René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Fix forced arg format in AC-processed modules with custom converters","articleBody":"There are custom Argument Clinic converters that define `format_unit` but omit `parse_arg`. As a result, generation of positional argument parsers is forced to back up from the fastest possible `_PyArg_CheckPositional` to slower `_PyArg_ParseStack`-based format strings.\r\n\r\nHere is a list of such classes (and fixing PRs except complex cases):\r\n\r\n- [x] Modules\\\\_multiprocessing\\multiprocessing.c (gh-94517)\r\n  - HANDLE_converter\r\n\r\n- [ ] Modules\\\\_multiprocessing\\semaphore.c\r\n  - SEM_HANDLE_converter\r\n\r\n- [x] Modules\\overlapped.c (gh-94516)\r\n  - OVERLAPPED_converter\r\n  - HANDLE_converter\r\n  - ULONG_PTR_converter\r\n  - DWORD_converter\r\n  - BOOL_converter\r\n\r\n- [x] Modules\\posixmodule.c (gh-122516)\r\n  - pid_t_converter\r\n  - idtype_t_converter\r\n  - id_t_converter\r\n  - intptr_t_converter\r\n  - Py_off_t_converter\r\n\r\n- [x] Modules\\resource.c (gh-94515)\r\n  - pid_t_converter\r\n\r\n- [x] PC\\msvcrtmodule.c (gh-94514)\r\n  - HANDLE_converter\r\n\r\n- [x] PC\\winreg.c (gh-94513)\r\n  - REGSAM_converter\r\n  - DWORD_converter\r\n  - HKEY_converter\r\n\r\nAn example of such a converter:\r\n\r\n```cpp\r\nclass BOOL_converter(CConverter):\r\n    type = 'BOOL'\r\n    format_unit = 'i'\r\n\r\nclass pid_t_converter(CConverter):\r\n    type = 'pid_t'\r\n    format_unit = '\" _Py_PARSE_PID \"'\r\n```\r\n\r\nI'm going to teach all of them about low-level generation by replacing manual `format_unit` definitions with:\r\n\r\n- inheritance from a corresponding builtin converter where possible\r\n- and custom `parse_arg`s in other places.\r\n\r\nFor the example it gives:\r\n\r\n```cpp\r\nclass BOOL_converter(int_converter):\r\n    type = 'BOOL'\r\n\r\nclass pid_t_converter(CConverter):\r\n    type = 'pid_t'\r\n    # Left as a backup for potential complex cases\r\n    format_unit = '\" _Py_PARSE_PID \"'\r\n\r\n    def parse_arg(self, argname, displayname):\r\n        return \"\"\"\r\n            {paramname} = PyLong_AsPid({argname});\r\n            if ({paramname} == -1 \u0026\u0026 PyErr_Occurred()) {{{{\r\n                goto exit;\r\n            }}}}\r\n            \"\"\".format(argname=argname, paramname=self.parser_name)\r\n```\r\n","author":{"url":"https://github.com/arhadthedev","@type":"Person","name":"arhadthedev"},"datePublished":"2022-07-02T19:31:36.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/94512/cpython/issues/94512"}