Don't screw up UTF-8 combined characters.

This commit is contained in:
Nicholas Marriott 2009-01-26 20:57:45 +00:00
parent 514adbdb4d
commit 4d7e555a48
4 changed files with 187 additions and 17 deletions

View File

@ -1,3 +1,8 @@
26 January 2009
* Don't balls up the terminal on UTF-8 combined characters. Don't support them
properly either - they are just discarded for the moment.
25 January 2009
* load-buffer command
@ -1008,7 +1013,7 @@
(including mutt, emacs). No status bar yet and no key remapping or other
customisation.
$Id: CHANGES,v 1.234 2009-01-25 19:00:10 tcunha Exp $
$Id: CHANGES,v 1.235 2009-01-26 20:57:42 nicm Exp $
LocalWords: showw utf UTF fulvio ciriaco joshe OSC APC gettime abc DEF OA clr
LocalWords: rivo nurges lscm Erdely eol smysession mysession ek dstname RB ms

3
TODO
View File

@ -75,7 +75,8 @@
- UTF-8 combining characters don't work. store in one cell - pointer to table
of chains. bit of a hack and would mean max of 65536 per screen. OR maybe
utf8 should work differently; could store as multiple cells, 1 of width >0
and n of width 0, then translate cursor indexes on-the-fly
and n of width 0, then translate cursor indexes on-the-fly would need to
adjust all cursor movement and also handle different width lines properly.
(hopefully) for 0.7, in no particular order:
- swap-pane-up, swap-pane-down (maybe move-pane-*)

View File

@ -1,4 +1,4 @@
/* $Id: screen-write.c,v 1.28 2009-01-19 17:15:19 nicm Exp $ */
/* $Id: screen-write.c,v 1.29 2009-01-26 20:57:44 nicm Exp $ */
/*
* Copyright (c) 2007 Nicholas Marriott <nicm@users.sourceforge.net>
@ -520,6 +520,10 @@ screen_write_cell(struct screen_write_ctx *ctx, const struct grid_cell *gc)
else
width = 1;
/* Discard zero-width characters. */
if (width == 0)
return;
/* If the character is wider than the screen, don't print it. */
if (width > screen_size_x(s)) {
memcpy(&tc, gc, sizeof tc);

188
utf8.c
View File

@ -1,4 +1,4 @@
/* $Id: utf8.c,v 1.3 2009-01-17 18:38:12 nicm Exp $ */
/* $Id: utf8.c,v 1.4 2009-01-26 20:57:45 nicm Exp $ */
/*
* Copyright (c) 2008 Nicholas Marriott <nicm@users.sourceforge.net>
@ -22,6 +22,169 @@
#include "tmux.h"
struct {
u_int first;
u_int last;
int width;
} utf8_width_table[] = {
{ 0x00300, 0x0036f, 0 },
{ 0x00483, 0x00486, 0 },
{ 0x00488, 0x00489, 0 },
{ 0x00591, 0x005bd, 0 },
{ 0x005bf, 0x005bf, 0 },
{ 0x005c1, 0x005c2, 0 },
{ 0x005c4, 0x005c5, 0 },
{ 0x005c7, 0x005c7, 0 },
{ 0x00600, 0x00603, 0 },
{ 0x00610, 0x00615, 0 },
{ 0x0064b, 0x0065e, 0 },
{ 0x00670, 0x00670, 0 },
{ 0x006d6, 0x006e4, 0 },
{ 0x006e7, 0x006e8, 0 },
{ 0x006ea, 0x006ed, 0 },
{ 0x0070f, 0x0070f, 0 },
{ 0x00711, 0x00711, 0 },
{ 0x00730, 0x0074a, 0 },
{ 0x007a6, 0x007b0, 0 },
{ 0x007eb, 0x007f3, 0 },
{ 0x00901, 0x00902, 0 },
{ 0x0093c, 0x0093c, 0 },
{ 0x00941, 0x00948, 0 },
{ 0x0094d, 0x0094d, 0 },
{ 0x00951, 0x00954, 0 },
{ 0x00962, 0x00963, 0 },
{ 0x00981, 0x00981, 0 },
{ 0x009bc, 0x009bc, 0 },
{ 0x009c1, 0x009c4, 0 },
{ 0x009cd, 0x009cd, 0 },
{ 0x009e2, 0x009e3, 0 },
{ 0x00a01, 0x00a02, 0 },
{ 0x00a3c, 0x00a3c, 0 },
{ 0x00a41, 0x00a42, 0 },
{ 0x00a47, 0x00a48, 0 },
{ 0x00a4b, 0x00a4d, 0 },
{ 0x00a70, 0x00a71, 0 },
{ 0x00a81, 0x00a82, 0 },
{ 0x00abc, 0x00abc, 0 },
{ 0x00ac1, 0x00ac5, 0 },
{ 0x00ac7, 0x00ac8, 0 },
{ 0x00acd, 0x00acd, 0 },
{ 0x00ae2, 0x00ae3, 0 },
{ 0x00b01, 0x00b01, 0 },
{ 0x00b3c, 0x00b3c, 0 },
{ 0x00b3f, 0x00b3f, 0 },
{ 0x00b41, 0x00b43, 0 },
{ 0x00b4d, 0x00b4d, 0 },
{ 0x00b56, 0x00b56, 0 },
{ 0x00b82, 0x00b82, 0 },
{ 0x00bc0, 0x00bc0, 0 },
{ 0x00bcd, 0x00bcd, 0 },
{ 0x00c3e, 0x00c40, 0 },
{ 0x00c46, 0x00c48, 0 },
{ 0x00c4a, 0x00c4d, 0 },
{ 0x00c55, 0x00c56, 0 },
{ 0x00cbc, 0x00cbc, 0 },
{ 0x00cbf, 0x00cbf, 0 },
{ 0x00cc6, 0x00cc6, 0 },
{ 0x00ccc, 0x00ccd, 0 },
{ 0x00ce2, 0x00ce3, 0 },
{ 0x00d41, 0x00d43, 0 },
{ 0x00d4d, 0x00d4d, 0 },
{ 0x00dca, 0x00dca, 0 },
{ 0x00dd2, 0x00dd4, 0 },
{ 0x00dd6, 0x00dd6, 0 },
{ 0x00e31, 0x00e31, 0 },
{ 0x00e34, 0x00e3a, 0 },
{ 0x00e47, 0x00e4e, 0 },
{ 0x00eb1, 0x00eb1, 0 },
{ 0x00eb4, 0x00eb9, 0 },
{ 0x00ebb, 0x00ebc, 0 },
{ 0x00ec8, 0x00ecd, 0 },
{ 0x00f18, 0x00f19, 0 },
{ 0x00f35, 0x00f35, 0 },
{ 0x00f37, 0x00f37, 0 },
{ 0x00f39, 0x00f39, 0 },
{ 0x00f71, 0x00f7e, 0 },
{ 0x00f80, 0x00f84, 0 },
{ 0x00f86, 0x00f87, 0 },
{ 0x00f90, 0x00f97, 0 },
{ 0x00f99, 0x00fbc, 0 },
{ 0x00fc6, 0x00fc6, 0 },
{ 0x0102d, 0x01030, 0 },
{ 0x01032, 0x01032, 0 },
{ 0x01036, 0x01037, 0 },
{ 0x01039, 0x01039, 0 },
{ 0x01058, 0x01059, 0 },
{ 0x01100, 0x0115f, 2 },
{ 0x01160, 0x011ff, 0 },
{ 0x0135f, 0x0135f, 0 },
{ 0x01712, 0x01714, 0 },
{ 0x01732, 0x01734, 0 },
{ 0x01752, 0x01753, 0 },
{ 0x01772, 0x01773, 0 },
{ 0x017b4, 0x017b5, 0 },
{ 0x017b7, 0x017bd, 0 },
{ 0x017c6, 0x017c6, 0 },
{ 0x017c9, 0x017d3, 0 },
{ 0x017dd, 0x017dd, 0 },
{ 0x0180b, 0x0180d, 0 },
{ 0x018a9, 0x018a9, 0 },
{ 0x01920, 0x01922, 0 },
{ 0x01927, 0x01928, 0 },
{ 0x01932, 0x01932, 0 },
{ 0x01939, 0x0193b, 0 },
{ 0x01a17, 0x01a18, 0 },
{ 0x01b00, 0x01b03, 0 },
{ 0x01b34, 0x01b34, 0 },
{ 0x01b36, 0x01b3a, 0 },
{ 0x01b3c, 0x01b3c, 0 },
{ 0x01b42, 0x01b42, 0 },
{ 0x01b6b, 0x01b73, 0 },
{ 0x01dc0, 0x01dca, 0 },
{ 0x01dfe, 0x01dff, 0 },
{ 0x0200b, 0x0200f, 0 },
{ 0x0202a, 0x0202e, 0 },
{ 0x02060, 0x02063, 0 },
{ 0x0206a, 0x0206f, 0 },
{ 0x020d0, 0x020ef, 0 },
{ 0x02329, 0x02329, 2 },
{ 0x0232a, 0x0232a, 2 },
{ 0x02e80, 0x03029, 2 },
{ 0x0302a, 0x0302f, 0 },
{ 0x03030, 0x0303e, 2 },
{ 0x03040, 0x0a4cf, 2 },
{ 0x03099, 0x0309a, 0 },
{ 0x0a806, 0x0a806, 0 },
{ 0x0a80b, 0x0a80b, 0 },
{ 0x0a825, 0x0a826, 0 },
{ 0x0ac00, 0x0d7a3, 2 },
{ 0x0f900, 0x0faff, 2 },
{ 0x0fb1e, 0x0fb1e, 0 },
{ 0x0fe00, 0x0fe0f, 0 },
{ 0x0fe10, 0x0fe19, 2 },
{ 0x0fe20, 0x0fe23, 0 },
{ 0x0fe30, 0x0fe6f, 2 },
{ 0x0feff, 0x0feff, 0 },
{ 0x0ff00, 0x0ff60, 2 },
{ 0x0ffe0, 0x0ffe6, 2 },
{ 0x0fff9, 0x0fffb, 0 },
{ 0x10a01, 0x10a03, 0 },
{ 0x10a05, 0x10a06, 0 },
{ 0x10a0c, 0x10a0f, 0 },
{ 0x10a38, 0x10a3a, 0 },
{ 0x10a3f, 0x10a3f, 0 },
{ 0x1d167, 0x1d169, 0 },
{ 0x1d173, 0x1d182, 0 },
{ 0x1d185, 0x1d18b, 0 },
{ 0x1d1aa, 0x1d1ad, 0 },
{ 0x1d242, 0x1d244, 0 },
{ 0x20000, 0x2fffd, 2 },
{ 0x30000, 0x3fffd, 2 },
{ 0xe0001, 0xe0001, 0 },
{ 0xe0020, 0xe007f, 0 },
{ 0xe0100, 0xe01ef, 0 }
};
u_int
utf8_combine(const u_char *data)
{
@ -70,18 +233,15 @@ utf8_split(u_int uv, u_char *data)
int
utf8_width(u_int uv)
{
if ((uv >= 0x1100 && uv <= 0x115f) ||
uv == 0x2329 ||
uv == 0x232a ||
(uv >= 0x2e80 && uv <= 0xa4cf && uv != 0x303f) ||
(uv >= 0xac00 && uv <= 0xd7a3) ||
(uv >= 0xf900 && uv <= 0xfaff) ||
(uv >= 0xfe10 && uv <= 0xfe19) ||
(uv >= 0xfe30 && uv <= 0xfe6f) ||
(uv >= 0xff00 && uv <= 0xff60) ||
(uv >= 0xffe0 && uv <= 0xffe6) ||
(uv >= 0x20000 && uv <= 0x2fffd) ||
(uv >= 0x30000 && uv <= 0x3fffd))
return (2);
u_int i;
for (i = 0; i < nitems(utf8_width_table); i++) {
/* XXX bsearch */
if (uv < utf8_width_table[i].first)
continue;
if (uv > utf8_width_table[i].last)
continue;
return (utf8_width_table[i].width);
}
return (1);
}