From 1e5c56c0d652ab4e663d0af7763a3d051f17893e Mon Sep 17 00:00:00 2001
From: Cynthia Foxwell <gamers@riseup.net>
Date: Fri, 28 Mar 2025 09:32:48 -0600
Subject: [PATCH] html to markdown fixes

---
 src/util/html.js | 32 ++++++++++++++++----------------
 1 file changed, 16 insertions(+), 16 deletions(-)

diff --git a/src/util/html.js b/src/util/html.js
index 85bb5ec..092a0f0 100644
--- a/src/util/html.js
+++ b/src/util/html.js
@@ -32,14 +32,14 @@ function parseHtmlEntities(str) {
 
 function htmlToMarkdown(str, images = true, embed = true) {
   str = str.replaceAll("\\", "\\\\");
-  str = str.replace(/<style(\s+[^>]+)?>(.|\n)*?<\/style>/gi, "");
-  str = str.replace(/<a (\s+[^>]+)?href="([^"]+?)"(\s+[^>]+)?>(.+?)<\/a>/gi, (_, __, url, ___, text) => {
+  str = str.replace(/<style(\s*[^>]+)?>(.|\n)*?<\/style>/gi, "");
+  str = str.replace(/<a (\s*[^>]+)?href="([^"]+?)"(\s*[^>]+)?>(.+?)<\/a>/gi, (_, __, url, ___, text) => {
     url = url.replace(/^\/\//, "https://").replace("\\#", "#");
     return url == text ? url : `[${text}](${embed ? "" : "&lt;"}${url}${embed ? "" : "&gt;"})`;
   });
   if (images)
     str = str.replace(
-      /<img (\s+[^>]+)?src="([^"]+?)"(\s+[^>]+)?(alt|title)="([^"]+?)"(\s+[^>]+)?\/>/gi,
+      /<img (\s*[^>]+)?src="([^"]+?)"(\s*[^>]+)?(alt|title)="([^"]+?)"(\s*[^>]+)?\/>/gi,
       `[$5](${embed ? "" : "&lt;"}$2${embed ? "" : "&gt;"})`
     );
   str = str.replace(/<\/?\s*br\s*\/?>/gi, "\n");
@@ -49,7 +49,7 @@ function htmlToMarkdown(str, images = true, embed = true) {
   );
   str = str.replace(/<\/?p>/gi, "\n");
   str = str.replace(/<dd>((.|\n)*?)<\/dd>/gi, (_, inner) => "\u3000\u3000" + inner.split("\n").join("\n\u3000\u3000"));
-  str = str.replace(/<ol(\s+[^>]+)?>((.|\n)*?)<\/ol>/gi, (_, __, inner) => {
+  str = str.replace(/<ol(\s*[^>]+)?>((.|\n)*?)<\/ol>/gi, (_, __, inner) => {
     let index = 0;
     return inner
       .replace(/<li>/gi, () => {
@@ -59,7 +59,7 @@ function htmlToMarkdown(str, images = true, embed = true) {
       .replace(/<\/li>/gi, "\n")
       .replaceAll("\n\n", "\n");
   });
-  str = str.replace(/<ul(\s+[^>]+)?>((.|\n)*?)<\/ul>/gi, (_, __, inner) => {
+  str = str.replace(/<ul(\s*[^>]+)?>((.|\n)*?)<\/ul>/gi, (_, __, inner) => {
     let index = 0;
     return inner
       .replace(/<li>/gi, () => {
@@ -69,17 +69,17 @@ function htmlToMarkdown(str, images = true, embed = true) {
       .replace(/<\/li>/gi, "\n")
       .replaceAll("\n\n", "\n");
   });
-  str = str.replace(/<\/?code(\s+[^>]+)?>/gi, "`");
-  str = str.replace(/<\/?em(\s+[^>]+)?>/gi, "_");
-  str = str.replace(/<\/?i(\s+[^>]+)?>/gi, "_");
-  str = str.replace(/<\/?b(\s+[^>]+)?>/gi, "**");
-  str = str.replace(/<\/?u(\s+[^>]+)?>/gi, "__");
-  str = str.replace(/<\/?s(\s+[^>]+)?>/gi, "~~");
-  str = str.replace(/<h1(\s+[^>]+)?>/gi, "# ");
-  str = str.replace(/<h2(\s+[^>]+)?>/gi, "## ");
-  str = str.replace(/<h3(\s+[^>]+)?>/gi, "### ");
-  str = str.replace(/<\/?h4(\s+[^>]+)?>/gi, "**");
-  str = str.replace(/<(math|noscript)(\s+[^>]+)?>((.|\n)*?)<\/(math|noscript)>/gi, "");
+  str = str.replace(/<\/?code(\s*[^>]+)?>/gi, "`");
+  str = str.replace(/<\/?em(\s*[^>]+)?>/gi, "_");
+  str = str.replace(/<\/?i(\s*[^>]+)?>/gi, "_");
+  str = str.replace(/<\/?b(\s*[^>]+)?>/gi, "**");
+  str = str.replace(/<\/?u(\s*[^>]+)?>/gi, "__");
+  str = str.replace(/<\/?s(\s*[^>]+)?>/gi, "~~");
+  str = str.replace(/<h1(\s*[^>]+)?>/gi, "# ");
+  str = str.replace(/<h2(\s*[^>]+)?>/gi, "## ");
+  str = str.replace(/<h3(\s*[^>]+)?>/gi, "### ");
+  str = str.replace(/<\/?h4(\s*[^>]+)?>/gi, "**");
+  str = str.replace(/<(math|noscript)(\s*[^>]+)?>((.|\n)*?)<\/(math|noscript)>/gi, "");
   str = str.replace(/<[^>]+?>/gi, "");
   str = parseHtmlEntities(str);
   // whyyyyyyyyyyyy