abra/bin/app-json.py

#!/usr/bin/env python3

# Usage: ./app-json.py
#
# Gather metadata from Co-op Cloud apps in $ABRA_DIR/apps (default
# ~/.abra/apps), and format it as JSON so that it can be hosted here:
# https://abra-apps.cloud.autonomic.zone

from json import dump
from logging import DEBUG, basicConfig, getLogger
from os import chdir, listdir, mkdir
from os.path import basename, exists, expanduser
from pathlib import Path
from re import findall, search
from shlex import split
from subprocess import check_output
from sys import exit

from requests import get

HOME_PATH = expanduser("~/")
CLONES_PATH = Path(f"{HOME_PATH}/.abra/apps").absolute()
YQ_PATH = Path(f"{HOME_PATH}/.abra/vendor/yq")
SCRIPT_PATH = Path(__file__).absolute().parent
REPOS_TO_SKIP = (
    "abra",
    "backup-bot",
    "cloud.autonomic.zone",
    "docs.cloud.autonomic.zone",
    "example",
    "organising",
    "pyabra",
    "stack-ssh-deploy",
)

log = getLogger(__name__)
basicConfig()
log.setLevel(DEBUG)


def _run_cmd(cmd, shell=False):
    """Run a shell command."""
    args = [split(cmd)]
    kwargs = {}

    if shell:
        args = [cmd]
        kwargs = {"shell": shell}

    try:
        return check_output(*args, **kwargs).decode("utf-8").strip()
    except Exception as exception:
        log.error(f"Failed to run {cmd}, saw {str(exception)}")
        exit(1)


def get_published_apps_json():
    """Retrieve already published apps json."""
    url = "https://abra-apps.cloud.autonomic.zone"

    log.info(f"Retrieving {url}")

    try:
        return get(url, timeout=5).json()
    except Exception as exception:
        log.error(f"Failed to retrieve {url}, saw {str(exception)}")
        return {}


def clone_all_apps():
    """Clone all Co-op Cloud apps to ~/.abra/apps."""
    if not exists(CLONES_PATH):
        mkdir(CLONES_PATH)
    url = "https://git.autonomic.zone/api/v1/orgs/coop-cloud/repos"

    log.info(f"Retrieving {url}")

    try:
        response = get(url, timeout=10)
    except Exception as exception:
        log.error(f"Failed to retrieve {url}, saw {str(exception)}")
        exit(1)

    repos = [[p["name"], p["ssh_url"]] for p in response.json()]

    for name, url in repos:
        if name in REPOS_TO_SKIP:
            continue

        if not exists(f"{CLONES_PATH}/{name}"):
            _run_cmd(f"git clone {url} {CLONES_PATH}/{name}")

            chdir(f"{CLONES_PATH}/{name}")
            if not int(_run_cmd("git branch --list | wc -l", shell=True)):
                log.info(f"Guessing main branch is HEAD for {name}")
                _run_cmd("git checkout main")
        else:
            _run_cmd("git fetch -a")


def generate_apps_json():
    """Generate the abra-apps.json application versions file."""
    apps_json = {}
    cached_apps_json = get_published_apps_json()

    for app in listdir(CLONES_PATH):
        if app in REPOS_TO_SKIP:
            log.info(f"Skipping {app}")
            continue

        app_path = f"{CLONES_PATH}/{app}"
        chdir(app_path)

        log.info(f"Processing {app}")
        apps_json[app] = {
            "category": "apps",
            "repository": f"https://git.autonomic.zone/coop-cloud/{app}.git",
            # Note(decentral1se): please note that the app features do not
            # correspond to version tags. We simply parse the latest features
            # list from HEAD. This may lead to unexpected situations where
            # users believe X feature is available under Y version but it is
            # not.
            "features": get_app_features(app_path),
            "versions": get_app_versions(app_path, cached_apps_json),
        }

    return apps_json


def get_app_features(app_path):
    """Parse features from app repo README files."""
    features = {}

    chdir(app_path)

    with open(f"{app_path}/README.md", "r") as handle:
        log.info(f"{app_path}/README.md")
        contents = handle.read()

    try:
        for match in findall(r"\*\*.*\s\*", contents):
            title = search(r"(?<=\*\*).*(?=\*\*)", match).group().lower()

            if title == "image":
                value = {
                    "image": search(r"(?<=`).*(?=`)", match).group(),
                    "url": search(r"(?<=\().*(?=\))", match).group(),
                    "rating": match.split(",")[1].strip(),
                    "source": match.split(",")[-1].replace("*", "").strip(),
                }
            else:
                value = match.split(":")[-1].replace("*", "").strip()

            features[title] = value
    except (IndexError, AttributeError):
        log.info(f"Can't parse {app_path}/README.md")
        return {}
    finally:
        _run_cmd("git checkout HEAD")

    log.info(f"Parsed {features}")

    return features


def get_app_versions(app_path, cached_apps_json):
    versions = {}

    chdir(app_path)

    tags = _run_cmd("git tag --list").split()

    if not tags:
        log.info(f"No tags discovered, moving on")
        return {}

    initial_branch = _run_cmd("git rev-parse --abbrev-ref HEAD")

    app_name = basename(app_path)

    try:
        existing_tags = cached_apps_json[app_name]["versions"].keys()
    except KeyError:
        existing_tags = []

    for tag in tags:
        _run_cmd(f"git checkout {tag}")

        services_cmd = f"{YQ_PATH} e '.services | keys | .[]' compose*.yml"
        services = _run_cmd(services_cmd, shell=True).split()

        parsed_services = []
        service_versions = {}
        for service in services:
            if service in ("null", "---"):
                continue

            if tag in existing_tags:
                log.info(f"Skipping {tag} because we've already processed it")
                existing_versions = cached_apps_json[app_name]["versions"][tag][service]
                service_versions[service] = existing_versions
                _run_cmd(f"git checkout {initial_branch}")
                continue

            if service in parsed_services:
                log.info(f"Skipped {service} asa we've already parsed it locally")
                continue

            services_cmd = f"{YQ_PATH} e '.services.{service}.image' compose*.yml"
            images = _run_cmd(services_cmd, shell=True).split()

            for image in images:
                if image in ("null", "---"):
                    continue

                images_cmd = f"skopeo inspect docker://{image} | jq '.Digest'"
                output = _run_cmd(images_cmd, shell=True)

                service_version_info = {
                    "image": image.split(":")[0],
                    "tag": image.split(":")[-1],
                    "digest": output.split(":")[-1][:8],
                }

                log.info(f"Parsed {service_version_info}")
                service_versions[service] = service_version_info

            parsed_services.append(service)

        versions[tag] = service_versions

    _run_cmd(f"git checkout {initial_branch}")

    return versions


def main():
    """Run the script."""
    clone_all_apps()

    target = f"{SCRIPT_PATH}/../deploy/abra-apps.cloud.autonomic.zone/abra-apps.json"
    with open(target, "w", encoding="utf-8") as handle:
        dump(generate_apps_json(), handle, ensure_ascii=False, indent=4)

    log.info(f"Successfully generated {target}")


main()