(chore): add bugbear and PIE lints (#1967)

Author: flying-sheep

pyproject.toml

@@ -181,6 +181,7 @@ docstring-code-format = true
 
 [tool.ruff.lint]
 select = [
+    "B",    # Likely bugs and design issues
     "BLE",  # Blind except
     "C4",   # Comprehensions
     "E",    # Error detected by Pycodestyle
@@ -191,6 +192,7 @@ select = [
     "ICN",  # Follow import conventions
     "ISC",  # Implicit string concatenation
     "PERF", # Performance
+    "PIE",  # Syntax simplifications
     "PTH",  # Pathlib instead of os.path
     "PT",   # Pytest conventions
     "PL",   # Pylint
@@ -221,6 +223,8 @@ allowed-confusables = [ "×", "’", "–", "α" ]
 [tool.ruff.lint.isort]
 known-first-party = [ "anndata" ]
 required-imports = [ "from __future__ import annotations" ]
+[tool.ruff.lint.flake8-bugbear]
+extend-immutable-calls = [ "slice" ]
 [tool.ruff.lint.flake8-tidy-imports.banned-api]
 "subprocess.call".msg = "Use `subprocess.run([…])` instead"
 "subprocess.check_call".msg = "Use `subprocess.run([…], check=True)` instead"

src/anndata/_core/aligned_df.py

@@ -80,7 +80,8 @@ def _gen_dataframe_df(
         raise _mk_df_error(source, attr, length, len(anno))
     anno = anno.copy(deep=False)
     if not is_string_dtype(anno.index):
-        warnings.warn("Transforming to str index.", ImplicitModificationWarning)
+        msg = "Transforming to str index."
+        warnings.warn(msg, ImplicitModificationWarning, stacklevel=2)
         anno.index = anno.index.astype(str)
     if not len(anno.columns):
         anno.columns = anno.columns.astype(str)

src/anndata/_core/anndata.py

@@ -391,10 +391,10 @@ def _init_as_actual(  # noqa: PLR0912, PLR0913, PLR0915
             _check_2d_shape(X)
             # if type doesn’t match, a copy is made, otherwise, use a view
             if dtype is not None:
-                warnings.warn(
-                    "The dtype argument is deprecated and will be removed in late 2024.",
-                    FutureWarning,
+                msg = (
+                    "The dtype argument is deprecated and will be removed in late 2024."
                 )
+                warnings.warn(msg, FutureWarning, stacklevel=3)
                 if issparse(X) or isinstance(X, ma.MaskedArray):
                     # TODO: maybe use view on data attribute of sparse matrix
                     #       as in readwrite.read_10x_h5
@@ -1300,11 +1300,11 @@ def obs_vector(self, k: str, *, layer: str | None = None) -> np.ndarray:
             if "X" in self.layers:
                 pass
             else:
-                warnings.warn(
+                msg = (
                     "In a future version of AnnData, access to `.X` by passing"
-                    " `layer='X'` will be removed. Instead pass `layer=None`.",
-                    FutureWarning,
+                    " `layer='X'` will be removed. Instead pass `layer=None`."
                 )
+                warnings.warn(msg, FutureWarning, stacklevel=2)
                 layer = None
         return get_vector(self, k, "obs", "var", layer=layer)
 
@@ -1332,11 +1332,11 @@ def var_vector(self, k, *, layer: str | None = None) -> np.ndarray:
             if "X" in self.layers:
                 pass
             else:
-                warnings.warn(
+                msg = (
                     "In a future version of AnnData, access to `.X` by passing "
-                    "`layer='X'` will be removed. Instead pass `layer=None`.",
-                    FutureWarning,
+                    "`layer='X'` will be removed. Instead pass `layer=None`."
                 )
+                warnings.warn(msg, FutureWarning, stacklevel=2)
                 layer = None
         return get_vector(self, k, "var", "obs", layer=layer)
 
@@ -2100,7 +2100,7 @@ def _infer_shape_for_axis(
     for elem in [xxx, xxxm, xxxp]:
         if elem is not None and hasattr(elem, "shape"):
             return elem.shape[0]
-    for elem, id in zip([layers, xxxm, xxxp], ["layers", "xxxm", "xxxp"]):
+    for elem, id in zip([layers, xxxm, xxxp], ["layers", "xxxm", "xxxp"], strict=True):
         if elem is not None:
             elem = cast("Mapping", elem)
             for sub_elem in elem.values():

src/anndata/_core/merge.py

@@ -801,7 +801,7 @@ def concat_arrays(arrays, reindexers, axis=0, index=None, fill_value=None):  # n
             raise NotImplementedError(msg)
         # TODO: behaviour here should be chosen through a merge strategy
         df = pd.concat(
-            unify_dtypes(f(x) for f, x in zip(reindexers, arrays)),
+            unify_dtypes(f(x) for f, x in zip(reindexers, arrays, strict=True)),
             axis=axis,
             ignore_index=True,
         )
@@ -816,7 +816,9 @@ def concat_arrays(arrays, reindexers, axis=0, index=None, fill_value=None):  # n
             msg = "Cannot concatenate an AwkwardArray with other array types."
             raise NotImplementedError(msg)
 
-        return ak.concatenate([f(a) for f, a in zip(reindexers, arrays)], axis=axis)
+        return ak.concatenate(
+            [f(a) for f, a in zip(reindexers, arrays, strict=True)], axis=axis
+        )
     elif any(isinstance(a, CupySparseMatrix) for a in arrays):
         import cupyx.scipy.sparse as cpsparse
 
@@ -829,7 +831,7 @@ def concat_arrays(arrays, reindexers, axis=0, index=None, fill_value=None):  # n
         return sparse_stack(
             [
                 f(as_cp_sparse(a), axis=1 - axis, fill_value=fill_value)
-                for f, a in zip(reindexers, arrays)
+                for f, a in zip(reindexers, arrays, strict=True)
             ],
             format="csr",
         )
@@ -842,7 +844,7 @@ def concat_arrays(arrays, reindexers, axis=0, index=None, fill_value=None):  # n
         return cp.concatenate(
             [
                 f(cp.asarray(x), fill_value=fill_value, axis=1 - axis)
-                for f, x in zip(reindexers, arrays)
+                for f, x in zip(reindexers, arrays, strict=True)
             ],
             axis=axis,
         )
@@ -856,7 +858,7 @@ def concat_arrays(arrays, reindexers, axis=0, index=None, fill_value=None):  # n
                     axis=1 - axis,
                     fill_value=fill_value,
                 )
-                for f, a in zip(reindexers, arrays)
+                for f, a in zip(reindexers, arrays, strict=True)
             ],
             format="csr",
         )
@@ -871,7 +873,7 @@ def concat_arrays(arrays, reindexers, axis=0, index=None, fill_value=None):  # n
         return np.concatenate(
             [
                 f(x, fill_value=fill_value, axis=1 - axis)
-                for f, x in zip(reindexers, arrays)
+                for f, x in zip(reindexers, arrays, strict=True)
             ],
             axis=axis,
         )
@@ -932,7 +934,7 @@ def gen_outer_reindexers(els, shapes, new_index: pd.Index, *, axis=0):
             (lambda x: x)
             if not_missing(el)
             else (lambda _, shape=shape: pd.DataFrame(index=range(shape)))
-            for el, shape in zip(els, shapes)
+            for el, shape in zip(els, shapes, strict=True)
         ]
     elif any(isinstance(el, AwkArray) for el in els if not_missing(el)):
         import awkward as ak
@@ -1025,7 +1027,7 @@ def outer_concat_aligned_mapping(
                     fill_value=fill_value,
                     off_axis_size=off_axis_size,
                 )
-                for el, n in zip(els, ns)
+                for el, n in zip(els, ns, strict=True)
             ],
             cur_reindexers,
             axis=concat_axis,
@@ -1046,7 +1048,8 @@ def concat_pairwise_mapping(
 
     for k in join_keys(mappings):
         els = [
-            m.get(k, sparse_class((s, s), dtype=bool)) for m, s in zip(mappings, shapes)
+            m.get(k, sparse_class((s, s), dtype=bool))
+            for m, s in zip(mappings, shapes, strict=True)
         ]
         if all(isinstance(el, CupySparseMatrix | CupyArray) for el in els):
             result[k] = _cp_block_diag(els, format="csr")
@@ -1075,7 +1078,7 @@ def merge_outer(mappings, batch_keys, *, join_index="-", merge=merge_unique):
     all_keys = union_keys(mappings)
     out = merge(mappings)
     for key in all_keys.difference(out.keys()):
-        for b, m in zip(batch_keys, mappings):
+        for b, m in zip(batch_keys, mappings, strict=True):
             val = m.get(key, None)
             if val is not None:
                 out[f"{key}{join_index}{b}"] = val
@@ -1633,7 +1636,7 @@ def concat(  # noqa: PLR0912, PLR0913, PLR0915
     alt_mapping = merge(
         [
             {k: r(v, axis=0) for k, v in getattr(a, f"{alt_axis_name}m").items()}
-            for r, a in zip(reindexers, adatas)
+            for r, a in zip(reindexers, adatas, strict=True)
         ],
     )
     alt_pairwise = merge(
@@ -1642,7 +1645,7 @@ def concat(  # noqa: PLR0912, PLR0913, PLR0915
                 k: r(r(v, axis=0), axis=1)
                 for k, v in getattr(a, f"{alt_axis_name}p").items()
             }
-            for r, a in zip(reindexers, adatas)
+            for r, a in zip(reindexers, adatas, strict=True)
         ]
     )
     uns = uns_merge([a.uns for a in adatas])
@@ -1668,11 +1671,11 @@ def concat(  # noqa: PLR0912, PLR0913, PLR0915
             axis=axis,
         )
     elif any(has_raw):
-        warn(
+        msg = (
             "Only some AnnData objects have `.raw` attribute, "
-            "not concatenating `.raw` attributes.",
-            UserWarning,
+            "not concatenating `.raw` attributes."
         )
+        warn(msg, UserWarning, stacklevel=2)
     return AnnData(
         **{
             "X": X,

src/anndata/_core/sparse_dataset.py

@@ -306,7 +306,7 @@ def get_compressed_vectors_for_slices(
     if len(slices) < 2:  # there is only one slice so no need to concatenate
         return data, indices, start_indptr
     end_indptr = np.concatenate(
-        [s[1:] - o for s, o in zip(indptr_indices[1:], offsets)]
+        [s[1:] - o for s, o in zip(indptr_indices[1:], offsets, strict=True)]
     )
     indptr = np.concatenate([start_indptr, end_indptr])
     return data, indices, indptr
@@ -499,10 +499,10 @@ def _normalize_index(
         return row, col
 
     def __setitem__(self, index: Index | tuple[()], value) -> None:
-        warnings.warn(
-            "__setitem__ for backed sparse will be removed in the next anndata release.",
-            FutureWarning,
+        msg = (
+            "__setitem__ for backed sparse will be removed in the next anndata release."
         )
+        warnings.warn(msg, FutureWarning, stacklevel=2)
         row, col = self._normalize_index(index)
         mock_matrix = self._to_backed()
         mock_matrix[row, col] = value

src/anndata/_core/storage.py

@@ -47,7 +47,7 @@ def __repr__():
     if isinstance(value, (*array_data_structure_types, Dataset2D)):
         if isinstance(value, np.matrix):
             msg = f"{name} should not be a np.matrix, use np.ndarray instead."
-            warnings.warn(msg, ImplicitModificationWarning)
+            warnings.warn(msg, ImplicitModificationWarning, stacklevel=3)
             value = value.A
         return value
     is_non_csc_r_array_or_matrix = (

src/anndata/_core/views.py

@@ -153,7 +153,7 @@ def convert_all(arrs: Iterable[np.ndarray]) -> Iterable[np.ndarray]:
             results = (results,)
         results = tuple(
             (np.asarray(result) if output is None else output)
-            for result, output in zip(results, outputs)
+            for result, output in zip(results, outputs, strict=True)
         )
         return results[0] if len(results) == 1 else results
 

src/anndata/_io/__init__.py

@@ -9,9 +9,9 @@ def __getattr__(key: str):
     from .. import io
 
     attr = getattr(io, key)
-    warnings.warn(
+    msg = (
         f"Importing {key} from `anndata._io` is deprecated. "
-        "Please use anndata.io instead.",
-        FutureWarning,
+        "Please use anndata.io instead."
     )
+    warnings.warn(msg, FutureWarning, stacklevel=2)
     return attr

src/anndata/_io/h5ad.py

@@ -296,11 +296,11 @@ def _read_raw(
 @report_read_key_on_error
 def read_dataframe_legacy(dataset: h5py.Dataset) -> pd.DataFrame:
     """Read pre-anndata 0.7 dataframes."""
-    warn(
-        f"'{dataset.name}' was written with a very old version of AnnData. "
-        "Consider rewriting it.",
-        OldFormatWarning,
+    msg = (
+        f"{dataset.name!r} was written with a very old version of AnnData. "
+        "Consider rewriting it."
     )
+    warn(msg, OldFormatWarning, stacklevel=2)
     if H5PY_V3:
         df = pd.DataFrame(
             _decode_structured_array(

src/anndata/_io/read.py

@@ -227,11 +227,11 @@ def read_loom(  # noqa: PLR0912, PLR0913
     """
     # Deprecations
     if obsm_names is not None:
-        warn(
+        msg = (
             "Argument obsm_names has been deprecated in favour of `obsm_mapping`. "
-            "In 0.9 this will be an error.",
-            FutureWarning,
+            "In 0.9 this will be an error."
         )
+        warn(msg, FutureWarning, stacklevel=2)
         if obsm_mapping != {}:
             msg = (
                 "Received values for both `obsm_names` and `obsm_mapping`. This is "
@@ -240,11 +240,11 @@ def read_loom(  # noqa: PLR0912, PLR0913
             raise ValueError(msg)
         obsm_mapping = obsm_names
     if varm_names is not None:
-        warn(
+        msg = (
             "Argument varm_names has been deprecated in favour of `varm_mapping`. "
-            "In 0.9 this will be an error.",
-            FutureWarning,
+            "In 0.9 this will be an error."
         )
+        warn(msg, FutureWarning, stacklevel=2)
         if varm_mapping != {}:
             msg = (
                 "Received values for both `varm_names` and `varm_mapping`. This is "