Fix for markup attributes (#1752)

This fixes some issues related to markup (XML/HTML) attributes. See the PR for more detail.
PrismJS · Feb 28, 2019 · c3862a2 · c3862a2
1 parent 5fd7577
commit c3862a2
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 18 deletions.
diff --git a/components/prism-markup.js b/components/prism-markup.js
@@ -4,7 +4,7 @@ Prism.languages.markup = {
 	'doctype': /<!DOCTYPE[\s\S]+?>/i,
 	'cdata': /<!\[CDATA\[[\s\S]*?]]>/i,
 	'tag': {
-		pattern: /<\/?(?!\d)[^\s>\/=$<%]+(?:\s+[^\s>\/=]+(?:=(?:("|')(?:\\[\s\S]|(?!\1)[^\\])*\1|[^\s'">=]+))?)*\s*\/?>/i,
+		pattern: /<\/?(?!\d)[^\s>\/=$<%]+(?:\s(?:\s*[^\s>\/=]+(?:\s*=\s*(?:"[^"]*"|'[^']*'|[^\s'">=]+(?=[\s>]))|(?=[\s/>])))+)?\s*\/?>/i,
 		greedy: true,
 		inside: {
 			'tag': {
@@ -15,12 +15,12 @@ Prism.languages.markup = {
 				}
 			},
 			'attr-value': {
-				pattern: /=(?:("|')(?:\\[\s\S]|(?!\1)[^\\])*\1|[^\s'">=]+)/i,
+				pattern: /=\s*(?:"[^"]*"|'[^']*'|[^\s'">=]+)/i,
 				inside: {
 					'punctuation': [
 						/^=/,
 						{
-							pattern: /(^|[^\\])["']/,
+							pattern: /^(\s*)["']|["']$/,
 							lookbehind: true
 						}
 					]

diff --git a/components/prism-markup.min.js b/components/prism-markup.min.js
diff --git a/prism.js b/prism.js
@@ -557,7 +557,7 @@ Prism.languages.markup = {
 	'doctype': /<!DOCTYPE[\s\S]+?>/i,
 	'cdata': /<!\[CDATA\[[\s\S]*?]]>/i,
 	'tag': {
-		pattern: /<\/?(?!\d)[^\s>\/=$<%]+(?:\s+[^\s>\/=]+(?:=(?:("|')(?:\\[\s\S]|(?!\1)[^\\])*\1|[^\s'">=]+))?)*\s*\/?>/i,
+		pattern: /<\/?(?!\d)[^\s>\/=$<%]+(?:\s(?:\s*[^\s>\/=]+(?:\s*=\s*(?:"[^"]*"|'[^']*'|[^\s'">=]+(?=[\s>]))|(?=[\s/>])))+)?\s*\/?>/i,
 		greedy: true,
 		inside: {
 			'tag': {
@@ -568,12 +568,12 @@ Prism.languages.markup = {
 				}
 			},
 			'attr-value': {
-				pattern: /=(?:("|')(?:\\[\s\S]|(?!\1)[^\\])*\1|[^\s'">=]+)/i,
+				pattern: /=\s*(?:"[^"]*"|'[^']*'|[^\s'">=]+)/i,
 				inside: {
 					'punctuation': [
 						/^=/,
 						{
-							pattern: /(^|[^\\])["']/,
+							pattern: /^(\s*)["']|["']$/,
 							lookbehind: true
 						}
 					]

diff --git a/tests/languages/markup/tag_attribute_feature.test b/tests/languages/markup/tag_attribute_feature.test
@@ -4,7 +4,8 @@
 bar
 baz">
 <div foo:bar=42>
-<div foo="=\"=">
+<div foo = 42 bar = "42">
+<div foo="=\"bar=baz/>
 
 ----------------------------------------------------
 
@@ -61,12 +62,12 @@ baz">
 		["attr-name", [
 			"class"
 		]],
-        ["attr-value", [
-            ["punctuation", "="],
-            ["punctuation", "\""],
-            "foo\r\nbar\r\nbaz",
-            ["punctuation", "\""]
-        ]],
+		["attr-value", [
+			["punctuation", "="],
+			["punctuation", "\""],
+			"foo\r\nbar\r\nbaz",
+			["punctuation", "\""]
+		]],
 		["punctuation", ">"]
 	]],
 
@@ -91,20 +92,43 @@ baz">
 			["punctuation", "<"],
 			"div"
 		]],
-		["attr-name", [
-			"foo"
+		["attr-name", ["foo"]],
+		["attr-value", [
+			["punctuation", "="],
+			" 42"
 		]],
+		["attr-name", ["bar"]],
 		["attr-value", [
 			["punctuation", "="],
 			["punctuation", "\""],
-			"=\\\"=",
+			"42",
 			["punctuation", "\""]
 		]],
 		["punctuation", ">"]
+	]],
+
+	["tag", [
+		["tag", [
+			["punctuation", "<"],
+			"div"
+		]],
+		["attr-name", ["foo"]],
+		["attr-value", [
+			["punctuation", "="],
+			["punctuation", "\""],
+			"=\\",
+			["punctuation", "\""]
+		]],
+		["attr-name", ["bar"]],
+		["attr-value", [
+			["punctuation", "="],
+			"baz/"
+		]],
+		["punctuation", ">"]
 	]]
 ]
 
 ----------------------------------------------------
 
 Checks for single-quoted, double-quoted and unquoted attributes, attributes without value and
-namespaced attributes.
+namespaced attributes.