Tests: Added new pattern check for octal escapes (#2189)

This adds a new pattern check to disallow octal character escapes (e.g. `\12`).

components/prism-bash.js

@@ -107,15 +107,15 @@
 		'string': [
 			// Support for Here-documents https://en.wikipedia.org/wiki/Here_document
 			{
-				pattern: /((?:^|[^<])<<-?\s*)(\w+?)\s*(?:\r?\n|\r)(?:[\s\S])*?(?:\r?\n|\r)\2/,
+				pattern: /((?:^|[^<])<<-?\s*)(\w+?)\s*(?:\r?\n|\r)[\s\S]*?(?:\r?\n|\r)\2/,
 				lookbehind: true,
 				greedy: true,
 				inside: insideString
 			},
 			// Here-document with quotes around the tag
 			// → No expansion (so no “inside”).
 			{
-				pattern: /((?:^|[^<])<<-?\s*)(["'])(\w+)\2\s*(?:\r?\n|\r)(?:[\s\S])*?(?:\r?\n|\r)\3/,
+				pattern: /((?:^|[^<])<<-?\s*)(["'])(\w+)\2\s*(?:\r?\n|\r)[\s\S]*?(?:\r?\n|\r)\3/,
 				lookbehind: true,
 				greedy: true
 			},

components/prism-shell-session.js

@@ -1,21 +1,42 @@
-Prism.languages['shell-session'] = {
-	'command': {
-		pattern: /\$(?:[^\r\n'"<]|(["'])(?:\\[\s\S]|\$\([^)]+\)|`[^`]+`|(?!\1)[^\\])*\1|(?:^|[^<])<<\s*["']?(?:\w+?)["']?\s*(?:\r\n?|\n)(?:[\s\S])*?(?:\r\n?|\n)\3)+/,
-		inside: {
-			'bash': {
-				pattern: /(\$\s*)[\s\S]+/,
-				lookbehind: true,
-				alias: 'language-bash',
-				inside: Prism.languages.bash
-			},
-			'sh': {
-				pattern: /^\$/,
-				alias: 'important'
+(function (Prism) {
+
+	// CAREFUL!
+	// The following patterns are concatenated, so the group referenced by a back reference is non-obvious!
+
+	var strings = [
+		// normal string
+		// 1 capturing group
+		/(["'])(?:\\[\s\S]|\$\([^)]+\)|`[^`]+`|(?!\1)[^\\])*\1/.source,
+
+		// here doc
+		// 1 capturing group
+		/<<-?\s*(\w+?)\s*(?:\r?\n|\r)[\s\S]*?(?:\r?\n|\r)\2/.source,
+
+		// here doc quoted
+		// 2 capturing group
+		/<<-?\s*(["'])(\w+)\3\s*(?:\r?\n|\r)[\s\S]*?(?:\r?\n|\r)\4/.source
+	].join('|');
+
+	Prism.languages['shell-session'] = {
+		'command': {
+			pattern: RegExp(/\$(?:[^\r\n'"<]|<<str>>)+/.source.replace(/<<str>>/g, strings)),
+			inside: {
+				'bash': {
+					pattern: /(\$\s*)[\s\S]+/,
+					lookbehind: true,
+					alias: 'language-bash',
+					inside: Prism.languages.bash
+				},
+				'sh': {
+					pattern: /^\$/,
+					alias: 'important'
+				}
 			}
+		},
+		'output': {
+			pattern: /.(?:.*(?:\r\n?|\n|.$))*/
+			// output highlighting?
 		}
-	},
-	'output': {
-		pattern: /.(?:.*(?:\r\n?|\n|.$))*/
-		// output highlighting?
-	}
-}
+	};
+
+}(Prism));

tests/languages/shell-session/command_string_feature.test

@@ -3,6 +3,16 @@ $ echo 'Foo
 $ echo "Foo
 > Bar"
 
+$ echo <<- STRING_END
+foo
+bar
+STRING_END
+
+$ echo <<- "STRING_END"
+foo
+bar
+STRING_END
+
 ----------------------------------------------------
 
 [
@@ -24,6 +34,30 @@ $ echo "Foo
 				"\"Foo\n> Bar\""
 			]]
 		]]
+	]],
+
+	["command", [
+		["sh", "$"],
+		["bash", [
+			["builtin", "echo"],
+			["operator", [
+				"<<-"
+			]],
+			["string", [
+				"STRING_END\nfoo\nbar\nSTRING_END"
+			]]
+		]]
+	]],
+
+	["command", [
+		["sh", "$"],
+		["bash", [
+			["builtin", "echo"],
+			["operator", [
+				"<<-"
+			]],
+			["string", "\"STRING_END\"\nfoo\nbar\nSTRING_END"]
+		]]
 	]]
 ]
 

tests/pattern-tests.js

@@ -264,4 +264,16 @@ function testPatterns(Prism) {
 		});
 	});
 
+	it('- should not use octal escapes', function () {
+		forEachPattern(({ ast, tokenPath, reportError }) => {
+			visitRegExpAST(ast.pattern, {
+				onCharacterEnter(node) {
+					if (/^\\(?:[1-9]|\d{2,})$/.test(node.raw)) {
+						reportError(`Token ${tokenPath}: Octal escape ${node.raw}.`);
+					}
+				}
+			});
+		});
+	});
+
 }