fix(MediaWiki): HTML entity not fragment

bhsd-harry · Aug 24, 2024 · fc43e71 · fc43e71
1 parent afe44d8
commit fc43e71
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 8 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,11 +1,15 @@
 ## 2.17.4
 
-*2024-08-20*
+*2024-08-24*
 
 **Added**
 
 - [Autocompletion](./README.md#autocompletion) for the `group` attribute of `<ref>` and `<references>` in the MediaWiki mode
 
+**Fixed**
+
+- `#` in an HTML entity should not be regarded as link fragment in the MediaWiki mode
+
 ## 2.17.3
 
 *2024-08-17*

diff --git a/dist/main.min.js b/dist/main.min.js
diff --git a/src/token.ts b/src/token.ts
@@ -830,7 +830,7 @@ export class MediaWiki {
 	@getTokenizer
 	inLink(file: boolean, section?: boolean): Tokenizer {
 		const style = section ? tokens[file ? 'error' : 'linkToSection'] : `${tokens.linkPageName} ${tokens.pageName}`,
-			re = section ? /^(?:[^|<[\]{}]|<(?!!--|\/?[a-z]))+/iu : /^[^#|<>[\]{}]+/u;
+			re = section ? /^(?:[^|<[\]{}]|<(?!!--|\/?[a-z]))+/iu : /^(?:&#(?:\d+|x[a-f\d]+);|[^#|<>[\]{}])+/iu;
 		let lt: number | undefined;
 		return (stream, state) => {
 			if (stream.sol() || lt && stream.pos > lt || stream.match(/^\s*\]\]/u)) {
@@ -1410,7 +1410,7 @@ export class MediaWiki {
 	inTemplatePageName(haveEaten?: boolean, anchor?: boolean): Tokenizer {
 		const style = anchor ? tokens.error : `${tokens.templateName} ${tokens.pageName}`,
 			chars = '{}<',
-			re = anchor ? this.templateRegex : /^[^|{}<>[\]#]+/u;
+			re = anchor ? this.templateRegex : /^(?:&#(?:\d+|x[a-f\d]+);|[^|{}<>[\]#])+/iu;
 		return (stream, state) => {
 			const sol = stream.sol(),
 				space = stream.eatSpace();
@@ -1782,7 +1782,7 @@ export class MediaWiki {
 	inGallery(section?: boolean): Tokenizer {
 		const style = section ? tokens.error : `${tokens.linkPageName} ${tokens.pageName}`,
 			regex = section ? /^(?:[[}\]]|\{(?!\{))+/u : /^(?:[>[}\]]|\{(?!\{)|<(?!!--))+/u,
-			re = section ? /^(?:[^|<[\]{}]|<(?!!--))+/u : /^[^#|<>[\]{}]+/u;
+			re = section ? /^(?:[^|<[\]{}]|<(?!!--))+/u : /^(?:&#(?:\d+|x[a-f\d]+);|[^#|<>[\]{}])+/u;
 		return (stream, state) => {
 			const space = stream.eatSpace();
 			if (!section && stream.match(/^#\s*/u)) {

diff --git a/test/parserTests.json b/test/parserTests.json
@@ -635,7 +635,7 @@
 	{
 		"desc": "T2337: Escaped self-links should be bold",
 		"wikitext": "[[Bu&#103;462]] [[Bug462]]",
-		"parsed": "<link-ground link-bracket>[[</><link-ground link-pagename pagename>Bu&amp;</><link-ground link-tosection>#103;462</><link-ground link-bracket>]]</> <link-ground link-bracket>[[</><link-ground link-pagename pagename>Bug462</><link-ground link-bracket>]]</>"
+		"parsed": "<link-ground link-bracket>[[</><link-ground link-pagename pagename>Bu&amp;#103;462</><link-ground link-bracket>]]</> <link-ground link-bracket>[[</><link-ground link-pagename pagename>Bug462</><link-ground link-bracket>]]</>"
 	},
 	{
 		"desc": "Self-link to section should not be bold",
@@ -2375,7 +2375,7 @@
 	{
 		"desc": "(T21451) Links should refer to the normalized form.",
 		"wikitext": "[[&#xFB2E;]]\n[[&#x5d0;&#x5b7;]]\n[[&#x5d0;ַ]]\n[[א&#x5b7;]]\n[[אַ]]",
-		"parsed": "<link-ground link-bracket>[[</><link-ground link-pagename pagename>&amp;</><link-ground link-tosection>#xFB2E;</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>&amp;</><link-ground link-tosection>#x5d0;&amp;#x5b7;</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>&amp;</><link-ground link-tosection>#x5d0;ַ</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>א&amp;</><link-ground link-tosection>#x5b7;</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>אַ</><link-ground link-bracket>]]</>"
+		"parsed": "<link-ground link-bracket>[[</><link-ground link-pagename pagename>&amp;#xFB2E;</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>&amp;#x5d0;&amp;#x5b7;</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>&amp;#x5d0;ַ</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>א&amp;#x5b7;</><link-ground link-bracket>]][[</><link-ground link-pagename pagename>אַ</><link-ground link-bracket>]]</>"
 	},
 	{
 		"desc": "Empty attribute crash test (T4067)",
@@ -9771,7 +9771,7 @@
 	{
 		"desc": "Entities in media filenames",
 		"wikitext": "[[File:Hi&#45;ho.jpg|thumb]]\n\n<gallery>\nHi&#45;ho.jpg\nFile:Hi&#45;ho.jpg\n</gallery>",
-		"parsed": "<link-ground link-bracket>[[</><link-ground link-pagename pagename>File:Hi&amp;</><link-ground error>#45;ho.jpg</><link-ground link-delimiter>|</><link-ground image-parameter>thumb</><link-ground link-bracket>]]</><exttag-bracket>&lt;</><exttag-name>gallery</><exttag-bracket>&gt</><tag-gallery link-pagename pagename>Hi&amp;</><tag-gallery error>#45;ho.jpg</><tag-gallery link-pagename pagename>File:Hi&amp;</><tag-gallery error>#45;ho.jpg</><exttag-bracket>&lt;/</><exttag-name>gallery</><exttag-bracket>&gt</>"
+		"parsed": "<link-ground link-bracket>[[</><link-ground link-pagename pagename>File:Hi&amp;#45;ho.jpg</><link-ground link-delimiter>|</><link-ground image-parameter>thumb</><link-ground link-bracket>]]</><exttag-bracket>&lt;</><exttag-name>gallery</><exttag-bracket>&gt</><tag-gallery link-pagename pagename>Hi&amp;#45;ho.jpgFile:Hi&amp;#45;ho.jpg</><exttag-bracket>&lt;/</><exttag-name>gallery</><exttag-bracket>&gt</>"
 	},
 	{
 		"desc": "Inline gallery attributes are appended to the defaults",