Auto-commited changes

src/content.js

@@ -5,7 +5,7 @@ window.__memosClipperLoaded = true;
 
 (function () {
   // ── Turndown-lite: a minimal but solid HTML→Markdown converter ──────────────
-  function htmlToMarkdown(element, isSelection = false) {
+  function htmlToMarkdown(element, isSelection = false, stripLinks = false) {
     const clone = element.cloneNode(true);
 
     // Remove unwanted elements — comprehensive list covering real-world sites
@@ -54,15 +54,40 @@ window.__memosClipperLoaded = true;
       // Also remove elements that are visually hidden via inline style
       clone.querySelectorAll('[style*="display:none"],[style*="display: none"],[style*="visibility:hidden"]')
         .forEach((el) => el.remove());
+
+      // Remove link-dense blocks (navigation menus, ad link lists, etc.)
+      // Collect candidates first to avoid mid-iteration detached-node issues.
+      // Only target outer chrome elements (nav, aside, header, footer, div, section)
+      // not content containers like article/main, to avoid stripping TOCs in prose.
+      const linkDenseCandidates = Array.from(
+        clone.querySelectorAll('nav, aside, header, footer, div, section')
+      ).filter((el) => {
+        // Skip if inside the primary content container
+        if (el.closest('article, main, [role="main"]')) return false;
+        const totalText = (el.textContent || '').trim().length;
+        if (totalText < 30) return false; // too short to judge
+        const linkText = Array.from(el.querySelectorAll('a'))
+          .reduce((sum, a) => sum + (a.textContent || '').trim().length, 0);
+        if (linkText / totalText <= 0.65) return false;
+        // Require that the element has little direct (non-link) text of its own
+        const directText = Array.from(el.childNodes)
+          .filter((n) => n.nodeType === Node.TEXT_NODE)
+          .reduce((sum, n) => sum + n.textContent.trim().length, 0);
+        return directText < totalText * 0.25;
+      });
+      // Remove outermost candidates only (skip those already inside a removed ancestor)
+      linkDenseCandidates.forEach((el) => {
+        if (el.isConnected) el.remove();
+      });
     } else {
       // In selection mode, we still want to remove script/style tags if any
       clone.querySelectorAll('script, style, noscript, template').forEach((el) => el.remove());
     }
 
-    return nodeToMd(clone).replace(/\n{3,}/g, "\n\n").trim();
+    return nodeToMd(clone, { listDepth: 0, ordered: false, index: 0 }, stripLinks).replace(/\n{3,}/g, "\n\n").trim();
   }
 
-  function nodeToMd(node, ctx = { listDepth: 0, ordered: false, index: 0 }) {
+  function nodeToMd(node, ctx = { listDepth: 0, ordered: false, index: 0 }, stripLinks = false) {
     if (node.nodeType === Node.TEXT_NODE) {
       return node.textContent.replace(/\s+/g, " ");
     }
@@ -71,7 +96,7 @@ window.__memosClipperLoaded = true;
     const tag = node.tagName.toLowerCase();
     const children = () =>
       Array.from(node.childNodes)
-        .map((c) => nodeToMd(c, ctx))
+        .map((c) => nodeToMd(c, ctx, stripLinks))
         .join("");
 
     switch (tag) {
@@ -93,7 +118,8 @@ window.__memosClipperLoaded = true;
       case "del": return `~~${children()}~~`;
       case "code": {
         const text = node.textContent;
-        return text.includes("`") ? `\`\`${text}\`\`` : `\`${text}\``;
+        if (text.includes("`")) return `\`\` ${text} \`\``;
+        return `\`${text}\``;
       }
       case "pre": {
         const codeEl = node.querySelector("code");
@@ -110,8 +136,9 @@ window.__memosClipperLoaded = true;
         .join("\n")}\n\n`;
 
       case "a": {
-        const href = node.getAttribute("href") || "";
         const text = children().trim();
+        if (stripLinks) return text; // just the anchor text, no URL
+        const href = node.getAttribute("href") || "";
         if (!text) return href;
         try {
           const abs = new URL(href, location.href).href;
@@ -134,24 +161,24 @@ window.__memosClipperLoaded = true;
 
       case "ul": {
         const lines = Array.from(node.children)
-          .map((li) => `${"  ".repeat(ctx.listDepth)}- ${nodeToMd(li, { ...ctx, listDepth: ctx.listDepth + 1 }).trim()}`)
+          .map((li) => `${"  ".repeat(ctx.listDepth)}- ${nodeToMd(li, { ...ctx, listDepth: ctx.listDepth + 1 }, stripLinks).trim()}`)
           .join("\n");
         return `\n\n${lines}\n\n`;
       }
       case "ol": {
         const lines = Array.from(node.children)
-          .map((li, i) => `${"  ".repeat(ctx.listDepth)}${i + 1}. ${nodeToMd(li, { ...ctx, listDepth: ctx.listDepth + 1 }).trim()}`)
+          .map((li, i) => `${"  ".repeat(ctx.listDepth)}${i + 1}. ${nodeToMd(li, { ...ctx, listDepth: ctx.listDepth + 1 }, stripLinks).trim()}`)
           .join("\n");
         return `\n\n${lines}\n\n`;
       }
       case "li": return children();
 
-      case "table": return convertTable(node);
+      case "table": return convertTable(node, stripLinks);
 
       case "figure": {
         const img = node.querySelector("img");
         const caption = node.querySelector("figcaption");
-        let md = img ? nodeToMd(img, ctx) : children();
+        let md = img ? nodeToMd(img, ctx, stripLinks) : children();
         if (caption) md += `\n*${caption.textContent.trim()}*`;
         return `\n\n${md}\n\n`;
       }
@@ -178,13 +205,16 @@ window.__memosClipperLoaded = true;
     }
   }
 
-  function convertTable(table) {
+  function convertTable(table, stripLinks = false) {
     const rows = Array.from(table.querySelectorAll("tr"));
     if (!rows.length) return "";
     const toRow = (tr) =>
       "| " +
       Array.from(tr.querySelectorAll("th,td"))
-        .map((c) => c.textContent.trim().replace(/\|/g, "\\|"))
+        .map((c) => {
+          const text = stripLinks ? c.textContent.trim() : nodeToMd(c).trim();
+          return text.replace(/\|/g, "\\|");
+        })
         .join(" | ") +
       " |";
     const header = toRow(rows[0]);
@@ -234,6 +264,7 @@ window.__memosClipperLoaded = true;
         let markdown = "";
         let images = [];
         let title = document.title || location.href;
+        const stripLinks = !!msg.stripLinks;
 
         if (msg.mode === "selection") {
           const sel = window.getSelection();
@@ -241,7 +272,7 @@ window.__memosClipperLoaded = true;
             const frag = sel.getRangeAt(0).cloneContents();
             const div = document.createElement("div");
             div.appendChild(frag);
-            markdown = htmlToMarkdown(div, true);
+            markdown = htmlToMarkdown(div, true, stripLinks);
             images = extractImages(div);
           } else {
             markdown = "";
@@ -253,7 +284,7 @@ window.__memosClipperLoaded = true;
             document.querySelector("main") ||
             document.querySelector('[role="main"]') ||
             document.body;
-          markdown = htmlToMarkdown(root);
+          markdown = htmlToMarkdown(root, false, stripLinks);
           images = extractImages(root);
         }