preserves/implementations/javascript/packages/core/src/encoder.ts

300 lines
9.1 KiB
TypeScript

import { Tag } from "./constants";
import { Bytes } from "./bytes";
import { Value } from "./values";
import { PreserveOn } from "./symbols";
import { EncodeError } from "./codec";
import { Record, Tuple } from "./record";
import { GenericEmbedded, EmbeddedTypeEncode } from "./embedded";
export type Encodable<T> =
Value<T> | Preservable<T> | Iterable<Value<T>> | ArrayBufferView;
export interface Preservable<T> {
[PreserveOn](encoder: Encoder<T>): void;
}
export function isPreservable<T>(v: any): v is Preservable<T> {
return typeof v === 'object' && v !== null && typeof v[PreserveOn] === 'function';
}
export interface EncoderOptions {
canonical?: boolean;
includeAnnotations?: boolean;
}
export interface EncoderEmbeddedOptions<T> extends EncoderOptions {
embeddedEncode?: EmbeddedTypeEncode<T>;
}
export function asLatin1(bs: Uint8Array): string {
return String.fromCharCode.apply(null, bs as any as number[]);
}
function isIterable<T>(v: any): v is Iterable<T> {
return typeof v === 'object' && v !== null && typeof v[Symbol.iterator] === 'function';
}
let _nextId = 0;
const _registry = new WeakMap<object, number>();
export function embeddedId(v: any): number {
let id = _registry.get(v);
if (id === void 0) {
id = _nextId++;
_registry.set(v, id);
}
return id;
}
export const identityEmbeddedTypeEncode: EmbeddedTypeEncode<any> = {
encode(s: EncoderState, v: any): void {
new Encoder(s, this).push(embeddedId(v));
},
toValue(v: any): Value<GenericEmbedded> {
return embeddedId(v);
}
};
export class EncoderState {
chunks: Array<Uint8Array>;
view: DataView;
index: number;
options: EncoderOptions;
constructor(options: EncoderOptions) {
this.chunks = [];
this.view = new DataView(new ArrayBuffer(256));
this.index = 0;
this.options = options;
}
get canonical(): boolean {
return this.options.canonical ?? true;
}
get includeAnnotations(): boolean {
return this.options.includeAnnotations ?? !this.canonical;
}
contents(): Bytes {
if (this.chunks.length === 0) {
const resultLength = this.index;
this.index = 0;
return new Bytes(this.view.buffer.slice(0, resultLength));
} else {
this.rotatebuffer(4096);
return Bytes.concat(this.chunks);
}
}
/* Like contents(), but hands back a string containing binary data "encoded" via latin-1 */
contentsString(): string {
if (this.chunks.length === 0) {
const s = asLatin1(new Uint8Array(this.view.buffer, 0, this.index));
this.index = 0;
return s;
} else {
this.rotatebuffer(4096);
return this.chunks.map(asLatin1).join('');
}
}
rotatebuffer(size: number) {
this.chunks.push(new Uint8Array(this.view.buffer, 0, this.index));
this.view = new DataView(new ArrayBuffer(size));
this.index = 0;
}
makeroom(amount: number) {
if (this.index + amount > this.view.byteLength) {
this.rotatebuffer(amount + 4096);
}
}
emitbyte(b: number) {
this.makeroom(1);
this.view.setUint8(this.index++, b);
}
emitbytes(bs: Uint8Array) {
this.makeroom(bs.length);
(new Uint8Array(this.view.buffer)).set(bs, this.index);
this.index += bs.length;
}
varint(v: number) {
while (v >= 128) {
this.emitbyte((v % 128) + 128);
v = Math.floor(v / 128);
}
this.emitbyte(v);
}
encodeint(v: number) {
// TODO: Bignums :-/
const plain_bitcount = Math.floor(Math.log2(v > 0 ? v : -(1 + v))) + 1;
const signed_bitcount = plain_bitcount + 1;
const bytecount = (signed_bitcount + 7) >> 3;
if (bytecount <= 16) {
this.emitbyte(Tag.MediumInteger_lo + bytecount - 1);
} else {
this.emitbyte(Tag.SignedInteger);
this.varint(bytecount);
}
const enc = (n: number, x: number) => {
if (n > 0) {
enc(n - 1, Math.floor(x / 256));
this.emitbyte(x & 255);
}
};
enc(bytecount, v);
}
encodebytes(tag: Tag, bs: Uint8Array) {
this.emitbyte(tag);
this.varint(bs.length);
this.emitbytes(bs);
}
}
export class Encoder<T = object> {
state: EncoderState;
embeddedEncode: EmbeddedTypeEncode<T>;
constructor(options: EncoderEmbeddedOptions<T>);
constructor(state: EncoderState, embeddedEncode?: EmbeddedTypeEncode<T>);
constructor(
state_or_options: (EncoderState | EncoderEmbeddedOptions<T>) = {},
embeddedEncode?: EmbeddedTypeEncode<T>)
{
if (state_or_options instanceof EncoderState) {
this.state = state_or_options;
this.embeddedEncode = embeddedEncode ?? identityEmbeddedTypeEncode;
} else {
this.state = new EncoderState(state_or_options);
this.embeddedEncode = state_or_options.embeddedEncode ?? identityEmbeddedTypeEncode;
}
}
withEmbeddedEncode<S>(
embeddedEncode: EmbeddedTypeEncode<S>,
body: (e: Encoder<S>) => void): this
{
body(new Encoder(this.state, embeddedEncode));
return this;
}
get canonical(): boolean {
return this.state.canonical;
}
get includeAnnotations(): boolean {
return this.state.includeAnnotations;
}
contents(): Bytes {
return this.state.contents();
}
contentsString(): string {
return this.state.contentsString();
}
encodevalues(tag: Tag, items: Iterable<Value<T>>) {
this.state.emitbyte(tag);
for (let i of items) { this.push(i); }
this.state.emitbyte(Tag.End);
}
push(v: Encodable<T>) {
if (isPreservable<never>(v)) {
v[PreserveOn](this as unknown as Encoder<never>);
}
else if (isPreservable<T>(v)) {
v[PreserveOn](this);
}
else if (typeof v === 'boolean') {
this.state.emitbyte(v ? Tag.True : Tag.False);
}
else if (typeof v === 'number') {
if (v >= -3 && v <= 12) {
this.state.emitbyte(Tag.SmallInteger_lo + ((v + 16) & 0xf));
} else {
this.state.encodeint(v);
}
}
else if (typeof v === 'string') {
this.state.encodebytes(Tag.String, new Bytes(v)._view);
}
else if (typeof v === 'symbol') {
const key = Symbol.keyFor(v);
if (key === void 0) throw new EncodeError("Cannot preserve non-global Symbol", v);
this.state.encodebytes(Tag.Symbol, new Bytes(key)._view);
}
else if (ArrayBuffer.isView(v)) {
if (v instanceof Uint8Array) {
this.state.encodebytes(Tag.ByteString, v);
} else {
const bs = new Uint8Array(v.buffer, v.byteOffset, v.byteLength);
this.state.encodebytes(Tag.ByteString, bs);
}
}
else if (Record.isRecord<Value<T>, Tuple<Value<T>>, T>(v)) {
this.state.emitbyte(Tag.Record);
this.push(v.label);
for (let i of v) { this.push(i); }
this.state.emitbyte(Tag.End);
}
else if (Array.isArray(v)) {
this.encodevalues(Tag.Sequence, v);
}
else if (isIterable<Value<T>>(v)) {
this.encodevalues(Tag.Sequence, v as Iterable<Value<T>>);
}
else {
this.state.emitbyte(Tag.Embedded);
this.embeddedEncode.encode(this.state, v.embeddedValue);
}
return this; // for chaining
}
}
export function encode<T>(
v: Encodable<T>,
options: EncoderEmbeddedOptions<T> = {}): Bytes
{
return new Encoder(options).push(v).contents();
}
const _canonicalEncoder = new Encoder({ canonical: true });
let _usingCanonicalEncoder = false;
export function canonicalEncode(v: Encodable<never>, options?: EncoderEmbeddedOptions<never>): Bytes;
export function canonicalEncode(v: Encodable<any>, options?: EncoderEmbeddedOptions<any>): Bytes;
export function canonicalEncode(v: any, options?: EncoderEmbeddedOptions<any>): Bytes {
if (options === void 0 && !_usingCanonicalEncoder) {
_usingCanonicalEncoder = true;
const bs = _canonicalEncoder.push(v).contents();
_usingCanonicalEncoder = false;
return bs;
} else {
return encode(v, { ... options, canonical: true });
}
}
export function canonicalString(v: Encodable<any>): string {
if (!_usingCanonicalEncoder) {
_usingCanonicalEncoder = true;
const s = _canonicalEncoder.push(v).contentsString();
_usingCanonicalEncoder = false;
return s;
} else {
return new Encoder({ canonical: true }).push(v).contentsString();
}
}
export function encodeWithAnnotations<T>(v: Encodable<T>,
options: EncoderEmbeddedOptions<T> = {}): Bytes {
return encode(v, { ... options, includeAnnotations: true });
}