Upload server.js

server.js

@@ -0,0 +1,372 @@
+const express = require('express');
+const cors = require('cors');
+const axios = require('axios');
+const fs = require('fs').promises;
+const path = require('path');
+require('dotenv').config();
+
+const app = express();
+
+// Security middleware
+app.use(cors({
+  origin: process.env.ALLOWED_ORIGINS?.split(',') || ['http://localhost', 'http://127.0.0.1', 'null'],
+  credentials: true
+}));
+
+app.use(express.json({ limit: '50mb' }));
+app.use(express.static('public'));
+
+const PORT = process.env.PORT || 5000;
+
+// Tera V2 Configuration
+const AI_CONFIG = {
+  url: process.env.AI_MODEL_URL || 'http://localhost:11434/api/generate',
+  model: process.env.AI_MODEL_NAME || 'qwen3.5:0.8b',
+  displayName: process.env.DISPLAY_NAME || 'Tera V2'
+};
+
+// Tera V2 System Prompt - Customizes the AI behavior
+const TERAV2_PROMPT = `You are Tera V2, an advanced AI document intelligence assistant created by Tera Labs.
+
+Your capabilities:
+- Analyze complex documents, contracts, and research papers
+- Extract key insights and summarize content
+- Answer questions based on provided documents
+- Generate podcast scripts and audio content
+- Support multiple document formats (PDF, TXT, MD, images)
+
+Response Style:
+- Professional yet approachable
+- Clear and concise
+- Always cite specific parts of documents when answering
+- Format responses for readability (bullet points, sections)
+- If information isn't in the documents, say so clearly
+
+When analyzing documents:
+1. Identify the document type and purpose
+2. Extract key information and data points
+3. Highlight important clauses or findings
+4. Flag any potential issues or concerns
+5. Provide actionable recommendations
+
+Remember: You are Tera V2. You are helpful, accurate, and efficient.`;
+
+// Dataset collection for future training
+const DATASET_PATH = path.join(__dirname, 'dataset');
+const LOGS_PATH = path.join(__dirname, 'logs');
+
+// Ensure directories exist
+async function ensureDirectories() {
+  try {
+    await fs.mkdir(DATASET_PATH, { recursive: true });
+    await fs.mkdir(LOGS_PATH, { recursive: true });
+  } catch (e) {}
+}
+ensureDirectories();
+
+// Logging function
+async function logConversation(data) {
+  try {
+    const date = new Date().toISOString().split('T')[0];
+    const logFile = path.join(LOGS_PATH, `conversations-${date}.jsonl`);
+    await fs.appendFile(logFile, JSON.stringify(data) + '\n');
+  } catch (e) {}
+}
+
+console.log('╔════════════════════════════════════════╗');
+console.log('║         Tera V2 - Production           ║');
+console.log('╚════════════════════════════════════════╝');
+console.log(`Model: ${AI_CONFIG.displayName} (${AI_CONFIG.model})`);
+console.log(`Status: Initializing...\n`);
+
+// Health check
+app.get('/health', async (req, res) => {
+  try {
+    const response = await axios.get('http://localhost:11434/api/tags', { timeout: 5000 });
+    const models = response.data.models || [];
+    const hasModel = models.some(m => m.name.includes(AI_CONFIG.model));
+    
+    res.json({ 
+      status: 'ok', 
+      service: 'Tera V2 Backend',
+      version: '2.0.0',
+      provider: 'Ollama (Local)',
+      model: AI_CONFIG.model,
+      displayName: AI_CONFIG.displayName,
+      modelLoaded: hasModel,
+      mode: 'production'
+    });
+  } catch (error) {
+    res.status(503).json({ 
+      status: 'error', 
+      message: 'Ollama not running',
+      solution: `Run: ollama run ${AI_CONFIG.model}`
+    });
+  }
+});
+
+// Get Tera V2 info
+app.get('/api/info', (req, res) => {
+  res.json({
+    name: 'Tera V2',
+    version: '2.0.0',
+    model: AI_CONFIG.model,
+    capabilities: ['document-analysis', 'chat', 'audio-generation', 'multimodal']
+  });
+});
+
+// Chat endpoint with streaming
+app.post('/api/chat', async (req, res) => {
+  const startTime = Date.now();
+  
+  try {
+    const { messages, sources, docId } = req.body;
+    
+    // Build context from sources
+    let context = '';
+    if (sources && sources.length > 0) {
+      context = '\n\nDOCUMENTS:\n' + sources.map((s, i) => 
+        `[${i + 1}] ${s.name}:\n${s.content?.substring(0, 3000) || '[Image content]'}`
+      ).join('\n\n');
+    }
+
+    const lastMessage = messages[messages.length - 1];
+    const prompt = `${TERAV2_PROMPT}${context}\n\nUser: ${lastMessage.content}\n\nTera V2:`;
+
+    // Log for dataset collection
+    logConversation({
+      timestamp: new Date().toISOString(),
+      type: 'chat',
+      docId,
+      hasSources: sources?.length > 0,
+      query: lastMessage.content
+    });
+
+    res.setHeader('Content-Type', 'text/event-stream');
+    res.setHeader('Cache-Control', 'no-cache');
+    res.setHeader('Connection', 'keep-alive');
+
+    const response = await axios.post(
+      AI_CONFIG.url,
+      {
+        model: AI_CONFIG.model,
+        prompt: prompt,
+        stream: true,
+        options: {
+          temperature: 0.7,
+          num_predict: 4096,
+          top_p: 0.9
+        }
+      },
+      {
+        responseType: 'stream',
+        timeout: 120000
+      }
+    );
+
+    let fullResponse = '';
+
+    response.data.on('data', (chunk) => {
+      const lines = chunk.toString().split('\n');
+      for (const line of lines) {
+        if (line.trim()) {
+          try {
+            const data = JSON.parse(line);
+            if (data.response) {
+              fullResponse += data.response;
+              res.write(`data: {"content": ${JSON.stringify(data.response)}}\n\n`);
+            }
+            if (data.done) {
+              const duration = Date.now() - startTime;
+              logConversation({
+                timestamp: new Date().toISOString(),
+                type: 'response',
+                docId,
+                responseLength: fullResponse.length,
+                duration
+              });
+              res.write('data: {"done": true}\n\n');
+            }
+          } catch (e) {}
+        }
+      }
+    });
+
+    response.data.on('end', () => {
+      res.write('data: {"done": true}\n\n');
+      res.end();
+    });
+
+    response.data.on('error', (error) => {
+      console.error('Stream error:', error);
+      res.write(`data: {"error": "${error.message}"}\n\n`);
+      res.end();
+    });
+
+  } catch (error) {
+    console.error('Chat error:', error.message);
+    if (error.code === 'ECONNREFUSED') {
+      res.status(503).json({ 
+        error: 'Tera V2 AI is not running',
+        solution: `Run: ollama run ${AI_CONFIG.model}`
+      });
+    } else {
+      res.status(500).json({ error: error.message });
+    }
+  }
+});
+
+// Audio script generation
+app.post('/api/generate-audio-script', async (req, res) => {
+  try {
+    const { sources, format, instructions, docId } = req.body;
+    
+    let content = '';
+    if (sources && sources.length > 0) {
+      content = sources.map(s => s.content?.substring(0, 4000) || '').join('\n\n');
+    }
+
+    const formatPrompts = {
+      deepdive: 'Create an in-depth podcast discussion between two hosts (Host 1: Alex, Host 2: Sam) analyzing the content. Include different perspectives and detailed insights.',
+      briefing: 'Create a brief news-style summary of the key points.',
+      story: 'Create a narrative-driven exploration of the content, telling it as an engaging story.'
+    };
+
+    const customInstr = instructions ? ` Additional instructions: ${instructions}` : '';
+    
+    const prompt = `${TERAV2_PROMPT}
+
+You are creating a podcast script. ${formatPrompts[format] || formatPrompts.deepdive}${customInstr}
+
+Format each speaker line as "Host 1: text" or "Host 2: text".
+
+Documents to analyze:
+${content}
+
+Generate a compelling podcast script:`;
+
+    // Log
+    logConversation({
+      timestamp: new Date().toISOString(),
+      type: 'audio-generation',
+      docId,
+      format,
+      hasCustomInstructions: !!instructions
+    });
+
+    const response = await axios.post(
+      AI_CONFIG.url,
+      {
+        model: AI_CONFIG.model,
+        prompt: prompt,
+        stream: false,
+        options: {
+          temperature: 0.8,
+          num_predict: 4096
+        }
+      },
+      { timeout: 120000 }
+    );
+
+    res.json({
+      script: response.data.response,
+      model: AI_CONFIG.displayName,
+      success: true
+    });
+
+  } catch (error) {
+    console.error('Audio script error:', error.message);
+    res.status(500).json({ error: error.message });
+  }
+});
+
+// Save training data endpoint (for future fine-tuning)
+app.post('/api/save-training-data', async (req, res) => {
+  try {
+    const { prompt, response, quality } = req.body;
+    
+    const trainingExample = {
+      instruction: prompt,
+      input: "",
+      output: response,
+      quality: quality || 'good',
+      timestamp: new Date().toISOString()
+    };
+
+    const date = new Date().toISOString().split('T')[0];
+    const file = path.join(DATASET_PATH, `training-${date}.jsonl`);
+    
+    await fs.appendFile(file, JSON.stringify(trainingExample) + '\n');
+    
+    res.json({ success: true });
+  } catch (error) {
+    res.status(500).json({ error: error.message });
+  }
+});
+
+// Export dataset
+app.get('/api/dataset', async (req, res) => {
+  try {
+    const files = await fs.readdir(DATASET_PATH);
+    const allData = [];
+    
+    for (const file of files) {
+      if (file.endsWith('.jsonl')) {
+        const content = await fs.readFile(path.join(DATASET_PATH, file), 'utf-8');
+        const lines = content.trim().split('\n');
+        lines.forEach(line => {
+          try {
+            allData.push(JSON.parse(line));
+          } catch (e) {}
+        });
+      }
+    }
+    
+    res.json({ 
+      count: allData.length,
+      data: allData
+    });
+  } catch (error) {
+    res.status(500).json({ error: error.message });
+  }
+});
+
+// List available Ollama models
+app.get('/api/models', async (req, res) => {
+  try {
+    const response = await axios.get('http://localhost:11434/api/tags');
+    res.json(response.data);
+  } catch (error) {
+    res.status(503).json({ error: 'Ollama not running' });
+  }
+});
+
+// Error handling
+app.use((err, req, res, next) => {
+  console.error(err.stack);
+  res.status(500).json({ error: 'Internal server error' });
+});
+
+app.listen(PORT, () => {
+  console.log('╔════════════════════════════════════════╗');
+  console.log('║     Tera V2 is Ready! 🚀               ║');
+  console.log('╚════════════════════════════════════════╝');
+  console.log(`Server: http://localhost:${PORT}`);
+  console.log(`AI Model: ${AI_CONFIG.displayName}`);
+  console.log(`Status: Production Ready`);
+  console.log('');
+  console.log('Features:');
+  console.log('  ✅ Document Analysis');
+  console.log('  ✅ Audio Generation');
+  console.log('  ✅ Dataset Collection (for training)');
+  console.log('  ✅ 100% Offline (No API keys)');
+  console.log('');
+  console.log('To use Tera V2:');
+  console.log(`1. Keep Ollama running: ollama run ${AI_CONFIG.model}`);
+  console.log('2. Open tera-docs.html in your browser');
+  console.log('');
+  console.log('Legal: Qwen license allows full commercial use & rebranding');
+  console.log('╔════════════════════════════════════════╗\n');
+});
+
+module.exports = app;