[update_lib] todo subcommand (#6823)

youknowone · web-flow · commit d690b2d65ea8 · 2026-01-21T09:00:45.000+09:00
* [update_lib] todo

* better CI comment
diff --git a/.github/workflows/lib-deps-check.yaml b/.github/workflows/lib-deps-check.yaml
@@ -28,6 +28,11 @@ jobs:
         run: |
           git fetch origin ${{ github.event.pull_request.head.sha }}
 
+      - name: Checkout PR Lib files
+        run: |
+          # Checkout only Lib/ directory from PR head for accurate comparison
+          git checkout ${{ github.event.pull_request.head.sha }} -- Lib/
+
       - name: Checkout CPython
         run: |
           git clone --depth 1 --branch v3.14.2 https://github.com/python/cpython.git cpython
@@ -104,7 +109,7 @@ jobs:
             </details>
 
             **Legend:**
-            - `[+]` path exists, `[-]` path missing
+            - `[+]` path exists in CPython
             - `[x]` up-to-date, `[ ]` outdated
             - `native:` Rust/C extension modules
 
diff --git a/scripts/update_lib/__main__.py b/scripts/update_lib/__main__.py
@@ -54,6 +54,11 @@ def main(argv: list[str] | None = None) -> int:
         help="Show dependency information for a module",
         add_help=False,
     )
+    subparsers.add_parser(
+        "todo",
+        help="Show prioritized list of modules to update",
+        add_help=False,
+    )
 
     args, remaining = parser.parse_known_args(argv)
 
@@ -87,6 +92,11 @@ def main(argv: list[str] | None = None) -> int:
 
         return show_deps_main(remaining)
 
+    if args.command == "todo":
+        from update_lib.show_todo import main as show_todo_main
+
+        return show_todo_main(remaining)
+
     return 0
 
 
diff --git a/scripts/update_lib/deps.py b/scripts/update_lib/deps.py
@@ -7,6 +7,7 @@
 - Test dependencies (auto-detected from 'from test import ...')
 """
 
+import functools
 import pathlib
 
 from update_lib.io_utils import read_python_files, safe_parse_ast, safe_read_text
@@ -145,15 +146,18 @@
 }
 
 
-def get_lib_paths(name: str, cpython_prefix: str = "cpython") -> list[pathlib.Path]:
+@functools.cache
+def get_lib_paths(
+    name: str, cpython_prefix: str = "cpython"
+) -> tuple[pathlib.Path, ...]:
     """Get all library paths for a module.
 
     Args:
         name: Module name (e.g., "datetime", "libregrtest")
         cpython_prefix: CPython directory prefix
 
     Returns:
-        List of paths to copy
+        Tuple of paths to copy
     """
     dep_info = DEPENDENCIES.get(name, {})
 
@@ -168,43 +172,49 @@ def get_lib_paths(name: str, cpython_prefix: str = "cpython") -> list[pathlib.Pa
     for dep in dep_info.get("hard_deps", []):
         paths.append(construct_lib_path(cpython_prefix, dep))
 
-    return paths
+    return tuple(paths)
 
 
-def get_test_paths(name: str, cpython_prefix: str = "cpython") -> list[pathlib.Path]:
+@functools.cache
+def get_test_paths(
+    name: str, cpython_prefix: str = "cpython"
+) -> tuple[pathlib.Path, ...]:
     """Get all test paths for a module.
 
     Args:
         name: Module name (e.g., "datetime", "libregrtest")
         cpython_prefix: CPython directory prefix
 
     Returns:
-        List of test paths
+        Tuple of test paths
     """
     if name in DEPENDENCIES and "test" in DEPENDENCIES[name]:
-        return [
+        return tuple(
             construct_lib_path(cpython_prefix, p) for p in DEPENDENCIES[name]["test"]
-        ]
+        )
 
     # Default: try directory first, then file
-    return [resolve_module_path(f"test/test_{name}", cpython_prefix, prefer="dir")]
+    return (resolve_module_path(f"test/test_{name}", cpython_prefix, prefer="dir"),)
 
 
-def get_data_paths(name: str, cpython_prefix: str = "cpython") -> list[pathlib.Path]:
+@functools.cache
+def get_data_paths(
+    name: str, cpython_prefix: str = "cpython"
+) -> tuple[pathlib.Path, ...]:
     """Get additional data paths for a module.
 
     Args:
         name: Module name
         cpython_prefix: CPython directory prefix
 
     Returns:
-        List of data paths (may be empty)
+        Tuple of data paths (may be empty)
     """
     if name in DEPENDENCIES and "data" in DEPENDENCIES[name]:
-        return [
+        return tuple(
             construct_lib_path(cpython_prefix, p) for p in DEPENDENCIES[name]["data"]
-        ]
-    return []
+        )
+    return ()
 
 
 def parse_test_imports(content: str) -> set[str]:
@@ -272,15 +282,16 @@ def parse_lib_imports(content: str) -> set[str]:
     return imports
 
 
-def get_all_imports(name: str, cpython_prefix: str = "cpython") -> set[str]:
+@functools.cache
+def get_all_imports(name: str, cpython_prefix: str = "cpython") -> frozenset[str]:
     """Get all imports from a library file.
 
     Args:
         name: Module name
         cpython_prefix: CPython directory prefix
 
     Returns:
-        Set of all imported module names
+        Frozenset of all imported module names
     """
     all_imports = set()
     for lib_path in get_lib_paths(name, cpython_prefix):
@@ -290,18 +301,19 @@ def get_all_imports(name: str, cpython_prefix: str = "cpython") -> set[str]:
 
     # Remove self
     all_imports.discard(name)
-    return all_imports
+    return frozenset(all_imports)
 
 
-def get_soft_deps(name: str, cpython_prefix: str = "cpython") -> set[str]:
+@functools.cache
+def get_soft_deps(name: str, cpython_prefix: str = "cpython") -> frozenset[str]:
     """Get soft dependencies by parsing imports from library file.
 
     Args:
         name: Module name
         cpython_prefix: CPython directory prefix
 
     Returns:
-        Set of imported stdlib module names (those that exist in cpython/Lib/)
+        Frozenset of imported stdlib module names (those that exist in cpython/Lib/)
     """
     all_imports = get_all_imports(name, cpython_prefix)
 
@@ -312,22 +324,23 @@ def get_soft_deps(name: str, cpython_prefix: str = "cpython") -> set[str]:
         if module_path.exists():
             stdlib_deps.add(imp)
 
-    return stdlib_deps
+    return frozenset(stdlib_deps)
 
 
-def get_rust_deps(name: str, cpython_prefix: str = "cpython") -> set[str]:
+@functools.cache
+def get_rust_deps(name: str, cpython_prefix: str = "cpython") -> frozenset[str]:
     """Get Rust/C dependencies (imports that don't exist in cpython/Lib/).
 
     Args:
         name: Module name
         cpython_prefix: CPython directory prefix
 
     Returns:
-        Set of imported module names that are built-in or C extensions
+        Frozenset of imported module names that are built-in or C extensions
     """
     all_imports = get_all_imports(name, cpython_prefix)
     soft_deps = get_soft_deps(name, cpython_prefix)
-    return all_imports - soft_deps
+    return frozenset(all_imports - soft_deps)
 
 
 def _dircmp_is_same(dcmp) -> bool:
@@ -350,6 +363,7 @@ def _dircmp_is_same(dcmp) -> bool:
     return True
 
 
+@functools.cache
 def is_up_to_date(
     name: str, cpython_prefix: str = "cpython", lib_prefix: str = "Lib"
 ) -> bool:
@@ -472,9 +486,9 @@ def resolve_all_paths(
         Dict with "lib", "test", "data", "test_deps" keys
     """
     result = {
-        "lib": get_lib_paths(name, cpython_prefix),
-        "test": get_test_paths(name, cpython_prefix),
-        "data": get_data_paths(name, cpython_prefix),
+        "lib": list(get_lib_paths(name, cpython_prefix)),
+        "test": list(get_test_paths(name, cpython_prefix)),
+        "data": list(get_data_paths(name, cpython_prefix)),
         "test_deps": [],
     }
 
diff --git a/scripts/update_lib/show_deps.py b/scripts/update_lib/show_deps.py
@@ -169,17 +169,17 @@ def format_deps(
 
     lines = []
 
-    # lib paths
+    # lib paths (only show existing)
     lib_paths = get_lib_paths(name, cpython_prefix)
     for p in lib_paths:
-        exists = "+" if p.exists() else "-"
-        lines.append(f"[{exists}] lib: {p}")
+        if p.exists():
+            lines.append(f"[+] lib: {p}")
 
-    # test paths
+    # test paths (only show existing)
     test_paths = get_test_paths(name, cpython_prefix)
     for p in test_paths:
-        exists = "+" if p.exists() else "-"
-        lines.append(f"[{exists}] test: {p}")
+        if p.exists():
+            lines.append(f"[+] test: {p}")
 
     # hard_deps (from DEPENDENCIES table)
     dep_info = DEPENDENCIES.get(name, {})
diff --git a/scripts/update_lib/show_todo.py b/scripts/update_lib/show_todo.py
diff --git a/scripts/update_lib/tests/test_deps.py b/scripts/update_lib/tests/test_deps.py
