Patch Attached....
On Wed, May 23, 2001 at 05:28:12PM +1000, [EMAIL PROTECTED] wrote:
> We probably need somewhere a table that takes an array of
> fallback encoding names. Most of the entries probably only
> need one fallback name but the Japanese Shift JIS ones
> might need more. Look in the encodings.def file in the
> libiconv source...
>
Does this patch meet the requirement?
--
Best regard
hashao
Index: xap_EncodingManager.cpp
===================================================================
RCS file: /cvsroot/abi/src/af/xap/xp/xap_EncodingManager.cpp,v
retrieving revision 1.32
diff -u -r1.32 xap_EncodingManager.cpp
--- xap_EncodingManager.cpp 2001/04/27 06:01:32 1.32
+++ xap_EncodingManager.cpp 2001/05/23 10:14:39
@@ -27,6 +27,41 @@
#include <stdio.h>
#include <string.h>
+/*!
+ This macros allow the use of an iconv fallback name if needed.
+ The fallbackname is an arrary of strings. NULL marks the end.
+ If no valid fallbackname is found, set the destination to 'name'.
+ Or else, what can we do?
+ */
+#define CPNAME_OR_FALLBACK(destination,name,fallbackname) \
+{ \
+ static char* fallback_macro_cpname = NULL; \
+ if (!fallback_macro_cpname) \
+ { \
+ iconv_t macro_cd = iconv_open(name,name); \
+ if (macro_cd == (iconv_t)-1) \
+ { \
+ const char** macro_cur = fallbackname; \
+ for(;*macro_cur; macro_cur++) \
+ { \
+ macro_cd = iconv_open(*macro_cur, *macro_cur); \
+ if(macro_cd != (iconv_t)-1) \
+ { \
+ fallback_macro_cpname = *macro_cur; \
+ iconv_close(macro_cd); \
+ break; \
+ } \
+ } \
+ fallback_macro_cpname = name ; \
+ } \
+ else \
+ { \
+ fallback_macro_cpname = name; \
+ iconv_close(macro_cd); \
+ } \
+ } \
+ destination = fallback_macro_cpname; \
+}
static iconv_t iconv_handle_N2U = NULL, iconv_handle_U2N = NULL,
iconv_handle_U2Latin1 = NULL,
@@ -375,6 +410,26 @@
return search_map_with_opt_suffix(m,fallback_key,fallback_key_final);
};
+struct _map_to_array
+{
+ const char* key;//it can't be NULL for non-special entries like last or first
+ const char** values;//NULL-teminated array of strings
+};
+
+/*
+ * Search a map from char* to an array of char*. Return the found char** if
+ * key is found in the mapping, else return NULL.
+ */
+static const char** search_map_to_array(const _map_to_array* m,const char* key)
+{
+ const _map_to_array* cur = m+1;
+ for(;cur->key;++cur)
+ {
+ if (!UT_stricmp(cur->key,key))
+ return cur->values;
+ }
+ return NULL;
+};
/* ************************* here begin tables *************************/
/* this array describes mapping form current encoding to Tex's encoding name.
@@ -431,7 +486,7 @@
Tested with GB2312 only.
*/
static const char* wincharsetcode_zh_GB2312[]= /* chinese*/
-{ "zh_CN.GB2312", "zh_TW.GB2312", NULL };
+{ "zh_CN.GBK", "zh_CN.GB2312", "zh_TW.GB2312", NULL };
static const char* wincharsetcode_zh_BIG5[]= /* chinese*/
{ "zh_CN.BIG5", "zh_TW.BIG5", NULL };
@@ -496,12 +551,14 @@
This table is useful since iconv implementations don't know some cpNNNN
charsets but under some different name.
*/
-static const _map MSCodepagename_to_charset_name_map[]=
+static const char* GB_encode[] = {"GBK", "GB2312", NULL};
+static const char* BIG5_encode[] = {"BIG5", NULL};
+static const _map_to_array MSCodepagename_to_charset_name_map[]=
{
/*key, value*/
{NULL,NULL},
- {"CP936","GB2312"},
- {"CP950","BIG5"},
+ {"CP936",GB_encode},
+ {"CP950",BIG5_encode},
{NULL,NULL}
};
@@ -514,6 +571,7 @@
{
/*key,value*/
{NULL,NULL},
+ {"GBK", "CP936"},
{"GB2312","CP936"},
{"BIG5","CP950"},
{NULL,NULL}
@@ -526,6 +584,7 @@
{NULL},
{"zh_CN.BIG5", "0x404"},
{"zh_CN.GB2312", "0x804"},
+ {"zh_CN.GBK", "0x804"},
{"zh_TW.BIG5", "0x404"},
{"zh_TW.GB2312", "0x804"},
{NULL}
@@ -879,9 +938,11 @@
static char buf[100];
sprintf(buf,"CP%d",lid);
char* cpname = buf;
- bool is_default;
- const char* ret =
search_map(MSCodepagename_to_charset_name_map,cpname,&is_default);
- return is_default ? cpname : ret;
+ char* destinate_charset = cpname;
+ const char** ret = search_map_to_array(MSCodepagename_to_charset_name_map,cpname);
+ if(ret)
+ CPNAME_OR_FALLBACK(destinate_charset, cpname, ret);
+ return !ret ? cpname : destinate_charset;
};
const char* XAP_EncodingManager::CodepageFromCharset(char *charset) const
@@ -895,9 +956,11 @@
const char* XAP_EncodingManager::WindowsCharsetName() const
{
char* cpname = wvLIDToCodePageConverter(getWinLanguageCode());
- bool is_default;
- const char* ret =
search_map(MSCodepagename_to_charset_name_map,cpname,&is_default);
- return is_default ? cpname : ret;
+ char* destinate_charset = cpname;
+ const char** ret = search_map_to_array(MSCodepagename_to_charset_name_map,cpname);
+ if(ret)
+ CPNAME_OR_FALLBACK(destinate_charset, cpname, ret);
+ return !ret ? cpname : destinate_charset;
};
UT_uint32 XAP_EncodingManager::getWinLanguageCode() const
