import arvados
import bz2
import copy
+import hashlib
import mock
import os
import pprint
+import re
import subprocess
import tempfile
import unittest
self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
self.check_manifest_file_sizes(
arvados.CollectionReader(
- cw.manifest_text()).normalize().manifest_text(),
+ cw.manifest_text()).manifest_text(normalize=True),
[0,1])
def check_manifest_file_sizes(self, manifest_text, expect_sizes):
. 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
"""
- self.assertEqual(arvados.CollectionReader(m1, self.api_client).normalize().manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
""". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
""")
m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
"""
- self.assertEqual(arvados.CollectionReader(m2, self.api_client).normalize().manifest_text(), m2)
+ self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
. 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
. 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
"""
- self.assertEqual(arvados.CollectionReader(m3, self.api_client).normalize().manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
""". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
""")
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
"""
- self.assertEqual(arvados.CollectionReader(m4, self.api_client).normalize().manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
"""./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
""")
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
"""
- self.assertEqual(arvados.CollectionReader(m5, self.api_client).normalize().manifest_text(),
+ self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
"""./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
""")
with self.data_file('1000G_ref_manifest') as f6:
m6 = f6.read()
- self.assertEqual(arvados.CollectionReader(m6, self.api_client).normalize().manifest_text(), m6)
+ self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
with self.data_file('jlake_manifest') as f7:
m7 = f7.read()
- self.assertEqual(arvados.CollectionReader(m7, self.api_client).normalize().manifest_text(), m7)
+ self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
"""
- self.assertEqual(arvados.CollectionReader(m8, self.api_client).normalize().manifest_text(), m8)
+ self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
def test_locators_and_ranges(self):
blocks2 = [['a', 10, 0],
. 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
"""
- m2 = arvados.CollectionReader(m1, self.api_client).normalize().manifest_text()
+ m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
self.assertEqual(m2,
". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt 43:41:md6sum.txt 84:43:md7sum.txt 6:37:md8sum.txt 84:43:md8sum.txt 83:1:md9sum.txt 0:43:md9sum.txt 84:36:md9sum.txt\n")
""". bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
+ def test_write_directory_tree_with_zero_recursion(self):
+ cwriter = arvados.CollectionWriter(self.api_client)
+ content = 'd1/d2/f3d1/f2f1'
+ blockhash = hashlib.md5(content).hexdigest() + '+' + str(len(content))
+ cwriter.write_directory_tree(
+ self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
+ max_manifest_depth=0)
+ self.assertEqual(
+ cwriter.manifest_text(),
+ ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
+
def test_write_one_file(self):
cwriter = arvados.CollectionWriter(self.api_client)
with self.make_test_file() as testfile:
DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
+ ALT_COLLECTION = API_COLLECTIONS['bar_file']
+ ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
+ ALT_MANIFEST = ALT_COLLECTION['manifest_text']
def _mock_api_call(self, mock_method, code, body):
mock_method = mock_method().execute
def test_locator_init(self):
client = self.api_client_mock(200)
# Ensure Keep will not return anything if asked.
- with tutil.mock_responses(None, 404):
+ with tutil.mock_get_responses(None, 404):
reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
api_client=client)
self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
# crunch-job needs this to read manifests that have only ever
# been written to Keep.
client = self.api_client_mock(200)
- with tutil.mock_responses(self.DEFAULT_MANIFEST, 404, 200):
+ self.mock_get_collection(client, 404, None)
+ with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
api_client=client)
self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
client = self.api_client_mock(404)
reader = arvados.CollectionReader(self.DEFAULT_UUID,
api_client=client)
- with tutil.mock_responses(self.DEFAULT_MANIFEST, 200):
+ with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
with self.assertRaises(arvados.errors.ApiError):
reader.manifest_text()
# To verify that CollectionReader tries Keep first here, we
# mock API server to return the wrong data.
client = self.api_client_mock(200)
- with tutil.mock_responses(self.DEFAULT_MANIFEST, 200):
+ with tutil.mock_get_responses(self.ALT_MANIFEST, 200):
self.assertEqual(
- self.DEFAULT_MANIFEST,
+ self.ALT_MANIFEST,
arvados.CollectionReader(
- self.DEFAULT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
+ self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
api_client=client).manifest_text())
def test_init_num_retries_propagated(self):
client = self.api_client_mock(200)
reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
num_retries=3)
- with tutil.mock_responses('foo', 500, 500, 200):
+ with tutil.mock_get_responses('foo', 500, 500, 200):
self.assertEqual('foo',
''.join(f.read(9) for f in reader.all_files()))
# client should be able to use CollectionReader on a manifest without normalizing it
client = self.api_client_mock(500)
nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
+ reader = arvados.CollectionReader(
+ nonnormal,
+ api_client=client, num_retries=0)
+ # Ensure stripped_manifest() doesn't mangle our manifest in
+ # any way other than stripping hints.
self.assertEqual(
- ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n",
- arvados.CollectionReader(
- nonnormal,
- api_client=client, num_retries=0).stripped_manifest())
+ re.sub('\+[^\d\s\+]+', '', nonnormal),
+ reader.stripped_manifest())
+ # Ensure stripped_manifest() didn't mutate our reader.
+ self.assertEqual(nonnormal, reader.manifest_text())
+ # Ensure the files appear in the order given in the manifest.
self.assertEqual(
[[6, '.', 'foo.txt'],
[0, '.', 'bar.txt']],
[[f.size(), f.stream_name(), f.name()]
- for f in
- arvados.CollectionReader(
- nonnormal,
- api_client=client, num_retries=0).all_streams()[0].all_files()])
+ for f in reader.all_streams()[0].all_files()])
@tutil.skip_sleep
class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
def mock_keep(self, body, *codes, **headers):
headers.setdefault('x-keep-replicas-stored', 2)
- return tutil.mock_responses(body, *codes, **headers)
+ return tutil.mock_put_responses(body, *codes, **headers)
def foo_writer(self, **kwargs):
api_client = self.api_client_mock()