If set to True, status messages will be printed to stdout.
+
+    `safe`
+        If set to False, load can be faster, but can corrupt the database if
+        it crashes or is interrupted.
     """
 
     # First take care of verbosity
     print_start, print_status, print_done = _get_verbose_prints(verbose)
 
 
-    if not directory:
-        directory = pkg_resources.resource_filename('pokedex', 'data/csv')
+    if directory is None:
+        directory = get_default_csv_dir()
 
     table_names = _get_table_names(metadata, tables)
     table_objs = [metadata.tables[name] for name in table_names]
     table_objs = sqlalchemy.sql.util.sort_tables(table_objs)
 
+    # SQLite speed tweaks
+    if not safe and session.connection().dialect.name == 'sqlite':
+        session.connection().execute("PRAGMA synchronous=OFF")
+        session.connection().execute("PRAGMA journal_mode=OFF")
 
     # Drop all tables if requested
     if drop_tables:
@@ -179,7 +187,7 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
             session.commit()
             new_rows[:] = []
 
-            progress = "{0}%".format(100 * csvfile.tell() // csvsize)
+            progress = "%d%%" % (100 * csvfile.tell() // csvsize)
             print_status(progress)
 
         for csvs in reader:
@@ -251,6 +259,10 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
 
         print_done()
 
+    # SQLite check
+    if session.connection().dialect.name == 'sqlite':
+        session.connection().execute("PRAGMA integrity_check")
+
 
 
 def dump(session, tables=[], directory=None, verbose=False):
@@ -276,7 +288,7 @@ def dump(session, tables=[], directory=None, verbose=False):
 
 
     if not directory:
-        directory = pkg_resources.resource_filename('pokedex', 'data/csv')
+        directory = get_default_csv_dir()
 
     table_names = _get_table_names(metadata, tables)
     table_names.sort()