Attached patch addes conversion from Japanese ISO-2022-JP html
to UTF-8 display_charset.
Screen captures:
http://www1.interq.or.jp/~deton/lynx-jis2utf8/
--
KIHARA, Hideto / de...@m1.interq.or.jp
diff --git a/WWW/Library/Implementation/SGML.c b/WWW/Library/Implementation/SGML.c
index 1f20ee8..5fbea95 100644
--- a/WWW/Library/Implementation/SGML.c
+++ b/WWW/Library/Implementation/SGML.c
@@ -93,6 +93,11 @@ static void fake_put_character(HTStream *p GCC_UNUSED,
/*the following macros are used for pretty source view. */
#define IS_C(attr) (attr.type == HTMLA_CLASS)
+#if defined(ISO2022JP_TOUTF8) && defined(EXP_JAPANESEUTF8_SUPPORT)
+# define UTF8_TTY_ISO2022JP (me->T.output_utf8)
+#else
+# define UTF8_TTY_ISO2022JP 0
+#endif
HTCJKlang HTCJK = NOCJK; /* CJK enum value. */
BOOL HTPassEightBitRaw = FALSE; /* Pass 161-172,174-255 raw. */
BOOL HTPassEightBitNum = FALSE; /* Pass ^ numeric entities raw. */
@@ -1659,6 +1664,34 @@ static void SGML_character(HTStream *me, int c_in)
/*
* If we want the raw input converted to Unicode, try that now. - FM
*/
+ /* Convert ISO-2022-JP to Unicode (charset=iso-2022-jp is unrecognized) */
+#define IS_JIS7_HILO(c) (0x20<(c)&&(c)<0x7F)
+ if (UTF8_TTY_ISO2022JP && (me->state == S_nonascii_text
+ || me->state == S_nonascii_text_sq
+ || me->state == S_nonascii_text_dq)) {
+ /* end of ISO-2022-JP? || not in ISO-2022-JP range */
+ if (TOASCII(c) == '\033' || !IS_JIS7_HILO(c)) {
+ me->kanji_buf = '\0';
+ goto top1;
+ }
+ if (me->kanji_buf == '\t') { /* flag for single byte kana in "ESC(I" */
+ if (conv_jisx0201kana) {
+ JISx0201TO0208_SJIS(c | 0200, me->U.utf_buf, me->U.utf_buf + 1);
+ clong = UCTransJPToUni(me->U.utf_buf, 2, UCGetLYhndl_byMIME("shift_jis"));
+ } else {
+ clong = UCTransToUni(c | 0200, UCGetLYhndl_byMIME("shift_jis"));
+ }
+ } else if (me->kanji_buf) {
+ me->U.utf_buf[0] = me->kanji_buf | 0200; /* to EUC-JP */
+ me->U.utf_buf[1] = c | 0200;
+ clong = UCTransJPToUni(me->U.utf_buf, 2, UCGetLYhndl_byMIME("euc-jp"));
+ me->kanji_buf = '\0';
+ } else {
+ me->kanji_buf = c;
+ clong = ucNeedMore;
+ }
+ goto top1;
+ }
if (me->T.trans_to_uni &&
#ifdef EXP_JAPANESEUTF8_SUPPORT
((strcmp(LYCharSet_UC[me->inUCLYhndl].MIMEname, "euc-jp") == 0) ||
@@ -1808,7 +1841,8 @@ static void SGML_character(HTStream *me, int c_in)
*/
if (TOASCII(clong) < 32 &&
c != '\t' && c != '\n' && c != '\r' &&
- !IS_CJK_TTY)
+ !IS_CJK_TTY &&
+ !(UTF8_TTY_ISO2022JP && TOASCII(c) == '\033'))
goto after_switch;
/*
@@ -1916,13 +1950,14 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_in_kanji;
me->kanji_buf = c;
break;
- } else if (IS_CJK_TTY && TOASCII(c) == '\033') { /* S/390 -- gil -- 0881 */
+ } else if ((IS_CJK_TTY || UTF8_TTY_ISO2022JP) && TOASCII(c) == '\033') { /* S/390 -- gil -- 0881 */
/*
* Setting up for CJK escape sequence handling (based on Takuya
* ASADA's (as...@three-a.co.jp) CJK Lynx). - FM
*/
me->state = S_esc;
- PUTC(c);
+ if (!UTF8_TTY_ISO2022JP)
+ PUTC(c);
break;
}
@@ -3649,7 +3684,8 @@ static void SGML_character(HTStream *me, int c_in)
* - Takuya ASADA (as...@three-a.co.jp)
*/
me->state = S_esc_sq;
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
} else if (me->T.decode_utf8 &&
*me->U.utf_buf) {
HTChunkPuts(string, me->U.utf_buf);
@@ -3693,7 +3729,8 @@ static void SGML_character(HTStream *me, int c_in)
* - Takuya ASADA (as...@three-a.co.jp)
*/
me->state = S_esc_dq;
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
} else if (me->T.decode_utf8 &&
*me->U.utf_buf) {
HTChunkPuts(string, me->U.utf_buf);
@@ -3956,8 +3993,11 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_paren;
} else {
me->state = S_text;
+ if (UTF8_TTY_ISO2022JP)
+ goto top1;
}
- PUTC(c);
+ if (!UTF8_TTY_ISO2022JP)
+ PUTC(c);
break;
case S_dollar: /* Expecting '@', 'B', 'A' or '(' after CJK "ESC$". */
@@ -3966,7 +4006,8 @@ static void SGML_character(HTStream *me, int c_in)
} else if (c == '(') {
me->state = S_dollar_paren;
}
- PUTC(c);
+ if (!UTF8_TTY_ISO2022JP)
+ PUTC(c);
break;
case S_dollar_paren: /* Expecting 'C' after CJK "ESC$(". */
@@ -3974,8 +4015,13 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_nonascii_text;
} else {
me->state = S_text;
+ if (UTF8_TTY_ISO2022JP) {
+ PUTS("$(");
+ goto top1;
+ }
}
- PUTC(c);
+ if (!UTF8_TTY_ISO2022JP)
+ PUTC(c);
break;
case S_paren: /* Expecting 'B', 'J', 'T' or 'I' after CJK "ESC(". */
@@ -3983,19 +4029,30 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_text;
} else if (c == 'I') {
me->state = S_nonascii_text;
+ if (UTF8_TTY_ISO2022JP)
+ me->kanji_buf = '\t'; /* flag for single byte katakana */
} else {
me->state = S_text;
+ if (UTF8_TTY_ISO2022JP) {
+ PUTC('(');
+ goto top1;
+ }
}
- PUTC(c);
+ if (!UTF8_TTY_ISO2022JP)
+ PUTC(c);
break;
case S_nonascii_text: /* Expecting CJK ESC after non-ASCII text. */
if (TOASCII(c) == '\033') { /* S/390 -- gil -- 1264 */
me->state = S_esc;
- }
- PUTC(c);
- if (c < 32)
+ } else if (c < 32) {
me->state = S_text;
+ }
+ if (UTF8_TTY_ISO2022JP) {
+ if (TOASCII(c) != '\033')
+ PUTUTF8(clong);
+ } else
+ PUTC(c);
break;
case S_esc_sq: /* Expecting '$'or '(' following CJK ESC. */
@@ -4005,8 +4062,11 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_paren_sq;
} else {
me->state = S_squoted;
+ if (UTF8_TTY_ISO2022JP)
+ goto top1;
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_dollar_sq: /* Expecting '@', 'B', 'A' or '(' after CJK "ESC$". */
@@ -4015,7 +4075,8 @@ static void SGML_character(HTStream *me, int c_in)
} else if (c == '(') {
me->state = S_dollar_paren_sq;
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_dollar_paren_sq: /* Expecting 'C' after CJK "ESC$(". */
@@ -4023,8 +4084,13 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_nonascii_text_sq;
} else {
me->state = S_squoted;
+ if (UTF8_TTY_ISO2022JP) {
+ HTChunkPuts(string, "$(");
+ goto top1;
+ }
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_paren_sq: /* Expecting 'B', 'J', 'T' or 'I' after CJK "ESC(". */
@@ -4032,17 +4098,28 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_squoted;
} else if (c == 'I') {
me->state = S_nonascii_text_sq;
+ if (UTF8_TTY_ISO2022JP)
+ me->kanji_buf = '\t'; /* flag for single byte katakana */
} else {
me->state = S_squoted;
+ if (UTF8_TTY_ISO2022JP) {
+ HTChunkPutc(string, '(');
+ goto top1;
+ }
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_nonascii_text_sq: /* Expecting CJK ESC after non-ASCII text. */
if (TOASCII(c) == '\033') { /* S/390 -- gil -- 1281 */
me->state = S_esc_sq;
}
- HTChunkPutc(string, c);
+ if (UTF8_TTY_ISO2022JP) {
+ if (TOASCII(c) != '\033')
+ HTChunkPutUtf8Char(string, clong);
+ } else
+ HTChunkPutc(string, c);
break;
case S_esc_dq: /* Expecting '$'or '(' following CJK ESC. */
@@ -4052,8 +4129,11 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_paren_dq;
} else {
me->state = S_dquoted;
+ if (UTF8_TTY_ISO2022JP)
+ goto top1;
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_dollar_dq: /* Expecting '@', 'B', 'A' or '(' after CJK "ESC$". */
@@ -4062,7 +4142,8 @@ static void SGML_character(HTStream *me, int c_in)
} else if (c == '(') {
me->state = S_dollar_paren_dq;
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_dollar_paren_dq: /* Expecting 'C' after CJK "ESC$(". */
@@ -4070,8 +4151,13 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_nonascii_text_dq;
} else {
me->state = S_dquoted;
+ if (UTF8_TTY_ISO2022JP) {
+ HTChunkPuts(string, "$(");
+ goto top1;
+ }
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_paren_dq: /* Expecting 'B', 'J', 'T' or 'I' after CJK "ESC(". */
@@ -4079,17 +4165,28 @@ static void SGML_character(HTStream *me, int c_in)
me->state = S_dquoted;
} else if (c == 'I') {
me->state = S_nonascii_text_dq;
+ if (UTF8_TTY_ISO2022JP)
+ me->kanji_buf = '\t'; /* flag for single byte katakana */
} else {
me->state = S_dquoted;
+ if (UTF8_TTY_ISO2022JP) {
+ HTChunkPutc(string, '(');
+ goto top1;
+ }
}
- HTChunkPutc(string, c);
+ if (!UTF8_TTY_ISO2022JP)
+ HTChunkPutc(string, c);
break;
case S_nonascii_text_dq: /* Expecting CJK ESC after non-ASCII text. */
if (TOASCII(c) == '\033') { /* S/390 -- gil -- 1298 */
me->state = S_esc_dq;
}
- HTChunkPutc(string, c);
+ if (UTF8_TTY_ISO2022JP) {
+ if (TOASCII(c) != '\033')
+ HTChunkPutUtf8Char(string, clong);
+ } else
+ HTChunkPutc(string, c);
break;
case S_junk_tag:
_______________________________________________
Lynx-dev mailing list
Lynx-dev@nongnu.org
https://lists.nongnu.org/mailman/listinfo/lynx-dev
