pythongh-117431: Adapt bytes and bytearray .find() and friends to Argument Clinic (python#117502)

This change gives a significant speedup, as the METH_FASTCALL calling
convention is now used. The following bytes and bytearray methods are adapted:

- count()
- find()
- index()
- rfind()
- rindex()

Co-authored-by: Inada Naoki <[email protected]>

Author: erlend-aasland

Include/internal/pycore_bytes_methods.h

@@ -26,11 +26,16 @@ extern void _Py_bytes_title(char *result, const char *s, Py_ssize_t len);
 extern void _Py_bytes_capitalize(char *result, const char *s, Py_ssize_t len);
 extern void _Py_bytes_swapcase(char *result, const char *s, Py_ssize_t len);
 
-extern PyObject *_Py_bytes_find(const char *str, Py_ssize_t len, PyObject *args);
-extern PyObject *_Py_bytes_index(const char *str, Py_ssize_t len, PyObject *args);
-extern PyObject *_Py_bytes_rfind(const char *str, Py_ssize_t len, PyObject *args);
-extern PyObject *_Py_bytes_rindex(const char *str, Py_ssize_t len, PyObject *args);
-extern PyObject *_Py_bytes_count(const char *str, Py_ssize_t len, PyObject *args);
+extern PyObject *_Py_bytes_find(const char *str, Py_ssize_t len, PyObject *sub,
+                                Py_ssize_t start, Py_ssize_t end);
+extern PyObject *_Py_bytes_index(const char *str, Py_ssize_t len, PyObject *sub,
+                                 Py_ssize_t start, Py_ssize_t end);
+extern PyObject *_Py_bytes_rfind(const char *str, Py_ssize_t len, PyObject *sub,
+                                 Py_ssize_t start, Py_ssize_t end);
+extern PyObject *_Py_bytes_rindex(const char *str, Py_ssize_t len, PyObject *sub,
+                                 Py_ssize_t start, Py_ssize_t end);
+extern PyObject *_Py_bytes_count(const char *str, Py_ssize_t len, PyObject *sub,
+                                 Py_ssize_t start, Py_ssize_t end);
 extern int _Py_bytes_contains(const char *str, Py_ssize_t len, PyObject *arg);
 extern PyObject *_Py_bytes_startswith(const char *str, Py_ssize_t len,
                                       PyObject *subobj, Py_ssize_t start,

Misc/NEWS.d/next/Core and Builtins/2024-04-12-09-09-11.gh-issue-117431.lxFEeJ.rst

@@ -0,0 +1,9 @@
+Improve the performance of the following :class:`bytes` and
+:class:`bytearray` methods by adapting them to the :c:macro:`METH_FASTCALL`
+calling convention:
+
+* :meth:`!count`
+* :meth:`!find`
+* :meth:`!index`
+* :meth:`!rfind`
+* :meth:`!rindex`

Objects/bytearrayobject.c

@@ -1121,16 +1121,44 @@ bytearray_dealloc(PyByteArrayObject *self)
 #include "stringlib/transmogrify.h"
 
 
+/*[clinic input]
+@text_signature "($self, sub[, start[, end]], /)"
+bytearray.find
+
+    sub: object
+    start: slice_index(accept={int, NoneType}, c_default='0') = None
+         Optional start position. Default: start of the bytes.
+    end: slice_index(accept={int, NoneType}, c_default='PY_SSIZE_T_MAX') = None
+         Optional stop position. Default: end of the bytes.
+    /
+
+Return the lowest index in B where subsection 'sub' is found, such that 'sub' is contained within B[start:end].
+
+Return -1 on failure.
+[clinic start generated code]*/
+
 static PyObject *
-bytearray_find(PyByteArrayObject *self, PyObject *args)
+bytearray_find_impl(PyByteArrayObject *self, PyObject *sub, Py_ssize_t start,
+                    Py_ssize_t end)
+/*[clinic end generated code: output=413e1cab2ae87da0 input=793dfad803e2952f]*/
 {
-    return _Py_bytes_find(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), args);
+    return _Py_bytes_find(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
+                          sub, start, end);
 }
 
+/*[clinic input]
+bytearray.count = bytearray.find
+
+Return the number of non-overlapping occurrences of subsection 'sub' in bytes B[start:end].
+[clinic start generated code]*/
+
 static PyObject *
-bytearray_count(PyByteArrayObject *self, PyObject *args)
+bytearray_count_impl(PyByteArrayObject *self, PyObject *sub,
+                     Py_ssize_t start, Py_ssize_t end)
+/*[clinic end generated code: output=a21ee2692e4f1233 input=4deb529db38deda8]*/
 {
-    return _Py_bytes_count(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), args);
+    return _Py_bytes_count(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
+                           sub, start, end);
 }
 
 /*[clinic input]
@@ -1162,22 +1190,55 @@ bytearray_copy_impl(PyByteArrayObject *self)
                                          PyByteArray_GET_SIZE(self));
 }
 
+/*[clinic input]
+bytearray.index = bytearray.find
+
+Return the lowest index in B where subsection 'sub' is found, such that 'sub' is contained within B[start:end].
+
+Raise ValueError if the subsection is not found.
+[clinic start generated code]*/
+
 static PyObject *
-bytearray_index(PyByteArrayObject *self, PyObject *args)
+bytearray_index_impl(PyByteArrayObject *self, PyObject *sub,
+                     Py_ssize_t start, Py_ssize_t end)
+/*[clinic end generated code: output=067a1e78efc672a7 input=8cbaf6836dbd2a9a]*/
 {
-    return _Py_bytes_index(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), args);
+    return _Py_bytes_index(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
+                           sub, start, end);
 }
 
+/*[clinic input]
+bytearray.rfind = bytearray.find
+
+Return the highest index in B where subsection 'sub' is found, such that 'sub' is contained within B[start:end].
+
+Return -1 on failure.
+[clinic start generated code]*/
+
 static PyObject *
-bytearray_rfind(PyByteArrayObject *self, PyObject *args)
+bytearray_rfind_impl(PyByteArrayObject *self, PyObject *sub,
+                     Py_ssize_t start, Py_ssize_t end)
+/*[clinic end generated code: output=51bf886f932b283c input=eaa107468a158423]*/
 {
-    return _Py_bytes_rfind(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), args);
+    return _Py_bytes_rfind(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
+                           sub, start, end);
 }
 
+/*[clinic input]
+bytearray.rindex = bytearray.find
+
+Return the highest index in B where subsection 'sub' is found, such that 'sub' is contained within B[start:end].
+
+Raise ValueError if the subsection is not found.
+[clinic start generated code]*/
+
 static PyObject *
-bytearray_rindex(PyByteArrayObject *self, PyObject *args)
+bytearray_rindex_impl(PyByteArrayObject *self, PyObject *sub,
+                      Py_ssize_t start, Py_ssize_t end)
+/*[clinic end generated code: output=38e1cf66bafb08b9 input=81cf49d0af4d5bd0]*/
 {
-    return _Py_bytes_rindex(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self), args);
+    return _Py_bytes_rindex(PyByteArray_AS_STRING(self), PyByteArray_GET_SIZE(self),
+                            sub, start, end);
 }
 
 static int
@@ -2236,17 +2297,15 @@ bytearray_methods[] = {
     STRINGLIB_CENTER_METHODDEF
     BYTEARRAY_CLEAR_METHODDEF
     BYTEARRAY_COPY_METHODDEF
-    {"count", (PyCFunction)bytearray_count, METH_VARARGS,
-     _Py_count__doc__},
+    BYTEARRAY_COUNT_METHODDEF
     BYTEARRAY_DECODE_METHODDEF
     BYTEARRAY_ENDSWITH_METHODDEF
     STRINGLIB_EXPANDTABS_METHODDEF
     BYTEARRAY_EXTEND_METHODDEF
-    {"find", (PyCFunction)bytearray_find, METH_VARARGS,
-     _Py_find__doc__},
+    BYTEARRAY_FIND_METHODDEF
     BYTEARRAY_FROMHEX_METHODDEF
     BYTEARRAY_HEX_METHODDEF
-    {"index", (PyCFunction)bytearray_index, METH_VARARGS, _Py_index__doc__},
+    BYTEARRAY_INDEX_METHODDEF
     BYTEARRAY_INSERT_METHODDEF
     {"isalnum", stringlib_isalnum, METH_NOARGS,
      _Py_isalnum__doc__},
@@ -2276,8 +2335,8 @@ bytearray_methods[] = {
     BYTEARRAY_REMOVEPREFIX_METHODDEF
     BYTEARRAY_REMOVESUFFIX_METHODDEF
     BYTEARRAY_REVERSE_METHODDEF
-    {"rfind", (PyCFunction)bytearray_rfind, METH_VARARGS, _Py_rfind__doc__},
-    {"rindex", (PyCFunction)bytearray_rindex, METH_VARARGS, _Py_rindex__doc__},
+    BYTEARRAY_RFIND_METHODDEF
+    BYTEARRAY_RINDEX_METHODDEF
     STRINGLIB_RJUST_METHODDEF
     BYTEARRAY_RPARTITION_METHODDEF
     BYTEARRAY_RSPLIT_METHODDEF

Objects/bytes_methods.c

@@ -453,31 +453,21 @@ stringlib_parse_args_finds().
 */
 
 Py_LOCAL_INLINE(int)
-parse_args_finds_byte(const char *function_name, PyObject *args,
-                      PyObject **subobj, char *byte,
-                      Py_ssize_t *start, Py_ssize_t *end)
+parse_args_finds_byte(const char *function_name, PyObject **subobj, char *byte)
 {
-    PyObject *tmp_subobj;
-    Py_ssize_t ival;
-
-    if(!stringlib_parse_args_finds(function_name, args, &tmp_subobj,
-                                   start, end))
-        return 0;
-
-    if (PyObject_CheckBuffer(tmp_subobj)) {
-        *subobj = tmp_subobj;
+    if (PyObject_CheckBuffer(*subobj)) {
         return 1;
     }
 
-    if (!_PyIndex_Check(tmp_subobj)) {
+    if (!_PyIndex_Check(*subobj)) {
         PyErr_Format(PyExc_TypeError,
                      "argument should be integer or bytes-like object, "
                      "not '%.200s'",
-                     Py_TYPE(tmp_subobj)->tp_name);
+                     Py_TYPE(*subobj)->tp_name);
         return 0;
     }
 
-    ival = PyNumber_AsSsize_t(tmp_subobj, NULL);
+    Py_ssize_t ival = PyNumber_AsSsize_t(*subobj, NULL);
     if (ival == -1 && PyErr_Occurred()) {
         return 0;
     }
@@ -508,19 +498,19 @@ parse_args_finds_byte(const char *function_name, PyObject *args,
 
 Py_LOCAL_INLINE(Py_ssize_t)
 find_internal(const char *str, Py_ssize_t len,
-              const char *function_name, PyObject *args, int dir)
+              const char *function_name, PyObject *subobj,
+              Py_ssize_t start, Py_ssize_t end,
+              int dir)
 {
-    PyObject *subobj;
     char byte;
     Py_buffer subbuf;
     const char *sub;
     Py_ssize_t sub_len;
-    Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
     Py_ssize_t res;
 
-    if (!parse_args_finds_byte(function_name, args,
-                               &subobj, &byte, &start, &end))
+    if (!parse_args_finds_byte(function_name, &subobj, &byte)) {
         return -2;
+    }
 
     if (subobj) {
         if (PyObject_GetBuffer(subobj, &subbuf, PyBUF_SIMPLE) != 0)
@@ -566,37 +556,21 @@ find_internal(const char *str, Py_ssize_t len,
     return res;
 }
 
-PyDoc_STRVAR_shared(_Py_find__doc__,
-"B.find(sub[, start[, end]]) -> int\n\
-\n\
-Return the lowest index in B where subsection sub is found,\n\
-such that sub is contained within B[start,end].  Optional\n\
-arguments start and end are interpreted as in slice notation.\n\
-\n\
-Return -1 on failure.");
-
 PyObject *
-_Py_bytes_find(const char *str, Py_ssize_t len, PyObject *args)
+_Py_bytes_find(const char *str, Py_ssize_t len, PyObject *sub,
+               Py_ssize_t start, Py_ssize_t end)
 {
-    Py_ssize_t result = find_internal(str, len, "find", args, +1);
+    Py_ssize_t result = find_internal(str, len, "find", sub, start, end, +1);
     if (result == -2)
         return NULL;
     return PyLong_FromSsize_t(result);
 }
 
-PyDoc_STRVAR_shared(_Py_index__doc__,
-"B.index(sub[, start[, end]]) -> int\n\
-\n\
-Return the lowest index in B where subsection sub is found,\n\
-such that sub is contained within B[start,end].  Optional\n\
-arguments start and end are interpreted as in slice notation.\n\
-\n\
-Raises ValueError when the subsection is not found.");
-
 PyObject *
-_Py_bytes_index(const char *str, Py_ssize_t len, PyObject *args)
+_Py_bytes_index(const char *str, Py_ssize_t len, PyObject *sub,
+                Py_ssize_t start, Py_ssize_t end)
 {
-    Py_ssize_t result = find_internal(str, len, "index", args, +1);
+    Py_ssize_t result = find_internal(str, len, "index", sub, start, end, +1);
     if (result == -2)
         return NULL;
     if (result == -1) {
@@ -607,37 +581,21 @@ _Py_bytes_index(const char *str, Py_ssize_t len, PyObject *args)
     return PyLong_FromSsize_t(result);
 }
 
-PyDoc_STRVAR_shared(_Py_rfind__doc__,
-"B.rfind(sub[, start[, end]]) -> int\n\
-\n\
-Return the highest index in B where subsection sub is found,\n\
-such that sub is contained within B[start,end].  Optional\n\
-arguments start and end are interpreted as in slice notation.\n\
-\n\
-Return -1 on failure.");
-
 PyObject *
-_Py_bytes_rfind(const char *str, Py_ssize_t len, PyObject *args)
+_Py_bytes_rfind(const char *str, Py_ssize_t len, PyObject *sub,
+                Py_ssize_t start, Py_ssize_t end)
 {
-    Py_ssize_t result = find_internal(str, len, "rfind", args, -1);
+    Py_ssize_t result = find_internal(str, len, "rfind", sub, start, end, -1);
     if (result == -2)
         return NULL;
     return PyLong_FromSsize_t(result);
 }
 
-PyDoc_STRVAR_shared(_Py_rindex__doc__,
-"B.rindex(sub[, start[, end]]) -> int\n\
-\n\
-Return the highest index in B where subsection sub is found,\n\
-such that sub is contained within B[start,end].  Optional\n\
-arguments start and end are interpreted as in slice notation.\n\
-\n\
-Raise ValueError when the subsection is not found.");
-
 PyObject *
-_Py_bytes_rindex(const char *str, Py_ssize_t len, PyObject *args)
+_Py_bytes_rindex(const char *str, Py_ssize_t len, PyObject *sub,
+                 Py_ssize_t start, Py_ssize_t end)
 {
-    Py_ssize_t result = find_internal(str, len, "rindex", args, -1);
+    Py_ssize_t result = find_internal(str, len, "rindex", sub, start, end, -1);
     if (result == -2)
         return NULL;
     if (result == -1) {
@@ -648,28 +606,20 @@ _Py_bytes_rindex(const char *str, Py_ssize_t len, PyObject *args)
     return PyLong_FromSsize_t(result);
 }
 
-PyDoc_STRVAR_shared(_Py_count__doc__,
-"B.count(sub[, start[, end]]) -> int\n\
-\n\
-Return the number of non-overlapping occurrences of subsection sub in\n\
-bytes B[start:end].  Optional arguments start and end are interpreted\n\
-as in slice notation.");
-
 PyObject *
-_Py_bytes_count(const char *str, Py_ssize_t len, PyObject *args)
+_Py_bytes_count(const char *str, Py_ssize_t len, PyObject *sub_obj,
+                Py_ssize_t start, Py_ssize_t end)
 {
-    PyObject *sub_obj;
     const char *sub;
     Py_ssize_t sub_len;
     char byte;
-    Py_ssize_t start = 0, end = PY_SSIZE_T_MAX;
 
     Py_buffer vsub;
     PyObject *count_obj;
 
-    if (!parse_args_finds_byte("count", args,
-                               &sub_obj, &byte, &start, &end))
+    if (!parse_args_finds_byte("count", &sub_obj, &byte)) {
         return NULL;
+    }
 
     if (sub_obj) {
         if (PyObject_GetBuffer(sub_obj, &vsub, PyBUF_SIMPLE) != 0)