refactor: fall back to non-overlay analysis when diff-informed analysis is unavailable

Author: sam-robson

lib/init-action.js

@@ -31,7 +31,10 @@ import {
   addNoLanguageDiagnostic,
   makeTelemetryDiagnostic,
 } from "./diagnostics";
-import { shouldPerformDiffInformedAnalysis } from "./diff-informed-analysis-utils";
+import {
+  computeAndPersistDiffRanges,
+  getDiffInformedAnalysisBranches,
+} from "./diff-informed-analysis-utils";
 import { EnvVar } from "./environment";
 import * as errorMessages from "./error-messages";
 import { Feature, FeatureEnablement } from "./feature-flags";
@@ -49,8 +52,12 @@ import {
   isAnalyzingDefaultBranch,
 } from "./git-utils";
 import { KnownLanguage, Language } from "./languages";
-import { Logger } from "./logging";
-import { CODEQL_OVERLAY_MINIMUM_VERSION, OverlayDatabaseMode } from "./overlay";
+import { Logger, withGroupAsync } from "./logging";
+import {
+  CODEQL_OVERLAY_MINIMUM_VERSION,
+  OverlayDatabaseMode,
+  revertOverlayModeIfDiffInformedUnavailable,
+} from "./overlay";
 import {
   addOverlayDisablementDiagnostics,
   OverlayDisabledReason,
@@ -1229,19 +1236,55 @@ export async function initConfig(
     );
   }
 
-  if (
-    config.overlayDatabaseMode === OverlayDatabaseMode.Overlay ||
-    (await shouldPerformDiffInformedAnalysis(
+  let shouldRunDiffInformedAnalysis = false;
+  let diffInformedRangesPrepared = false;
+  try {
+    const diffInformedAnalysisBranches = await getDiffInformedAnalysisBranches(
       inputs.codeql,
       inputs.features,
       logger,
-    ))
+    );
+    shouldRunDiffInformedAnalysis = diffInformedAnalysisBranches !== undefined;
+
+    if (diffInformedAnalysisBranches) {
+      await withGroupAsync("Computing PR diff ranges", async () => {
+        try {
+          diffInformedRangesPrepared = await computeAndPersistDiffRanges(
+            diffInformedAnalysisBranches,
+            logger,
+          );
+        } catch (e) {
+          logger.warning(
+            `Failed to compute diff-informed analysis ranges: ${getErrorMessage(e)}`,
+          );
+        }
+      });
+    }
+  } catch (e) {
+    logger.warning(
+      `Failed to determine diff-informed analysis availability: ${getErrorMessage(e)}`,
+    );
+    // Treat errors conservatively so overlay falls back if diff-informed
+    // availability could not be checked.
+    shouldRunDiffInformedAnalysis = true;
+  }
+
+  if (
+    config.overlayDatabaseMode === OverlayDatabaseMode.Overlay ||
+    shouldRunDiffInformedAnalysis
   ) {
     config.extraQueryExclusions.push({
       exclude: { tags: "exclude-from-incremental" },
     });
   }
 
+  revertOverlayModeIfDiffInformedUnavailable(
+    config,
+    shouldRunDiffInformedAnalysis,
+    diffInformedRangesPrepared,
+    logger,
+  );
+
   if (await isTrapCachingEnabled(features, config.overlayDatabaseMode)) {
     const { trapCaches, trapCacheDownloadTime } = await downloadCacheWithTime(
       inputs.codeql,

src/config-utils.ts

@@ -151,6 +151,33 @@ export async function getPullRequestEditedDiffRanges(
   return results;
 }
 
+/**
+ * Compute and persist the diff ranges for a pull request. This fetches the
+ * diff from the GitHub API and writes it to the diff ranges JSON file so that
+ * CodeQL can use it for diff-informed analysis.
+ *
+ * @param branches The base and head branches of the pull request, as returned
+ *   by `getDiffInformedAnalysisBranches`.
+ * @param logger
+ * @returns `true` if the diff ranges were successfully computed and persisted,
+ *   otherwise `false`.
+ */
+export async function computeAndPersistDiffRanges(
+  branches: PullRequestBranches,
+  logger: Logger,
+): Promise<boolean> {
+  const ranges = await getPullRequestEditedDiffRanges(branches, logger);
+  if (ranges === undefined) {
+    return false;
+  }
+  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).`,
+  );
+  return true;
+}
+
 async function getFileDiffsWithBasehead(
   branches: PullRequestBranches,
   logger: Logger,

src/diff-informed-analysis-utils.ts

@@ -37,11 +37,6 @@ 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 {
@@ -59,7 +54,7 @@ import {
   runDatabaseInitCluster,
 } from "./init";
 import { JavaEnvVars, KnownLanguage } from "./languages";
-import { getActionsLogger, Logger, withGroupAsync } from "./logging";
+import { getActionsLogger, Logger } from "./logging";
 import {
   downloadOverlayBaseDatabaseFromCache,
   OverlayBaseDatabaseDownloadStats,
@@ -427,7 +422,6 @@ 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);
@@ -818,42 +812,6 @@ 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> {
-  await withGroupAsync("Computing PR diff ranges", async () => {
-    try {
-      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/init-action.ts

@@ -24,6 +24,7 @@ import {
   getCacheRestoreKeyPrefix,
   getCacheSaveKey,
   OverlayDatabaseMode,
+  revertOverlayModeIfDiffInformedUnavailable,
   writeBaseDatabaseOidsFile,
   writeOverlayChangesFile,
 } from ".";
@@ -602,3 +603,54 @@ test.serial("overlay-base database cache keys remain stable", async (t) => {
     `Expected save key "${saveKey}" to start with restore key prefix "${restoreKeyPrefix}"`,
   );
 });
+
+test.serial(
+  "revertOverlayModeIfDiffInformedUnavailable: no-op when mode is not Overlay",
+  (t) => {
+    const config = createTestConfig({});
+    config.overlayDatabaseMode = OverlayDatabaseMode.None;
+    const logger = getRunnerLogger(true);
+
+    revertOverlayModeIfDiffInformedUnavailable(config, true, false, logger);
+
+    t.is(config.overlayDatabaseMode, OverlayDatabaseMode.None);
+  },
+);
+
+test.serial(
+  "revertOverlayModeIfDiffInformedUnavailable: no-op when diff-informed analysis is available",
+  (t) => {
+    const config = createTestConfig({});
+    config.overlayDatabaseMode = OverlayDatabaseMode.Overlay;
+    const logger = getRunnerLogger(true);
+
+    revertOverlayModeIfDiffInformedUnavailable(config, true, true, logger);
+
+    t.is(config.overlayDatabaseMode, OverlayDatabaseMode.Overlay);
+  },
+);
+
+test.serial(
+  "revertOverlayModeIfDiffInformedUnavailable: reverts to None when diff-informed analysis is unavailable",
+  (t) => {
+    const config = createTestConfig({});
+    config.overlayDatabaseMode =
+      OverlayDatabaseMode.Overlay as OverlayDatabaseMode;
+    const logger = getRunnerLogger(true);
+
+    revertOverlayModeIfDiffInformedUnavailable(config, true, false, logger);
+    t.is(config.overlayDatabaseMode, OverlayDatabaseMode.None);
+  },
+);
+
+test.serial(
+  "revertOverlayModeIfDiffInformedUnavailable: no-op when diff-informed analysis is disabled",
+  (t) => {
+    const config = createTestConfig({});
+    config.overlayDatabaseMode = OverlayDatabaseMode.Overlay;
+    const logger = getRunnerLogger(true);
+
+    revertOverlayModeIfDiffInformedUnavailable(config, false, false, logger);
+    t.is(config.overlayDatabaseMode, OverlayDatabaseMode.Overlay);
+  },
+);