def test_check(self):
cfg = {
"max_bag_size": 1000000000,
"max_headbag_size": 5000000,
"target_bag_size": 800000000
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
self.assertFalse( self.spltr.check() )
cfg = {
"max_bag_size": 10000,
"max_headbag_size": 5000000
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
self.assertTrue( self.spltr.check() )
cfg = {
"max_bag_size": 10000
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
self.assertTrue( self.spltr.check() )
cfg = { }
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
self.assertFalse( self.spltr.check() )
def test_verify_complete(self):
cfg = {
"max_bag_size": 400000,
"max_headbag_size": 50000,
"verify_complete": False
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
bags = self.spltr.split(self.workdir)
self.spltr._verify_complete(self.bagdir, bags)
with open(os.path.join(bags[0],"manifest-sha256.txt")) as fd:
datafile = fd.readline().strip().split()[-1]
os.remove(os.path.join(bags[0], datafile))
try:
self.spltr._verify_complete(self.bagdir, bags)
self.fail("Failed to raise Validation exception")
except Exception as ex:
self.assertEqual(len(ex.errors), 1)
datafile = "meta"+datafile
os.remove(os.path.join(bags[-1], datafile))
try:
errors = self.spltr._verify_complete(self.bagdir, bags)
self.fail("Failed to raise Validation exception")
except Exception as ex:
self.assertEqual(len(ex.errors), 2)
def test_check_and_split(self):
cfg = {
"max_bag_size": 400000,
"max_headbag_size": 50000
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
bags = self.spltr.check_and_split(self.workdir)
self.assertEqual(len(bags), 4)
self.assertTrue(os.path.exists(os.path.join(bags[-1], "multibag")))
def test_check_and_split_too_small(self):
cfg = {
"max_bag_size": 400000000,
"max_headbag_size": 50000000,
"verify_complete": False
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
bags = self.spltr.check_and_split(self.workdir)
self.assertEqual(len(bags), 1)
self.assertFalse(os.path.exists(os.path.join(bags[0], "multibag")))
def test_split_replace(self):
cfg = {
"max_bag_size": 400000,
"max_headbag_size": 50000,
"replace": True,
"verify_complete": False
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
bags = self.spltr.split(self.workdir)
self.assertEqual([os.path.basename(b) for b in bags],
["dataset-0", "dataset-1", "dataset-2", "dataset-3"])
self.assertTrue(os.path.isdir(os.path.join(bags[-1], "multibag")))
def test_confirm_found(self):
cfg = {
"max_bag_size": 400000,
"max_headbag_size": 50000,
"verify_complete": False
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
bags = self.spltr.split(self.workdir)
headbag = multibag.multibag.open_headbag(bags[-1])
error = self.spltr._confirm_found("data/goober", bags, headbag)
self.assertIsNotNone(error)
self.assertIn("Failed to find", error)
def test_ctor(self):
self.spltr = multibag.MultibagSplitter(self.bagdir)
self.assertEqual(self.spltr.maxsz, 0)
self.assertEqual(self.spltr.maxhbsz, 0)
self.assertEqual(self.spltr.trgsz, 0)
cfg = {
"max_bag_size": 1000000000,
"max_headbag_size": 5000000,
"target_bag_size": 800000000
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
self.assertEqual(self.spltr.maxsz, 1000000000)
self.assertEqual(self.spltr.maxhbsz, 5000000)
self.assertEqual(self.spltr.trgsz, 800000000)
cfg = {
"max_bag_size": 10000000
}
self.spltr = multibag.MultibagSplitter(self.bagdir, cfg)
self.assertEqual(self.spltr.maxsz, 10000000)
self.assertEqual(self.spltr.maxhbsz, 10000000)
self.assertEqual(self.spltr.trgsz, 10000000)