+ with self.assertRaises(arvados.errors.ArgumentError):
+ self.assertEqual(m1, CollectionReader(m1))
+
+ def test_init_manifest_with_error(self):
+ m1 = """. 0:43:md5sum.txt"""
+ with self.assertRaises(arvados.errors.ArgumentError):
+ self.assertEqual(m1, CollectionReader(m1))
+
+ def test_remove(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
+ self.assertIn("count1.txt", c)
+ c.remove("count1.txt")
+ self.assertNotIn("count1.txt", c)
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
+ with self.assertRaises(arvados.errors.ArgumentError):
+ c.remove("")
+
+ def test_find(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
+ self.assertIs(c.find("."), c)
+ self.assertIs(c.find("./count1.txt"), c["count1.txt"])
+ self.assertIs(c.find("count1.txt"), c["count1.txt"])
+ with self.assertRaises(IOError):
+ c.find("/.")
+ with self.assertRaises(arvados.errors.ArgumentError):
+ c.find("")
+ self.assertIs(c.find("./nonexistant.txt"), None)
+ self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
+
+ def test_remove_in_subdir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+ c.remove("foo/count2.txt")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
+
+ def test_remove_empty_subdir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+ c.remove("foo/count2.txt")
+ c.remove("foo")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
+
+ def test_remove_nonempty_subdir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+ with self.assertRaises(IOError):
+ c.remove("foo")
+ c.remove("foo", recursive=True)
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
+
+ def test_copy_to_file_in_dir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c.copy("count1.txt", "foo/count2.txt")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
+
+ def test_copy_file(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c.copy("count1.txt", "count2.txt")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
+
+ def test_copy_to_existing_dir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+ c.copy("count1.txt", "foo")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
+
+ def test_copy_to_new_dir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c.copy("count1.txt", "foo/")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
+
+ def test_rename_file(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c.rename("count1.txt", "count2.txt")
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
+
+ def test_move_file_to_dir(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c.mkdirs("foo")
+ c.rename("count1.txt", "foo/count2.txt")
+ self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
+
+ def test_move_file_to_other(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection()
+ c2.rename("count1.txt", "count2.txt", source_collection=c1)
+ self.assertEqual("", c1.manifest_text())
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
+
+ def test_clone(self):
+ c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+ cl = c.clone()
+ self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
+
+ def test_diff_del_add(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+ d = c2.diff(c1)
+ self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
+ ('add', './count1.txt', c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
+ ('add', './count2.txt', c2["count2.txt"])])
+ self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_diff_same(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ d = c2.diff(c1)
+ self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_diff_mod(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
+ d = c2.diff(c1)
+ self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
+
+ self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_diff_add(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
+ d = c2.diff(c1)
+ self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+
+ self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_diff_add_in_subcollection(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+ d = c2.diff(c1)
+ self.assertEqual(d, [('del', './foo', c2["foo"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('add', './foo', c2["foo"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+
+ self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_diff_del_add_in_subcollection(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+ c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
+
+ d = c2.diff(c1)
+ self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
+ ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
+ ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+
+ self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_diff_mod_in_subcollection(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+ c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
+ d = c2.diff(c1)
+ self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+ d = c1.diff(c2)
+ self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"]),
+ ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
+
+ self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
+
+ def test_conflict_keep_local_change(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
+ c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
+ d = c1.diff(c2)
+ self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
+ ('add', './count2.txt', c2["count2.txt"])])
+ f = c1.open("count1.txt", "w")
+ f.write("zzzzz")
+
+ # c1 changed, so it should not be deleted.
+ c1.apply(d)
+ self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
+
+ def test_conflict_mod(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
+ c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
+ d = c1.diff(c2)
+ self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
+ f = c1.open("count1.txt", "w")
+ f.write("zzzzz")
+
+ # c1 changed, so c2 mod will go to a conflict file
+ c1.apply(d)
+ self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
+
+ def test_conflict_add(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
+ c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
+ d = c1.diff(c2)
+ self.assertEqual(d, [('del', './count2.txt', c1["count2.txt"]),
+ ('add', './count1.txt', c2["count1.txt"])])
+ f = c1.open("count1.txt", "w")
+ f.write("zzzzz")
+
+ # c1 added count1.txt, so c2 add will go to a conflict file
+ c1.apply(d)
+ self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
+
+ def test_conflict_del(self):
+ c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
+ c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
+ d = c1.diff(c2)
+ self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
+ c1.remove("count1.txt")
+
+ # c1 deleted, so c2 mod will go to a conflict file
+ c1.apply(d)
+ self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
+
+ def test_notify(self):
+ c1 = Collection()
+ events = []
+ c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
+ f = c1.open("foo.txt", "w")
+ self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
+
+ def test_open_w(self):
+ c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
+ self.assertEqual(c1["count1.txt"].size(), 10)
+ c1.open("count1.txt", "w").close()
+ self.assertEqual(c1["count1.txt"].size(), 0)
+
+
+class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
+ def test_get_manifest_text_only_committed(self):
+ c = Collection()
+ with c.open("count.txt", "w") as f:
+ # One file committed
+ with c.open("foo.txt", "w") as foo:
+ foo.write("foo")
+ foo.flush() # Force block commit
+ f.write("0123456789")
+ # Other file not committed. Block not written to keep yet.
+ self.assertEqual(
+ c._get_manifest_text(".",
+ strip=False,
+ normalize=False,
+ only_committed=True),
+ '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
+ # And now with the file closed...
+ f.flush() # Force block commit
+ self.assertEqual(
+ c._get_manifest_text(".",
+ strip=False,
+ normalize=False,
+ only_committed=True),
+ ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
+
+ def test_only_small_blocks_are_packed_together(self):
+ c = Collection()
+ # Write a couple of small files,
+ f = c.open("count.txt", "w")
+ f.write("0123456789")
+ f.close(flush=False)
+ foo = c.open("foo.txt", "w")
+ foo.write("foo")
+ foo.close(flush=False)
+ # Then, write a big file, it shouldn't be packed with the ones above
+ big = c.open("bigfile.txt", "w")
+ big.write("x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
+ big.close(flush=False)
+ self.assertEqual(
+ c.manifest_text("."),
+ '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
+
+
+class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
+ MAIN_SERVER = {}
+ KEEP_SERVER = {}
+
+ def create_count_txt(self):
+ # Create an empty collection, save it to the API server, then write a
+ # file, but don't save it.
+
+ c = Collection()
+ c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
+ self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
+ self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
+
+ with c.open("count.txt", "w") as f:
+ f.write("0123456789")
+
+ self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
+
+ return c
+
+ def test_create_and_save(self):
+ c = self.create_count_txt()
+ c.save()
+ self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
+
+ def test_create_and_save_new(self):
+ c = self.create_count_txt()
+ c.save_new()
+ self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
+
+ def test_create_diff_apply(self):
+ c1 = self.create_count_txt()
+ c1.save()
+
+ c2 = Collection(c1.manifest_locator())
+ with c2.open("count.txt", "w") as f:
+ f.write("abcdefg")
+
+ diff = c1.diff(c2)
+
+ self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
+
+ c1.apply(diff)
+ self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
+
+ def test_diff_apply_with_token(self):
+ baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
+ c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
+ other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
+
+ diff = baseline.diff(other)
+ self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
+
+ c.apply(diff)
+
+ self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
+
+
+ def test_create_and_update(self):
+ c1 = self.create_count_txt()
+ c1.save()
+
+ c2 = arvados.collection.Collection(c1.manifest_locator())
+ with c2.open("count.txt", "w") as f:
+ f.write("abcdefg")
+
+ c2.save()
+
+ self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
+ c1.update()
+ self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
+
+
+ def test_create_and_update_with_conflict(self):
+ c1 = self.create_count_txt()
+ c1.save()
+
+ with c1.open("count.txt", "w") as f:
+ f.write("XYZ")
+
+ c2 = arvados.collection.Collection(c1.manifest_locator())
+ with c2.open("count.txt", "w") as f:
+ f.write("abcdefg")
+
+ c2.save()
+
+ c1.update()
+ self.assertRegexpMatches(c1.manifest_text(), r"\. e65075d550f9b5bf9992fa1d71a131be\+3\S* 7ac66c0f148de9519b8bd264312c4d64\+7\S* 0:3:count\.txt 3:7:count\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
+