1 /* unicode.c - convert between Unicode and UTF-8
2 *
3 * Copyright 2020 The Android Open Source Project.
4 *
5 * Loosely based on the Plan9/Inferno unicode(1).
6
7 USE_UNICODE(NEWTOY(unicode, "<1", TOYFLAG_USR|TOYFLAG_BIN))
8
9 config UNICODE
10 bool "unicode"
11 default n
12 help
13 usage: unicode [[min]-max]
14
15 Convert between Unicode code points and UTF-8, in both directions.
16 */
17
18 #define FOR_unicode
19 #include "toys.h"
20
codepoint(unsigned wc)21 static void codepoint(unsigned wc) {
22 char *low="NULSOHSTXETXEOTENQACKBELBS HT LF VT FF CR SO SI DLEDC1DC2DC3DC4"
23 "NAKSYNETBCANEM SUBESCFS GS RS US ";
24 unsigned n, i;
25
26 printf("U+%04X : ", wc);
27 if (wc < ' ') printf("%.3s", low+(wc*3));
28 else if (wc == 0x7f) printf("DEL");
29 else {
30 toybuf[n = wctoutf8(toybuf, wc)] = 0;
31 printf("%s%s", toybuf, n>1 ? " :":"");
32 if (n>1) for (i = 0; i < n; i++) printf(" %#02x", toybuf[i]);
33 }
34 xputc('\n');
35 }
36
unicode_main(void)37 void unicode_main(void)
38 {
39 unsigned from, to;
40 char next, **args;
41
42 for (args = toys.optargs; *args; args++) {
43 // unicode 660-666 => table of `U+0600 : ٠ : 0xd9 0xa0` etc.
44 if (sscanf(*args, "%x-%x%c", &from, &to, &next) == 2) {
45 while (from <= to) codepoint(from++);
46
47 // unicode 666 => just `U+0666 : ٦ : 0xd9 0xa6`.
48 } else if (sscanf(*args, "%x%c", &from, &next) == 1) {
49 codepoint(from);
50
51 // unicode hello => table showing every character in the string.
52 } else {
53 char *s = *args;
54 size_t l = strlen(s);
55 wchar_t wc;
56 int n;
57
58 while ((n = utf8towc(&wc, s, l)) > 0) {
59 codepoint(wc);
60 s += n;
61 l -= n;
62 }
63 }
64 }
65 }
66