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

import { Tag } from "./constants";
import { Bytes } from "./bytes";
import { Value } from "./values";
import { EncodeError } from "./codec";
import { Record, Tuple } from "./record";
import { EmbeddedTypeEncode } from "./embedded";
import type { Embedded } from "./embedded";
import * as IO from "./iolist";

export type Encodable<T> =
    Value<T> | Preservable<T> | Iterable<Value<T>> | ArrayBufferView;

export interface Preservable<T> {
    __preserve_on__(encoder: Encoder<T>): IO.IOList;
}

export function isPreservable<T>(v: any): v is Preservable<T> {
    return typeof v === 'object' && v !== null && '__preserve_on__' in v && typeof v.__preserve_on__ === '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: EncoderOptions, v: any): IO.IOList {
        return new Encoder(s, this)._encode(embeddedId(v));
    }
};

export function encodeVarint(v: number): IO.IOList {
    function wr(v: number, d: number): IO.IOList {
        if (v < 128) {
            return v + d;
        } else {
            return [wr(Math.floor(v / 128), 0), (v & 127) + d];
        }
    }
    return wr(v, 128);
}

export function encodeInt(v: number): IO.IOList {
    // TODO: Bignums :-/

    if (v === 0) return false;
    if (v === -1) return 255;

    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;

    function enc(n: number, x: number): IO.IOList {
        return (n > 0) && [enc(n - 1, Math.floor(x / 256)), x & 255];
    };

    return enc(bytecount, v);
}

export class Encoder<T = object> {
    options: EncoderOptions;
    embeddedEncode: EmbeddedTypeEncode<T>;

    constructor(options: EncoderEmbeddedOptions<T>);
    constructor(options: EncoderOptions, embeddedEncode?: EmbeddedTypeEncode<T>);
    constructor(options: (EncoderOptions | EncoderEmbeddedOptions<T>) = {},
                embeddedEncode?: EmbeddedTypeEncode<T>)
    {
        this.options = options;
        if ('embeddedEncode' in options) {
            this.embeddedEncode = options.embeddedEncode ?? identityEmbeddedTypeEncode;
        } else {
            this.embeddedEncode = embeddedEncode ?? identityEmbeddedTypeEncode;
        }
    }

    get canonical(): boolean {
        return this.options.canonical ?? true;
    }

    get includeAnnotations(): boolean {
        return this.options.includeAnnotations ?? !this.canonical;
    }

    encode(v: Encodable<T>): Bytes {
        return IO.ioListBytes(this._encode(v));
    }

    encodeString(v: Encodable<T>): string {
        return asLatin1(this.encode(v)._view);
    }

    _encodevalues(items: Iterable<Value<T>>): IO.IOList {
        const ios: IO.IOList = [];
        for (let i of items) {
            const c = IO.countIOList(this._encode(i));
            ios.push(encodeVarint(c.length));
            ios.push(c);
        }
        return ios;
    }

    _encode(v: Encodable<T>): IO.IOList {
        if (isPreservable<T>(v)) return v.__preserve_on__(this);
        if (typeof v === 'boolean') return v ? Tag.True : Tag.False;
        if (typeof v === 'number') return [Tag.SignedInteger, encodeInt(v)];
        if (typeof v === 'string') return [Tag.String, new Bytes(v)._view, 0];
        if (typeof v === 'symbol') {
            const key = Symbol.keyFor(v);
            if (key === void 0) throw new EncodeError("Cannot preserve non-global Symbol", v);
            return [Tag.Symbol, new Bytes(key)._view];
        }
        if (ArrayBuffer.isView(v)) {
            if (v instanceof Uint8Array) {
                return [Tag.ByteString, v];
            } else {
                return [Tag.ByteString, new Uint8Array(v.buffer, v.byteOffset, v.byteLength)];
            }
        }
        if (Record.isRecord<Value<T>, Tuple<Value<T>>, T>(v)) {
            return [Tag.Record, this._encodevalues([v.label, ... v])];
        }
        if (isIterable<Value<T>>(v)) return [Tag.Sequence, this._encodevalues(v)];
        return ((v: Embedded<T>) =>
            [Tag.Embedded, this.embeddedEncode.encode(this.options, v.embeddedValue)])(v);
    }
}

export function encode<T>(
    v: Encodable<T>,
    options: EncoderEmbeddedOptions<T> = {}): Bytes
{
    return new Encoder(options).encode(v);
}

const _canonicalEncoder = new Encoder({ canonical: true });

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) {
        return _canonicalEncoder.encode(v);
    } else {
        return encode(v, { ... options, canonical: true });
    }
}

export function canonicalString(v: Encodable<any>): string {
    return _canonicalEncoder.encodeString(v);
}

export function encodeWithAnnotations<T>(v: Encodable<T>,
                                         options: EncoderEmbeddedOptions<T> = {}): Bytes {
    return encode(v, { ... options, includeAnnotations: true });
}