Merge pull request #3374 from github/henrymercer/scan-debug-artifacts

CI: Perform a best-effort scan of the debug artifacts during release validation

Author: henrymercer

.github/actions/verify-debug-artifact-scan-completed/action.yml

@@ -0,0 +1,6 @@
+name: Verify that the best-effort debug artifact scan completed
+description: Verifies that the best-effort debug artifact scan completed successfully during tests
+runs:
+  using: node24
+  main: index.js
+  post: post.js

.github/actions/verify-debug-artifact-scan-completed/index.js

@@ -0,0 +1,2 @@
+// The main step is a no-op, since we can only verify artifact scan completion in the post step.
+console.log("Will verify artifact scan completion in the post step.");

.github/actions/verify-debug-artifact-scan-completed/post.js

@@ -0,0 +1,11 @@
+// Post step - runs after the workflow completes, when artifact scan has finished
+const process = require("process");
+
+const scanFinished = process.env.CODEQL_ACTION_ARTIFACT_SCAN_FINISHED;
+
+if (scanFinished !== "true") {
+  console.error("Error: Best-effort artifact scan did not complete. Expected CODEQL_ACTION_ARTIFACT_SCAN_FINISHED=true");
+  process.exit(1);
+}
+
+console.log("✓ Best-effort artifact scan completed successfully");

.github/workflows/debug-artifacts-failure-safe.yml

@@ -58,6 +58,8 @@ jobs:
         uses: actions/setup-dotnet@v5
         with:
           dotnet-version: '9.x'
+      - name: Assert best-effort artifact scan completed
+        uses: ./../action/.github/actions/verify-debug-artifact-scan-completed
       - uses: ./../action/init
         with:
           tools: ${{ steps.prepare-test.outputs.tools-url }}

.github/workflows/debug-artifacts-safe.yml

@@ -54,6 +54,8 @@ jobs:
         uses: actions/setup-dotnet@v5
         with:
           dotnet-version: '9.x'
+      - name: Assert best-effort artifact scan completed
+        uses: ./../action/.github/actions/verify-debug-artifact-scan-completed
       - uses: ./../action/init
         id: init
         with:

lib/analyze-action-post.js

@@ -0,0 +1,98 @@
+import * as fs from "fs";
+import * as os from "os";
+import * as path from "path";
+
+import test from "ava";
+
+import { scanArtifactsForTokens } from "./artifact-scanner";
+import { getRunnerLogger } from "./logging";
+import { getRecordingLogger, LoggedMessage } from "./testing-utils";
+
+test("scanArtifactsForTokens detects GitHub tokens in files", async (t) => {
+  const logger = getRunnerLogger(true);
+  const tempDir = fs.mkdtempSync(path.join(os.tmpdir(), "scanner-test-"));
+
+  try {
+    // Create a test file with a fake GitHub token
+    const testFile = path.join(tempDir, "test.txt");
+    fs.writeFileSync(
+      testFile,
+      "This is a test file with token ghp_1234567890123456789012345678901234AB",
+    );
+
+    const error = await t.throwsAsync(
+      async () => await scanArtifactsForTokens([testFile], logger),
+    );
+
+    t.regex(
+      error?.message || "",
+      /Found 1 potential GitHub token.*Personal Access Token/,
+    );
+    t.regex(error?.message || "", /test\.txt/);
+  } finally {
+    // Clean up
+    fs.rmSync(tempDir, { recursive: true, force: true });
+  }
+});
+
+test("scanArtifactsForTokens handles files without tokens", async (t) => {
+  const logger = getRunnerLogger(true);
+  const tempDir = fs.mkdtempSync(path.join(os.tmpdir(), "scanner-test-"));
+
+  try {
+    // Create a test file without tokens
+    const testFile = path.join(tempDir, "test.txt");
+    fs.writeFileSync(
+      testFile,
+      "This is a test file without any sensitive data",
+    );
+
+    await t.notThrowsAsync(
+      async () => await scanArtifactsForTokens([testFile], logger),
+    );
+  } finally {
+    // Clean up
+    fs.rmSync(tempDir, { recursive: true, force: true });
+  }
+});
+
+if (os.platform() !== "win32") {
+  test("scanArtifactsForTokens finds token in debug artifacts", async (t) => {
+    t.timeout(15000); // 15 seconds
+    const messages: LoggedMessage[] = [];
+    const logger = getRecordingLogger(messages, { logToConsole: false });
+    // The zip here is a regression test based on
+    // https://github.com/github/codeql-action/security/advisories/GHSA-vqf5-2xx6-9wfm
+    const testZip = path.join(
+      __dirname,
+      "..",
+      "src",
+      "testdata",
+      "debug-artifacts-with-fake-token.zip",
+    );
+
+    // This zip file contains a nested structure with a fake token in:
+    // my-db-java-partial.zip/trap/java/invocations/kotlin.9017231652989744319.trap
+    const error = await t.throwsAsync(
+      async () => await scanArtifactsForTokens([testZip], logger),
+    );
+
+    t.regex(
+      error?.message || "",
+      /Found.*potential GitHub token/,
+      "Should detect token in nested zip",
+    );
+    t.regex(
+      error?.message || "",
+      /kotlin\.9017231652989744319\.trap/,
+      "Should report the .trap file containing the token",
+    );
+
+    const logOutput = messages.map((msg) => msg.message).join("\n");
+    t.regex(
+      logOutput,
+      /^Extracting gz file: .*\.gz$/m,
+      "Logs should show that .gz files were extracted",
+    );
+  });
+}