Merge pull request #11 from giuli007/exclude_tables

Add an --exclude option to the dump command

Author: simonw

README.md

@@ -24,6 +24,10 @@ To dump out every table in that database, use `--all`:
 
     sqlite-diffable dump fixtures.db dump/ --all
 
+To dump all table except some specific ones, use `--exclude` one or more times:
+
+    sqlite-diffable dump fixtures.db dump/ --all --exclude unwanted_first_table --exclude unwanted_second_table
+
 ## Loading a database
 
 To load a previously dumped database, run the following:

sqlite_diffable/cli.py

@@ -24,25 +24,32 @@ def cli():
 )
 @click.argument("tables", nargs=-1, required=False)
 @click.option("--all", is_flag=True, help="Dump all tables")
-def dump(dbpath, output, tables, all):
+@click.option("--exclude", multiple=True, help="Tables to exclude from the dump")
+def dump(dbpath, output, tables, all, exclude):
     """
     Dump a SQLite database out as flat files in the directory
 
     Usage:
 
         sqlite-diffable dump my.db output/ --all
 
-    --all dumps ever table. Or specify tables like this:
+    --all dumps every table. Or specify tables like this:
 
         sqlite-diffable dump my.db output/ entries tags
+
+    Exclude specific tables:
+
+        sqlite-diffable dump my.db output/ --all --exclude table1 --exclude table2
     """
     if not tables and not all:
         raise click.ClickException("You must pass --all or specify some tables")
     output = pathlib.Path(output)
     output.mkdir(exist_ok=True)
     conn = sqlite_utils.Database(dbpath)
     if all:
-        tables = conn.table_names()
+        tables = set(conn.table_names()) - set(exclude)
+    else:
+        tables = set(tables) - set(exclude)
     for table in tables:
         tablename = table.replace("/", "")
         filepath = output / "{}.ndjson".format(tablename)

tests/test_dump.py

@@ -37,6 +37,18 @@ def test_dump_all(two_tables_db, tmpdir):
     assert (output_dir / "second_table.metadata.json").exists()
 
 
+def test_dump_exclude(two_tables_db, tmpdir):
+    output_dir = tmpdir / "out"
+    result = CliRunner().invoke(
+        cli.cli, ["dump", two_tables_db, str(output_dir), "--all", "--exclude", "second_table"]
+    )
+    assert result.exit_code == 0, result.output
+    assert (output_dir / "one_table.ndjson").exists()
+    assert (output_dir / "one_table.metadata.json").exists()
+    assert not (output_dir / "second_table.ndjson").exists()
+    assert not (output_dir / "second_table.metadata.json").exists()
+
+
 def test_load(two_tables_db, tmpdir):
     output_dir = tmpdir / "out"
     restore_db = tmpdir / "restore.db"