PDCLIB-3: Add _PDCLIB_unpackwint to enable UTF-16 platforms to pass a "packed UTF...

[pdclib] / internals / _PDCLIB_locale.h

diff --git a/internals/_PDCLIB_locale.h b/internals/_PDCLIB_locale.h
index 9c008e0ad43c745afc6ef5abc77a5c84fb4aba9a..8d28f2d976d2f2c202231b966dc929168f5fd098 100644 (file)
--- a/internals/_PDCLIB_locale.h
+++ b/internals/_PDCLIB_locale.h
@@ -67,13 +67,37 @@ typedef struct _PDCLIB_ctype
     unsigned char collation;
 } _PDCLIB_ctype_t;
 
-typedef struct _PDCLIB_wctype
+typedef struct _PDCLIB_wcinfo
 {
     _PDCLIB_uint32_t num;
     _PDCLIB_uint16_t flags;
     _PDCLIB_uint32_t lower;
     _PDCLIB_uint32_t upper;
-} _PDCLIB_wctype_t;
+} _PDCLIB_wcinfo_t;
+
+static inline _PDCLIB_wint_t _PDCLIB_unpackwint( _PDCLIB_wint_t wc )
+{
+    if( sizeof(_PDCLIB_wchar_t) == 2 && sizeof(_PDCLIB_wint_t) == 4 ) {
+        /* On UTF-16 platforms, as an extension accept a "packed surrogate"
+         * encoding. We accept the surrogate pairs either way
+         */
+
+        _PDCLIB_wint_t c = (wc & 0xF800F800);
+        if(c == (_PDCLIB_wint_t) 0xD800DC00) {
+            // MSW: Lead, LSW: Trail
+            _PDCLIB_wint_t lead  = wc >> 16 & 0x3FF;
+            _PDCLIB_wint_t trail = wc       & 0x3FF;
+            wc = lead << 10 | trail;
+        } else if(c == (_PDCLIB_wint_t) 0xDC00D800) {
+            // MSW: Trail, LSW: Lead
+            _PDCLIB_wint_t trail = wc >> 16 & 0x3FF;
+            _PDCLIB_wint_t lead  = wc       & 0x3FF;
+            wc = lead << 10 | trail;
+        }
+
+    }
+    return wc;
+}
 
 struct _PDCLIB_locale {
     _PDCLIB_charcodec_t          _Codec;