Change PyUnicode_KIND to 1,2,4. Drop _KIND_SIZE and _CHARACTER_SIZE.

Author: loewis

Doc/c-api/unicode.rst

@@ -99,7 +99,7 @@ access internal read-only data of Unicode objects:
 
    .. deprecated-removed:: 3.3 4.0
       Part of the old-style Unicode API, please migrate to using
-      :c:func:`PyUnicode_GET_LENGTH` or :c:func:`PyUnicode_KIND_SIZE`.
+      :c:func:`PyUnicode_GET_LENGTH`.
 
 
 .. c:function:: Py_UNICODE* PyUnicode_AS_UNICODE(PyObject *o)
@@ -149,9 +149,8 @@ access internal read-only data of Unicode objects:
    Return a pointer to the canonical representation cast to UCS1, UCS2 or UCS4
    integer types for direct character access.  No checks are performed if the
    canonical representation has the correct character size; use
-   :c:func:`PyUnicode_CHARACTER_SIZE` or :c:func:`PyUnicode_KIND` to select the
-   right macro.  Make sure :c:func:`PyUnicode_READY` has been called before
-   accessing this.
+   :c:func:`PyUnicode_KIND` to select the right macro.  Make sure 
+   :c:func:`PyUnicode_READY` has been called before accessing this.
 
    .. versionadded:: 3.3
 
@@ -176,15 +175,6 @@ access internal read-only data of Unicode objects:
    .. versionadded:: 3.3
 
 
-.. c:function:: int PyUnicode_CHARACTER_SIZE(PyObject *o)
-
-   Return the number of bytes the string uses to represent single characters;
-   this can be 1, 2 or 4.  *o* has to be a Unicode object in the "canonical"
-   representation (not checked).
-
-   .. versionadded:: 3.3
-
-
 .. c:function:: void* PyUnicode_DATA(PyObject *o)
 
    Return a void pointer to the raw unicode buffer.  *o* has to be a Unicode
@@ -193,14 +183,6 @@ access internal read-only data of Unicode objects:
    .. versionadded:: 3.3
 
 
-.. c:function:: int PyUnicode_KIND_SIZE(int kind, Py_ssize_t index)
-
-   Compute ``index * char_size`` where ``char_size`` is ``2**(kind - 1)``.  The
-   index is a character index, the result is a size in bytes.
-
-   .. versionadded:: 3.3
-
-
 .. c:function:: void PyUnicode_WRITE(int kind, void *data, Py_ssize_t index, \
                                      Py_UCS4 value)
 

Include/unicodeobject.h

@@ -305,12 +305,12 @@ typedef struct {
              * character type = Py_UCS2 (16 bits, unsigned)
              * at least one character must be in range U+0100-U+FFFF
 
-           - PyUnicode_4BYTE_KIND (3):
+           - PyUnicode_4BYTE_KIND (4):
 
              * character type = Py_UCS4 (32 bits, unsigned)
              * at least one character must be in range U+10000-U+10FFFF
          */
-        unsigned int kind:2;
+        unsigned int kind:3;
         /* Compact is with respect to the allocation scheme. Compact unicode
            objects only require one memory block while non-compact objects use
            one block for the PyUnicodeObject struct and another for its data
@@ -424,29 +424,21 @@ PyAPI_DATA(PyTypeObject) PyUnicodeIter_Type;
 #define PyUnicode_IS_COMPACT_ASCII(op)                 \
     (PyUnicode_IS_ASCII(op) && PyUnicode_IS_COMPACT(op))
 
+enum PyUnicode_Kind {
 /* String contains only wstr byte characters.  This is only possible
    when the string was created with a legacy API and _PyUnicode_Ready()
    has not been called yet.  */
-#define PyUnicode_WCHAR_KIND 0
-
+    PyUnicode_WCHAR_KIND = 0,
 /* Return values of the PyUnicode_KIND() macro: */
-
-#define PyUnicode_1BYTE_KIND 1
-#define PyUnicode_2BYTE_KIND 2
-#define PyUnicode_4BYTE_KIND 3
-
-
-/* Return the number of bytes the string uses to represent single characters,
-   this can be 1, 2 or 4.
-
-   See also PyUnicode_KIND_SIZE(). */
-#define PyUnicode_CHARACTER_SIZE(op) \
-    (((Py_ssize_t)1 << (PyUnicode_KIND(op) - 1)))
+    PyUnicode_1BYTE_KIND = 1,
+    PyUnicode_2BYTE_KIND = 2,
+    PyUnicode_4BYTE_KIND = 4
+};
 
 /* Return pointers to the canonical representation cast to unsigned char,
    Py_UCS2, or Py_UCS4 for direct character access.
-   No checks are performed, use PyUnicode_CHARACTER_SIZE or
-   PyUnicode_KIND() before to ensure these will work correctly. */
+   No checks are performed, use PyUnicode_KIND() before to ensure
+   these will work correctly. */
 
 #define PyUnicode_1BYTE_DATA(op) ((Py_UCS1*)PyUnicode_DATA(op))
 #define PyUnicode_2BYTE_DATA(op) ((Py_UCS2*)PyUnicode_DATA(op))
@@ -473,13 +465,6 @@ PyAPI_DATA(PyTypeObject) PyUnicodeIter_Type;
      PyUnicode_IS_COMPACT(op) ? _PyUnicode_COMPACT_DATA(op) :   \
      _PyUnicode_NONCOMPACT_DATA(op))
 
-/* Compute (index * char_size) where char_size is 2 ** (kind - 1).
-   The index is a character index, the result is a size in bytes.
-
-   See also PyUnicode_CHARACTER_SIZE(). */
-#define PyUnicode_KIND_SIZE(kind, index) \
-    (((Py_ssize_t)(index)) << ((kind) - 1))
-
 /* 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). */

Modules/_io/textio.c

@@ -291,9 +291,7 @@ _PyIncrementalNewlineDecoder_decode(PyObject *_self,
         kind = PyUnicode_KIND(modified);
         out = PyUnicode_DATA(modified);
         PyUnicode_WRITE(kind, PyUnicode_DATA(modified), 0, '\r');
-        memcpy(out + PyUnicode_KIND_SIZE(kind, 1),
-               PyUnicode_DATA(output),
-               PyUnicode_KIND_SIZE(kind, output_len));
+        memcpy(out + kind, PyUnicode_DATA(output), kind * output_len);
         Py_DECREF(output);
         output = modified; /* output remains ready */
         self->pendingcr = 0;
@@ -336,15 +334,15 @@ _PyIncrementalNewlineDecoder_decode(PyObject *_self,
            for the \r *byte* with the libc's optimized memchr.
            */
         if (seennl == SEEN_LF || seennl == 0) {
-            only_lf = (memchr(in_str, '\r', PyUnicode_KIND_SIZE(kind, len)) == NULL);
+            only_lf = (memchr(in_str, '\r', kind * len) == NULL);
         }
 
         if (only_lf) {
             /* If not already seen, quick scan for a possible "\n" character.
                (there's nothing else to be done, even when in translation mode)
             */
             if (seennl == 0 &&
-                memchr(in_str, '\n', PyUnicode_KIND_SIZE(kind, len)) != NULL) {
+                memchr(in_str, '\n', kind * len) != NULL) {
                 Py_ssize_t i = 0;
                 for (;;) {
                     Py_UCS4 c;
@@ -403,7 +401,7 @@ _PyIncrementalNewlineDecoder_decode(PyObject *_self,
                when there is something to translate. On the other hand,
                we already know there is a \r byte, so chances are high
                that something needs to be done. */
-            translated = PyMem_Malloc(PyUnicode_KIND_SIZE(kind, len));
+            translated = PyMem_Malloc(kind * len);
             if (translated == NULL) {
                 PyErr_NoMemory();
                 goto error;
@@ -1576,15 +1574,14 @@ textiowrapper_read(textio *self, PyObject *args)
 static char *
 find_control_char(int kind, char *s, char *end, Py_UCS4 ch)
 {
-    int size = PyUnicode_KIND_SIZE(kind, 1);
     for (;;) {
         while (PyUnicode_READ(kind, s, 0) > ch)
-            s += size;
+            s += kind;
         if (PyUnicode_READ(kind, s, 0) == ch)
             return s;
         if (s == end)
             return NULL;
-        s += size;
+        s += kind;
     }
 }
 
@@ -1593,14 +1590,13 @@ _PyIO_find_line_ending(
     int translated, int universal, PyObject *readnl,
     int kind, char *start, char *end, Py_ssize_t *consumed)
 {
-    int size = PyUnicode_KIND_SIZE(kind, 1);
-    Py_ssize_t len = ((char*)end - (char*)start)/size;
+    Py_ssize_t len = ((char*)end - (char*)start)/kind;
 
     if (translated) {
         /* Newlines are already translated, only search for \n */
         char *pos = find_control_char(kind, start, end, '\n');
         if (pos != NULL)
-            return (pos - start)/size + 1;
+            return (pos - start)/kind + 1;
         else {
             *consumed = len;
             return -1;
@@ -1616,20 +1612,20 @@ _PyIO_find_line_ending(
             /* Fast path for non-control chars. The loop always ends
                since the Unicode string is NUL-terminated. */
             while (PyUnicode_READ(kind, s, 0) > '\r')
-                s += size;
+                s += kind;
             if (s >= end) {
                 *consumed = len;
                 return -1;
             }
             ch = PyUnicode_READ(kind, s, 0);
-            s += size;
+            s += kind;
             if (ch == '\n')
-                return (s - start)/size;
+                return (s - start)/kind;
             if (ch == '\r') {
                 if (PyUnicode_READ(kind, s, 0) == '\n')
-                    return (s - start)/size + 1;
+                    return (s - start)/kind + 1;
                 else
-                    return (s - start)/size;
+                    return (s - start)/kind;
             }
         }
     }
@@ -1642,13 +1638,13 @@ _PyIO_find_line_ending(
         if (readnl_len == 1) {
             char *pos = find_control_char(kind, start, end, nl[0]);
             if (pos != NULL)
-                return (pos - start)/size + 1;
+                return (pos - start)/kind + 1;
             *consumed = len;
             return -1;
         }
         else {
             char *s = start;
-            char *e = end - (readnl_len - 1)*size;
+            char *e = end - (readnl_len - 1)*kind;
             char *pos;
             if (e < s)
                 e = s;
@@ -1662,14 +1658,14 @@ _PyIO_find_line_ending(
                         break;
                 }
                 if (i == readnl_len)
-                    return (pos - start)/size + readnl_len;
-                s = pos + size;
+                    return (pos - start)/kind + readnl_len;
+                s = pos + kind;
             }
             pos = find_control_char(kind, e, end, nl[0]);
             if (pos == NULL)
                 *consumed = len;
             else
-                *consumed = (pos - start)/size;
+                *consumed = (pos - start)/kind;
             return -1;
         }
     }
@@ -1738,8 +1734,8 @@ _textiowrapper_readline(textio *self, Py_ssize_t limit)
         endpos = _PyIO_find_line_ending(
             self->readtranslate, self->readuniversal, self->readnl,
             kind,
-            ptr + PyUnicode_KIND_SIZE(kind, start),
-            ptr + PyUnicode_KIND_SIZE(kind, line_len),
+            ptr + kind * start,
+            ptr + kind * line_len,
             &consumed);
         if (endpos >= 0) {
             endpos += start;

Modules/_json.c

@@ -365,7 +365,7 @@ scanstring_unicode(PyObject *pystr, Py_ssize_t end, int strict, Py_ssize_t *next
             APPEND_OLD_CHUNK
                 chunk = PyUnicode_FromKindAndData(
                     kind,
-                    (char*)buf + PyUnicode_KIND_SIZE(kind, end),
+                    (char*)buf + kind * end,
                     next - end);
             if (chunk == NULL) {
                 goto bail;
@@ -931,7 +931,7 @@ _match_number_unicode(PyScannerObject *s, PyObject *pystr, Py_ssize_t start, Py_
     if (custom_func) {
         /* copy the section we determined to be a number */
         numstr = PyUnicode_FromKindAndData(kind,
-                                           (char*)str + PyUnicode_KIND_SIZE(kind, start),
+                                           (char*)str + kind * start,
                                            idx - start);
         if (numstr == NULL)
             return NULL;

Modules/_sre.c

@@ -1669,7 +1669,7 @@ getstring(PyObject* string, Py_ssize_t* p_length,
             return NULL;
         ptr = PyUnicode_DATA(string);
         *p_length = PyUnicode_GET_LENGTH(string);
-        *p_charsize = PyUnicode_CHARACTER_SIZE(string);
+        *p_charsize = PyUnicode_KIND(string);
         *p_logical_charsize = 4;
         return ptr;
     }

Objects/stringlib/eq.h

@@ -30,5 +30,5 @@ unicode_eq(PyObject *aa, PyObject *bb)
         PyUnicode_GET_LENGTH(a) == 1)
         return 1;
     return memcmp(PyUnicode_1BYTE_DATA(a), PyUnicode_1BYTE_DATA(b),
-                  PyUnicode_GET_LENGTH(a) * PyUnicode_CHARACTER_SIZE(a)) == 0;
+                  PyUnicode_GET_LENGTH(a) * PyUnicode_KIND(a)) == 0;
 }