Merge pull request #59 from eX-Mech/pydantic-to-attrs

Pydantic to attrs

.github/environment.yml

@@ -3,4 +3,3 @@ channels:
 
 dependencies:
   - xarray
-  - pydantic

.github/environment_pypy.yml

@@ -2,6 +2,5 @@ channels:
   - conda-forge
 
 dependencies:
-  - python=3.7[build=*pypy]
+  - python=3.9[build=*pypy]
   - xarray
-  - pydantic

.github/workflows/tests.yml

@@ -112,3 +112,8 @@ jobs:
       shell: bash -l {0}
       run: |
         python -m pytest -s -v -n auto
+
+    - name: Run mypy
+      shell: bash -l {0}
+      run: |
+        python -m mypy ./pymech --show-error-codes

.pre-commit-config.yaml

@@ -7,10 +7,10 @@ repos:
     -   id: check-yaml
     -   id: check-added-large-files
 -   repo: https://github.com/psf/black
-    rev: 22.3.0
+    rev: 22.10.0
     hooks:
     -   id: black
--   repo: https://gitlab.com/pycqa/flake8
-    rev: 3.9.2  # pick a git hash / tag to point to
+-   repo: https://github.com/pycqa/flake8
+    rev: 5.0.4
     hooks:
     -   id: flake8

CHANGELOG.md

@@ -29,7 +29,7 @@ Security in case of vulnerabilities.
 
 ### Added
 
-- Dataclass {class}`pymech.neksuite.field.Header` using [`pydantic`](https://pydantic-docs.helpmanual.io/) for reading file headers
+- Dataclass {class}`pymech.neksuite.field.Header` using [`attrs`](https://www.attrs.org/en/stable/) for reading file headers
 - Function {func}`pymech.neksuite.field.read_header`
 - Module {mod}`pymech.meshtools` for manipulating meshes
 - Xarray backend ``pymech``. See {ref}`dataset`

docs/asv_bench/results/benchmarks.json

@@ -6,8 +6,8 @@
         "number": 0,
         "param_names": [],
         "params": [],
-        "processes": 2,
         "repeat": 0,
+        "rounds": 2,
         "sample_time": 0.01,
         "timeout": 60.0,
         "type": "time",
@@ -22,8 +22,8 @@
         "number": 0,
         "param_names": [],
         "params": [],
-        "processes": 2,
         "repeat": 0,
+        "rounds": 2,
         "sample_time": 0.01,
         "timeout": 60.0,
         "type": "time",

docs/asv_bench/results/tuxpulse/2cb0b50e-virtualenv-py3.8-dask[array]-numpy-pandas.json

@@ -0,0 +1 @@
+{"commit_hash": "2cb0b50e87c98df6df575f61d9b244abc6886d62", "env_name": "virtualenv-py3.8-dask[array]-numpy-pandas", "date": 1655662291000, "params": {"arch": "x86_64", "cpu": "AMD Ryzen 7 4800H with Radeon Graphics", "machine": "tuxpulse", "num_cpu": "16", "os": "Linux 5.15.0-10033-tuxedo", "ram": "16GB", "python": "3.8", "numpy": "", "dask[array]": "", "pandas": ""}, "python": "3.8", "requirements": {"numpy": "", "dask[array]": "", "pandas": ""}, "env_vars": {}, "result_columns": ["result", "params", "version", "started_at", "duration", "stats_ci_99_a", "stats_ci_99_b", "stats_q_25", "stats_q_75", "stats_number", "stats_repeat", "samples", "profile"], "results": {"bench_open_dataset.OpenDatasetSuite.time_read": [[0.6610401119996823], [], "d8736c9050c8133f745cc20f6d35fcbac47a834c114c581fff6c8b4123f99d6a", 1655665708407, 8.1294, [0.65864], [0.6679], [0.66033], [0.66239], [1], [10]], "bench_readnek.ReadnekSuite.time_read": [[0.0306186504985817], [], "4247c82d15af70d066d71b05656858fa8f9088d8df24e2be77eb30b987a05a4b", 1655665712473, 0.61015, [0.030559], [0.031918], [0.03059], [0.030657], [1], [10]]}, "durations": {}, "version": 2}

docs/asv_bench/results/tuxpulse/2cb0b50e-virtualenv-py3.8-numpy-pandas.json

@@ -0,0 +1 @@
+{"commit_hash": "2cb0b50e87c98df6df575f61d9b244abc6886d62", "env_name": "virtualenv-py3.8-numpy-pandas", "date": 1655662291000, "params": {"arch": "x86_64", "cpu": "AMD Ryzen 7 4800H with Radeon Graphics", "machine": "tuxpulse", "num_cpu": "16", "os": "Linux 5.15.0-10033-tuxedo", "ram": "16GB", "python": "3.8", "numpy": "", "pandas": ""}, "python": "3.8", "requirements": {"numpy": "", "pandas": ""}, "env_vars": {}, "result_columns": ["result", "params", "version", "started_at", "duration", "stats_ci_99_a", "stats_ci_99_b", "stats_q_25", "stats_q_75", "stats_number", "stats_repeat", "samples", "profile"], "results": {"bench_open_dataset.OpenDatasetSuite.time_read": [[0.6675610369984497], [], "d8736c9050c8133f745cc20f6d35fcbac47a834c114c581fff6c8b4123f99d6a", 1655665724254, 8.1799, [0.66404], [0.67107], [0.66708], [0.66929], [1], [10]], "bench_readnek.ReadnekSuite.time_read": [[0.031290766501115286], [], "4247c82d15af70d066d71b05656858fa8f9088d8df24e2be77eb30b987a05a4b", 1655665728370, 0.60922, [0.031064], [0.036323], [0.031117], [0.031526], [1], [10]]}, "durations": {"<build>": 6.579625368118286}, "version": 2}

docs/asv_bench/results/tuxpulse/53a66ce5-virtualenv-py3.8-dask[array]-numpy-pandas.json

@@ -0,0 +1 @@
+{"commit_hash": "53a66ce515a41d5667d9452082c4da17f6791442", "env_name": "virtualenv-py3.8-dask[array]-numpy-pandas", "date": 1655624617000, "params": {"machine": "tuxpulse", "os": "Linux 5.15.0-10033-tuxedo", "arch": "x86_64", "cpu": "AMD Ryzen 7 4800H with Radeon Graphics", "num_cpu": "16", "ram": "16GB", "python": "3.8", "numpy": "", "dask[array]": "", "pandas": ""}, "python": "3.8", "requirements": {"numpy": "", "dask[array]": "", "pandas": ""}, "env_vars": {}, "result_columns": ["result", "params", "version", "started_at", "duration", "stats_ci_99_a", "stats_ci_99_b", "stats_q_25", "stats_q_75", "stats_number", "stats_repeat", "samples", "profile"], "results": {"bench_open_dataset.OpenDatasetSuite.time_read": [[0.7063027359981788], [], "d8736c9050c8133f745cc20f6d35fcbac47a834c114c581fff6c8b4123f99d6a", 1655662111786, 8.6814, [0.68268], [0.79311], [0.69674], [0.72925], [1], [10]], "bench_readnek.ReadnekSuite.time_read": [[0.037250710500302375], [], "4247c82d15af70d066d71b05656858fa8f9088d8df24e2be77eb30b987a05a4b", 1655662116052, 0.71059, [0.032594], [0.039794], [0.034223], [0.039192], [1], [10]]}, "durations": {}, "version": 2}

docs/asv_bench/results/tuxpulse/53a66ce5-virtualenv-py3.8-numpy-pandas.json

@@ -0,0 +1 @@
+{"commit_hash": "53a66ce515a41d5667d9452082c4da17f6791442", "env_name": "virtualenv-py3.8-numpy-pandas", "date": 1655624617000, "params": {"machine": "tuxpulse", "os": "Linux 5.15.0-10033-tuxedo", "arch": "x86_64", "cpu": "AMD Ryzen 7 4800H with Radeon Graphics", "num_cpu": "16", "ram": "16GB", "python": "3.8", "numpy": "", "pandas": ""}, "python": "3.8", "requirements": {"numpy": "", "pandas": ""}, "env_vars": {}, "result_columns": ["result", "params", "version", "started_at", "duration", "stats_ci_99_a", "stats_ci_99_b", "stats_q_25", "stats_q_75", "stats_number", "stats_repeat", "samples", "profile"], "results": {"bench_open_dataset.OpenDatasetSuite.time_read": [[0.6838599720013008], [], "d8736c9050c8133f745cc20f6d35fcbac47a834c114c581fff6c8b4123f99d6a", 1655662134561, 8.4276, [0.66956], [0.70753], [0.68118], [0.69425], [1], [10]], "bench_readnek.ReadnekSuite.time_read": [[0.031951281998772174], [], "4247c82d15af70d066d71b05656858fa8f9088d8df24e2be77eb30b987a05a4b", 1655662138775, 0.66082, [0.031676], [0.032222], [0.031877], [0.032062], [1], [10]]}, "durations": {"<build>": 13.125392198562622}, "version": 2}

docs/asv_bench/results/tuxpulse/machine.json

@@ -0,0 +1,9 @@
+{
+    "arch": "x86_64",
+    "cpu": "AMD Ryzen 7 4800H with Radeon Graphics",
+    "machine": "tuxpulse",
+    "num_cpu": "16",
+    "os": "Linux 5.15.0-10033-tuxedo",
+    "ram": "16GB",
+    "version": 1
+}

docs/index.rst

@@ -61,7 +61,7 @@ interpreter of your choice:
 
    ::
 
-      conda create -n my-env -c conda-forge pypy pip xarray pydantic
+      conda create -n my-env -c conda-forge pypy pip xarray
       conda activate my-env
       pypy -m pip install pymech
 

docs/neksuite.rst

@@ -20,6 +20,12 @@ readnek() is clever enough to figure out the rest.
 The output is a single :py:class:`pymech.core.HexaData` data structure containing all
 the information that was stored in the file.
 
+.. note::
+
+    If you are only interested in the metadata (for example, the simulation
+    time) of a solution file a quicker solution would be to execute
+    :py:func:`pymech.neksuite.field.read_header`
+
 writenek
 --------
 This function writes a binary ``.f%05d`` file in the same format that *Nek5000*

pymech/neksuite/field.py

@@ -1,57 +1,77 @@
 from __future__ import annotations
 
+import os
 import io
 import struct
 import sys
-from typing import Optional, Tuple
+from pathlib import Path
+from typing import Optional, Tuple, Union, BinaryIO
 
 import numpy as np
-from pydantic.dataclasses import ValidationError, dataclass
+
+from attrs import define, field
+
 from pymech.core import HexaData
 from pymech.log import logger
 
 
-@dataclass
+def _as_unicode(string: StringOrBytes) -> str:
+    if isinstance(string, bytes):
+        return string.decode()
+    else:
+        return string
+
+
+def _as_tuple_of_ints(seq):
+    return tuple(int(s) for s in seq)
+
+
+StringOrBytes = Union[str, bytes]
+PathLike = Union[str, os.PathLike]
+
+
+@define
 class Header:
     """Dataclass for Nek5000 field file header. This relies on the package
-    pydantic_ and its ability to do type-checking and type-coercion of the
+    attrs_ and its ability to do type-validation and type-conversion of the
     header metadata.
 
-    .. _pydantic: https://pydantic-docs.helpmanual.io/
+    .. _attrs: https://www.attrs.org/en/stable/
 
     """
 
     # get word size: single or double precision
-    wdsz: int
+    wdsz: int = field(converter=int)
     # get polynomial order
-    orders: Tuple[int, ...]
+    orders: Tuple[int, ...] = field(converter=_as_tuple_of_ints)
     # get number of elements
-    nb_elems: int
+    nb_elems: int = field(converter=int)
     # get number of elements in the file
-    nb_elems_file: int
+    nb_elems_file: int = field(converter=int)
     # get current time
-    time: float
+    time: float = field(converter=float)
     # get current time step
-    istep: int
+    istep: int = field(converter=int)
     # get file id
-    fid: int
+    fid: int = field(converter=int)
     # get tot number of files
-    nb_files: int
+    nb_files: int = field(converter=int)
 
-    # get variables [XUPTS[01-99]]
-    variables: Optional[str] = None
+    # NOTE: field(factory=...) specifies the default value for the field.
+    # https://www.attrs.org/en/stable/init.html#defaults
 
+    # get variables [XUPTS[01-99]]
+    variables: str = field(converter=_as_unicode, factory=str)
     # floating point precision
-    realtype: Optional[str] = None
-
+    realtype: str = field(factory=str)
     # compute total number of points per element
-    nb_pts_elem: Optional[int] = None
+    nb_pts_elem: int = field(factory=int)
     # get number of physical dimensions
-    nb_dims: Optional[int] = None
+    nb_dims: int = field(factory=int)
     # get number of variables
-    nb_vars: Optional[Tuple[int, ...]] = None
+    nb_vars: Tuple[int, ...] = field(factory=tuple)
 
-    def __post_init_post_parse__(self):
+    def __attrs_post_init__(self):
         # get word size: single or double precision
         wdsz = self.wdsz
         if not self.realtype:
@@ -70,7 +90,7 @@ def __post_init_post_parse__(self):
             self.nb_dims = 2 + int(orders[2] > 1)
 
         if not self.variables and not self.nb_vars:
-            raise ValidationError("Both variables and nb_vars cannot be None", self)
+            raise ValueError("Both variables and nb_vars cannot be uninitialized.")
         elif self.variables:
             self.nb_vars = self._variables_to_nb_vars()
         elif self.nb_vars:
@@ -132,17 +152,26 @@ def as_bytestring(self) -> bytes:
         return header.ljust(132).encode("utf-8")
 
 
-def read_header(fp: io.BufferedReader) -> Header:
+def read_header(path_or_file_obj: Union[PathLike, BinaryIO]) -> Header:
     """Make a :class:`pymech.neksuite.Header` instance from a file buffer
     opened in binary mode.
 
     """
-    header = fp.read(132).split()
+    if isinstance(path_or_file_obj, (str, os.PathLike)):
+        with Path(path_or_file_obj).open("rb") as fp:
+            header = fp.read(132).split()
+    elif isinstance(path_or_file_obj, io.BufferedReader):
+        fp = path_or_file_obj
+        header = fp.read(132).split()
+    else:
+        raise ValueError("Should be a path or opened file object in 'rb' mode.")
+
     logger.debug(b"Header: " + b" ".join(header))
     if len(header) < 12:
         raise IOError("Header of the file was too short.")
 
-    return Header(header[1], header[2:5], *header[5:12])
+    # Relying on attrs converter to type-cast. Mypy will complain
+    return Header(header[1], header[2:5], *header[5:12])  # type: ignore[arg-type]
 
 
 # ==============================================================================

setup.cfg

@@ -29,7 +29,8 @@ python_requires = >= 3.7
 packages = find:
 install_requires =
     xarray >= 0.19.0, != 0.20.0, != 0.20.1
-    pydantic >= 1.8.1
+    importlib_metadata < 5.0.0 ; python_version < "3.8"
+    attrs >= 21.4.0
 
 setup_requires =
     setuptools>=39.2.0
@@ -47,11 +48,11 @@ docs =
     sphinx-copybutton
     sphinx-inline-tabs
     myst-nb
-    importlib_metadata; python_version < "3.8"
     asv >= 0.5.1
 
 tests =
     %(full)s
+    mypy
     pytest >= 6.2.5
     pytest-cov >= 3.0.0
     pytest-xdist >= 2.4.0
@@ -69,7 +70,7 @@ xarray.backends =
 
 [flake8]
 max-line-length = 88
-ignore = W503, E203, E501
+extend-ignore = W503, E203, E501
 
 [coverage:run]
 source = pymech
@@ -97,4 +98,4 @@ ignore_errors = True
 addopts = "--cov=pymech"
 
 [mypy]
-plugins = pydantic.mypy
+ignore_missing_imports = True

tests/test_header.py

@@ -0,0 +1,14 @@
+import pytest
+
+from pymech.neksuite.field import Header, read_header
+
+
+def test_header_var_nb_vars_uninitialized():
+    with pytest.raises(ValueError):
+        Header(4, [8, 8, 8], 2, 2, 100, 10000, 3, 1)
+
+
+def test_header_time(test_data_dir):
+    fname = f"{test_data_dir}/nek/channel3D_0.f00001"
+    header = read_header(fname)
+    assert header.time == 0.2