Merge remote-tracking branch 'upstream/262-shallow-extraction' into develop

Author: pombredanne

src/scancode/extract_cli.py

@@ -86,12 +86,13 @@ class ExtractCommand(utils.BaseCommand):
 
 @click.option('--verbose', is_flag=True, default=False, help='Print verbose file-by-file progress messages.')
 @click.option('--quiet', is_flag=True, default=False, help='Do not print any summary or progress message.')
+@click.option('--shallow', is_flag=True, default=False, help='Do not extract recursively nested archives (e.g. not archives in archives).')
 
 @click.help_option('-h', '--help')
 @click.option('--about', is_flag=True, is_eager=True, callback=print_about, help='Show information about ScanCode and licensing and exit.')
 @click.option('--version', is_flag=True, is_eager=True, callback=print_version, help='Show the version and exit.')
 
-def extractcode(ctx, input, verbose, quiet, *args, **kwargs):  # @ReservedAssignment
+def extractcode(ctx, input, verbose, quiet, shallow, *args, **kwargs):  # @ReservedAssignment
     """extract archives and compressed files found in the <input> file or directory tree.
 
     Use this command before scanning proper as an <input> preparation step.
@@ -156,7 +157,7 @@ def display_extract_summary():
     if not quiet:
         echo_stderr('Extracting archives...', fg='green')
 
-    with utils.progressmanager(extract_archives(abs_location), item_show_func=extract_event,
+    with utils.progressmanager(extract_archives(abs_location, recurse=not shallow), item_show_func=extract_event,
                                verbose=verbose, quiet=quiet) as extraction_events:
         for xev in extraction_events:
             if xev.done and (xev.warnings or xev.errors):

tests/scancode/data/extract_shallow/top.zip

@@ -209,3 +209,21 @@ def test_extractcode_command_can_extract_archive_with_unicode_names(monkeypatch)
         '/unicodepath/koristenjem_Karkkainen_-_Sander.pdf'
     ]
     assert sorted(expected) == sorted(file_result)
+
+
+def test_extractcode_command_can_extract_shallow(monkeypatch):
+    test_dir = test_env.get_test_loc('extract_shallow', copy=True)
+    monkeypatch.setattr(click._termui_impl, 'isatty', lambda _: True)
+    runner = CliRunner()
+    result = runner.invoke(extract_cli.extractcode, ['--shallow', test_dir])
+    assert result.exit_code == 0
+    file_result = [f for f in map(as_posixpath, file_iter(test_dir)) if not f.endswith('unicodepath.tgz')]
+    file_result = [''.join(f.partition('/top.zip-extract/')[1:]) for f in file_result]
+    file_result = [f for f in file_result if f]
+    # this checks that the zip in top.zip are not extracted
+    expected = [
+        '/top.zip-extract/some3.zip',
+        '/top.zip-extract/some2.zip',
+        '/top.zip-extract/some1.zip',
+    ]
+    assert sorted(expected) == sorted(file_result)