Marking student's code is difficult and time consuming, and while there are many code judges online, they are all closed and often difficult to utilise.
PyMark was developed as a response to this. PyMark is (supposedly) easy to setup and is completely open source.
PyMark is coded in Python and is currently only capable of running Python code.
PyMark allows you to create your own programming and informatics problems, similar in format to those at the IOI. PyMark should be easy to set up.
PyMark is not intended to be a platform for running competitions.
PyMark runs in python, and uses mongodb for storing data on it's back end. It requires the python modules bottle and pymongo, which can be installed by running pip.
PyMark has some very limited sandboxing features, which will hopefully prevent people doing anything stupid. However, there are well known methods of escaping the sandbox.
Feel free to host your own database.
PyMark has been tested with mongolab, and it works. Make an account, then create a (free) database, and add the following collections:
accountscodesessionssubmissions
You'll then need to create a user, then connect PyMark to the database with the URI, which will look something like this: mongodb://<dbuser>:<dbpassword>@ds031213.mongolab.com:31213/<dbname>
Using the command line, navigate to a folder in which you wish to install PyMark, and run git clone https://github.com/DXsmiley/PyMark.git. Then naviate to the folder where you wish to store your data files. From there, run python ../path_to_pymark/server.py. PyMark will automatically create a number of files and folders. You should then be able to access PyMark on your machine from localhost:8080.
If you open the newly created settings.json file, you will be able to specify the following parameters.
- "invoke_code": The command line argument used to run python code. Use
{}to specify the place for the code filename. - "invoke_shell": Either 0 or 1. Use to specify whether to gain access to the shell environment. Exact results will vary depending on system.
- "db_login": The login URI used for the database.
- "db_name": The name of the database.
- "port": The port on which the server should listen. 8080 by default.
These parameters can also be set through environment variables, which have the same name, but shuld be prefixed by pymark_.
Do this manually, though the database.
After logging in to the admin (or a tutor) account, go to the page /problem_new. It will have a number of fields you have to fill out.
The first attribute is the 'idenifier', or short name. This is an identifier unique to the problem. The problem will be placed at /statement/identifier. This is usually the same as the problem name, but in all lowercase and without any whitespace or special symbols.
The name if the problem is just that - its name. This will be shown on the problem listing page.
The problem statement is a description of what the student should do. You can write HTML code, using the paragraph and headder tags for formatting.
To begin, create a folder in the directory 'problems'. Are create a the following files settings.json and statement.html within it.
Finally, you specify the test data.
Example test data.
[
{
"name": "Subtask 1",
"weight": 30,
"cases": [
{
"input": [
"1 3 2",
"8 3 7 2 49 3"
],
"output": [
"2 1",
"7"
],
"show result": true
}
]
},
{
"name": "Subtask 2",
"weight": 70,
"cases": [
{
"input": [
"4 70 4",
"1 76 34 865 382 12 53"
],
"output": [
"4 8",
"30"
]
}
]
}
]
Note that the "show result" flag can be used to show the user what their code produced, the correct output and the input.
**Warning**
"show result is currently not functional"
You can send me a message via GitHub, or through my email dxsmiley@hotmail.com.
Alternately, you might be able to find what your looking for by poking around in the code, although it's not well commented.