Run CI with two different uv resolution strategies: highest and lowest-direct (#3852)

* fix typing_extensions ImportError

#3752 (comment)

* bump min ase to 3.23.0

* test both lowest-direct and highest uv package resolutions in CI

* establish some loose lower bounds on all deps

needed for deps install with --resolution=lowest-direct to work

* netcdf4>=1.6

* pin chgnet==0.3.6

* try chgnet==0.3.5

* bump pre-commit hooks

* imperative doc strings

* doc str white space

* test_string->test_str

* ruff auto fixes

* matgl>=1.1.1 numba>=0.55

* try fix netcdf4 install

* h5py>=3.9.0

* monty>=2024.5.24

fixes TypeError: deprecated() got an unexpected keyword argument 'deadline'

* revert TestWavecar to again inherit from PymatgenTest

* h5py>=3.11.0

* phonopy>=2.23

* scipy>=1.13.0

* matplotlib>=3.8

* requests>=2.32

* don't run oldest/newest supported python twice on linux AND windows

* fix bad refactor in get_atom_map

fixes KeyError: 'Co' ipot = self.pot_dict[el.symbol]

Author: janosh

.github/workflows/test.yml

@@ -25,22 +25,18 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
+        # maximize CI coverage of different platforms and python versions while minimizing the
+        # total number of jobs. We run all pytest splits with the oldest supported python
+        # version (currently 3.9) on windows (seems most likely to surface errors) and with
+        # newest version (currently 3.12) on ubuntu (to get complete coverage on unix). We
+        # ignore mac-os, which is assumed to be similar to ubuntu.
+        config:
+          - { os: windows-latest, python: "3.9", resolution: highest }
+          - { os: ubuntu-latest, python: "3.12", resolution: lowest-direct }
         # pytest-split automatically distributes work load so parallel jobs finish in similar time
-        os: [ubuntu-latest, windows-latest]
-        python-version: ["3.9", "3.12"]
         split: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
-        # include/exclude is meant to maximize CI coverage of different platforms and python
-        # versions while minimizing the total number of jobs. We run all pytest splits with the
-        # oldest supported python version (currently 3.9) on windows (seems most likely to surface
-        # errors) and with newest version (currently 3.12) on ubuntu (to get complete and speedy
-        # coverage on unix). We ignore mac-os, which is assumed to be similar to ubuntu.
-        exclude:
-          - os: windows-latest
-            python-version: "3.12"
-          - os: ubuntu-latest
-            python-version: "3.9"
 
-    runs-on: ${{ matrix.os }}
+    runs-on: ${{ matrix.config.os }}
 
     env:
       PMG_MAPI_KEY: ${{ secrets.PMG_MAPI_KEY }}
@@ -55,13 +51,13 @@ jobs:
 
       - name: Create mamba environment
         run: |
-          micromamba create -n pmg python=${{ matrix.python-version }} --yes
+          micromamba create -n pmg python=${{ matrix.config.python }} --yes
 
       - name: Install uv
         run: micromamba run -n pmg pip install uv
 
       - name: Install ubuntu-only conda dependencies
-        if: matrix.os == 'ubuntu-latest'
+        if: matrix.config.os == 'ubuntu-latest'
         run: |
           micromamba install -n pmg -c conda-forge enumlib packmol bader openbabel openff-toolkit --yes
 
@@ -73,11 +69,7 @@ jobs:
           pip install torch
 
           uv pip install numpy cython
-
-          uv pip install --editable '.[dev,optional]'
-
-          # TODO remove next line installing ase from main branch when FrechetCellFilter is released
-          uv pip install --upgrade 'git+https://gitlab.com/ase/ase'
+          uv pip install --editable '.[dev,optional]' --resolution=${{ matrix.config.resolution }}
 
       - name: pytest split ${{ matrix.split }}
         run: |

.pre-commit-config.yaml

@@ -8,7 +8,7 @@ ci:
 
 repos:
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.4.3
+    rev: v0.4.7
     hooks:
       - id: ruff
         args: [--fix, --unsafe-fixes]
@@ -27,7 +27,7 @@ repos:
       - id: mypy
 
   - repo: https://github.com/codespell-project/codespell
-    rev: v2.2.6
+    rev: v2.3.0
     hooks:
       - id: codespell
         stages: [commit, commit-msg]
@@ -47,7 +47,7 @@ repos:
       - id: blacken-docs
 
   - repo: https://github.com/igorshubovych/markdownlint-cli
-    rev: v0.40.0
+    rev: v0.41.0
     hooks:
       - id: markdownlint
         # MD013: line too long
@@ -64,6 +64,6 @@ repos:
         args: [--drop-empty-cells, --keep-output]
 
   - repo: https://github.com/RobertCraigie/pyright-python
-    rev: v1.1.361
+    rev: v1.1.365
     hooks:
       - id: pyright

pymatgen/alchemy/materials.py

@@ -291,7 +291,7 @@ def from_cif_str(
             TransformedStructure
         """
         parser = CifParser.from_str(cif_string, occupancy_tolerance=occupancy_tolerance)
-        raw_string = re.sub(r"'", '"', cif_string)
+        raw_str = re.sub(r"'", '"', cif_string)
         cif_dict = parser.as_dict()
         cif_keys = list(cif_dict)
         struct = parser.parse_structures(primitive=primitive)[0]
@@ -303,7 +303,7 @@ def from_cif_str(
         source_info = {
             "source": source,
             "datetime": str(datetime.datetime.now()),
-            "original_file": raw_string,
+            "original_file": raw_str,
             "cif_data": cif_dict[cif_keys[0]],
         }
         return cls(struct, transformations, history=[source_info])
@@ -326,12 +326,12 @@ def from_poscar_str(
             raise ValueError(
                 "Transformation can be created only from POSCAR strings with proper VASP5 element symbols."
             )
-        raw_string = re.sub(r"'", '"', poscar_string)
+        raw_str = re.sub(r"'", '"', poscar_string)
         struct = poscar.structure
         source_info = {
             "source": "POSCAR",
             "datetime": str(datetime.datetime.now()),
-            "original_file": raw_string,
+            "original_file": raw_str,
         }
         return cls(struct, transformations, history=[source_info])
 

pymatgen/analysis/chemenv/connectivity/structure_connectivity.py

@@ -153,12 +153,12 @@ def setup_environment_subgraph(self, environments_symbols, only_atoms=None):
         if not isinstance(environments_symbols, collections.abc.Iterable):
             environments_symbols = [environments_symbols]
         environments_symbols = sorted(environments_symbols)
-        envs_string = "-".join(environments_symbols)
+        envs_str = "-".join(environments_symbols)
         if only_atoms is not None:
-            envs_string += "#" + "-".join(sorted(only_atoms))
+            envs_str += "#" + "-".join(sorted(only_atoms))
         # Get it directly if it was already computed
-        if envs_string in self.environment_subgraphs:
-            self._environment_subgraph = self.environment_subgraphs[envs_string]
+        if envs_str in self.environment_subgraphs:
+            self._environment_subgraph = self.environment_subgraphs[envs_str]
             return
 
         # Initialize graph for a subset of environments
@@ -248,7 +248,7 @@ def setup_environment_subgraph(self, environments_symbols, only_atoms=None):
                         delta=conn,
                         ligands=ligands,
                     )
-        self.environment_subgraphs[envs_string] = self._environment_subgraph
+        self.environment_subgraphs[envs_str] = self._environment_subgraph
 
     def setup_connectivity_description(self):
         pass

pymatgen/analysis/elasticity/elastic.py

@@ -728,8 +728,8 @@ def get_compliance_expansion(self):
         if not self.order <= 4:
             raise ValueError("Compliance tensor expansion only supported for fourth-order and lower")
         ce_exp = [ElasticTensor(self[0]).compliance_tensor]
-        ein_string = "ijpq,pqrsuv,rskl,uvmn->ijklmn"
-        ce_exp.append(np.einsum(ein_string, -ce_exp[-1], self[1], ce_exp[-1], ce_exp[-1]))
+        ein_str = "ijpq,pqrsuv,rskl,uvmn->ijklmn"
+        ce_exp.append(np.einsum(ein_str, -ce_exp[-1], self[1], ce_exp[-1], ce_exp[-1]))
         if self.order == 4:
             # Four terms in the Fourth-Order compliance tensor
             einstring_1 = "pqab,cdij,efkl,ghmn,abcdefgh"

pymatgen/core/structure.py

@@ -36,7 +36,6 @@
 from scipy.linalg import expm, polar
 from scipy.spatial.distance import squareform
 from tabulate import tabulate
-from typing_extensions import Self
 
 from pymatgen.core.bonds import CovalentBond, get_bond_length
 from pymatgen.core.composition import Composition
@@ -60,6 +59,7 @@
     from ase.optimize.optimize import Optimizer
     from matgl.ext.ase import TrajectoryObserver
     from numpy.typing import ArrayLike, NDArray
+    from typing_extensions import Self
 
     from pymatgen.util.typing import CompositionLike, MillerIndex, PathLike, PbcLike, SpeciesLike
 
@@ -3363,14 +3363,12 @@ def break_bond(self, ind1: int, ind2: int, tol: float = 0.2) -> tuple[Self, Self
         Args:
             ind1 (int): 1st site index
             ind2 (int): 2nd site index
-            tol (float): Relative tolerance to test. Basically, the code
-                checks if the distance between the sites is less than (1 +
-                tol) * typical bond distances. Defaults to 0.2, i.e.,
-                20% longer.
+            tol (float): Relative tolerance to test. Basically, the code checks if the distance
+                between the sites is less than (1 + tol) * typical bond distances.
+                Defaults to 0.2, i.e. 20% longer.
 
         Returns:
-            Two IMolecule representing the clusters formed from
-            breaking the bond.
+            tuple[IMolecule, IMolecule]: The clusters formed from breaking the bond.
         """
         clusters = ([self[ind1]], [self[ind2]])
 
@@ -3393,7 +3391,8 @@ def belongs_to_cluster(site, cluster):
                 raise ValueError("Not all sites are matched!")
             sites = unmatched
 
-        return cast(tuple[Self, Self], tuple(map(type(self).from_sites, clusters)))
+        from_sites = type(self).from_sites
+        return from_sites(clusters[0]), from_sites(clusters[1])
 
     def get_covalent_bonds(self, tol: float = 0.2) -> list[CovalentBond]:
         """Determine the covalent bonds in a molecule.

pymatgen/io/abinit/inputs.py

@@ -1286,8 +1286,7 @@ def has_same_variable(kref, vref, other_inp):
         return self[0].to_str(with_pseudos=with_pseudos)
 
     def write(self, filepath="run.abi"):
-        """
-        Write ndset input files to disk. The name of the file
+        """Write ndset input files to disk. The name of the file
         is constructed from the dataset index e.g. run0.abi.
         """
         root, ext = os.path.splitext(filepath)

pymatgen/io/abinit/netcdf.py

@@ -92,7 +92,7 @@ def __init__(self, path):
         self.ngroups = len(list(self.walk_tree()))
 
         # Always return non-masked numpy arrays.
-        # Slicing a ncvar returns a MaskedArrray and this is really annoying
+        # Slicing a ncvar returns a MaskedArray and this is really annoying
         # because it can lead to unexpected behavior in e.g. calls to np.matmul!
         # See also https://github.com/Unidata/netcdf4-python/issues/785
         self.rootgrp.set_auto_mask(False)

pymatgen/io/adf.py

@@ -533,8 +533,7 @@ def __init__(self, task):
         self.task = task
 
     def write_file(self, molecule, inp_file):
-        """
-        Write an ADF input file.
+        """Write an ADF input file.
 
         Args:
             molecule (Molecule): The molecule for this task.

pymatgen/io/common.py

@@ -266,8 +266,7 @@ def get_average_along_axis(self, ind):
         return total / ng[(ind + 1) % 3] / ng[(ind + 2) % 3]
 
     def to_hdf5(self, filename):
-        """
-        Writes the VolumetricData to a HDF5 format, which is a highly optimized
+        """Write the VolumetricData to a HDF5 format, which is a highly optimized
         format for reading storing large data. The mapping of the VolumetricData
         to this file format is as follows:
 
@@ -323,8 +322,7 @@ def from_hdf5(cls, filename: str, **kwargs) -> Self:
             return cls(structure, data=data, data_aug=data_aug, **kwargs)
 
     def to_cube(self, filename, comment: str = ""):
-        """
-        Write the total volumetric data to a cube file format, which consists of two comment lines,
+        """Write the total volumetric data to a cube file format, which consists of two comment lines,
         a header section defining the structure IN BOHR, and the data.
 
         Args: