Add some typings informations and typecheck

Mostly make it a bit clearer what is allowed/expected where.
In particular many of the decode functions allow already decoded
objects.

Add Typing to store.

Note in particular that some function can get bytes/str/anything that
can be stringified though others only can get str/None as they don't
call normalize on path.

Author: Carreau

.travis.yml

@@ -29,6 +29,7 @@ matrix:
     - python: 3.8
       dist: xenial
       sudo: true
+      env: MYPY_CHECK='true'
 
 before_install:
   - docker pull arafato/azurite
@@ -38,11 +39,12 @@ before_script:
     - mongo mydb_test --eval 'db.createUser({user:"travis",pwd:"test",roles:["readWrite"]});'
 
 install:
-  - pip install -U pip setuptools wheel tox-travis coveralls
+  - pip install -U pip setuptools wheel tox-travis coveralls mypy
 
 script:
   - tox
   - if [[ $BUILD_DOCS == 'true' ]]; then tox -e docs; fi
+  - if [[ $MYPY_CHECK == 'true' ]]; then mypy zarr; fi
 
 after_success:
   - coveralls --service=travis-pro

docs/release.rst

@@ -14,6 +14,8 @@ Next release
   without ``ipytree`` installed.
   By :user:`Zain Patel <mzjp2>`; :issue:`537`
 
+* Add typing informations to many of the core functions :issue:`589`
+
 * Explicitly close stores during testing.
   By :user:`Elliott Sales de Andrade <QuLogic>`; :issue:`442`
 

mypy.ini

@@ -0,0 +1,4 @@
+[mypy]
+python_version = 3.6
+ignore_missing_imports = True
+follow_imports = silent

pytest.ini

@@ -3,4 +3,4 @@ doctest_optionflags = NORMALIZE_WHITESPACE ELLIPSIS IGNORE_EXCEPTION_DETAIL
 addopts = --durations=10
 filterwarnings =
     error::DeprecationWarning:zarr.*
-
+    ignore:PY_SSIZE_T_CLEAN will be required.*:DeprecationWarning

requirements_dev_optional.txt

@@ -13,7 +13,7 @@ pymongo==3.9.0
 tox==3.14.0
 coverage==5.0.3
 coveralls==1.11.1
-flake8==3.7.8
+flake8==3.8.3
 pytest-cov==2.7.1
 pytest-doctestplus==0.4.0
 pytest-remotedata==0.3.2

zarr/codecs.py

@@ -1,4 +1,5 @@
 # -*- coding: utf-8 -*-
 # flake8: noqa
 from numcodecs import *
+from numcodecs import get_codec, Blosc, Zlib, Delta, AsType, BZ2
 from numcodecs.registry import codec_registry

zarr/indexing.py

@@ -468,7 +468,7 @@ def __iter__(self):
             yield ChunkDimProjection(dim_chunk_ix, dim_chunk_sel, dim_out_sel)
 
 
-def slice_to_range(s, l):
+def slice_to_range(s, l):  # noqa: E741
     return range(*s.indices(l))
 
 

zarr/meta.py

@@ -7,10 +7,12 @@
 from zarr.errors import MetadataError
 from zarr.util import json_dumps, json_loads
 
+from typing import Union, Any, List, Mapping as MappingType
+
 ZARR_FORMAT = 2
 
 
-def parse_metadata(s):
+def parse_metadata(s: Union[MappingType, str]) -> MappingType[str, Any]:
 
     # Here we allow that a store may return an already-parsed metadata object,
     # or a string of JSON that we will parse here. We allow for an already-parsed
@@ -28,7 +30,7 @@ def parse_metadata(s):
     return meta
 
 
-def decode_array_metadata(s):
+def decode_array_metadata(s: Union[MappingType, str]) -> MappingType[str, Any]:
     meta = parse_metadata(s)
 
     # check metadata format
@@ -56,7 +58,7 @@ def decode_array_metadata(s):
         return meta
 
 
-def encode_array_metadata(meta):
+def encode_array_metadata(meta: MappingType[str, Any]) -> bytes:
     dtype = meta['dtype']
     sdshape = ()
     if dtype.subdtype is not None:
@@ -74,27 +76,27 @@ def encode_array_metadata(meta):
     return json_dumps(meta)
 
 
-def encode_dtype(d):
+def encode_dtype(d: np.dtype) -> str:
     if d.fields is None:
         return d.str
     else:
         return d.descr
 
 
-def _decode_dtype_descr(d):
+def _decode_dtype_descr(d) -> List[Any]:
     # need to convert list of lists to list of tuples
     if isinstance(d, list):
         # recurse to handle nested structures
         d = [(k[0], _decode_dtype_descr(k[1])) + tuple(k[2:]) for k in d]
     return d
 
 
-def decode_dtype(d):
+def decode_dtype(d) -> np.dtype:
     d = _decode_dtype_descr(d)
     return np.dtype(d)
 
 
-def decode_group_metadata(s):
+def decode_group_metadata(s: Union[MappingType, str]) -> MappingType[str, Any]:
     meta = parse_metadata(s)
 
     # check metadata format version
@@ -108,7 +110,7 @@ def decode_group_metadata(s):
 
 # N.B., keep `meta` parameter as a placeholder for future
 # noinspection PyUnusedLocal
-def encode_group_metadata(meta=None):
+def encode_group_metadata(meta=None) -> bytes:
     meta = dict(
         zarr_format=ZARR_FORMAT,
     )
@@ -161,7 +163,7 @@ def decode_fill_value(v, dtype):
         return np.array(v, dtype=dtype)[()]
 
 
-def encode_fill_value(v, dtype):
+def encode_fill_value(v: Any, dtype: np.dtype) -> Any:
     # early out
     if v is None:
         return v