import { prettyByte } from "./utils/prettyByte.ts";import { ExtensionCodec, ExtensionCodecType } from "./ExtensionCodec.ts";import { getInt64, getUint64 } from "./utils/int.ts";import { TEXT_DECODER_THRESHOLD, utf8DecodeJs, utf8DecodeTD,} from "./utils/utf8.ts";import { createDataView, ensureUint8Array } from "./utils/typedArrays.ts";import { CachedKeyDecoder, KeyDecoder } from "./CachedKeyDecoder.ts";
const enum State { ARRAY, MAP_KEY, MAP_VALUE,}
type MapKeyType = string | number;
const isValidMapKeyType = (key: unknown): key is MapKeyType => { const keyType = typeof key;
return keyType === "string" || keyType === "number";};
type StackMapState = { type: State.MAP_KEY | State.MAP_VALUE; size: number; key: MapKeyType | null; readCount: number; map: Record<string, unknown>;};
type StackArrayState = { type: State.ARRAY; size: number; array: Array<unknown>; position: number;};
type StackState = StackArrayState | StackMapState;
const HEAD_BYTE_REQUIRED = -1;
const EMPTY_VIEW = new DataView(new ArrayBuffer(0));const EMPTY_BYTES = new Uint8Array(EMPTY_VIEW.buffer);
export const DataViewIndexOutOfBoundsError: typeof Error = (() => { try { EMPTY_VIEW.getInt8(0); } catch (e) { return e.constructor; } throw new Error("never reached");})();
const MORE_DATA = new DataViewIndexOutOfBoundsError("Insufficient data");
const DEFAULT_MAX_LENGTH = 0xffff_ffff;
const sharedCachedKeyDecoder = new CachedKeyDecoder();
export class Decoder<ContextType> { private totalPos = 0; private pos = 0;
private view = EMPTY_VIEW; private bytes = EMPTY_BYTES; private headByte = HEAD_BYTE_REQUIRED; private readonly stack: Array<StackState> = [];
public constructor( private readonly extensionCodec: ExtensionCodecType<ContextType> = ExtensionCodec.defaultCodec as any, private readonly context: ContextType = undefined as any, private readonly maxStrLength = DEFAULT_MAX_LENGTH, private readonly maxBinLength = DEFAULT_MAX_LENGTH, private readonly maxArrayLength = DEFAULT_MAX_LENGTH, private readonly maxMapLength = DEFAULT_MAX_LENGTH, private readonly maxExtLength = DEFAULT_MAX_LENGTH, private readonly keyDecoder: KeyDecoder | null = sharedCachedKeyDecoder, ) {}
private reinitializeState() { this.totalPos = 0; this.headByte = HEAD_BYTE_REQUIRED; }
private setBuffer( buffer: ArrayLike<number> | ArrayBuffer | BufferSource, ): void { this.bytes = ensureUint8Array(buffer); this.view = createDataView(this.bytes); this.pos = 0; }
private appendBuffer(buffer: ArrayLike<number> | BufferSource) { buffer = ensureUint8Array(buffer).slice(); if (this.headByte === HEAD_BYTE_REQUIRED && !this.hasRemaining()) { this.setBuffer(buffer); } else { const remainingData = this.bytes.subarray(this.pos); const newData = ensureUint8Array(buffer); const concated = new Uint8Array(remainingData.length + newData.length); concated.set(remainingData); concated.set(newData, remainingData.length); this.setBuffer(concated); } }
private hasRemaining(size = 1) { return this.view.byteLength - this.pos >= size; }
private createNoExtraBytesError(posToShow: number): Error { const { view, pos } = this; return new RangeError( `Extra ${view.byteLength - pos} of ${view.byteLength} byte(s) found at buffer[${posToShow}]`, ); }
public decode(buffer: ArrayLike<number> | ArrayBuffer): unknown { this.reinitializeState(); this.setBuffer(buffer); return this.doDecodeSingleSync(); }
private doDecodeSingleSync(): unknown { const object = this.doDecodeSync(); if (this.hasRemaining()) { throw this.createNoExtraBytesError(this.pos); } return object; }
public async decodeAsync( stream: AsyncIterable<ArrayLike<number> | BufferSource>, ): Promise<unknown> { let decoded = false; let object: unknown; for await (const buffer of stream) { if (decoded) { throw this.createNoExtraBytesError(this.totalPos); }
this.appendBuffer(buffer);
try { object = this.doDecodeSync(); decoded = true; } catch (e) { if (!(e instanceof DataViewIndexOutOfBoundsError)) { throw e; } } this.totalPos += this.pos; }
if (decoded) { if (this.hasRemaining()) { throw this.createNoExtraBytesError(this.totalPos); } return object; }
const { headByte, pos, totalPos } = this; throw new RangeError( `Insufficient data in parcing ${ prettyByte(headByte) } at ${totalPos} (${pos} in the current buffer)`, ); }
public decodeArrayStream(stream: AsyncIterable<ArrayLike<number>>) { return this.decodeMultiAsync(stream, true); }
public decodeStream(stream: AsyncIterable<ArrayLike<number>>) { return this.decodeMultiAsync(stream, false); }
private async *decodeMultiAsync( stream: AsyncIterable<ArrayLike<number>>, isArray: boolean, ) { let isArrayHeaderRequired = isArray; let arrayItemsLeft = -1;
for await (const buffer of stream) { if (isArray && arrayItemsLeft === 0) { throw this.createNoExtraBytesError(this.totalPos); }
this.appendBuffer(buffer);
if (isArrayHeaderRequired) { arrayItemsLeft = this.readArraySize(); isArrayHeaderRequired = false; this.complete(); }
try { while (true) { yield this.doDecodeSync(); if (--arrayItemsLeft === 0) { break; } } } catch (e) { if (!(e instanceof DataViewIndexOutOfBoundsError)) { throw e; } } this.totalPos += this.pos; } }
private doDecodeSync(): unknown { DECODE: while (true) { const headByte = this.readHeadByte(); let object: unknown;
if (headByte >= 0xe0) { object = headByte - 0x100; } else if (headByte < 0xc0) { if (headByte < 0x80) { object = headByte; } else if (headByte < 0x90) { const size = headByte - 0x80; if (size !== 0) { this.pushMapState(size); this.complete(); continue DECODE; } else { object = {}; } } else if (headByte < 0xa0) { const size = headByte - 0x90; if (size !== 0) { this.pushArrayState(size); this.complete(); continue DECODE; } else { object = []; } } else { const byteLength = headByte - 0xa0; object = this.decodeUtf8String(byteLength, 0); } } else if (headByte === 0xc0) { object = null; } else if (headByte === 0xc2) { object = false; } else if (headByte === 0xc3) { object = true; } else if (headByte === 0xca) { object = this.readF32(); } else if (headByte === 0xcb) { object = this.readF64(); } else if (headByte === 0xcc) { object = this.readU8(); } else if (headByte === 0xcd) { object = this.readU16(); } else if (headByte === 0xce) { object = this.readU32(); } else if (headByte === 0xcf) { object = this.readU64(); } else if (headByte === 0xd0) { object = this.readI8(); } else if (headByte === 0xd1) { object = this.readI16(); } else if (headByte === 0xd2) { object = this.readI32(); } else if (headByte === 0xd3) { object = this.readI64(); } else if (headByte === 0xd9) { const byteLength = this.lookU8(); object = this.decodeUtf8String(byteLength, 1); } else if (headByte === 0xda) { const byteLength = this.lookU16(); object = this.decodeUtf8String(byteLength, 2); } else if (headByte === 0xdb) { const byteLength = this.lookU32(); object = this.decodeUtf8String(byteLength, 4); } else if (headByte === 0xdc) { const size = this.readU16(); if (size !== 0) { this.pushArrayState(size); this.complete(); continue DECODE; } else { object = []; } } else if (headByte === 0xdd) { const size = this.readU32(); if (size !== 0) { this.pushArrayState(size); this.complete(); continue DECODE; } else { object = []; } } else if (headByte === 0xde) { const size = this.readU16(); if (size !== 0) { this.pushMapState(size); this.complete(); continue DECODE; } else { object = {}; } } else if (headByte === 0xdf) { const size = this.readU32(); if (size !== 0) { this.pushMapState(size); this.complete(); continue DECODE; } else { object = {}; } } else if (headByte === 0xc4) { const size = this.lookU8(); object = this.decodeBinary(size, 1); } else if (headByte === 0xc5) { const size = this.lookU16(); object = this.decodeBinary(size, 2); } else if (headByte === 0xc6) { const size = this.lookU32(); object = this.decodeBinary(size, 4); } else if (headByte === 0xd4) { object = this.decodeExtension(1, 0); } else if (headByte === 0xd5) { object = this.decodeExtension(2, 0); } else if (headByte === 0xd6) { object = this.decodeExtension(4, 0); } else if (headByte === 0xd7) { object = this.decodeExtension(8, 0); } else if (headByte === 0xd8) { object = this.decodeExtension(16, 0); } else if (headByte === 0xc7) { const size = this.lookU8(); object = this.decodeExtension(size, 1); } else if (headByte === 0xc8) { const size = this.lookU16(); object = this.decodeExtension(size, 2); } else if (headByte === 0xc9) { const size = this.lookU32(); object = this.decodeExtension(size, 4); } else { throw new Error(`Unrecognized type byte: ${prettyByte(headByte)}`); }
this.complete();
const stack = this.stack; while (stack.length > 0) { const state = stack[stack.length - 1]; if (state.type === State.ARRAY) { state.array[state.position] = object; state.position++; if (state.position === state.size) { stack.pop(); object = state.array; } else { continue DECODE; } } else if (state.type === State.MAP_KEY) { if (!isValidMapKeyType(object)) { throw new Error( "The type of key must be string or number but " + typeof object, ); }
state.key = object; state.type = State.MAP_VALUE; continue DECODE; } else {
state.map[state.key!] = object; state.readCount++;
if (state.readCount === state.size) { stack.pop(); object = state.map; } else { state.key = null; state.type = State.MAP_KEY; continue DECODE; } } }
return object; } }
private readHeadByte(): number { if (this.headByte === HEAD_BYTE_REQUIRED) { this.headByte = this.readU8(); }
return this.headByte; }
private complete(): void { this.headByte = HEAD_BYTE_REQUIRED; }
private readArraySize(): number { const headByte = this.readHeadByte();
switch (headByte) { case 0xdc: return this.readU16(); case 0xdd: return this.readU32(); default: { if (headByte < 0xa0) { return headByte - 0x90; } else { throw new Error( `Unrecognized array type byte: ${prettyByte(headByte)}`, ); } } } }
private pushMapState(size: number) { if (size > this.maxMapLength) { throw new Error( `Max length exceeded: map length (${size}) > maxMapLengthLength (${this.maxMapLength})`, ); }
this.stack.push({ type: State.MAP_KEY, size, key: null, readCount: 0, map: {}, }); }
private pushArrayState(size: number) { if (size > this.maxArrayLength) { throw new Error( `Max length exceeded: array length (${size}) > maxArrayLength (${this.maxArrayLength})`, ); }
this.stack.push({ type: State.ARRAY, size, array: new Array<unknown>(size), position: 0, }); }
private decodeUtf8String(byteLength: number, headerOffset: number): string { if (byteLength > this.maxStrLength) { throw new Error( `Max length exceeded: UTF-8 byte length (${byteLength}) > maxStrLength (${this.maxStrLength})`, ); }
if (this.bytes.byteLength < this.pos + headerOffset + byteLength) { throw MORE_DATA; }
const offset = this.pos + headerOffset; let object: string; if (this.stateIsMapKey() && this.keyDecoder?.canBeCached(byteLength)) { object = this.keyDecoder.decode(this.bytes, offset, byteLength); } else if (byteLength > TEXT_DECODER_THRESHOLD) { object = utf8DecodeTD(this.bytes, offset, byteLength); } else { object = utf8DecodeJs(this.bytes, offset, byteLength); } this.pos += headerOffset + byteLength; return object; }
private stateIsMapKey(): boolean { if (this.stack.length > 0) { const state = this.stack[this.stack.length - 1]; return state.type === State.MAP_KEY; } return false; }
private decodeBinary(byteLength: number, headOffset: number): Uint8Array { if (byteLength > this.maxBinLength) { throw new Error( `Max length exceeded: bin length (${byteLength}) > maxBinLength (${this.maxBinLength})`, ); }
if (!this.hasRemaining(byteLength + headOffset)) { throw MORE_DATA; }
const offset = this.pos + headOffset; const object = this.bytes.subarray(offset, offset + byteLength); this.pos += headOffset + byteLength; return object; }
private decodeExtension(size: number, headOffset: number): unknown { if (size > this.maxExtLength) { throw new Error( `Max length exceeded: ext length (${size}) > maxExtLength (${this.maxExtLength})`, ); }
const extType = this.view.getInt8(this.pos + headOffset); const data = this.decodeBinary(size, headOffset + 1 ); return this.extensionCodec.decode(data, extType, this.context); }
private lookU8() { return this.view.getUint8(this.pos); }
private lookU16() { return this.view.getUint16(this.pos); }
private lookU32() { return this.view.getUint32(this.pos); }
private readU8(): number { const value = this.view.getUint8(this.pos); this.pos++; return value; }
private readI8(): number { const value = this.view.getInt8(this.pos); this.pos++; return value; }
private readU16(): number { const value = this.view.getUint16(this.pos); this.pos += 2; return value; }
private readI16(): number { const value = this.view.getInt16(this.pos); this.pos += 2; return value; }
private readU32(): number { const value = this.view.getUint32(this.pos); this.pos += 4; return value; }
private readI32(): number { const value = this.view.getInt32(this.pos); this.pos += 4; return value; }
private readU64(): number { const value = getUint64(this.view, this.pos); this.pos += 8; return value; }
private readI64(): number { const value = getInt64(this.view, this.pos); this.pos += 8; return value; }
private readF32() { const value = this.view.getFloat32(this.pos); this.pos += 4; return value; }
private readF64() { const value = this.view.getFloat64(this.pos); this.pos += 8; return value; }}