Add feature flag to roll out JAR minimization in the Java extractor

src/analyze-action.ts

@@ -29,7 +29,7 @@ import { uploadDatabases } from "./database-upload";
 import { uploadDependencyCaches } from "./dependency-caching";
 import { getDiffInformedAnalysisBranches } from "./diff-informed-analysis-utils";
 import { EnvVar } from "./environment";
-import { Features } from "./feature-flags";
+import { Feature, Features } from "./feature-flags";
 import { KnownLanguage } from "./languages";
 import { getActionsLogger, Logger } from "./logging";
 import { uploadOverlayBaseDatabaseToCache } from "./overlay-database-utils";
@@ -384,7 +384,11 @@ async function run() {
 
     // Store dependency cache(s) if dependency caching is enabled.
     if (shouldStoreCache(config.dependencyCachingEnabled)) {
-      await uploadDependencyCaches(config, logger);
+      const minimizeJavaJars = await features.getValue(
+        Feature.JavaMinimizeDependencyJars,
+        codeql,
+      );
+      await uploadDependencyCaches(config, logger, minimizeJavaJars);
     }
 
     // We don't upload results in test mode, so don't wait for processing

src/dependency-caching.ts

@@ -8,7 +8,7 @@ import { getTemporaryDirectory } from "./actions-util";
 import { getTotalCacheSize } from "./caching-utils";
 import { Config } from "./config-utils";
 import { EnvVar } from "./environment";
-import { Language } from "./languages";
+import { KnownLanguage, Language } from "./languages";
 import { Logger } from "./logging";
 import { getRequiredEnvParam } from "./util";
 
@@ -89,11 +89,13 @@ async function makeGlobber(patterns: string[]): Promise<glob.Globber> {
  *
  * @param languages The languages being analyzed.
  * @param logger A logger to record some informational messages to.
+ * @param minimizeJavaJars Whether the Java extractor should rewrite downloaded JARs to minimize their size.
  * @returns A list of languages for which dependency caches were restored.
  */
 export async function downloadDependencyCaches(
   languages: Language[],
   logger: Logger,
+  minimizeJavaJars: boolean,
 ): Promise<Language[]> {
   const restoredCaches: Language[] = [];
 
@@ -118,8 +120,10 @@ export async function downloadDependencyCaches(
       continue;
     }
 
-    const primaryKey = await cacheKey(language, cacheConfig);
-    const restoreKeys: string[] = [await cachePrefix(language)];
+    const primaryKey = await cacheKey(language, cacheConfig, minimizeJavaJars);
+    const restoreKeys: string[] = [
+      await cachePrefix(language, minimizeJavaJars),
+    ];
 
     logger.info(
       `Downloading cache for ${language} with key ${primaryKey} and restore keys ${restoreKeys.join(
@@ -149,8 +153,13 @@ export async function downloadDependencyCaches(
  *
  * @param config The configuration for this workflow.
  * @param logger A logger to record some informational messages to.
+ * @param minimizeJavaJars Whether the Java extractor should rewrite downloaded JARs to minimize their size.
  */
-export async function uploadDependencyCaches(config: Config, logger: Logger) {
+export async function uploadDependencyCaches(
+  config: Config,
+  logger: Logger,
+  minimizeJavaJars: boolean,
+): Promise<void> {
   for (const language of config.languages) {
     const cacheConfig = getDefaultCacheConfig()[language];
 
@@ -192,7 +201,7 @@ export async function uploadDependencyCaches(config: Config, logger: Logger) {
       continue;
     }
 
-    const key = await cacheKey(language, cacheConfig);
+    const key = await cacheKey(language, cacheConfig, minimizeJavaJars);
 
     logger.info(
       `Uploading cache of size ${size} for ${language} with key ${key}...`,
@@ -222,14 +231,16 @@ export async function uploadDependencyCaches(config: Config, logger: Logger) {
  *
  * @param language The language being analyzed.
  * @param cacheConfig The cache configuration for the language.
+ * @param minimizeJavaJars Whether the Java extractor should rewrite downloaded JARs to minimize their size.
  * @returns A cache key capturing information about the project(s) being analyzed in the specified language.
  */
 async function cacheKey(
   language: Language,
   cacheConfig: CacheConfig,
+  minimizeJavaJars: boolean = false,
 ): Promise<string> {
   const hash = await glob.hashFiles(cacheConfig.hash.join("\n"));
-  return `${await cachePrefix(language)}${hash}`;
+  return `${await cachePrefix(language, minimizeJavaJars)}${hash}`;
 }
 
 /**
@@ -237,9 +248,13 @@ async function cacheKey(
  * can be changed to invalidate old caches, the runner's operating system, and the specified language name.
  *
  * @param language The language being analyzed.
+ * @param minimizeJavaJars Whether the Java extractor should rewrite downloaded JARs to minimize their size.
  * @returns The prefix that identifies what a cache is for.
  */
-async function cachePrefix(language: Language): Promise<string> {
+async function cachePrefix(
+  language: Language,
+  minimizeJavaJars: boolean,
+): Promise<string> {
   const runnerOs = getRequiredEnvParam("RUNNER_OS");
   const customPrefix = process.env[EnvVar.DEPENDENCY_CACHING_PREFIX];
   let prefix = CODEQL_DEPENDENCY_CACHE_PREFIX;
@@ -248,5 +263,10 @@ async function cachePrefix(language: Language): Promise<string> {
     prefix = `${prefix}-${customPrefix}`;
   }
 
+  // To ensure a safe rollout of JAR minimization, we change the key when the feature is enabled.
+  if (language === KnownLanguage.java && minimizeJavaJars) {
+    prefix = `minify-${prefix}`;
+  }
+
   return `${prefix}-${CODEQL_DEPENDENCY_CACHE_VERSION}-${runnerOs}-${language}-`;
 }

src/environment.ts

@@ -115,6 +115,9 @@ export enum EnvVar {
    */
   DEPENDENCY_CACHING_PREFIX = "CODEQL_ACTION_DEPENDENCY_CACHE_PREFIX",
 
+  /** Used by the Java extractor option to enable minimizing dependency JARs. */
+  JAVA_EXTRACTOR_MINIMIZE_DEPENDENCY_JARS = "CODEQL_EXTRACTOR_JAVA_OPTION_MINIMIZE_DEPENDENCY_JARS",
+
   /**
    * Whether to enable experimental extractors for CodeQL.
    */

src/feature-flags.ts

@@ -50,6 +50,7 @@ export enum Feature {
   DisableJavaBuildlessEnabled = "disable_java_buildless_enabled",
   DisableKotlinAnalysisEnabled = "disable_kotlin_analysis_enabled",
   ExportDiagnosticsEnabled = "export_diagnostics_enabled",
+  JavaMinimizeDependencyJars = "java_minimize_dependency_jars",
   OverlayAnalysis = "overlay_analysis",
   OverlayAnalysisActions = "overlay_analysis_actions",
   OverlayAnalysisCodeScanningActions = "overlay_analysis_code_scanning_actions",
@@ -269,6 +270,11 @@ export const featureConfig: Record<
     legacyApi: true,
     minimumVersion: undefined,
   },
+  [Feature.JavaMinimizeDependencyJars]: {
+    defaultValue: false,
+    envVar: "CODEQL_ACTION_JAVA_MINIMIZE_DEPENDENCY_JARS",
+    minimumVersion: "2.23.0",
+  },
 };
 
 /**

src/init-action.ts

@@ -78,6 +78,7 @@ import {
   wrapError,
   checkActionVersion,
   getErrorMessage,
+  BuildMode,
 } from "./util";
 import { validateWorkflow } from "./workflow";
 
@@ -546,8 +547,16 @@ async function run() {
     }
 
     // Restore dependency cache(s), if they exist.
+    const minimizeJavaJars = await features.getValue(
+      Feature.JavaMinimizeDependencyJars,
+      codeql,
+    );
     if (shouldRestoreCache(config.dependencyCachingEnabled)) {
-      await downloadDependencyCaches(config.languages, logger);
+      await downloadDependencyCaches(
+        config.languages,
+        logger,
+        minimizeJavaJars,
+      );
     }
 
     // Suppress warnings about disabled Python library extraction.
@@ -597,6 +606,24 @@ async function run() {
       }
     }
 
+    // If the feature flag to minimize Java dependency jars is enabled, and we are doing a Java
+    // `build-mode: none` analysis (i.e. the flag is relevant), then set the environment variable
+    // that enables the corresponding option in the Java extractor.
+    if (process.env[EnvVar.JAVA_EXTRACTOR_MINIMIZE_DEPENDENCY_JARS]) {
+      logger.debug(
+        `${EnvVar.JAVA_EXTRACTOR_MINIMIZE_DEPENDENCY_JARS} is already set to '${process.env[EnvVar.JAVA_EXTRACTOR_MINIMIZE_DEPENDENCY_JARS]}', so the Action will not override it.`,
+      );
+    } else if (
+      minimizeJavaJars &&
+      config.buildMode === BuildMode.None &&
+      config.languages.includes(KnownLanguage.java)
+    ) {
+      core.exportVariable(
+        EnvVar.JAVA_EXTRACTOR_MINIMIZE_DEPENDENCY_JARS,
+        "true",
+      );
+    }
+
     const { registriesAuthTokens, qlconfigFile } =
       await configUtils.generateRegistries(
         getOptionalInput("registries"),