Merge branch 'encukou'

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

diff --git a/pokedex/db/load.py b/pokedex/db/load.py
index e6b3287..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')
-        for table in reversed(table_objs):
+        for n, table in enumerate(reversed(table_objs)):
             table.drop(checkfirst=True)
+            print_status('%s/%s' % (n, len(table_objs)))
         print_done()
 
-    for table in table_objs:
+    print_start('Creating tables')
+    for n, table in enumerate(table_objs):
         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
@@ -168,6 +172,32 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False, s
         reader = csv.reader(csvfile, lineterminator='\n')
         column_names = [unicode(column) for column in reader.next()]
 
+        if not safe and session.connection().dialect.name == 'postgresql':
+            """
+            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]
+            if not_null_cols:
+                force_not_null = 'FORCE NOT NULL ' + ','.join('"%s"' % c for c in not_null_cols)
+            else:
+                force_not_null = ''
+            command = "COPY {table_name} ({columns}) FROM '{csvpath}' CSV HEADER {force_not_null}"
+            session.connection().execute(
+                    command.format(
+                            table_name=table_name,
+                            csvpath=csvpath,
+                            columns=','.join('"%s"' % c for c in column_names),
+                            force_not_null=force_not_null,
+                        )
+                )
+            session.commit()
+            print_done()
+            continue
+
         # Self-referential tables may contain rows with foreign keys of other
         # rows in the same table that do not yet exist.  Pull these out and add
         # them to the session last
@@ -183,6 +213,8 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False, s
 
         new_rows = []
         def insert_and_commit():
+            if not new_rows:
+                return
             session.connection().execute(insert_stmt, new_rows)
             session.commit()
             new_rows[:] = []