Changeset 696b405 in mainline for common

Timestamp:

2025-04-13T12:54:10Z (6 months ago)

Author:

Jiří Zárevúcky <zarevucky.jiri@…>

Branches:

master

Children:

9daee3de

Parents:

7d1497c

git-author:

Jiří Zárevúcky <zarevucky.jiri@…> (2025-04-13 10:13:16)

git-committer:

Jiří Zárevúcky <zarevucky.jiri@…> (2025-04-13 12:54:10)

Message:

Implement standard character conversion functions from <uchar.h> and <wchar.h>

These are more somewhat more flexible than the existing functions in <str.h> since
they can process incomplete stream and save unfinished character for next call.

Location:

common

Files:

• include/uchar.h (moved) (moved from uspace/lib/c/include/uchar.h ) (2 diffs)
• include/wchar.h (added)
• stdc/uchar.c (added)
• stdc/wchar.c (moved) (moved from uspace/lib/c/include/wchar.h ) (2 diffs)

common/include/uchar.h

@@ -1 +1 @@
 /*
- * Copyright (c) 2020 Martin Decky
+ * Copyright (c) 2025 Jiří Zárevúcky
  * All rights reserved.
  *
@@ -36 +36 @@
 #define _LIBC_UCHAR_H_
 
+#include <_bits/mbstate_t.h>
+#include <_bits/size_t.h>
 #include <_bits/uchar.h>
+#include <stdint.h>
+
+size_t mbrtoc8(char8_t *__restrict pc8, const char *__restrict s, size_t n,
+    mbstate_t *__restrict ps);
+size_t c8rtomb(char *__restrict s, char8_t c8, mbstate_t *__restrict ps);
+size_t mbrtoc16(char16_t *__restrict pc16, const char *__restrict s, size_t n,
+    mbstate_t *__restrict ps);
+size_t c16rtomb(char *__restrict s, char16_t c16, mbstate_t *__restrict ps);
+size_t mbrtoc32(char32_t *__restrict pc32, const char *__restrict s, size_t n,
+    mbstate_t *__restrict ps);
+size_t c32rtomb(char *__restrict s, char32_t c32, mbstate_t *__restrict ps);
+
+#ifdef _HELENOS_SOURCE
+#define UCHAR_ILSEQ      ((size_t) -1)
+#define UCHAR_INCOMPLETE ((size_t) -2)
+#define UCHAR_CONTINUED  ((size_t) -3)
+#endif
 
 #endif

common/stdc/wchar.c

@@ -1 +1 @@
 /*
- * Copyright (c) 2017 CZ.NIC, z.s.p.o.
+ * Copyright (c) 2025 Jiří Zárevúcky
  * All rights reserved.
  *
@@ -27 +27 @@
  */
 
-/*
- * Authors:
- *      Jiří Zárevúcky (jzr) <zarevucky.jiri@gmail.com>
- */
+#include <uchar.h>
+#include <wchar.h>
 
-/** @addtogroup libc
- * @{
- */
-/** @file
- */
-
-#ifndef _LIBC_WCHAR_H_
-#define _LIBC_WCHAR_H_
-
-#include <_bits/size_t.h>
-#include <_bits/wchar_t.h>
-#include <_bits/wchar_limits.h>
-#include <_bits/wint_t.h>
-#include <_bits/NULL.h>
-#include <_bits/WEOF.h>
-
+#if __STDC_HOSTED__
+#include <fibril.h>
 #endif
 
-/** @}
- */
+wint_t btowc(int c)
+{
+        return (c < 0x80) ? c : WEOF;
+}
+
+int wctob(wint_t c)
+{
+        return c;
+}
+
+int mbsinit(const mbstate_t *ps)
+{
+        return ps == NULL || ps->continuation == 0;
+}
+
+size_t mbrlen(const char *s, size_t n, mbstate_t *ps)
+{
+#if __STDC_HOSTED__
+        static fibril_local mbstate_t global_state;
+        if (!ps)
+                ps = &global_state;
+#endif
+
+        return mbrtowc(NULL, s, n, ps);
+}
+
+_Static_assert(sizeof(wchar_t) == sizeof(char16_t) || sizeof(wchar_t) == sizeof(char32_t));
+
+size_t mbrtowc(wchar_t *pwc, const char *s, size_t n, mbstate_t *ps)
+{
+#if __STDC_HOSTED__
+        static fibril_local mbstate_t global_state;
+        if (!ps)
+                ps = &global_state;
+#endif
+
+        if (sizeof(wchar_t) == sizeof(char16_t))
+                return mbrtoc16((char16_t *) pwc, s, n, ps);
+        else
+                return mbrtoc32((char32_t *) pwc, s, n, ps);
+}
+
+size_t wcrtomb(char *s, wchar_t wc, mbstate_t * ps)
+{
+#if __STDC_HOSTED__
+        static fibril_local mbstate_t global_state;
+        if (!ps)
+                ps = &global_state;
+#endif
+
+        if (sizeof(wchar_t) == sizeof(char16_t))
+                return c16rtomb(s, (char16_t) wc, ps);
+        else
+                return c32rtomb(s, (char32_t) wc, ps);
+}