Updated refa + fixed 2 cases of exp backtracking (#2774)

components/prism-asciidoc.js

@@ -129,7 +129,7 @@
 				'function': /^[a-z\d-]+(?=:)/,
 				'punctuation': /^::?/,
 				'attributes': {
-					pattern: /(?:\[(?:[^\]\\"]|(["'])(?:(?!\1)[^\\]|\\.)*\1|\\.)*\])/,
+					pattern: /(?:\[(?:[^\]\\"']|(["'])(?:(?!\1)[^\\]|\\.)*\1|\\.)*\])/,
 					inside: attributes.inside
 				}
 			}
@@ -149,7 +149,7 @@
 			Those do not have the restrictions of the constrained quotes.
 			They are, in order: __emphasis__, **strong**, ++monospace++, +++passthrough+++, ##unquoted##, $$passthrough$$, ~subscript~, ^superscript^, {attribute-reference}, [[anchor]], [[[bibliography anchor]]], <<xref>>, (((indexes))) and ((indexes))
 			 */
-			pattern: /(^|[^\\])(?:(?:\B\[(?:[^\]\\"]|(["'])(?:(?!\2)[^\\]|\\.)*\2|\\.)*\])?(?:\b_(?!\s)(?: _|[^_\\\r\n]|\\.)+(?:(?:\r?\n|\r)(?: _|[^_\\\r\n]|\\.)+)*_\b|\B``(?!\s).+?(?:(?:\r?\n|\r).+?)*''\B|\B`(?!\s)(?:[^`'\s]|\s+\S)+['`]\B|\B(['*+#])(?!\s)(?: \3|(?!\3)[^\\\r\n]|\\.)+(?:(?:\r?\n|\r)(?: \3|(?!\3)[^\\\r\n]|\\.)+)*\3\B)|(?:\[(?:[^\]\\"]|(["'])(?:(?!\4)[^\\]|\\.)*\4|\\.)*\])?(?:(__|\*\*|\+\+\+?|##|\$\$|[~^]).+?(?:(?:\r?\n|\r).+?)*\5|\{[^}\r\n]+\}|\[\[\[?.+?(?:(?:\r?\n|\r).+?)*\]?\]\]|<<.+?(?:(?:\r?\n|\r).+?)*>>|\(\(\(?.+?(?:(?:\r?\n|\r).+?)*\)?\)\)))/m,
+			pattern: /(^|[^\\])(?:(?:\B\[(?:[^\]\\"']|(["'])(?:(?!\2)[^\\]|\\.)*\2|\\.)*\])?(?:\b_(?!\s)(?: _|[^_\\\r\n]|\\.)+(?:(?:\r?\n|\r)(?: _|[^_\\\r\n]|\\.)+)*_\b|\B``(?!\s).+?(?:(?:\r?\n|\r).+?)*''\B|\B`(?!\s)(?:[^`'\s]|\s+\S)+['`]\B|\B(['*+#])(?!\s)(?: \3|(?!\3)[^\\\r\n]|\\.)+(?:(?:\r?\n|\r)(?: \3|(?!\3)[^\\\r\n]|\\.)+)*\3\B)|(?:\[(?:[^\]\\"']|(["'])(?:(?!\4)[^\\]|\\.)*\4|\\.)*\])?(?:(__|\*\*|\+\+\+?|##|\$\$|[~^]).+?(?:(?:\r?\n|\r).+?)*\5|\{[^}\r\n]+\}|\[\[\[?.+?(?:(?:\r?\n|\r).+?)*\]?\]\]|<<.+?(?:(?:\r?\n|\r).+?)*>>|\(\(\(?.+?(?:(?:\r?\n|\r).+?)*\)?\)\)))/m,
 			lookbehind: true,
 			inside: {
 				'attributes': attributes,

components/prism-perl.js

@@ -161,7 +161,7 @@ Prism.languages.perl = {
 		// ${...}
 		/[&*$@%]#?(?=\{)/,
 		// $foo
-		/[&*$@%]#?(?:(?:::)*'?(?!\d)[\w$]+)+(?:::)*/i,
+		/[&*$@%]#?(?:(?:::)*'?(?!\d)[\w$]+(?![\w$]))+(?:::)*/i,
 		// $1
 		/[&*$@%]\d+/,
 		// $_, @_, %!

package.json

@@ -50,7 +50,7 @@
     "jsdom": "^13.0.0",
     "mocha": "^6.2.0",
     "pump": "^3.0.0",
-    "refa": "0.5.0",
+    "refa": "^0.7.0",
     "regexpp": "^2.0.1",
     "scslre": "^0.1.1",
     "simple-git": "^1.107.0",

tests/pattern-tests.js

@@ -5,7 +5,7 @@ const PrismLoader = require('./helper/prism-loader');
 const { BFS, parseRegex } = require('./helper/util');
 const { languages } = require('../components.json');
 const { visitRegExpAST } = require('regexpp');
-const { JS, Words, NFA } = require('refa');
+const { transform, combineTransformers, JS, Words, NFA, Transformers } = require('refa');
 const scslre = require('scslre');
 
 /**
@@ -383,6 +383,22 @@ function testPatterns(Prism) {
 	});
 
 	it('- should not cause exponential backtracking', function () {
+		/** @type {Transformers.CreationOptions} */
+		const options = {
+			ignoreOrder: true,
+			ignoreAmbiguity: true
+		};
+		const transformer = combineTransformers([
+			Transformers.inline(options),
+			Transformers.removeDeadBranches(options),
+			Transformers.unionCharacters(options),
+			Transformers.moveUpEmpty(options),
+			Transformers.nestedQuantifiers(options),
+			Transformers.sortAssertions(options),
+			Transformers.removeUnnecessaryAssertions(options),
+			Transformers.applyAssertions(options),
+		]);
+
 		forEachPattern(({ pattern, ast, tokenPath }) => {
 			const patternStr = String(pattern);
 			if (expoSafeRegexes.has(patternStr)) {
@@ -398,11 +414,15 @@ function testPatterns(Prism) {
 			 * @returns {NFA}
 			 */
 			function toNFA(element) {
-				const { expression, maxCharacter } = parser.parseElement(element, {
-					backreferences: "resolve",
-					lookarounds: "disable",
+				let { expression, maxCharacter } = parser.parseElement(element, {
+					maxBackreferenceWords: 1000,
+					backreferences: "disable"
 				});
-				return NFA.fromRegex(expression, { maxCharacter });
+
+				// try to remove assertions
+				expression = transform(transformer, expression);
+
+				return NFA.fromRegex(expression, { maxCharacter }, { assertions: "disable" });
 			}
 
 			/**