From c7f6383e34c30b63e0e0b86f320f42f2c9a0bdb7 Mon Sep 17 00:00:00 2001
From: "(quasar) nebula" <qznebula@protonmail.com>
Date: Wed, 3 Sep 2025 16:55:19 -0300
Subject: content, replacer: match inline links, auto-provide custom label

Changes in matchMarkdownLinks here are refactoring only, not new
behavior.
---
 src/common-util/wiki-data.js | 48 ++++++++++++++++++++++++++++++++------------
 1 file changed, 35 insertions(+), 13 deletions(-)

(limited to 'src/common-util')

diff --git a/src/common-util/wiki-data.js b/src/common-util/wiki-data.js
index cb024022..e0f41ee1 100644
--- a/src/common-util/wiki-data.js
+++ b/src/common-util/wiki-data.js
@@ -535,24 +535,46 @@ export function combineWikiDataArrays(arrays) {
 export function* matchMarkdownLinks(markdownSource, {marked}) {
   const plausibleLinkRegexp = /\[(?=.*?\))/g;
 
+  // Pedantic rules use more particular parentheses detection in link
+  // destinations - they allow one level of balanced parentheses, and
+  // otherwise, parentheses must be escaped. This allows for entire links
+  // to be wrapped in parentheses, e.g below:
+  //
+  //   This is so cool. ([You know??](https://example.com))
+  //
+  const definiteLinkRegexp = marked.Lexer.rules.inline.pedantic.link;
+
   let plausibleMatch = null;
   while (plausibleMatch = plausibleLinkRegexp.exec(markdownSource)) {
-    // Pedantic rules use more particular parentheses detection in link
-    // destinations - they allow one level of balanced parentheses, and
-    // otherwise, parentheses must be escaped. This allows for entire links
-    // to be wrapped in parentheses, e.g below:
-    //
-    //   This is so cool. ([You know??](https://example.com))
-    //
     const definiteMatch =
-      marked.Lexer.rules.inline.pedantic.link
-        .exec(markdownSource.slice(plausibleMatch.index));
+      definiteLinkRegexp.exec(markdownSource.slice(plausibleMatch.index));
 
-    if (definiteMatch) {
-      const [{length}, label, href] = definiteMatch;
-      const index = plausibleMatch.index + definiteMatch.index;
+    if (!definiteMatch) {
+      continue;
+    }
+
+    const [{length}, label, href] = definiteMatch;
+    const index = plausibleMatch.index + definiteMatch.index;
 
-      yield {label, href, index, length};
+    yield {label, href, index, length};
+  }
+}
+
+export function* matchInlineLinks(source) {
+  const plausibleLinkRegexp = /\b[a-z]*:\/\/[^ ]*?(?=(?:[,.!?]*)(?:\s|$))/gm;
+
+  let plausibleMatch = null;
+  while (plausibleMatch = plausibleLinkRegexp.exec(source)) {
+    const [href] = plausibleMatch;
+    const {index} = plausibleMatch;
+    const [{length}] = plausibleMatch;
+
+    try {
+      new URL(href);
+    } catch {
+      continue;
     }
+
+    yield {href, length, index};
   }
 }
-- 
cgit 1.3.0-6-gf8a5