handle iso-2022-jp-2 as a superset of iso-2022-jp.  this fixes the decoding
issue with 4657D0DD-A119-4E19-B50D-EBCE5861F9F8@gmail.com from 31 dec 2013

Reference: /n/atom/patch/applied/tcs2022-jp-2
Date: Wed Jan  1 23:09:19 CET 2014
Signed-off-by: quanstro@quanstro.net

--- /sys/src/cmd/tcs/mkfile	Wed Jan  1 23:07:53 2014
+++ /sys/src/cmd/tcs/mkfile	Wed Jan  1 23:07:54 2014
@@ -10,6 +10,7 @@
 	utf.$O\
 	html.$O\
 	kuten208.$O\
+	kuten212.$O\
 	gb.$O\
 	gbk.$O\
 	ksc.$O\
--- /sys/src/cmd/tcs/conv.h	Wed Jan  1 23:07:55 2014
+++ /sys/src/cmd/tcs/conv.h	Wed Jan  1 23:07:56 2014
@@ -1,5 +1,5 @@
 void jis_in(int fd, long *notused, struct convert *out);
-void jisjis_in(int fd, long *notused, struct convert *out);
+void jis2_in(int fd, long *notused, struct convert *out);
 void msjis_in(int fd, long *notused, struct convert *out);
 void ujis_in(int fd, long *notused, struct convert *out);
 void jisjis_out(Rune *base, int n, long *notused);
--- /sys/src/cmd/tcs/tcs.c	Wed Jan  1 23:07:58 2014
+++ /sys/src/cmd/tcs/tcs.c	Wed Jan  1 23:07:59 2014
@@ -525,8 +525,10 @@
 	{ "ibm862", "IBM Code Page 862 (Hebrew)", Table, (void*)tabcp862 },
 	{ "ibm866", "IBM Code Page 866 (Russian)", Table, (void*)tabcp866 },
 	{ "ibm874", "IBM Code Page 874 (Thai)", Table, (void*)tabcp874 },
-	{ "iso-2022-jp", "alias for jis-kanji (MIME)", From|Func, 0, (Fnptr)jisjis_in },
+	{ "iso-2022-jp", "alias for jis-kanji (MIME)", From|Func, 0, (Fnptr)jis2_in },
 	{ "iso-2022-jp", "alias for jis-kanji (MIME)", Func, 0, (Fnptr)jisjis_out },
+	{ "iso-2022-jp-2", "alias for jis-kanji (MIME)", From|Func, 0, (Fnptr)jis2_in },
+	{ "iso-2022-jp-2", "alias for jis-kanji (MIME)", Func, 0, (Fnptr)jisjis_out },
 	{ "iso-8859-1", "alias for 8859-1 (MIME)", Table, (void *)tab8859_1 },
 	{ "iso-8859-2", "alias for 8859-2 (MIME)", Table, (void *)tab8859_2 },
 	{ "iso-8859-3", "alias for 8859-3 (MIME)", Table, (void *)tab8859_3 },
@@ -539,7 +541,7 @@
 	{ "iso-8859-10", "alias for 8859-10 (MIME)", Table, (void *)tab8859_10 },
 	{ "iso-8859-15", "alias for 8859-15 (MIME)", Table, (void *)tab8859_15 },
 	{ "jis", "guesses at the JIS encoding", From|Func, 0, (Fnptr)jis_in },
-	{ "jis-kanji", "ISO 2022-JP (Japanese)", From|Func, 0, (Fnptr)jisjis_in },
+	{ "jis-kanji", "ISO 2022-JP (Japanese)", From|Func, 0, (Fnptr)jis2_in },
 	{ "jis-kanji", "ISO 2022-JP (Japanese)", Func, 0, (Fnptr)jisjis_out },
 	{ "koi8", "KOI-8 (GOST 19769-74)", Table, (void *)tabkoi8 },
 	{ "koi8-r", "alias for koi8 (MIME)", Table, (void *)tabkoi8 },
--- /sys/src/cmd/tcs/conv_jis.c	Wed Jan  1 23:08:00 2014
+++ /sys/src/cmd/tcs/conv_jis.c	Wed Jan  1 23:08:02 2014
@@ -10,7 +10,10 @@
 #include	"hdr.h"
 #include	"conv.h"
 #include	"kuten208.h"
+#include	"kuten212.h"
 #include	"jis.h"
+#include	"gb.h"
+#include	"ksc.h"
 
 /*
 	a state machine for interpreting all sorts of encodings
@@ -254,56 +257,176 @@
 }
 
 /*
-	a state machine for interpreting jis-kanji == 2022-JP
-*/
+ * a partial state machine for interpreting iso-2022-2
+ */
+extern long tab8859_7[];		/* consider a bit of maintence on tcs! */
+
 static void
-jis(int c, Rune **r, long input_loc)
+jis2(int c, Rune **r, long input_loc)
 {
-	static enum { state0, state1, state2, state3, state4 } state = state0;
-	static int set8 = 0;
-	static int japan646 = 0;
-	static int lastc;
 	int n;
 	long l;
+	static enum { state0, state1, staten, state1dot, state2, state21, state3, state4, } state = state0;
+	static int lastc, japan646;
+	static enum {iNone, i88591, i88597, } g2 = iNone;
+	static enum {Cnone, Ckuten208, Ckuten212, Cgb, Cksc, } cs = Cnone;
 
 again:
-	switch(state)
-	{
+
+	if(c == '\r' || c == '\n')
+		g2 = iNone;
+
+	switch(state){
 	case state0:	/* idle state */
-		if(c == ESC){ state = state1; return; }
-		if(c < 0) return;
-		if(!set8 && (c < 128)){
+		if(c == ESC){
+			state = state1;
+			return;
+		}
+		if(c < 0)
+			return;
+		if(cs == Cnone && (c < 128)){
 			if(japan646){
-				switch(c)
-				{
-				case '\\':	emit(0xA5); return;	/* yen */
-				case '~':	emit(0xAF); return;	/* spacing macron */
-				default:	emit(c); return;
+				switch(c){
+				case '\\':
+					emit(0xA5);	/* yen */
+					return;
+				case '~':
+					emit(0xAF);	/* spacing macron */
+					return;
 				}
-			} else {
-				emit(c);
-				return;
 			}
+			emit(c);
+			return;
 		}
-		lastc = c; state = state4; return;
+		lastc = c;
+		state = state4;
+		return;
 
-	case state1:	/* seen an escape */
-		if(c == '$'){ state = state2; return; }
-		if(c == '('){ state = state3; return; }
-		emit(ESC); state = state0; goto again;
+	case state1:	/* ESC */
+		switch(c){
+		case '$':
+			state = state2; 
+			return; 
+		case '(':
+			state = state3; 
+			return; 
+		case 'N':
+			state = staten;
+			return;
+		case '.':
+			state = state1dot;
+			return;
+		default:
+			emit(ESC); 
+			state = state0; 
+			goto again;
+		}
 
-	case state2:	/* may be shifting into JIS */
-		if((c == '@') || (c == 'B')){
-			set8 = 1; state = state0; return;
+	case staten:	/* ESC N */
+		if((uint)c > 0x80)
+			goto bad;
+		switch(g2){
+		default:
+		case iNone:
+			if(squawk)
+				EPR "%s: bad char %#x %s\n", argv0, c, file);
+			goto error;
+
+		case i88591:
+			emit(c+0x80);
+			state = state0;
+			return;
+		case i88597:
+			c = tab8859_7[c | 0x80];
+			if(c < 0){
+		bad:
+				if(squawk)
+					EPR "%s: bad char %#x %s\n", argv0, c, file);
+		error:
+				nerrors++;
+				if(!clean)
+					emit(BADMAP);
+				emit('(');
+				emit(c);
+				emit(')');
+			}else
+				emit(c);
+			state = state0;
+			return;
+		}
+
+	case state1dot:	/* ESC . */
+		switch(c){
+		case 'A':
+			g2 = i88591;
+			state = state0;
+			return;
+		case 'F':
+			g2 = i88597;
+			state = state0;
+			return;
+		default:
+			state = state0;
+			emit(ESC);
+			emit('$');
+			emit('.');
+			goto again;
+		}
+
+	case state2:	/* ESC $ ... may be shifting into JIS */
+		switch(c){
+		case '@':
+		case 'B':
+			cs = Ckuten208;
+			state = state0;
+			return;
+		case 'A':
+			/* handle gb2312-1980 */
+			cs = Cgb;
+			state = state0;
+			return;
+		case '(':
+			state = state21;
+			return;
+		default:
+			emit(ESC); 
+			emit('$'); 
+			state = state0; 
+			goto again;
+		}
+
+	case state21:	/* ESC $ ( */
+		switch(c){
+		case 'C':
+			/* handle ksc5601-1987 */
+			cs = Cksc;
+			state = state0;
+			return;
+		case 'D':
+			/* handle jis x 0212-1990 */
+			cs = Ckuten212;
+			japan646 = 0;			/* guess */
+			state = state0;
+			return;
+		default:
+			emit(ESC);
+			emit('$');
+			emit('(');
+			state = state0;
+			goto again;
 		}
-		emit(ESC); emit('$'); state = state0; goto again;
 
 	case state3:	/* may be shifting out of JIS */
-		if((c == 'J') || (c == 'H') || (c == 'B')){
-			japan646 = (c == 'J');
-			set8 = 0; state = state0; return;
+		if(c == 'J' || c == 'H' || c == 'B'){
+			japan646 = c == 'J';
+			cs = Cnone; 
+			state = state0; 
+			return;
 		}
-		emit(ESC); emit('('); state = state0; goto again;
+		emit(ESC);
+		emit('(');
+		state = state0;
+		goto again;
 
 	case state4:	/* two part char */
 		if(c < 0){
@@ -316,8 +439,29 @@
 			state = state0;
 			goto again;
 		}
-		n = (lastc&0x7F)*100 + (c&0x7f) - 3232;	/* kuten208 */
-		if((n >= KUTEN208MAX) || ((l = tabkuten208[n]) == -1)){
+		n = (lastc&0x7F)*100 + (c&0x7f) - 3232;		/* kuten208 */
+		l = -1;
+		switch(cs){
+		default:
+			abort();
+		case Cgb:
+			if(n < GBMAX)
+				l = tabgb[n];
+			break;
+		case Cksc:
+			if(n < ksc5601max)
+				l = tabksc5601[n];
+			break;
+		case Ckuten208:
+			if(n < KUTEN208MAX)
+				l = tabkuten208[n];
+			break;
+		case Ckuten212:
+			if(n < KUTEN212MAX)
+				l = tabkuten212[n];
+			break;
+		}
+		if(l == -1){
 			nerrors++;
 			if(squawk)
 				EPR "%s: unknown kuten208 %d (from 0x%x,0x%x) near byte %ld in %s\n", argv0, n, lastc, c, input_loc, file);
@@ -388,10 +532,9 @@
 }
 
 void
-jisjis_in(int fd, long *notused, struct convert *out)
+jis2_in(int fd, long*, struct convert *out)
 {
-	USED(notused);
-	do_in(fd, jis, out);
+	do_in(fd, jis2, out);
 }
 
 static int first = 1;