Changeset f94a11f in mainline

Timestamp:

2025-04-15T18:38:48Z (19 hours ago)

Author:

Jiří Zárevúcky <zarevucky.jiri@…>

Branches:

master

Children:

65bf084

Parents:

5d2bdaa

Message:

Rename mbstate_t field to prevent confusion with continuation bytes

Files:

• abi/include/_bits/mbstate_t.h (modified) (1 diff)
• common/stdc/uchar.c (modified) (13 diffs)
• common/stdc/wchar.c (modified) (1 diff)

abi/include/_bits/mbstate_t.h

@@ -37 +37 @@
 
 typedef struct {
-        unsigned short continuation;
+        unsigned short state;
 } mbstate_t;
 

common/stdc/uchar.c

@@ -111 +111 @@
         if (!s) {
                 // Equivalent to mbrtoc32(NULL, "", 1, mb).
-                if (mb->continuation) {
+                if (mb->state) {
                         _set_ilseq();
                         return UCHAR_ILSEQ;
@@ -121 +121 @@
         size_t i = 0;
 
-        if (!mb->continuation) {
+        if (!mb->state) {
                 /* Clean slate, read initial byte. */
 
@@ -152 +152 @@
 
                         /* 2 byte encoding               110xxxxx */
-                        mb->continuation = b ^ 0b0000000011000000;
+                        mb->state = b ^ 0b0000000011000000;
 
                 } else if (_is_3_byte(b)) {
                         /* 3 byte encoding               1110xxxx */
-                        mb->continuation = b ^ 0b1111110011100000;
+                        mb->state = b ^ 0b1111110011100000;
 
                 } else if (_is_4_byte(b)) {
                         /* 4 byte encoding               11110xxx */
-                        mb->continuation = b ^ 0b1111111100000000;
+                        mb->state = b ^ 0b1111111100000000;
                 }
         }
@@ -168 +168 @@
                 uint8_t b = s[i];
 
-                if (!_is_continuation(b) || _is_non_shortest(mb->continuation, b)) {
+                if (!_is_continuation(b) || _is_non_shortest(mb->state, b)) {
                         _set_ilseq();
                         return UCHAR_ILSEQ;
@@ -174 +174 @@
 
                 /* Top bit becomes zero just before the last byte is shifted in. */
-                if (!(mb->continuation & 0x8000)) {
-                        *c = ((char32_t) mb->continuation) << 6 | (b & 0x3f);
-                        mb->continuation = 0;
+                if (!(mb->state & 0x8000)) {
+                        *c = ((char32_t) mb->state) << 6 | (b & 0x3f);
+                        mb->state = 0;
                         return ++i;
                 }
 
-                mb->continuation = mb->continuation << 6 | (b & 0x3f);
+                mb->state = mb->state << 6 | (b & 0x3f);
         }
 
@@ -253 +253 @@
         if (!s) {
                 /* Equivalent to mbrtoc16(NULL, "", 1, mb). */
-                if (mb->continuation) {
+                if (mb->state) {
                         _set_ilseq();
                         return UCHAR_ILSEQ;
@@ -261 +261 @@
         }
 
-        if ((mb->continuation & 0xD000) == 0xD000) {
+        if ((mb->state & 0xD000) == 0xD000) {
                 /* mbstate_t contains the second surrogate character. */
                 /* mbrtoc32() will never set it to such value.        */
-                *c = mb->continuation;
-                mb->continuation = 0;
+                *c = mb->state;
+                mb->state = 0;
                 return UCHAR_CONTINUED;
         }
@@ -276 +276 @@
                 } else {
                         /* Encode UTF-16 surrogates. */
-                        mb->continuation = (c32 & 0x3FF) + 0xDC00;
+                        mb->state = (c32 & 0x3FF) + 0xDC00;
                         *c = (c32 >> 10) + 0xD7C0;
                 }
@@ -298 +298 @@
         if (!s) {
                 // Equivalent to c16rtomb(buf, L’\0’, mb).
-                if (mb->continuation) {
+                if (mb->state) {
                         _set_ilseq();
                         return UCHAR_ILSEQ;
@@ -307 +307 @@
 
         if (!_is_surrogate(c)) {
-                if (mb->continuation) {
+                if (mb->state) {
                         _set_ilseq();
                         return UCHAR_ILSEQ;
@@ -315 +315 @@
         }
 
-        if (!mb->continuation) {
-                mb->continuation = c;
+        if (!mb->state) {
+                mb->state = c;
                 return 0;
         }
@@ -323 +323 @@
 
         /* Decode UTF-16 surrogates. */
-        if (_is_low_surrogate(mb->continuation) && _is_high_surrogate(c)) {
-                c32 = ((c - 0xD7C0) << 10) | (mb->continuation - 0xDC00);
-        } else if (_is_high_surrogate(mb->continuation) && _is_low_surrogate(c)) {
-                c32 = ((mb->continuation - 0xD7C0) << 10) | (c - 0xDC00);
+        if (_is_low_surrogate(mb->state) && _is_high_surrogate(c)) {
+                c32 = ((c - 0xD7C0) << 10) | (mb->state - 0xDC00);
+        } else if (_is_high_surrogate(mb->state) && _is_low_surrogate(c)) {
+                c32 = ((mb->state - 0xD7C0) << 10) | (c - 0xDC00);
         } else {
                 _set_ilseq();
@@ -332 +332 @@
         }
 
-        mb->continuation = 0;
+        mb->state = 0;
         return c32rtomb(s, c32, mb);
 }

common/stdc/wchar.c

@@ -46 +46 @@
 int mbsinit(const mbstate_t *ps)
 {
-        return ps == NULL || ps->continuation == 0;
+        return ps == NULL || ps->state == 0;
 }