Compress the Unicode data into runs. This results in a significant size reduction.

[pdclib.old] / internals / _PDCLIB_locale.h

diff --git a/internals/_PDCLIB_locale.h b/internals/_PDCLIB_locale.h
index dbedc8093e1829b1ed32e1c956aeca2573483830..9644eb4b19c3b836b43505d2358f999093eb3799 100644 (file)
--- a/internals/_PDCLIB_locale.h
+++ b/internals/_PDCLIB_locale.h
@@ -2,6 +2,7 @@
 #define __PDCLIB_LOCALE_H __PDCLIB_LOCALE_H
 #include <_PDCLIB_int.h>
 #include <locale.h>
+#include <wctype.h>
 #include <threads.h>
 #include <stdlib.h>
 
@@ -9,25 +10,26 @@
 #define _PDCLIB_LOCALE_METHOD_THREAD_LOCAL  'T'
 
 #if !defined(_PDCLIB_LOCALE_METHOD)
-    #error _PDCLIB_LOCALE_METHOD undefined: don't know where I'm storing the thread locale
+    /* If undefined, no POSIX per thread locales */
+    #define _PDCLIB_threadlocale() (&_PDCLIB_global_locale)
 #elif _PDCLIB_LOCALE_METHOD == _PDCLIB_LOCALE_METHOD_TSS
     extern tss_t _PDCLIB_locale_tss;
     static inline locale_t _PDCLIB_threadlocale( void )
     {
         locale_t l = tss_get(_PDCLIB_locale_tss);
-        if(l == NULL)
+        if ( l == NULL )
             l = &_PDCLIB_global_locale;
         return l;
     }
 
     static inline void _PDCLIB_setthreadlocale( locale_t l )
     {
-        if(tss_set(_PDCLIB_locale_tss, l) != thrd_success)
+        if ( tss_set( _PDCLIB_locale_tss, l ) != thrd_success )
             abort();
     }
 #elif _PDCLIB_LOCALE_METHOD == _PDCLIB_LOCALE_METHOD_THREAD_LOCAL
     extern thread_local locale_t _PDCLIB_locale_tls;
-    #define _PDCLIB_threadlocale() (_PDCLIB_locale_tls || &_PDCLIB_global_locale)
+    #define _PDCLIB_threadlocale() ( _PDCLIB_locale_tls || &_PDCLIB_global_locale )
     static inline locale_t _PDCLIB_threadlocale( void )
     {
         locale_t l = _PDCLIB_locale_tls;
@@ -59,6 +61,9 @@
 #define _PDCLIB_CTYPE_DIGIT 256
 #define _PDCLIB_CTYPE_XDIGT 512
 
+#define _PDCLIB_WCTRANS_TOLOWER 1
+#define _PDCLIB_WCTRANS_TOUPPER 2
+
 typedef struct _PDCLIB_ctype
 {
     _PDCLIB_uint16_t flags;
@@ -69,48 +74,74 @@ typedef struct _PDCLIB_ctype
 
 typedef struct _PDCLIB_wcinfo
 {
-    _PDCLIB_wint_t   num;
+    _PDCLIB_wint_t   start;
+    _PDCLIB_uint16_t length;
     _PDCLIB_uint16_t flags;
-    _PDCLIB_wint_t   lower;
-    _PDCLIB_wint_t   upper;
+    _PDCLIB_wint_t   lower_delta;
+    _PDCLIB_wint_t   upper_delta;
 } _PDCLIB_wcinfo_t;
 
-extern _PDCLIB_wcinfo_t _PDCLIB_wcinfo[];
-extern size_t           _PDCLIB_wcinfo_size;
+struct _PDCLIB_locale {
+    const _PDCLIB_charcodec_t    _Codec;
+    struct lconv                 _Conv;
+
+    /* ctype / wctype */
+    /* XXX: Maybe re-evaluate constness of these later on? */
+    const _PDCLIB_wcinfo_t      *_WCType;
+    _PDCLIB_size_t               _WCTypeSize;
+    const _PDCLIB_ctype_t       *_CType; 
+
+    /* perror/strerror */
+    char                        *_ErrnoStr[_PDCLIB_ERRNO_MAX];
+};
+
+extern const _PDCLIB_wcinfo_t _PDCLIB_wcinfo[];
+extern const size_t           _PDCLIB_wcinfo_size;
 
 static inline int _PDCLIB_wcinfo_cmp( const void * _key, const void * _obj )
 {
-    _PDCLIB_uint32_t * key = (_PDCLIB_uint32_t *) _key;
+    _PDCLIB_int32_t * key = (_PDCLIB_int32_t *) _key;
     _PDCLIB_wcinfo_t * obj = (_PDCLIB_wcinfo_t *) _obj;
-    return *key - obj->num;
+    if ( *key < obj->start ) 
+    {
+        return -1;
+    } 
+    else if ( *key >= obj->start + obj->length )
+    {
+        return 1;
+    }
+    else
+    {
+        return 0;
+    }
 }
 
-static inline _PDCLIB_wcinfo_t * _PDCLIB_wcgetinfo( _PDCLIB_uint32_t num )
+static inline _PDCLIB_wcinfo_t * _PDCLIB_wcgetinfo( locale_t l, _PDCLIB_int32_t num )
 {
     _PDCLIB_wcinfo_t *info = (_PDCLIB_wcinfo_t*) 
-        bsearch( &num, _PDCLIB_wcinfo, _PDCLIB_wcinfo_size, 
-                 sizeof( _PDCLIB_wcinfo[0] ), _PDCLIB_wcinfo_cmp );
+        bsearch( &num, l->_WCType, l->_WCTypeSize, 
+                 sizeof( l->_WCType[0] ), _PDCLIB_wcinfo_cmp );
 
     return info;
 }
 
-static inline _PDCLIB_wint_t _PDCLIB_unpackwint( _PDCLIB_wint_t wc )
+static inline wint_t _PDCLIB_unpackwint( 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);
+        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;
+            wint_t lead  = wc >> 16 & 0x3FF;
+            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;
+            wint_t trail = wc >> 16 & 0x3FF;
+            wint_t lead  = wc       & 0x3FF;
             wc = lead << 10 | trail;
         }
 
@@ -118,15 +149,22 @@ static inline _PDCLIB_wint_t _PDCLIB_unpackwint( _PDCLIB_wint_t wc )
     return wc;
 }
 
-struct _PDCLIB_locale {
-    _PDCLIB_charcodec_t          _Codec;
-    struct lconv                 _Conv;
-
-    /* ctype */
-    _PDCLIB_ctype_t             *_CType; 
-
-    /* perror/strerror */
-    char                        *_ErrnoStr[_PDCLIB_ERRNO_MAX];
-};
+/* Internal xlocale-style WCType API */
+int _PDCLIB_iswalnum_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswalpha_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswblank_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswcntrl_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswdigit_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswgraph_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswlower_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswprint_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswpunct_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswspace_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswupper_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswxdigit_l( wint_t _Wc, locale_t l );
+int _PDCLIB_iswctype_l( wint_t _Wc, wctype_t _Desc, locale_t l );
+wint_t _PDCLIB_towlower_l( wint_t _Wc, locale_t l );
+wint_t _PDCLIB_towupper_l( wint_t _Wc, locale_t l );
+wint_t _PDCLIB_towctrans_l( wint_t _Wc, wctrans_t _Desc, locale_t l );
 
 #endif