perf: improve the perf of Anonymous GitHub

src/AnonymizedFile.ts

@@ -38,7 +38,7 @@ export default class AnonymizedFile {
 
   repository: Repository;
   anonymizedPath: string;
-  sha?: string;
+  _sha?: string;
 
   constructor(data: { repository: Repository; anonymizedPath: string }) {
     this.repository = data.repository;
@@ -50,6 +50,12 @@ export default class AnonymizedFile {
     this.anonymizedPath = data.anonymizedPath;
   }
 
+  async sha() {
+    if (this._sha) return this._sha;
+    await this.originalPath();
+    return this._sha;
+  }
+
   /**
    * De-anonymize the path
    *
@@ -118,7 +124,7 @@ export default class AnonymizedFile {
 
     const file: TreeFile = currentAnonymized as TreeFile;
     this.fileSize = file.size;
-    this.sha = file.sha;
+    this._sha = file.sha;
 
     if (isAmbiguous) {
       // it should never happen
@@ -134,16 +140,15 @@ export default class AnonymizedFile {
     } else {
       this._originalPath = currentOriginalPath;
     }
-
     return this._originalPath;
   }
-  async extension() {
-    const filename = basename(await this.originalPath());
+  extension() {
+    const filename = basename(this.anonymizedPath);
     const extensions = filename.split(".").reverse();
     return extensions[0].toLowerCase();
   }
-  async isImage(): Promise<boolean> {
-    const extension = await this.extension();
+  isImage() {
+    const extension = this.extension();
     return [
       "png",
       "jpg",
@@ -160,18 +165,21 @@ export default class AnonymizedFile {
       "heic",
     ].includes(extension);
   }
-  async isFileSupported() {
-    const extension = await this.extension();
+  isFileSupported() {
+    const extension = this.extension();
     if (!this.repository.options.pdf && extension == "pdf") {
       return false;
     }
-    if (!this.repository.options.image && (await this.isImage())) {
+    if (!this.repository.options.image && this.isImage()) {
       return false;
     }
     return true;
   }
 
   async content(): Promise<Readable> {
+    if (this.anonymizedPath.includes(config.ANONYMIZATION_MASK)) {
+      await this.originalPath();
+    }
     if (this.fileSize && this.fileSize > config.MAX_FILE_SIZE) {
       throw new AnonymousError("file_too_big", {
         object: this,
@@ -185,9 +193,8 @@ export default class AnonymizedFile {
   }
 
   async anonymizedContent() {
-    await this.originalPath();
     const rs = await this.content();
-    return rs.pipe(anonymizeStream(await this.originalPath(), this.repository));
+    return rs.pipe(anonymizeStream(this));
   }
 
   get originalCachePath() {
@@ -196,14 +203,25 @@ export default class AnonymizedFile {
         object: this,
         httpStatus: 400,
       });
+    if (!this._originalPath) {
+      if (this.anonymizedPath.includes(config.ANONYMIZATION_MASK)) {
+        throw new AnonymousError("path_not_defined", {
+          object: this,
+          httpStatus: 400,
+        });
+      } else {
+        return join(this.repository.originalCachePath, this.anonymizedPath);
+      }
+    }
+
     return join(this.repository.originalCachePath, this._originalPath);
   }
 
   async send(res: Response): Promise<void> {
     const pipe = promisify(pipeline);
     try {
-      if (await this.extension()) {
-        res.contentType(await this.extension());
+      if (this.extension()) {
+        res.contentType(this.extension());
       }
       await pipe(await this.anonymizedContent(), res);
     } catch (error) {

src/Repository.ts

@@ -16,6 +16,7 @@ import ConferenceModel from "./database/conference/conferences.model";
 import AnonymousError from "./AnonymousError";
 import { downloadQueue } from "./queue";
 import { isConnected } from "./database/database";
+import AnonymizedFile from "./AnonymizedFile";
 
 export default class Repository {
   private _model: IAnonymizedRepositoryDocument;
@@ -48,12 +49,17 @@ export default class Repository {
    * @param opt force to get an updated list of files
    * @returns The anonymized file tree
    */
-  async anonymizedFiles(opt?: {
-    /** Force to refresh the file tree */
-    force?: boolean;
-    /** Include the file sha in the response */
-    includeSha: boolean;
-  }): Promise<Tree> {
+  async anonymizedFiles(
+    opt: {
+      /** Force to refresh the file tree */
+      force?: boolean;
+      /** Include the file sha in the response */
+      includeSha: boolean;
+    } = {
+      force: false,
+      includeSha: false,
+    }
+  ): Promise<Tree> {
     const terms = this._model.options.terms || [];
 
     function anonymizeTreeRecursive(tree: TreeElement): TreeElement {
@@ -78,11 +84,11 @@ export default class Repository {
    * @param opt force to get an updated list of files
    * @returns The file tree
    */
-  async files(opt?: { force?: boolean }) {
+  async files(opt: { force?: boolean } = { force: false }): Promise<Tree> {
     if (
       this._model.originalFiles &&
       Object.keys(this._model.originalFiles).length !== 0 &&
-      !opt?.force
+      !opt.force
     ) {
       return this._model.originalFiles;
     }
@@ -90,9 +96,6 @@ export default class Repository {
     this._model.originalFiles = files;
     this._model.size = { storage: 0, file: 0 };
     await this.computeSize();
-    await this._model.save();
-
-    this._model.originalFiles = files;
     return files;
   }
 
@@ -140,8 +143,13 @@ export default class Repository {
   zip(): Readable {
     return storage.archive(this.originalCachePath, {
       format: "zip",
-      fileTransformer: (filename) =>
-        anonymizeStream(filename, this) as Transformer,
+      fileTransformer: (filename: string) =>
+        anonymizeStream(
+          new AnonymizedFile({
+            repository: this,
+            anonymizedPath: filename,
+          })
+        ) as Transformer,
     });
   }
 

src/User.ts

@@ -119,9 +119,14 @@ export default class User {
    */
   async getRepositories() {
     const repositories = (
-      await AnonymizedRepositoryModel.find({
-        owner: this.id,
-      }).exec()
+      await AnonymizedRepositoryModel.find(
+        {
+          owner: this.id,
+        },
+        {
+          originalFiles: 0,
+        }
+      ).exec()
     ).map((d) => new Repository(d));
     const promises = [];
     for (let repo of repositories) {

src/anonymize-utils.ts

@@ -5,6 +5,7 @@ import { isText } from "istextorbinary";
 import { basename } from "path";
 import { Transform } from "stream";
 import { Readable } from "stream";
+import AnonymizedFile from "./AnonymizedFile";
 
 const urlRegex =
   /<?\b((https?|ftp|file):\/\/)[-A-Za-z0-9+&@#/%?=~_|!:,.;]+[-A-Za-z0-9+&@#/%=~_|]\b\/?>?/g;
@@ -31,7 +32,7 @@ export function isTextFile(filePath: string, content: Buffer) {
   return isText(filename, content);
 }
 
-export function anonymizeStream(filename: string, repository: Repository) {
+export function anonymizeStream(file: AnonymizedFile) {
   const ts = new Transform();
   var chunks = [],
     len = 0,
@@ -43,8 +44,8 @@ export function anonymizeStream(filename: string, repository: Repository) {
 
     if (pos === 1) {
       let data: any = Buffer.concat(chunks, len);
-      if (isTextFile(filename, data)) {
-        data = anonymizeContent(data.toString(), repository);
+      if (isTextFile(file.anonymizedPath, data)) {
+        data = anonymizeContent(data.toString(), file.repository);
       }
 
       chunks = [];
@@ -60,8 +61,8 @@ export function anonymizeStream(filename: string, repository: Repository) {
   ts._flush = function _flush(cb) {
     if (chunks.length) {
       let data: any = Buffer.concat(chunks, len);
-      if (isText(filename, data)) {
-        data = anonymizeContent(data.toString(), repository);
+      if (isText(file.anonymizedPath, data)) {
+        data = anonymizeContent(data.toString(), file.repository);
       }
 
       this.push(data);

src/database/database.ts

@@ -21,14 +21,25 @@ export async function connect() {
   return database;
 }
 
-export async function getRepository(repoId: string) {
+export async function getRepository(
+  repoId: string,
+  opts: {
+    includeFiles: boolean;
+  } = {
+    includeFiles: true,
+  }
+) {
   if (!repoId || repoId == "undefined") {
     throw new AnonymousError("repo_not_found", {
       object: repoId,
       httpStatus: 404,
     });
   }
-  const data = await AnonymizedRepositoryModel.findOne({ repoId });
+  const project: any = {};
+  if (!opts.includeFiles) {
+    project.originalFiles = 0;
+  }
+  const data = await AnonymizedRepositoryModel.findOne({ repoId }, project);
   if (!data)
     throw new AnonymousError("repo_not_found", {
       object: repoId,

src/routes/file.ts

@@ -19,12 +19,12 @@ router.get(
 
     try {
       await repo.countView();
-    
+
       const f = new AnonymizedFile({
         repository: repo,
         anonymizedPath,
       });
-      if (!(await f.isFileSupported())) {
+      if (!f.isFileSupported()) {
         throw new AnonymousError("file_not_supported", {
           httpStatus: 403,
           object: f,
@@ -34,7 +34,7 @@ router.get(
         anonymizedPath.substring(anonymizedPath.lastIndexOf("/") + 1)
       );
       // cache the file for 5min
-      res.header('Cache-Control', 'max-age=300');
+      res.header("Cache-Control", "max-age=300");
       await f.send(res);
     } catch (error) {
       return handleError(error, res, req);

src/routes/repository-private.ts

@@ -109,7 +109,10 @@ router.post(
   "/:repoId/refresh",
   async (req: express.Request, res: express.Response) => {
     try {
-      const repo = await getRepo(req, res, { nocheck: true });
+      const repo = await getRepo(req, res, {
+        nocheck: true,
+        includeFiles: false,
+      });
       if (!repo) return;
 
       if (
@@ -133,7 +136,10 @@ router.post(
 router.delete(
   "/:repoId/",
   async (req: express.Request, res: express.Response) => {
-    const repo = await getRepo(req, res, { nocheck: true });
+    const repo = await getRepo(req, res, {
+      nocheck: true,
+      includeFiles: false,
+    });
     if (!repo) return;
     // if (repo.status == "removing") return res.json({ status: repo.status });
     try {
@@ -237,7 +243,10 @@ router.get(
 // get repository information
 router.get("/:repoId/", async (req: express.Request, res: express.Response) => {
   try {
-    const repo = await getRepo(req, res, { nocheck: true });
+    const repo = await getRepo(req, res, {
+      nocheck: true,
+      includeFiles: false,
+    });
     if (!repo) return;
 
     const user = await getUser(req);
@@ -327,7 +336,10 @@ router.post(
   "/:repoId/",
   async (req: express.Request, res: express.Response) => {
     try {
-      const repo = await getRepo(req, res, { nocheck: true });
+      const repo = await getRepo(req, res, {
+        nocheck: true,
+        includeFiles: false,
+      });
       if (!repo) return;
       const user = await getUser(req);
 

src/routes/repository-public.ts

@@ -76,7 +76,7 @@ router.get(
   async (req: express.Request, res: express.Response) => {
     try {
       res.header("Cache-Control", "no-cache");
-      const repo = await getRepo(req, res, { nocheck: true });
+      const repo = await getRepo(req, res, { nocheck: true, includeFiles: false });
       if (!repo) return;
       let redirectURL = null;
       if (

src/routes/route-utils.ts

@@ -37,11 +37,16 @@ export async function getPullRequest(
 export async function getRepo(
   req: express.Request,
   res: express.Response,
-  opt?: { nocheck?: boolean }
+  opt: { nocheck?: boolean; includeFiles?: boolean } = {
+    nocheck: false,
+    includeFiles: true,
+  }
 ) {
   try {
-    const repo = await db.getRepository(req.params.repoId);
-    if (opt?.nocheck == true) {
+    const repo = await db.getRepository(req.params.repoId, {
+      includeFiles: opt.includeFiles,
+    });
+    if (opt.nocheck == true) {
     } else {
       // redirect if the repository is expired
       if (

src/routes/webview.ts

@@ -78,8 +78,7 @@ async function webView(req: express.Request, res: express.Response) {
       }
 
       let best_match = null;
-      indexSelector:
-      for (const p of indexPriority) {
+      indexSelector: for (const p of indexPriority) {
         for (let filename in currentAnonymized) {
           if (filename.toLowerCase() == p) {
             best_match = filename;
@@ -96,13 +95,13 @@ async function webView(req: express.Request, res: express.Response) {
       }
     }
 
-    if (!(await f.isFileSupported())) {
+    if (!f.isFileSupported()) {
       throw new AnonymousError("file_not_supported", {
         httpStatus: 400,
         object: f,
       });
     }
-    if ((await f.extension()) == "md") {
+    if (f.extension() == "md") {
       const content = await streamToString(await f.anonymizedContent());
       res.contentType("html").send(marked.marked(content));
     } else {

src/source/GitHubStream.ts

@@ -39,7 +39,7 @@ export default class GitHubStream extends GitHubBase implements SourceBase {
       const ghRes = await octokit.rest.git.getBlob({
         owner: this.githubRepository.owner,
         repo: this.githubRepository.repo,
-        file_sha: file.sha,
+        file_sha: await file.sha(),
       });
       if (!ghRes.data.content && ghRes.data.size != 0) {
         throw new AnonymousError("file_not_accessible", {

src/storage/S3.ts

@@ -27,6 +27,9 @@ export default class S3Storage implements StorageBase {
       endpoint: config.S3_ENDPOINT,
       accessKeyId: config.S3_CLIENT_ID,
       secretAccessKey: config.S3_CLIENT_SECRET,
+      httpOptions: {
+        timeout: 1000 * 60 * 60 * 2, // 2 hour
+      }
     });
   }