Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/sohamratnaparkhi/geni-ai-travel-planner


https://github.com/sohamratnaparkhi/geni-ai-travel-planner

Last synced: about 1 month ago
JSON representation

Awesome Lists containing this project

README 

          
# Travel Planner API



A comprehensive FastAPI backend for AI-powered travel planning using Google Gemini AI.



## Features



- 🚀 **Parallel Processing**: Process multiple destinations simultaneously

- 🤖 **Google Gemini AI**: Powered by Google's advanced AI for travel recommendations

- 📊 **Comprehensive Data**: Activities, restaurants, and accommodation for each destination

- 💰 **Budget-aware**: All recommendations respect your budget constraints

- âš¡ **Fast**: Parallel processing for quick results



## Setup



### 1. Install Dependencies

```bash

pip install -r requirements.txt

```



### 2. Set up Google Gemini AI

1. Get your API key from [Google AI Studio](https://makersuite.google.com/app/apikey)

2. Set environment variable:

```bash

export GEMINI_API_KEY=your_api_key_here

# or on Windows

set GEMINI_API_KEY=your_api_key_here

```



### 3. Start the Server

```bash

uvicorn main:app --reload --host 0.0.0.0 --port 8000

```



## API Endpoints



### Process Destinations (Background Processing)

**POST** `/travel/process-destinations`



Start background processing of multiple destinations with AI-powered recommendations.



**Request Body:**

```json

{

  "destinations": [

    {

      "place": "Jaipur",

      "days": 3,

      "budget": 15000,

      "custom_ins": "vegetarian food, historic sites, no clubs, budget-friendly"

    },

    {

      "place": "Goa",

      "days": 5,

      "budget": 25000,

      "custom_ins": "beach activities, seafood, luxury hotels, adventure sports"

    }

  ]

}

```



**Request Fields:**

- `place`: Destination name (required)

- `days`: Number of days for the trip (required)

- `budget`: Total budget in INR (required)

- `custom_ins`: Custom user preferences like "vegetarian food, historic sites, luxury hotels" (optional)



**Response (Immediate):**

```json

{

  "task_id": "550e8400-e29b-41d4-a716-446655440000",

  "status": "processing",

  "message": "Started processing 2 destinations in background",

  "created_at": 1699123456.789,

  "destinations": [

    {

      "place": "Jaipur",

      "days": 3,

      "budget": 15000,

      "processing_status": "processing"

    }

  ]

}

```



### Get Task Status

**GET** `/travel/task-status/{task_id}`



Poll the server to check the status of a background processing task.



**Response:**

```json

{

  "task_id": "550e8400-e29b-41d4-a716-446655440000",

  "status": "completed",

  "message": "Successfully processed Jaipur",

  "created_at": 1699123456.789,

  "destinations": [

    {

      "place": "Jaipur",

      "days": 3,

      "budget": 15000,

      "processing_status": "completed",

      "activities": { ... },

      "restaurants": { ... },

      "accommodation": { ... }

    }

  ]

}

```



### Available Endpoints

- `GET /` - API information

- `GET /health` - Health check

- `POST /travel/process-destinations` - Start background processing

- `GET /travel/task-status/{task_id}` - Check processing status

- `GET /travel/plans` - Get travel plans

- `POST /travel/plans` - Create travel plan

- `GET /travel/destinations` - Popular destinations

- `GET /user/users` - Get users

- `POST /user/users` - Create user



## Architecture



### Background Processing Architecture

- **Immediate Response**: Returns task ID instantly for frontend polling

- **Background Processing**: Each destination processed in background tasks

- **Parallel AI Calls**: For each destination, 3 AI tasks run in parallel:

  - Activities & things to do

  - Restaurants & local foods

  - Accommodation recommendations

- **Status Tracking**: Real-time status updates via task polling



### AI Prompts

The system uses 3 detailed prompts for different aspects:

1. **Activities**: Comprehensive activity recommendations with timing, costs, and tips

2. **Restaurants**: Restaurant recommendations with cuisines, pricing, and local delicacies

3. **Accommodation**: Hotel recommendations based on budget and duration



## Example Usage



### Step 1: Start Background Processing

```bash

# Start processing destinations in background

curl -X POST "http://localhost:8000/travel/process-destinations" \

  -H "Content-Type: application/json" \

  -d @example_request.json

```



**Response:**

```json

{

  "task_id": "550e8400-e29b-41d4-a716-446655440000",

  "status": "processing",

  "message": "Started processing 2 destinations in background",

  "created_at": 1699123456.789,

  "destinations": [...]

}

```



### Step 2: Poll for Status

```bash

# Check processing status

curl "http://localhost:8000/travel/task-status/550e8400-e29b-41d4-a716-446655440000"

```



**Response (while processing):**

```json

{

  "task_id": "550e8400-e29b-41d4-a716-446655440000",

  "status": "processing",

  "message": "Processing Jaipur",

  "destinations": [...]

}

```



**Response (when completed):**

```json

{

  "task_id": "550e8400-e29b-41d4-a716-446655440000",

  "status": "completed",

  "message": "Successfully processed Jaipur",

  "destinations": [

    {

      "place": "Jaipur",

      "processing_status": "completed",

      "activities": { ... },

      "restaurants": { ... },

      "accommodation": { ... }

    }

  ]

}

```



## Response Structure



Each destination response includes:

- **Activities**: Categorized activities with costs, timing, and tips

- **Restaurants**: Restaurant recommendations with cuisines and pricing

- **Accommodation**: Hotel options with amenities and location details

- **Budget Breakdown**: Detailed cost analysis

- **Processing Status**: Real-time status tracking



## Simple & Robust Design



The API uses a simplified approach for maximum reliability:

- AI returns simple objects with 3 arrays (activities, food, accommodations)

- No complex JSON parsing required

- Detailed prompts ensure specific, real recommendations

- Each response contains actual places and restaurants, not generic suggestions

- Background processing with task status polling

- Custom user preferences support through `custom_ins` field

- Comprehensive debugging and error handling



## Debugging



The API includes extensive debugging features to help troubleshoot issues:



### Debug Endpoints

- `GET /travel/test-gemini` - Test if Gemini API is working

- Console logs show detailed processing information

- Background task status updates in real-time



### Debug Information

When processing requests, you'll see console output like:

```

DEBUG: API Key configured: your-key...

DEBUG: Starting background processing for Pune with custom preferences: vegetarian food, historic sites

DEBUG: Task initialized for Pune

DEBUG: Making AI call for activities...

DEBUG: Activities response received, length: 1234

DEBUG: Task status updated for Pune. Current tasks in storage: 1

```



### Troubleshooting



1. **API Key Issues**: Use `GET /travel/test-gemini` to check if your API key is working

2. **Processing Delays**: Check console logs for detailed progress information

3. **Stuck Processing**: The debug logs will show exactly where processing is failing



## Error Handling



The API includes comprehensive error handling for:

- Invalid destinations

- AI processing failures

- Budget constraints

- Network issues

- Custom preference parsing errors



## Technology Stack



- **FastAPI**: Modern Python web framework

- **Google Gemini AI**: Advanced AI for content generation

- **AsyncIO**: Parallel processing capabilities

- **Pydantic**: Data validation and serialization



Frontend repo -  https://github.com/Soham1803/travel-planner-app