mirror of
https://github.com/tmux/tmux.git
synced 2024-11-12 23:28:52 +00:00
f09cde2542
position rather than keeping the last character from the input stream, this is how most terminals work and fixes problems with displaying these characters in vim. GitHub issue 3600.
788 lines
15 KiB
C
788 lines
15 KiB
C
/* $OpenBSD$ */
|
|
|
|
/*
|
|
* Copyright (c) 2008 Nicholas Marriott <nicholas.marriott@gmail.com>
|
|
*
|
|
* Permission to use, copy, modify, and distribute this software for any
|
|
* purpose with or without fee is hereby granted, provided that the above
|
|
* copyright notice and this permission notice appear in all copies.
|
|
*
|
|
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
|
|
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
|
|
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
|
|
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
|
|
* WHATSOEVER RESULTING FROM LOSS OF MIND, USE, DATA OR PROFITS, WHETHER
|
|
* IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
|
|
* OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
|
|
*/
|
|
|
|
#include <sys/types.h>
|
|
|
|
#include <ctype.h>
|
|
#include <errno.h>
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <vis.h>
|
|
|
|
#include "tmux.h"
|
|
|
|
static const wchar_t utf8_force_wide[] = {
|
|
0x0261D,
|
|
0x026F9,
|
|
0x0270A,
|
|
0x0270B,
|
|
0x0270C,
|
|
0x0270D,
|
|
0x1F1E6,
|
|
0x1F1E7,
|
|
0x1F1E8,
|
|
0x1F1E9,
|
|
0x1F1EA,
|
|
0x1F1EB,
|
|
0x1F1EC,
|
|
0x1F1ED,
|
|
0x1F1EE,
|
|
0x1F1EF,
|
|
0x1F1F0,
|
|
0x1F1F1,
|
|
0x1F1F2,
|
|
0x1F1F3,
|
|
0x1F1F4,
|
|
0x1F1F5,
|
|
0x1F1F6,
|
|
0x1F1F7,
|
|
0x1F1F8,
|
|
0x1F1F9,
|
|
0x1F1FA,
|
|
0x1F1FB,
|
|
0x1F1FC,
|
|
0x1F1FD,
|
|
0x1F1FE,
|
|
0x1F1FF,
|
|
0x1F385,
|
|
0x1F3C2,
|
|
0x1F3C3,
|
|
0x1F3C4,
|
|
0x1F3C7,
|
|
0x1F3CA,
|
|
0x1F3CB,
|
|
0x1F3CC,
|
|
0x1F3FB,
|
|
0x1F3FC,
|
|
0x1F3FD,
|
|
0x1F3FE,
|
|
0x1F3FF,
|
|
0x1F442,
|
|
0x1F443,
|
|
0x1F446,
|
|
0x1F447,
|
|
0x1F448,
|
|
0x1F449,
|
|
0x1F44A,
|
|
0x1F44B,
|
|
0x1F44C,
|
|
0x1F44D,
|
|
0x1F44E,
|
|
0x1F44F,
|
|
0x1F450,
|
|
0x1F466,
|
|
0x1F467,
|
|
0x1F468,
|
|
0x1F469,
|
|
0x1F46B,
|
|
0x1F46C,
|
|
0x1F46D,
|
|
0x1F46E,
|
|
0x1F470,
|
|
0x1F471,
|
|
0x1F472,
|
|
0x1F473,
|
|
0x1F474,
|
|
0x1F475,
|
|
0x1F476,
|
|
0x1F477,
|
|
0x1F478,
|
|
0x1F47C,
|
|
0x1F481,
|
|
0x1F482,
|
|
0x1F483,
|
|
0x1F485,
|
|
0x1F486,
|
|
0x1F487,
|
|
0x1F48F,
|
|
0x1F491,
|
|
0x1F4AA,
|
|
0x1F574,
|
|
0x1F575,
|
|
0x1F57A,
|
|
0x1F590,
|
|
0x1F595,
|
|
0x1F596,
|
|
0x1F645,
|
|
0x1F646,
|
|
0x1F647,
|
|
0x1F64B,
|
|
0x1F64C,
|
|
0x1F64D,
|
|
0x1F64E,
|
|
0x1F64F,
|
|
0x1F6A3,
|
|
0x1F6B4,
|
|
0x1F6B5,
|
|
0x1F6B6,
|
|
0x1F6C0,
|
|
0x1F6CC,
|
|
0x1F90C,
|
|
0x1F90F,
|
|
0x1F918,
|
|
0x1F919,
|
|
0x1F91A,
|
|
0x1F91B,
|
|
0x1F91C,
|
|
0x1F91D,
|
|
0x1F91E,
|
|
0x1F91F,
|
|
0x1F926,
|
|
0x1F930,
|
|
0x1F931,
|
|
0x1F932,
|
|
0x1F933,
|
|
0x1F934,
|
|
0x1F935,
|
|
0x1F936,
|
|
0x1F937,
|
|
0x1F938,
|
|
0x1F939,
|
|
0x1F93D,
|
|
0x1F93E,
|
|
0x1F977,
|
|
0x1F9B5,
|
|
0x1F9B6,
|
|
0x1F9B8,
|
|
0x1F9B9,
|
|
0x1F9BB,
|
|
0x1F9CD,
|
|
0x1F9CE,
|
|
0x1F9CF,
|
|
0x1F9D1,
|
|
0x1F9D2,
|
|
0x1F9D3,
|
|
0x1F9D4,
|
|
0x1F9D5,
|
|
0x1F9D6,
|
|
0x1F9D7,
|
|
0x1F9D8,
|
|
0x1F9D9,
|
|
0x1F9DA,
|
|
0x1F9DB,
|
|
0x1F9DC,
|
|
0x1F9DD,
|
|
0x1FAC3,
|
|
0x1FAC4,
|
|
0x1FAC5,
|
|
0x1FAF0,
|
|
0x1FAF1,
|
|
0x1FAF2,
|
|
0x1FAF3,
|
|
0x1FAF4,
|
|
0x1FAF5,
|
|
0x1FAF6,
|
|
0x1FAF7,
|
|
0x1FAF8
|
|
};
|
|
|
|
struct utf8_item {
|
|
RB_ENTRY(utf8_item) index_entry;
|
|
u_int index;
|
|
|
|
RB_ENTRY(utf8_item) data_entry;
|
|
char data[UTF8_SIZE];
|
|
u_char size;
|
|
};
|
|
|
|
static int
|
|
utf8_data_cmp(struct utf8_item *ui1, struct utf8_item *ui2)
|
|
{
|
|
if (ui1->size < ui2->size)
|
|
return (-1);
|
|
if (ui1->size > ui2->size)
|
|
return (1);
|
|
return (memcmp(ui1->data, ui2->data, ui1->size));
|
|
}
|
|
RB_HEAD(utf8_data_tree, utf8_item);
|
|
RB_GENERATE_STATIC(utf8_data_tree, utf8_item, data_entry, utf8_data_cmp);
|
|
static struct utf8_data_tree utf8_data_tree = RB_INITIALIZER(utf8_data_tree);
|
|
|
|
static int
|
|
utf8_index_cmp(struct utf8_item *ui1, struct utf8_item *ui2)
|
|
{
|
|
if (ui1->index < ui2->index)
|
|
return (-1);
|
|
if (ui1->index > ui2->index)
|
|
return (1);
|
|
return (0);
|
|
}
|
|
RB_HEAD(utf8_index_tree, utf8_item);
|
|
RB_GENERATE_STATIC(utf8_index_tree, utf8_item, index_entry, utf8_index_cmp);
|
|
static struct utf8_index_tree utf8_index_tree = RB_INITIALIZER(utf8_index_tree);
|
|
|
|
static u_int utf8_next_index;
|
|
|
|
#define UTF8_GET_SIZE(uc) (((uc) >> 24) & 0x1f)
|
|
#define UTF8_GET_WIDTH(uc) (((uc) >> 29) - 1)
|
|
|
|
#define UTF8_SET_SIZE(size) (((utf8_char)(size)) << 24)
|
|
#define UTF8_SET_WIDTH(width) ((((utf8_char)(width)) + 1) << 29)
|
|
|
|
/* Get a UTF-8 item from data. */
|
|
static struct utf8_item *
|
|
utf8_item_by_data(const u_char *data, size_t size)
|
|
{
|
|
struct utf8_item ui;
|
|
|
|
memcpy(ui.data, data, size);
|
|
ui.size = size;
|
|
|
|
return (RB_FIND(utf8_data_tree, &utf8_data_tree, &ui));
|
|
}
|
|
|
|
/* Get a UTF-8 item from data. */
|
|
static struct utf8_item *
|
|
utf8_item_by_index(u_int index)
|
|
{
|
|
struct utf8_item ui;
|
|
|
|
ui.index = index;
|
|
|
|
return (RB_FIND(utf8_index_tree, &utf8_index_tree, &ui));
|
|
}
|
|
|
|
/* Add a UTF-8 item. */
|
|
static int
|
|
utf8_put_item(const u_char *data, size_t size, u_int *index)
|
|
{
|
|
struct utf8_item *ui;
|
|
|
|
ui = utf8_item_by_data(data, size);
|
|
if (ui != NULL) {
|
|
*index = ui->index;
|
|
log_debug("%s: found %.*s = %u", __func__, (int)size, data,
|
|
*index);
|
|
return (0);
|
|
}
|
|
|
|
if (utf8_next_index == 0xffffff + 1)
|
|
return (-1);
|
|
|
|
ui = xcalloc(1, sizeof *ui);
|
|
ui->index = utf8_next_index++;
|
|
RB_INSERT(utf8_index_tree, &utf8_index_tree, ui);
|
|
|
|
memcpy(ui->data, data, size);
|
|
ui->size = size;
|
|
RB_INSERT(utf8_data_tree, &utf8_data_tree, ui);
|
|
|
|
*index = ui->index;
|
|
log_debug("%s: added %.*s = %u", __func__, (int)size, data, *index);
|
|
return (0);
|
|
}
|
|
|
|
static int
|
|
utf8_table_cmp(const void *vp1, const void *vp2)
|
|
{
|
|
const wchar_t *wc1 = vp1, *wc2 = vp2;
|
|
|
|
if (*wc1 < *wc2)
|
|
return (-1);
|
|
if (*wc1 > *wc2)
|
|
return (1);
|
|
return (0);
|
|
}
|
|
|
|
/* Check if character in table. */
|
|
int
|
|
utf8_in_table(wchar_t find, const wchar_t *table, u_int count)
|
|
{
|
|
wchar_t *found;
|
|
|
|
found = bsearch(&find, table, count, sizeof *table, utf8_table_cmp);
|
|
return (found != NULL);
|
|
}
|
|
|
|
/* Get UTF-8 character from data. */
|
|
enum utf8_state
|
|
utf8_from_data(const struct utf8_data *ud, utf8_char *uc)
|
|
{
|
|
u_int index;
|
|
|
|
if (ud->width > 2)
|
|
fatalx("invalid UTF-8 width: %u", ud->width);
|
|
|
|
if (ud->size > UTF8_SIZE)
|
|
goto fail;
|
|
if (ud->size <= 3) {
|
|
index = (((utf8_char)ud->data[2] << 16)|
|
|
((utf8_char)ud->data[1] << 8)|
|
|
((utf8_char)ud->data[0]));
|
|
} else if (utf8_put_item(ud->data, ud->size, &index) != 0)
|
|
goto fail;
|
|
*uc = UTF8_SET_SIZE(ud->size)|UTF8_SET_WIDTH(ud->width)|index;
|
|
log_debug("%s: (%d %d %.*s) -> %08x", __func__, ud->width, ud->size,
|
|
(int)ud->size, ud->data, *uc);
|
|
return (UTF8_DONE);
|
|
|
|
fail:
|
|
if (ud->width == 0)
|
|
*uc = UTF8_SET_SIZE(0)|UTF8_SET_WIDTH(0);
|
|
else if (ud->width == 1)
|
|
*uc = UTF8_SET_SIZE(1)|UTF8_SET_WIDTH(1)|0x20;
|
|
else
|
|
*uc = UTF8_SET_SIZE(1)|UTF8_SET_WIDTH(1)|0x2020;
|
|
return (UTF8_ERROR);
|
|
}
|
|
|
|
/* Get UTF-8 data from character. */
|
|
void
|
|
utf8_to_data(utf8_char uc, struct utf8_data *ud)
|
|
{
|
|
struct utf8_item *ui;
|
|
u_int index;
|
|
|
|
memset(ud, 0, sizeof *ud);
|
|
ud->size = ud->have = UTF8_GET_SIZE(uc);
|
|
ud->width = UTF8_GET_WIDTH(uc);
|
|
|
|
if (ud->size <= 3) {
|
|
ud->data[2] = (uc >> 16);
|
|
ud->data[1] = ((uc >> 8) & 0xff);
|
|
ud->data[0] = (uc & 0xff);
|
|
} else {
|
|
index = (uc & 0xffffff);
|
|
if ((ui = utf8_item_by_index(index)) == NULL)
|
|
memset(ud->data, ' ', ud->size);
|
|
else
|
|
memcpy(ud->data, ui->data, ud->size);
|
|
}
|
|
|
|
log_debug("%s: %08x -> (%d %d %.*s)", __func__, uc, ud->width, ud->size,
|
|
(int)ud->size, ud->data);
|
|
}
|
|
|
|
/* Get UTF-8 character from a single ASCII character. */
|
|
u_int
|
|
utf8_build_one(u_char ch)
|
|
{
|
|
return (UTF8_SET_SIZE(1)|UTF8_SET_WIDTH(1)|ch);
|
|
}
|
|
|
|
/* Set a single character. */
|
|
void
|
|
utf8_set(struct utf8_data *ud, u_char ch)
|
|
{
|
|
static const struct utf8_data empty = { { 0 }, 1, 1, 1 };
|
|
|
|
memcpy(ud, &empty, sizeof *ud);
|
|
*ud->data = ch;
|
|
}
|
|
|
|
/* Copy UTF-8 character. */
|
|
void
|
|
utf8_copy(struct utf8_data *to, const struct utf8_data *from)
|
|
{
|
|
u_int i;
|
|
|
|
memcpy(to, from, sizeof *to);
|
|
|
|
for (i = to->size; i < sizeof to->data; i++)
|
|
to->data[i] = '\0';
|
|
}
|
|
|
|
/* Get width of Unicode character. */
|
|
static enum utf8_state
|
|
utf8_width(struct utf8_data *ud, int *width)
|
|
{
|
|
wchar_t wc;
|
|
|
|
if (utf8_towc(ud, &wc) != UTF8_DONE)
|
|
return (UTF8_ERROR);
|
|
if (utf8_in_table(wc, utf8_force_wide, nitems(utf8_force_wide))) {
|
|
*width = 2;
|
|
return (UTF8_DONE);
|
|
}
|
|
|
|
*width = wcwidth(wc);
|
|
log_debug("wcwidth(%05X) returned %d", (u_int)wc, *width);
|
|
if (*width < 0) {
|
|
/*
|
|
* C1 control characters are nonprintable, so they are always
|
|
* zero width.
|
|
*/
|
|
*width = (wc >= 0x80 && wc <= 0x9f) ? 0 : 1;
|
|
}
|
|
if (*width >= 0 && *width <= 0xff)
|
|
return (UTF8_DONE);
|
|
return (UTF8_ERROR);
|
|
}
|
|
|
|
/* Convert UTF-8 character to wide character. */
|
|
enum utf8_state
|
|
utf8_towc(const struct utf8_data *ud, wchar_t *wc)
|
|
{
|
|
switch (mbtowc(wc, ud->data, ud->size)) {
|
|
case -1:
|
|
log_debug("UTF-8 %.*s, mbtowc() %d", (int)ud->size, ud->data,
|
|
errno);
|
|
mbtowc(NULL, NULL, MB_CUR_MAX);
|
|
return (UTF8_ERROR);
|
|
case 0:
|
|
return (UTF8_ERROR);
|
|
}
|
|
log_debug("UTF-8 %.*s is %05X", (int)ud->size, ud->data, (u_int)*wc);
|
|
return (UTF8_DONE);
|
|
}
|
|
|
|
/*
|
|
* Open UTF-8 sequence.
|
|
*
|
|
* 11000010-11011111 C2-DF start of 2-byte sequence
|
|
* 11100000-11101111 E0-EF start of 3-byte sequence
|
|
* 11110000-11110100 F0-F4 start of 4-byte sequence
|
|
*/
|
|
enum utf8_state
|
|
utf8_open(struct utf8_data *ud, u_char ch)
|
|
{
|
|
memset(ud, 0, sizeof *ud);
|
|
if (ch >= 0xc2 && ch <= 0xdf)
|
|
ud->size = 2;
|
|
else if (ch >= 0xe0 && ch <= 0xef)
|
|
ud->size = 3;
|
|
else if (ch >= 0xf0 && ch <= 0xf4)
|
|
ud->size = 4;
|
|
else
|
|
return (UTF8_ERROR);
|
|
utf8_append(ud, ch);
|
|
return (UTF8_MORE);
|
|
}
|
|
|
|
/* Append character to UTF-8, closing if finished. */
|
|
enum utf8_state
|
|
utf8_append(struct utf8_data *ud, u_char ch)
|
|
{
|
|
int width;
|
|
|
|
if (ud->have >= ud->size)
|
|
fatalx("UTF-8 character overflow");
|
|
if (ud->size > sizeof ud->data)
|
|
fatalx("UTF-8 character size too large");
|
|
|
|
if (ud->have != 0 && (ch & 0xc0) != 0x80)
|
|
ud->width = 0xff;
|
|
|
|
ud->data[ud->have++] = ch;
|
|
if (ud->have != ud->size)
|
|
return (UTF8_MORE);
|
|
|
|
if (ud->width == 0xff)
|
|
return (UTF8_ERROR);
|
|
if (utf8_width(ud, &width) != UTF8_DONE)
|
|
return (UTF8_ERROR);
|
|
ud->width = width;
|
|
|
|
return (UTF8_DONE);
|
|
}
|
|
|
|
/*
|
|
* Encode len characters from src into dst, which is guaranteed to have four
|
|
* bytes available for each character from src (for \abc or UTF-8) plus space
|
|
* for \0.
|
|
*/
|
|
int
|
|
utf8_strvis(char *dst, const char *src, size_t len, int flag)
|
|
{
|
|
struct utf8_data ud;
|
|
const char *start = dst, *end = src + len;
|
|
enum utf8_state more;
|
|
size_t i;
|
|
|
|
while (src < end) {
|
|
if ((more = utf8_open(&ud, *src)) == UTF8_MORE) {
|
|
while (++src < end && more == UTF8_MORE)
|
|
more = utf8_append(&ud, *src);
|
|
if (more == UTF8_DONE) {
|
|
/* UTF-8 character finished. */
|
|
for (i = 0; i < ud.size; i++)
|
|
*dst++ = ud.data[i];
|
|
continue;
|
|
}
|
|
/* Not a complete, valid UTF-8 character. */
|
|
src -= ud.have;
|
|
}
|
|
if (src[0] == '$' && src < end - 1) {
|
|
if (isalpha((u_char)src[1]) ||
|
|
src[1] == '_' ||
|
|
src[1] == '{')
|
|
*dst++ = '\\';
|
|
*dst++ = '$';
|
|
} else if (src < end - 1)
|
|
dst = vis(dst, src[0], flag, src[1]);
|
|
else if (src < end)
|
|
dst = vis(dst, src[0], flag, '\0');
|
|
src++;
|
|
}
|
|
*dst = '\0';
|
|
return (dst - start);
|
|
}
|
|
|
|
/* Same as utf8_strvis but allocate the buffer. */
|
|
int
|
|
utf8_stravis(char **dst, const char *src, int flag)
|
|
{
|
|
char *buf;
|
|
int len;
|
|
|
|
buf = xreallocarray(NULL, 4, strlen(src) + 1);
|
|
len = utf8_strvis(buf, src, strlen(src), flag);
|
|
|
|
*dst = xrealloc(buf, len + 1);
|
|
return (len);
|
|
}
|
|
|
|
/* Same as utf8_strvis but allocate the buffer. */
|
|
int
|
|
utf8_stravisx(char **dst, const char *src, size_t srclen, int flag)
|
|
{
|
|
char *buf;
|
|
int len;
|
|
|
|
buf = xreallocarray(NULL, 4, srclen + 1);
|
|
len = utf8_strvis(buf, src, srclen, flag);
|
|
|
|
*dst = xrealloc(buf, len + 1);
|
|
return (len);
|
|
}
|
|
|
|
/* Does this string contain anything that isn't valid UTF-8? */
|
|
int
|
|
utf8_isvalid(const char *s)
|
|
{
|
|
struct utf8_data ud;
|
|
const char *end;
|
|
enum utf8_state more;
|
|
|
|
end = s + strlen(s);
|
|
while (s < end) {
|
|
if ((more = utf8_open(&ud, *s)) == UTF8_MORE) {
|
|
while (++s < end && more == UTF8_MORE)
|
|
more = utf8_append(&ud, *s);
|
|
if (more == UTF8_DONE)
|
|
continue;
|
|
return (0);
|
|
}
|
|
if (*s < 0x20 || *s > 0x7e)
|
|
return (0);
|
|
s++;
|
|
}
|
|
return (1);
|
|
}
|
|
|
|
/*
|
|
* Sanitize a string, changing any UTF-8 characters to '_'. Caller should free
|
|
* the returned string. Anything not valid printable ASCII or UTF-8 is
|
|
* stripped.
|
|
*/
|
|
char *
|
|
utf8_sanitize(const char *src)
|
|
{
|
|
char *dst = NULL;
|
|
size_t n = 0;
|
|
enum utf8_state more;
|
|
struct utf8_data ud;
|
|
u_int i;
|
|
|
|
while (*src != '\0') {
|
|
dst = xreallocarray(dst, n + 1, sizeof *dst);
|
|
if ((more = utf8_open(&ud, *src)) == UTF8_MORE) {
|
|
while (*++src != '\0' && more == UTF8_MORE)
|
|
more = utf8_append(&ud, *src);
|
|
if (more == UTF8_DONE) {
|
|
dst = xreallocarray(dst, n + ud.width,
|
|
sizeof *dst);
|
|
for (i = 0; i < ud.width; i++)
|
|
dst[n++] = '_';
|
|
continue;
|
|
}
|
|
src -= ud.have;
|
|
}
|
|
if (*src > 0x1f && *src < 0x7f)
|
|
dst[n++] = *src;
|
|
else
|
|
dst[n++] = '_';
|
|
src++;
|
|
}
|
|
dst = xreallocarray(dst, n + 1, sizeof *dst);
|
|
dst[n] = '\0';
|
|
return (dst);
|
|
}
|
|
|
|
/* Get UTF-8 buffer length. */
|
|
size_t
|
|
utf8_strlen(const struct utf8_data *s)
|
|
{
|
|
size_t i;
|
|
|
|
for (i = 0; s[i].size != 0; i++)
|
|
/* nothing */;
|
|
return (i);
|
|
}
|
|
|
|
/* Get UTF-8 string width. */
|
|
u_int
|
|
utf8_strwidth(const struct utf8_data *s, ssize_t n)
|
|
{
|
|
ssize_t i;
|
|
u_int width = 0;
|
|
|
|
for (i = 0; s[i].size != 0; i++) {
|
|
if (n != -1 && n == i)
|
|
break;
|
|
width += s[i].width;
|
|
}
|
|
return (width);
|
|
}
|
|
|
|
/*
|
|
* Convert a string into a buffer of UTF-8 characters. Terminated by size == 0.
|
|
* Caller frees.
|
|
*/
|
|
struct utf8_data *
|
|
utf8_fromcstr(const char *src)
|
|
{
|
|
struct utf8_data *dst = NULL;
|
|
size_t n = 0;
|
|
enum utf8_state more;
|
|
|
|
while (*src != '\0') {
|
|
dst = xreallocarray(dst, n + 1, sizeof *dst);
|
|
if ((more = utf8_open(&dst[n], *src)) == UTF8_MORE) {
|
|
while (*++src != '\0' && more == UTF8_MORE)
|
|
more = utf8_append(&dst[n], *src);
|
|
if (more == UTF8_DONE) {
|
|
n++;
|
|
continue;
|
|
}
|
|
src -= dst[n].have;
|
|
}
|
|
utf8_set(&dst[n], *src);
|
|
n++;
|
|
src++;
|
|
}
|
|
dst = xreallocarray(dst, n + 1, sizeof *dst);
|
|
dst[n].size = 0;
|
|
return (dst);
|
|
}
|
|
|
|
/* Convert from a buffer of UTF-8 characters into a string. Caller frees. */
|
|
char *
|
|
utf8_tocstr(struct utf8_data *src)
|
|
{
|
|
char *dst = NULL;
|
|
size_t n = 0;
|
|
|
|
for(; src->size != 0; src++) {
|
|
dst = xreallocarray(dst, n + src->size, 1);
|
|
memcpy(dst + n, src->data, src->size);
|
|
n += src->size;
|
|
}
|
|
dst = xreallocarray(dst, n + 1, 1);
|
|
dst[n] = '\0';
|
|
return (dst);
|
|
}
|
|
|
|
/* Get width of UTF-8 string. */
|
|
u_int
|
|
utf8_cstrwidth(const char *s)
|
|
{
|
|
struct utf8_data tmp;
|
|
u_int width;
|
|
enum utf8_state more;
|
|
|
|
width = 0;
|
|
while (*s != '\0') {
|
|
if ((more = utf8_open(&tmp, *s)) == UTF8_MORE) {
|
|
while (*++s != '\0' && more == UTF8_MORE)
|
|
more = utf8_append(&tmp, *s);
|
|
if (more == UTF8_DONE) {
|
|
width += tmp.width;
|
|
continue;
|
|
}
|
|
s -= tmp.have;
|
|
}
|
|
if (*s > 0x1f && *s != 0x7f)
|
|
width++;
|
|
s++;
|
|
}
|
|
return (width);
|
|
}
|
|
|
|
/* Pad UTF-8 string to width on the left. Caller frees. */
|
|
char *
|
|
utf8_padcstr(const char *s, u_int width)
|
|
{
|
|
size_t slen;
|
|
char *out;
|
|
u_int n, i;
|
|
|
|
n = utf8_cstrwidth(s);
|
|
if (n >= width)
|
|
return (xstrdup(s));
|
|
|
|
slen = strlen(s);
|
|
out = xmalloc(slen + 1 + (width - n));
|
|
memcpy(out, s, slen);
|
|
for (i = n; i < width; i++)
|
|
out[slen++] = ' ';
|
|
out[slen] = '\0';
|
|
return (out);
|
|
}
|
|
|
|
/* Pad UTF-8 string to width on the right. Caller frees. */
|
|
char *
|
|
utf8_rpadcstr(const char *s, u_int width)
|
|
{
|
|
size_t slen;
|
|
char *out;
|
|
u_int n, i;
|
|
|
|
n = utf8_cstrwidth(s);
|
|
if (n >= width)
|
|
return (xstrdup(s));
|
|
|
|
slen = strlen(s);
|
|
out = xmalloc(slen + 1 + (width - n));
|
|
for (i = 0; i < width - n; i++)
|
|
out[i] = ' ';
|
|
memcpy(out + i, s, slen);
|
|
out[i + slen] = '\0';
|
|
return (out);
|
|
}
|
|
|
|
int
|
|
utf8_cstrhas(const char *s, const struct utf8_data *ud)
|
|
{
|
|
struct utf8_data *copy, *loop;
|
|
int found = 0;
|
|
|
|
copy = utf8_fromcstr(s);
|
|
for (loop = copy; loop->size != 0; loop++) {
|
|
if (loop->size != ud->size)
|
|
continue;
|
|
if (memcmp(loop->data, ud->data, loop->size) == 0) {
|
|
found = 1;
|
|
break;
|
|
}
|
|
}
|
|
free(copy);
|
|
|
|
return (found);
|
|
}
|