Merge pull request #1827 from github/koesie10/filter-sarif-snippets

Limit SARIF code snippet size

extensions/ql-vscode/src/pure/sarif-utils.ts

@@ -1,5 +1,5 @@
 import * as Sarif from "sarif";
-import { HighlightedRegion } from "../remote-queries/shared/analysis-result";
+import type { HighlightedRegion } from "../remote-queries/shared/analysis-result";
 import { ResolvableLocationValue } from "./bqrs-cli-types";
 
 export interface SarifLink {

extensions/ql-vscode/src/remote-queries/sarif-processing.ts

@@ -1,5 +1,6 @@
 import * as sarif from "sarif";
 import {
+  parseHighlightedLine,
   parseSarifPlainTextMessage,
   parseSarifRegion,
 } from "../pure/sarif-utils";
@@ -15,6 +16,11 @@ import {
   HighlightedRegion,
 } from "./shared/analysis-result";
 
+// A line of more than 8k characters is probably generated.
+const CODE_SNIPPET_LARGE_LINE_SIZE_LIMIT = 8192;
+// If less than 1% of the line is highlighted, we consider it a small snippet.
+const CODE_SNIPPET_HIGHLIGHTED_REGION_MINIMUM_RATIO = 0.01;
+
 const defaultSeverity = "Warning";
 
 export function extractAnalysisAlerts(
@@ -163,17 +169,44 @@ export function tryGetRule(
 }
 
 function getCodeSnippet(
+  contextRegion?: sarif.Region,
   region?: sarif.Region,
-  alternateRegion?: sarif.Region,
 ): CodeSnippet | undefined {
-  region = region ?? alternateRegion;
+  const actualRegion = contextRegion ?? region;
 
-  if (!region) {
+  if (!actualRegion) {
     return undefined;
   }
 
-  const text = region.snippet?.text || "";
-  const { startLine, endLine } = parseSarifRegion(region);
+  const text = actualRegion.snippet?.text || "";
+  const { startLine, endLine } = parseSarifRegion(actualRegion);
+
+  if (
+    contextRegion &&
+    region &&
+    text.length > CODE_SNIPPET_LARGE_LINE_SIZE_LIMIT
+  ) {
+    const code = text.split("\n");
+
+    const highlightedRegion = parseSarifRegion(region);
+
+    const highlightedLines = code.map((line, index) => {
+      return parseHighlightedLine(line, startLine + index, highlightedRegion);
+    });
+
+    const highlightedCharactersCount = highlightedLines.reduce(
+      (a, line) => a + line.highlightedSection.length,
+      0,
+    );
+
+    const highlightedRatio = highlightedCharactersCount / text.length;
+
+    if (highlightedRatio < CODE_SNIPPET_HIGHLIGHTED_REGION_MINIMUM_RATIO) {
+      // If not enough is highlighted and the snippet is large, it's probably generated or bundled code and
+      // we don't want to show it.
+      return undefined;
+    }
+  }
 
   return {
     startLine,

extensions/ql-vscode/test/pure-tests/sarif-processing.test.ts

@@ -637,6 +637,43 @@ describe("SARIF processing", () => {
       expect(message.tokens[2].t).toBe("text");
       expect(message.tokens[2].text).toBe(".");
     });
+
+    it("should not include snippets for large snippets", () => {
+      const sarif = buildValidSarifLog();
+      // Build string of 10 kilobytes
+      const snippet = new Array(10 * 1024).fill("a").join("");
+      sarif.runs![0]!.results![0]!.locations![0]!.physicalLocation!.contextRegion!.snippet =
+        {
+          text: snippet,
+        };
+
+      const result = extractAnalysisAlerts(sarif, fakefileLinkPrefix);
+
+      const actualCodeSnippet = result.alerts[0].codeSnippet;
+
+      expect(result).toBeTruthy();
+      expectNoParsingError(result);
+      expect(actualCodeSnippet).toBeUndefined();
+    });
+
+    it("should include snippets for large snippets which are relevant", () => {
+      const sarif = buildValidSarifLog();
+      // Build string of 10 kilobytes
+      const snippet = new Array(10 * 1024).fill("a").join("");
+      sarif.runs![0]!.results![0]!.locations![0]!.physicalLocation!.contextRegion!.snippet =
+        {
+          text: snippet,
+        };
+      sarif.runs![0]!.results![0]!.locations![0]!.physicalLocation!.region!.endColumn = 1000;
+
+      const result = extractAnalysisAlerts(sarif, fakefileLinkPrefix);
+
+      const actualCodeSnippet = result.alerts[0].codeSnippet;
+
+      expect(result).toBeTruthy();
+      expectNoParsingError(result);
+      expect(actualCodeSnippet).not.toBeUndefined();
+    });
   });
 
   function expectResultParsingError(msg: string) {