work-helpers/work_helpers/sg_frm_list.py

import os
import re

from pathlib import Path

import click
import gitea


GITEA_URL = "https://git.cpi.imtek.uni-freiburg.de"
GITEA_OWNER = "Safeguard"
GITEA_REPO = "MBP-issues"

DIR_MBP_WORKBOOKS = "/mnt/e/Safeguard MBP Workbooks"
FILE_CHANGELOG = "/mnt/e/Safeguard-MBP-issues/CHANGELOG.txt"
FILE_OUTPUT = "List of FRMs.txt"

RE_ISSUE_CHANGES = re.compile("#(\d+)")
RE_ISSUE_FILENAME = re.compile("Issue (\d+)", re.IGNORECASE)

WORKBOOK_TYPES = ["hyb", "regen", "dry", "qc"]


def connect_to_gitea(url=GITEA_URL):
    token = os.getenv("GITEA_SG_FRM_TOKEN")
    return gitea.Gitea(url, token)


def get_gitea_issues(url=GITEA_URL, owner=GITEA_OWNER, repo=GITEA_REPO):
    client = connect_to_gitea(url)
    repo = gitea.Repository.request(client, owner, repo)
    return repo.get_issues_state(gitea.Issue.CLOSED)


def get_issue_numbers_from_changelog(filepath=FILE_CHANGELOG):
    with Path(FILE_CHANGELOG).open("r") as fh:
        # remove starting lines
        for line in fh:
            if line.startswith("v3"):
                break

        # this are now the changes, until a blank line occures
        issues = set()
        for line in fh:
            if not line.strip():
                break
            results = (r.group(1) for r in RE_ISSUE_CHANGES.finditer(line))
            # hash_stripped = (issue[1:] for issue in results)
            issues.update(int(issue) for issue in results)
        return issues


def _issue_and_type_from_file_path(file_path):
    raw_issue = RE_ISSUE_FILENAME.search(file_path.stem)
    issue = int(raw_issue.group(1))
    rest, raw_kind = file_path.stem.rsplit(" ", 1)
    kind = raw_kind.lower()
    if kind == "reg":
        kind = "regen"
    if kind not in WORKBOOK_TYPES:
        raise ValueError(f"Unknown workbook type: {file_path.name}")
    return issue, kind


def get_frms(issues, workbook_dir=DIR_MBP_WORKBOOKS):
    directory = Path(DIR_MBP_WORKBOOKS)
    search_result = directory.glob("**/*.docx")
    visible = (i for i in search_result if not i.stem.startswith("."))
    file_paths = (i for i in visible if i.is_file())
    result = {}
    for file_path in file_paths:
        if file_path.parent == directory:
            continue
        try:
            issue, kind = _issue_and_type_from_file_path(file_path)
        except ValueError():
            continue
        if issue in issues:
            item = result.setdefault(issue, dict.fromkeys(WORKBOOK_TYPES, ""))
            item[kind] = file_path.name
    return result


def process(
    url=GITEA_URL,
    owner=GITEA_OWNER,
    repo=GITEA_REPO,
    changes=FILE_CHANGELOG,
    frm_dir=DIR_MBP_WORKBOOKS,
):
    gitea_issues = get_gitea_issues(url, owner, repo)
    closed_issues = {i.number: i.title for i in gitea_issues}
    change_issues = get_issue_numbers_from_changelog(changes)
    frm_issues = get_frms(change_issues)
    for issue, info in frm_issues.items():
        info["title"] = closed_issues.get(issue, "")
    return frm_issues


def write(frms, output_path=FILE_OUTPUT):
    with Path(output_path).open("w") as fh:
        for issue in sorted(frms.keys()):
            frm = frms[issue]
            title = frm["title"]
            fh.write(f"Issue {issue}: {title}\n")
            for kind in WORKBOOK_TYPES:
                frm_file_name = frm[kind]
                if frm_file_name:
                    fh.write(f" - {frm_file_name}\n")
            fh.write("\n")


@click.command()
@click.option("-o", "--output", default=FILE_OUTPUT)
def cli(output=FILE_OUTPUT):
    frms = process()
    write(frms, output)