Speed up `import pokedex.db` slightly.

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

diff --git a/pokedex/db/load.py b/pokedex/db/load.py
index 4512bf6..6a548b8 100644 (file)
--- a/pokedex/db/load.py
+++ b/pokedex/db/load.py
@@ -140,12 +140,16 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False, s
     # Drop all tables if requested
     if drop_tables:
         print_start('Dropping tables')
     # Drop all tables if requested
     if drop_tables:
         print_start('Dropping tables')

-        for table in reversed(table_objs):
+        for n, table in enumerate(reversed(table_objs)):

             table.drop(checkfirst=True)
             table.drop(checkfirst=True)

+            print_status('%s/%s' % (n, len(table_objs)))

         print_done()
 
         print_done()
 

-    for table in table_objs:
+    print_start('Creating tables')
+    for n, table in enumerate(table_objs):

         table.create()
         table.create()

+        print_status('%s/%s' % (n, len(table_objs)))
+    print_done()

     connection = session.connection()
 
     # Okay, run through the tables and actually load the data now
     connection = session.connection()
 
     # Okay, run through the tables and actually load the data now


@@ -170,14 +174,10 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False, s
 
         if not safe and session.connection().dialect.name == 'postgresql':
             """
 
         if not safe and session.connection().dialect.name == 'postgresql':
             """

-            Postgres' CSV dialect is nearly the same as ours, except that it
-            treats completely empty values as NULL, and empty quoted
-            strings ("") as an empty strings.
-            Pokedex dump does not quote empty strings. So, both empty strings
-            and NULLs are read in as NULL.
-            For an empty string in a NOT NULL column, the load will fail, and
-            load will fall back to the cross-backend row-by-row loading. And in
-            nullable columns, we already load empty stings as NULL.
+            Postgres' CSV dialect works with our data, if we mark the not-null
+            columns with FORCE NOT NULL.
+            COPY is only allowed for DB superusers. If you're not one, use safe
+            loading (pokedex load -S).

             """
             session.commit()
             not_null_cols = [c for c in column_names if not table_obj.c[c].nullable]
             """
             session.commit()
             not_null_cols = [c for c in column_names if not table_obj.c[c].nullable]


@@ -213,6 +213,8 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False, s
 
         new_rows = []
         def insert_and_commit():
 
         new_rows = []
         def insert_and_commit():

+            if not new_rows:
+                return

             session.connection().execute(insert_stmt, new_rows)
             session.commit()
             new_rows[:] = []
             session.connection().execute(insert_stmt, new_rows)
             session.commit()
             new_rows[:] = []