🧠 CAPTCHA Solver API (FastAPI + Docker)

A simple, containerized FastAPI service that uses a trained SVM model to solve 6‑character uppercase CAPTCHA images.

---

📁 Project Structure

unlock/
├── datasource/
│   └── zeta/
│       └── run.py          # decrypt logic for 'zeta' model
├── routes/
│   └── decripto.py         # API route definitions
├── schemas/
│   └── decripto.py         # Pydantic request schemas
├── services/
│   └── decripto_service.py # dynamic loader for run.py
├── resources/
│   └── tmp/                # saved uploads, renamed as <decoded>.png
├── postman_decripto.json   # Postman collection for /decripto
├── main.py                 # FastAPI app entrypoint
├── docker-compose.yml
└── IaC/
    ├── fastAPI/
    │   └── Dockerfile      # FastAPI service image
    ├── python/
    │   └── Dockerfile      # Python helper container image
    └── requirements.txt    # all Python dependencies

---

🚀 Quick Start

1. Build & run all services:

   docker compose up --build

2. API docs: Visit http://localhost:8000/docs for the interactive Swagger UI.

3. MongoDB ping: GET http://localhost:8000/ping-db

---

🔗 /decripto Endpoint

POST /decripto Accepts JSON with one of:

• `` (Base64 payload)
• `` (link to Base64 string or PNG image)

Field	Type	Description
modelo	string	Model folder name under datasource/ (e.g. "zeta")
type	string	"base64" or "image"
content	string	Optional Base64 string (data:image/png;base64,...)
url	string	Optional URL returning Base64 or PNG image

Behavior

• If `` is provided, it is used directly.

• If `` is provided:

  • type="base64" → response body text treated as Base64.
  • type="image" → response bytes converted to Base64 automatically.

Success response:

{ "resultado": "ABCXYZ" }

---

📦 Postman Collection

Import the postman_decripto.json file in Postman:

1. File → Import
2. Select postman_decripto.json
3. Run the provided examples for each scenario.

---

🔧 Tech Stack

• Python 3 & FastAPI
• Docker & docker-compose
• Machine Learning: scikit‑learn, scikit‑image, joblib
• CV: opencv‑python‑headless, numpy

---

📚 References

• FastAPI: https://fastapi.tiangolo.com/
• Pydantic: https://docs.pydantic.dev/
• Docker Compose: https://docs.docker.com/compose/
• scikit‑image HOG: https://scikit-image.org/