ClickHouse/utils/test_history/test-history

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

# Note: should work with python 2 and 3

import argparse
import datetime
import sys

import tabulate
from github import Github
from termcolor import colored

COLORMAP = {
    "success": colored("success", "green"),
    "failure": colored("failure", "red"),
    "error": colored("error", "red"),
    "pending": colored("pending", "yellow"),
    "not run": colored("not run", "white"),
}


def _filter_statuses(statuses):
    """
    Squash statuses to latest state
    1. context="first", state="success", update_time=1
    2. context="second", state="success", update_time=2
    3. context="first", stat="failure", update_time=3
    =========>
    1. context="second", state="success"
    2. context="first", stat="failure"
    """
    filt = {}
    for status in sorted(statuses, key=lambda x: x.updated_at):
        filt[status.context] = status
    return list(filt.values())


def get_filtered_statuses(commit):
    return _filter_statuses(commit.get_statuses())


def get_commits(repo, since):
    return sorted(repo.get_commits(since=since), key=lambda x: x.commit.author.date)


def process_one_commit(commit):
    commit_statuses = get_filtered_statuses(commit)

    # very dirty, but don't require additional dependencies
    commit_modified = commit.commit.author.date + datetime.timedelta(hours=3)
    commit_sha = commit.sha
    checks_result = {}
    for commit_status in commit_statuses:
        state = commit_status.state
        check_name = commit_status.context
        checks_result[check_name] = state

    return commit_sha, commit_modified, checks_result


if __name__ == "__main__":
    three_days_ago = datetime.datetime.now() - datetime.timedelta(days=3)
    parser = argparse.ArgumentParser("ClickHouse commits history parser")
    parser.add_argument("--token", required=True)
    parser.add_argument("--since", default=three_days_ago.strftime("%Y-%m-%d %H:%M:%S"))
    parser.add_argument("--substr", default="Functional stateful")

    args = parser.parse_args()

    date_since = datetime.datetime.strptime(args.since, "%Y-%m-%d %H:%M:%S")

    gh = Github(args.token)
    repo = gh.get_repo("ClickHouse/ClickHouse")
    commits = get_commits(repo, date_since)

    longest_header = []
    all_data = []
    for num, commit in enumerate(commits):
        commit_sha, commit_modified, check_results = process_one_commit(commit)
        mapped_keys = [key for key in list(check_results.keys()) if args.substr in key]
        if len(mapped_keys) > len(longest_header):
            longest_header = mapped_keys
        all_data.append((commit_modified, commit_sha, check_results))
        if (num + 1) % 10 == 0:
            print("Processed", num + 1, "commits")

    longest_header = ["Date", "SHA"] + longest_header

    result_data = []
    for row in all_data:
        current_result = [row[0].strftime("%Y-%m-%d %H:%M:%S"), row[1][0:7]]
        for check_name in longest_header[2:]:
            if check_name in row[2]:
                check_result = row[2][check_name]
            else:
                check_result = "not run"

            if sys.stdout.isatty():
                current_result.append(COLORMAP[check_result])
            else:
                current_result.append(check_result)
        result_data.append(current_result)

    if sys.stdout.isatty():
        longest_header = [colored(h, "white", attrs=["bold"]) for h in longest_header]

    print(tabulate.tabulate(result_data, headers=longest_header, tablefmt="grid"))