fix: enforce content policies using original file types

src/core/AnonymizedFile.ts

@@ -343,7 +343,7 @@ export default class AnonymizedFile {
 
   async anonymizedContent() {
     const anonymizer = this.repository.generateAnonymizeTransformer(
-      this.anonymizedPath
+      await this.originalPath()
     );
     if (!config.STREAMER_ENTRYPOINT) {
       // collect the content locally
@@ -385,7 +385,7 @@ export default class AnonymizedFile {
 
   async send(res: Response): Promise<void> {
     const anonymizer = this.repository.generateAnonymizeTransformer(
-      this.anonymizedPath
+      await this.originalPath()
     );
     // eslint-disable-next-line no-async-promise-executor
     return new Promise<void>(async (resolve, reject) => {

src/core/zipStream.ts

@@ -149,7 +149,7 @@ export async function streamAnonymizedZip(
             entry.path.substring(entry.path.indexOf("/") + 1),
             compiledTerms
           );
-          if (!isEntryAllowed(fileName, opt.contentOptions)) {
+          if (!isEntryAllowed(entry.path, opt.contentOptions)) {
             entry.autodrain();
             return;
           }
@@ -159,7 +159,7 @@ export async function streamAnonymizedZip(
           // isText=false for every file, so the zip ships unanonymized.
           const anonymizer = new AnonymizeTransformer({
             ...opt.anonymizerOptions,
-            filePath: fileName,
+            filePath: entry.path,
           });
           const st = entry.pipe(anonymizer);
           archive.append(st, { name: fileName });

src/server/routes/file.ts

@@ -105,6 +105,7 @@ router.get(
         repository: repo,
         anonymizedPath,
       });
+      const originalPath = await f.originalPath();
       if (!f.isFileSupported()) {
         throw new AnonymousError("file_not_supported", {
           httpStatus: 403,
@@ -115,7 +116,7 @@ router.get(
         res.attachment(
           anonymizedPath.substring(anonymizedPath.lastIndexOf("/") + 1)
         );
-      } else if (isScriptableDocument(anonymizedPath)) {
+      } else if (isScriptableDocument(originalPath) || isScriptableDocument(anonymizedPath)) {
         // A repository's own .html/.svg is untrusted content served from our
         // origin: opening it renders it as a document, and any script in it
         // would run as the site itself (session cookie, same-origin fetches

src/server/routes/webview.ts

@@ -151,6 +151,7 @@ async function webView(req: express.Request, res: express.Response) {
       }
     }
 
+    await f.originalPath();
     if (!f.isFileSupported()) {
       throw new AnonymousError("file_not_supported", {
         httpStatus: 400,

test/production-regressions.test.js

@@ -119,6 +119,34 @@ describe("production regressions", function () {
       throw new Error("expected rejection");
     } catch (error) { expect(error.message).to.equal("token_expired"); }
   });
+
+  function response() {
+    return { headers: {}, header(key, value) { this.headers[key] = value; return this; },
+      contentType() { return this; }, status(value) { this.statusCode = value; return this; }, end() {}, send() {} };
+  }
+  function fileRoute(originalName) {
+    const File = require("../src/core/AnonymizedFile").default;
+    const utils = require("../src/server/routes/route-utils");
+    const repo = { options: { pdf: false, image: false, terms: [] }, model: { source: { commit: "commit-1" }, options: {} }, isReady: async () => true, countView: async () => {} };
+    stub(utils, "getRepo", async () => repo);
+    stub(File.prototype, "originalPath", async function () { this._file = { name: originalName, path: "" }; return originalName; });
+    stub(File.prototype, "sha", async () => "sha");
+    stub(File.prototype, "send", async () => {});
+    const handler = require("../src/server/routes/file").default.stack[0].route.stack[0].handle;
+    return { repo, handler, req: { url: "/repo/file/page.txt?v=old", protocol: "https", hostname: "host", params: { repoId: "repo" }, query: { v: "old" }, headers: {} } };
+  }
+  it("gates PDFs using their original extension", async function () {
+    const { handler, req } = fileRoute("report.pdf");
+    let failure;
+    stub(require("../src/server/routes/route-utils"), "handleError", error => { failure = error; });
+    await handler(req, response());
+    expect(failure.message).to.equal("file_not_supported");
+  });
+  it("sandboxes renamed HTML documents", async function () {
+    const { handler, req } = fileRoute("page.html");
+    const first = response(); await handler(req, first);
+    expect(first.headers["Content-Security-Policy"]).to.include("sandbox");
+  });
   it("omits an upstream length when later text is rewritten", async function () {
     const File = require("../src/core/AnonymizedFile").default;
     stub(config, "STREAMER_ENTRYPOINT", "");