def test_vcf_data_file_identical():
with tempdir() as temp:
localizer1 = StringLocalizer("##fileformat=VCFv4.2\n"
"chr1\t1111\t.\tA\tG\t1000\tPASS\t.\tGT")
v1 = VcfDataFile(temp / "foo1.vcf", localizer1)
localizer2 = StringLocalizer("##fileformat=VCFv4.2\n"
"chr1\t1111\t.\tA\tG\t999.9\tPASS\t.")
v2 = VcfDataFile(temp / "foo2.vcf", localizer2)
v1.assert_contents_equal(v2)
def test_vcf_data_file_different():
with tempdir() as temp:
localizer1 = StringLocalizer("##fileformat=VCFv4.2\n"
"chr1\t1111\t.\tA\tG\t1000\tPASS\t.")
v1 = VcfDataFile(temp / "foo1.vcf", localizer1, allowed_diff_lines=0)
localizer2 = StringLocalizer("##fileformat=VCFv4.2\n"
"chr1\t1111\t.\tA\tG\t999.9\tPASS\t.\n"
"chr1\t2222\t.\tT\tC\t500\tPASS\t.")
v2 = VcfDataFile(temp / "foo2.vcf", localizer2)
with pytest.raises(AssertionError):
v1.assert_contents_equal(v2)
v1.set_compare_opts(allowed_diff_lines=1)
v1.assert_contents_equal(v2)
def create_data_file(user_config: UserConfiguration,
type: Optional[Union[str, dict]] = DEFAULT_TYPE,
name: Optional[str] = None,
path: Optional[Union[str, Path]] = None,
url: Optional[str] = None,
contents: Optional[Union[str, dict]] = None,
env: Optional[str] = None,
http_headers: Optional[dict] = None,
digests: Optional[dict] = None,
datadirs: Optional[DataDirs] = None,
**kwargs) -> DataFile:
if isinstance(type, dict):
data_file_opts = cast(dict, type)
type = data_file_opts.pop("name")
else:
data_file_opts = {}
data_file_opts.update(kwargs)
local_path = None
localizer = None
if path:
local_path = ensure_path(path, [user_config.cache_dir])
if local_path and local_path.exists():
pass
elif env and env in os.environ:
env_path = ensure_path(os.environ[env], exists=True)
if not local_path:
local_path = env_path
else:
localizer = LinkLocalizer(env_path)
elif url:
localizer = UrlLocalizer(url, user_config, http_headers, digests)
if not local_path:
if name:
local_path = ensure_path(user_config.cache_dir / name)
else:
filename = url.rsplit("/", 1)[1]
local_path = ensure_path(user_config.cache_dir / filename)
elif contents:
if isinstance(contents, str):
localizer = StringLocalizer(cast(str, contents))
else:
localizer = JsonLocalizer(cast(dict, contents))
if type == DEFAULT_TYPE:
type = "json"
if not local_path:
if name:
local_path = ensure_path(user_config.cache_dir / name)
else:
local_path = ensure_path(
tempfile.mktemp(dir=user_config.cache_dir))
elif name and datadirs:
for dd in datadirs.paths:
dd_path = dd / name
if dd_path.exists():
break
else:
raise FileNotFoundError(
f"File {name} not found in any of the following datadirs: "
f"{datadirs.paths}")
if not local_path:
local_path = dd_path
else:
localizer = LinkLocalizer(dd_path)
else:
raise FileNotFoundError(
f"File {path or name} does not exist. Either a url, file contents, "
f"or a local file must be provided.")
data_file_class = DATA_TYPES.get(type, DefaultDataFile)
return data_file_class(local_path, localizer, **data_file_opts)
def test_string_localizer():
with tempdir() as d:
foo = d / "foo"
StringLocalizer("foo").localize(foo)
with open(foo, "rt") as inp:
assert inp.read() == "foo"