From cb1cb610ac08ba78d9f8735402929722d7253088 Mon Sep 17 00:00:00 2001
From: dklimpel <5740567+dklimpel@users.noreply.github.com>
Date: Sun, 2 Jun 2024 15:14:02 +0200
Subject: [PATCH 1/2] feat: anchor link checks support HTML tags like `<a
 name="foo"></a>`

---
 index.js                         | 27 ++++++++++++++++++++++--
 test/hash-links.md               | 35 ++++++++++++++++++++++++++++++--
 test/markdown-link-check.test.js |  9 +++++++-
 3 files changed, 66 insertions(+), 5 deletions(-)

diff --git a/index.js b/index.js
index cc4c9ae..98c76ab 100644
--- a/index.js
+++ b/index.js
@@ -43,9 +43,32 @@ function performSpecialReplacements(str, opts) {
     return str;
 }
 
+function removeCodeBlocks(markdown) {
+    return markdown.replace(/^```[\S\s]+?^```$/gm, '');
+}
+
+function extractHtmlSections(markdown) {
+    markdown =
+        // remove code blocks
+        removeCodeBlocks(markdown)
+        // remove HTML comments
+        .replace(/<!--[\S\s]+?-->/gm, '')
+        // remove single line code (if not escaped with "\")
+        .replace(/(?<!\\)`[\S\s]+?(?<!\\)`/gm, '');
+
+    const regexAllId = /<(?<tag>[^\s]+).*?id="(?<id>[^"]*?)".*?>/gmi;
+    const regexAName = /<a.*?name="(?<name>[^"]*?)".*?>/gmi;
+
+    const sections = []
+        .concat(Array.from(markdown.matchAll(regexAllId), (match) => match.groups.id))
+        .concat(Array.from(markdown.matchAll(regexAName), (match) => match.groups.name));
+
+    return sections
+}
+
 function extractSections(markdown) {
     // First remove code blocks.
-    markdown = markdown.replace(/^```[\S\s]+?^```$/mg, '');
+    markdown = removeCodeBlocks(markdown);
 
     const sectionTitles = markdown.match(/^#+ .*$/gm) || [];
 
@@ -85,7 +108,7 @@ module.exports = function markdownLinkCheck(markdown, opts, callback) {
     }
 
     const links = markdownLinkExtractor(markdown);
-    const sections = extractSections(markdown);
+    const sections = extractSections(markdown).concat(extractHtmlSections(markdown));
     const linksCollection = _.uniq(links);
     const bar = (opts.showProgressBar) ?
         new ProgressBar('Checking... [:bar] :percent', {
diff --git a/test/hash-links.md b/test/hash-links.md
index 4db835b..e9e53de 100644
--- a/test/hash-links.md
+++ b/test/hash-links.md
@@ -1,11 +1,28 @@
 # Foo
 
 <!-- markdownlint-disable MD033 -->
-<a id="tomato"></a>
+<a id="tomato_id"></a>
+
+<a name="tomato_name"></a>
+
+<div id="onion"></div>
+<div id="onion_outer">
+    <div id="onion_inner"></div>
+</div>
+
+<!--
+<a id="tomato_comment"></a>
+-->
+
 <!-- markdownlint-enable MD033 -->
 
 This is a test.
 
+HTML anchor in code `<a id="tomato_code"></a>` should be ignored.
+
+<!-- markdownlint-disable-next-line MD033 -->
+Ignore escaped backticks \`<a id="tomato_escaped_backticks"></a>\`. Link should work.
+
 ## Bar
 
 The title is [Foo](#foo).
@@ -18,7 +35,21 @@ The second section is [Bar](#bar).
 
 There is no section named [Potato](#potato).
 
-There is an anchor named [Tomato](#tomato).
+There is an anchor named with `id` [Tomato](#tomato_id).
+
+There is an anchor named with `name` [Tomato](#tomato_name).
+
+There is an anchor in code [Tomato in code](#tomato_code).
+
+There is an anchor in escaped code [Tomato in escaped backticks](#tomato_escaped_backticks).
+
+There is an anchor in HTML comment [Tomato in comment](#tomato_comment).
+
+There is an anchor in single div [Onion](#onion).
+
+There is an anchor in outer div [Onion outer](#onion_outer).
+
+There is an anchor in inner div [Onion inner](#onion_inner).
 
 ## Header with special char ✨
 
diff --git a/test/markdown-link-check.test.js b/test/markdown-link-check.test.js
index a46cfc5..eb77212 100644
--- a/test/markdown-link-check.test.js
+++ b/test/markdown-link-check.test.js
@@ -377,7 +377,14 @@ describe('markdown-link-check', function () {
                 { link: '#foo', statusCode: 200, err: null, status: 'alive' },
                 { link: '#bar', statusCode: 200, err: null, status: 'alive' },
                 { link: '#potato', statusCode: 404, err: null, status: 'dead' },
-                { link: '#tomato', statusCode: 404, err: null, status: 'dead' },
+                { link: '#tomato_id', statusCode: 200, err: null, status: 'alive' },
+                { link: '#tomato_name', statusCode: 200, err: null, status: 'alive' },
+                { link: '#tomato_code', statusCode: 404, err: null, status: 'dead' },
+                { link: '#tomato_escaped_backticks', statusCode: 200, err: null, status: 'alive' },
+                { link: '#tomato_comment', statusCode: 404, err: null, status: 'dead' },
+                { link: '#onion', statusCode: 200, err: null, status: 'alive' },
+                { link: '#onion_outer', statusCode: 200, err: null, status: 'alive' },
+                { link: '#onion_inner', statusCode: 200, err: null, status: 'alive' },
                 { link: '#header-with-special-char-', statusCode: 404, err: null, status: 'dead' },
             ]);
             done();

From 181d704e94bde2ad43c4d9c90bbe5a6383536c8e Mon Sep 17 00:00:00 2001
From: dklimpel <5740567+dklimpel@users.noreply.github.com>
Date: Mon, 3 Jun 2024 16:16:22 +0200
Subject: [PATCH 2/2] add support and tests for single quotes

---
 index.js                         | 4 ++--
 test/hash-links.md               | 8 +++++++-
 test/markdown-link-check.test.js | 2 ++
 3 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/index.js b/index.js
index 98c76ab..aad6ea3 100644
--- a/index.js
+++ b/index.js
@@ -56,8 +56,8 @@ function extractHtmlSections(markdown) {
         // remove single line code (if not escaped with "\")
         .replace(/(?<!\\)`[\S\s]+?(?<!\\)`/gm, '');
 
-    const regexAllId = /<(?<tag>[^\s]+).*?id="(?<id>[^"]*?)".*?>/gmi;
-    const regexAName = /<a.*?name="(?<name>[^"]*?)".*?>/gmi;
+    const regexAllId = /<(?<tag>[^\s]+).*?id=["'](?<id>[^"']*?)["'].*?>/gmi;
+    const regexAName = /<a.*?name=["'](?<name>[^"']*?)["'].*?>/gmi;
 
     const sections = []
         .concat(Array.from(markdown.matchAll(regexAllId), (match) => match.groups.id))
diff --git a/test/hash-links.md b/test/hash-links.md
index e9e53de..33e3d70 100644
--- a/test/hash-links.md
+++ b/test/hash-links.md
@@ -2,9 +2,11 @@
 
 <!-- markdownlint-disable MD033 -->
 <a id="tomato_id"></a>
-
 <a name="tomato_name"></a>
 
+<a id='tomato_id_single_quote'></a>
+<a name='tomato_name_single_quote'></a>
+
 <div id="onion"></div>
 <div id="onion_outer">
     <div id="onion_inner"></div>
@@ -39,6 +41,10 @@ There is an anchor named with `id` [Tomato](#tomato_id).
 
 There is an anchor named with `name` [Tomato](#tomato_name).
 
+There is an anchor named with `id` [Tomato in single quote](#tomato_id_single_quote).
+
+There is an anchor named with `name` [Tomato in single quote](#tomato_name_single_quote).
+
 There is an anchor in code [Tomato in code](#tomato_code).
 
 There is an anchor in escaped code [Tomato in escaped backticks](#tomato_escaped_backticks).
diff --git a/test/markdown-link-check.test.js b/test/markdown-link-check.test.js
index eb77212..2d043f5 100644
--- a/test/markdown-link-check.test.js
+++ b/test/markdown-link-check.test.js
@@ -379,6 +379,8 @@ describe('markdown-link-check', function () {
                 { link: '#potato', statusCode: 404, err: null, status: 'dead' },
                 { link: '#tomato_id', statusCode: 200, err: null, status: 'alive' },
                 { link: '#tomato_name', statusCode: 200, err: null, status: 'alive' },
+                { link: '#tomato_id_single_quote', statusCode: 200, err: null, status: 'alive' },
+                { link: '#tomato_name_single_quote', statusCode: 200, err: null, status: 'alive' },
                 { link: '#tomato_code', statusCode: 404, err: null, status: 'dead' },
                 { link: '#tomato_escaped_backticks', statusCode: 200, err: null, status: 'alive' },
                 { link: '#tomato_comment', statusCode: 404, err: null, status: 'dead' },