Extract explicit CodeQL bundle URL classification

Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

src/codeql-bundle.test.ts

@@ -0,0 +1,56 @@
+import test from "ava";
+
+import { getCodeQLBundleFromUrl } from "./codeql-bundle";
+import { BuiltInLanguage } from "./languages";
+
+for (const [assetName, language] of [
+  ["codeql-bundle-java-linux64.tar.zst", BuiltInLanguage.java],
+  ["codeql-bundle-swift-osx64.tar.zst", BuiltInLanguage.swift],
+  // Recognize unpublished language/platform combinations to keep them out of the toolcache.
+  ["codeql-bundle-csharp-win64.tar.gz", BuiltInLanguage.csharp],
+  ["codeql-bundle-java-kotlin-linux64.tar.zst", BuiltInLanguage.java],
+  ["codeql-bundle-%70ython-linux64.tar.zst", BuiltInLanguage.python],
+] as const) {
+  test(`getCodeQLBundleFromUrl identifies ${assetName} without adding a fallback`, (t) => {
+    const url = `https://github.com/github/codeql-action/releases/download/codeql-bundle-v1.2.3/${assetName}`;
+    t.deepEqual(getCodeQLBundleFromUrl(url), {
+      kind: "per-language",
+      url,
+      language,
+    });
+  });
+}
+
+test("getCodeQLBundleFromUrl preserves encoding, query parameters and fragments", (t) => {
+  const url =
+    "https://github.com/github/codeql-action/releases/download/codeql-bundle-v1.2.3/codeql-bundle-%70ython-linux64.tar.zst?download=1#asset";
+  t.deepEqual(getCodeQLBundleFromUrl(url), {
+    kind: "per-language",
+    url,
+    language: BuiltInLanguage.python,
+  });
+});
+
+test("getCodeQLBundleFromUrl treats unrecognized assets as combined bundles", (t) => {
+  for (const name of [
+    "codeql-bundle-linux64.tar.zst",
+    "codeql-bundle-osx64.tar.gz",
+    "codeql-bundle-win64.tar.zst",
+    // The all-platform bundle.
+    "codeql-bundle.tar.gz",
+    // A platform we do not publish per-language bundles for, whose name also contains a hyphen.
+    "codeql-bundle-linux-arm64.tar.zst",
+    // Not a language we know about.
+    "codeql-bundle-cobol-linux64.tar.zst",
+    // A name we cannot decode must not be mistaken for a language either.
+    "codeql-bundle-%zz-linux64.tar.zst",
+  ]) {
+    const url = `https://github.com/github/codeql-action/releases/download/codeql-bundle-v1.2.3/${name}`;
+    t.deepEqual(getCodeQLBundleFromUrl(url), { kind: "combined", url });
+  }
+});
+
+test("getCodeQLBundleFromUrl preserves URLs it cannot parse", (t) => {
+  const url = "not a url";
+  t.deepEqual(getCodeQLBundleFromUrl(url), { kind: "combined", url });
+});

src/codeql-bundle.ts

@@ -0,0 +1,33 @@
+import { BuiltInLanguage, parseBuiltInLanguage } from "./languages";
+
+/** Describes the contents and location of a downloadable CodeQL bundle. */
+export type CodeQLBundle =
+  | { kind: "combined"; url: string }
+  | {
+      kind: "per-language";
+      url: string;
+      language: BuiltInLanguage;
+      /** Only set when the Action selected the bundle, allowing a same-version fallback. */
+      combinedBundleURL?: string;
+    };
+
+const PER_LANGUAGE_BUNDLE_NAME =
+  /^codeql-bundle-(.+)-(?:linux64|osx64|win64)\.tar\.(?:gz|zst)$/;
+
+/** Classifies an explicit tools URL without changing it or adding a fallback. */
+export function getCodeQLBundleFromUrl(url: string): CodeQLBundle {
+  let assetName: string;
+  try {
+    const pathname = new URL(url).pathname;
+    // URL-encoded names must not bypass the toolcache safeguard.
+    assetName = decodeURIComponent(pathname.split("/").pop() ?? "");
+  } catch {
+    return { kind: "combined", url };
+  }
+
+  const match = assetName.match(PER_LANGUAGE_BUNDLE_NAME);
+  const language = match ? parseBuiltInLanguage(match[1]) : undefined;
+  return language === undefined
+    ? { kind: "combined", url }
+    : { kind: "per-language", url, language };
+}

src/per-language-bundles.test.ts

@@ -8,7 +8,6 @@ import {
   getPerLanguageBundleLanguage,
   MIN_PER_LANGUAGE_BUNDLE_CLI_VERSION,
   PerLanguageBundleOptions,
-  tryGetBundleLanguageFromUrl,
 } from "./per-language-bundles";
 import {
   createFeatures,
@@ -172,51 +171,3 @@ test("getPerLanguageBundleLanguage skips only the release version check for the
     undefined,
   );
 });
-
-test("tryGetBundleLanguageFromUrl recognizes per-language bundle URLs", (t) => {
-  const url = (name: string) =>
-    `https://github.com/github/codeql-action/releases/download/codeql-bundle-v1.2.3/${name}`;
-
-  t.is(
-    tryGetBundleLanguageFromUrl(url("codeql-bundle-java-linux64.tar.zst")),
-    BuiltInLanguage.java,
-  );
-  t.is(
-    tryGetBundleLanguageFromUrl(url("codeql-bundle-swift-osx64.tar.zst")),
-    BuiltInLanguage.swift,
-  );
-  // We do not publish these, but should still recognize them if we ever do.
-  t.is(
-    tryGetBundleLanguageFromUrl(url("codeql-bundle-csharp-win64.tar.gz")),
-    BuiltInLanguage.csharp,
-  );
-  // A percent-encoded name resolves to the same asset, so it must not let a bundle that contains a
-  // single language pass for one that contains them all and end up in the toolcache.
-  t.is(
-    tryGetBundleLanguageFromUrl(url("codeql-bundle-%70ython-linux64.tar.zst")),
-    BuiltInLanguage.python,
-  );
-});
-
-test("tryGetBundleLanguageFromUrl rejects other bundle URLs", (t) => {
-  const url = (name: string) =>
-    `https://github.com/github/codeql-action/releases/download/codeql-bundle-v1.2.3/${name}`;
-
-  for (const name of [
-    "codeql-bundle-linux64.tar.zst",
-    "codeql-bundle-osx64.tar.gz",
-    "codeql-bundle-win64.tar.zst",
-    // The all-platform bundle.
-    "codeql-bundle.tar.gz",
-    // A platform we do not publish per-language bundles for, whose name also contains a hyphen.
-    "codeql-bundle-linux-arm64.tar.zst",
-    // Not a language we know about.
-    "codeql-bundle-cobol-linux64.tar.zst",
-    // A name we cannot decode must not be mistaken for a language either.
-    "codeql-bundle-%zz-linux64.tar.zst",
-  ]) {
-    t.is(tryGetBundleLanguageFromUrl(url(name)), undefined, name);
-  }
-
-  t.is(tryGetBundleLanguageFromUrl("not a url"), undefined);
-});

src/per-language-bundles.ts

@@ -11,26 +11,6 @@ import { GitHubVariant } from "./util";
 /** Minimum CLI version for selecting a per-language release bundle. */
 export const MIN_PER_LANGUAGE_BUNDLE_CLI_VERSION = "2.27.1";
 
-const PER_LANGUAGE_BUNDLE_NAME =
-  /^codeql-bundle-(.+)-(?:linux64|osx64|win64)\.tar\.(?:gz|zst)$/;
-
-/** Returns the language in a per-language tools URL, or undefined for other URLs. */
-export function tryGetBundleLanguageFromUrl(
-  url: string,
-): BuiltInLanguage | undefined {
-  let assetName: string;
-  try {
-    const pathname = new URL(url).pathname;
-    // URL-encoded names must not bypass the toolcache safeguard.
-    assetName = decodeURIComponent(pathname.split("/").pop() ?? "");
-  } catch {
-    return undefined;
-  }
-
-  const match = assetName.match(PER_LANGUAGE_BUNDLE_NAME);
-  return match ? parseBuiltInLanguage(match[1]) : undefined;
-}
-
 /** Languages with per-language bundles published for each platform. */
 const PER_LANGUAGE_BUNDLE_LANGUAGES: Readonly<
   Record<BundlePlatform, ReadonlySet<BuiltInLanguage>>

src/setup-codeql.ts

@@ -18,6 +18,7 @@ import {
 } from "./actions-util";
 import * as api from "./api-client";
 import { getBundlePlatform } from "./bundle-platform";
+import { CodeQLBundle, getCodeQLBundleFromUrl } from "./codeql-bundle";
 import * as defaults from "./defaults.json";
 import {
   addNoLanguageDiagnostic,
@@ -35,10 +36,7 @@ import {
 import { BuiltInLanguage } from "./languages";
 import { Logger } from "./logging";
 import { getCodeQlVersionsForOverlayBaseDatabases } from "./overlay/caching";
-import {
-  getPerLanguageBundleLanguage,
-  tryGetBundleLanguageFromUrl,
-} from "./per-language-bundles";
+import { getPerLanguageBundleLanguage } from "./per-language-bundles";
 import * as tar from "./tar";
 import {
   deleteToolcacheBundles,
@@ -225,17 +223,6 @@ export function convertToSemVer(version: string, logger: Logger): string {
   return s;
 }
 
-/** Describes the contents and location of a downloadable CodeQL bundle. */
-type CodeQLBundle =
-  | { kind: "combined"; url: string }
-  | {
-      kind: "per-language";
-      url: string;
-      language: BuiltInLanguage;
-      /** Only set when the Action selected the bundle, allowing a same-version fallback. */
-      combinedBundleURL?: string;
-    };
-
 /** A resolved download, including its bundle identity and version. */
 export interface CodeQLDownloadSource {
   /** Distinguishes downloads from local archives and cached installations. */
@@ -816,14 +803,7 @@ export async function getCodeQLSource(
     }
     compressionMethod = method;
 
-    if (bundle === undefined) {
-      // Explicit per-language URLs must also stay out of the toolcache, but have no fallback.
-      const language = tryGetBundleLanguageFromUrl(url);
-      bundle =
-        language === undefined
-          ? { kind: "combined", url }
-          : { kind: "per-language", url, language };
-    }
+    bundle ??= getCodeQLBundleFromUrl(url);
     if (bundle.kind === "per-language") {
       logger.info(
         `${url} appears to be a CodeQL bundle that contains only ${bundle.language}.`,