Xinrui/deepseekmodel

.github/workflows/deploy.yml

@@ -1,29 +1,35 @@
-name: Deploy AI Archives to EC2
+# .github/workflows/deploy.yml
+name: Deploy to EC2
 
 on:
   push:
-    branches: [main]
+    branches:
+      - xinrui/deepseekmodel
 
 jobs:
   deploy:
     runs-on: ubuntu-latest
     env:
-      SSH_KEY: ${{ secrets.SSH_PRIVATE_KEY }}
+      SSH_KEY: ${{ secrets.EC2_KEY_BASE64 }}
       HOST: ${{ secrets.SERVER_HOST }}
       USER: ${{ secrets.SERVER_USER }}
 
     steps:
-      - uses: actions/checkout@v3
+      - name: Checkout code
+        uses: actions/checkout@v3
 
+      - name: Decode and save EC2 SSH key
+        run: |
+          echo "$SSH_KEY" | base64 -d > deploy_key
+          chmod 600 deploy_key
+
       - name: Build
         run: |
           npm ci
           npm run build
 
       - name: Sync build to EC2
         run: |
-          echo "$SSH_KEY" > deploy_key
-          chmod 600 deploy_key
           rsync -avz --delete \
             --exclude 'node_modules' --exclude '.git' --exclude '.env' \
             -e "ssh -i deploy_key -o StrictHostKeyChecking=no" \
@@ -32,4 +38,4 @@ jobs:
       - name: Restart AI Archives service
         run: |
           ssh -i deploy_key -o StrictHostKeyChecking=no $USER@$HOST \
-            "sudo systemctl restart aiarchives"
+            "sudo systemctl restart aiarchives"

app/api/conversation/route.ts

@@ -72,9 +72,22 @@ export async function POST(req: NextRequest) {
       return NextResponse.json({ error: '`htmlDoc` must be a file field' }, { status: 400 });
     }
 
-    // Parse the conversation from HTML
+    // Parse the conversation from HTML if not skipScraping, otherwise skip parsing
+    const skipScraping = formData.has('skipScraping');
+    let conversation;
     const html = await file.text();
-    const conversation = await parseHtmlToConversation(html, model);
+    if (!skipScraping) {
+      conversation = await parseHtmlToConversation(html, model);
+    } else {
+      conversation = {
+        model: model,
+        content: html,
+        scrapedAt: new Date().toISOString(),
+        sourceHtmlBytes: html.length,
+      };
+    }
+
+    console.log('conversation =>', conversation.model, conversation.content);
 
     // Generate a unique ID for the conversation
     const conversationId = randomUUID();

app/page.tsx

@@ -27,7 +27,7 @@ interface ConversationCardData {
  */
 async function fetchConversations(): Promise<ConversationRecord[]> {
   try {
-    const baseUrl = process.env.NEXT_PUBLIC_BASE_URL || 'http://localhost:3000';
+    const baseUrl = process.env.NEXT_PUBLIC_BASE_URL || 'http://localhost:3000'; 
     const response = await fetch(`${baseUrl}/api/conversation?limit=50`, {
       method: 'GET',
       headers: {

extension/manifest.json

@@ -13,7 +13,8 @@
     "https://claude.ai/*",
     "https://gemini.google.com/*",
     "https://chatgpt.com/*",
-    "https://chat.openai.com/*"
+    "https://chat.openai.com/*",
+    "https://chat.deepseek.com/*"
   ],
   "content_security_policy": {
     "extension_pages": "script-src 'self'; object-src 'self'"
@@ -26,7 +27,8 @@
         "https://bard.google.com/*",
         "https://gemini.google.com/*",
         "https://claude.ai/*",
-        "https://www.meta.ai/*"
+        "https://www.meta.ai/*",
+        "https://chat.deepseek.com/*"
       ],
       "js": ["config.js", "index.js"],
       "run_at": "document_end",

extension/popup.js

@@ -41,7 +41,7 @@ chrome.tabs.query({ active: true, currentWindow: true, url: 'https://bard.google
 });
 chrome.tabs.query({ active: true, currentWindow: true, url: 'https://www.meta.ai/*' }, (tabs) => {
   if (tabs?.length) {
-    chrome.tabs.sendMessage(tabs[0].id, { action: 'model', model: 'Meta' }, function (_) {
+    chrome.tabs.sendMessage(tabs[0].id, { action: 'model', model: 'meta' }, function (_) {
       console.log('is Meta');
     });
   }
@@ -53,5 +53,12 @@ chrome.tabs.query({ active: true, currentWindow: true, url: 'https://claude.ai/*
     });
   }
 });
+chrome.tabs.query({ active: true, currentWindow: true, url: 'https://chat.deepseek.com/*' }, (tabs) => {
+  if (tabs?.length) {
+    chrome.tabs.sendMessage(tabs[0].id, { action: 'model', model: 'deepSeek' }, function (_) {
+      console.log('is DeepSeek');
+    });
+  }
+});
 
 window.onload = initApp;

lib/parsers/deepseek.ts

@@ -1,14 +1,50 @@
 import type { Conversation } from '@/types/conversation';
+import { JSDOM } from 'jsdom';
+import fs from 'fs';
+import path from 'path';
 
 /**
  * Extracts a DeepSeek share page into a structured Conversation.
  * @param html - Raw HTML content from the DeepSeek share page
  * @returns Promise resolving to a structured Conversation object
  */
 export async function parseDeepSeek(html: string): Promise<Conversation> {
+  const dom = new JSDOM(html);
+  const document = dom.window.document;
+
+  // 1. Inject <style> from deepseek.css
+  const cssPath = path.resolve(process.cwd(), 'lib/parsers/assets/deepseek.css');
+  const css = fs.readFileSync(cssPath, 'utf-8');
+
+  const styleTag = document.createElement('style');
+  styleTag.textContent = css;
+  document.head.appendChild(styleTag);
+
+  // 2. Clean up icons and extra UI elements
+  const iconSelectors = ['.ds-icon-button'];
+  iconSelectors.forEach(selector => {
+    document.querySelectorAll(selector).forEach(el => el.remove());
+  });
+
+  // 3. Grab main conversation container
+  const chatContainer = document.querySelector('div.dad65929');
+  if (!chatContainer) {
+    throw new Error('Conversation container not found');
+  }
+
+  // 4. Prepare the final HTML output
+  const htmlContent = `
+    <html>
+      <head>${document.head.innerHTML}</head>
+      <body class="light-theme theme-host">
+        ${chatContainer.outerHTML}
+      </body>
+    </html>
+  `;
+
   return {
-    model: 'DeepSeek',
-    content: html,
+    model: 'deepSeek',
+    content: htmlContent,
     scrapedAt: new Date().toISOString(),
     sourceHtmlBytes: html.length,
   };