Script for checking version numbering in test files for exercism/python

check_test_versions.py

import json
import os
import re


# There are probably others that don't require canonical data
NO_DATA = {"beer-song", "house", "robot-name", "zebra-puzzle", "twelve-days"}



def check_versions():
    status = {"up-to-date": [], "up-to-date (x-common)": [],
              "out-of-date": {}, "unknown": [], "unimplemented": [],
              "no-canonical-data": [], "no-data-required": [],
              "deprecated": []}
    avail_exercises = set(os.listdir("./problem-specifications/exercises"))
    impl_exercises = set(os.listdir("./python/exercises"))

    status["unimplemented"] = sorted(avail_exercises - impl_exercises)

    for exercise in sorted(impl_exercises):
        if exercise in NO_DATA:
            status["no-data-required"].append(exercise)
            continue
        if is_deprecated(exercise):
            status["deprecated"].append(exercise)
            continue

        test_file, data_file = get_filepaths(exercise)
        test_version, version_string  = get_version(test_file)
        try:
            data_version = get_canonical_version(data_file)
        except FileNotFoundError:
            status["no-canonical-data"].append(exercise)
            continue

        if test_version == data_version:
            if "x-common" in version_string:
                status["up-to-date (x-common)"].append(exercise)
            else:
                status["up-to-date"].append(exercise)
        elif not test_version or not data_version:
            status["unknown"].append(exercise)
        else:
            status["out-of-date"].update({exercise: (test_version,
                                                     data_version)})

    with open("version_status.json", "w") as outfile:
        json.dump(status, outfile, indent=2)


def get_filepaths(exercise):
    """Return a tuple of the filepaths for *_test.py and canonical-data.json"""
    test_file = os.path.join("./python/exercises",
                             exercise,
                             exercise.replace("-", "_") + "_test.py")
    data_file = os.path.join("./problem-specifications/exercises",
                             exercise,
                             "canonical-data.json")
    return (test_file, data_file)


def get_version(test_file):
    """Return a tuple of (version, version_string)"""
    version = ""
    version_string = ""
    with open(test_file, "r") as f:
        for line in f:
            if line.startswith("#"):
                version = parse_version(line)
            if version:
                version_string = line
                break
    return (version, version_string)


def parse_version(line):
    """Parse a possible version string, return None if no version found"""
    version = re.findall("[0-9][.][0-9][.][0-9]", line)
    return version[-1] if version else ""


def get_canonical_version(data_file):
    """Get the current version from the canonical-data file"""
    with open(data_file, "r") as data:
        return json.load(data)["version"]


def is_deprecated(exercise):
    with open("./python/config.json", "r") as config:
        data = json.load(config)
        ex_data = [ex for ex in data["exercises"] if ex["slug"] == exercise]
        try:
            deprecated = ex_data[0]["deprecated"]
        except KeyError:
            deprecated = False
        except IndexError:
            print("{} not found in config.json".format(exercise))
            deprecated = False

        return deprecated


if __name__ == '__main__':
    check_versions()