MathLearningMachine

This repo holds the backend application code for MathLearningMachine.

Contributing

For new feature development, branch of development with a named feature branch. To merge back into development, use a pull request. This way, we can get integrated unit and acceptance test reports in github (once we write them).

We'll use Flask micro web framework for this application.

Other technologies may or may not include:

Hypothesis and PyTest for testing
Pylint for static analysis

and who knows what else.

Tests

The test infrastructure will use PyTest for discovery and running and Hypothesis for property and model based testing.

Tools

VS Code is recommended for development. Install the "Remote - Containers" extension, the "Live Share" extension, and the Python extension, all by Microsoft.

Developing in VS Code

Clone the repo from Github and switch to your branch.
Open the project folder in VS Code.
Before the first launch, Open the terminal and run docker-compose build to build the image.
Ctrl+Shift+P to open the command palatte (or Click the green icon at the bottom left of the window) and run "Remote-Containers: Open in Container".

The project will open in a docker container, with VS Code attached.

Your git credentials will be shared from your local host to the container, so git will work as expected and you can commit and push from inside the container.

Open the integrated terminal to run commands in the container.

Running Application

See Docker Usage for Docker-specific instructions.

$ flask run will start the flask development server and send requests to the flask application.

Running Tests

To run the tests:

$ docker-compose build
$ docker-compose run flask bash -c "cd /app/flask/flaskapp && python -m pytest"

Public API

Solve Image

URI: /solve-image

Method: POST

Request Body:

{
    "b64_img": (str) "BASE64ENCODEDIMAGE"
}

Response Body:

{
    "confidence": (number) CONFIDENCE_VALUE,
    "input_detected": (str) "LATEX_STRING_OF_INPUT_DETECTED",
    "solved": (str) "LATEX_STRING_OF_CAS_SOLUTION_TO_PROBLEM",
    "detections": (object) ALL_DETECTED_OBJECTS_IN_THE_IMAGE,
}

Example Response

{
  "confidence": 0.46090906455275005,
  "detections": [
    {
      "box": {
        "endX": "226",
        "endY": "248",
        "startX": "85",
        "startY": "89"
      },
      "confidence": "0.7222325",
      "label": "\\mathcal{F}"
    },
    {
      "box": {
        "endX": "315",
        "endY": "263",
        "startX": "263",
        "startY": "80"
      },
      "confidence": "0.6381727",
      "label": "\\pm"
    }
  ],
  "input_detected": "\\mathcal{F}\\pm",
  "solved": "F \\mathcal{math} pm "
}

Solve Latex

URI: /solve-latex