mypy file selection in pre-commit

Author: evertlammerts

.pre-commit-config.yaml

@@ -23,6 +23,10 @@ repos:
     rev: v1.17.1
     hooks:
       - id: mypy
+        # Match all files under duckdb/ or _duckdb-stubs/,
+        # but not under duckdb/experimental/ or duckdb/query_graph/
+        files: ^(duckdb/(?!experimental/|query_graph/)|_duckdb-stubs/)
+        args: ["--config-file", "pyproject.toml"]
 
   - repo: local
     hooks:

duckdb/polars_io.py

@@ -277,8 +277,8 @@ def source_generator(
         for record_batch in iter(results.read_next_batch, None):
             if predicate is not None and duck_predicate is None:
                 # We have a predicate, but did not manage to push it down, we fallback here
-                yield pl.from_arrow(record_batch).filter(predicate)  # type: ignore[arg-type,misc]
+                yield pl.from_arrow(record_batch).filter(predicate)  # type: ignore[arg-type,misc,unused-ignore]
             else:
-                yield pl.from_arrow(record_batch)  # type: ignore[misc]
+                yield pl.from_arrow(record_batch)  # type: ignore[misc,unused-ignore]
 
     return register_io_source(source_generator, schema=schema)

pyproject.toml

@@ -316,20 +316,26 @@ filterwarnings = [
 
 [tool.mypy]
 packages = ["duckdb", "_duckdb"]
+mypy_path = ["_duckdb-stubs"]
 strict = true
 warn_unreachable = true
 pretty = true
 python_version = "3.9"
 exclude = [
     "duckdb/experimental/", # not checking the pyspark API
     "duckdb/query_graph/",  # old and unmaintained (should probably remove)
+    "tests",
+    "scripts",
+    "sqlogic",
 ]
 
 [[tool.mypy.overrides]]
 module = [
     "fsspec.*",
+    "numpy",
     "pandas",
     "polars",
+    "polars.io.plugins",
     "pyarrow.*",
     "torch",
     "tensorflow",