Prune results of Ruby query from SARIF

src/upload-lib.test.ts

@@ -3,14 +3,16 @@ import * as path from "path";
 
 import test from "ava";
 
-import { getRunnerLogger } from "./logging";
+import { getRunnerLogger, Logger } from "./logging";
 import { setupTests } from "./testing-utils";
 import * as uploadLib from "./upload-lib";
+import { pruneInvalidResults } from "./upload-lib";
 import {
+  GitHubVariant,
+  GitHubVersion,
   initializeEnvironment,
   Mode,
-  GitHubVersion,
-  GitHubVariant,
+  SarifFile,
   withTmpDir,
 } from "./util";
 
@@ -344,6 +346,116 @@ test("validateUniqueCategory for multiple runs", (t) => {
   t.throws(() => uploadLib.validateUniqueCategory(sarif2));
 });
 
+test("pruneInvalidResults", (t) => {
+  const loggedMessages: string[] = [];
+  const mockLogger = {
+    info: (message: string) => {
+      loggedMessages.push(message);
+    },
+  } as Logger;
+
+  const sarif: SarifFile = {
+    runs: [
+      {
+        tool: otherTool,
+        results: [resultWithBadMessage1, resultWithGoodMessage],
+      },
+      {
+        tool: affectedCodeQLVersion,
+        results: [
+          resultWithOtherRuleId,
+          resultWithBadMessage1,
+          resultWithBadMessage2,
+          resultWithGoodMessage,
+        ],
+      },
+      {
+        tool: unaffectedCodeQLVersion,
+        results: [resultWithBadMessage1, resultWithGoodMessage],
+      },
+    ],
+  };
+  const result = pruneInvalidResults(sarif, mockLogger);
+
+  const expected: SarifFile = {
+    runs: [
+      {
+        tool: otherTool,
+        results: [resultWithBadMessage1, resultWithGoodMessage],
+      },
+      {
+        tool: affectedCodeQLVersion,
+        results: [resultWithOtherRuleId, resultWithGoodMessage],
+      },
+      {
+        tool: unaffectedCodeQLVersion,
+        results: [resultWithBadMessage1, resultWithGoodMessage],
+      },
+    ],
+  };
+
+  t.deepEqual(result, expected);
+  t.deepEqual(loggedMessages.length, 1);
+  t.assert(loggedMessages[0].includes("Pruned 2 results"));
+});
+
+const affectedCodeQLVersion = {
+  driver: {
+    name: "CodeQL",
+    semanticVersion: "2.11.2",
+  },
+};
+
+const unaffectedCodeQLVersion = {
+  driver: {
+    name: "CodeQL",
+    semanticVersion: "2.11.3",
+  },
+};
+
+const otherTool = {
+  driver: {
+    name: "Some other tool",
+    semanticVersion: "2.11.2",
+  },
+};
+
+const resultWithOtherRuleId = {
+  ruleId: "doNotPrune",
+  message: {
+    text: "should not be pruned even though it says MD5 in it",
+  },
+  locations: [],
+  partialFingerprints: {},
+};
+
+const resultWithGoodMessage = {
+  ruleId: "rb/weak-cryptographic-algorithm",
+  message: {
+    text: "should not be pruned SHA128 is not a FP",
+  },
+  locations: [],
+  partialFingerprints: {},
+};
+
+const resultWithBadMessage1 = {
+  ruleId: "rb/weak-cryptographic-algorithm",
+  message: {
+    text: "should be pruned MD5 is a FP",
+  },
+  locations: [],
+  partialFingerprints: {},
+};
+
+const resultWithBadMessage2 = {
+  ruleId: "rb/weak-cryptographic-algorithm",
+  message: {
+    text: "should be pruned SHA1 is a FP",
+  },
+  locations: [],
+  partialFingerprints: {},
+};
+
 function createMockSarif(id?: string, tool?: string) {
   return {
     runs: [

src/upload-lib.ts

@@ -1,5 +1,6 @@
 import * as fs from "fs";
 import * as path from "path";
+import { env } from "process";
 import zlib from "zlib";
 
 import * as core from "@actions/core";
@@ -15,7 +16,7 @@ import { Logger } from "./logging";
 import { parseRepositoryNwo, RepositoryNwo } from "./repository";
 import * as sharedEnv from "./shared-environment";
 import * as util from "./util";
-import { SarifFile } from "./util";
+import { SarifFile, SarifResult, SarifRun } from "./util";
 
 // Takes a list of paths to sarif files and combines them together,
 // returning the contents of the combined sarif file.
@@ -396,6 +397,9 @@ async function uploadFiles(
     environment
   );
 
+  if (env["CODEQL_DISABLE_SARIF_PRUNING"] !== "true")
+    sarif = pruneInvalidResults(sarif, logger);
+
   const toolNames = util.getToolNames(sarif);
 
   validateUniqueCategory(sarif);
@@ -546,3 +550,42 @@ export function validateUniqueCategory(sarif: SarifFile): void {
 function sanitize(str?: string) {
   return (str ?? "_").replace(/[^a-zA-Z0-9_]/g, "_").toLocaleUpperCase();
 }
+
+export function pruneInvalidResults(
+  sarif: SarifFile,
+  logger: Logger
+): SarifFile {
+  let pruned = 0;
+  const newRuns: SarifRun[] = [];
+  for (const run of sarif.runs || []) {
+    if (
+      run.tool?.driver?.name === "CodeQL" &&
+      run.tool?.driver?.semanticVersion === "2.11.2"
+    ) {
+      // Version 2.11.2 of the CodeQL CLI had many false positives in the
+      // rb/weak-cryptographic-algorithm query which we prune here. The
+      // issue is tracked in https://github.com/github/codeql/issues/11107.
+      const newResults: SarifResult[] = [];
+      for (const result of run.results || []) {
+        if (
+          result.ruleId === "rb/weak-cryptographic-algorithm" &&
+          (result.message?.text?.includes(" MD5 ") ||
+            result.message?.text?.includes(" SHA1 "))
+        ) {
+          pruned += 1;
+          continue;
+        }
+        newResults.push(result);
+      }
+      newRuns.push({ ...run, results: newResults });
+    } else {
+      newRuns.push(run);
+    }
+  }
+  if (pruned > 0) {
+    logger.info(
+      `Pruned ${pruned} results believed to be invalid from SARIF file.`
+    );
+  }
+  return { ...sarif, runs: newRuns };
+}

src/util.ts

@@ -52,21 +52,28 @@ export const DID_AUTOBUILD_GO_ENV_VAR_NAME =
 
 export interface SarifFile {
   version?: string | null;
-  runs: Array<{
-    tool?: {
-      driver?: {
-        name?: string;
-      };
+  runs: SarifRun[];
+}
+
+export interface SarifRun {
+  tool?: {
+    driver?: {
+      name?: string;
+      semanticVersion?: string;
     };
-    automationDetails?: {
-      id?: string;
-    };
-    artifacts?: string[];
-    results?: SarifResult[];
-  }>;
+  };
+  automationDetails?: {
+    id?: string;
+  };
+  artifacts?: string[];
+  results?: SarifResult[];
 }
 
 export interface SarifResult {
+  ruleId?: string;
+  message?: {
+    text?: string;
+  };
   locations: Array<{
     physicalLocation: {
       artifactLocation: {