generate_full_changelog.py

from subprocess import getoutput
import os
from datetime import datetime
from github_activity import generate_activity_md
import re
import sys
from ghapi.core import GhApi

name = sys.argv[-1]
os.chdir(name)

getoutput('git fetch upstream --tags')

url = getoutput('git remote get-url upstream')
_, target = url.replace('.git', '').split(':')

data = getoutput('git log --tags --simplify-by-decoration --pretty="format:%h | %D"').splitlines()

pattern = 'tag: (v?\d+\.\d+\.\d+)$'


def filter(datum):
    sha, tag = datum
    return re.match(pattern, tag) is not None

data = [d.split(' | ') for (i, d) in enumerate(data)]
data = [d for d in data if filter(d)]

print(data, file=sys.stderr)

output = ''
auth = os.environ['GITHUB_ACCESS_TOKEN']


def format_pr_entry(target, number):
    """Format a PR entry in the style used by our changelogs.

    Parameters
    ----------
    target : str
        The GitHub owner/repo
    number : int
        The PR number to resolve
    auth : str, optional
        The GitHub authorization token

    Returns
    -------
    str
        A formatted PR entry
    """
    owner, repo = target.split("/")
    gh = GhApi(owner=owner, repo=repo, token=auth)
    pull = gh.pulls.get(number)
    title = pull.title
    url = pull.url
    user_name = pull.user.login
    user_url = pull.user.html_url
    return f"- {title} [#{number}]({url}) [@{user_name}]({user_url})"


for i in range(len(data) - 1):
    curr_data = data[i]
    prev_data = data[i + 1]

    since = prev_data[0]
    until = curr_data[0]

    match = re.search(pattern, curr_data[1])
    tag = match.groups()[0]

    print(f'\n({i + 1}/{len(data)})', since, until, tag, file=sys.stderr)
    md = generate_activity_md(
            target,
            since=since,
            heading_level=2,
            until=until,
            auth=auth,
            kind="pr"
        )

    if not md:
        continue

    entry = md.replace("[full changelog]", "[Full Changelog]")

    entry = entry.splitlines()[2:]

    for (ind, line) in enumerate(entry):
        if re.search(r"\[@meeseeksmachine\]", line) is not None:
            match = re.search(r"Backport PR #(\d+)", line)
            if match:
                print('resolving', match.groups()[0], file=sys.stderr)
                entry[ind] = format_pr_entry(target, match.groups()[0])

    entry = "\n".join(entry).strip()

    output += f"""
## {tag}

{entry}
"""

print(output)