Fix spacing on MD upload

lib/mdx.js

@@ -135,58 +135,133 @@ function generateHeadingId(text) {
 
 /**
  * Convert markdown to HTML with enhanced formatting
+ * - Preserves extra blank lines between paragraphs as visible gaps
+ *   (each extra blank line becomes <p class="md-gap">&nbsp;</p>)
  * @param {string} markdown - Raw markdown content
  * @returns {string} HTML content
  */
 function markdownToHtml(markdown) {
   if (!markdown) return "";
 
+  // Normalize line endings
+  const GAP_TOKEN = "<GAP/>";
+  const src = String(markdown).replace(/\r\n?/g, "\n");
+
+  // For 3+ consecutive newlines, keep 2 for the paragraph break and
+  // emit a counted gap token for additional blank lines to preserve spacing.
+  const withGaps = src.replace(/\n{3,}/g, (m) => {
+    const extra = m.length - 2;
+    return `\n\n<GAP:${extra}/>`;
+  });
+
   return (
-    markdown
+    withGaps
       // Headers with IDs
-      .replace(/^### (.*$)/gim, (match, text) => {
-        const id = generateHeadingId(text);
-        return `<h3 id="${id}">${text}</h3>`;
-      })
-      .replace(/^## (.*$)/gim, (match, text) => {
-        const id = generateHeadingId(text);
-        return `<h2 id="${id}">${text}</h2>`;
-      })
-      .replace(/^# (.*$)/gim, (match, text) => {
-        const id = generateHeadingId(text);
-        return `<h1 id="${id}">${text}</h1>`;
-      })
-      // Bold and italic
-      .replace(/\*\*(.*?)\*\*/g, "<strong>$1</strong>")
-      .replace(/\*(.*?)\*/g, "<em>$1</em>")
-      // Code blocks
+      .replace(
+        /^###### (.*$)/gim,
+        (m, t) => `<h6 id="${generateHeadingId(t)}">${t}</h6>`
+      )
+      .replace(
+        /^##### (.*$)/gim,
+        (m, t) => `<h5 id="${generateHeadingId(t)}">${t}</h5>`
+      )
+      .replace(
+        /^#### (.*$)/gim,
+        (m, t) => `<h4 id="${generateHeadingId(t)}">${t}</h4>`
+      )
+      .replace(
+        /^### (.*$)/gim,
+        (m, t) => `<h3 id="${generateHeadingId(t)}">${t}</h3>`
+      )
+      .replace(
+        /^## (.*$)/gim,
+        (m, t) => `<h2 id="${generateHeadingId(t)}">${t}</h2>`
+      )
+      .replace(
+        /^# (.*$)/gim,
+        (m, t) => `<h1 id="${generateHeadingId(t)}">${t}</h1>`
+      )
+
+      // Code fences (block) and inline code
       .replace(
         /```(\w+)?\n([\s\S]*?)\n```/g,
-        '<pre><code class="language-$1">$2</code></pre>'
+        (m, lang = "", code) =>
+          `<pre><code class="language-${lang}">${code}</code></pre>`
       )
       .replace(/`([^`]+)`/g, "<code>$1</code>")
-      // Links
-      .replace(/\[([^\]]+)\]\(([^)]+)\)/g, '<a href="$2">$1</a>')
-      // Lists
-      .replace(/^\* (.*$)/gim, "<li>$1</li>")
-      .replace(/^- (.*$)/gim, "<li>$1</li>")
-      .replace(/(<li>.*<\/li>)/gim, "<ul>$1</ul>")
+
+      // Bold and italic (strong before em to avoid overlap issues)
+      .replace(/\*\*(.+?)\*\*/g, "<strong>$1</strong>")
+      .replace(/\*(.+?)\*/g, "<em>$1</em>")
+
+      // Links and images
+      .replace(
+        /!\[([^\]]*)\]\(([^)\s]+)(?:\s+"([^"]+)")?\)/g,
+        (m, alt, src, title = "") =>
+          `<img alt="${alt}" src="${src}"${title ? ` title="${title}"` : ""}>`
+      )
+      .replace(
+        /\[([^\]]+)\]\(([^)\s]+)(?:\s+"([^"]+)")?\)/g,
+        (m, text, href, title = "") =>
+          `<a href="${href}"${title ? ` title="${title}"` : ""}>${text}</a>`
+      )
+
       // Blockquotes
-      .replace(/^> (.*$)/gim, "<blockquote><p>$1</p></blockquote>")
-      // Horizontal rules
-      .replace(/^---$/gm, "<hr>")
-      .replace(/^\*\*\*$/gm, "<hr>")
-      // Paragraphs
-      .replace(/\n\n/g, "</p><p>")
-      .replace(/^(?!<[h|u|li|blockquote|hr|pre])(.*$)/gim, "<p>$1</p>")
-      // Clean up empty paragraphs and fix list wrapping
-      .replace(/<p><\/p>/g, "")
-      .replace(/<p>(.*?)<\/p>/g, (match, content) => {
-        return content.trim() ? match : "";
+      .replace(/^(>\s?.+)(\n(>\s?.+))*$/gim, (m) => {
+        const inner = m.replace(/^>\s?/gm, "");
+        return `<blockquote><p>${inner.replace(
+          /\n{2,}/g,
+          "</p><p>"
+        )}</p></blockquote>`;
       })
-      .replace(/<\/ul>\s*<ul>/g, "") // Merge consecutive ul elements
-      .replace(/<ul>\s*<\/ul>/g, "")
-  ); // Remove empty ul elements
+
+      // Lists (ul/ol)
+      .replace(/^(\s*[-*]\s.+(?:\n\s*[-*]\s.+)*)/gim, (m) => {
+        const items = m
+          .trim()
+          .split(/\n/)
+          .map((l) => l.replace(/^\s*[-*]\s+/, ""))
+          .map((t) => `<li>${t}</li>`)
+          .join("");
+        return `<ul>${items}</ul>`;
+      })
+      .replace(/^(\s*\d+\.\s.+(?:\n\s*\d+\.\s.+)*)/gim, (m) => {
+        const items = m
+          .trim()
+          .split(/\n/)
+          .map((l) => l.replace(/^\s*\d+\.\s+/, ""))
+          .map((t) => `<li>${t}</li>`)
+          .join("");
+        return `<ol>${items}</ol>`;
+      })
+
+      // Horizontal rules
+      .replace(/^\s*(?:-{3,}|\*{3,})\s*$/gm, "<hr>")
+
+      // Paragraphs:
+      // 1) Convert double newlines to paragraph boundaries
+      .replace(/\n\n/g, "</p><p>")
+      // 2) Convert single line breaks to <br> tags within paragraphs
+      .replace(/(?<!\n)\n(?!\n)/g, "<br>")
+      // 3) Wrap remaining bare lines that are not already block-level elements.
+      //    (Also skip our GAP_TOKEN so we can turn it into gap paragraphs later.)
+      .replace(
+        /^(?!\s*<(h[1-6]|ul|ol|li|blockquote|hr|pre|code|table|img)\b)(?!\s*<\/)(?!\s*<GAP\/>)(.+)$/gim,
+        "<p>$2</p>"
+      )
+
+      // Clean up truly empty paragraphs but keep &nbsp; gap paragraphs
+      .replace(/<p>\s*<\/p>/g, "")
+
+      // Turn counted GAP tokens into explicit, styleable gap elements
+      .replace(
+        /<GAP:(\d+)\/>/g,
+        (m, n) =>
+          `<div class="md-gap" style="--gap:${Number(
+            n
+          )}" aria-hidden="true"></div>`
+      )
+  );
 }
 
 /**