From 128d6241176a879ea9b6e34ea67f1d113d22a1b8 Mon Sep 17 00:00:00 2001 From: Victor Stinner Date: Thu, 21 Apr 2022 21:53:18 +0200 Subject: [PATCH] gh-89653: PEP 670: Convert unicodeobject.h macros to functions (#91773) Convert unicodeobject.h macros to static inline functions: * PyUnicode_MAX_CHAR_VALUE() * PyUnicode_READ() * PyUnicode_READY() * PyUnicode_READ_CHAR() * PyUnicode_WRITE() Move PyUnicode_READY() after _PyUnicode_Ready(), since it uses _PyUnicode_Ready(). Static inline functions are wrapped by macros which casts arguments with _PyObject_CAST() and casts 'kind' arguments to "unsigned int" to prevent introducing new compiler warnings when passing "const PyObject*". --- Include/cpython/unicodeobject.h | 131 ++++++++++++++++++-------------- 1 file changed, 74 insertions(+), 57 deletions(-) diff --git a/Include/cpython/unicodeobject.h b/Include/cpython/unicodeobject.h index 9a993a1c7ff..c1aebce9012 100644 --- a/Include/cpython/unicodeobject.h +++ b/Include/cpython/unicodeobject.h @@ -45,7 +45,7 @@ #define Py_UNICODE_ISALPHA(ch) _PyUnicode_IsAlpha(ch) #define Py_UNICODE_ISALNUM(ch) \ - (Py_UNICODE_ISALPHA(ch) || \ + (Py_UNICODE_ISALPHA(ch) || \ Py_UNICODE_ISDECIMAL(ch) || \ Py_UNICODE_ISDIGIT(ch) || \ Py_UNICODE_ISNUMERIC(ch)) @@ -379,80 +379,84 @@ static inline Py_ssize_t PyUnicode_GET_LENGTH(PyObject *op) { } #define PyUnicode_GET_LENGTH(op) PyUnicode_GET_LENGTH(_PyObject_CAST(op)) -/* In the access macros below, "kind" may be evaluated more than once. - All other macro parameters are evaluated exactly once, so it is safe - to put side effects into them (such as increasing the index). */ - -/* Write into the canonical representation, this macro does not do any sanity +/* Write into the canonical representation, this function does not do any sanity checks and is intended for usage in loops. The caller should cache the - kind and data pointers obtained from other macro calls. + kind and data pointers obtained from other function calls. index is the index in the string (starts at 0) and value is the new code point value which should be written to that location. */ +static inline void PyUnicode_WRITE(unsigned int kind, void *data, + Py_ssize_t index, Py_UCS4 value) +{ + if (kind == PyUnicode_1BYTE_KIND) { + ((Py_UCS1 *)data)[index] = (Py_UCS1)value; + } + else if (kind == PyUnicode_2BYTE_KIND) { + ((Py_UCS2 *)data)[index] = (Py_UCS2)value; + } + else { + assert(kind == PyUnicode_4BYTE_KIND); + ((Py_UCS4 *)data)[index] = value; + } +} #define PyUnicode_WRITE(kind, data, index, value) \ - do { \ - switch ((kind)) { \ - case PyUnicode_1BYTE_KIND: { \ - ((Py_UCS1 *)(data))[(index)] = (Py_UCS1)(value); \ - break; \ - } \ - case PyUnicode_2BYTE_KIND: { \ - ((Py_UCS2 *)(data))[(index)] = (Py_UCS2)(value); \ - break; \ - } \ - default: { \ - assert((kind) == PyUnicode_4BYTE_KIND); \ - ((Py_UCS4 *)(data))[(index)] = (Py_UCS4)(value); \ - } \ - } \ - } while (0) + PyUnicode_WRITE((unsigned int)(kind), (void*)(data), (index), (Py_UCS4)(value)) /* Read a code point from the string's canonical representation. No checks or ready calls are performed. */ +static inline Py_UCS4 PyUnicode_READ(unsigned int kind, + const void *data, Py_ssize_t index) +{ + if (kind == PyUnicode_1BYTE_KIND) { + return ((const Py_UCS1 *)data)[index]; + } + if (kind == PyUnicode_2BYTE_KIND) { + return ((const Py_UCS2 *)data)[index]; + } + return ((const Py_UCS4 *)data)[index]; +} #define PyUnicode_READ(kind, data, index) \ - ((Py_UCS4) \ - ((kind) == PyUnicode_1BYTE_KIND ? \ - ((const Py_UCS1 *)(data))[(index)] : \ - ((kind) == PyUnicode_2BYTE_KIND ? \ - ((const Py_UCS2 *)(data))[(index)] : \ - ((const Py_UCS4 *)(data))[(index)] \ - ) \ - )) + PyUnicode_READ((unsigned int)(kind), (const void*)(data), (index)) /* PyUnicode_READ_CHAR() is less efficient than PyUnicode_READ() because it calls PyUnicode_KIND() and might call it twice. For single reads, use PyUnicode_READ_CHAR, for multiple consecutive reads callers should cache kind and use PyUnicode_READ instead. */ +static inline Py_UCS4 PyUnicode_READ_CHAR(PyObject *unicode, Py_ssize_t index) +{ + assert(PyUnicode_IS_READY(unicode)); + unsigned int kind = PyUnicode_KIND(unicode); + if (kind == PyUnicode_1BYTE_KIND) { + return PyUnicode_1BYTE_DATA(unicode)[index]; + } + if (kind == PyUnicode_2BYTE_KIND) { + return PyUnicode_2BYTE_DATA(unicode)[index]; + } + return PyUnicode_4BYTE_DATA(unicode)[index]; +} #define PyUnicode_READ_CHAR(unicode, index) \ - (assert(PyUnicode_IS_READY(unicode)), \ - (Py_UCS4) \ - (PyUnicode_KIND((unicode)) == PyUnicode_1BYTE_KIND ? \ - ((const Py_UCS1 *)(PyUnicode_DATA((unicode))))[(index)] : \ - (PyUnicode_KIND((unicode)) == PyUnicode_2BYTE_KIND ? \ - ((const Py_UCS2 *)(PyUnicode_DATA((unicode))))[(index)] : \ - ((const Py_UCS4 *)(PyUnicode_DATA((unicode))))[(index)] \ - ) \ - )) - -/* PyUnicode_READY() does less work than _PyUnicode_Ready() in the best - case. If the canonical representation is not yet set, it will still call - _PyUnicode_Ready(). - Returns 0 on success and -1 on errors. */ -#define PyUnicode_READY(op) \ - ((PyUnicode_IS_READY(op) ? \ - 0 : _PyUnicode_Ready(_PyObject_CAST(op)))) + PyUnicode_READ_CHAR(_PyObject_CAST(unicode), (index)) /* Return a maximum character value which is suitable for creating another string based on op. This is always an approximation but more efficient than iterating over the string. */ +static inline Py_UCS4 PyUnicode_MAX_CHAR_VALUE(PyObject *op) +{ + assert(PyUnicode_IS_READY(op)); + if (PyUnicode_IS_ASCII(op)) { + return 0x7fU; + } + + unsigned int kind = PyUnicode_KIND(op); + if (kind == PyUnicode_1BYTE_KIND) { + return 0xffU; + } + if (kind == PyUnicode_2BYTE_KIND) { + return 0xffffU; + } + return 0x10ffffU; +} #define PyUnicode_MAX_CHAR_VALUE(op) \ - (assert(PyUnicode_IS_READY(op)), \ - (PyUnicode_IS_ASCII(op) ? \ - (0x7f) : \ - (PyUnicode_KIND(op) == PyUnicode_1BYTE_KIND ? \ - (0xffU) : \ - (PyUnicode_KIND(op) == PyUnicode_2BYTE_KIND ? \ - (0xffffU) : \ - (0x10ffffU))))) + PyUnicode_MAX_CHAR_VALUE(_PyObject_CAST(op)) Py_DEPRECATED(3.3) static inline Py_ssize_t PyUnicode_WSTR_LENGTH(PyObject *op) { @@ -479,12 +483,25 @@ PyAPI_FUNC(PyObject*) PyUnicode_New( objects which were created using the old API to the new flexible format introduced with PEP 393. - Don't call this function directly, use the public PyUnicode_READY() macro + Don't call this function directly, use the public PyUnicode_READY() function instead. */ PyAPI_FUNC(int) _PyUnicode_Ready( PyObject *unicode /* Unicode object */ ); +/* PyUnicode_READY() does less work than _PyUnicode_Ready() in the best + case. If the canonical representation is not yet set, it will still call + _PyUnicode_Ready(). + Returns 0 on success and -1 on errors. */ +static inline int PyUnicode_READY(PyObject *op) +{ + if (PyUnicode_IS_READY(op)) { + return 0; + } + return _PyUnicode_Ready(op); +} +#define PyUnicode_READY(op) PyUnicode_READY(_PyObject_CAST(op)) + /* Get a copy of a Unicode string. */ PyAPI_FUNC(PyObject*) _PyUnicode_Copy( PyObject *unicode