feat(#53): use JSDOM to parse internal anchor tags

Author: carbontwelve

src/html-link-parser.js

@@ -1,3 +1,4 @@
+import {JSDOM} from 'jsdom';
 export default class HTMLLinkParser {
 
   /**
@@ -24,7 +25,7 @@ export default class HTMLLinkParser {
    */
   parseSingle(link, pageDirectory) {
     const meta = {
-      href: link.slice(6, -1)
+      href: link
         .replace(/.(md|markdown)\s?$/i, "")
         .replace("\\", "")
         .trim()
@@ -61,8 +62,19 @@ export default class HTMLLinkParser {
    * @return {Array<import('@photogabble/eleventy-plugin-interlinker').LinkMeta>}
    */
   find(document, pageDirectory) {
+    const dom = new JSDOM(document);
+    const anchors = dom.window.document.getElementsByTagName('a');
+    const toParse = [];
+
+    for (const anchor of anchors) {
+      // Ignore any anchor tags within either code or pre tags
+      if (anchor.closest('code,pre')) continue;
+      // Ignore any links that don't begin with / denoting internal links
+      if (anchor.href.startsWith('/')) toParse.push(anchor.href);
+    }
+
     return this.parseMultiple(
-      (document.match(this.internalLinkRegex) || []),
+      toParse,
       pageDirectory
     )
   }

tests/html-internal-link-parser.test.js

@@ -2,8 +2,12 @@ import HTMLLinkParser from '../src/html-link-parser.js'
 import DeadLinks from '../src/dead-links.js';
 import {pageLookup} from '../src/find-page.js';
 import test from 'ava';
+import fs from "node:fs";
+import path from "node:path";
+import {fileURLToPath} from "node:url";
 
 const pageDirectory = pageLookup([]);
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
 
 test('html link parser grabs multiple href, ignoring external links', t => {
     const parser = new HTMLLinkParser(new DeadLinks());
@@ -25,8 +29,8 @@ test('html link parser grabs multiple href, ignoring external links', t => {
 });
 
 test('html link parser ignores href within code blocks', t => {
-  const parser = new HTMLLinkParser(new DeadLinks());
-  const links = parser.find('<code><a href="/home">this is a link home</a></code>', pageDirectory);
+  t.is(0, ((new HTMLLinkParser(new DeadLinks())).find('<code><a href="/home">this is a link home</a></code>', pageDirectory)).length);
 
-  t.is(0, links.length);
+  const html = fs.readFileSync(__dirname + '/fixtures/within-code.html', {encoding:'utf8', flag:'r'});
+  t.is(1, ((new HTMLLinkParser(new DeadLinks())).find(html, pageDirectory)).length);
 });