Last change
on this file since 76712b2 was 6a3a178, checked in by Ema <ema_spirova@…>, 3 years ago |
initial commit
|
-
Property mode
set to
100644
|
File size:
1.4 KB
|
Line | |
---|
1 | function con(b) {
|
---|
2 | if ((b & 0xc0) === 0x80) {
|
---|
3 | return b & 0x3f;
|
---|
4 | } else {
|
---|
5 | throw new Error("invalid UTF-8 encoding");
|
---|
6 | }
|
---|
7 | }
|
---|
8 |
|
---|
9 | function code(min, n) {
|
---|
10 | if (n < min || (0xd800 <= n && n < 0xe000) || n >= 0x10000) {
|
---|
11 | throw new Error("invalid UTF-8 encoding");
|
---|
12 | } else {
|
---|
13 | return n;
|
---|
14 | }
|
---|
15 | }
|
---|
16 |
|
---|
17 | export function decode(bytes) {
|
---|
18 | return _decode(bytes)
|
---|
19 | .map(x => String.fromCharCode(x))
|
---|
20 | .join("");
|
---|
21 | }
|
---|
22 |
|
---|
23 | function _decode(bytes) {
|
---|
24 | if (bytes.length === 0) {
|
---|
25 | return [];
|
---|
26 | }
|
---|
27 |
|
---|
28 | /**
|
---|
29 | * 1 byte
|
---|
30 | */
|
---|
31 | {
|
---|
32 | const [b1, ...bs] = bytes;
|
---|
33 |
|
---|
34 | if (b1 < 0x80) {
|
---|
35 | return [code(0x0, b1), ..._decode(bs)];
|
---|
36 | }
|
---|
37 |
|
---|
38 | if (b1 < 0xc0) {
|
---|
39 | throw new Error("invalid UTF-8 encoding");
|
---|
40 | }
|
---|
41 | }
|
---|
42 |
|
---|
43 | /**
|
---|
44 | * 2 bytes
|
---|
45 | */
|
---|
46 | {
|
---|
47 | const [b1, b2, ...bs] = bytes;
|
---|
48 |
|
---|
49 | if (b1 < 0xe0) {
|
---|
50 | return [code(0x80, ((b1 & 0x1f) << 6) + con(b2)), ..._decode(bs)];
|
---|
51 | }
|
---|
52 | }
|
---|
53 |
|
---|
54 | /**
|
---|
55 | * 3 bytes
|
---|
56 | */
|
---|
57 | {
|
---|
58 | const [b1, b2, b3, ...bs] = bytes;
|
---|
59 |
|
---|
60 | if (b1 < 0xf0) {
|
---|
61 | return [
|
---|
62 | code(0x800, ((b1 & 0x0f) << 12) + (con(b2) << 6) + con(b3)),
|
---|
63 | ..._decode(bs)
|
---|
64 | ];
|
---|
65 | }
|
---|
66 | }
|
---|
67 |
|
---|
68 | /**
|
---|
69 | * 4 bytes
|
---|
70 | */
|
---|
71 | {
|
---|
72 | const [b1, b2, b3, b4, ...bs] = bytes;
|
---|
73 |
|
---|
74 | if (b1 < 0xf8) {
|
---|
75 | return [
|
---|
76 | code(
|
---|
77 | 0x10000,
|
---|
78 | ((((b1 & 0x07) << 18) + con(b2)) << 12) + (con(b3) << 6) + con(b4)
|
---|
79 | ),
|
---|
80 | ..._decode(bs)
|
---|
81 | ];
|
---|
82 | }
|
---|
83 | }
|
---|
84 |
|
---|
85 | throw new Error("invalid UTF-8 encoding");
|
---|
86 | }
|
---|
Note:
See
TracBrowser
for help on using the repository browser.