CI: run stubtest (#178)

* CI: run stubtest

* maybe fix windows bug?

* remove allowlist

* isort

* remove non-existing files

* slightly nicer solution?

* fix most default arguments

* remove very few of the many non-existing things

* read_json

Author: twoertwein

docs/tests.md

@@ -11,5 +11,6 @@ Here are the most important options. Fore more details, please use `poe --help`.
   - Run only pytest: `poe pytest`
   - Run only pre-commit: `poe style`
 - Run tests against the installed stubs (this will install and uninstall the stubs): `poe test_dist`
+- Optional: Run stubtest to compare the installed pandas-stubs against pandas (this will fail): `poe stubtest`. If you have created an allowlist to ignore certain errors: `poe stubtest path_to_the_allow_list`
 
 These tests originally came from https://github.com/VirtusLab/pandas-stubs.

pandas-stubs/core/arrays/_arrow_utils.pyi

@@ -67,15 +67,15 @@ class Categorical(ExtensionArray, PandasObject):
     @overload
     def set_ordered(self, value, inplace: Literal[True]) -> None: ...
     @overload
-    def set_ordered(self, value, inplace: Literal[False]) -> Categorical: ...
+    def set_ordered(self, value, inplace: Literal[False] = ...) -> Categorical: ...
     @overload
-    def set_ordered(self, value, inplace: bool) -> Categorical | None: ...
+    def set_ordered(self, value, inplace: bool = ...) -> Categorical | None: ...
     @overload
     def as_ordered(self, inplace: Literal[True]) -> None: ...
     @overload
-    def as_ordered(self, inplace: Literal[False]) -> Categorical: ...
+    def as_ordered(self, inplace: Literal[False] = ...) -> Categorical: ...
     @overload
-    def as_ordered(self, inplace: bool) -> Categorical | None: ...
+    def as_ordered(self, inplace: bool = ...) -> Categorical | None: ...
     @overload
     def as_unordered(self, inplace: Literal[True]) -> None: ...
     @overload
@@ -155,7 +155,7 @@ class Categorical(ExtensionArray, PandasObject):
     def __ge__(self, other) -> bool: ...
     @property
     def shape(self): ...
-    def shift(self, periods, fill_value=...): ...
+    def shift(self, periods=..., fill_value=...): ...
     def __array__(self, dtype=...) -> np.ndarray: ...
     def __array_ufunc__(self, ufunc, method, *inputs, **kwargs): ...
     @property

pandas-stubs/core/arrays/categorical.pyi

@@ -1,6 +1,6 @@
 def set_use_numexpr(v: bool = ...) -> None: ...
 def set_numexpr_threads(n=...) -> None: ...
-def evaluate(op, op_str, a, b, use_numexpr: bool = ...): ...
+def evaluate(op, a, b, use_numexpr: bool = ...): ...
 def where(cond, a, b, use_numexpr: bool = ...): ...
 def set_test_mode(v: bool = ...) -> None: ...
 def get_test_result(): ...

pandas-stubs/core/computation/expressions.pyi

@@ -1,7 +1,7 @@
 import numpy as np
 
 class UndefinedVariableError(NameError):
-    def __init__(self, name, is_local: bool) -> None: ...
+    def __init__(self, name, is_local: bool = ...) -> None: ...
 
 class Term:
     def __new__(cls, name, env, side=..., encoding=...): ...

pandas-stubs/core/computation/ops.pyi

@@ -1291,7 +1291,7 @@ class DataFrame(NDFrame, OpsMixin):
         fill_value: float | None = ...,
     ) -> DataFrame: ...
     def droplevel(
-        self, level: Level | list[Level] = ..., axis: AxisType = ...
+        self, level: Level | list[Level], axis: AxisType = ...
     ) -> DataFrame: ...
     def eq(
         self, other, axis: AxisType = ..., level: Level | None = ...

pandas-stubs/core/frame.pyi

@@ -191,7 +191,7 @@ class Index(IndexOpsMixin, PandasObject):
     def map(self, mapper, na_action=...) -> Index: ...
     def isin(self, values, level=...) -> np_ndarray_bool: ...
     def slice_indexer(self, start=..., end=..., step=..., kind=...): ...
-    def get_slice_bound(self, label, side, kind): ...
+    def get_slice_bound(self, label, side, kind=...): ...
     def slice_locs(self, start=..., end=..., step=..., kind=...): ...
     def delete(self, loc): ...
     def insert(self, loc, item): ...

pandas-stubs/core/indexes/base.pyi

@@ -25,7 +25,7 @@ class CategoricalIndex(ExtensionIndex, accessor.PandasDelegate):
     def __contains__(self, key) -> bool: ...
     def __array__(self, dtype=...) -> np.ndarray: ...
     def astype(self, dtype: DtypeArg, copy: bool = ...) -> Index: ...
-    def fillna(self, value, downcast=...): ...
+    def fillna(self, value=..., downcast=...): ...
     def is_unique(self) -> bool: ...
     @property
     def is_monotonic_increasing(self) -> bool: ...

pandas-stubs/core/indexes/category.pyi

@@ -117,7 +117,7 @@ class MultiIndex(Index):
     def get_indexer_non_unique(self, target): ...
     def reindex(self, target, method=..., level=..., limit=..., tolerance=...): ...
     def get_slice_bound(
-        self, label: Hashable | Sequence[Hashable], side: str, kind: str
+        self, label: Hashable | Sequence[Hashable], side: str, kind: str = ...
     ) -> int: ...
     def slice_locs(self, start=..., end=..., step=..., kind=...): ...
     def get_loc(self, key, method=...): ...

pandas-stubs/core/indexes/multi.pyi

@@ -6,17 +6,15 @@ from pandas.core.groupby.grouper import Grouper as Grouper
 from pandas._typing import FrameOrSeriesUnion
 
 class Resampler(BaseGroupBy, ShallowMixin):
-    def __init__(
-        self, obj, groupby=..., axis: int = ..., kind=..., **kwargs
-    ) -> None: ...
+    def __init__(self, obj, groupby, axis: int = ..., kind=..., **kwargs) -> None: ...
     def __getattr__(self, attr: str): ...
     def __iter__(self): ...
     @property
     def obj(self): ...
     @property
     def ax(self): ...
     def pipe(self, func, *args, **kwargs): ...
-    def aggregate(self, func, *args, **kwargs): ...
+    def aggregate(self, func=..., *args, **kwargs): ...
     agg = aggregate
     def transform(self, arg, *args, **kwargs): ...
     def pad(self, limit=...): ...

pandas-stubs/core/ops/roperator.pyi

@@ -1,33 +1,6 @@
 import numpy as np
 from pandas.core.frame import DataFrame
 
-class _Unstacker:
-    values = ...
-    value_columns = ...
-    fill_value = ...
-    constructor = ...
-    index = ...
-    level = ...
-    lift = ...
-    new_index_levels = ...
-    new_index_names = ...
-    removed_name = ...
-    removed_level = ...
-    removed_level_full = ...
-    def __init__(
-        self,
-        values: np.ndarray,
-        index,
-        level=...,
-        value_columns=...,
-        fill_value=...,
-        constructor=...,
-    ) -> None: ...
-    def get_result(self): ...
-    def get_new_values(self): ...
-    def get_new_columns(self): ...
-    def get_new_index(self): ...
-
 def unstack(obj, level, fill_value=...): ...
 def stack(frame, level: int = ..., dropna: bool = ...): ...
 def stack_multiple(frame, level, dropna: bool = ...): ...

pandas-stubs/core/resample.pyi

@@ -393,7 +393,7 @@ class Series(IndexOpsMixin, NDFrame, Generic[S1]):
     def count(self, level: None = ...) -> int: ...
     @overload
     def count(self, level: Hashable) -> Series[S1]: ...
-    def mode(self, dropna) -> Series[S1]: ...
+    def mode(self, dropna=...) -> Series[S1]: ...
     def unique(self) -> np.ndarray: ...
     @overload
     def drop_duplicates(

pandas-stubs/core/reshape/reshape.pyi

@@ -46,7 +46,7 @@ class _Window(PandasObject, ShallowMixin, SelectionMixin):
     def __getattr__(self, attr: str): ...
     def __iter__(self): ...
     def aggregate(
-        self, func: Callable | None = ..., *args, **kwargs
+        self, func: Callable | None, *args, **kwargs
     ) -> Scalar | FrameOrSeries: ...
     agg = aggregate
 

pandas-stubs/core/series.pyi

@@ -43,7 +43,7 @@ class FastParquetImpl(BaseImpl):
 
 def to_parquet(
     df: DataFrame,
-    path,
+    path=...,
     engine: str = ...,
     compression=...,
     index: bool | None = ...,

pandas-stubs/core/window/rolling.pyi

@@ -7,7 +7,6 @@ from typing import (
 from pandas.core.base import PandasObject
 from pandas.core.frame import DataFrame
 
-class SQLAlchemyRequired(ImportError): ...
 class DatabaseError(IOError): ...
 
 def execute(sql, con, cur=..., params=...): ...
@@ -33,7 +32,7 @@ def read_sql_query(
 ) -> DataFrame: ...
 def read_sql(
     sql: str | Any,
-    con: str | Any = ...,
+    con: str | Any,
     index_col: str | Sequence[str] | None = ...,
     coerce_float: bool = ...,
     params: Sequence[str] | tuple[str, ...] | Mapping[str, str] | None = ...,

pandas-stubs/io/gcs.pyi

@@ -1,3 +1,2 @@
-def get_sys_info() -> list[tuple[str, str | int | None]]: ...
 def show_versions(as_json: bool = ...) -> None: ...
 def main() -> int: ...

pandas-stubs/io/parquet.pyi

@@ -1,18 +1,13 @@
 from typing import Callable
 
 def safe_import(mod_name: str, min_version: str | None = ...): ...
-
-tables = ...
-xfail_non_writeable = ...
-
 def skip_if_installed(package: str) -> Callable: ...
 def skip_if_no(package: str, min_version: str | None = ...) -> Callable: ...
 
 skip_if_no_mpl = ...
 skip_if_mpl = ...
 skip_if_32bit = ...
 skip_if_windows = ...
-skip_if_windows_python_3 = ...
 skip_if_has_locale = ...
 skip_if_not_us_locale = ...
 skip_if_no_scipy = ...

pandas-stubs/io/s3.pyi

@@ -88,6 +88,10 @@ script = "scripts.test.run:pyright_src"
 help = "Run pyright on 'tests' using the installed stubs"
 script = "scripts.test:test(dist=True, type_checker='pyright')"
 
+[tool.poe.tasks.stubtest]
+script = "scripts.test:stubtest(allowlist)"
+help = "Run stubtest to compare the installed stubs against pandas"
+args = [{ name = "allowlist", positional = true, default = "", required = false, help= "Path to an allowlist (optional)" }]
 
 
 [tool.black]

pandas-stubs/io/sql.pyi

@@ -63,5 +63,8 @@ def run_job(steps: List[Step]) -> None:
         end = time.perf_counter()
         logger.success(f"End: '{step.name}', runtime: {end - start:.3f} seconds.")
 
+    if not failed:
+        __rollback_job(rollback_steps)
+
     if failed:
         sys.exit(1)

pandas-stubs/util/_print_versions.pyi

@@ -1,3 +1,5 @@
+import dataclasses
+from functools import partial
 from typing import Literal
 
 from scripts._job import run_job
@@ -10,8 +12,6 @@
     _step.rename_src,
     _step.mypy_dist,
     _step.pyright_dist,
-    _step.uninstall_dist,
-    _step.restore_src,
 ]
 
 
@@ -33,3 +33,10 @@ def test(
         steps = [step for step in steps if remove not in step.name]
 
     run_job(steps)
+
+
+def stubtest(allowlist: str):
+    stubtest = dataclasses.replace(
+        _step.stubtest, run=partial(_step.stubtest.run, allowlist=allowlist)
+    )
+    run_job(_DIST_STEPS[:-2] + [stubtest])

pandas-stubs/util/_test_decorators.pyi

@@ -26,5 +26,6 @@
 pyright_dist = Step(
     name="Run pyright on 'tests' using the installed stubs", run=run.pyright_dist
 )
-uninstall_dist = Step(name="Uninstall pandas-stubs", run=run.uninstall_dist)
-restore_src = Step(name="Restore local stubs", run=run.restore_src)
+stubtest = Step(
+    name="Run stubtest to compare the installed stubs against pandas", run=run.stubtest
+)

pyproject.toml

@@ -1,5 +1,6 @@
 from pathlib import Path
 import subprocess
+import sys
 
 
 def mypy_src():
@@ -22,6 +23,22 @@ def style():
     subprocess.run(cmd, check=True)
 
 
+def stubtest(allowlist: str = ""):
+    cmd = [
+        sys.executable,
+        "-m",
+        "mypy.stubtest",
+        "pandas",
+        "--concise",
+        "--ignore-missing-stub",
+        "--mypy-config-file",
+        "pyproject.toml",
+    ]
+    if allowlist:
+        cmd += ["--allowlist", allowlist]
+    subprocess.run(cmd, check=True)
+
+
 def build_dist():
     cmd = ["poetry", "build", "-f", "wheel"]
     subprocess.run(cmd, check=True)