@cloudpss/ubjson 0.5.39 → 0.5.41

package/src/helper/decode-ae.ts

@@ -0,0 +1,621 @@
+import { constants } from './constants.js';
+import { unsupportedType } from './errors.js';
+import { decode } from './string-decoder.js';
+import { toUint8Array } from './utils.js';
+const { defineProperty } = Object;
+const { fromCharCode } = String;
+
+/**
+ * 数据读取函数
+ * `yield` 返回需要的字节数，`return` 返回解析结果
+ */
+export type DecodeFuncAe<T = unknown> = Generator<number, T, void>;
+
+/** 数据包装 */
+export interface DecodeCursorAe {
+    /** 当前数据块 */
+    readonly view: DataView;
+    /** 当前数据块 */
+    readonly data: Uint8Array;
+    /** 当前数据大小 */
+    readonly size: number;
+    /** 当前读指针位置 */
+    offset: number;
+}
+
+/** 创建数据包装 */
+export function DecodeCursor(data: BinaryData): DecodeCursorAe {
+    const d = toUint8Array(data);
+    const v = new DataView(d.buffer, d.byteOffset, d.byteLength);
+    return {
+        data: d,
+        view: v,
+        size: v.byteLength,
+        offset: 0,
+    };
+}
+
+/**
+ * 读取第一个非 NOOP 的字节
+ */
+export function* readMarker(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    let marker: number;
+    do {
+        if (cursor.offset + 1 > cursor.size) {
+            yield cursor.offset - cursor.size + 1;
+        }
+        marker = cursor.data[cursor.offset++]!;
+    } while (marker === constants.NO_OP);
+    return marker;
+}
+
+/** 读取一个大于 0 的整数 */
+export function* readLength(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    let marker: number;
+    do {
+        if (cursor.offset + 1 > cursor.size) {
+            yield cursor.offset - cursor.size + 1;
+        }
+        marker = cursor.data[cursor.offset++]!;
+    } while (marker === constants.NO_OP);
+    let length: number;
+    switch (marker) {
+        case constants.INT8:
+            if (cursor.offset + 1 > cursor.size) {
+                yield cursor.offset - cursor.size + 1;
+            }
+            length = cursor.view.getInt8(cursor.offset++);
+            break;
+        case constants.UINT8:
+            length = yield* readUint8Data(cursor);
+            break;
+        case constants.INT16:
+            length = yield* readInt16Data(cursor);
+            break;
+        case constants.INT32:
+            length = yield* readInt32Data(cursor);
+            break;
+        case constants.INT64: {
+            const l = yield* readInt64Data(cursor);
+            if (l < 0 || l > Number.MAX_SAFE_INTEGER) {
+                throw new Error('Invalid length');
+            }
+            length = Number(l);
+            break;
+        }
+        default:
+            throw new Error(`Unexpected marker '${fromCharCode(marker)}'(${marker}) for int length`);
+    }
+    if (length < 0) {
+        throw new Error('Invalid length');
+    }
+    return length;
+}
+
+/** readInt8Data */
+export function* readInt8Data(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    if (cursor.offset + 1 > cursor.size) {
+        yield cursor.offset - cursor.size + 1;
+    }
+    return cursor.view.getInt8(cursor.offset++);
+}
+/** readUint8Data */
+export function* readUint8Data(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    if (cursor.offset + 1 > cursor.size) {
+        yield cursor.offset - cursor.size + 1;
+    }
+    return cursor.data[cursor.offset++]!;
+}
+/** readInt16Data */
+export function* readInt16Data(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    if (cursor.offset + 2 > cursor.size) {
+        yield cursor.offset - cursor.size + 2;
+    }
+    const value = cursor.view.getInt16(cursor.offset);
+    cursor.offset += 2;
+    return value;
+}
+/** readInt32Data */
+export function* readInt32Data(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    if (cursor.offset + 4 > cursor.size) {
+        yield cursor.offset - cursor.size + 4;
+    }
+    const value = cursor.view.getInt32(cursor.offset);
+    cursor.offset += 4;
+    return value;
+}
+/** readInt64Data */
+export function* readInt64Data(cursor: DecodeCursorAe): DecodeFuncAe<bigint> {
+    if (cursor.offset + 8 > cursor.size) {
+        yield cursor.offset - cursor.size + 8;
+    }
+    const value = cursor.view.getBigInt64(cursor.offset);
+    cursor.offset += 8;
+    return value;
+}
+/** readFloat32Data */
+export function* readFloat32Data(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    if (cursor.offset + 4 > cursor.size) {
+        yield cursor.offset - cursor.size + 4;
+    }
+    const value = cursor.view.getFloat32(cursor.offset);
+    cursor.offset += 4;
+    return value;
+}
+/** readFloat64Data */
+export function* readFloat64Data(cursor: DecodeCursorAe): DecodeFuncAe<number> {
+    if (cursor.offset + 8 > cursor.size) {
+        yield cursor.offset - cursor.size + 8;
+    }
+    const value = cursor.view.getFloat64(cursor.offset);
+    cursor.offset += 8;
+    return value;
+}
+
+/** 读取数据 */
+export function* read(cursor: DecodeCursorAe): DecodeFuncAe<unknown> {
+    let marker: number;
+    do {
+        if (cursor.offset + 1 > cursor.size) {
+            yield cursor.offset - cursor.size + 1;
+        }
+        marker = cursor.data[cursor.offset++]!;
+    } while (marker === constants.NO_OP);
+    return yield* readData(cursor, marker);
+}
+
+/** 读取优化对象数据 */
+function* readObjectOptimizedData(cursor: DecodeCursorAe, marker: OptimizedFormatMarkers): DecodeFuncAe<unknown> {
+    const { count, type } = marker;
+    const object: Record<string, unknown> = {};
+    for (let i = 0; i < count; i++) {
+        const key = yield* readKey(cursor);
+        const value = yield* readData(cursor, type ?? (yield* readMarker(cursor)));
+        if (key === '__proto__') {
+            defineProperty(object, key, { value, enumerable: true, configurable: true, writable: true });
+        }
+        object[key] = value;
+    }
+    return object;
+}
+
+/** 根据标签读取后续数据 */
+export function* readData(cursor: DecodeCursorAe, marker: number): DecodeFuncAe<unknown> {
+    // 按照出现频率排序
+    switch (marker) {
+        case constants.STRING: {
+            let marker: number;
+            do {
+                if (cursor.offset + 1 > cursor.size) {
+                    yield cursor.offset - cursor.size + 1;
+                }
+                marker = cursor.data[cursor.offset++]!;
+            } while (marker === constants.NO_OP);
+            let length: number;
+            switch (marker) {
+                case constants.INT8:
+                    if (cursor.offset + 1 > cursor.size) {
+                        yield cursor.offset - cursor.size + 1;
+                    }
+                    length = cursor.view.getInt8(cursor.offset++);
+                    break;
+                case constants.UINT8:
+                    length = yield* readUint8Data(cursor);
+                    break;
+                case constants.INT16:
+                    length = yield* readInt16Data(cursor);
+                    break;
+                case constants.INT32:
+                    length = yield* readInt32Data(cursor);
+                    break;
+                case constants.INT64: {
+                    const l = yield* readInt64Data(cursor);
+                    if (l < 0 || l > Number.MAX_SAFE_INTEGER) {
+                        throw new Error('Invalid length');
+                    }
+                    length = Number(l);
+                    break;
+                }
+                default:
+                    throw new Error(`Unexpected marker '${fromCharCode(marker)}'(${marker}) for int length`);
+            }
+            if (length < 0) {
+                throw new Error('Invalid length');
+            }
+            if (cursor.offset + length > cursor.size) {
+                yield cursor.offset - cursor.size + length;
+            }
+            const begin = cursor.offset;
+            const end = begin + length;
+            cursor.offset = end;
+            const { data } = cursor;
+            return decode(data, begin, end);
+        }
+        case constants.OBJECT: {
+            let type;
+            let count;
+            let marker: number;
+            do {
+                if (cursor.offset + 1 > cursor.size) {
+                    yield cursor.offset - cursor.size + 1;
+                }
+                marker = cursor.data[cursor.offset++]!;
+            } while (marker === constants.NO_OP);
+            switch (marker) {
+                case constants.TYPE_MARKER: {
+                    do {
+                        if (cursor.offset + 1 > cursor.size) {
+                            yield cursor.offset - cursor.size + 1;
+                        }
+                        type = cursor.data[cursor.offset++]!;
+                    } while (type === constants.NO_OP);
+                    let marker2;
+                    do {
+                        if (cursor.offset + 1 > cursor.size) {
+                            yield cursor.offset - cursor.size + 1;
+                        }
+                        marker2 = cursor.data[cursor.offset++]!;
+                    } while (marker2 === constants.NO_OP);
+                    if (marker2 !== constants.COUNT_MARKER) {
+                        throw new Error('Expected count marker');
+                    }
+                }
+                /* fall through */
+                case constants.COUNT_MARKER: {
+                    count = yield* readLength(cursor);
+                    break;
+                }
+                default: {
+                    // 不是 '$' 或 '#'
+                    // 直到 '}'
+                    const object: Record<string, unknown> = {};
+                    for (;;) {
+                        while (marker === constants.NO_OP) {
+                            if (cursor.offset + 1 > cursor.size) {
+                                yield cursor.offset - cursor.size + 1;
+                            }
+                            marker = cursor.data[cursor.offset++]!;
+                        }
+                        if (marker === constants.OBJECT_END) break;
+                        let length: number;
+                        switch (marker) {
+                            case constants.INT8:
+                                if (cursor.offset + 1 > cursor.size) {
+                                    yield cursor.offset - cursor.size + 1;
+                                }
+                                length = cursor.view.getInt8(cursor.offset++);
+                                break;
+                            case constants.UINT8:
+                                length = yield* readUint8Data(cursor);
+                                break;
+                            case constants.INT16:
+                                length = yield* readInt16Data(cursor);
+                                break;
+                            case constants.INT32:
+                                length = yield* readInt32Data(cursor);
+                                break;
+                            case constants.INT64: {
+                                const l = yield* readInt64Data(cursor);
+                                if (l < 0 || l > Number.MAX_SAFE_INTEGER) {
+                                    throw new Error('Invalid length');
+                                }
+                                length = Number(l);
+                                break;
+                            }
+                            default:
+                                throw new Error(
+                                    `Unexpected marker '${fromCharCode(marker)}'(${marker}) for int length`,
+                                );
+                        }
+                        if (length < 0) {
+                            throw new Error('Invalid length');
+                        }
+                        if (cursor.offset + length > cursor.size) {
+                            yield cursor.offset - cursor.size + length;
+                        }
+                        marker = constants.NO_OP;
+                        const begin = cursor.offset;
+                        const end = begin + length;
+                        cursor.offset = end;
+                        const { data } = cursor;
+                        const key = decode(data, begin, end);
+                        let valueMarker: number;
+                        do {
+                            if (cursor.offset + 1 > cursor.size) {
+                                yield cursor.offset - cursor.size + 1;
+                            }
+                            valueMarker = cursor.data[cursor.offset++]!;
+                        } while (valueMarker === constants.NO_OP);
+                        const value = yield* readData(cursor, valueMarker);
+                        if (key === '__proto__') {
+                            defineProperty(object, key, {
+                                value,
+                                enumerable: true,
+                                configurable: true,
+                                writable: true,
+                            });
+                        }
+                        object[key] = value;
+                    }
+                    return object;
+                }
+            }
+            return yield* readObjectOptimizedData(cursor, { type, count });
+        }
+        case constants.ARRAY: {
+            let type;
+            let count;
+            let marker: number;
+            do {
+                if (cursor.offset + 1 > cursor.size) {
+                    yield cursor.offset - cursor.size + 1;
+                }
+                marker = cursor.data[cursor.offset++]!;
+            } while (marker === constants.NO_OP);
+            switch (marker) {
+                case constants.TYPE_MARKER: {
+                    do {
+                        if (cursor.offset + 1 > cursor.size) {
+                            yield cursor.offset - cursor.size + 1;
+                        }
+                        type = cursor.data[cursor.offset++]!;
+                    } while (type === constants.NO_OP);
+                    let marker2;
+                    do {
+                        if (cursor.offset + 1 > cursor.size) {
+                            yield cursor.offset - cursor.size + 1;
+                        }
+                        marker2 = cursor.data[cursor.offset++]!;
+                    } while (marker2 === constants.NO_OP);
+                    if (marker2 !== constants.COUNT_MARKER) {
+                        throw new Error('Expected count marker');
+                    }
+                }
+                /* fall through */
+                case constants.COUNT_MARKER: {
+                    count = yield* readLength(cursor);
+                    break;
+                }
+                default: {
+                    // 不是 '$' 或 '#'
+                    const array = [];
+                    for (;;) {
+                        while (marker === constants.NO_OP) {
+                            if (cursor.offset + 1 > cursor.size) {
+                                yield cursor.offset - cursor.size + 1;
+                            }
+                            marker = cursor.data[cursor.offset++]!;
+                        }
+                        // 直到 ']'
+                        if (marker === constants.ARRAY_END) break;
+                        array.push(yield* readData(cursor, marker));
+                        marker = constants.NO_OP;
+                    }
+                    return array;
+                }
+            }
+            switch (type) {
+                case constants.UINT8: {
+                    if (cursor.offset + count > cursor.size) {
+                        yield cursor.offset - cursor.size + count;
+                    }
+                    const buf = new Uint8Array(
+                        cursor.data.buffer,
+                        cursor.data.byteOffset + cursor.offset,
+                        count,
+                    ).slice();
+                    cursor.offset += count;
+                    return buf;
+                }
+                case constants.INT8: {
+                    if (cursor.offset + count > cursor.size) {
+                        yield cursor.offset - cursor.size + count;
+                    }
+                    const buf = new Int8Array(
+                        cursor.data.buffer,
+                        cursor.data.byteOffset + cursor.offset,
+                        count,
+                    ).slice();
+                    cursor.offset += count;
+                    return buf;
+                }
+                case constants.INT16: {
+                    if (cursor.offset + count * 2 > cursor.size) {
+                        yield cursor.offset - cursor.size + count * 2;
+                    }
+                    const result = new Int16Array(count);
+                    for (let i = 0; i < count; i++) {
+                        result[i] = yield* readInt16Data(cursor);
+                    }
+                    return result;
+                }
+                case constants.INT32: {
+                    if (cursor.offset + count * 4 > cursor.size) {
+                        yield cursor.offset - cursor.size + count * 4;
+                    }
+                    const result = new Int32Array(count);
+                    for (let i = 0; i < count; i++) {
+                        result[i] = yield* readInt32Data(cursor);
+                    }
+                    return result;
+                }
+                case constants.FLOAT32: {
+                    if (cursor.offset + count * 4 > cursor.size) {
+                        yield cursor.offset - cursor.size + count * 4;
+                    }
+                    const result = new Float32Array(count);
+                    for (let i = 0; i < count; i++) {
+                        result[i] = yield* readFloat32Data(cursor);
+                    }
+                    return result;
+                }
+                case constants.FLOAT64: {
+                    if (cursor.offset + count * 8 > cursor.size) {
+                        yield cursor.offset - cursor.size + count * 8;
+                    }
+                    const result = new Float64Array(count);
+                    for (let i = 0; i < count; i++) {
+                        result[i] = yield* readFloat64Data(cursor);
+                    }
+                    return result;
+                }
+                case constants.INT64: {
+                    if (cursor.offset + count * 8 > cursor.size) {
+                        yield cursor.offset - cursor.size + count * 8;
+                    }
+                    const result = new BigInt64Array(count);
+                    for (let i = 0; i < count; i++) {
+                        result[i] = yield* readInt64Data(cursor);
+                    }
+                    return result;
+                }
+                case constants.NULL:
+                    return Array.from({ length: count }).fill(null);
+                case constants.TRUE:
+                    return Array.from({ length: count }).fill(true);
+                case constants.FALSE:
+                    return Array.from({ length: count }).fill(false);
+                default:
+                    break;
+            }
+            const array: unknown[] = [];
+            array.length = count;
+            for (let i = 0; i < count; i++) {
+                array[i] = type === undefined ? yield* read(cursor) : yield* readData(cursor, type);
+            }
+            return array;
+        }
+        case constants.FLOAT64: {
+            if (cursor.offset + 8 > cursor.size) {
+                yield cursor.offset - cursor.size + 8;
+            }
+            const value = cursor.view.getFloat64(cursor.offset);
+            cursor.offset += 8;
+            return value;
+        }
+        case constants.UINT8: {
+            if (cursor.offset + 1 > cursor.size) {
+                yield cursor.offset - cursor.size + 1;
+            }
+            return cursor.data[cursor.offset++]!;
+        }
+        case constants.INT16: {
+            if (cursor.offset + 2 > cursor.size) {
+                yield cursor.offset - cursor.size + 2;
+            }
+            const value = cursor.view.getInt16(cursor.offset);
+            cursor.offset += 2;
+            return value;
+        }
+        case constants.FLOAT32: {
+            if (cursor.offset + 4 > cursor.size) {
+                yield cursor.offset - cursor.size + 4;
+            }
+            const value = cursor.view.getFloat32(cursor.offset);
+            cursor.offset += 4;
+            return value;
+        }
+        case constants.CHAR: {
+            if (cursor.offset + 1 > cursor.size) {
+                yield cursor.offset - cursor.size + 1;
+            }
+            return fromCharCode(cursor.data[cursor.offset++]!);
+        }
+        case constants.INT32: {
+            if (cursor.offset + 4 > cursor.size) {
+                yield cursor.offset - cursor.size + 4;
+            }
+            const value = cursor.view.getInt32(cursor.offset);
+            cursor.offset += 4;
+            return value;
+        }
+        case constants.INT8: {
+            if (cursor.offset + 1 > cursor.size) {
+                yield cursor.offset - cursor.size + 1;
+            }
+            return cursor.view.getInt8(cursor.offset++);
+        }
+        case constants.NULL:
+            return null;
+        case constants.TRUE:
+            return true;
+        case constants.FALSE:
+            return false;
+        case constants.INT64: {
+            if (cursor.offset + 8 > cursor.size) {
+                yield cursor.offset - cursor.size + 8;
+            }
+            const value = cursor.view.getBigInt64(cursor.offset);
+            cursor.offset += 8;
+            if (value < Number.MIN_SAFE_INTEGER || value > Number.MAX_SAFE_INTEGER) {
+                return value;
+            }
+            return Number(value);
+        }
+        case constants.HIGH_PRECISION_NUMBER: {
+            const length = yield* readLength(cursor);
+            if (cursor.offset + length > cursor.size) {
+                yield cursor.offset - cursor.size + length;
+            }
+            const begin = cursor.offset;
+            const _buffer = new Uint8Array(cursor.data.buffer, begin, length).slice();
+            cursor.offset = begin + length;
+            // return _buffer
+            unsupportedType('high precision number');
+        }
+    }
+    throw new Error(`Unexpected marker '${fromCharCode(marker)}'(${marker})`);
+}
+
+/** readKey */
+export function* readKey(cursor: DecodeCursorAe): DecodeFuncAe<string> {
+    let marker: number;
+    do {
+        if (cursor.offset + 1 > cursor.size) {
+            yield cursor.offset - cursor.size + 1;
+        }
+        marker = cursor.data[cursor.offset++]!;
+    } while (marker === constants.NO_OP);
+    let length: number;
+    switch (marker) {
+        case constants.INT8:
+            if (cursor.offset + 1 > cursor.size) {
+                yield cursor.offset - cursor.size + 1;
+            }
+            length = cursor.view.getInt8(cursor.offset++);
+            break;
+        case constants.UINT8:
+            length = yield* readUint8Data(cursor);
+            break;
+        case constants.INT16:
+            length = yield* readInt16Data(cursor);
+            break;
+        case constants.INT32:
+            length = yield* readInt32Data(cursor);
+            break;
+        case constants.INT64: {
+            const l = yield* readInt64Data(cursor);
+            if (l < 0 || l > Number.MAX_SAFE_INTEGER) {
+                throw new Error('Invalid length');
+            }
+            length = Number(l);
+            break;
+        }
+        default:
+            throw new Error(`Unexpected marker '${fromCharCode(marker)}'(${marker}) for int length`);
+    }
+    if (length < 0) {
+        throw new Error('Invalid length');
+    }
+    if (cursor.offset + length > cursor.size) {
+        yield cursor.offset - cursor.size + length;
+    }
+    const begin = cursor.offset;
+    const end = begin + length;
+    cursor.offset = end;
+    const { data } = cursor;
+    return decode(data, begin, end);
+}
+
+/** Optimized Format 数据 */
+export type OptimizedFormatMarkers = { type?: number; count: number };