def test_add_absolute_archive(self):
# Given
archive_extractor = ArchiveExtractor(oozie_node=Element("fake"), props=self.props)
# When
archive_extractor.add_archive("/test_archive.zip")
# Then
self.assertEqual(["/test_archive.zip"], archive_extractor.archives)
self.assertEqual(["hdfs:///test_archive.zip"], archive_extractor.hdfs_archives)
def test_add_absolute_archive(self):
# Given
archive_extractor = ArchiveExtractor(oozie_node=Element("fake"),
params=self.default_params)
# When
archive_extractor.add_archive("/test_archive.zip")
# Then
self.assertEqual(archive_extractor.archives, ["/test_archive.zip"])
self.assertEqual(archive_extractor.hdfs_archives,
["hdfs:///test_archive.zip"])
def test_add_archive_extra_hash(self):
# Given
archive_extractor = ArchiveExtractor(oozie_node=Element("fake"), props=self.props)
# When
with self.assertRaises(Exception) as context:
archive_extractor.add_archive("/test_archive.zip#4rarear#")
# Then
self.assertEqual(
"There should be maximum one '#' in the path /test_archive.zip#4rarear#", str(context.exception)
)
def test_add_relative_archive(self):
# Given
archive_extractor = ArchiveExtractor(oozie_node=Element("fake"), props=self.props)
# When
archive_extractor.add_archive("test_archive.zip")
# Then
self.assertEqual(["test_archive.zip"], archive_extractor.archives)
self.assertEqual(
["hdfs:///user/pig/examples/pig_test_node/test_archive.zip"], archive_extractor.hdfs_archives
)
def test_add_multiple_archives(self):
# Given
archive_extractor = ArchiveExtractor(oozie_node=Element("fake"), props=self.props)
# When
archive_extractor.add_archive("/test_archive.zip")
archive_extractor.add_archive("test_archive2.tar")
archive_extractor.add_archive("/test_archive3.tar.gz")
# Then
self.assertEqual(
archive_extractor.archives, ["/test_archive.zip", "test_archive2.tar", "/test_archive3.tar.gz"]
)
self.assertEqual(
archive_extractor.hdfs_archives,
[
"hdfs:///test_archive.zip",
"hdfs:///user/pig/examples/pig_test_node/test_archive2.tar",
"hdfs:///test_archive3.tar.gz",
],
)
def test_add_hash_archives(self):
# Given
archive_extractor = ArchiveExtractor(oozie_node=Element("fake"),
params=self.default_params)
# When
archive_extractor.add_archive("/test_archive.zip#test3_link")
archive_extractor.add_archive("test_archive2.tar#test_link")
archive_extractor.add_archive("/test_archive3.tar.gz")
# Then
self.assertEqual(
archive_extractor.archives,
[
"/test_archive.zip#test3_link", "test_archive2.tar#test_link",
"/test_archive3.tar.gz"
],
)
self.assertEqual(
archive_extractor.hdfs_archives,
[
"hdfs:///test_archive.zip#test3_link",
"hdfs:///user/pig/examples/pig_test_node/test_archive2.tar#test_link",
"hdfs:///test_archive3.tar.gz",
],
)