mirror of
https://github.com/run-llama/invoice-extraction-vibe-coding.git
synced 2026-06-30 21:27:54 -04:00
refactor (#1)
This commit is contained in:
Jerry Liu
@@ -0,0 +1,21 @@
|
||||
# LlamaCloud API Configuration
|
||||
# Get these values from https://cloud.llamaindex.ai/
|
||||
|
||||
# Your LlamaCloud API key (required)
|
||||
LLAMA_CLOUD_API_KEY=your_api_key_here
|
||||
|
||||
# Your LlamaCloud project ID (required)
|
||||
LLAMA_CLOUD_PROJECT_ID=your_project_id_here
|
||||
|
||||
# Your LlamaCloud organization ID (required)
|
||||
LLAMA_CLOUD_ORGANIZATION_ID=your_organization_id_here
|
||||
|
||||
# Name for your extraction agent (optional - defaults to "invoice_extraction_agent")
|
||||
LLAMA_CLOUD_AGENT_NAME=invoice_extraction_agent
|
||||
|
||||
# Instructions:
|
||||
# 1. Copy this file to .env: cp .env.template .env
|
||||
# 2. Replace the placeholder values with your actual LlamaCloud credentials
|
||||
# 3. Save the .env file
|
||||
# 4. Run: python create_agent.py
|
||||
# 5. Run: python sample.py
|
||||
+83
@@ -0,0 +1,83 @@
|
||||
# Environment variables
|
||||
.env
|
||||
.env.local
|
||||
.env.production
|
||||
|
||||
# Python
|
||||
__pycache__/
|
||||
*.py[cod]
|
||||
*$py.class
|
||||
*.so
|
||||
.Python
|
||||
build/
|
||||
develop-eggs/
|
||||
dist/
|
||||
downloads/
|
||||
eggs/
|
||||
.eggs/
|
||||
lib/
|
||||
lib64/
|
||||
parts/
|
||||
sdist/
|
||||
var/
|
||||
wheels/
|
||||
*.egg-info/
|
||||
.installed.cfg
|
||||
*.egg
|
||||
MANIFEST
|
||||
|
||||
# Virtual environments
|
||||
.venv/
|
||||
venv/
|
||||
ENV/
|
||||
env/
|
||||
.env/
|
||||
|
||||
# IDEs
|
||||
.vscode/
|
||||
.idea/
|
||||
*.swp
|
||||
*.swo
|
||||
*~
|
||||
|
||||
# OS
|
||||
.DS_Store
|
||||
.DS_Store?
|
||||
._*
|
||||
.Spotlight-V100
|
||||
.Trashes
|
||||
ehthumbs.db
|
||||
Thumbs.db
|
||||
|
||||
# Streamlit
|
||||
.streamlit/
|
||||
|
||||
# Temporary files and development folders
|
||||
tmp_*/
|
||||
temp/
|
||||
*.tmp
|
||||
*.log
|
||||
|
||||
# Node modules (for any JS tooling)
|
||||
node_modules/
|
||||
*.log
|
||||
npm-debug.log*
|
||||
|
||||
# Next.js (excluding from repo)
|
||||
nextjs_app/
|
||||
nextjs_app2/
|
||||
|
||||
# Coverage reports
|
||||
htmlcov/
|
||||
.coverage
|
||||
.coverage.*
|
||||
coverage.xml
|
||||
*.cover
|
||||
|
||||
# Pytest
|
||||
.pytest_cache/
|
||||
|
||||
# mypy
|
||||
.mypy_cache/
|
||||
.dmypy.json
|
||||
dmypy.json
|
||||
@@ -1,62 +1,177 @@
|
||||
# Invoice Extraction Streamlit App
|
||||
# Invoice Processing App - Vibe Coding Example
|
||||
|
||||
A simple Streamlit application that allows users to upload invoice images and extract structured data using LlamaCloud's extraction agent.
|
||||
This repository demonstrates how to "vibe code" a full-stack invoice processing application using Cursor/Claude Code from a simple starting file. This showcases building a complete Streamlit application with LlamaCloud's document extraction capabilities.
|
||||
|
||||
## Features
|
||||
## 🚀 Quick Start - Vibe Coding Your Own App
|
||||
|
||||
- 📄 Upload invoice images (JPG, JPEG, PNG, BMP, TIFF)
|
||||
- 🔍 Automatic data extraction using LlamaCloud's kaggle_invoice_agent
|
||||
- 📊 Display extracted data in both JSON and formatted views
|
||||
- 🎨 Clean, modern UI with progress indicators
|
||||
This repository demonstrates **vibe coding** - rapidly building applications through iterative AI-assisted development. The goal is to transform a simple script into a full-featured application using Cursor/Claude Code.
|
||||
|
||||
## Setup
|
||||
### Prerequisites
|
||||
- Python 3.11+
|
||||
- LlamaCloud account and API key ([Get one here](https://cloud.llamaindex.ai/))
|
||||
|
||||
1. **Install dependencies:**
|
||||
```bash
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
### Setup Process
|
||||
|
||||
2. **Set up environment variables:**
|
||||
Create a `.env` file in the project root with your LlamaCloud API credentials:
|
||||
```
|
||||
LLAMA_CLOUD_API_KEY=your_api_key_here
|
||||
```
|
||||
|
||||
3. **Run the app:**
|
||||
```bash
|
||||
streamlit run app.py
|
||||
```
|
||||
|
||||
4. **Open your browser:**
|
||||
The app will be available at `http://localhost:8501`
|
||||
|
||||
## Usage
|
||||
|
||||
1. Upload an invoice image using the file uploader
|
||||
2. Click "Extract Data" to run the extraction
|
||||
3. View the extracted data in both JSON and formatted formats
|
||||
4. The sidebar shows configuration details and app information
|
||||
|
||||
## Configuration
|
||||
|
||||
The app uses the following configuration (from `sample.py`):
|
||||
- **Project ID:** `2fef999e-1073-40e6-aeb3-1f3c0e64d99b`
|
||||
- **Organization ID:** `43b88c8f-e488-46f6-9013-698e3d2e374a`
|
||||
- **Agent:** `kaggle_invoice_agent`
|
||||
|
||||
## File Structure
|
||||
|
||||
```
|
||||
jerry_invoice_streamlit/
|
||||
├── app.py # Main Streamlit application
|
||||
├── sample.py # Original sample code
|
||||
├── requirements.txt # Python dependencies
|
||||
├── README.md # This file
|
||||
└── .env # Environment variables (create this)
|
||||
#### 1. **Clone and Install**
|
||||
```bash
|
||||
git clone <your-repo-url>
|
||||
cd invoice-extraction-vibe-coding
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
|
||||
## Troubleshooting
|
||||
#### 2. **Set Up LlamaCloud**
|
||||
1. Go to [LlamaCloud](https://cloud.llamaindex.ai/) and log in
|
||||
2. Create a new project and note your `project_id` and `organization_id`
|
||||
3. Get your API key from your account settings
|
||||
|
||||
- **API Key Issues:** Make sure your `.env` file contains the correct `LLAMA_CLOUD_API_KEY`
|
||||
- **Import Errors:** Ensure all dependencies are installed with `pip install -r requirements.txt`
|
||||
- **File Upload Issues:** Check that your image file is in a supported format
|
||||
#### 3. **Configure Environment**
|
||||
Copy the template and add your credentials:
|
||||
```bash
|
||||
cp .env.template .env
|
||||
```
|
||||
Then edit `.env` with your LlamaCloud credentials:
|
||||
```env
|
||||
LLAMA_CLOUD_API_KEY=your_actual_api_key
|
||||
LLAMA_CLOUD_PROJECT_ID=your_actual_project_id
|
||||
LLAMA_CLOUD_ORGANIZATION_ID=your_actual_organization_id
|
||||
LLAMA_CLOUD_AGENT_NAME=invoice_extraction_agent
|
||||
```
|
||||
|
||||
#### 4. **Create Extraction Agent**
|
||||
```bash
|
||||
python create_agent.py
|
||||
```
|
||||
This automatically creates an extraction agent in LlamaCloud using the invoice schema from `sample_data/sample_schema.py`.
|
||||
|
||||
#### 5. **Test Your Setup**
|
||||
```bash
|
||||
python sample.py
|
||||
```
|
||||
This validates your configuration and runs extraction on the sample invoice. You should see structured JSON output.
|
||||
|
||||
#### 6. **Start Vibe Coding!**
|
||||
Now comes the fun part - use the `cursor_prompt.md` template to transform `sample.py` into a full application:
|
||||
|
||||
1. Open your project in Cursor or Claude Code
|
||||
2. Copy the prompt from `cursor_prompt.md`
|
||||
3. Start iterating: "Transform this simple script into a professional Streamlit app with..."
|
||||
4. Build features incrementally through natural language prompts
|
||||
|
||||
#### 7. **Compare Your Result**
|
||||
When you're done vibe coding, compare your creation with our `app.py` to see different approaches!
|
||||
|
||||
## 📁 Project Structure
|
||||
|
||||
```
|
||||
invoice-extraction-vibe-coding/
|
||||
├── app.py # Full-featured Streamlit application (generated)
|
||||
├── sample.py # Starting point - simple extraction script
|
||||
├── create_agent.py # Script to create LlamaCloud extraction agent
|
||||
├── .env.template # Environment variables template
|
||||
├── requirements.txt # Python dependencies
|
||||
├── cursor_prompt.md # Cursor prompt template for vibe coding
|
||||
└── sample_data/ # Sample invoice data
|
||||
├── batch1-0274.jpg # Sample invoice image
|
||||
└── sample_schema.py # Pydantic models for data structure
|
||||
```
|
||||
|
||||
## 🎯 What This Demonstrates
|
||||
|
||||
### From Simple Script to Full App
|
||||
- **Starting Point**: `sample.py` - A basic 20-line script that calls LlamaCloud extraction
|
||||
- **End Result**: `app.py` - A fully-featured Streamlit application with:
|
||||
- Professional UI with custom CSS
|
||||
- File upload functionality
|
||||
- Real-time data extraction
|
||||
- Structured data display
|
||||
- Invoice history tracking
|
||||
- Status indicators and metrics
|
||||
|
||||
### LlamaCloud Integration
|
||||
- Document parsing and extraction using LlamaExtract
|
||||
- Structured data output using predefined schemas
|
||||
- Integration with kaggle_invoice_agent
|
||||
|
||||
## 🛠 Vibe Coding with the Prompt Template
|
||||
|
||||
This project demonstrates **vibe coding** - rapidly building applications through conversational AI development.
|
||||
|
||||
### Start with `cursor_prompt.md`
|
||||
The `cursor_prompt.md` file contains a comprehensive prompt template that can potentially **one-shot** the entire application transformation:
|
||||
|
||||
1. **Copy the main prompt** from `cursor_prompt.md`
|
||||
2. **Paste it into Cursor/Claude Code** with your `sample.py` file open
|
||||
3. **Watch the magic happen** - the AI may build the entire Streamlit app in one go!
|
||||
|
||||
### The Vibe Coding Process:
|
||||
- **Start Simple**: `sample.py` - a working 20-line script
|
||||
- **Use Natural Language**: Describe what you want, not how to build it
|
||||
- **Iterate if Needed**: Add features through follow-up prompts
|
||||
- **Build Incrementally**: Test and refine each addition
|
||||
|
||||
### Why the Template Works:
|
||||
- **Comprehensive Scope**: Covers UI, functionality, and user experience
|
||||
- **Clear Context**: References your actual files and sample data
|
||||
- **Specific Examples**: Shows exactly what features to build
|
||||
- **Production-Ready**: Asks for professional-quality output
|
||||
|
||||
Try the full prompt first - you might be surprised how much gets built in a single interaction!
|
||||
|
||||
## 🔧 Features
|
||||
|
||||
### Current Application (`app.py`)
|
||||
- 📄 **Multi-format Support**: JPG, JPEG, PNG, BMP, TIFF
|
||||
- 🎨 **Professional UI**: Custom CSS styling and responsive design
|
||||
- 📊 **Data Visualization**: Structured invoice data display
|
||||
- 📈 **Analytics**: Processing metrics and status tracking
|
||||
- 🔄 **Real-time Processing**: Live extraction with progress indicators
|
||||
- 💾 **Session Storage**: Invoice history within session
|
||||
|
||||
### Data Schema
|
||||
The application extracts structured invoice data including:
|
||||
- Invoice metadata (number, date)
|
||||
- Seller and client information
|
||||
- Line items with pricing details
|
||||
- VAT calculations and summaries
|
||||
- Total amounts
|
||||
|
||||
See `sample_data/sample_schema.py` for the complete Pydantic model definitions.
|
||||
|
||||
## 📝 Usage
|
||||
|
||||
1. **Upload Invoice**: Drag and drop or select an invoice image
|
||||
2. **Extract Data**: Click the extraction button to process with LlamaCloud
|
||||
3. **View Results**: See structured data in formatted tables and JSON
|
||||
4. **Track History**: View processed invoices in the history tab
|
||||
|
||||
## 🔑 Configuration
|
||||
|
||||
Update these values in your application for your own LlamaCloud setup:
|
||||
```python
|
||||
project_id = "your-project-id"
|
||||
organization_id = "your-organization-id"
|
||||
agent_name = "your-agent-name"
|
||||
```
|
||||
|
||||
## 🐛 Troubleshooting
|
||||
|
||||
### Common Issues
|
||||
- **API Key**: Ensure your LlamaCloud API key is valid and in `.env`
|
||||
- **Dependencies**: Run `pip install -r requirements.txt`
|
||||
- **File Formats**: Only image formats are supported
|
||||
- **File Size**: Keep images under 10MB for best performance
|
||||
|
||||
## 🤝 Contributing
|
||||
|
||||
This is an educational example demonstrating the vibe coding approach. Feel free to:
|
||||
- Fork and modify for your use case
|
||||
- Share improvements and variations
|
||||
- Use as a starting point for your own projects
|
||||
|
||||
## 📄 License
|
||||
|
||||
Open source - use freely for learning and development.
|
||||
|
||||
---
|
||||
|
||||
**Built with ❤️ using LlamaIndex, LlamaCloud, and the power of vibe coding with Cursor/Claude Code!**
|
||||
+126
@@ -0,0 +1,126 @@
|
||||
"""
|
||||
Create LlamaCloud extraction agent for invoice processing.
|
||||
|
||||
Run this script to automatically create an extraction agent with the invoice schema
|
||||
before testing with sample.py or building your Streamlit app.
|
||||
"""
|
||||
|
||||
import os
|
||||
import sys
|
||||
from dotenv import load_dotenv
|
||||
from llama_cloud_services import LlamaExtract
|
||||
from llama_cloud.core.api_error import ApiError
|
||||
from sample_data.sample_schema import Invoice
|
||||
|
||||
# Load environment variables
|
||||
load_dotenv()
|
||||
|
||||
def create_invoice_agent():
|
||||
"""Create or update the invoice extraction agent."""
|
||||
|
||||
# Get configuration
|
||||
project_id = os.getenv("LLAMA_CLOUD_PROJECT_ID")
|
||||
organization_id = os.getenv("LLAMA_CLOUD_ORGANIZATION_ID")
|
||||
agent_name = os.getenv("LLAMA_CLOUD_AGENT_NAME", "invoice_extraction_agent")
|
||||
|
||||
# Validate configuration
|
||||
if not project_id or project_id == "your-project-id-here":
|
||||
print("❌ Error: LLAMA_CLOUD_PROJECT_ID not configured!")
|
||||
print("Please set your project ID in the .env file")
|
||||
return False
|
||||
|
||||
if not organization_id or organization_id == "your-organization-id-here":
|
||||
print("❌ Error: LLAMA_CLOUD_ORGANIZATION_ID not configured!")
|
||||
print("Please set your organization ID in the .env file")
|
||||
return False
|
||||
|
||||
try:
|
||||
print("🚀 Initializing LlamaCloud extraction service...")
|
||||
|
||||
# Initialize LlamaExtract
|
||||
extract = LlamaExtract(
|
||||
show_progress=True,
|
||||
check_interval=5,
|
||||
project_id=project_id,
|
||||
organization_id=organization_id
|
||||
)
|
||||
|
||||
print(f"✅ Connected to LlamaCloud")
|
||||
print(f"📋 Project ID: {project_id}")
|
||||
print(f"🏢 Organization ID: {organization_id}")
|
||||
print(f"🤖 Agent name: {agent_name}")
|
||||
|
||||
# Check if agent already exists
|
||||
try:
|
||||
existing_agent = extract.get_agent(name=agent_name)
|
||||
if existing_agent:
|
||||
print(f"✅ Agent '{agent_name}' already exists!")
|
||||
print(f"🤖 Agent ID: {existing_agent.id}")
|
||||
print("📝 You can use this existing agent for extraction.")
|
||||
return True
|
||||
except ApiError as e:
|
||||
if e.status_code == 404:
|
||||
print(f"📝 Agent '{agent_name}' does not exist, creating new one...")
|
||||
else:
|
||||
raise
|
||||
|
||||
# Create new agent with invoice schema
|
||||
print("🔧 Creating new extraction agent...")
|
||||
print("📊 Using invoice schema from sample_data/sample_schema.py")
|
||||
|
||||
agent = extract.create_agent(
|
||||
name=agent_name,
|
||||
data_schema=Invoice
|
||||
)
|
||||
|
||||
print("🎉 Success! Extraction agent created successfully!")
|
||||
print(f"🤖 Agent ID: {agent.id}")
|
||||
print(f"📝 Agent Name: {agent.name}")
|
||||
|
||||
print("\n📋 Schema Summary:")
|
||||
print(" • Invoice metadata (number, date)")
|
||||
print(" • Seller and client information")
|
||||
print(" • Line items with pricing details")
|
||||
print(" • VAT calculations and summaries")
|
||||
print(" • Total amounts")
|
||||
|
||||
print(f"\n✅ Setup complete! Your agent '{agent_name}' is ready to use.")
|
||||
print("\n🚀 Next steps:")
|
||||
print("1. Run 'python sample.py' to test extraction")
|
||||
print("2. Use 'cursor_prompt.md' to build your Streamlit app")
|
||||
|
||||
return True
|
||||
|
||||
except Exception as e:
|
||||
print(f"❌ Error creating agent: {e}")
|
||||
print("\n🔧 Troubleshooting:")
|
||||
print("1. Check your .env file has correct credentials")
|
||||
print("2. Verify your LlamaCloud account has necessary permissions")
|
||||
print("3. Ensure you have a valid project and organization setup")
|
||||
return False
|
||||
|
||||
def main():
|
||||
"""Main function."""
|
||||
print("🔨 LlamaCloud Invoice Extraction Agent Setup")
|
||||
print("=" * 50)
|
||||
|
||||
# Check if .env file exists
|
||||
if not os.path.exists(".env"):
|
||||
print("❌ No .env file found!")
|
||||
print("\nPlease create a .env file with:")
|
||||
print("LLAMA_CLOUD_API_KEY=your_api_key_here")
|
||||
print("LLAMA_CLOUD_PROJECT_ID=your_project_id_here")
|
||||
print("LLAMA_CLOUD_ORGANIZATION_ID=your_organization_id_here")
|
||||
print("LLAMA_CLOUD_AGENT_NAME=invoice_extraction_agent")
|
||||
sys.exit(1)
|
||||
|
||||
# Create agent
|
||||
success = create_invoice_agent()
|
||||
|
||||
if success:
|
||||
sys.exit(0)
|
||||
else:
|
||||
sys.exit(1)
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
@@ -0,0 +1,111 @@
|
||||
# Cursor Prompt for Invoice Processing App
|
||||
|
||||
Use this prompt to transform the simple `sample.py` script into a full-featured Streamlit invoice processing application.
|
||||
|
||||
## Main Prompt
|
||||
|
||||
```
|
||||
I have a simple piece of sample code (sample.py) that initializes a LlamaCloud extraction agent and runs extraction over a sample invoice image. I want to transform this into a professional Streamlit web application.
|
||||
|
||||
Here's what I'm starting with:
|
||||
- sample.py: Basic extraction script using LlamaExtract with kaggle_invoice_agent
|
||||
- sample_data/batch1-0274.jpg: Sample invoice image for testing
|
||||
- sample_data/sample_schema.py: Pydantic models defining the expected invoice data structure
|
||||
|
||||
Transform this into a comprehensive Streamlit app with:
|
||||
|
||||
## Core Features
|
||||
1. **File Upload Interface**
|
||||
- Support multiple document formats:
|
||||
- PDF files (.pdf)
|
||||
- Word documents (.docx)
|
||||
- Image files (.jpg, .jpeg, .png, .bmp, .tiff)
|
||||
- And other common document formats supported by LlamaCloud
|
||||
- Drag-and-drop functionality
|
||||
- Document preview before processing
|
||||
- File validation and size limits
|
||||
|
||||
2. **Data Extraction & Display**
|
||||
- Use the same LlamaExtract agent from sample.py
|
||||
- Process uploaded invoices in real-time
|
||||
- Display extracted data using the schema structure from sample_schema.py
|
||||
- Show both formatted view and raw JSON
|
||||
|
||||
3. **Professional UI Design**
|
||||
- Modern, clean interface with custom CSS
|
||||
- Responsive layout with proper spacing
|
||||
- Professional color scheme and typography
|
||||
- Loading states and progress indicators
|
||||
|
||||
## Data Presentation
|
||||
Based on the sample_schema.py structure, display:
|
||||
- Invoice overview (number, date, status)
|
||||
- Seller and client information in organized cards
|
||||
- Line items in a structured table
|
||||
- VAT summary with calculations
|
||||
- Total amounts prominently displayed
|
||||
|
||||
## Additional Features
|
||||
- Session-based invoice history
|
||||
- Processing status and metrics in sidebar
|
||||
- Error handling with user-friendly messages
|
||||
- System status indicators
|
||||
- Usage statistics and tips
|
||||
|
||||
## Technical Requirements
|
||||
- Use the same project_id and organization_id from sample.py
|
||||
- Maintain the kaggle_invoice_agent configuration
|
||||
- Include proper environment variable handling for API keys
|
||||
- Add comprehensive error handling for API failures
|
||||
|
||||
Make it production-ready with clean code structure, proper documentation, and professional presentation. The app should feel like a commercial invoice processing tool while maintaining the simplicity of the original extraction logic.
|
||||
|
||||
Test with the provided sample_data/batch1-0274.jpg to ensure everything works correctly.
|
||||
```
|
||||
|
||||
## Follow-up Enhancement Prompts
|
||||
|
||||
### UI Polish
|
||||
```
|
||||
Enhance the visual design with:
|
||||
- Custom CSS for a more professional look
|
||||
- Better color scheme (consider invoice/business themes)
|
||||
- Improved typography and spacing
|
||||
- Status badges and progress indicators
|
||||
- Responsive design for different screen sizes
|
||||
```
|
||||
|
||||
### Feature Expansion
|
||||
```
|
||||
Add these capabilities:
|
||||
- Export results to JSON/CSV
|
||||
- Batch processing of multiple invoices
|
||||
- Invoice data validation against schema
|
||||
- Search and filter processed invoices
|
||||
- Comparison between different invoices
|
||||
```
|
||||
|
||||
### User Experience
|
||||
```
|
||||
Improve the user experience with:
|
||||
- Better error messages and recovery options
|
||||
- Helpful tooltips and guidance
|
||||
- Keyboard shortcuts and accessibility
|
||||
- Auto-save of processing results
|
||||
- Undo/redo functionality where applicable
|
||||
```
|
||||
|
||||
## Key Context Files
|
||||
|
||||
When using this prompt, make sure to include:
|
||||
- `sample.py` - The starting extraction script
|
||||
- `sample_data/sample_schema.py` - Pydantic models for data structure
|
||||
- `sample_data/batch1-0274.jpg` - Test invoice image
|
||||
- Current `requirements.txt` - Dependencies list
|
||||
|
||||
## Expected Transformation
|
||||
|
||||
**From:** 20-line extraction script
|
||||
**To:** Full-featured web application with professional UI, comprehensive functionality, file handling, data visualization, and production-ready code quality.
|
||||
|
||||
The result should demonstrate the power of "vibe coding" - rapidly building complex applications through iterative AI-assisted development.
|
||||
+3
-3
@@ -1,4 +1,4 @@
|
||||
streamlit>=1.28.0
|
||||
python-dotenv>=1.0.0
|
||||
llama-cloud-services>=0.1.0
|
||||
llama-cloud>=0.1.0
|
||||
python-dotenv>=1.1.1
|
||||
llama-cloud-services==0.6.49
|
||||
llama-cloud==0.1.34
|
||||
@@ -1,22 +1,83 @@
|
||||
"""
|
||||
Simple invoice extraction script using LlamaCloud.
|
||||
|
||||
This is the starting point for vibe coding a full Streamlit application.
|
||||
Configure your LlamaCloud credentials and run this script to test extraction
|
||||
before building the full web application.
|
||||
"""
|
||||
|
||||
import os
|
||||
import json
|
||||
from dotenv import load_dotenv
|
||||
from llama_cloud_services import LlamaExtract
|
||||
from llama_cloud.core.api_error import ApiError
|
||||
|
||||
# Load environment variables
|
||||
load_dotenv()
|
||||
|
||||
project_id = "2fef999e-1073-40e6-aeb3-1f3c0e64d99b"
|
||||
organization_id = "43b88c8f-e488-46f6-9013-698e3d2e374a"
|
||||
# Configuration - Update these with your LlamaCloud details
|
||||
PROJECT_ID = os.getenv("LLAMA_CLOUD_PROJECT_ID", "your-project-id-here")
|
||||
ORGANIZATION_ID = os.getenv("LLAMA_CLOUD_ORGANIZATION_ID", "your-organization-id-here")
|
||||
AGENT_NAME = os.getenv("LLAMA_CLOUD_AGENT_NAME", "your-agent-name-here")
|
||||
|
||||
# Optionally, add your project id/organization id
|
||||
extract = LlamaExtract(
|
||||
show_progress=False,
|
||||
check_interval=5,
|
||||
project_id=project_id,
|
||||
organization_id=organization_id
|
||||
)
|
||||
# Sample image path
|
||||
SAMPLE_IMAGE = "sample_data/batch1-0274.jpg"
|
||||
|
||||
def main():
|
||||
"""Run invoice extraction on sample image."""
|
||||
|
||||
# Validate configuration
|
||||
if PROJECT_ID == "your-project-id-here" or ORGANIZATION_ID == "your-organization-id-here":
|
||||
print("❌ Please configure your LlamaCloud credentials!")
|
||||
print("Update your .env file with:")
|
||||
print("LLAMA_CLOUD_PROJECT_ID=your-actual-project-id")
|
||||
print("LLAMA_CLOUD_ORGANIZATION_ID=your-actual-organization-id")
|
||||
print("LLAMA_CLOUD_AGENT_NAME=your-actual-agent-name")
|
||||
return
|
||||
|
||||
try:
|
||||
print("🚀 Initializing LlamaCloud extraction agent...")
|
||||
|
||||
# Initialize the extraction service
|
||||
extract = LlamaExtract(
|
||||
show_progress=True, # Show progress for better UX
|
||||
check_interval=5,
|
||||
project_id=PROJECT_ID,
|
||||
organization_id=ORGANIZATION_ID
|
||||
)
|
||||
|
||||
# Get the configured agent
|
||||
agent = extract.get_agent(name=AGENT_NAME)
|
||||
print(f"✅ Successfully connected to agent: {AGENT_NAME}")
|
||||
|
||||
# Run extraction on sample image
|
||||
print(f"📄 Processing sample invoice: {SAMPLE_IMAGE}")
|
||||
result = agent.extract(SAMPLE_IMAGE)
|
||||
|
||||
# Display results
|
||||
print("✅ Extraction completed successfully!")
|
||||
print("\n📊 Extracted Data:")
|
||||
print("=" * 50)
|
||||
print(json.dumps(result.data, indent=2))
|
||||
|
||||
# Save results for reference
|
||||
output_file = "sample_output.json"
|
||||
with open(output_file, "w") as f:
|
||||
json.dump(result.data, f, indent=2)
|
||||
print(f"\n💾 Results saved to: {output_file}")
|
||||
|
||||
print("\n🎉 Success! You're ready to start vibe coding the Streamlit app!")
|
||||
print("Use cursor_prompt.md to transform this script into a full application.")
|
||||
|
||||
except ApiError as e:
|
||||
print(f"❌ API Error: {e}")
|
||||
print("Check your API key and agent configuration.")
|
||||
except FileNotFoundError:
|
||||
print(f"❌ Sample image not found: {SAMPLE_IMAGE}")
|
||||
print("Make sure the sample_data directory exists with the sample image.")
|
||||
except Exception as e:
|
||||
print(f"❌ Unexpected error: {e}")
|
||||
|
||||
agent = extract.get_agent(name="kaggle_invoice_agent")
|
||||
|
||||
result = agent.extract("batch1-0274.jpg")
|
||||
result.data # outputs json of the extracted output according to the schema defined in the agent
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
|
||||
|
||||
Reference in New Issue
Block a user