Contents

Student Usage

Otter provides an IPython API and a command line tool that allow students to run checks and export notebooks within the assignment environment.

The Notebook API

Otter supports in-notebook checks so that students can check their progress when working through assignments via the otter.Notebook class. The Notebook takes one optional parameter that corresponds to the path from the current working directory to the directory of tests; the default for this path is ./tests.

import otter
grader = otter.Notebook()

If my tests were in ./hw00-tests, then I would instantiate with

grader = otter.Notebook("hw00-tests")

Students can run tests in the test directory using Notebook.check which takes in a question identifier (the file name without the .py extension). For example,

grader.check("q1")

will run the test q1.py in the tests directory. If a test passes, then the cell displays “All tests passed!” If the test fails, then the details of the first failing test are printed out, including the test code, expected output, and actual output:

Students can also run all tests in the tests directory at once using Notebook.check_all:

grader.check_all()

This will rerun all tests against the current global environment and display the results for each tests concatenated into a single HTML output. It is recommended that this cell is put at the end of a notebook for students to run before they submit so that students can ensure that there are no variable name collisions, propagating errors, or other things that would cause the autograder to fail a test they should be passing.

Exporting Submissions

Students can also use the Notebook class to generate their own PDFs for manual grading using the method Notebook.export. This function takes an optional argument of the path to the notebook; if unspecified, it will infer the path by trying to read the config file (if present), using the path of the only notebook in the working directory if there is only one, or it will raise an error telling you to provide the path. This method creates a submission zip file that includes the notebook file, the log, and, optionally, a PDF of the notebook (set pdf=False to disable this last).

As an example, if I wanted to export hw01.ipynb with cell filtering, my call would be

grader.export("hw01.ipynb")

as filtering is by defult on. If I instead wanted no filtering, I would use

grader.export("hw01.ipynb", filtering=False)

To generate just a PDF of the notebook, use Notebook.to_pdf.

Running on Non-standard Python Environments

When running on non-standard Python notebook environments (which use their own interpreters, such as Colab or Jupyterlite), some Otter features are disabled due differences in file system access, the unavailability of compatible versions of packages, etc. When you instantiate a Notebook, Otter automatically tries to determine if you’re running on one of these environments, but you can manually indicate which you’re running on by setting either the colab or jupyterlite argument to True.

Command Line Script Checker

Otter also features a command line tool that allows students to run checks on Python files from the command line. otter check takes one required argument, the path to the file that is being checked, and three optional flags:

  • -t is the path to the directory of tests. If left unspecified, it is assumed to be ./tests

  • -q is the identifier of a specific question to check (the file name without the .py extension). If left unspecified, all tests in the tests directory are run.

  • --seed is an optional random seed for execution seeding

The recommended file structure for using the checker is something like the one below:

hw00
├── hw00.py
└── tests
    ├── q1.py
    └── q2.py  # etc.

After a cd into hw00, if I wanted to run the test q2.py, I would run

$ otter check hw00.py -q q2
All tests passed!

In the example above, I passed all of the tests. If I had failed any of them, I would get an output like that below:

$ otter check hw00.py -q q2
1 of 2 tests passed

Tests passed:
    possible


Tests failed:
*********************************************************************
Line 2, in tests/q2.py 0
Failed example:
    1 == 1
Expected:
    False
Got:
    True

To run all tests at once, I would run

$ otter check hw00.py
Tests passed:
    q1  q3  q4  q5


Tests failed:
*********************************************************************
Line 2, in tests/q2.py 0
Failed example:
    1 == 1
Expected:
    False
Got:
    True

As you can see, I passed for of the five tests above, and filed q2.py.

If I instead had the directory structure below (note the new tests directory name)

hw00
├── hw00.py
└── hw00-tests
    ├── q1.py
    └── q2.py  # etc.

then all of my commands would be changed by adding -t hw00-tests to each call. As an example, let’s rerun all of the tests again:

$ otter check hw00.py -t hw00-tests
Tests passed:
    q1  q3  q4  q5


Tests failed:
*********************************************************************
Line 2, in hw00-tests/q2.py 0
Failed example:
    1 == 1
Expected:
    False
Got:
    True

otter.Notebook Reference

class otter.check.notebook.Notebook(nb_path=None, tests_dir='./tests', tests_url_prefix=None, colab=None, jupyterlite=None)

Notebook class for in-notebook autograding

Parameters

  • nb_path (str, optional) – path to the notebook being run

  • tests_dir (str, optional) – path to tests directory

  • colab (bool, optional) – whether this notebook is being run on Google Colab; if None, this information is automatically parsed from IPython on creation

  • jupyterlite (bool, optional) – whether this notebook is being run on JupyterLite; if None, this information is automatically parsed from IPython on creation

add_plugin_files(plugin_name, *args, nb_path=None, **kwargs)

Runs the notebook_export event of the plugin plugin_name and tracks the file paths it returns to be included when calling Notebook.export.

Parameters

  • plugin_name (str) – importable name of an Otter plugin that implements the from_notebook hook

  • *args – arguments to be passed to the plugin

  • nb_path (str, optional) – path to the notebook

  • **kwargs – keyword arguments to be passed to the plugin

check(question, global_env=None)

Runs tests for a specific question against a global environment. If no global environment is provided, the test is run against the calling frame’s environment.

Parameters

  • question (str) – name of question being graded

  • global_env (dict, optional) – global environment resulting from execution of a single notebook

Returns

the grade for the question

Return type

otter.test_files.abstract_test.TestFile

check_all()

Runs all tests on this notebook. Tests are run against the current global environment, so any tests with variable name collisions will fail.

export(nb_path=None, export_path=None, pdf=True, filtering=True, pagebreaks=True, files=[], display_link=True, force_save=False, run_tests=False)

Exports a submission to a zip file. Creates a submission zipfile from a notebook at nb_path, optionally including a PDF export of the notebook and any files in files.

Parameters

  • nb_path (str, optional) – path to the notebook we want to export; will attempt to infer if not provided

  • export_path (str, optional) – path at which to write zipfile; defaults to notebook’s name + .zip

  • pdf (bool, optional) – whether a PDF should be included

  • filtering (bool, optional) – whether the PDF should be filtered; ignored if pdf is False

  • pagebreaks (bool, optional) – whether pagebreaks should be included between questions in the PDF

  • files (list of str, optional) – paths to other files to include in the zip file

  • display_link (bool, optional) – whether or not to display a download link

  • force_save (bool, optional) – whether or not to display JavaScript that force-saves the notebook (only works in Jupyter Notebook classic, not JupyterLab)

classmethod grading_mode(tests_dir)

A context manager for the Notebook grading mode. Yields a pointer to the list of results that will be populated during grading.

It is the caller’s responsibility to maintain the pointer. The pointer in the Checker class will be overwritten when the context exits.

run_plugin(plugin_name, *args, nb_path=None, **kwargs)

Runs the plugin plugin_name with the specified arguments. Use nb_path if the path to the notebook is not configured.

Parameters

  • plugin_name (str) – importable name of an Otter plugin that implements the from_notebook hook

  • *args – arguments to be passed to the plugin

  • nb_path (str, optional) – path to the notebook

  • **kwargs – keyword arguments to be passed to the plugin

to_pdf(nb_path=None, filtering=True, pagebreaks=True, display_link=True, force_save=False)

Exports a notebook to a PDF using Otter Export

Parameters

  • nb_path (str, optional) – path to the notebook we want to export; will attempt to infer if not provided

  • filtering (bool, optional) – set true if only exporting a subset of notebook cells to PDF

  • pagebreaks (bool, optional) – if true, pagebreaks are included between questions

  • display_link (bool, optional) – whether or not to display a download link

  • force_save (bool, optional) – whether or not to display JavaScript that force-saves the notebook (only works in Jupyter Notebook classic, not JupyterLab)

previous

Usage and Output

next

Otter Configuration Files