feat: introduce streamers that handle the stream and anonymization from github

Dockerfile

@@ -13,10 +13,8 @@ COPY healthcheck.js .
 
 COPY src ./src
 COPY public ./public
-COPY index.ts .
-COPY config.ts .
 
 RUN npm install && npm run build && npm cache clean --force
 COPY opentelemetry.js .
 
-CMD [ "node", "--require", "./opentelemetry.js", "./build/index.js"]
+CMD [ "node", "--require", "./opentelemetry.js", "./build/server/index.js"]

docker-compose.yml

@@ -5,17 +5,17 @@ services:
     build: .
     restart: always
     image: tdurieux/anonymous_github:v2
-    deploy:
-      mode: replicated
-      replicas: 4
-      endpoint_mode: vip
+    ports:
+      - $EXPOSED_PORT:5000
     env_file:
       - ./.env
     volumes:
       - ./repositories:/app/build/repositories/
     environment:
+      - PORT=5000
       - REDIS_HOSTNAME=redis
       - DB_HOSTNAME=mongodb
+      - STREAMER_ENTRYPOINT=http://streamer:5000/
     healthcheck:
       test:
         - CMD
@@ -27,22 +27,37 @@ services:
     links:
       - mongodb
       - redis
-      - opentelemetry
+      - streamer
 
-  nginx:
-    image: nginx:stable-alpine
+  streamer:
+    build: .
+    restart: always
+    image: tdurieux/anonymous_github:v2
+    deploy:
+      mode: replicated
+      replicas: 4
+      endpoint_mode: vip
+    entrypoint: ["node", "./build/streamer/index.js"]
     env_file:
       - ./.env
     volumes:
-      - ./nginx.conf:/etc/nginx/conf.d/default.conf:ro
-    depends_on:
-      - anonymous_github
-    ports:
-      - "$EXPOSED_PORT:4000"
+      - ./repositories:/app/build/repositories/
+    environment:
+      - PORT=5000
+    healthcheck:
+      test:
+        - CMD
+        - node
+        - healthcheck.js
+      interval: 10s
+      timeout: 10s
+      retries: 5
 
   redis:
     image: "redis:alpine"
     restart: always
+    ports:
+      - 127.0.0.1:6379:6379
     healthcheck:
       test:
         - CMD
@@ -101,13 +116,15 @@ services:
     image: tiredofit/db-backup
     links:
       - mongodb
+    env_file:
+      - ./.env
     volumes:
       - ./db_backups:/backup
     environment:
       - DB_TYPE=mongo
       - DB_HOST=mongodb
-      - DB_DUMP_FREQ=60
-      - DB_CLEANUP_TIME=240
+      - DB_DUMP_FREQ=120
+      - DB_CLEANUP_TIME=500
       - COMPRESSION=XZ
       - DB_USER=$DB_USERNAME
       - DB_PASS=$DB_PASSWORD

index.ts

@@ -1,7 +0,0 @@
-import { config } from "dotenv";
-config();
-
-import server from "./src/server";
-
-// start the server
-server();

nginx.conf

@@ -1,15 +0,0 @@
-upstream backend {
-    server anonymous_github:5000;    
-}
-
-server {
-    listen 4000;
-
-    resolver 127.0.0.11 valid=5s;
-    
-    include /etc/nginx/mime.types;
-
-    location / {
-        proxy_pass http://backend/;
-    }
-}

package.json

@@ -3,13 +3,12 @@
   "version": "2.2.0",
   "description": "Anonymise Github repositories for double-anonymous reviews",
   "bin": {
-    "anonymous_github": "build/cli.js"
+    "anonymous_github": "build/cli/index.js"
   },
   "scripts": {
     "test": "mocha --reporter spec",
-    "start": "node --inspect=5858 -r ts-node/register ./index.ts",
-    "dev": "nodemon --transpile-only index.ts",
-    "migrateDB": "ts-node --transpile-only migrateDB.ts",
+    "start": "node --inspect=5858 -r ts-node/register ./src/server/index.ts",
+    "dev": "nodemon --transpile-only ./src/server/index.ts",
     "build": "rm -rf build && tsc",
     "knip": "knip"
   },

src/cli/index.ts

@@ -10,12 +10,12 @@ import { join } from "path";
 import * as gh from "parse-github-url";
 import * as inquirer from "inquirer";
 
-import server from "./src/server";
-import config from "./config";
-import GitHubDownload from "./src/source/GitHubDownload";
-import Repository from "./src/Repository";
-import AnonymizedRepositoryModel from "./src/database/anonymizedRepositories/anonymizedRepositories.model";
-import { getRepositoryFromGitHub } from "./src/source/GitHubRepository";
+import server from "../server";
+import config from "../config";
+import GitHubDownload from "../core/source/GitHubDownload";
+import Repository from "../core/Repository";
+import AnonymizedRepositoryModel from "../core/model/anonymizedRepositories/anonymizedRepositories.model";
+import { getRepositoryFromGitHub } from "../core/source/GitHubRepository";
 
 function generateRandomFileName(size: number) {
   const characters =

src/config.ts

@@ -18,6 +18,7 @@ interface Config {
    * Allow to download repository and files
    */
   ENABLE_DOWNLOAD: boolean;
+  STREAMER_ENTRYPOINT: string | null;
   ANONYMIZATION_MASK: string;
   PORT: number;
   APP_HOSTNAME: string;
@@ -26,11 +27,11 @@ interface Config {
   DB_HOSTNAME: string;
   FOLDER: string;
   additionalExtensions: string[];
-  S3_BUCKET?: string;
-  S3_CLIENT_ID?: string;
-  S3_CLIENT_SECRET?: string;
-  S3_ENDPOINT?: string;
-  S3_REGION?: string;
+  S3_BUCKET: string | null;
+  S3_CLIENT_ID: string | null;
+  S3_CLIENT_SECRET: string | null;
+  S3_ENDPOINT: string | null;
+  S3_REGION: string | null;
   STORAGE: "filesystem" | "s3";
   TRUST_PROXY: number;
   RATE_LIMIT: number;
@@ -58,7 +59,7 @@ const config: Config = {
   DB_HOSTNAME: "mongodb",
   REDIS_HOSTNAME: "redis",
   REDIS_PORT: 6379,
-  FOLDER: resolve(__dirname, "repositories"),
+  FOLDER: resolve(__dirname, "..", "repositories"),
   additionalExtensions: [
     "license",
     "dockerfile",
@@ -70,11 +71,12 @@ const config: Config = {
     "in",
   ],
   STORAGE: "filesystem",
-  S3_BUCKET: process.env.S3_BUCKET,
-  S3_CLIENT_ID: process.env.S3_CLIENT_ID,
-  S3_CLIENT_SECRET: process.env.S3_CLIENT_SECRET,
-  S3_ENDPOINT: process.env.S3_ENDPOINT,
-  S3_REGION: process.env.S3_REGION,
+  STREAMER_ENTRYPOINT: null,
+  S3_BUCKET: null,
+  S3_CLIENT_ID: null,
+  S3_CLIENT_SECRET: null,
+  S3_ENDPOINT: null,
+  S3_REGION: null,
 };
 
 for (let conf in process.env) {

src/core/AnonymizedFile.ts

@@ -2,13 +2,15 @@ import { join, basename } from "path";
 import { Response } from "express";
 import { Readable } from "stream";
 import { trace } from "@opentelemetry/api";
+import { lookup } from "mime-types";
+
 import Repository from "./Repository";
 import { RepositoryStatus, Tree, TreeElement, TreeFile } from "./types";
 import config from "../config";
 import { anonymizePath, isTextFile } from "./anonymize-utils";
 import AnonymousError from "./AnonymousError";
-import { handleError } from "./routes/route-utils";
-import { lookup } from "mime-types";
+import { handleError } from "../server/routes/route-utils";
+import got from "got";
 
 /**
  * Represent a file in a anonymized repository
@@ -197,7 +199,7 @@ export default class AnonymizedFile {
               httpStatus: 403,
             });
           }
-          const out = await this.repository.source?.getFileContent(this);
+          const content = await this.repository.source?.getFileContent(this);
           if (
             !this.repository.model.isReseted ||
             this.repository.status != RepositoryStatus.READY
@@ -205,7 +207,7 @@ export default class AnonymizedFile {
             this.repository.model.isReseted = false;
             await this.repository.updateStatus(RepositoryStatus.READY);
           }
-          return out;
+          return content;
         } finally {
           span.end();
         }
@@ -213,19 +215,32 @@ export default class AnonymizedFile {
   }
 
   async anonymizedContent() {
-    return trace
-      .getTracer("ano-file")
-      .startActiveSpan("anonymizedContent", async (span) => {
-        span.setAttribute("anonymizedPath", this.anonymizedPath);
-        const content = await this.content();
-        return content
-          .pipe(
-            this.repository.generateAnonymizeTransformer(this.anonymizedPath)
-          )
-          .on("close", () => {
-            span.end();
-          });
+    const span = trace.getTracer("ano-file").startSpan("Repository.conference");
+    span.setAttribute("anonymizedPath", this.anonymizedPath);
+    const anonymizer = this.repository.generateAnonymizeTransformer(
+      this.anonymizedPath
+    );
+    if (!config.STREAMER_ENTRYPOINT) {
+      // collect the content locally
+      const content = await this.content();
+      return content.pipe(anonymizer).on("close", () => {
+        span.end();
       });
+    }
+    // use the streamer service
+    return got.stream(join(config.STREAMER_ENTRYPOINT, "api"), {
+      method: "POST",
+      json: {
+        token: await this.repository.getToken(),
+        repoFullName: this.repository.model.source.repositoryName,
+        commit: this.repository.model.source.commit,
+        branch: this.repository.model.source.branch,
+        repoId: this.repository.repoId,
+        filePath: this.filePath,
+        sha: await this.sha(),
+        anonymizerOptions: anonymizer.opt,
+      },
+    });
   }
 
   get filePath() {
@@ -243,6 +258,9 @@ export default class AnonymizedFile {
   }
 
   async send(res: Response): Promise<void> {
+    const anonymizer = this.repository.generateAnonymizeTransformer(
+      this.anonymizedPath
+    );
     return trace
       .getTracer("ano-file")
       .startActiveSpan("AnonymizedFile.send", async (span) => {
@@ -250,6 +268,39 @@ export default class AnonymizedFile {
         span.setAttribute("anonymizedPath", this.anonymizedPath);
         return new Promise<void>(async (resolve, reject) => {
           try {
+            if (config.STREAMER_ENTRYPOINT) {
+              // use the streamer service
+              got
+                .stream(join(config.STREAMER_ENTRYPOINT, "api"), {
+                  method: "POST",
+                  json: {
+                    token: await this.repository.getToken(),
+                    repoFullName: this.repository.model.source.repositoryName,
+                    commit: this.repository.model.source.commit,
+                    branch: this.repository.model.source.branch,
+                    repoId: this.repository.repoId,
+                    filePath: this.filePath,
+                    sha: await this.sha(),
+                    anonymizerOptions: anonymizer.opt,
+                  },
+                })
+                .on("error", () => {
+                  handleError(
+                    new AnonymousError("file_not_found", {
+                      object: this,
+                      httpStatus: 404,
+                    }),
+                    res
+                  );
+                })
+                .pipe(res)
+                .on("close", () => {
+                  span.end();
+                  resolve();
+                });
+              return;
+            }
+
             const mime = lookup(this.anonymizedPath);
             if (mime && this.extension() != "ts") {
               res.contentType(mime);
@@ -257,9 +308,6 @@ export default class AnonymizedFile {
               res.contentType("text/plain");
             }
             res.header("Accept-Ranges", "none");
-            const anonymizer = this.repository.generateAnonymizeTransformer(
-              this.anonymizedPath
-            );
             anonymizer.once("transform", (data) => {
               if (!mime && data.isText) {
                 res.contentType("text/plain");
@@ -269,26 +317,27 @@ export default class AnonymizedFile {
                 res.header("Content-Length", this.fileSize.toString());
               }
             });
-
             const content = await this.content();
+            function handleStreamError(error: Error) {
+              if (!content.closed && !content.destroyed) {
+                content.destroy();
+              }
+              span.recordException(error);
+              span.end();
+              reject(error);
+              // handleError(error, res);
+            }
             content
+              .on("error", handleStreamError)
               .pipe(anonymizer)
               .pipe(res)
+              .on("error", handleStreamError)
               .on("close", () => {
                 if (!content.closed && !content.destroyed) {
                   content.destroy();
                 }
                 span.end();
                 resolve();
-              })
-              .on("error", (error) => {
-                if (!content.closed && !content.destroyed) {
-                  content.destroy();
-                }
-                span.recordException(error);
-                span.end();
-                reject(error);
-                handleError(error, res);
               });
           } catch (error) {
             handleError(error, res);

src/core/Conference.ts

@@ -1,5 +1,5 @@
-import AnonymizedRepositoryModel from "./database/anonymizedRepositories/anonymizedRepositories.model";
-import { IConferenceDocument } from "./database/conference/conferences.types";
+import AnonymizedRepositoryModel from "./model/anonymizedRepositories/anonymizedRepositories.model";
+import { IConferenceDocument } from "./model/conference/conferences.types";
 import Repository from "./Repository";
 import { ConferenceStatus } from "./types";
 

src/core/GitHubUtils.ts

@@ -1,7 +1,8 @@
 import { trace } from "@opentelemetry/api";
-import Repository from "./Repository";
 import { Octokit } from "@octokit/rest";
-import UserModel from "./database/users/users.model";
+
+import Repository from "./Repository";
+import UserModel from "./model/users/users.model";
 import config from "../config";
 
 export function octokit(token: string) {

src/core/PullRequest.ts

@@ -1,10 +1,10 @@
 import { RepositoryStatus } from "./types";
 import User from "./User";
-import UserModel from "./database/users/users.model";
+import UserModel from "./model/users/users.model";
 import Conference from "./Conference";
-import ConferenceModel from "./database/conference/conferences.model";
+import ConferenceModel from "./model/conference/conferences.model";
 import AnonymousError from "./AnonymousError";
-import { IAnonymizedPullRequestDocument } from "./database/anonymizedPullRequests/anonymizedPullRequests.types";
+import { IAnonymizedPullRequestDocument } from "./model/anonymizedPullRequests/anonymizedPullRequests.types";
 import config from "../config";
 import got from "got";
 import { octokit } from "./GitHubUtils";

src/core/Repository.ts

@@ -6,16 +6,16 @@ import GitHubStream from "./source/GitHubStream";
 import GitHubDownload from "./source/GitHubDownload";
 import Zip from "./source/Zip";
 import { anonymizePath } from "./anonymize-utils";
-import UserModel from "./database/users/users.model";
-import { IAnonymizedRepositoryDocument } from "./database/anonymizedRepositories/anonymizedRepositories.types";
+import UserModel from "./model/users/users.model";
+import { IAnonymizedRepositoryDocument } from "./model/anonymizedRepositories/anonymizedRepositories.types";
 import { AnonymizeTransformer } from "./anonymize-utils";
 import GitHubBase from "./source/GitHubBase";
 import Conference from "./Conference";
-import ConferenceModel from "./database/conference/conferences.model";
+import ConferenceModel from "./model/conference/conferences.model";
 import AnonymousError from "./AnonymousError";
-import { downloadQueue } from "./queue";
-import { isConnected } from "./database/database";
-import AnonymizedRepositoryModel from "./database/anonymizedRepositories/anonymizedRepositories.model";
+import { downloadQueue } from "../queue";
+import { isConnected } from "../server/database";
+import AnonymizedRepositoryModel from "./model/anonymizedRepositories/anonymizedRepositories.model";
 import { GitHubRepository } from "./source/GitHubRepository";
 import { trace } from "@opentelemetry/api";
 import { getToken } from "./GitHubUtils";
@@ -59,7 +59,7 @@ export default class Repository {
 
   private checkedToken: boolean = false;
 
-  private async getToken() {
+  async getToken() {
     if (this.checkedToken) return this._model.source.accessToken as string;
     const originalToken = this._model.source.accessToken;
     const token = await getToken(this);

src/core/User.ts

@@ -1,11 +1,12 @@
-import AnonymizedRepositoryModel from "./database/anonymizedRepositories/anonymizedRepositories.model";
-import RepositoryModel from "./database/repositories/repositories.model";
-import { IUserDocument } from "./database/users/users.types";
+import { trace } from "@opentelemetry/api";
+
+import AnonymizedRepositoryModel from "./model/anonymizedRepositories/anonymizedRepositories.model";
+import RepositoryModel from "./model/repositories/repositories.model";
+import { IUserDocument } from "./model/users/users.types";
 import Repository from "./Repository";
 import { GitHubRepository } from "./source/GitHubRepository";
 import PullRequest from "./PullRequest";
-import AnonymizedPullRequestModel from "./database/anonymizedPullRequests/anonymizedPullRequests.model";
-import { trace } from "@opentelemetry/api";
+import AnonymizedPullRequestModel from "./model/anonymizedPullRequests/anonymizedPullRequests.model";
 import { octokit } from "./GitHubUtils";
 
 /**

src/core/anonymize-utils.ts

@@ -1,10 +1,10 @@
-import config from "../config";
-import { isText } from "istextorbinary";
 import { basename } from "path";
-import { Transform } from "stream";
-import { Readable } from "stream";
+import { Transform, Readable } from "stream";
+import { isText } from "istextorbinary";
 import { trace } from "@opentelemetry/api";
 
+import config from "../config";
+
 const urlRegex =
   /<?\b((https?|ftp|file):\/\/)[-A-Za-z0-9+&@#/%?=~_|!:,.;]+[-A-Za-z0-9+&@#/%=~_|]\b\/?>?/g;
 
@@ -35,7 +35,7 @@ export class AnonymizeTransformer extends Transform {
   anonimizer: ContentAnonimizer;
 
   constructor(
-    private readonly opt: {
+    readonly opt: {
       filePath: string;
     } & ConstructorParameters<typeof ContentAnonimizer>[0]
   ) {
@@ -56,7 +56,6 @@ export class AnonymizeTransformer extends Transform {
         if (this.isText === null) {
           this.isText = isTextFile(this.opt.filePath, chunk);
         }
-
         if (this.isText) {
           const content = this.anonimizer.anonymize(chunk.toString());
           if (this.anonimizer.wasAnonymized) {
@@ -171,7 +170,7 @@ export class ContentAnonimizer {
       });
 
       // remove the term in the text
-      return content.replace(new RegExp(`\\b${term}\\b`, "gi"), () => {
+      content = content.replace(new RegExp(`\\b${term}\\b`, "gi"), () => {
         this.wasAnonymized = true;
         return mask;
       });

src/core/source/GitHubBase.ts

@@ -2,6 +2,7 @@ import { Readable } from "stream";
 
 import AnonymizedFile from "../AnonymizedFile";
 import { Tree } from "../types";
+import { SourceBase } from "./Source";
 
 export interface GitHubBaseData {
   getToken: () => string | Promise<string>;
@@ -11,21 +12,6 @@ export interface GitHubBaseData {
   commit: string;
 }
 
-export interface SourceBase {
-  readonly type: string;
-
-  /**
-   * Retrieve the fie content
-   * @param file the file of the content to retrieve
-   */
-  getFileContent(file: AnonymizedFile): Promise<Readable>;
-
-  /**
-   * Get all the files from a specific source
-   */
-  getFiles(progress?: (status: string) => void): Promise<Tree>;
-}
-
 export default abstract class GitHubBase implements SourceBase {
   abstract type: "GitHubDownload" | "GitHubStream" | "Zip";
   accessToken: string | undefined;

src/core/source/GitHubDownload.ts

@@ -4,11 +4,11 @@ import { OctokitResponse } from "@octokit/types";
 
 import storage from "../storage";
 import GitHubBase, { GitHubBaseData } from "./GitHubBase";
-import AnonymizedFile from "../AnonymizedFile";
-import AnonymousError from "../AnonymousError";
 import { trace } from "@opentelemetry/api";
 import { FILE_TYPE } from "../storage/Storage";
 import { octokit } from "../GitHubUtils";
+import AnonymousError from "../AnonymousError";
+import AnonymizedFile from "../AnonymizedFile";
 
 export default class GitHubDownload extends GitHubBase {
   type: "GitHubDownload" | "GitHubStream" | "Zip" = "GitHubDownload";

src/core/source/GitHubRepository.ts

@@ -1,12 +1,13 @@
 import { Branch } from "../types";
 import * as gh from "parse-github-url";
-import { IRepositoryDocument } from "../database/repositories/repositories.types";
 import { RestEndpointMethodTypes } from "@octokit/rest";
-import RepositoryModel from "../database/repositories/repositories.model";
-import AnonymousError from "../AnonymousError";
-import { isConnected } from "../database/database";
 import { trace } from "@opentelemetry/api";
+
+import AnonymousError from "../AnonymousError";
+import { isConnected } from "../../server/database";
 import { octokit } from "../GitHubUtils";
+import { IRepositoryDocument } from "../model/repositories/repositories.types";
+import RepositoryModel from "../model/repositories/repositories.model";
 
 export class GitHubRepository {
   private _data: Partial<{

src/core/source/GitHubStream.ts

@@ -82,36 +82,24 @@ export default class GitHubStream extends GitHubBase {
           object: file,
         });
       }
-      try {
-        const content = this.downloadFile(await this.data.getToken(), file_sha);
+      const content = this.downloadFile(await this.data.getToken(), file_sha);
 
-        // duplicate the stream to write it to the storage
-        const stream1 = content.pipe(new stream.PassThrough());
-        const stream2 = content.pipe(new stream.PassThrough());
-        storage.write(
-          file.repository.repoId,
-          file.filePath,
-          stream1,
-          this.type
-        );
-        return stream2;
-      } catch (error) {
-        if (
-          (error as any).status === 404 ||
-          (error as any).httpStatus === 404
-        ) {
-          throw new AnonymousError("file_not_found", {
-            httpStatus: (error as any).status || (error as any).httpStatus,
-            cause: error as Error,
-            object: file,
-          });
-        }
-        throw new AnonymousError("file_too_big", {
+      // duplicate the stream to write it to the storage
+      const stream1 = content.pipe(new stream.PassThrough());
+      const stream2 = content.pipe(new stream.PassThrough());
+
+      content.on("error", (error) => {
+        error = new AnonymousError("file_not_found", {
           httpStatus: (error as any).status || (error as any).httpStatus,
           cause: error as Error,
           object: file,
         });
-      }
+        stream1.emit("error", error);
+        stream2.emit("error", error);
+      });
+
+      storage.write(file.repository.repoId, file.filePath, stream1, this.type);
+      return stream2;
     } finally {
       span.end();
     }

src/core/source/Source.ts

@@ -0,0 +1,24 @@
+import { Readable } from "stream";
+
+import AnonymizedFile from "../AnonymizedFile";
+import { Tree } from "../types";
+import GitHubDownload from "./GitHubDownload";
+import GitHubStream from "./GitHubStream";
+import Zip from "./Zip";
+
+export type Source = GitHubDownload | GitHubStream | Zip;
+
+export interface SourceBase {
+  readonly type: string;
+
+  /**
+   * Retrieve the fie content
+   * @param file the file of the content to retrieve
+   */
+  getFileContent(file: AnonymizedFile): Promise<Readable>;
+
+  /**
+   * Get all the files from a specific source
+   */
+  getFiles(progress?: (status: string) => void): Promise<Tree>;
+}

src/core/source/Zip.ts

@@ -1,7 +1,8 @@
+import * as stream from "stream";
+
 import AnonymizedFile from "../AnonymizedFile";
 import storage from "../storage";
-import * as stream from "stream";
-import { SourceBase } from "./GitHubBase";
+import { SourceBase } from "./Source";
 
 export default class Zip implements SourceBase {
   type = "Zip";

src/core/storage/FileSystem.ts

@@ -83,7 +83,15 @@ export default class FileSystem extends StorageBase {
     span.setAttribute("path", fullPath);
     try {
       await this.mk(repoId, dirname(p));
+      if (data instanceof Readable) {
+        data.on("error", (err) => {
+          this.rm(repoId, p);
+        });
+      }
       return await fs.promises.writeFile(fullPath, data, "utf-8");
+    } catch (err: any) {
+      span.recordException(err);
+      // throw err;
     } finally {
       span.end();
     }

src/core/storage/S3.ts

@@ -32,6 +32,8 @@ export default class S3Storage extends StorageBase {
   private client(timeout = 10000) {
     if (!config.S3_CLIENT_ID) throw new Error("S3_CLIENT_ID not set");
     if (!config.S3_CLIENT_SECRET) throw new Error("S3_CLIENT_SECRET not set");
+    if (!config.S3_REGION) throw new Error("S3_REGION not set");
+    if (!config.S3_ENDPOINT) throw new Error("S3_ENDPOINT not set");
     return new S3({
       credentials: {
         accessKeyId: config.S3_CLIENT_ID,
@@ -212,6 +214,14 @@ export default class S3Storage extends StorageBase {
     try {
       if (!config.S3_BUCKET) throw new Error("S3_BUCKET not set");
 
+      if (data instanceof Readable) {
+        data.on("error", (err) => {
+          console.error(`[ERROR] S3 write ${path}`, err);
+          span.recordException(err as Error);
+          this.rm(repoId, path);
+        });
+      }
+
       const params: PutObjectCommandInput = {
         Bucket: config.S3_BUCKET,
         Key: join(this.repoPath(repoId), path),
@@ -229,7 +239,6 @@ export default class S3Storage extends StorageBase {
       });
 
       await parallelUploads3.done();
-      return;
     } finally {
       span.end();
     }

src/core/storage/Storage.ts

@@ -4,6 +4,10 @@ import * as archiver from "archiver";
 import { Response } from "express";
 
 import { Tree } from "../types";
+import S3Storage from "./S3";
+import FileSystem from "./FileSystem";
+
+export type Storage = S3Storage | FileSystem;
 
 export enum FILE_TYPE {
   FILE = "file",

src/core/types.ts

@@ -1,13 +1,3 @@
-import GitHubDownload from "./source/GitHubDownload";
-import GitHubStream from "./source/GitHubStream";
-import Zip from "./source/Zip";
-import S3Storage from "./storage/S3";
-import FileSystem from "./storage/FileSystem";
-
-export type Source = GitHubDownload | GitHubStream | Zip;
-
-export type Storage = S3Storage | FileSystem;
-
 export interface Branch {
   name: string;
   commit: string;

src/queue/index.ts

@@ -1,6 +1,6 @@
 import { Queue, Worker } from "bullmq";
 import config from "../config";
-import Repository from "./Repository";
+import Repository from "../core/Repository";
 import * as path from "path";
 
 export let cacheQueue: Queue<Repository>;
@@ -37,7 +37,7 @@ export function startWorker() {
   });
   const cacheWorker = new Worker<Repository>(
     cacheQueue.name,
-    path.resolve("build/src/processes/removeCache.js"),
+    path.resolve("build/queue/processes/removeCache.js"),
     {
       concurrency: 5,
       connection,
@@ -49,7 +49,7 @@ export function startWorker() {
   });
   const removeWorker = new Worker<Repository>(
     removeQueue.name,
-    path.resolve("build/src/processes/removeRepository.js"),
+    path.resolve("build/queue/processes/removeRepository.js"),
     {
       concurrency: 5,
       connection,
@@ -62,7 +62,7 @@ export function startWorker() {
 
   const downloadWorker = new Worker<Repository>(
     downloadQueue.name,
-    path.resolve("build/src/processes/downloadRepository.js"),
+    path.resolve("build/queue/processes/downloadRepository.js"),
     {
       concurrency: 3,
       connection,

src/queue/processes/downloadRepository.ts

@@ -1,10 +1,10 @@
+import { Exception, trace } from "@opentelemetry/api";
 import { SandboxedJob } from "bullmq";
 import { config } from "dotenv";
 config();
-import Repository from "../Repository";
-import { getRepository as getRepositoryImport } from "../database/database";
-import { RepositoryStatus } from "../types";
-import { Exception, trace } from "@opentelemetry/api";
+import Repository from "../../core/Repository";
+import { getRepository as getRepositoryImport } from "../../server/database";
+import { RepositoryStatus } from "../../core/types";
 
 export default async function (job: SandboxedJob<Repository, void>) {
   const {

src/queue/processes/removeCache.ts

@@ -1,7 +1,7 @@
-import { SandboxedJob } from "bullmq";
-import Repository from "../Repository";
-import { getRepository as getRepositoryImport } from "../database/database";
 import { Exception, trace } from "@opentelemetry/api";
+import { SandboxedJob } from "bullmq";
+import Repository from "../../core/Repository";
+import { getRepository as getRepositoryImport } from "../../server/database";
 
 export default async function (job: SandboxedJob<Repository, void>) {
   const {

src/queue/processes/removeRepository.ts

@@ -1,8 +1,8 @@
-import { SandboxedJob } from "bullmq";
-import Repository from "../Repository";
-import { getRepository as getRepositoryImport } from "../database/database";
-import { RepositoryStatus } from "../types";
 import { trace } from "@opentelemetry/api";
+import { SandboxedJob } from "bullmq";
+import Repository from "../../core/Repository";
+import { getRepository as getRepositoryImport } from "../../server/database";
+import { RepositoryStatus } from "../../core/types";
 
 export default async function (job: SandboxedJob<Repository, void>) {
   const {

src/server/database.ts

@@ -1,10 +1,10 @@
 import mongoose, { ConnectOptions } from "mongoose";
-import Repository from "../Repository";
-import config from "../../config";
-import AnonymizedRepositoryModel from "./anonymizedRepositories/anonymizedRepositories.model";
-import AnonymousError from "../AnonymousError";
-import AnonymizedPullRequestModel from "./anonymizedPullRequests/anonymizedPullRequests.model";
-import PullRequest from "../PullRequest";
+import Repository from "../core/Repository";
+import config from "../config";
+import AnonymizedRepositoryModel from "../core/model/anonymizedRepositories/anonymizedRepositories.model";
+import AnonymousError from "../core/AnonymousError";
+import AnonymizedPullRequestModel from "../core/model/anonymizedPullRequests/anonymizedPullRequests.model";
+import PullRequest from "../core/PullRequest";
 
 const MONGO_URL = `mongodb://${config.DB_USERNAME}:${config.DB_PASSWORD}@${config.DB_HOSTNAME}:27017/`;
 

src/server/index.ts

@@ -1,3 +1,6 @@
+import { config as dotenv } from "dotenv";
+dotenv();
+
 import { createClient } from "redis";
 import { resolve, join } from "path";
 import { existsSync } from "fs";
@@ -7,16 +10,15 @@ import RedisStore from "rate-limit-redis";
 import * as express from "express";
 import * as compression from "compression";
 import * as passport from "passport";
-
-import config from "../config";
-import { connect } from "./database/database";
+import { connect } from "./database";
 import { initSession, router as connectionRouter } from "./routes/connection";
 import router from "./routes";
-import AnonymizedRepositoryModel from "./database/anonymizedRepositories/anonymizedRepositories.model";
+import AnonymizedRepositoryModel from "../core/model/anonymizedRepositories/anonymizedRepositories.model";
 import { conferenceStatusCheck, repositoryStatusCheck } from "./schedule";
-import { startWorker } from "./queue";
-import AnonymizedPullRequestModel from "./database/anonymizedPullRequests/anonymizedPullRequests.model";
+import { startWorker } from "../queue";
+import AnonymizedPullRequestModel from "../core/model/anonymizedPullRequests/anonymizedPullRequests.model";
 import { getUser } from "./routes/route-utils";
+import config from "../config";
 
 function indexResponse(req: express.Request, res: express.Response) {
   if (
@@ -214,3 +216,5 @@ export default async function start() {
   app.listen(config.PORT);
   console.log("Database connected and Server started on port: " + config.PORT);
 }
+
+start();

src/server/routes/admin.ts

@@ -1,12 +1,12 @@
 import { Queue } from "bullmq";
 import * as express from "express";
-import AnonymousError from "../AnonymousError";
-import AnonymizedRepositoryModel from "../database/anonymizedRepositories/anonymizedRepositories.model";
-import ConferenceModel from "../database/conference/conferences.model";
-import UserModel from "../database/users/users.model";
-import { cacheQueue, downloadQueue, removeQueue } from "../queue";
-import Repository from "../Repository";
-import User from "../User";
+import AnonymousError from "../../core/AnonymousError";
+import AnonymizedRepositoryModel from "../../core/model/anonymizedRepositories/anonymizedRepositories.model";
+import ConferenceModel from "../../core/model/conference/conferences.model";
+import UserModel from "../../core/model/users/users.model";
+import { cacheQueue, downloadQueue, removeQueue } from "../../queue";
+import Repository from "../../core/Repository";
+import User from "../../core/User";
 import { ensureAuthenticated } from "./connection";
 import { handleError, getUser, isOwnerOrAdmin, getRepo } from "./route-utils";
 

src/server/routes/conference.ts

@@ -1,10 +1,10 @@
 import * as express from "express";
-import AnonymousError from "../AnonymousError";
-import Conference from "../Conference";
-import ConferenceModel from "../database/conference/conferences.model";
+import AnonymousError from "../../core/AnonymousError";
+import Conference from "../../core/Conference";
+import ConferenceModel from "../../core/model/conference/conferences.model";
 import { ensureAuthenticated } from "./connection";
 import { handleError, getUser, isOwnerOrAdmin } from "./route-utils";
-import { IConferenceDocument } from "../database/conference/conferences.types";
+import { IConferenceDocument } from "../../core/model/conference/conferences.types";
 
 const router = express.Router();
 

src/server/routes/connection.ts

@@ -7,9 +7,9 @@ import { Profile, Strategy } from "passport-github2";
 import * as express from "express";
 
 import config from "../../config";
-import UserModel from "../database/users/users.model";
-import { IUserDocument } from "../database/users/users.types";
-import AnonymousError from "../AnonymousError";
+import UserModel from "../../core/model/users/users.model";
+import { IUserDocument } from "../../core/model/users/users.types";
+import AnonymousError from "../../core/AnonymousError";
 
 export function ensureAuthenticated(
   req: express.Request,

src/server/routes/file.ts

@@ -1,6 +1,6 @@
 import * as express from "express";
-import AnonymizedFile from "../AnonymizedFile";
-import AnonymousError from "../AnonymousError";
+import AnonymizedFile from "../../core/AnonymizedFile";
+import AnonymousError from "../../core/AnonymousError";
 import { getRepo, handleError } from "./route-utils";
 
 export const router = express.Router();
@@ -49,7 +49,7 @@ router.get(
       // cache the file for 5min
       res.header("Cache-Control", "max-age=300");
       await repo.countView();
-      f.send(res);
+      await f.send(res);
     } catch (error) {
       return handleError(error, res, req);
     }

src/server/routes/pullRequest-private.ts

@@ -7,11 +7,11 @@ import {
   handleError,
   isOwnerOrAdmin,
 } from "./route-utils";
-import AnonymousError from "../AnonymousError";
-import { IAnonymizedPullRequestDocument } from "../database/anonymizedPullRequests/anonymizedPullRequests.types";
-import PullRequest from "../PullRequest";
-import AnonymizedPullRequestModel from "../database/anonymizedPullRequests/anonymizedPullRequests.model";
-import { RepositoryStatus } from "../types";
+import AnonymousError from "../../core/AnonymousError";
+import { IAnonymizedPullRequestDocument } from "../../core/model/anonymizedPullRequests/anonymizedPullRequests.types";
+import PullRequest from "../../core/PullRequest";
+import AnonymizedPullRequestModel from "../../core/model/anonymizedPullRequests/anonymizedPullRequests.model";
+import { RepositoryStatus } from "../../core/types";
 
 const router = express.Router();
 

src/server/routes/pullRequest-public.ts

@@ -1,7 +1,7 @@
 import * as express from "express";
 
 import { getPullRequest, handleError } from "./route-utils";
-import AnonymousError from "../AnonymousError";
+import AnonymousError from "../../core/AnonymousError";
 
 const router = express.Router();
 

src/server/routes/repository-private.ts

@@ -1,22 +1,22 @@
 import * as express from "express";
 import { ensureAuthenticated } from "./connection";
 
-import * as db from "../database/database";
+import * as db from "../database";
 import { getRepo, getUser, handleError, isOwnerOrAdmin } from "./route-utils";
-import { getRepositoryFromGitHub } from "../source/GitHubRepository";
+import { getRepositoryFromGitHub } from "../../core/source/GitHubRepository";
 import gh = require("parse-github-url");
-import AnonymizedRepositoryModel from "../database/anonymizedRepositories/anonymizedRepositories.model";
-import { IAnonymizedRepositoryDocument } from "../database/anonymizedRepositories/anonymizedRepositories.types";
-import Repository from "../Repository";
-import UserModel from "../database/users/users.model";
-import ConferenceModel from "../database/conference/conferences.model";
-import AnonymousError from "../AnonymousError";
-import { downloadQueue, removeQueue } from "../queue";
-import RepositoryModel from "../database/repositories/repositories.model";
-import User from "../User";
-import { RepositoryStatus } from "../types";
-import { IUserDocument } from "../database/users/users.types";
-import { checkToken } from "../GitHubUtils";
+import AnonymizedRepositoryModel from "../../core/model/anonymizedRepositories/anonymizedRepositories.model";
+import { IAnonymizedRepositoryDocument } from "../../core/model/anonymizedRepositories/anonymizedRepositories.types";
+import Repository from "../../core/Repository";
+import UserModel from "../../core/model/users/users.model";
+import ConferenceModel from "../../core/model/conference/conferences.model";
+import AnonymousError from "../../core/AnonymousError";
+import { downloadQueue, removeQueue } from "../../queue";
+import RepositoryModel from "../../core/model/repositories/repositories.model";
+import User from "../../core/User";
+import { RepositoryStatus } from "../../core/types";
+import { IUserDocument } from "../../core/model/users/users.types";
+import { checkToken } from "../../core/GitHubUtils";
 
 const router = express.Router();
 

src/server/routes/repository-public.ts

@@ -4,9 +4,9 @@ import * as stream from "stream";
 import config from "../../config";
 
 import { getRepo, handleError } from "./route-utils";
-import AnonymousError from "../AnonymousError";
-import { downloadQueue } from "../queue";
-import { RepositoryStatus } from "../types";
+import AnonymousError from "../../core/AnonymousError";
+import { downloadQueue } from "../../queue";
+import { RepositoryStatus } from "../../core/types";
 
 const router = express.Router();
 

src/server/routes/route-utils.ts

@@ -1,8 +1,9 @@
 import * as express from "express";
-import AnonymousError from "../AnonymousError";
-import * as db from "../database/database";
-import UserModel from "../database/users/users.model";
-import User from "../User";
+import AnonymousError from "../../core/AnonymousError";
+import * as db from "../database";
+import UserModel from "../../core/model/users/users.model";
+import User from "../../core/User";
+import { HTTPError } from "got";
 
 export async function getPullRequest(
   req: express.Request,
@@ -85,6 +86,11 @@ function printError(error: any, req?: express.Request) {
       if (req.originalUrl === "/api/repo/undefined/options") return;
     }
     console.error(message);
+  } else if (error instanceof HTTPError) {
+    let message = `[ERROR] HTTP.${
+      error.code
+    } ${error.message.toString()}  ${error.stack?.split("\n")[1].trim()}`;
+    console.error(message);
   } else if (error instanceof Error) {
     console.error(error);
   } else {
@@ -107,7 +113,10 @@ export function handleError(
     status = error.httpStatus;
   } else if (error.$metadata?.httpStatusCode) {
     status = error.$metadata.httpStatusCode;
-  } else if (message && message.indexOf("not_found") > -1) {
+  } else if (
+    message &&
+    (message.indexOf("not_found") > -1 || message.indexOf("(Not Found)") > -1)
+  ) {
     status = 404;
   } else if (message && message.indexOf("not_connected") > -1) {
     status = 401;

src/server/routes/user.ts

@@ -2,8 +2,8 @@ import * as express from "express";
 import config from "../../config";
 import { ensureAuthenticated } from "./connection";
 import { handleError, getUser, isOwnerOrAdmin } from "./route-utils";
-import UserModel from "../database/users/users.model";
-import User from "../User";
+import UserModel from "../../core/model/users/users.model";
+import User from "../../core/User";
 
 const router = express.Router();
 

src/server/routes/webview.ts

@@ -1,11 +1,11 @@
 import * as express from "express";
 import { getRepo, handleError } from "./route-utils";
 import * as path from "path";
-import AnonymizedFile from "../AnonymizedFile";
-import AnonymousError from "../AnonymousError";
-import { Tree, TreeElement } from "../types";
+import AnonymizedFile from "../../core/AnonymizedFile";
+import AnonymousError from "../../core/AnonymousError";
+import { Tree, TreeElement } from "../../core/types";
 import * as marked from "marked";
-import { streamToString } from "../anonymize-utils";
+import { streamToString } from "../../core/anonymize-utils";
 
 const router = express.Router();
 

src/server/schedule.ts

@@ -1,8 +1,8 @@
 import * as schedule from "node-schedule";
-import Conference from "./Conference";
-import AnonymizedRepositoryModel from "./database/anonymizedRepositories/anonymizedRepositories.model";
-import ConferenceModel from "./database/conference/conferences.model";
-import Repository from "./Repository";
+import Conference from "../core/Conference";
+import AnonymizedRepositoryModel from "../core/model/anonymizedRepositories/anonymizedRepositories.model";
+import ConferenceModel from "../core/model/conference/conferences.model";
+import Repository from "../core/Repository";
 
 export function conferenceStatusCheck() {
   // check every 6 hours the status of the conferences

src/streamer/index.ts

@@ -0,0 +1,31 @@
+import { config as dotenv } from "dotenv";
+dotenv();
+
+import * as express from "express";
+import * as compression from "compression";
+
+import config from "../config";
+import router from "./route";
+import { handleError } from "../server/routes/route-utils";
+import AnonymousError from "../core/AnonymousError";
+
+const app = express();
+app.use(express.json());
+
+app.use(compression());
+
+app.use("/api", router);
+
+app.all("*", (req, res) => {
+  handleError(
+    new AnonymousError("file_not_found", {
+      httpStatus: 404,
+      object: req.originalUrl,
+    }),
+    res,
+    req
+  );
+});
+app.listen(config.PORT, () => {
+  console.log(`Server started on http://streamer:${config.PORT}`);
+});

src/streamer/route.ts

@@ -0,0 +1,63 @@
+import * as express from "express";
+import GitHubStream from "../core/source/GitHubStream";
+import { AnonymizeTransformer, isTextFile } from "../core/anonymize-utils";
+import { handleError } from "../server/routes/route-utils";
+import { contentType } from "mime-types";
+
+export const router = express.Router();
+
+router.post("/", async (req: express.Request, res: express.Response) => {
+  req.body = req.body || {};
+  const token: string = req.body.token;
+  const repoFullName = req.body.repoFullName.split("/");
+  const repoId = req.body.repoId;
+  const branch = req.body.branch;
+  const fileSha = req.body.sha;
+  const commit = req.body.commit;
+  const filePath = req.body.filePath;
+  const anonymizerOptions = req.body.anonymizerOptions;
+  const anonymizer = new AnonymizeTransformer(anonymizerOptions);
+
+  const source = new GitHubStream({
+    repoId,
+    organization: repoFullName[0],
+    repoName: repoFullName[1],
+    commit: commit,
+    getToken: () => token,
+  });
+  const content = source.downloadFile(token, fileSha);
+  try {
+    const mime = contentType(filePath);
+    if (mime && !filePath.endsWith(".ts")) {
+      res.contentType(mime);
+    } else if (isTextFile(filePath)) {
+      res.contentType("text/plain");
+    }
+    res.header("Accept-Ranges", "none");
+    anonymizer.once("transform", (data) => {
+      if (!mime && data.isText) {
+        res.contentType("text/plain");
+      }
+    });
+    function handleStreamError(error: Error) {
+      if (!content.closed && !content.destroyed) {
+        content.destroy();
+      }
+      handleError(error, res);
+    }
+    content
+      .on("error", handleStreamError)
+      .pipe(anonymizer)
+      .pipe(res)
+      .on("error", handleStreamError)
+      .on("close", () => {
+        if (!content.closed && !content.destroyed) {
+          content.destroy();
+        }
+      });
+  } catch (error) {
+    handleError(error, res);
+  }
+});
+
+export default router;

tsconfig.json

@@ -17,6 +17,6 @@
     "esModuleInterop": false,
     "incremental": true
   },
-  "include": ["src/**/*.ts", "index.ts", "cli.ts"],
+  "include": ["src/**/*.ts", "index.ts", "src/cli/index.ts"],
   "exclude": ["node_modules", ".vscode"]
 }