diff --git a/test-suite.html b/test-suite.html
index 2e2e171b3b..e1079901ab 100644
--- a/test-suite.html
+++ b/test-suite.html
@@ -93,8 +93,11 @@ <h2 id="writing-tests-writing-your-test">Writing your test</h2>
 
 	<p>Your file is built up of two or three sections, separated by ten or more dashes <code>-</code>, starting at the begin of the line:</p>
 	<ol>
-		<li>Your language snippet. The code you want to compile using Prism. (<strong>required</strong>)</li>
-		<li>The simplified token stream you expect. Needs to be valid JSON. (<strong>required</strong>)</li>
+		<li>Your language snippet. The code you want to tokenize using Prism. (<strong>required</strong>)</li>
+		<li>
+			The simplified token stream you expect. Needs to be valid JSON. (<em>optional</em>) <br>
+			If there no token stream defined, the test case will fail unless the <code>--accept</code> flag is present when running the test command (e.g. <code>npm run test:languages -- --accept</code>). If the flag is present and there is no expected token stream, the runner will insert the actual token stream into the test case file, changing it.
+		</li>
 		<li>A comment explaining the test case. (<em>optional</em>)</li>
 	</ol>
 	<p>The easiest way would be to look at an existing test file:</p>
@@ -114,10 +117,25 @@ <h2 id="writing-tests-writing-your-test">Writing your test</h2>
 
 This is a comment explaining this test case.</code></pre>
 
+	<h2 id="writing-tests-the-easy-way">The easy way</h2>
+	<p>The easy way to create one or multiple new test case(s) is this:</p>
+
+	<ol>
+		<li>Create a new file for a new test case in <code>tests/languages/${language}</code>.</li>
+		<li>Insert the code you want to test (and nothing more).</li>
+		<li>Repeat the first two steps for as many test cases as you want.</li>
+		<li>Run <code>npm run test:languages -- --accept</code>.</li>
+		<li>Done.</li>
+	</ol>
+
+	<p>This works by making the test runner insert the actual token stream of you test code as the expected token stream. <strong>Carefully check that the inserted token stream is actually what you expect or else the test is meaningless!</strong></p>
+
+	<p>Optionally, you can then also add comments to test cases.</p>
+
 
 	<h2 id="writing-tests-explaining-the-simplified-token-stream">Explaining the simplified token stream</h2>
 
-	<p>While compiling, Prism transforms your source code into a token stream. This is basically a tree of nested tokens (or arrays, or strings).</p>
+	<p>While highlighting, Prism transforms your source code into a token stream. This is basically a tree of nested tokens (or arrays, or strings).</p>
 	<p>As these trees are hard to write by hand, the test runner uses a simplified version of it.</p>
 	<p>It uses the following rules:</p>
 	<ul>
diff --git a/tests/helper/test-case.js b/tests/helper/test-case.js
index 0ebd277e8c..a00ceda1aa 100644
--- a/tests/helper/test-case.js
+++ b/tests/helper/test-case.js
@@ -49,43 +49,66 @@ module.exports = {
 	 *
 	 * @param {string} languageIdentifier
 	 * @param {string} filePath
+	 * @param {boolean} acceptEmpty
 	 */
-	runTestCase(languageIdentifier, filePath) {
+	runTestCase(languageIdentifier, filePath, acceptEmpty) {
 		const testCase = this.parseTestCaseFile(filePath);
 		const usedLanguages = this.parseLanguageNames(languageIdentifier);
 
-		if (null === testCase) {
-			throw new Error("Test case file has invalid format (or the provided token stream is invalid JSON), please read the docs.");
-		}
-
 		const Prism = PrismLoader.createInstance(usedLanguages.languages);
 
 		// the first language is the main language to highlight
-		const simplifiedTokenStream = this.simpleTokenize(Prism, testCase.testSource, usedLanguages.mainLanguage);
+		const simplifiedTokenStream = this.simpleTokenize(Prism, testCase.code, usedLanguages.mainLanguage);
+
+		if (testCase.expectedTokenStream === null) {
+			// the test case doesn't have an expected value
+			if (!acceptEmpty) {
+				throw new Error('This test case doesn\'t have an expected toke n stream.'
+					+ ' Either add the JSON of a token stream or run \`npm run test:languages -- --accept\`'
+					+ ' to automatically add the current token stream.');
+			}
+
+			// change the file
+			const lineEnd = (/\r\n/.test(testCase.code) || !/\n/.test(testCase.code)) ? '\r\n' : '\n';
+			const separator = "\n\n----------------------------------------------------\n\n";
+			const pretty = TokenStreamTransformer.prettyprint(simplifiedTokenStream)
+				.replace(/^( +)/gm, m => {
+					return "\t".repeat(m.length / 4);
+				});
+
+			let content = testCase.code + separator + pretty;
+			if (testCase.comment) {
+				content += separator + testCase.comment;
+			}
+			content = content.replace(/\r?\n/g, lineEnd);
 
-		const actual = JSON.stringify(simplifiedTokenStream);
-		const expected = JSON.stringify(testCase.expectedTokenStream);
+			fs.writeFileSync(filePath, content, "utf-8");
+		} else {
+			// there is an expected value
+			const actual = JSON.stringify(simplifiedTokenStream);
+			const expected = JSON.stringify(testCase.expectedTokenStream);
 
-		if (actual === expected) {
-			// no difference
-			return;
-		}
+			if (actual === expected) {
+				// no difference
+				return;
+			}
 
-		// The index of the first difference between the expected token stream and the actual token stream.
-		// The index is in the raw expected token stream JSON of the test case.
-		const diffIndex = translateIndexIgnoreSpaces(testCase.expectedJson, expected, firstDiff(expected, actual));
-		const expectedJsonLines = testCase.expectedJson.substr(0, diffIndex).split(/\r\n?|\n/g);
-		const columnNumber = expectedJsonLines.pop().length + 1;
-		const lineNumber = testCase.expectedLineOffset + expectedJsonLines.length;
-
-		const tokenStreamStr = TokenStreamTransformer.prettyprint(simplifiedTokenStream);
-		const message = "\n\nActual Token Stream:" +
-			"\n-----------------------------------------\n" +
-			tokenStreamStr +
-			"\n-----------------------------------------\n" +
-			"File: " + filePath + ":" + lineNumber + ":" + columnNumber + "\n\n";
-
-		assert.deepEqual(simplifiedTokenStream, testCase.expectedTokenStream, testCase.comment + message);
+			// The index of the first difference between the expected token stream and the actual token stream.
+			// The index is in the raw expected token stream JSON of the test case.
+			const diffIndex = translateIndexIgnoreSpaces(testCase.expectedJson, expected, firstDiff(expected, actual));
+			const expectedJsonLines = testCase.expectedJson.substr(0, diffIndex).split(/\r\n?|\n/g);
+			const columnNumber = expectedJsonLines.pop().length + 1;
+			const lineNumber = testCase.expectedLineOffset + expectedJsonLines.length;
+
+			const tokenStreamStr = TokenStreamTransformer.prettyprint(simplifiedTokenStream);
+			const message = "\n\nActual Token Stream:" +
+				"\n-----------------------------------------\n" +
+				tokenStreamStr +
+				"\n-----------------------------------------\n" +
+				"File: " + filePath + ":" + lineNumber + ":" + columnNumber + "\n\n";
+
+			assert.deepEqual(simplifiedTokenStream, testCase.expectedTokenStream, testCase.comment + message);
+		}
 	},
 
 	/**
@@ -160,33 +183,36 @@ module.exports = {
 	 *
 	 * @private
 	 * @param {string} filePath
-	 * @returns {{testSource: string, expectedTokenStream: Array<string[]>, comment:string?}|null}
+	 * @returns {ParsedTestCase}
+	 *
+	 * @typedef ParsedTestCase
+	 * @property {string} code
+	 * @property {string} expectedJson
+	 * @property {number} expectedLineOffset
+	 * @property {Array | null} expectedTokenStream
+	 * @property {string} comment
 	 */
 	parseTestCaseFile(filePath) {
 		const testCaseSource = fs.readFileSync(filePath, "utf8");
-		const testCaseParts = testCaseSource.split(/^-{10,}\w*$/m);
-
-		try {
-			const testCase = {
-				testSource: testCaseParts[0].trim(),
-				expectedJson: testCaseParts[1],
-				expectedLineOffset: testCaseParts[0].split(/\r\n?|\n/g).length,
-				expectedTokenStream: JSON.parse(testCaseParts[1]),
-				comment: null
-			};
-
-			// if there are three parts, the third one is the comment
-			// explaining the test case
-			if (testCaseParts[2]) {
-				testCase.comment = testCaseParts[2].trim();
-			}
+		const testCaseParts = testCaseSource.split(/^-{10,}[ \t]*$/m);
 
-			return testCase;
-		}
-		catch (e) {
-			// the JSON can't be parsed (e.g. it could be empty)
-			return null;
+		if (testCaseParts.length > 3) {
+			throw new Error("Invalid test case format: Too many sections.");
 		}
+
+		const code = testCaseParts[0].trim();
+		const expected = (testCaseParts[1] || '').trim();
+		const comment = (testCaseParts[2] || '').trim();
+
+		const testCase = {
+			code,
+			expectedJson: expected,
+			expectedLineOffset: code.split(/\r\n?|\n/g).length,
+			expectedTokenStream: expected ? JSON.parse(expected) : null,
+			comment
+		};
+
+		return testCase;
 	},
 
 	/**
diff --git a/tests/run.js b/tests/run.js
index b9c8893ffe..4803dde72b 100644
--- a/tests/run.js
+++ b/tests/run.js
@@ -12,6 +12,8 @@ const testSuite =
 		// load complete test suite
 		: TestDiscovery.loadAllTests(__dirname + "/languages");
 
+const accept = !!argv.accept;
+
 // define tests for all tests in all languages in the test suite
 for (const language in testSuite) {
 	if (!testSuite.hasOwnProperty(language)) {
@@ -27,7 +29,7 @@ for (const language in testSuite) {
 
 				it("– should pass test case '" + fileName + "'", function () {
 					if (path.extname(filePath) === '.test') {
-						TestCase.runTestCase(language, filePath);
+						TestCase.runTestCase(language, filePath, accept);
 					} else {
 						TestCase.runTestsWithHooks(language, require(filePath));
 					}