feat: always include files from diff in overlay changed files

src/analyze-action.ts

@@ -28,7 +28,6 @@ import {
   DependencyCacheUploadStatusReport,
   uploadDependencyCaches,
 } from "./dependency-caching";
-import { getDiffInformedAnalysisBranches } from "./diff-informed-analysis-utils";
 import { EnvVar } from "./environment";
 import { initFeatures } from "./feature-flags";
 import { KnownLanguage } from "./languages";
@@ -305,14 +304,8 @@ async function run(startedAt: Date) {
       logger,
     );
 
-    const branches = await getDiffInformedAnalysisBranches(
-      codeql,
-      features,
-      logger,
-    );
-    const diffRangePackDir = branches
-      ? await setupDiffInformedQueryRun(branches, logger)
-      : undefined;
+    // Setup diff informed analysis if needed (based on whether init created the file)
+    const diffRangePackDir = await setupDiffInformedQueryRun(logger);
 
     await warnIfGoInstalledAfterInit(config, logger);
     await runAutobuildIfLegacyGoWorkflow(config, logger);

src/analyze.ts

@@ -5,11 +5,7 @@ import { performance } from "perf_hooks";
 import * as io from "@actions/io";
 import * as yaml from "js-yaml";
 
-import {
-  getTemporaryDirectory,
-  getRequiredInput,
-  PullRequestBranches,
-} from "./actions-util";
+import { getTemporaryDirectory, getRequiredInput } from "./actions-util";
 import * as analyses from "./analyses";
 import { setupCppAutobuild } from "./autobuild";
 import { type CodeQL } from "./codeql";
@@ -21,8 +17,7 @@ import {
 import { addDiagnostic, makeDiagnostic } from "./diagnostics";
 import {
   DiffThunkRange,
-  writeDiffRangesJsonFile,
-  getPullRequestEditedDiffRanges,
+  readDiffRangesJsonFile,
 } from "./diff-informed-analysis-utils";
 import { EnvVar } from "./environment";
 import { FeatureEnablement, Feature } from "./feature-flags";
@@ -237,16 +232,19 @@ async function finalizeDatabaseCreation(
  * the diff range information, or `undefined` if the feature is disabled.
  */
 export async function setupDiffInformedQueryRun(
-  branches: PullRequestBranches,
   logger: Logger,
 ): Promise<string | undefined> {
   return await withGroupAsync(
     "Generating diff range extension pack",
     async () => {
-      logger.info(
-        `Calculating diff ranges for ${branches.base}...${branches.head}`,
-      );
-      const diffRanges = await getPullRequestEditedDiffRanges(branches, logger);
+      const diffRanges = readDiffRangesJsonFile(logger);
+      if (diffRanges === undefined) {
+        logger.info(
+          "No precomputed diff ranges found; skipping diff-informed analysis stage.",
+        );
+        return undefined;
+      }
+
       const checkoutPath = getRequiredInput("checkout_path");
       const packDir = writeDiffRangeDataExtensionPack(
         logger,
@@ -368,10 +366,6 @@ dataExtensions:
     `Wrote pr-diff-range extension pack to ${extensionFilePath}:\n${extensionContents}`,
   );
 
-  // Write the diff ranges to a JSON file, for action-side alert filtering by the
-  // upload-lib module.
-  writeDiffRangesJsonFile(logger, ranges);
-
   return diffRangeDir;
 }
 

src/diff-informed-analysis-utils.ts

@@ -105,7 +105,14 @@ export function readDiffRangesJsonFile(
   logger.debug(
     `Read pr-diff-range JSON file from ${jsonFilePath}:\n${jsonContents}`,
   );
-  return JSON.parse(jsonContents) as DiffThunkRange[];
+  try {
+    return JSON.parse(jsonContents) as DiffThunkRange[];
+  } catch (e) {
+    logger.warning(
+      `Failed to parse diff ranges JSON file at ${jsonFilePath}: ${e}`,
+    );
+    return undefined;
+  }
 }
 
 /**

src/init-action.ts

@@ -37,6 +37,11 @@ import {
   makeDiagnostic,
   makeTelemetryDiagnostic,
 } from "./diagnostics";
+import {
+  getDiffInformedAnalysisBranches,
+  getPullRequestEditedDiffRanges,
+  writeDiffRangesJsonFile,
+} from "./diff-informed-analysis-utils";
 import { EnvVar } from "./environment";
 import { Feature, FeatureEnablement, initFeatures } from "./feature-flags";
 import {
@@ -54,7 +59,7 @@ import {
   runDatabaseInitCluster,
 } from "./init";
 import { JavaEnvVars, KnownLanguage } from "./languages";
-import { getActionsLogger, Logger } from "./logging";
+import { getActionsLogger, Logger, withGroupAsync } from "./logging";
 import {
   downloadOverlayBaseDatabaseFromCache,
   OverlayBaseDatabaseDownloadStats,
@@ -413,6 +418,7 @@ async function run(startedAt: Date) {
     }
 
     await checkInstallPython311(config.languages, codeql);
+    await computeAndPersistDiffRanges(codeql, features, logger);
   } catch (unwrappedError) {
     const error = wrapError(unwrappedError);
     core.setFailed(error.message);
@@ -833,6 +839,42 @@ async function loadRepositoryProperties(
   }
 }
 
+/**
+ * Compute and persist diff ranges when diff-informed analysis is enabled
+ * (feature flag + PR context). This writes the standard pr-diff-range.json
+ * file for later reuse in the analyze step. Failures are logged but non-fatal.
+ */
+async function computeAndPersistDiffRanges(
+  codeql: CodeQL,
+  features: FeatureEnablement,
+  logger: Logger,
+): Promise<void> {
+  try {
+    await withGroupAsync("Compute PR diff ranges", async () => {
+      const branches = await getDiffInformedAnalysisBranches(
+        codeql,
+        features,
+        logger,
+      );
+      if (!branches) {
+        return;
+      }
+      const ranges = await getPullRequestEditedDiffRanges(branches, logger);
+      if (ranges === undefined) {
+        return;
+      }
+      writeDiffRangesJsonFile(logger, ranges);
+      const distinctFiles = new Set(ranges.map((r) => r.path)).size;
+      logger.info(
+        `Persisted ${ranges.length} diff range(s) across ${distinctFiles} file(s).`,
+      );
+    });
+  } catch (e) {
+    logger.warning(
+      `Failed to compute and persist PR diff ranges: ${getErrorMessage(e)}`,
+    );
+  }
+}
 async function recordZstdAvailability(
   config: configUtils.Config,
   zstdAvailability: ZstdAvailability,

src/overlay/index.test.ts

@@ -34,12 +34,14 @@ test.serial(
   "writeOverlayChangesFile generates correct changes file",
   async (t) => {
     await withTmpDir(async (tmpDir) => {
-      const dbLocation = path.join(tmpDir, "db");
-      await fs.promises.mkdir(dbLocation, { recursive: true });
-      const sourceRoot = path.join(tmpDir, "src");
-      await fs.promises.mkdir(sourceRoot, { recursive: true });
-      const tempDir = path.join(tmpDir, "temp");
-      await fs.promises.mkdir(tempDir, { recursive: true });
+      const [dbLocation, sourceRoot, tempDir] = ["db", "src", "temp"].map((d) =>
+        path.join(tmpDir, d),
+      );
+      await Promise.all(
+        [dbLocation, sourceRoot, tempDir].map((d) =>
+          fs.promises.mkdir(d, { recursive: true }),
+        ),
+      );
 
       const logger = getRunnerLogger(true);
       const config = createTestConfig({ dbLocation });
@@ -73,6 +75,9 @@ test.serial(
       const getTempDirStub = sinon
         .stub(actionsUtil, "getTemporaryDirectory")
         .returns(tempDir);
+      const getGitRootStub = sinon
+        .stub(gitUtils, "getGitRoot")
+        .resolves(sourceRoot);
       const changesFilePath = await writeOverlayChangesFile(
         config,
         sourceRoot,
@@ -80,6 +85,7 @@ test.serial(
       );
       getFileOidsStubForOverlay.restore();
       getTempDirStub.restore();
+      getGitRootStub.restore();
 
       const fileContent = await fs.promises.readFile(changesFilePath, "utf-8");
       const parsedContent = JSON.parse(fileContent) as { changes: string[] };
@@ -93,6 +99,232 @@ test.serial(
   },
 );
 
+test.serial(
+  "writeOverlayChangesFile merges additional diff files into overlay changes",
+  async (t) => {
+    await withTmpDir(async (tmpDir) => {
+      const [dbLocation, sourceRoot, tempDir] = ["db", "src", "temp"].map((d) =>
+        path.join(tmpDir, d),
+      );
+      await Promise.all(
+        [dbLocation, sourceRoot, tempDir].map((d) =>
+          fs.promises.mkdir(d, { recursive: true }),
+        ),
+      );
+
+      const logger = getRunnerLogger(true);
+      const config = createTestConfig({ dbLocation });
+
+      // Mock the getFileOidsUnderPath function to return base OIDs
+      // "reverted.js" has the same OID in both base and current, simulating
+      // a revert PR where the file content matches the overlay-base
+      const baseOids = {
+        "unchanged.js": "aaa111",
+        "modified.js": "bbb222",
+        "reverted.js": "eee555",
+      };
+      const getFileOidsStubForBase = sinon
+        .stub(gitUtils, "getFileOidsUnderPath")
+        .resolves(baseOids);
+
+      // Write the base database OIDs file
+      await writeBaseDatabaseOidsFile(config, sourceRoot);
+      getFileOidsStubForBase.restore();
+
+      // Mock the getFileOidsUnderPath function to return overlay OIDs
+      // "reverted.js" has the same OID as the base -- OID comparison alone
+      // would NOT include it, only additionalChangedFiles causes it to appear
+      const currentOids = {
+        "unchanged.js": "aaa111",
+        "modified.js": "ddd444", // Changed OID
+        "reverted.js": "eee555", // Same OID as base -- not detected by OID comparison
+      };
+      const getFileOidsStubForOverlay = sinon
+        .stub(gitUtils, "getFileOidsUnderPath")
+        .resolves(currentOids);
+
+      const getTempDirStub = sinon
+        .stub(actionsUtil, "getTemporaryDirectory")
+        .returns(tempDir);
+      const getGitRootStub = sinon
+        .stub(gitUtils, "getGitRoot")
+        .resolves(sourceRoot);
+
+      // Write a pr-diff-range.json file with diff ranges including
+      // "reverted.js" (unchanged OIDs) and "modified.js" (already in OID changes)
+      await fs.promises.writeFile(
+        path.join(tempDir, "pr-diff-range.json"),
+        JSON.stringify([
+          { path: "reverted.js", startLine: 1, endLine: 10 },
+          { path: "modified.js", startLine: 1, endLine: 5 },
+          { path: "diff-only.js", startLine: 1, endLine: 3 },
+        ]),
+      );
+
+      const changesFilePath = await writeOverlayChangesFile(
+        config,
+        sourceRoot,
+        logger,
+      );
+      getFileOidsStubForOverlay.restore();
+      getTempDirStub.restore();
+      getGitRootStub.restore();
+
+      const fileContent = await fs.promises.readFile(changesFilePath, "utf-8");
+      const parsedContent = JSON.parse(fileContent) as { changes: string[] };
+
+      t.deepEqual(
+        parsedContent.changes.sort(),
+        ["diff-only.js", "modified.js", "reverted.js"],
+        "Should include OID-changed files, diff-only files, and deduplicate overlapping files",
+      );
+    });
+  },
+);
+
+test.serial(
+  "writeOverlayChangesFile works without additional diff files",
+  async (t) => {
+    await withTmpDir(async (tmpDir) => {
+      const [dbLocation, sourceRoot, tempDir] = ["db", "src", "temp"].map((d) =>
+        path.join(tmpDir, d),
+      );
+      await Promise.all(
+        [dbLocation, sourceRoot, tempDir].map((d) =>
+          fs.promises.mkdir(d, { recursive: true }),
+        ),
+      );
+
+      const logger = getRunnerLogger(true);
+      const config = createTestConfig({ dbLocation });
+
+      // Mock the getFileOidsUnderPath function to return base OIDs
+      const baseOids = {
+        "unchanged.js": "aaa111",
+        "modified.js": "bbb222",
+      };
+      const getFileOidsStubForBase = sinon
+        .stub(gitUtils, "getFileOidsUnderPath")
+        .resolves(baseOids);
+
+      await writeBaseDatabaseOidsFile(config, sourceRoot);
+      getFileOidsStubForBase.restore();
+
+      const currentOids = {
+        "unchanged.js": "aaa111",
+        "modified.js": "ddd444",
+      };
+      const getFileOidsStubForOverlay = sinon
+        .stub(gitUtils, "getFileOidsUnderPath")
+        .resolves(currentOids);
+
+      const getTempDirStub = sinon
+        .stub(actionsUtil, "getTemporaryDirectory")
+        .returns(tempDir);
+      const getGitRootStub = sinon
+        .stub(gitUtils, "getGitRoot")
+        .resolves(sourceRoot);
+
+      // No pr-diff-range.json file exists - should work the same as before
+      const changesFilePath = await writeOverlayChangesFile(
+        config,
+        sourceRoot,
+        logger,
+      );
+      getFileOidsStubForOverlay.restore();
+      getTempDirStub.restore();
+      getGitRootStub.restore();
+
+      const fileContent = await fs.promises.readFile(changesFilePath, "utf-8");
+      const parsedContent = JSON.parse(fileContent) as { changes: string[] };
+
+      t.deepEqual(
+        parsedContent.changes.sort(),
+        ["modified.js"],
+        "Should only include OID-changed files when no additional files provided",
+      );
+    });
+  },
+);
+
+test.serial(
+  "writeOverlayChangesFile converts diff range paths to sourceRoot-relative when sourceRoot is a subdirectory",
+  async (t) => {
+    await withTmpDir(async (tmpDir) => {
+      // Simulate: repo root = tmpDir, sourceRoot = tmpDir/src
+      const repoRoot = tmpDir;
+      const sourceRoot = path.join(tmpDir, "src");
+      const [dbLocation, tempDir] = ["db", "temp"].map((d) =>
+        path.join(tmpDir, d),
+      );
+      await Promise.all(
+        [dbLocation, sourceRoot, tempDir].map((d) =>
+          fs.promises.mkdir(d, { recursive: true }),
+        ),
+      );
+
+      const logger = getRunnerLogger(true);
+      const config = createTestConfig({ dbLocation });
+
+      // Base OIDs (sourceRoot-relative paths)
+      const baseOids = {
+        "app.js": "aaa111",
+        "lib/util.js": "bbb222",
+      };
+      const getFileOidsStubForBase = sinon
+        .stub(gitUtils, "getFileOidsUnderPath")
+        .resolves(baseOids);
+      await writeBaseDatabaseOidsFile(config, sourceRoot);
+      getFileOidsStubForBase.restore();
+
+      // Current OIDs — same as base (no OID changes)
+      const currentOids = {
+        "app.js": "aaa111",
+        "lib/util.js": "bbb222",
+      };
+      const getFileOidsStubForOverlay = sinon
+        .stub(gitUtils, "getFileOidsUnderPath")
+        .resolves(currentOids);
+
+      const getTempDirStub = sinon
+        .stub(actionsUtil, "getTemporaryDirectory")
+        .returns(tempDir);
+      // getGitRoot returns the repo root (parent of sourceRoot)
+      const getGitRootStub = sinon
+        .stub(gitUtils, "getGitRoot")
+        .resolves(repoRoot);
+
+      // Diff ranges use repo-root-relative paths (as returned by the GitHub compare API)
+      await fs.promises.writeFile(
+        path.join(tempDir, "pr-diff-range.json"),
+        JSON.stringify([
+          { path: "src/app.js", startLine: 1, endLine: 10 },
+          { path: "src/lib/util.js", startLine: 5, endLine: 8 },
+          { path: "other/outside.js", startLine: 1, endLine: 3 }, // not under sourceRoot
+        ]),
+      );
+
+      const changesFilePath = await writeOverlayChangesFile(
+        config,
+        sourceRoot,
+        logger,
+      );
+      getFileOidsStubForOverlay.restore();
+      getTempDirStub.restore();
+      getGitRootStub.restore();
+
+      const fileContent = await fs.promises.readFile(changesFilePath, "utf-8");
+      const parsedContent = JSON.parse(fileContent) as { changes: string[] };
+
+      t.deepEqual(
+        parsedContent.changes.sort(),
+        ["app.js", "lib/util.js"],
+        "Should convert repo-root-relative paths to sourceRoot-relative and filter out files outside sourceRoot",
+      );
+    });
+  },
+);
+
 interface DownloadOverlayBaseDatabaseTestCase {
   overlayDatabaseMode: OverlayDatabaseMode;
   useOverlayDatabaseCaching: boolean;

src/overlay/index.ts

@@ -13,7 +13,7 @@ import { getAutomationID } from "../api-client";
 import { createCacheKeyHash } from "../caching-utils";
 import { type CodeQL } from "../codeql";
 import { type Config } from "../config-utils";
-import { getCommitOid, getFileOidsUnderPath } from "../git-utils";
+import { getCommitOid, getFileOidsUnderPath, getGitRoot } from "../git-utils";
 import { Logger, withGroupAsync } from "../logging";
 import {
   CleanupLevel,
@@ -130,11 +130,17 @@ export async function writeOverlayChangesFile(
 ): Promise<string> {
   const baseFileOids = await readBaseDatabaseOidsFile(config, logger);
   const overlayFileOids = await getFileOidsUnderPath(sourceRoot);
-  const changedFiles = computeChangedFiles(baseFileOids, overlayFileOids);
+  const oidChangedFiles = computeChangedFiles(baseFileOids, overlayFileOids);
   logger.info(
-    `Found ${changedFiles.length} changed file(s) under ${sourceRoot}.`,
+    `Found ${oidChangedFiles.length} changed file(s) under ${sourceRoot} from OID comparison.`,
   );
 
+  // Merge in any file paths from precomputed PR diff ranges to ensure the
+  // overlay always includes all files from the PR diff, even in edge cases
+  // like revert PRs where OID comparison shows no change.
+  const diffRangeFiles = await getDiffRangeFilePaths(sourceRoot, logger);
+  const changedFiles = [...new Set([...oidChangedFiles, ...diffRangeFiles])];
+
   const changedFilesJson = JSON.stringify({ changes: changedFiles });
   const overlayChangesFile = path.join(
     getTemporaryDirectory(),
@@ -165,6 +171,65 @@ function computeChangedFiles(
   return changes;
 }
 
+async function getDiffRangeFilePaths(
+  sourceRoot: string,
+  logger: Logger,
+): Promise<string[]> {
+  const jsonFilePath = path.join(getTemporaryDirectory(), "pr-diff-range.json");
+  if (!fs.existsSync(jsonFilePath)) {
+    return [];
+  }
+  let diffRanges: Array<{ path: string }>;
+  try {
+    diffRanges = JSON.parse(fs.readFileSync(jsonFilePath, "utf8")) as Array<{
+      path: string;
+    }>;
+  } catch (e) {
+    logger.warning(
+      `Failed to parse diff ranges JSON file at ${jsonFilePath}: ${e}`,
+    );
+    return [];
+  }
+  logger.debug(
+    `Read ${diffRanges.length} diff range(s) from ${jsonFilePath} for overlay changes.`,
+  );
+  const repoRelativePaths = [...new Set(diffRanges.map((r) => r.path))];
+
+  // Diff-range paths are relative to the repo root (from the GitHub compare
+  // API), but overlay changed files must be relative to sourceRoot (to match
+  // getFileOidsUnderPath output). Convert and filter accordingly.
+  const repoRoot = await getGitRoot(sourceRoot);
+  if (repoRoot === undefined) {
+    logger.warning(
+      "Cannot determine git root; returning diff range paths as-is.",
+    );
+    return repoRelativePaths;
+  }
+
+  // e.g. if repoRoot=/workspace and sourceRoot=/workspace/src, prefix="src"
+  const sourceRootRelPrefix = path
+    .relative(repoRoot, sourceRoot)
+    .replaceAll(path.sep, "/");
+
+  // If sourceRoot IS the repo root, prefix is "" and all paths pass through.
+  if (sourceRootRelPrefix === "") {
+    return repoRelativePaths;
+  }
+
+  const prefixWithSlash = `${sourceRootRelPrefix}/`;
+  const result: string[] = [];
+  for (const p of repoRelativePaths) {
+    if (p.startsWith(prefixWithSlash)) {
+      result.push(p.slice(prefixWithSlash.length));
+    } else {
+      logger.debug(
+        `Skipping diff range path "${p}" (not under source root "${sourceRootRelPrefix}").`,
+      );
+    }
+  }
+  return result;
+}
+
 // Constants for database caching
 const CACHE_VERSION = 1;
 const CACHE_PREFIX = "codeql-overlay-base-database";