microsoft / ai.ed Goto Github PK

In WebServer, rename .gitignore.txt to .gitignore.

Create a test suite

Commit existing code to the repo.

Make a new directory called ai-compile/samples. Move samples like ExampleScript1.py (currently in Symbolic) to there. We can add more samples to this directory in the future.

These versions are stale. Can you update them and see what they can be updated to? Ideally we will want to be able to work in a version higher than Python 3.6.

antlr4-python3-runtime==4.9.2
clang==6.0.0.2
pandas==0.24.2
scikit-learn==0.21.1
keras==2.2.4
prettytable==0.7.2
edlib==1.2.4
tensorflow==1.14
numpy==1.16.2

I don't believe this package is needed: "tsc": "^1.20150623.0".

In the branch tbarik/flask, I've added a simple server.py that runs a Flask server on port 5000. We should modify this to add another app route that calls the PyMacer functionality directly. We should be able to then test this route as an HTTP request using a tool like Insomnia.

This can be done with base Flask, without needing flask-restful.

From looking at Brain.py, this looks like mostly calling repairProgram at the REST endpoint.

Use this to format the vscode-extension code.

I'd like to change the JSON schema so that it does not directly return an array. While an array is technically valid JSON, it causes a number of issues. As a first pass, let's change this to:

[
   {
      "lineNo":1,
      "repairLine":"print ( \"You chose \" , number )",
      "repair_classes":[
         "+ COMMA"
      ],
      "feedbacks":[
         "Seems like there was some error on this line. You seem to be missing some token(s). Try adding - COMMA (',')."
      ]
   }
]

to have an outer element:

{
   "repairs": [ { "lineNo": 1 } ]
}

At the same time, rename feedbacks to feedback and repair_classes to repairClasses.

Add two tests for flask. The first test is test_syntaxerror_request that shows what MACER would return on a program with an syntax error. The second test is test_noerror_request, that shows what MACER would return with no errors.

You can use:

pytest --disable-pytest-warnings .\test_server.py

to run the tests. The --disable-pytest-warnings is there to suppress TensorFlow deprecation warnings.

The branch name is tbarik/unittest.

Perhaps it can go here: in a new branch.

Add the .vs directory in webserver to gitignore. Also delete the existing files from .vs in a PR. I don't believe the entire folder should be checked in.