gh-110964: Remove private _PyArg functions

[vstinner]

Move the following private functions and structures to pycore_modsupport.h internal C API:

• _PyArg_BadArgument()
• _PyArg_CheckPositional()
• _PyArg_NoKeywords()
• _PyArg_NoPositional()
• _PyArg_ParseStack()
• _PyArg_ParseStackAndKeywords()
• _PyArg_Parser structure
• _PyArg_UnpackKeywords()
• _PyArg_UnpackKeywordsWithVararg()
• _PyArg_UnpackStack()
• _Py_ANY_VARARGS()

Changes:

• Python/getargs.h now includes pycore_modsupport.h to export functions.

• clinic.py now adds pycore_modsupport.h when one of these functions is used.

• Add pycore_modsupport.h includes when a C extension uses one of these functions.

• Define Py_BUILD_CORE_MODULE in C extensions which now include directly or indirectly (via code generated by Argument Clinic) pycore_modsupport.h:

  • _csv
  • _curses_panel
  • _dbm
  • _gdbm
  • _multiprocessing.posixshmem
  • _sqlite.row
  • _statistics
  • grp
  • resource
  • syslog

• _testcapi: bad_get() no longer uses METH_FASTCALL calling convention but METH_VARARGS. Replace _PyArg_UnpackStack() with PyArg_ParseTuple().

• _testcapi: add PYTESTCAPI_NEED_INTERNAL_API macro which is defined by _testcapi sub-modules which need the internal C API (pycore_modsupport.h): exceptions.c, float.c, vectorcall.c, watchers.c.

• Remove Include/cpython/modsupport.h header file. Include/modsupport.h no longer includes the removed header file.

• Issue: [C API] Move undocumented private _PyArg C API to pycore_modsupport.h internal C API #110964

[vstinner]

This change is big because it changes 122 files generated by Argument Clinic:

$ git show --stat|grep 'clinic/.*.c.h'|wc -l
122

I prefer to write a single PR to move all private _PyArg functions at once, because each moved function touch many files generated by AC.

[AlexWaygood]

The AC changes look fine to me! (I'll leave the question of whether the changes are worth making to those more skilled in C.)

[vstinner]

For clinic.py, I was lazy and used the clinic global variable in bad_argument(). The correct fix is to pass a clinic argument to add bad_argument() methods and calling sites. Problem: there are many. Second problem: I wrote such change many times, but I lost my work since the overall change was blocked for different reasons. Also, I tried to keep this PR as small as possible.

In short, clinic.py should be reworked later, once this change lands.

[AlexWaygood]

or perhaps this?

Suggested change

	if clinic is not None:
	clinic.add_include('pycore_modsupport.h', '_PyArg_BadArgument()')
	assert clinic is not None
	clinic.add_include('pycore_modsupport.h', '_PyArg_BadArgument()')

[vstinner]

As explained in my previous comment, I plan to write a follow-up for this code. We should not use the global variable, but pass an argument which cannot be None.

[AlexWaygood]

Yes, I agree that's the more principled action in the longer term, and I'm fine with using the easier solution for now. But this is the same number of lines, and I think makes it clearer that we never expect clinic to be None here (if it is None, something has gone horribly wrong :-)

[vstinner]

PR to refactor this code, to avoid if clinic is not None: PR #110982.

[vstinner]

For clinic.py, I was lazy and used the clinic global variable in bad_argument().

I wrote PR #110984 for that.

[vstinner]

I merged my PR. @AlexWaygood: thanks for reviewing clinic.py changes.