gh-133583: Add support for fixed size unsigned integers in argument parsing (GH-133584)

* Add Argument Clinic converters: uint8, uint16, uint32, uint64.
* Add private C API: _PyLong_UInt8_Converter(),
  _PyLong_UInt16_Converter(), _PyLong_UInt32_Converter(),
  _PyLong_UInt64_Converter().

Include/internal/pycore_long.h

@@ -158,6 +158,11 @@ PyAPI_FUNC(int) _PyLong_UnsignedLongLong_Converter(PyObject *, void *);
 // Export for '_testclinic' shared extension (Argument Clinic code)
 PyAPI_FUNC(int) _PyLong_Size_t_Converter(PyObject *, void *);
 
+PyAPI_FUNC(int) _PyLong_UInt8_Converter(PyObject *, void *);
+PyAPI_FUNC(int) _PyLong_UInt16_Converter(PyObject *, void *);
+PyAPI_FUNC(int) _PyLong_UInt32_Converter(PyObject *, void *);
+PyAPI_FUNC(int) _PyLong_UInt64_Converter(PyObject *, void *);
+
 /* Long value tag bits:
  * 0-1: Sign bits value = (1-sign), ie. negative=2, positive=0, zero=1.
  * 2: Set to 1 for the small ints

Lib/test/test_clinic.py

@@ -2835,6 +2835,10 @@ class ClinicExternalTest(TestCase):
             "size_t",
             "slice_index",
             "str",
+            "uint16",
+            "uint32",
+            "uint64",
+            "uint8",
             "unicode",
             "unsigned_char",
             "unsigned_int",

Modules/_lzmamodule.c

@@ -17,6 +17,7 @@
 
 #include <lzma.h>
 
+#include "pycore_long.h"          // _PyLong_UInt32_Converter()
 // Blocks output buffer wrappers
 #include "pycore_blocks_output_buffer.h"
 
@@ -223,8 +224,6 @@ FUNCNAME(PyObject *obj, void *ptr)                                  \
     return 1;                                                       \
 }
 
-INT_TYPE_CONVERTER_FUNC(uint32_t, uint32_converter)
-INT_TYPE_CONVERTER_FUNC(uint64_t, uint64_converter)
 INT_TYPE_CONVERTER_FUNC(lzma_vli, lzma_vli_converter)
 INT_TYPE_CONVERTER_FUNC(lzma_mode, lzma_mode_converter)
 INT_TYPE_CONVERTER_FUNC(lzma_match_finder, lzma_mf_converter)
@@ -254,7 +253,7 @@ parse_filter_spec_lzma(_lzma_state *state, PyObject *spec)
         return NULL;
     }
     if (preset_obj != NULL) {
-        int ok = uint32_converter(preset_obj, &preset);
+        int ok = _PyLong_UInt32_Converter(preset_obj, &preset);
         Py_DECREF(preset_obj);
         if (!ok) {
             return NULL;
@@ -275,14 +274,14 @@ parse_filter_spec_lzma(_lzma_state *state, PyObject *spec)
     if (!PyArg_ParseTupleAndKeywords(state->empty_tuple, spec,
                                      "|OOO&O&O&O&O&O&O&O&", optnames,
                                      &id, &preset_obj,
-                                     uint32_converter, &options->dict_size,
-                                     uint32_converter, &options->lc,
-                                     uint32_converter, &options->lp,
-                                     uint32_converter, &options->pb,
+                                     _PyLong_UInt32_Converter, &options->dict_size,
+                                     _PyLong_UInt32_Converter, &options->lc,
+                                     _PyLong_UInt32_Converter, &options->lp,
+                                     _PyLong_UInt32_Converter, &options->pb,
                                      lzma_mode_converter, &options->mode,
-                                     uint32_converter, &options->nice_len,
+                                     _PyLong_UInt32_Converter, &options->nice_len,
                                      lzma_mf_converter, &options->mf,
-                                     uint32_converter, &options->depth)) {
+                                     _PyLong_UInt32_Converter, &options->depth)) {
         PyErr_SetString(PyExc_ValueError,
                         "Invalid filter specifier for LZMA filter");
         PyMem_Free(options);
@@ -301,7 +300,7 @@ parse_filter_spec_delta(_lzma_state *state, PyObject *spec)
     lzma_options_delta *options;
 
     if (!PyArg_ParseTupleAndKeywords(state->empty_tuple, spec, "|OO&", optnames,
-                                     &id, uint32_converter, &dist)) {
+                                     &id, _PyLong_UInt32_Converter, &dist)) {
         PyErr_SetString(PyExc_ValueError,
                         "Invalid filter specifier for delta filter");
         return NULL;
@@ -325,7 +324,7 @@ parse_filter_spec_bcj(_lzma_state *state, PyObject *spec)
     lzma_options_bcj *options;
 
     if (!PyArg_ParseTupleAndKeywords(state->empty_tuple, spec, "|OO&", optnames,
-                                     &id, uint32_converter, &start_offset)) {
+                                     &id, _PyLong_UInt32_Converter, &start_offset)) {
         PyErr_SetString(PyExc_ValueError,
                         "Invalid filter specifier for BCJ filter");
         return NULL;
@@ -806,7 +805,7 @@ Compressor_new(PyTypeObject *type, PyObject *args, PyObject *kwargs)
         return NULL;
     }
 
-    if (preset_obj != Py_None && !uint32_converter(preset_obj, &preset)) {
+    if (preset_obj != Py_None && !_PyLong_UInt32_Converter(preset_obj, &preset)) {
         return NULL;
     }
 
@@ -1226,7 +1225,7 @@ _lzma_LZMADecompressor_impl(PyTypeObject *type, int format,
                             "Cannot specify memory limit with FORMAT_RAW");
             return NULL;
         }
-        if (!uint64_converter(memlimit, &memlimit_)) {
+        if (!_PyLong_UInt64_Converter(memlimit, &memlimit_)) {
             return NULL;
         }
     }

Modules/clinic/socketmodule.c.h

@@ -6,6 +6,7 @@ preserve
 #  include "pycore_gc.h"          // PyGC_Head
 #  include "pycore_runtime.h"     // _Py_ID()
 #endif
+#include "pycore_long.h"          // _PyLong_UInt16_Converter()
 #include "pycore_modsupport.h"    // _PyArg_UnpackKeywords()
 
 PyDoc_STRVAR(_socket_socket_close__doc__,
@@ -369,4 +370,4 @@ exit:
 #ifndef _SOCKET_IF_INDEXTONAME_METHODDEF
     #define _SOCKET_IF_INDEXTONAME_METHODDEF
 #endif /* !defined(_SOCKET_IF_INDEXTONAME_METHODDEF) */
-/*[clinic end generated code: output=c971b79d2193b426 input=a9049054013a1b77]*/
+/*[clinic end generated code: output=07776dd21d1e3b56 input=a9049054013a1b77]*/

Modules/socketmodule.c

@@ -638,33 +638,22 @@ _PyLong_##NAME##_Converter(PyObject *obj, void *ptr)                \
     return 1;                                                       \
 }
 
-UNSIGNED_INT_CONVERTER(UInt16, uint16_t)
-UNSIGNED_INT_CONVERTER(UInt32, uint32_t)
-
 #if defined(HAVE_IF_NAMEINDEX) || defined(MS_WINDOWS)
 # ifdef MS_WINDOWS
     UNSIGNED_INT_CONVERTER(NetIfindex, NET_IFINDEX)
 # else
-    UNSIGNED_INT_CONVERTER(NetIfindex, unsigned int)
+#   define _PyLong_NetIfindex_Converter _PyLong_UnsignedInt_Converter
 #   define NET_IFINDEX unsigned int
 # endif
 #endif // defined(HAVE_IF_NAMEINDEX) || defined(MS_WINDOWS)
 
 /*[python input]
-class uint16_converter(CConverter):
-    type = "uint16_t"
-    converter = '_PyLong_UInt16_Converter'
-
-class uint32_converter(CConverter):
-    type = "uint32_t"
-    converter = '_PyLong_UInt32_Converter'
-
 class NET_IFINDEX_converter(CConverter):
     type = "NET_IFINDEX"
     converter = '_PyLong_NetIfindex_Converter'
 
 [python start generated code]*/
-/*[python end generated code: output=da39a3ee5e6b4b0d input=3de2e4a03fbf83b8]*/
+/*[python end generated code: output=da39a3ee5e6b4b0d input=1cf809c40a407c34]*/
 
 /*[clinic input]
 module _socket

Objects/longobject.c

@@ -1760,6 +1760,10 @@ UNSIGNED_INT_CONVERTER(UnsignedInt, unsigned int)
 UNSIGNED_INT_CONVERTER(UnsignedLong, unsigned long)
 UNSIGNED_INT_CONVERTER(UnsignedLongLong, unsigned long long)
 UNSIGNED_INT_CONVERTER(Size_t, size_t)
+UNSIGNED_INT_CONVERTER(UInt8, uint8_t)
+UNSIGNED_INT_CONVERTER(UInt16, uint16_t)
+UNSIGNED_INT_CONVERTER(UInt32, uint32_t)
+UNSIGNED_INT_CONVERTER(UInt64, uint64_t)
 
 
 #define CHECK_BINOP(v,w)                                \

Tools/clinic/libclinic/converters.py

@@ -17,6 +17,54 @@ from libclinic.converter import (
 TypeSet = set[bltns.type[object]]
 
 
+class BaseUnsignedIntConverter(CConverter):
+
+    def use_converter(self) -> None:
+        if self.converter:
+            self.add_include('pycore_long.h',
+                             f'{self.converter}()')
+
+    def parse_arg(self, argname: str, displayname: str, *, limited_capi: bool) -> str | None:
+        if not limited_capi:
+            return super().parse_arg(argname, displayname, limited_capi=limited_capi)
+        return self.format_code("""
+            {{{{
+                Py_ssize_t _bytes = PyLong_AsNativeBytes({argname}, &{paramname}, sizeof({type}),
+                        Py_ASNATIVEBYTES_NATIVE_ENDIAN |
+                        Py_ASNATIVEBYTES_ALLOW_INDEX |
+                        Py_ASNATIVEBYTES_REJECT_NEGATIVE |
+                        Py_ASNATIVEBYTES_UNSIGNED_BUFFER);
+                if (_bytes < 0) {{{{
+                    goto exit;
+                }}}}
+                if ((size_t)_bytes > sizeof({type})) {{{{
+                    PyErr_SetString(PyExc_OverflowError,
+                                    "Python int too large for C {type}");
+                    goto exit;
+                }}}}
+            }}}}
+            """,
+            argname=argname,
+            type=self.type)
+
+
+class uint8_converter(BaseUnsignedIntConverter):
+    type = "uint8_t"
+    converter = '_PyLong_UInt8_Converter'
+
+class uint16_converter(BaseUnsignedIntConverter):
+    type = "uint16_t"
+    converter = '_PyLong_UInt16_Converter'
+
+class uint32_converter(BaseUnsignedIntConverter):
+    type = "uint32_t"
+    converter = '_PyLong_UInt32_Converter'
+
+class uint64_converter(BaseUnsignedIntConverter):
+    type = "uint64_t"
+    converter = '_PyLong_UInt64_Converter'
+
+
 class bool_converter(CConverter):
     type = 'int'
     default_type = bool
@@ -211,29 +259,7 @@ class short_converter(CConverter):
         return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
-def format_inline_unsigned_int_converter(self: CConverter, argname: str) -> str:
-    return self.format_code("""
-        {{{{
-            Py_ssize_t _bytes = PyLong_AsNativeBytes({argname}, &{paramname}, sizeof({type}),
-                    Py_ASNATIVEBYTES_NATIVE_ENDIAN |
-                    Py_ASNATIVEBYTES_ALLOW_INDEX |
-                    Py_ASNATIVEBYTES_REJECT_NEGATIVE |
-                    Py_ASNATIVEBYTES_UNSIGNED_BUFFER);
-            if (_bytes < 0) {{{{
-                goto exit;
-            }}}}
-            if ((size_t)_bytes > sizeof({type})) {{{{
-                PyErr_SetString(PyExc_OverflowError,
-                                "Python int too large for C {type}");
-                goto exit;
-            }}}}
-        }}}}
-        """,
-        argname=argname,
-        type=self.type)
-
-
-class unsigned_short_converter(CConverter):
+class unsigned_short_converter(BaseUnsignedIntConverter):
     type = 'unsigned short'
     default_type = int
     c_ignored_default = "0"
@@ -244,11 +270,6 @@ class unsigned_short_converter(CConverter):
         else:
             self.converter = '_PyLong_UnsignedShort_Converter'
 
-    def use_converter(self) -> None:
-        if self.converter == '_PyLong_UnsignedShort_Converter':
-            self.add_include('pycore_long.h',
-                             '_PyLong_UnsignedShort_Converter()')
-
     def parse_arg(self, argname: str, displayname: str, *, limited_capi: bool) -> str | None:
         if self.format_unit == 'H':
             return self.format_code("""
@@ -258,9 +279,7 @@ class unsigned_short_converter(CConverter):
                 }}}}
                 """,
                 argname=argname)
-        if not limited_capi:
-            return super().parse_arg(argname, displayname, limited_capi=limited_capi)
-        return format_inline_unsigned_int_converter(self, argname)
+        return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
 @add_legacy_c_converter('C', accept={str})
@@ -311,7 +330,7 @@ class int_converter(CConverter):
         return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
-class unsigned_int_converter(CConverter):
+class unsigned_int_converter(BaseUnsignedIntConverter):
     type = 'unsigned int'
     default_type = int
     c_ignored_default = "0"
@@ -322,11 +341,6 @@ class unsigned_int_converter(CConverter):
         else:
             self.converter = '_PyLong_UnsignedInt_Converter'
 
-    def use_converter(self) -> None:
-        if self.converter == '_PyLong_UnsignedInt_Converter':
-            self.add_include('pycore_long.h',
-                             '_PyLong_UnsignedInt_Converter()')
-
     def parse_arg(self, argname: str, displayname: str, *, limited_capi: bool) -> str | None:
         if self.format_unit == 'I':
             return self.format_code("""
@@ -336,9 +350,7 @@ class unsigned_int_converter(CConverter):
                 }}}}
                 """,
                 argname=argname)
-        if not limited_capi:
-            return super().parse_arg(argname, displayname, limited_capi=limited_capi)
-        return format_inline_unsigned_int_converter(self, argname)
+        return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
 class long_converter(CConverter):
@@ -359,7 +371,7 @@ class long_converter(CConverter):
         return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
-class unsigned_long_converter(CConverter):
+class unsigned_long_converter(BaseUnsignedIntConverter):
     type = 'unsigned long'
     default_type = int
     c_ignored_default = "0"
@@ -370,11 +382,6 @@ class unsigned_long_converter(CConverter):
         else:
             self.converter = '_PyLong_UnsignedLong_Converter'
 
-    def use_converter(self) -> None:
-        if self.converter == '_PyLong_UnsignedLong_Converter':
-            self.add_include('pycore_long.h',
-                             '_PyLong_UnsignedLong_Converter()')
-
     def parse_arg(self, argname: str, displayname: str, *, limited_capi: bool) -> str | None:
         if self.format_unit == 'k':
             return self.format_code("""
@@ -387,9 +394,7 @@ class unsigned_long_converter(CConverter):
                 argname=argname,
                 bad_argument=self.bad_argument(displayname, 'int', limited_capi=limited_capi),
             )
-        if not limited_capi:
-            return super().parse_arg(argname, displayname, limited_capi=limited_capi)
-        return format_inline_unsigned_int_converter(self, argname)
+        return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
 class long_long_converter(CConverter):
@@ -410,7 +415,7 @@ class long_long_converter(CConverter):
         return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
-class unsigned_long_long_converter(CConverter):
+class unsigned_long_long_converter(BaseUnsignedIntConverter):
     type = 'unsigned long long'
     default_type = int
     c_ignored_default = "0"
@@ -421,11 +426,6 @@ class unsigned_long_long_converter(CConverter):
         else:
             self.converter = '_PyLong_UnsignedLongLong_Converter'
 
-    def use_converter(self) -> None:
-        if self.converter == '_PyLong_UnsignedLongLong_Converter':
-            self.add_include('pycore_long.h',
-                             '_PyLong_UnsignedLongLong_Converter()')
-
     def parse_arg(self, argname: str, displayname: str, *, limited_capi: bool) -> str | None:
         if self.format_unit == 'K':
             return self.format_code("""
@@ -438,9 +438,7 @@ class unsigned_long_long_converter(CConverter):
                 argname=argname,
                 bad_argument=self.bad_argument(displayname, 'int', limited_capi=limited_capi),
             )
-        if not limited_capi:
-            return super().parse_arg(argname, displayname, limited_capi=limited_capi)
-        return format_inline_unsigned_int_converter(self, argname)
+        return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
 class Py_ssize_t_converter(CConverter):
@@ -557,15 +555,11 @@ class slice_index_converter(CConverter):
                 argname=argname)
 
 
-class size_t_converter(CConverter):
+class size_t_converter(BaseUnsignedIntConverter):
     type = 'size_t'
     converter = '_PyLong_Size_t_Converter'
     c_ignored_default = "0"
 
-    def use_converter(self) -> None:
-        self.add_include('pycore_long.h',
-                         '_PyLong_Size_t_Converter()')
-
     def parse_arg(self, argname: str, displayname: str, *, limited_capi: bool) -> str | None:
         if self.format_unit == 'n':
             return self.format_code("""
@@ -575,9 +569,7 @@ class size_t_converter(CConverter):
                 }}}}
                 """,
                 argname=argname)
-        if not limited_capi:
-            return super().parse_arg(argname, displayname, limited_capi=limited_capi)
-        return format_inline_unsigned_int_converter(self, argname)
+        return super().parse_arg(argname, displayname, limited_capi=limited_capi)
 
 
 class fildes_converter(CConverter):