improvement

This commit is contained in:
Metazla
2024-04-06 16:16:16 +02:00
parent 06bdd8402a
commit 5045120fbb
21 changed files with 152 additions and 251 deletions
+8 -3
View File
@@ -5,15 +5,15 @@ import {stat} from "fs/promises";
import path from "path";
export class ComputeHashIndexCache implements ComputeInterface {
hashIndexManager: HashIndexManager;
hashComputer: HashComputer;
private hashIndexManager: HashIndexManager;
private hashComputer: HashComputer;
constructor(indexFilePath: string,private targetHash: CID_ALGORITHM_NAMES[] = [CID_ALGORITHM_NAMES.sha1, CID_ALGORITHM_NAMES.sha256],workerPath?:string) {
this.hashComputer = new HashComputer(targetHash,workerPath);
this.hashIndexManager = new HashIndexManager(indexFilePath,targetHash);
}
async computeMissingHash(filePath: string, metadata: MultiHashData): Promise<void> {
public async computeMissingHash(filePath: string, metadata: MultiHashData): Promise<void> {
await this.hashIndexManager.init();
let stats = await stat(filePath);
if (this.hashIndexManager.getCache().has(path.basename(filePath))) {
@@ -29,4 +29,9 @@ export class ComputeHashIndexCache implements ComputeInterface {
await this.hashComputer.computeMissingHash(filePath, metadata);
this.hashIndexManager.addFileCid(filePath, stats.size, stats.mtime.toISOString(), metadata);
}
async getHashIndexManager(): Promise<HashIndexManager> {
await this.hashIndexManager.init();
return this.hashIndexManager;
}
}
+30 -1
View File
@@ -6,6 +6,7 @@ import {existsAsync} from "./ExistsAsync";
import path from "path";
import {clearInterval} from "node:timers";
import {CID_ALGORITHM_NAMES} from "./MultiHashData";
import {stat} from "fs/promises";
interface IndexLine extends Partial<Record<CID_ALGORITHM_NAMES, string>> {
path: string;
@@ -163,7 +164,7 @@ export class HashIndexManager {
});
}
private async saveCacheToFile(): Promise<void> {
public async saveCacheToFile(): Promise<void> {
if (this.indexOpsInProgress || !this.hasChanged) {
return;
}
@@ -205,6 +206,34 @@ export class HashIndexManager {
this.indexOpsInProgress = false;
}
public async getCidForFileAsync(filePath: string): Promise<IndexLine> {
const fileName = path.basename(filePath);
const stats = await stat(filePath);
let fileNameIndex = this.cache.get(fileName);
if (fileNameIndex) {
if (fileNameIndex.mtime) {
//if we have a mtime, we need to check it
if (fileNameIndex.size === (stats.size + "") && fileNameIndex.mtime === stats.mtime.toISOString()) {
return fileNameIndex;
}
} else {
//mtime is optional
if (fileNameIndex.size === (stats.size + "")) {
return fileNameIndex;
}
}
}
// 3 - if not found, delete the entry (keeps the index clean)
/*if (fileNameIndex && fileNameIndex.size !== (fileSize + "") && pathIndex && pathIndex.size !== (fileSize + "")) {
this.cache.delete(fileName);
this.cache.delete(filePath);
}*/
return null;
}
public getCidForFile(filePath: string, fileSize: number, mtime: string): IndexLine {
const fileName = path.basename(filePath);
let fileNameIndex = this.cache.get(fileName);
+56 -1
View File
@@ -1,7 +1,8 @@
import {CID_ALGORITHM_NAMES, HashComputer} from "../lib";
import {CID_ALGORITHM_NAMES, ComputeHashIndexCache, HashComputer} from "../lib";
import {describe, expect, it} from 'vitest';
import * as process from "process";
import { mkdir, rm, rmdir } from "fs/promises";
process.env.WORKER_URL = "./dist/worker.js";
@@ -39,4 +40,58 @@ describe('add', () => {
"cid_sha3-384": 'baekrkmeglipkjf33exqexdvijl4twipruay4wsbvt44akce43lq5rgccxgr7v6dwksmj3sxedakevc53g6ga'
});
});
it('meta data compute with index', async () => {
await rm("./tmp",{recursive:true});
await mkdir("./tmp",{recursive:true});
let hashComputer = new ComputeHashIndexCache("./tmp/index.csv",[
CID_ALGORITHM_NAMES.sha1,
CID_ALGORITHM_NAMES.sha256,
CID_ALGORITHM_NAMES.md5,
CID_ALGORITHM_NAMES.sha3_256,
CID_ALGORITHM_NAMES.crc32,
CID_ALGORITHM_NAMES.sha3_384
]);
let metadata = {};
await hashComputer.computeMissingHash('./src/tests/test.txt', metadata);
console.log(metadata);
/**
* test.txt CF20B62A (CRC32)
* 26f1157f4349a6fb5d207ac5f5995140 *test.txt (MD5)
* bb242b7b821074f4b9eb9a94d439ab668705aac4 ?SHA1*test.txt
* 2aa7a80d2fd1eb96d31afc7cdda9211a97524202c63f04151d6163fba35061a0 ?SHA256*test.txt
* 865a1ea4977b25e04b8ea84af93b21f1a031cb48359f3805089cdae1d89842b9a3faf87654989dcae418144a8bbb378c ?SHA3-384*test.txt
*/
expect(metadata).toEqual({
"cid_crc32": "bagzafmqcathsbnrk",
"cid_sha1": 'baeircff3eqvxxaqqot2lt242stkdtk3gq4c2vra',
"cid_sha2-256": 'baejbeibku6ua2l6r5olnggx4pto2sii2s5jeeawgh4cbkhlbmp52gudbua',
"cid_md5": 'bahkqdvibcatpcfl7ine2n625eb5ml5mzkfaa',
"cid_sha3-256": 'baelbmidner7x3dd7t2kldh6wiq4w6lsye7qrwo6uoqot2t52zkhqk56fnq',
"cid_sha3-384": 'baekrkmeglipkjf33exqexdvijl4twipruay4wsbvt44akce43lq5rgccxgr7v6dwksmj3sxedakevc53g6ga'
});
await (await hashComputer.getHashIndexManager()).saveCacheToFile();
});
it('cache', async () => {
await mkdir("./tmp",{recursive:true});
let hashComputer = new ComputeHashIndexCache("./tmp/index.csv",[
CID_ALGORITHM_NAMES.sha1,
CID_ALGORITHM_NAMES.sha256,
CID_ALGORITHM_NAMES.md5,
CID_ALGORITHM_NAMES.sha3_256,
CID_ALGORITHM_NAMES.crc32,
CID_ALGORITHM_NAMES.sha3_384
]);
let indexLine = await (await hashComputer.getHashIndexManager()).getCidForFileAsync('./src/tests/test.txt');
delete indexLine['mtime'];
expect(indexLine).toEqual({
"cid_sha1": 'baeircff3eqvxxaqqot2lt242stkdtk3gq4c2vra',
"cid_sha2-256": 'baejbeibku6ua2l6r5olnggx4pto2sii2s5jeeawgh4cbkhlbmp52gudbua',
"cid_md5": 'bahkqdvibcatpcfl7ine2n625eb5ml5mzkfaa',
"cid_sha3-256": 'baelbmidner7x3dd7t2kldh6wiq4w6lsye7qrwo6uoqot2t52zkhqk56fnq',
"cid_crc32": "bagzafmqcathsbnrk",
"cid_sha3-384": 'baekrkmeglipkjf33exqexdvijl4twipruay4wsbvt44akce43lq5rgccxgr7v6dwksmj3sxedakevc53g6ga',
"path": "test.txt",
"size": "499",
});
});
});