[3.14] gh-133403: Check Tools/build/deepfreeze.py with mypy (GH-133802) (#134038)

miss-islington · sobolevn · web-flow · commit 620065c33e11 · 2025-05-15T09:39:20.000Z
gh-133403: Check `Tools/build/deepfreeze.py` with mypy (GH-133802) (cherry picked from commit 7eaa097) Co-authored-by: sobolevn <mail@sobolevn.me>
diff --git a/.github/workflows/mypy.yml b/.github/workflows/mypy.yml
@@ -14,10 +14,12 @@ on:
       - "Lib/tomllib/**"
       - "Misc/mypy/**"
       - "Tools/build/compute-changes.py"
+      - "Tools/build/deepfreeze.py"
       - "Tools/build/generate_sbom.py"
       - "Tools/build/generate-build-details.py"
       - "Tools/build/verify_ensurepip_wheels.py"
       - "Tools/build/update_file.py"
+      - "Tools/build/umarshal.py"
       - "Tools/cases_generator/**"
       - "Tools/clinic/**"
       - "Tools/jit/**"
diff --git a/Tools/build/.ruff.toml b/Tools/build/.ruff.toml
@@ -1,7 +1,7 @@
 extend = "../../.ruff.toml"  # Inherit the project-wide settings
 
 [per-file-target-version]
-"deepfreeze.py" = "py310"
+"deepfreeze.py" = "py311"  # requires `code.co_exceptiontable`
 "stable_abi.py" = "py311"  # requires 'tomllib'
 
 [format]
diff --git a/Tools/build/deepfreeze.py b/Tools/build/deepfreeze.py
@@ -2,9 +2,12 @@
 
 The script may be executed by _bootstrap_python interpreter.
 Shared library extension modules are not available in that case.
-On Windows, and in cross-compilation cases, it is executed
-by Python 3.10, and 3.11 features are not available.
+Requires 3.11+ to be executed,
+because relies on `code.co_qualname` and `code.co_exceptiontable`.
 """
+
+from __future__ import annotations
+
 import argparse
 import builtins
 import collections
@@ -13,10 +16,14 @@
 import re
 import time
 import types
-from typing import TextIO
 
 import umarshal
 
+TYPE_CHECKING = False
+if TYPE_CHECKING:
+    from collections.abc import Iterator
+    from typing import Any, TextIO
+
 ROOT = os.path.dirname(os.path.dirname(os.path.dirname(__file__)))
 
 verbose = False
@@ -45,8 +52,8 @@ def make_string_literal(b: bytes) -> str:
 
 next_code_version = 1
 
-def get_localsplus(code: types.CodeType):
-    a = collections.defaultdict(int)
+def get_localsplus(code: types.CodeType) -> tuple[tuple[str, ...], bytes]:
+    a: collections.defaultdict[str, int] = collections.defaultdict(int)
     for name in code.co_varnames:
         a[name] |= CO_FAST_LOCAL
     for name in code.co_cellvars:
@@ -136,7 +143,7 @@ def get_identifiers_and_strings(self) -> tuple[set[str], dict[str, str]]:
         return identifiers, strings
 
     @contextlib.contextmanager
-    def indent(self) -> None:
+    def indent(self) -> Iterator[None]:
         save_level = self.level
         try:
             self.level += 1
@@ -148,7 +155,7 @@ def write(self, arg: str) -> None:
         self.file.writelines(("    "*self.level, arg, "\n"))
 
     @contextlib.contextmanager
-    def block(self, prefix: str, suffix: str = "") -> None:
+    def block(self, prefix: str, suffix: str = "") -> Iterator[None]:
         self.write(prefix + " {")
         with self.indent():
             yield
@@ -250,9 +257,17 @@ def generate_code(self, name: str, code: types.CodeType) -> str:
         co_names = self.generate(name + "_names", code.co_names)
         co_filename = self.generate(name + "_filename", code.co_filename)
         co_name = self.generate(name + "_name", code.co_name)
-        co_qualname = self.generate(name + "_qualname", code.co_qualname)
         co_linetable = self.generate(name + "_linetable", code.co_linetable)
-        co_exceptiontable = self.generate(name + "_exceptiontable", code.co_exceptiontable)
+        # We use 3.10 for type checking, but this module requires 3.11
+        # TODO: bump python version for this script.
+        co_qualname = self.generate(
+            name + "_qualname",
+            code.co_qualname,  # type: ignore[attr-defined]
+        )
+        co_exceptiontable = self.generate(
+            name + "_exceptiontable",
+            code.co_exceptiontable,  # type: ignore[attr-defined]
+        )
         # These fields are not directly accessible
         localsplusnames, localspluskinds = get_localsplus(code)
         co_localsplusnames = self.generate(name + "_localsplusnames", localsplusnames)
@@ -379,13 +394,13 @@ def generate_complex(self, name: str, z: complex) -> str:
             self.write(f".cval = {{ {z.real}, {z.imag} }},")
         return f"&{name}.ob_base"
 
-    def generate_frozenset(self, name: str, fs: frozenset[object]) -> str:
+    def generate_frozenset(self, name: str, fs: frozenset[Any]) -> str:
         try:
-            fs = sorted(fs)
+            fs_sorted = sorted(fs)
         except TypeError:
             # frozen set with incompatible types, fallback to repr()
-            fs = sorted(fs, key=repr)
-        ret = self.generate_tuple(name, tuple(fs))
+            fs_sorted = sorted(fs, key=repr)
+        ret = self.generate_tuple(name, tuple(fs_sorted))
         self.write("// TODO: The above tuple should be a frozenset")
         return ret
 
@@ -402,7 +417,7 @@ def generate(self, name: str, obj: object) -> str:
             # print(f"Cache hit {key!r:.40}: {self.cache[key]!r:.40}")
             return self.cache[key]
         self.misses += 1
-        if isinstance(obj, (types.CodeType, umarshal.Code)) :
+        if isinstance(obj, types.CodeType) :
             val = self.generate_code(name, obj)
         elif isinstance(obj, tuple):
             val = self.generate_tuple(name, obj)
@@ -458,7 +473,7 @@ def decode_frozen_data(source: str) -> types.CodeType:
         if re.match(FROZEN_DATA_LINE, line):
             values.extend([int(x) for x in line.split(",") if x.strip()])
     data = bytes(values)
-    return umarshal.loads(data)
+    return umarshal.loads(data)  # type: ignore[no-any-return]
 
 
 def generate(args: list[str], output: TextIO) -> None:
@@ -494,12 +509,12 @@ def generate(args: list[str], output: TextIO) -> None:
                    help="Input file and module name (required) in file:modname format")
 
 @contextlib.contextmanager
-def report_time(label: str):
-    t0 = time.time()
+def report_time(label: str) -> Iterator[None]:
+    t0 = time.perf_counter()
     try:
         yield
     finally:
-        t1 = time.time()
+        t1 = time.perf_counter()
     if verbose:
         print(f"{label}: {t1-t0:.3f} sec")
 
diff --git a/Tools/build/mypy.ini b/Tools/build/mypy.ini
@@ -4,10 +4,12 @@
 # .github/workflows/mypy.yml
 files =
     Tools/build/compute-changes.py,
+    Tools/build/deepfreeze.py,
     Tools/build/generate-build-details.py,
     Tools/build/generate_sbom.py,
     Tools/build/verify_ensurepip_wheels.py,
-    Tools/build/update_file.py
+    Tools/build/update_file.py,
+    Tools/build/umarshal.py
 
 pretty = True
 
diff --git a/Tools/build/umarshal.py b/Tools/build/umarshal.py
@@ -145,12 +145,12 @@ def r_PyLong(self) -> int:
     def r_float_bin(self) -> float:
         buf = self.r_string(8)
         import struct  # Lazy import to avoid breaking UNIX build
-        return struct.unpack("d", buf)[0]
+        return struct.unpack("d", buf)[0]  # type: ignore[no-any-return]
 
     def r_float_str(self) -> float:
         n = self.r_byte()
         buf = self.r_string(n)
-        return ast.literal_eval(buf.decode("ascii"))
+        return ast.literal_eval(buf.decode("ascii"))  # type: ignore[no-any-return]
 
     def r_ref_reserve(self, flag: int) -> int:
         if flag:
@@ -306,16 +306,17 @@ def loads(data: bytes) -> Any:
     return r.r_object()
 
 
-def main():
+def main() -> None:
     # Test
     import marshal
     import pprint
     sample = {'foo': {(42, "bar", 3.14)}}
     data = marshal.dumps(sample)
     retval = loads(data)
     assert retval == sample, retval
-    sample = main.__code__
-    data = marshal.dumps(sample)
+
+    sample2 = main.__code__
+    data = marshal.dumps(sample2)
     retval = loads(data)
     assert isinstance(retval, Code), retval
     pprint.pprint(retval.__dict__)
