Revert-Railway troubleshooting

Procfile

@@ -1 +1 @@
-web: bash railway-start.sh 
+web: uvicorn api.main:app --host 0.0.0.0 --port $PORT 

RAILWAY-FIXES.md

@@ -1,74 +0,0 @@
-# 🔧 Railway Deployment Fixes Applied
-
-## Issues Found & Fixed
-
-### ✅ 1. Port Configuration Fixed
-**Problem:** `api/main.py` was hardcoded to port 8000
-**Fix:** Updated to use `$PORT` environment variable
-```python
-port = int(os.getenv("PORT", 8000))
-uvicorn.run(app, host="0.0.0.0", port=port)
-```
-
-### ✅ 2. NLTK Data Download Added
-**Problem:** NLTK data not available on Railway
-**Fix:** Created `railway-start.sh` to download NLTK data on startup
-```bash
-python -c "import nltk; nltk.download('punkt'); nltk.download('stopwords')"
-```
-
-### ✅ 3. Startup Script Created
-**Problem:** No proper startup sequence
-**Fix:** Created `railway-start.sh` and updated `Procfile`
-```
-web: bash railway-start.sh
-```
-
-### ✅ 4. Memory Optimization
-**Problem:** Large AI model causing memory issues
-**Fix:** Changed default model from `facebook/bart-large-cnn` to `t5-small`
-- Smaller model (60MB vs 1.6GB)
-- Faster loading
-- Less memory usage
-
-## Files Modified
-
-1. **`api/main.py`** - Added PORT environment variable support
-2. **`api/summarizer.py`** - Changed default model to t5-small
-3. **`Procfile`** - Updated to use startup script
-4. **`railway-start.sh`** - Created new startup script
-5. **`railway-troubleshooting.md`** - Created troubleshooting guide
-
-## Next Steps
-
-1. **Commit and push** these changes to GitHub
-2. **Redeploy** on Railway
-3. **Check logs** for any remaining issues
-4. **Test the API** at your Railway URL + `/docs`
-
-## Expected Results
-
-- ✅ Build should succeed
-- ✅ App should start properly
-- ✅ API should be accessible
-- ✅ Model loading should work (with smaller model)
-- ✅ NLTK data should be available
-
-## If Still Having Issues
-
-1. **Check Railway logs** for specific error messages
-2. **Upgrade Railway plan** if memory issues persist
-3. **Consider Hugging Face Spaces** as alternative
-4. **Use even smaller model** if needed
-
-## Testing Your Deployment
-
-Once deployed, test these endpoints:
-- `GET /` - Basic info
-- `GET /health` - Health check
-- `GET /models` - Available models
-- `POST /upload-pdf` - PDF validation
-- `POST /summarize` - Full summarization
-
-Your Railway URL will be something like:
-`https://your-app-name.up.railway.app` 

api/main.py

@@ -221,8 +221,4 @@ async def change_model(model_name: str):
 
 if __name__ == "__main__":
     import uvicorn
-    import os
-    
-    # Get port from environment variable (for Railway deployment)
-    port = int(os.getenv("PORT", 8000))
-    uvicorn.run(app, host="0.0.0.0", port=port) 
+    uvicorn.run(app, host="0.0.0.0", port=8000) 

api/summarizer.py

@@ -11,7 +11,7 @@ class BookSummarizer:
     Handles AI-powered text summarization using transformer models.
     """
     
-    def __init__(self, model_name: str = "t5-small"):
+    def __init__(self, model_name: str = "facebook/bart-large-cnn"):
         """
         Initialize the summarizer with a specific model.
         

railway-start.sh

@@ -1,9 +0,0 @@
-#!/bin/bash
-
-# Download NLTK data
-echo "📥 Downloading NLTK data..."
-python -c "import nltk; nltk.download('punkt'); nltk.download('stopwords')"
-
-# Start the FastAPI server
-echo "🚀 Starting Book Summarizer API..."
-uvicorn api.main:app --host 0.0.0.0 --port $PORT 

railway-troubleshooting.md

@@ -1,97 +0,0 @@
-# 🚂 Railway Deployment Troubleshooting
-
-## Your Build Was Successful! ✅
-The log shows "Successfully Built!" which means your Docker image was created correctly.
-
-## Common Runtime Issues & Solutions
-
-### 1. **Port Configuration Issue**
-**Problem:** App not listening on the correct port
-**Solution:** Make sure your FastAPI app uses the PORT environment variable
-
-Check your `api/main.py` - it should look like this:
-```python
-import os
-import uvicorn
-
-if __name__ == "__main__":
-    port = int(os.getenv("PORT", 8000))
-    uvicorn.run("api.main:app", host="0.0.0.0", port=port, reload=False)
-```
-
-### 2. **Memory Issues (Most Common)**
-**Problem:** AI models require more memory than Railway's free tier provides
-**Solution:** 
-- Upgrade to Railway's paid plan ($5/month)
-- Or use a smaller model in `api/summarizer.py`
-
-### 3. **Model Download Timeout**
-**Problem:** Hugging Face models take too long to download
-**Solution:** Add model caching or use smaller models
-
-### 4. **Missing Dependencies**
-**Problem:** Some packages not installed correctly
-**Solution:** Check your `requirements.txt` is complete
-
-## Quick Fixes to Try
-
-### Fix 1: Update your Procfile
-Make sure your `Procfile` contains:
-```
-web: uvicorn api.main:app --host 0.0.0.0 --port $PORT
-```
-
-### Fix 2: Add startup script
-Create `start.sh`:
-```bash
-#!/bin/bash
-python -c "import nltk; nltk.download('punkt'); nltk.download('stopwords')"
-uvicorn api.main:app --host 0.0.0.0 --port $PORT
-```
-
-### Fix 3: Use smaller model
-In `api/summarizer.py`, change default model to:
-```python
-def __init__(self, model_name: str = "t5-small"):  # Smaller, faster model
-```
-
-## How to Check Railway Logs
-
-1. Go to your Railway dashboard
-2. Click on your project
-3. Go to "Deployments" tab
-4. Click on the latest deployment
-5. Check "Logs" for runtime errors
-
-## Common Error Messages & Solutions
-
-| Error | Solution |
-|-------|----------|
-| `Port already in use` | Check Procfile uses `$PORT` |
-| `Out of memory` | Upgrade Railway plan or use smaller model |
-| `Model download failed` | Use smaller model or add retry logic |
-| `Module not found` | Check requirements.txt |
-| `NLTK data missing` | Add NLTK download to startup |
-
-## Quick Test
-
-Try this minimal test first:
-1. Temporarily comment out model loading in `api/main.py`
-2. Deploy to Railway
-3. If it works, the issue is with the AI model loading
-4. Then gradually add back the AI functionality
-
-## Alternative: Use Hugging Face Spaces
-
-If Railway continues to have issues, consider deploying to Hugging Face Spaces:
-- Free GPU support
-- Optimized for AI models
-- Automatic deployment from GitHub
-- Better for transformer models
-
-## Need Help?
-
-1. Check Railway logs for specific error messages
-2. Try the quick fixes above
-3. Consider upgrading Railway plan for more resources
-4. Or switch to Hugging Face Spaces for AI-specific hosting 

start.sh

@@ -25,12 +25,4 @@ echo "✅ Dependencies installed"
 
 echo ""
 echo "🚀 Starting Book Summarizer AI..."
-python3 start.py
-
-# Download NLTK data
-echo "📥 Downloading NLTK data..."
-python -c "import nltk; nltk.download('punkt'); nltk.download('stopwords')"
-
-# Start the FastAPI server
-echo "🚀 Starting Book Summarizer API..."
-uvicorn api.main:app --host 0.0.0.0 --port $PORT 
+python3 start.py