forge-tools/frg/context.py

"""
Repo/git-aware contextual data

This module takes care of anything that needs to be pulled from the
invocation environment (i.e. git context, ...) to make things work.
"""

import pydantic

import logging
import subprocess

logger = logging.getLogger(__name__)


class GitContext(pydantic.BaseModel):
    """
    Contextual data based on where the invocation
    to the tool is made.
    """

    repo_name: str
    repo_author: str
    current_branch: str
    host: str


def parse_remote_url(remote_url: str) -> tuple[str, str, str]:
    """Parses the remote URL attached to a repository to detect the host, repo owner and name."""
    if "@" in remote_url:
        _, url = remote_url.removesuffix(".git").split("@")
        host, full_repo_name = url.split(":")
        owner, repo = full_repo_name.split("/")

        return ("https://" + host, owner, repo)

    if remote_url.startswith("http"):
        url = (
            remote_url.removeprefix("http://")
            .removeprefix("https://")
            .removesuffix(".git")
        )
        url_parts = url.split("/")

        repo = url_parts[-1]
        owner = url_parts[-2]
        host = "https://" + "/".join(url_parts[:-2])

        return (host, owner, repo)

    return ("", "", "")


def get_git_context(*, domain_aliases: dict[str, str] | None = None) -> GitContext:
    """Gathers contextual data from the environment."""
    if not domain_aliases:
        domain_aliases = dict()

    current_branch_cmd = subprocess.run(
        ["git", "branch", "--show-current"], capture_output=True
    )
    current_branch = current_branch_cmd.stdout.decode("utf8").strip()

    remote_url_cmd = subprocess.run(
        ["git", "config", "--get", "remote.origin.url"], capture_output=True
    )
    remote_url = remote_url_cmd.stdout.decode("utf8").strip()

    host, owner, repo = parse_remote_url(remote_url)

    if host in domain_aliases:
        logger.info(f"Remapped domain: {host} -> {domain_aliases[host]}")
        host = domain_aliases[host]

    return GitContext(
        repo_name=repo, repo_author=owner, host=host, current_branch=current_branch
    )