fix(zhihu/download): author extraction, zhida links, redirect detection

adapters/zhihu/download.yaml

@@ -1,5 +1,6 @@
 # Adapter logic derived from OpenCLI (https://github.com/jackwener/opencli)
 # Original author: jackwener | License: Apache-2.0
+# Override: fix author extraction (issue #40) + strip zhida links
 site: zhihu
 name: download
 description: 导出知乎文章为 Markdown 格式
@@ -17,21 +18,45 @@ args:
     default: ./zhihu-articles
     description: Output directory
 
-columns: [title, author, status, size]
+columns: [title, author, status, path, size]
 
 pipeline:
   - navigate:
       url: "${{ args.url }}"
-      settleMs: 3000
+      settleMs: 5000
 
   - evaluate: |
       (() => {
+        // Detect redirect to homepage (article not found / removed)
+        if (location.hostname === 'www.zhihu.com' && !location.pathname.startsWith('/p/') && !location.pathname.startsWith('/question/')) {
+          return { title: 'error', author: '', status: 'failed', size: 'Article not found (redirected to homepage)', filename: 'error.md', imageUrls: [], content: 'Article not found or removed', output: args.output || './zhihu-articles' };
+        }
+
         const title = document.querySelector('.Post-Title, h1.ContentItem-title, .ArticleTitle, .QuestionHeader-title')?.textContent?.trim() || 'untitled';
-        const author = document.querySelector('.AuthorInfo-name, .UserLink-link')?.textContent?.trim() || 'unknown';
+
+        // Fixed author extraction with fallbacks (issue #40)
+        const author =
+          document.querySelector('.AuthorInfo-name')?.textContent?.trim() ||
+          document.querySelector('.UserLink-link')?.textContent?.trim() ||
+          document.querySelector('meta[itemprop="author"]')?.getAttribute('content') ||
+          document.querySelector('meta[name="author"]')?.getAttribute('content') ||
+          (() => {
+            try {
+              const s = document.querySelector('script#js-initialData');
+              if (s) {
+                const data = JSON.parse(s.textContent);
+                const users = data?.initialState?.entities?.users;
+                if (users) return Object.values(users)[0]?.name || '';
+              }
+            } catch(e) {}
+            return '';
+          })() ||
+          'unknown';
+
         const timeEl = document.querySelector('.ContentItem-time, .Post-Time');
         const publishTime = timeEl?.textContent?.trim() || '';
         const contentEl = document.querySelector('.Post-RichTextContainer, .RichContent-inner, .RichText');
-        if (!contentEl) return [{ title, author, status: 'failed', size: 'No content found' }];
+        if (!contentEl) return { title, author, status: 'failed', size: 'No content found', filename: 'error.md', imageUrls: [], content: 'Failed to extract content', output: args.output || './zhihu-articles' };
 
         function htmlToMd(el) {
           let md = '';
@@ -46,7 +71,15 @@ pipeline:
             else if (tag === 'br') md += '\n';
             else if (tag === 'strong' || tag === 'b') md += '**' + node.textContent + '**';
             else if (tag === 'em' || tag === 'i') md += '*' + node.textContent + '*';
-            else if (tag === 'a') md += '[' + node.textContent + '](' + (node.href || '') + ')';
+            else if (tag === 'a') {
+              const href = node.href || '';
+              // Strip zhida.zhihu.com links, keep display text only
+              if (href.includes('zhida.zhihu.com')) {
+                md += node.textContent;
+              } else {
+                md += '[' + node.textContent + '](' + href + ')';
+              }
+            }
             else if (tag === 'img') {
               const src = node.getAttribute('data-original') || node.getAttribute('data-actualsrc') || node.getAttribute('data-src') || node.getAttribute('data-lazy-src') || node.src || '';
               if (src && !src.includes('data:image')) {
@@ -56,7 +89,6 @@ pipeline:
               }
             }
             else if (tag === 'noscript') {
-              // zhihu wraps real img inside noscript for lazy loading
               const inner = node.textContent || node.innerHTML || '';
               const srcMatch = inner.match(/src=["']([^"']+)["']/);
               if (srcMatch) {
@@ -83,7 +115,6 @@ pipeline:
 
         const markdown = htmlToMd(contentEl);
 
-        // Extract image URLs from the generated markdown to ensure exact match
         const imageUrls = [];
         const imgRe = /!\[([^\]]*)\]\(([^)]+)\)/g;
         let m;