Init indexed file
This commit is contained in:
92
lib/dat/binary-helpers.ts
Normal file
92
lib/dat/binary-helpers.ts
Normal file
@@ -0,0 +1,92 @@
|
||||
export class BinaryWriter {
|
||||
private chunks: Buffer[] = [];
|
||||
|
||||
writeUInt32(value: number): this {
|
||||
const buf = Buffer.alloc(4);
|
||||
buf.writeUInt32LE(value, 0);
|
||||
this.chunks.push(buf);
|
||||
return this;
|
||||
}
|
||||
|
||||
writeDouble(value: number): this {
|
||||
const buf = Buffer.alloc(8);
|
||||
buf.writeDoubleLE(value, 0);
|
||||
this.chunks.push(buf);
|
||||
return this;
|
||||
}
|
||||
|
||||
writeBigUInt64(value: bigint): this {
|
||||
const buf = Buffer.alloc(8);
|
||||
buf.writeBigUInt64LE(value, 0);
|
||||
this.chunks.push(buf);
|
||||
return this;
|
||||
}
|
||||
|
||||
writeString(value: string): this {
|
||||
const strBuf = Buffer.from(value, 'utf8');
|
||||
this.writeUInt32(strBuf.length);
|
||||
this.chunks.push(strBuf);
|
||||
return this;
|
||||
}
|
||||
|
||||
writeNumberArray(values: number[]): this {
|
||||
this.writeUInt32(values.length);
|
||||
for (const v of values) this.writeDouble(v);
|
||||
return this;
|
||||
}
|
||||
|
||||
writeStringArray(values: string[]): this {
|
||||
this.writeUInt32(values.length);
|
||||
for (const v of values) this.writeString(v);
|
||||
return this;
|
||||
}
|
||||
|
||||
toBuffer(): Buffer {
|
||||
return Buffer.concat(this.chunks);
|
||||
}
|
||||
}
|
||||
|
||||
export class BinaryReader {
|
||||
private offset = 0;
|
||||
|
||||
constructor(private buf: Buffer) {}
|
||||
|
||||
readUInt32(): number {
|
||||
const v = this.buf.readUInt32LE(this.offset);
|
||||
this.offset += 4;
|
||||
return v;
|
||||
}
|
||||
|
||||
readDouble(): number {
|
||||
const v = this.buf.readDoubleLE(this.offset);
|
||||
this.offset += 8;
|
||||
return v;
|
||||
}
|
||||
|
||||
readBigUInt64(): bigint {
|
||||
const v = this.buf.readBigUInt64LE(this.offset);
|
||||
this.offset += 8;
|
||||
return v;
|
||||
}
|
||||
|
||||
readString(): string {
|
||||
const len = this.readUInt32();
|
||||
const v = this.buf.toString('utf8', this.offset, this.offset + len);
|
||||
this.offset += len;
|
||||
return v;
|
||||
}
|
||||
|
||||
readNumberArray(): number[] {
|
||||
const len = this.readUInt32();
|
||||
const arr: number[] = [];
|
||||
for (let i = 0; i < len; i++) arr.push(this.readDouble());
|
||||
return arr;
|
||||
}
|
||||
|
||||
readStringArray(): string[] {
|
||||
const len = this.readUInt32();
|
||||
const arr: string[] = [];
|
||||
for (let i = 0; i < len; i++) arr.push(this.readString());
|
||||
return arr;
|
||||
}
|
||||
}
|
||||
5
lib/dat/constants.ts
Normal file
5
lib/dat/constants.ts
Normal file
@@ -0,0 +1,5 @@
|
||||
// src/data-file/constants.ts
|
||||
export const DATA_MAGIC = 'DATA';
|
||||
export const DATA_VERSION = 1;
|
||||
export const DATA_HEADER_SIZE = 64;
|
||||
export const RECORD_HEADER_SIZE = 8;
|
||||
7
lib/dat/index.ts
Normal file
7
lib/dat/index.ts
Normal file
@@ -0,0 +1,7 @@
|
||||
// src/data-file/index.ts
|
||||
export { DataWriter } from './writer.js';
|
||||
export { DataReader } from './reader.js';
|
||||
export { DataProtocol } from './protocol.js';
|
||||
export * from './types.js';
|
||||
export * from './constants.js';
|
||||
export * from './serializers.js';
|
||||
80
lib/dat/protocol.ts
Normal file
80
lib/dat/protocol.ts
Normal file
@@ -0,0 +1,80 @@
|
||||
// src/data-file/protocol.ts
|
||||
import { DATA_MAGIC, DATA_VERSION, DATA_HEADER_SIZE, RECORD_HEADER_SIZE } from './constants.js';
|
||||
import { crc32 } from '../idx/index.js';
|
||||
import type { Serializer } from './types.js';
|
||||
|
||||
export interface DataHeader {
|
||||
magic: string;
|
||||
version: number;
|
||||
createdAt: bigint;
|
||||
fileSize: bigint;
|
||||
recordCount: number;
|
||||
reserved: Buffer;
|
||||
}
|
||||
|
||||
export class DataProtocol {
|
||||
static createHeader(): Buffer {
|
||||
const buf = Buffer.alloc(DATA_HEADER_SIZE);
|
||||
buf.write(DATA_MAGIC, 0, 4, 'ascii');
|
||||
buf.writeUInt32LE(DATA_VERSION, 4);
|
||||
buf.writeBigUInt64LE(BigInt(Date.now()) * 1000000n, 8);
|
||||
buf.writeBigUInt64LE(BigInt(DATA_HEADER_SIZE), 16);
|
||||
buf.writeUInt32LE(0, 24);
|
||||
return buf;
|
||||
}
|
||||
|
||||
static readHeader(buf: Buffer): DataHeader {
|
||||
return {
|
||||
magic: buf.toString('ascii', 0, 4),
|
||||
version: buf.readUInt32LE(4),
|
||||
createdAt: buf.readBigUInt64LE(8),
|
||||
fileSize: buf.readBigUInt64LE(16),
|
||||
recordCount: buf.readUInt32LE(24),
|
||||
reserved: buf.subarray(28, 64),
|
||||
};
|
||||
}
|
||||
|
||||
static updateHeader(buf: Buffer, fileSize: bigint, recordCount: number): void {
|
||||
buf.writeBigUInt64LE(fileSize, 16);
|
||||
buf.writeUInt32LE(recordCount, 24);
|
||||
}
|
||||
|
||||
static serializeRecord<T>(data: T, serializer: Serializer<T>): Buffer {
|
||||
const dataBytes = serializer.serialize(data);
|
||||
const totalLen = RECORD_HEADER_SIZE + dataBytes.length;
|
||||
|
||||
const buf = Buffer.alloc(totalLen);
|
||||
|
||||
dataBytes.copy(buf, RECORD_HEADER_SIZE);
|
||||
|
||||
buf.writeUInt32LE(dataBytes.length, 0);
|
||||
const checksum = crc32(buf, RECORD_HEADER_SIZE, totalLen);
|
||||
buf.writeUInt32LE(checksum, 4);
|
||||
|
||||
return buf;
|
||||
}
|
||||
|
||||
static deserializeRecord<T>(
|
||||
buf: Buffer,
|
||||
offset: number,
|
||||
serializer: Serializer<T>
|
||||
): { data: T; length: number } | null {
|
||||
if (offset + RECORD_HEADER_SIZE > buf.length) return null;
|
||||
|
||||
const dataLen = buf.readUInt32LE(offset);
|
||||
const storedChecksum = buf.readUInt32LE(offset + 4);
|
||||
|
||||
const totalLen = RECORD_HEADER_SIZE + dataLen;
|
||||
if (offset + totalLen > buf.length) return null;
|
||||
|
||||
const calcChecksum = crc32(buf, offset + RECORD_HEADER_SIZE, offset + totalLen);
|
||||
if (calcChecksum !== storedChecksum) {
|
||||
throw new Error(`Checksum mismatch at offset ${offset}`);
|
||||
}
|
||||
|
||||
const dataBytes = buf.subarray(offset + RECORD_HEADER_SIZE, offset + totalLen);
|
||||
const data = serializer.deserialize(dataBytes);
|
||||
|
||||
return { data, length: totalLen };
|
||||
}
|
||||
}
|
||||
213
lib/dat/reader.ts
Normal file
213
lib/dat/reader.ts
Normal file
@@ -0,0 +1,213 @@
|
||||
// src/data-file/reader.ts
|
||||
import * as fs from 'node:fs';
|
||||
import mmap from '@elilee/mmap-native';
|
||||
import { DATA_HEADER_SIZE } from './constants.js';
|
||||
import { DataProtocol, DataHeader } from './protocol.js';
|
||||
import { IndexReader } from '../idx/index.js';
|
||||
import type { Serializer, DataEntry } from './types.js';
|
||||
|
||||
export class DataReader<T> {
|
||||
private fd: number | null = null;
|
||||
private buffer: Buffer | null = null;
|
||||
private header: DataHeader | null = null;
|
||||
|
||||
private indexReader: IndexReader;
|
||||
private serializer: Serializer<T>;
|
||||
|
||||
readonly dataPath: string;
|
||||
readonly indexPath: string;
|
||||
|
||||
constructor(basePath: string, serializer: Serializer<T>) {
|
||||
this.dataPath = `${basePath}.dat`;
|
||||
this.indexPath = `${basePath}.idx`;
|
||||
this.serializer = serializer;
|
||||
this.indexReader = new IndexReader(this.indexPath);
|
||||
}
|
||||
|
||||
open(): void {
|
||||
const stats = fs.statSync(this.dataPath);
|
||||
this.fd = fs.openSync(this.dataPath, 'r');
|
||||
|
||||
this.buffer = mmap.map(
|
||||
stats.size,
|
||||
mmap.PROT_READ,
|
||||
mmap.MAP_SHARED,
|
||||
this.fd,
|
||||
0
|
||||
);
|
||||
|
||||
this.header = DataProtocol.readHeader(this.buffer);
|
||||
this.indexReader.open();
|
||||
}
|
||||
|
||||
getHeader(): DataHeader {
|
||||
if (!this.header) throw new Error('Data file not opened');
|
||||
return this.header;
|
||||
}
|
||||
|
||||
getBySequence(sequence: number): DataEntry<T> | null {
|
||||
if (!this.buffer) throw new Error('Data file not opened');
|
||||
|
||||
const found = this.indexReader.binarySearchBySequence(sequence);
|
||||
if (!found) return null;
|
||||
|
||||
const result = DataProtocol.deserializeRecord(
|
||||
this.buffer,
|
||||
Number(found.entry.offset),
|
||||
this.serializer
|
||||
);
|
||||
if (!result) return null;
|
||||
|
||||
return {
|
||||
sequence: found.entry.sequence,
|
||||
timestamp: found.entry.timestamp,
|
||||
data: result.data,
|
||||
};
|
||||
}
|
||||
|
||||
getByIndex(index: number): DataEntry<T> | null {
|
||||
if (!this.buffer) throw new Error('Data file not opened');
|
||||
|
||||
const entry = this.indexReader.getEntry(index);
|
||||
if (!entry) return null;
|
||||
|
||||
const result = DataProtocol.deserializeRecord(
|
||||
this.buffer,
|
||||
Number(entry.offset),
|
||||
this.serializer
|
||||
);
|
||||
if (!result) return null;
|
||||
|
||||
return {
|
||||
sequence: entry.sequence,
|
||||
timestamp: entry.timestamp,
|
||||
data: result.data,
|
||||
};
|
||||
}
|
||||
|
||||
getBulkData(startSeq: number, endSeq: number): DataEntry<T>[] {
|
||||
if (!this.buffer) throw new Error('Data file not opened');
|
||||
|
||||
const results: DataEntry<T>[] = [];
|
||||
const indexHeader = this.indexReader.getHeader();
|
||||
let startIdx = this.findStartIndex(startSeq, indexHeader.validCount);
|
||||
|
||||
for (let i = startIdx; i < indexHeader.validCount; i++) {
|
||||
const entry = this.indexReader.getEntry(i);
|
||||
if (!entry) continue;
|
||||
|
||||
if (entry.sequence > endSeq) break;
|
||||
|
||||
if (entry.sequence >= startSeq) {
|
||||
const result = DataProtocol.deserializeRecord(
|
||||
this.buffer,
|
||||
Number(entry.offset),
|
||||
this.serializer
|
||||
);
|
||||
if (result) {
|
||||
results.push({
|
||||
sequence: entry.sequence,
|
||||
timestamp: entry.timestamp,
|
||||
data: result.data,
|
||||
});
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return results;
|
||||
}
|
||||
|
||||
private findStartIndex(targetSeq: number, validCount: number): number {
|
||||
let left = 0;
|
||||
let right = validCount - 1;
|
||||
let result = 0;
|
||||
|
||||
while (left <= right) {
|
||||
const mid = Math.floor((left + right) / 2);
|
||||
const entry = this.indexReader.getEntry(mid);
|
||||
|
||||
if (!entry) {
|
||||
right = mid - 1;
|
||||
continue;
|
||||
}
|
||||
|
||||
if (entry.sequence >= targetSeq) {
|
||||
result = mid;
|
||||
right = mid - 1;
|
||||
} else {
|
||||
left = mid + 1;
|
||||
}
|
||||
}
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
getBulkDataByTime(startTs: bigint, endTs: bigint): DataEntry<T>[] {
|
||||
if (!this.buffer) throw new Error('Data file not opened');
|
||||
|
||||
const indexResults = this.indexReader.findByTimeRange(startTs, endTs);
|
||||
const results: DataEntry<T>[] = [];
|
||||
|
||||
for (const { entry } of indexResults) {
|
||||
const result = DataProtocol.deserializeRecord(
|
||||
this.buffer,
|
||||
Number(entry.offset),
|
||||
this.serializer
|
||||
);
|
||||
if (result) {
|
||||
results.push({
|
||||
sequence: entry.sequence,
|
||||
timestamp: entry.timestamp,
|
||||
data: result.data,
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
return results;
|
||||
}
|
||||
|
||||
getAllData(): DataEntry<T>[] {
|
||||
if (!this.buffer) throw new Error('Data file not opened');
|
||||
|
||||
const entries = this.indexReader.getAllEntries();
|
||||
const results: DataEntry<T>[] = [];
|
||||
|
||||
for (const entry of entries) {
|
||||
const result = DataProtocol.deserializeRecord(
|
||||
this.buffer,
|
||||
Number(entry.offset),
|
||||
this.serializer
|
||||
);
|
||||
if (result) {
|
||||
results.push({
|
||||
sequence: entry.sequence,
|
||||
timestamp: entry.timestamp,
|
||||
data: result.data,
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
return results;
|
||||
}
|
||||
|
||||
getRecordCount(): number {
|
||||
return this.indexReader.getHeader().validCount;
|
||||
}
|
||||
|
||||
getLastSequence(): number {
|
||||
return this.indexReader.getHeader().lastSequence;
|
||||
}
|
||||
|
||||
close(): void {
|
||||
if (this.buffer) {
|
||||
mmap.unmap(this.buffer);
|
||||
this.buffer = null;
|
||||
}
|
||||
if (this.fd !== null) {
|
||||
fs.closeSync(this.fd);
|
||||
this.fd = null;
|
||||
}
|
||||
this.header = null;
|
||||
this.indexReader.close();
|
||||
}
|
||||
}
|
||||
32
lib/dat/serializers.ts
Normal file
32
lib/dat/serializers.ts
Normal file
@@ -0,0 +1,32 @@
|
||||
// src/data-file/serializers.ts
|
||||
import type { Serializer } from './types.js';
|
||||
|
||||
export function jsonSerializer<T>(): Serializer<T> {
|
||||
return {
|
||||
serialize(data: T): Buffer {
|
||||
return Buffer.from(JSON.stringify(data), 'utf8');
|
||||
},
|
||||
deserialize(buf: Buffer): T {
|
||||
return JSON.parse(buf.toString('utf8'));
|
||||
},
|
||||
};
|
||||
}
|
||||
|
||||
export function msgpackSerializer<T>(): Serializer<T> {
|
||||
const { encode, decode } = require('@msgpack/msgpack');
|
||||
return {
|
||||
serialize(data: T): Buffer {
|
||||
return Buffer.from(encode(data));
|
||||
},
|
||||
deserialize(buf: Buffer): T {
|
||||
return decode(buf) as T;
|
||||
},
|
||||
};
|
||||
}
|
||||
|
||||
export function createSerializer<T>(
|
||||
serialize: (data: T) => Buffer,
|
||||
deserialize: (buf: Buffer) => T
|
||||
): Serializer<T> {
|
||||
return { serialize, deserialize };
|
||||
}
|
||||
16
lib/dat/types.ts
Normal file
16
lib/dat/types.ts
Normal file
@@ -0,0 +1,16 @@
|
||||
// src/data-file/types.ts
|
||||
export interface Serializer<T> {
|
||||
serialize(data: T): Buffer;
|
||||
deserialize(buf: Buffer): T;
|
||||
}
|
||||
|
||||
export interface DataEntry<T> {
|
||||
sequence: number;
|
||||
timestamp: bigint;
|
||||
data: T;
|
||||
}
|
||||
|
||||
export interface DataFileOptions<T> {
|
||||
serializer: Serializer<T>;
|
||||
maxEntries?: number;
|
||||
}
|
||||
120
lib/dat/writer.ts
Normal file
120
lib/dat/writer.ts
Normal file
@@ -0,0 +1,120 @@
|
||||
// src/data-file/writer.ts
|
||||
import * as fs from 'node:fs';
|
||||
import { DATA_HEADER_SIZE } from './constants.js';
|
||||
import { DataProtocol } from './protocol.js';
|
||||
import { IndexWriter } from '../idx/index.js';
|
||||
import type { Serializer, DataFileOptions } from './types.js';
|
||||
|
||||
export class DataWriter<T> {
|
||||
private fd: number | null = null;
|
||||
private headerBuf: Buffer | null = null;
|
||||
private currentOffset: bigint = BigInt(DATA_HEADER_SIZE);
|
||||
private recordCount = 0;
|
||||
|
||||
private indexWriter: IndexWriter;
|
||||
private serializer: Serializer<T>;
|
||||
|
||||
readonly dataPath: string;
|
||||
readonly indexPath: string;
|
||||
|
||||
constructor(basePath: string, options: DataFileOptions<T>) {
|
||||
this.dataPath = `${basePath}.dat`;
|
||||
this.indexPath = `${basePath}.idx`;
|
||||
this.serializer = options.serializer;
|
||||
|
||||
const maxEntries = options.maxEntries ?? 10_000_000;
|
||||
this.indexWriter = new IndexWriter(this.indexPath, { maxEntries });
|
||||
}
|
||||
|
||||
open(): void {
|
||||
const isNew = !fs.existsSync(this.dataPath);
|
||||
|
||||
this.fd = fs.openSync(this.dataPath, isNew ? 'w+' : 'r+');
|
||||
this.headerBuf = Buffer.alloc(DATA_HEADER_SIZE);
|
||||
|
||||
if (isNew) {
|
||||
const header = DataProtocol.createHeader();
|
||||
fs.writeSync(this.fd, header, 0, DATA_HEADER_SIZE, 0);
|
||||
this.currentOffset = BigInt(DATA_HEADER_SIZE);
|
||||
this.recordCount = 0;
|
||||
} else {
|
||||
fs.readSync(this.fd, this.headerBuf, 0, DATA_HEADER_SIZE, 0);
|
||||
const header = DataProtocol.readHeader(this.headerBuf);
|
||||
this.currentOffset = header.fileSize;
|
||||
this.recordCount = header.recordCount;
|
||||
}
|
||||
|
||||
this.indexWriter.open();
|
||||
}
|
||||
|
||||
append(data: T, timestamp?: bigint): number {
|
||||
if (this.fd === null) throw new Error('Data file not opened');
|
||||
|
||||
const buf = DataProtocol.serializeRecord(data, this.serializer);
|
||||
const offset = this.currentOffset;
|
||||
|
||||
fs.writeSync(this.fd, buf, 0, buf.length, Number(offset));
|
||||
|
||||
const sequence = this.indexWriter.getNextSequence();
|
||||
const ts = timestamp ?? BigInt(Date.now()) * 1000000n;
|
||||
|
||||
this.indexWriter.append(offset, buf.length, ts);
|
||||
|
||||
this.currentOffset += BigInt(buf.length);
|
||||
this.recordCount++;
|
||||
|
||||
return sequence;
|
||||
}
|
||||
|
||||
appendBulk(records: T[], timestamp?: bigint): number[] {
|
||||
const sequences: number[] = [];
|
||||
const ts = timestamp ?? BigInt(Date.now()) * 1000000n;
|
||||
|
||||
for (const record of records) {
|
||||
const seq = this.append(record, ts);
|
||||
sequences.push(seq);
|
||||
}
|
||||
|
||||
return sequences;
|
||||
}
|
||||
|
||||
getLastSequence(): number {
|
||||
return this.indexWriter.getLastSequence();
|
||||
}
|
||||
|
||||
getNextSequence(): number {
|
||||
return this.indexWriter.getNextSequence();
|
||||
}
|
||||
|
||||
sync(): void {
|
||||
if (this.fd === null || !this.headerBuf) return;
|
||||
|
||||
DataProtocol.updateHeader(this.headerBuf, this.currentOffset, this.recordCount);
|
||||
fs.writeSync(this.fd, this.headerBuf, 0, DATA_HEADER_SIZE, 0);
|
||||
fs.fsyncSync(this.fd);
|
||||
|
||||
this.indexWriter.syncAll();
|
||||
}
|
||||
|
||||
close(): void {
|
||||
this.sync();
|
||||
|
||||
if (this.fd !== null) {
|
||||
fs.closeSync(this.fd);
|
||||
this.fd = null;
|
||||
}
|
||||
|
||||
this.indexWriter.close();
|
||||
this.headerBuf = null;
|
||||
}
|
||||
|
||||
getStats() {
|
||||
return {
|
||||
dataPath: this.dataPath,
|
||||
indexPath: this.indexPath,
|
||||
currentOffset: this.currentOffset,
|
||||
recordCount: this.recordCount,
|
||||
lastSequence: this.indexWriter.getLastSequence(),
|
||||
};
|
||||
}
|
||||
}
|
||||
8
lib/idx/constants.ts
Normal file
8
lib/idx/constants.ts
Normal file
@@ -0,0 +1,8 @@
|
||||
// src/index-file/constants.ts
|
||||
export const INDEX_MAGIC = 'INDX';
|
||||
export const INDEX_VERSION = 1;
|
||||
export const INDEX_HEADER_SIZE = 64;
|
||||
export const INDEX_ENTRY_SIZE = 32;
|
||||
|
||||
export const FLAG_VALID = 0x0001;
|
||||
export const FLAG_DELETED = 0x0002;
|
||||
6
lib/idx/index.ts
Normal file
6
lib/idx/index.ts
Normal file
@@ -0,0 +1,6 @@
|
||||
// src/index-file/index.ts
|
||||
export { IndexWriter } from './writer.js';
|
||||
export { IndexReader } from './reader.js';
|
||||
export { IndexProtocol, crc32 } from './protocol.js';
|
||||
export * from './types.js';
|
||||
export * from './constants.js';
|
||||
106
lib/idx/protocol.ts
Normal file
106
lib/idx/protocol.ts
Normal file
@@ -0,0 +1,106 @@
|
||||
// src/index-file/protocol.ts
|
||||
import {
|
||||
INDEX_MAGIC,
|
||||
INDEX_VERSION,
|
||||
INDEX_HEADER_SIZE,
|
||||
INDEX_ENTRY_SIZE,
|
||||
FLAG_VALID,
|
||||
} from './constants.js';
|
||||
import type { IndexHeader, IndexEntry } from './types.js';
|
||||
|
||||
const CRC_TABLE = new Uint32Array(256);
|
||||
for (let i = 0; i < 256; i++) {
|
||||
let c = i;
|
||||
for (let j = 0; j < 8; j++) {
|
||||
c = (c & 1) ? (0xEDB88320 ^ (c >>> 1)) : (c >>> 1);
|
||||
}
|
||||
CRC_TABLE[i] = c >>> 0;
|
||||
}
|
||||
|
||||
export function crc32(buf: Buffer, start = 0, end?: number): number {
|
||||
let crc = 0xFFFFFFFF;
|
||||
const len = end ?? buf.length;
|
||||
for (let i = start; i < len; i++) {
|
||||
crc = CRC_TABLE[(crc ^ buf[i]) & 0xFF] ^ (crc >>> 8);
|
||||
}
|
||||
return (~crc) >>> 0;
|
||||
}
|
||||
|
||||
export class IndexProtocol {
|
||||
static createHeader(entryCount: number, magic = INDEX_MAGIC): Buffer {
|
||||
const buf = Buffer.alloc(INDEX_HEADER_SIZE);
|
||||
buf.write(magic, 0, 4, 'ascii');
|
||||
buf.writeUInt32LE(INDEX_VERSION, 4);
|
||||
buf.writeBigUInt64LE(BigInt(Date.now()) * 1000000n, 8);
|
||||
buf.writeUInt32LE(INDEX_ENTRY_SIZE, 16);
|
||||
buf.writeUInt32LE(entryCount, 20);
|
||||
buf.writeUInt32LE(0, 24);
|
||||
buf.writeBigUInt64LE(0n, 28);
|
||||
buf.writeUInt32LE(0, 36);
|
||||
return buf;
|
||||
}
|
||||
|
||||
static readHeader(buf: Buffer): IndexHeader {
|
||||
return {
|
||||
magic: buf.toString('ascii', 0, 4),
|
||||
version: buf.readUInt32LE(4),
|
||||
createdAt: buf.readBigUInt64LE(8),
|
||||
entrySize: buf.readUInt32LE(16),
|
||||
entryCount: buf.readUInt32LE(20),
|
||||
validCount: buf.readUInt32LE(24),
|
||||
dataFileSize: buf.readBigUInt64LE(28),
|
||||
lastSequence: buf.readUInt32LE(36),
|
||||
reserved: buf.subarray(40, 64),
|
||||
};
|
||||
}
|
||||
|
||||
static updateHeaderCounts(
|
||||
buf: Buffer,
|
||||
validCount: number,
|
||||
dataFileSize: bigint,
|
||||
lastSequence: number
|
||||
): void {
|
||||
buf.writeUInt32LE(validCount, 24);
|
||||
buf.writeBigUInt64LE(dataFileSize, 28);
|
||||
buf.writeUInt32LE(lastSequence, 36);
|
||||
}
|
||||
|
||||
static writeEntry(buf: Buffer, index: number, entry: Omit<IndexEntry, 'checksum'>): void {
|
||||
const off = INDEX_HEADER_SIZE + index * INDEX_ENTRY_SIZE;
|
||||
|
||||
buf.writeUInt32LE(entry.sequence, off);
|
||||
buf.writeBigUInt64LE(entry.timestamp, off + 4);
|
||||
buf.writeBigUInt64LE(entry.offset, off + 12);
|
||||
buf.writeUInt32LE(entry.length, off + 20);
|
||||
buf.writeUInt32LE(entry.flags | FLAG_VALID, off + 24);
|
||||
|
||||
const checksum = crc32(buf, off, off + 28);
|
||||
buf.writeUInt32LE(checksum, off + 28);
|
||||
}
|
||||
|
||||
static readEntry(buf: Buffer, index: number): IndexEntry | null {
|
||||
const off = INDEX_HEADER_SIZE + index * INDEX_ENTRY_SIZE;
|
||||
const flags = buf.readUInt32LE(off + 24);
|
||||
|
||||
if (!(flags & FLAG_VALID)) return null;
|
||||
|
||||
return {
|
||||
sequence: buf.readUInt32LE(off),
|
||||
timestamp: buf.readBigUInt64LE(off + 4),
|
||||
offset: buf.readBigUInt64LE(off + 12),
|
||||
length: buf.readUInt32LE(off + 20),
|
||||
flags,
|
||||
checksum: buf.readUInt32LE(off + 28),
|
||||
};
|
||||
}
|
||||
|
||||
static isValidEntry(buf: Buffer, index: number): boolean {
|
||||
const off = INDEX_HEADER_SIZE + index * INDEX_ENTRY_SIZE;
|
||||
const flags = buf.readUInt32LE(off + 24);
|
||||
return (flags & FLAG_VALID) !== 0;
|
||||
}
|
||||
|
||||
static calcFileSize(entryCount: number): number {
|
||||
return INDEX_HEADER_SIZE + INDEX_ENTRY_SIZE * entryCount;
|
||||
}
|
||||
}
|
||||
131
lib/idx/reader.ts
Normal file
131
lib/idx/reader.ts
Normal file
@@ -0,0 +1,131 @@
|
||||
// src/index-file/reader.ts
|
||||
import * as fs from 'node:fs';
|
||||
import mmap from '@elilee/mmap-native';
|
||||
import { IndexProtocol } from './protocol.js';
|
||||
import type { IndexHeader, IndexEntry } from './types.js';
|
||||
|
||||
export class IndexReader {
|
||||
private fd: number | null = null;
|
||||
private buffer: Buffer | null = null;
|
||||
private header: IndexHeader | null = null;
|
||||
|
||||
readonly path: string;
|
||||
|
||||
constructor(path: string) {
|
||||
this.path = path;
|
||||
}
|
||||
|
||||
open(): void {
|
||||
const stats = fs.statSync(this.path);
|
||||
this.fd = fs.openSync(this.path, 'r');
|
||||
|
||||
this.buffer = mmap.map(
|
||||
stats.size,
|
||||
mmap.PROT_READ,
|
||||
mmap.MAP_SHARED,
|
||||
this.fd,
|
||||
0
|
||||
);
|
||||
|
||||
this.header = IndexProtocol.readHeader(this.buffer);
|
||||
}
|
||||
|
||||
getHeader(): IndexHeader {
|
||||
if (!this.header) throw new Error('Index file not opened');
|
||||
return this.header;
|
||||
}
|
||||
|
||||
getEntry(index: number): IndexEntry | null {
|
||||
if (!this.buffer || !this.header) throw new Error('Index file not opened');
|
||||
if (index < 0 || index >= this.header.entryCount) return null;
|
||||
return IndexProtocol.readEntry(this.buffer, index);
|
||||
}
|
||||
|
||||
findBySequence(sequence: number): { index: number; entry: IndexEntry } | null {
|
||||
if (!this.buffer || !this.header) throw new Error('Index file not opened');
|
||||
|
||||
for (let i = 0; i < this.header.validCount; i++) {
|
||||
const entry = IndexProtocol.readEntry(this.buffer, i);
|
||||
if (entry && entry.sequence === sequence) {
|
||||
return { index: i, entry };
|
||||
}
|
||||
}
|
||||
return null;
|
||||
}
|
||||
|
||||
findBySequenceRange(startSeq: number, endSeq: number): { index: number; entry: IndexEntry }[] {
|
||||
if (!this.buffer || !this.header) throw new Error('Index file not opened');
|
||||
|
||||
const results: { index: number; entry: IndexEntry }[] = [];
|
||||
for (let i = 0; i < this.header.validCount; i++) {
|
||||
const entry = IndexProtocol.readEntry(this.buffer, i);
|
||||
if (entry && entry.sequence >= startSeq && entry.sequence <= endSeq) {
|
||||
results.push({ index: i, entry });
|
||||
}
|
||||
}
|
||||
return results;
|
||||
}
|
||||
|
||||
getAllEntries(): IndexEntry[] {
|
||||
if (!this.buffer || !this.header) throw new Error('Index file not opened');
|
||||
|
||||
const entries: IndexEntry[] = [];
|
||||
for (let i = 0; i < this.header.validCount; i++) {
|
||||
const entry = IndexProtocol.readEntry(this.buffer, i);
|
||||
if (entry) entries.push(entry);
|
||||
}
|
||||
return entries;
|
||||
}
|
||||
|
||||
findByTimeRange(startTs: bigint, endTs: bigint): { index: number; entry: IndexEntry }[] {
|
||||
if (!this.buffer || !this.header) throw new Error('Index file not opened');
|
||||
|
||||
const results: { index: number; entry: IndexEntry }[] = [];
|
||||
for (let i = 0; i < this.header.validCount; i++) {
|
||||
const entry = IndexProtocol.readEntry(this.buffer, i);
|
||||
if (entry && entry.timestamp >= startTs && entry.timestamp <= endTs) {
|
||||
results.push({ index: i, entry });
|
||||
}
|
||||
}
|
||||
return results;
|
||||
}
|
||||
|
||||
binarySearchBySequence(targetSeq: number): { index: number; entry: IndexEntry } | null {
|
||||
if (!this.buffer || !this.header) throw new Error('Index file not opened');
|
||||
|
||||
let left = 0;
|
||||
let right = this.header.validCount - 1;
|
||||
|
||||
while (left <= right) {
|
||||
const mid = Math.floor((left + right) / 2);
|
||||
const entry = IndexProtocol.readEntry(this.buffer, mid);
|
||||
|
||||
if (!entry) {
|
||||
right = mid - 1;
|
||||
continue;
|
||||
}
|
||||
|
||||
if (entry.sequence === targetSeq) {
|
||||
return { index: mid, entry };
|
||||
} else if (entry.sequence < targetSeq) {
|
||||
left = mid + 1;
|
||||
} else {
|
||||
right = mid - 1;
|
||||
}
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
close(): void {
|
||||
if (this.buffer) {
|
||||
mmap.unmap(this.buffer);
|
||||
this.buffer = null;
|
||||
}
|
||||
if (this.fd !== null) {
|
||||
fs.closeSync(this.fd);
|
||||
this.fd = null;
|
||||
}
|
||||
this.header = null;
|
||||
}
|
||||
}
|
||||
26
lib/idx/types.ts
Normal file
26
lib/idx/types.ts
Normal file
@@ -0,0 +1,26 @@
|
||||
// src/index-file/types.ts
|
||||
export interface IndexHeader {
|
||||
magic: string;
|
||||
version: number;
|
||||
createdAt: bigint;
|
||||
entrySize: number;
|
||||
entryCount: number;
|
||||
validCount: number;
|
||||
dataFileSize: bigint;
|
||||
lastSequence: number;
|
||||
reserved: Buffer;
|
||||
}
|
||||
|
||||
export interface IndexEntry {
|
||||
sequence: number;
|
||||
timestamp: bigint;
|
||||
offset: bigint;
|
||||
length: number;
|
||||
flags: number;
|
||||
checksum: number;
|
||||
}
|
||||
|
||||
export interface IndexFileOptions {
|
||||
maxEntries: number;
|
||||
magic?: string;
|
||||
}
|
||||
141
lib/idx/writer.ts
Normal file
141
lib/idx/writer.ts
Normal file
@@ -0,0 +1,141 @@
|
||||
// src/index-file/writer.ts
|
||||
import * as fs from 'node:fs';
|
||||
import mmap from '@elilee/mmap-native';
|
||||
import { INDEX_HEADER_SIZE, FLAG_VALID } from './constants.js';
|
||||
import { IndexProtocol } from './protocol.js';
|
||||
import type { IndexFileOptions } from './types.js';
|
||||
|
||||
export class IndexWriter {
|
||||
private fd: number | null = null;
|
||||
private buffer: Buffer | null = null;
|
||||
private validCount = 0;
|
||||
private dataFileSize = 0n;
|
||||
private lastSequence = 0;
|
||||
|
||||
readonly path: string;
|
||||
readonly maxEntries: number;
|
||||
readonly fileSize: number;
|
||||
|
||||
constructor(path: string, options: IndexFileOptions) {
|
||||
this.path = path;
|
||||
this.maxEntries = options.maxEntries;
|
||||
this.fileSize = IndexProtocol.calcFileSize(options.maxEntries);
|
||||
}
|
||||
|
||||
open(): void {
|
||||
const isNew = !fs.existsSync(this.path);
|
||||
|
||||
this.fd = fs.openSync(this.path, isNew ? 'w+' : 'r+');
|
||||
|
||||
if (isNew) {
|
||||
fs.ftruncateSync(this.fd, this.fileSize);
|
||||
}
|
||||
|
||||
this.buffer = mmap.map(
|
||||
this.fileSize,
|
||||
mmap.PROT_READ | mmap.PROT_WRITE,
|
||||
mmap.MAP_SHARED,
|
||||
this.fd,
|
||||
0
|
||||
);
|
||||
|
||||
if (isNew) {
|
||||
const header = IndexProtocol.createHeader(this.maxEntries);
|
||||
header.copy(this.buffer, 0);
|
||||
this.syncHeader();
|
||||
} else {
|
||||
const header = IndexProtocol.readHeader(this.buffer);
|
||||
this.validCount = header.validCount;
|
||||
this.dataFileSize = header.dataFileSize;
|
||||
this.lastSequence = header.lastSequence;
|
||||
}
|
||||
}
|
||||
|
||||
write(
|
||||
index: number,
|
||||
sequence: number,
|
||||
offset: bigint,
|
||||
length: number,
|
||||
timestamp?: bigint
|
||||
): boolean {
|
||||
if (!this.buffer) throw new Error('Index file not opened');
|
||||
if (index < 0 || index >= this.maxEntries) return false;
|
||||
|
||||
const ts = timestamp ?? BigInt(Date.now()) * 1000000n;
|
||||
|
||||
IndexProtocol.writeEntry(this.buffer, index, {
|
||||
sequence,
|
||||
timestamp: ts,
|
||||
offset,
|
||||
length,
|
||||
flags: FLAG_VALID,
|
||||
});
|
||||
|
||||
this.validCount++;
|
||||
if (sequence > this.lastSequence) {
|
||||
this.lastSequence = sequence;
|
||||
}
|
||||
|
||||
const newDataEnd = offset + BigInt(length);
|
||||
if (newDataEnd > this.dataFileSize) {
|
||||
this.dataFileSize = newDataEnd;
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
append(offset: bigint, length: number, timestamp?: bigint): number {
|
||||
const index = this.validCount;
|
||||
if (index >= this.maxEntries) return -1;
|
||||
|
||||
const sequence = this.lastSequence + 1;
|
||||
this.write(index, sequence, offset, length, timestamp);
|
||||
return index;
|
||||
}
|
||||
|
||||
getLastSequence(): number {
|
||||
return this.lastSequence;
|
||||
}
|
||||
|
||||
getNextSequence(): number {
|
||||
return this.lastSequence + 1;
|
||||
}
|
||||
|
||||
syncHeader(): void {
|
||||
if (!this.buffer) return;
|
||||
IndexProtocol.updateHeaderCounts(
|
||||
this.buffer,
|
||||
this.validCount,
|
||||
this.dataFileSize,
|
||||
this.lastSequence
|
||||
);
|
||||
mmap.sync(this.buffer, 0, INDEX_HEADER_SIZE, mmap.MS_ASYNC);
|
||||
}
|
||||
|
||||
syncAll(): void {
|
||||
if (!this.buffer) return;
|
||||
this.syncHeader();
|
||||
mmap.sync(this.buffer, 0, this.fileSize, mmap.MS_SYNC);
|
||||
}
|
||||
|
||||
close(): void {
|
||||
if (!this.buffer || this.fd === null) return;
|
||||
|
||||
this.syncAll();
|
||||
mmap.unmap(this.buffer);
|
||||
fs.closeSync(this.fd);
|
||||
|
||||
this.buffer = null;
|
||||
this.fd = null;
|
||||
}
|
||||
|
||||
getStats() {
|
||||
return {
|
||||
path: this.path,
|
||||
maxEntries: this.maxEntries,
|
||||
validCount: this.validCount,
|
||||
dataFileSize: this.dataFileSize,
|
||||
lastSequence: this.lastSequence,
|
||||
};
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user