def test_trailing_slash(self):
archive_dir = self.makedirs('archive') + os.sep
runner = InlineMRJobRunner()
archive_path = runner._dir_archive_path(archive_dir)
self.assertEqual(os.path.basename(archive_path), 'archive.tar.gz')
def test_doesnt_actually_create_archive(self):
archive_dir = self.makedirs('archive')
runner = InlineMRJobRunner()
archive_path = runner._dir_archive_path(archive_dir)
self.assertFalse(os.path.exists(archive_path))
def test_doesnt_actually_create_archive(self):
archive_dir = self.makedirs('archive')
runner = InlineMRJobRunner()
archive_path = runner._dir_archive_path(archive_dir)
self.assertFalse(os.path.exists(archive_path))
def test_trailing_slash(self):
archive_dir = self.makedirs('archive') + os.sep
runner = InlineMRJobRunner()
archive_path = runner._dir_archive_path(archive_dir)
self.assertEqual(os.path.basename(archive_path), 'archive.tar.gz')
def test_same_dir_twice(self):
archive_dir = self.makedirs('archive')
runner = InlineMRJobRunner()
archive_path_1 = runner._dir_archive_path(archive_dir)
archive_path_2 = runner._dir_archive_path(archive_dir)
self.assertEqual(os.path.basename(archive_path_1), 'archive.tar.gz')
self.assertEqual(archive_path_1, archive_path_2)
def test_same_dir_twice(self):
archive_dir = self.makedirs('archive')
runner = InlineMRJobRunner()
archive_path_1 = runner._dir_archive_path(archive_dir)
archive_path_2 = runner._dir_archive_path(archive_dir)
self.assertEqual(os.path.basename(archive_path_1), 'archive.tar.gz')
self.assertEqual(archive_path_1, archive_path_2)
def test_dirs_with_same_name(self):
foo_archive = self.makedirs(os.path.join('foo', 'archive'))
bar_archive = self.makedirs(os.path.join('bar', 'archive'))
runner = InlineMRJobRunner()
foo_archive_path = runner._dir_archive_path(foo_archive)
bar_archive_path = runner._dir_archive_path(bar_archive)
self.assertEqual(os.path.basename(foo_archive_path), 'archive.tar.gz')
self.assertNotEqual(foo_archive_path, bar_archive_path)
def test_dirs_with_same_name(self):
foo_archive = self.makedirs(os.path.join('foo', 'archive'))
bar_archive = self.makedirs(os.path.join('bar', 'archive'))
runner = InlineMRJobRunner()
foo_archive_path = runner._dir_archive_path(foo_archive)
bar_archive_path = runner._dir_archive_path(bar_archive)
self.assertEqual(os.path.basename(foo_archive_path),
'archive.tar.gz')
self.assertNotEqual(foo_archive_path, bar_archive_path)
def test_empty_dir(self):
runner = InlineMRJobRunner()
empty_dir = self.makedirs('empty')
tar_gz_path = runner._dir_archive_path(empty_dir)
self.assertEqual(os.path.basename(tar_gz_path), 'empty.tar.gz')
runner._create_dir_archive(empty_dir)
with tarfile.open(tar_gz_path, 'r:gz') as tar_gz:
self.assertEqual(sorted(tar_gz.getnames()), [])
def test_empty_dir(self):
runner = InlineMRJobRunner()
empty_dir = self.makedirs('empty')
tar_gz_path = runner._dir_archive_path(empty_dir)
self.assertEqual(os.path.basename(tar_gz_path), 'empty.tar.gz')
runner._create_dir_archive(empty_dir)
with tarfile.open(tar_gz_path, 'r:gz') as tar_gz:
self.assertEqual(sorted(tar_gz.getnames()), [])
def test_only_create_archive_once(self):
runner = InlineMRJobRunner()
tar_gz_path = runner._dir_archive_path(self._to_archive)
runner._create_dir_archive(self._to_archive)
mtime_1 = os.stat(tar_gz_path).st_mtime
sleep(1)
runner._create_dir_archive(self._to_archive)
mtime_2 = os.stat(tar_gz_path).st_mtime
self.assertEqual(mtime_1, mtime_2)
def test_only_create_archive_once(self):
runner = InlineMRJobRunner()
tar_gz_path = runner._dir_archive_path(self._to_archive)
runner._create_dir_archive(self._to_archive)
mtime_1 = os.stat(tar_gz_path).st_mtime
sleep(1)
runner._create_dir_archive(self._to_archive)
mtime_2 = os.stat(tar_gz_path).st_mtime
self.assertEqual(mtime_1, mtime_2)
def test_archive(self):
runner = InlineMRJobRunner()
tar_gz_path = runner._dir_archive_path(self._to_archive)
self.assertEqual(os.path.basename(tar_gz_path), 'archive.tar.gz')
runner._create_dir_archive(self._to_archive)
tar_gz = tarfile.open(tar_gz_path, 'r:gz')
try:
self.assertEqual(sorted(tar_gz.getnames()),
[os.path.join('bar', 'baz'), 'foo'])
finally:
tar_gz.close()
def test_archive(self):
runner = InlineMRJobRunner()
tar_gz_path = runner._dir_archive_path(self._to_archive)
self.assertEqual(os.path.basename(tar_gz_path), 'archive.tar.gz')
runner._create_dir_archive(self._to_archive)
tar_gz = tarfile.open(tar_gz_path, 'r:gz')
try:
self.assertEqual(sorted(tar_gz.getnames()),
[os.path.join('bar', 'baz'), 'foo'])
finally:
tar_gz.close()
def test_uri(self):
# we don't check whether URIs exist or are directories
runner = InlineMRJobRunner()
archive_path = runner._dir_archive_path('s3://bucket/stuff')
self.assertEqual(os.path.basename(archive_path), 'stuff.tar.gz')
def test_uri(self):
# we don't check whether URIs exist or are directories
runner = InlineMRJobRunner()
archive_path = runner._dir_archive_path('s3://bucket/stuff')
self.assertEqual(os.path.basename(archive_path), 'stuff.tar.gz')