PDCLIB-2 PDCLIB-8: Implement mbrtowc using mbrtoc32/mbrtoc16 depending upon definitio...

diff --git a/functions/uchar/mbrtoc32.c b/functions/uchar/mbrtoc32.c
index 05d9184fdce8c6fb9b0c4f4e4cedf187befced47..bb269cdcb593a23a12d686496a5a7e7a1d58a456 100644 (file)
--- a/functions/uchar/mbrtoc32.c
+++ b/functions/uchar/mbrtoc32.c
@@ -33,6 +33,7 @@ size_t mbrtoc32_l(
             // A character was output
             if(nr == n) {
                 // The output character resulted entirely from stored state
+                // With UTF-32, this shouldn't be possible?
                 return (size_t) -3;
             } else if(pc32[-1] == 0) {
                 // Was null character

diff --git a/internals/_PDCLIB_encoding.h b/internals/_PDCLIB_encoding.h
index 9b31735423a84dc20af05713b07caf08d435e352..4d17123edb1dc20f64c9fc9e42b9ef3b51ac97f4 100644 (file)
--- a/internals/_PDCLIB_encoding.h
+++ b/internals/_PDCLIB_encoding.h
@@ -6,7 +6,7 @@
 
 #ifndef __PDCLIB_ENCODING_H
 #define __PDCLIB_ENCODING_H __PDCLIB_ENCODING_H
-#include "_PDCLIB_int.h"
+#include <uchar.h>
 
 /* Must be cauued with bufsize >= 1, in != NULL, out != NULL, ps != NULL
  *
@@ -142,4 +142,60 @@ struct _PDCLIB_charcodec {
     size_t __mb_max;
 };
 
+/* mbstate _PendState values */
+enum {
+    /* Nothing pending; _PendChar ignored */
+    _PendClear = 0, 
+
+    /* Process the character stored in _PendChar before reading the buffer 
+     * passed for the conversion
+     */
+    _PendPrefix = 1,
+};
+
+/* XXX Defining these here is temporary - will move to xlocale in future */
+size_t mbrtoc16_l(
+        char16_t    *_PDCLIB_restrict   pc16,
+        const char  *_PDCLIB_restrict   s, 
+        size_t                          n,
+        mbstate_t   *_PDCLIB_restrict   ps,
+_PDCLIB_locale_t     _PDCLIB_restrict   l);
+
+size_t c16rtomb_l(
+        char        *_PDCLIB_restrict   s, 
+        char16_t                        c16, 
+        mbstate_t   *_PDCLIB_restrict   ps,
+_PDCLIB_locale_t     _PDCLIB_restrict   l);
+
+size_t mbrtoc32_l(
+        char32_t    *_PDCLIB_restrict   pc32,
+        const char  *_PDCLIB_restrict   s, 
+        size_t                          n,
+        mbstate_t   *_PDCLIB_restrict   ps,
+_PDCLIB_locale_t     _PDCLIB_restrict   l);
+
+size_t c32rtomb_l(
+        char        *_PDCLIB_restrict   s, 
+        char32_t                        c32,
+        mbstate_t   *_PDCLIB_restrict   ps,
+_PDCLIB_locale_t     _PDCLIB_restrict   l);
+
+#define _PDCLIB_WCHAR_ENCODING_UTF16 16
+#define _PDCLIB_WCHAR_ENCODING_UCS4  32
+
+#if !defined(_PDCLIB_WCHAR_ENCODING)
+    #define _PDCLIB_WCHAR_ENCODING 0
+#endif
+
+#if _PDCLIB_WCHAR_ENCODING == _PDCLIB_WCHAR_ENCODING_UTF16
+    #define _PDCLIB_mbrtocwc_l mbrtoc16_l
+    #define _PDCLIB_cwcrtomb_l c16rtomb_l
+#elif _PDCLIB_WCHAR_ENCODING == _PDCLIB_WCHAR_ENCODING_UCS4
+    #define _PDCLIB_mbrtocwc_l mbrtoc32_l
+    #define _PDCLIB_cwcrtomb_l c32rtomb_l
+#else
+    #error _PDCLIB_WCHAR_ENCODING not defined correctly
+    #error Define to one of _PDCLIB_WCHAR_ENCODING_UCS4 or _PDCLIB_WCHAR_ENCODING_UTF16
+#endif
+
 #endif

diff --git a/internals/_PDCLIB_int.h b/internals/_PDCLIB_int.h
index 52ba8b35f607b82cd5133562abaa042b045dbea9..9ff9e69e321b3a1809747449bed85d697984e1ed 100644 (file)
--- a/internals/_PDCLIB_int.h
+++ b/internals/_PDCLIB_int.h
@@ -348,16 +348,23 @@ typedef struct _PDCLIB_mbstate {
                  char    _StC [124];
     };
 
-    union {
-        /* c16/related functions: Surrogate storage
-         *
-         * If zero, no surrogate pending. If nonzero, surrogate.
-         */
-        _PDCLIB_uint16_t     _Surrogate;
-
-        /* Reserved for potential mbtoutf8/etc functions */
-        unsigned char        _U8[4];
-    };
+    /* c16/related functions: Surrogate storage
+     *
+     * If zero, no surrogate pending. If nonzero, surrogate.
+     */
+    _PDCLIB_uint16_t     _Surrogate;
+
+    /* In cases where the underlying codec is capable of regurgitating a 
+     * character without consuming any extra input (e.g. a surrogate pair in a
+     * UCS-4 to UTF-16 conversion) then these fields are used to track that 
+     * state. In particular, they are used to buffer/fake the input for mbrtowc
+     * and similar functions.
+     *
+     * See _PDCLIB_encoding.h for values of _PendState and the resultant value 
+     * in _PendChar.
+     */
+    unsigned char _PendState;
+             char _PendChar;
 } _PDCLIB_mbstate_t;
 
 typedef struct _PDCLIB_charcodec *_PDCLIB_charcodec_t;

diff --git a/platform/posix/internals/_PDCLIB_config.h b/platform/posix/internals/_PDCLIB_config.h
index 273f2d2f5997f55382880556406fa8a791bed0a3..576b347c253df2a2984c38426dca97b0f0c0c197 100644 (file)
--- a/platform/posix/internals/_PDCLIB_config.h
+++ b/platform/posix/internals/_PDCLIB_config.h
@@ -327,6 +327,9 @@ struct _PDCLIB_imaxdiv_t
 /* Locale method. See _PDCLIB_locale.h */
 #define _PDCLIB_LOCALE_METHOD _PDCLIB_LOCALE_METHOD_TSS
 
+/* wchar_t encoding */
+#define _PDCLIB_WCHAR_ENCODING _PDCLIB_WCHAR_ENCODING_UCS4
+
 /* I/O ---------------------------------------------------------------------- */
 
 /* The default size for file buffers. Must be at least 256. */

diff --git a/platform/win32/internals/_PDCLIB_config.h b/platform/win32/internals/_PDCLIB_config.h
index 4e4050153f399e7dc7e5bd623ce26d0d06613bb9..c6da3ef5a02afea68a3c4ab3aa36722bd7ea1398 100644 (file)
--- a/platform/win32/internals/_PDCLIB_config.h
+++ b/platform/win32/internals/_PDCLIB_config.h
@@ -338,6 +338,9 @@ struct _PDCLIB_imaxdiv_t
 /* Locale method. See _PDCLIB_locale.h */
 #define _PDCLIB_LOCALE_METHOD _PDCLIB_LOCALE_METHOD_TSS
 
+/* wchar_t encoding */
+#define _PDCLIB_WCHAR_ENCODING _PDCLIB_WCHAR_ENCODING_UTF16
+
 /* I/O ---------------------------------------------------------------------- */
 
 /* The default size for file buffers. Must be at least 256. */