1
0
Fork 0
mirror of https://github.com/denoland/deno.git synced 2024-12-17 12:52:26 -05:00
denoland-deno/std/encoding/base32.ts
2020-07-14 15:24:17 -04:00

207 lines
6.6 KiB
TypeScript

// Modified from https://github.com/beatgammit/base64-js
// Copyright (c) 2014 Jameson Little. MIT License.
const lookup: string[] = [];
const revLookup: number[] = [];
// RFC4648 base32
const code = "ABCDEFGHIJKLMNOPQRSTUVWXYZ234567";
for (let i = 0, len = code.length; i < len; ++i) {
lookup[i] = code[i];
revLookup[code.charCodeAt(i)] = i;
}
const placeHolderPadLookup = [0, 1, , 2, 3, , 4];
function _getPadLen(placeHoldersLen: number): number {
const maybeLen = placeHolderPadLookup[placeHoldersLen];
if (typeof maybeLen !== "number") {
throw new Error("Invalid pad length");
}
return maybeLen;
}
function getLens(b32: string): [number, number] {
const len = b32.length;
if (len % 8 > 0) {
throw new Error("Invalid string. Length must be a multiple of 8");
}
let validLen = b32.indexOf("=");
if (validLen === -1) validLen = len;
const placeHoldersLen = validLen === len ? 0 : 8 - (validLen % 8);
return [validLen, placeHoldersLen];
}
/**
* Returns number of bytes encoded in the given RFC4648 base32 string input.
* @param b32
*/
export function byteLength(b32: string): number {
const [validLen, placeHoldersLen] = getLens(b32);
return _byteLength(validLen, placeHoldersLen);
}
function _byteLength(validLen: number, placeHoldersLen: number): number {
return ((validLen + placeHoldersLen) * 5) / 8 - _getPadLen(placeHoldersLen);
}
/**
* Decodes a given RFC4648 base32 encoded string.
* @param b32
*/
export function decode(b32: string): Uint8Array {
let tmp: number;
const [validLen, placeHoldersLen] = getLens(b32);
const arr = new Uint8Array(_byteLength(validLen, placeHoldersLen));
let curByte = 0;
// if there are placeholders, only get up to the last complete 8 chars
const len = placeHoldersLen > 0 ? validLen - 8 : validLen;
let i: number;
for (i = 0; i < len; i += 8) {
tmp = (revLookup[b32.charCodeAt(i)] << 20) |
(revLookup[b32.charCodeAt(i + 1)] << 15) |
(revLookup[b32.charCodeAt(i + 2)] << 10) |
(revLookup[b32.charCodeAt(i + 3)] << 5) |
revLookup[b32.charCodeAt(i + 4)];
arr[curByte++] = (tmp >> 17) & 0xff;
arr[curByte++] = (tmp >> 9) & 0xff;
arr[curByte++] = (tmp >> 1) & 0xff;
tmp = ((tmp & 1) << 15) |
(revLookup[b32.charCodeAt(i + 5)] << 10) |
(revLookup[b32.charCodeAt(i + 6)] << 5) |
revLookup[b32.charCodeAt(i + 7)];
arr[curByte++] = (tmp >> 8) & 0xff;
arr[curByte++] = tmp & 0xff;
}
if (placeHoldersLen === 1) {
tmp = (revLookup[b32.charCodeAt(i)] << 20) |
(revLookup[b32.charCodeAt(i + 1)] << 15) |
(revLookup[b32.charCodeAt(i + 2)] << 10) |
(revLookup[b32.charCodeAt(i + 3)] << 5) |
revLookup[b32.charCodeAt(i + 4)];
arr[curByte++] = (tmp >> 17) & 0xff;
arr[curByte++] = (tmp >> 9) & 0xff;
arr[curByte++] = (tmp >> 1) & 0xff;
tmp = ((tmp & 1) << 7) |
(revLookup[b32.charCodeAt(i + 5)] << 2) |
(revLookup[b32.charCodeAt(i + 6)] >> 3);
arr[curByte++] = tmp & 0xff;
} else if (placeHoldersLen === 3) {
tmp = (revLookup[b32.charCodeAt(i)] << 19) |
(revLookup[b32.charCodeAt(i + 1)] << 14) |
(revLookup[b32.charCodeAt(i + 2)] << 9) |
(revLookup[b32.charCodeAt(i + 3)] << 4) |
(revLookup[b32.charCodeAt(i + 4)] >> 1);
arr[curByte++] = (tmp >> 16) & 0xff;
arr[curByte++] = (tmp >> 8) & 0xff;
arr[curByte++] = tmp & 0xff;
} else if (placeHoldersLen === 4) {
tmp = (revLookup[b32.charCodeAt(i)] << 11) |
(revLookup[b32.charCodeAt(i + 1)] << 6) |
(revLookup[b32.charCodeAt(i + 2)] << 1) |
(revLookup[b32.charCodeAt(i + 3)] >> 4);
arr[curByte++] = (tmp >> 8) & 0xff;
arr[curByte++] = tmp & 0xff;
} else if (placeHoldersLen === 6) {
tmp = (revLookup[b32.charCodeAt(i)] << 3) |
(revLookup[b32.charCodeAt(i + 1)] >> 2);
arr[curByte++] = tmp & 0xff;
}
return arr;
}
function encodeChunk(uint8: Uint8Array, start: number, end: number): string {
let tmp: number;
const output = [];
for (let i = start; i < end; i += 5) {
tmp = ((uint8[i] << 16) & 0xff0000) |
((uint8[i + 1] << 8) & 0xff00) |
(uint8[i + 2] & 0xff);
output.push(lookup[(tmp >> 19) & 0x1f]);
output.push(lookup[(tmp >> 14) & 0x1f]);
output.push(lookup[(tmp >> 9) & 0x1f]);
output.push(lookup[(tmp >> 4) & 0x1f]);
tmp = ((tmp & 0xf) << 16) |
((uint8[i + 3] << 8) & 0xff00) |
(uint8[i + 4] & 0xff);
output.push(lookup[(tmp >> 15) & 0x1f]);
output.push(lookup[(tmp >> 10) & 0x1f]);
output.push(lookup[(tmp >> 5) & 0x1f]);
output.push(lookup[tmp & 0x1f]);
}
return output.join("");
}
/**
* Encodes a given Uint8Array into RFC4648 base32 representation
* @param uint8
*/
export function encode(uint8: Uint8Array): string {
let tmp: number;
const len = uint8.length;
const extraBytes = len % 5;
const parts = [];
const maxChunkLength = 16385; // must be multiple of 5
const len2 = len - extraBytes;
// go through the array every 5 bytes, we'll deal with trailing stuff later
for (let i = 0; i < len2; i += maxChunkLength) {
parts.push(
encodeChunk(
uint8,
i,
i + maxChunkLength > len2 ? len2 : i + maxChunkLength,
),
);
}
// pad the end with zeros, but make sure to not forget the extra bytes
if (extraBytes === 4) {
tmp = ((uint8[len2] & 0xff) << 16) |
((uint8[len2 + 1] & 0xff) << 8) |
(uint8[len2 + 2] & 0xff);
parts.push(lookup[(tmp >> 19) & 0x1f]);
parts.push(lookup[(tmp >> 14) & 0x1f]);
parts.push(lookup[(tmp >> 9) & 0x1f]);
parts.push(lookup[(tmp >> 4) & 0x1f]);
tmp = ((tmp & 0xf) << 11) | (uint8[len2 + 3] << 3);
parts.push(lookup[(tmp >> 10) & 0x1f]);
parts.push(lookup[(tmp >> 5) & 0x1f]);
parts.push(lookup[tmp & 0x1f]);
parts.push("=");
} else if (extraBytes === 3) {
tmp = ((uint8[len2] & 0xff) << 17) |
((uint8[len2 + 1] & 0xff) << 9) |
((uint8[len2 + 2] & 0xff) << 1);
parts.push(lookup[(tmp >> 20) & 0x1f]);
parts.push(lookup[(tmp >> 15) & 0x1f]);
parts.push(lookup[(tmp >> 10) & 0x1f]);
parts.push(lookup[(tmp >> 5) & 0x1f]);
parts.push(lookup[tmp & 0x1f]);
parts.push("===");
} else if (extraBytes === 2) {
tmp = ((uint8[len2] & 0xff) << 12) | ((uint8[len2 + 1] & 0xff) << 4);
parts.push(lookup[(tmp >> 15) & 0x1f]);
parts.push(lookup[(tmp >> 10) & 0x1f]);
parts.push(lookup[(tmp >> 5) & 0x1f]);
parts.push(lookup[tmp & 0x1f]);
parts.push("====");
} else if (extraBytes === 1) {
tmp = (uint8[len2] & 0xff) << 2;
parts.push(lookup[(tmp >> 5) & 0x1f]);
parts.push(lookup[tmp & 0x1f]);
parts.push("======");
}
return parts.join("");
}