fix: comment stripping should remove multiline comments (#279)

src/syntax.ts

@@ -1,4 +1,5 @@
 import { promises as fsp } from "node:fs";
+import { parse } from "acorn";
 import { extname } from "pathe";
 import { readPackageJSON } from "pkg-types";
 import { ResolveOptions, resolvePath } from "./resolve";
@@ -10,8 +11,6 @@
 const CJS_RE =
   /([\s;]|^)(module.exports\b|exports\.\w|require\s*\(|global\.\w)/m;
 
-const COMMENT_RE = /\/\*.+?\*\/|\/\/.*(?=[nr])/g;
-
 const BUILTIN_EXTENSIONS = new Set([".mjs", ".cjs", ".node", ".wasm"]);
 
 /**
@@ -25,6 +24,38 @@
   stripComments?: boolean;
 };
 
+interface TokenLocation {
+  start: number;
+  end: number;
+}
+
+function _getCommentLocations(code: string) {
+  const locations: TokenLocation[] = [];
+  parse(code, {
+    ecmaVersion: "latest",
+    allowHashBang: true,
+    allowAwaitOutsideFunction: true,
+    allowImportExportEverywhere: true,
+    onComment(_isBlock, _text, start, end) {
+      locations.push({ start, end });
+    },
+  });
+  return locations;
+}
+
+/**
+ * Strip comments from a string of source code
+ *
+ * @param code - The source code to remove comments from.
+ */
+export function stripComments(code: string) {
+  const locations = _getCommentLocations(code);
+  for (const location of locations.reverse()) {
+    code = code.slice(0, location.start) + code.slice(location.end);
+  }
+  return code;
+}
+
 /**
  * Determines if a given code string contains ECMAScript module syntax.
  *
@@ -37,7 +68,7 @@
   opts: DetectSyntaxOptions = {},
 ): boolean {
   if (opts.stripComments) {
-    code = code.replace(COMMENT_RE, "");
+    code = stripComments(code);
   }
   return ESM_RE.test(code);
 }
@@ -54,7 +85,7 @@
   opts: DetectSyntaxOptions = {},
 ): boolean {
   if (opts.stripComments) {
-    code = code.replace(COMMENT_RE, "");
+    code = stripComments(code);
   }
   return CJS_RE.test(code);
 }
@@ -68,7 +99,7 @@
  */
 export function detectSyntax(code: string, opts: DetectSyntaxOptions = {}) {
   if (opts.stripComments) {
-    code = code.replace(COMMENT_RE, "");
+    code = stripComments(code);
   }
   // We strip comments once hence not passing opts down to hasESMSyntax and hasCJSSyntax
   const hasESM = hasESMSyntax(code, {});

test/syntax.test.ts

@@ -1,6 +1,6 @@
 import { join } from "pathe";
 import { describe, it, expect } from "vitest";
-import { detectSyntax, isValidNodeImport } from "../src";
+import { detectSyntax, isValidNodeImport, stripComments } from "../src";
 
 const staticTests = {
   // ESM
@@ -92,6 +92,21 @@ const staticTests = {
 const staticTestsWithComments = {
   '// They\'re exposed using "export import" so that types are passed along as expected\nmodule.exports={};':
     { hasESM: false, hasCJS: true, isMixed: false },
+  "/* export * */": { hasESM: false, hasCJS: false, isMixed: false },
+  "/* \n export * \n */": { hasESM: false, hasCJS: false, isMixed: false },
+  "/* \n export * \n */ export * from 'foo' /* \n export * \n */": {
+    hasESM: true,
+    hasCJS: false,
+    isMixed: false,
+  },
+};
+
+const commentStrippingTests = {
+  '// They\'re exposed using "export import" so that types are passed along as expected\nmodule.exports={};':
+    "\nmodule.exports={};",
+  "/* export * */": "",
+  "/* \n export * \n */": "",
+  "/* \n export * \n */ export * from 'foo' /* \n export * \n */": ` export * from 'foo' `,
 };
 
 describe("detectSyntax", () => {
@@ -112,6 +127,14 @@ describe("detectSyntax (with comment)", () => {
   }
 });
 
+describe("stripComments", () => {
+  for (const [input, result] of Object.entries(commentStrippingTests)) {
+    it(input, () => {
+      expect(stripComments(input)).to.deep.equal(result);
+    });
+  }
+});
+
 const nodeImportTests = {
   "node:fs": true,
   fs: true,