statistics.py

#!/usr/bin/env python3

import argparse
import os
import sys
import typing

from lib.detector.common import *

# Will be initialized by the ArgumentParser
ARGS: argparse.Namespace = None

def init_argument_parser() -> None:
    parser = argparse.ArgumentParser()

    parser.add_argument("--report-directory", type=str, default="reports", help="""
                         A directory path to load the reports from
                        [default: reports]""")

    parser.add_argument("--qmeta-file", type=str, default="qmeta.csv", help="""
                         A file to store the query metadata in
                        [default: qmeta.csv]""")

    parser.add_argument("--print-args", action="store_true", help="""
                         Print the given command line arguments to stderr
                        [default: disabled]""")

    global ARGS
    ARGS = parser.parse_args()

    if ARGS.print_args:
        print(ARGS, file=sys.stderr)

def main() -> None:
    init_argument_parser()

    repositories = get_repositories(ARGS.report_directory)

    qmeta_dict = {}
    languages_dict = {}
    global_min, global_max = 2e64, 0
    global_repo_count = 0

    with open(ARGS.qmeta_file, "r", encoding="utf-8") as f:
        for line in sorted(f.readlines())[::-1]:
            repository, primary_language, stars_count = line.split(",")

            if repository not in repositories:
                continue
            if repository in qmeta_dict:
                continue

            global_repo_count += 1
            stars_count = int(stars_count)

            qmeta_dict[repository] = (primary_language, stars_count)

            repo_count, old_min, old_max = languages_dict.get(primary_language, (0, 2e64, 0))
            languages_dict[primary_language] = (repo_count + 1,
                                                min(old_min, stars_count),
                                                max(old_max, stars_count))

            global_min, global_max = min(global_min, stars_count), max(global_max, stars_count)

    print("| Language             | Repositories |           Stars |")
    print("|======================|==============|=================|")

    for language, (repo_count, lang_min, lang_max) in sorted(languages_dict.items(),
                                                             key=lambda kv: kv[0],
                                                             reverse=False):
        print(f"| {language:20} | {repo_count:12d} | {lang_min:6d} - {lang_max:6d} |")

    print("|======================|==============|=================|")
    print(f"| Global               | {global_repo_count:12d} | {global_min:6d} - {global_max:6d} |")


if __name__ == "__main__":
    main()