def test_init_internal_data(self, tmpdir):
# A bucket with same path won't overwrite existing data...
bucket1 = Bucket(str(tmpdir))
assert bucket1._get_internal_data() == dict(version=1,
curr_src_num=0,
curr_repr_num={})
to_set = dict(version=1, curr_src_num=1, curr_repr_num={'1': 2})
bucket1._set_internal_data(to_set)
assert bucket1._get_internal_data() == to_set
bucket2 = Bucket(str(tmpdir))
assert bucket2._get_internal_data() == to_set
def test_get_stored_repr_num(self, tmpdir):
# we can get a representation number if the repective key is
# stored in the bucket already.
bucket = Bucket(str(tmpdir.join("cache")))
key_path1 = tmpdir / "cache" / "keys" / "1" / "1.key"
key_path2 = tmpdir / "cache" / "keys" / "1" / "2.key"
key_path3 = tmpdir / "cache" / "keys" / "2" / "1.key"
assert bucket.get_stored_repr_num(1, 'somekey') is None
assert bucket.get_stored_repr_num(1, 'otherkey') is None
assert bucket.get_stored_repr_num(2, 'somekey') is None
assert bucket.get_stored_repr_num(2, 'otherkey') is None
key_path1.write('otherkey', ensure=True)
assert bucket.get_stored_repr_num(1, 'somekey') is None
assert bucket.get_stored_repr_num(1, 'otherkey') == 1
assert bucket.get_stored_repr_num(2, 'somekey') is None
assert bucket.get_stored_repr_num(2, 'otherkey') is None
key_path2.write('somekey', ensure=True)
assert bucket.get_stored_repr_num(1, 'somekey') == 2
assert bucket.get_stored_repr_num(1, 'otherkey') == 1
assert bucket.get_stored_repr_num(2, 'somekey') is None
assert bucket.get_stored_repr_num(2, 'otherkey') is None
key_path3.write('somekey', ensure=True)
assert bucket.get_stored_repr_num(1, 'somekey') == 2
assert bucket.get_stored_repr_num(1, 'otherkey') == 1
assert bucket.get_stored_repr_num(2, 'somekey') == 1
assert bucket.get_stored_repr_num(2, 'otherkey') is None
def test_curr_repr_num(self, tmpdir):
# we can get/set current representation number
bucket = Bucket(str(tmpdir))
assert bucket.get_current_repr_num(1) == 0
assert bucket.get_current_repr_num('2') == 0
bucket.set_current_repr_num('1', 12)
assert bucket.get_current_repr_num('1') == 12
assert bucket.get_current_repr_num('2') == 0
def test_init_sets_attributes(self, tmpdir):
# Main attributes are set properly...
bucket = Bucket(str(tmpdir))
assert bucket.srcdir == tmpdir / "sources"
assert bucket.resultdir == tmpdir / "repr"
assert bucket.keysdir == tmpdir / "keys"
assert bucket._data == dict(version=1,
curr_src_num=0,
curr_repr_num=dict())
def test_get_representation_stored(self, cache_env):
# we can get paths of representations
bucket = Bucket(str(cache_env.join("cache")))
res1 = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result1.txt"),
repr_key=b'mykey')
res2 = bucket.get_representation(res1)
assert res1 == "1_1"
assert res2 == cache_env / "cache" / "repr" / "1" / "1" / "result1.txt"
def test_store_representation_string_key(self, cache_env):
# we can store sources with their representations and a string key
bucket = Bucket(str(cache_env.join("cache")))
res = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result1.txt"),
repr_key="somekey")
assert res == "1_1"
assert (cache_env / "cache" / "keys" / "1" /
"1.key").read() == 'somekey'
def test_get_stored_source_num(self, cache_env):
# we can test whether a source file is stored in a bucket already.
bucket = Bucket(str(cache_env.join("cache")))
src1 = cache_env / "src1.txt"
src2 = cache_env / "src2.txt"
assert bucket.get_stored_source_num(str(src1)) is None
assert bucket.get_stored_source_num(str(src2)) is None
shutil.copyfile(str(src1), os.path.join(bucket.srcdir, "source_1"))
assert bucket.get_stored_source_num(str(src1)) == 1
assert bucket.get_stored_source_num(str(src2)) is None
shutil.copyfile(str(src2), os.path.join(bucket.srcdir, "source_2"))
assert bucket.get_stored_source_num(str(src1)) == 1
assert bucket.get_stored_source_num(str(src2)) == 2
def test_store_representation_no_key(self, cache_env):
# we can store sources with their representations
bucket = Bucket(str(cache_env.join("cache")))
res = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result1.txt"))
source_path = cache_env / "cache" / "sources" / "source_1"
result_path = cache_env / "cache" / "repr" / "1" / "1" / "result1.txt"
assert res == "1_1"
assert source_path.isfile()
assert source_path.read() == "source1\n"
assert result_path.dirpath().isdir()
assert result_path.isfile()
assert result_path.read() == "result1\n"
assert (cache_env / "cache" / "keys" / "1" / "1.key").isfile()
assert (cache_env / "cache" / "keys" / "1" / "1.key").read() == ""
def test_store_representation_update_result(self, cache_env):
# if we send a different representation for the same source
# and key, the old representation will be replaced.
bucket = Bucket(str(cache_env / "cache"))
res1 = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result1.txt"),
repr_key='mykey')
res2 = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result2.txt"),
repr_key='mykey')
assert res1 == "1_1"
assert res2 == "1_1"
result_dir = cache_env / "cache" / "repr" / "1" / "1"
assert result_dir.join("result1.txt").exists() is False
assert result_dir.join("result2.txt").exists() is True
assert result_dir.join("result2.txt").read() == "result2\n"
def test_keys(self, cache_env):
# we can get a list of all bucket keys in a bucket.
bucket = Bucket(str(cache_env))
assert list(bucket.keys()) == []
key1 = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result1.txt"),
repr_key='foo')
assert list(bucket.keys()) == [
key1,
]
key2 = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result2.txt"),
repr_key='bar')
assert sorted(list(bucket.keys())) == [key1, key2]
key3 = bucket.store_representation(str(cache_env / "src1.txt"),
str(cache_env / "result3.txt"),
repr_key='baz')
assert sorted(list(bucket.keys())) == [key1, key2, key3]
def test_curr_src_num(self, tmpdir):
# we can get/set current source number
bucket = Bucket(str(tmpdir))
assert bucket.get_current_source_num() == 0
bucket.set_current_source_num(12)
assert bucket.get_current_source_num() == 12
def test_init_creates_subdirs(self, tmpdir):
# a new bucket contains certain subdirs and a file
Bucket(str(tmpdir))
for filename in ['sources', 'repr', 'keys', 'data']:
assert tmpdir.join(filename).exists()
def test_get_representation_unstored(self, tmpdir):
# we cannot get unstored representations
bucket = Bucket(str(tmpdir.join("cache")))
assert bucket.get_representation("1_1") is None