Config generation

As a temporary solution for vcspull not being able to generate Configuration through scanning directories or fetching them via API (e.g. gitlab, github, etc), you can write scripts to generate configs in the mean time.

Collect repos from Gitlab

Contributed by Andreas Schleifer (a.schleifer@bigpoint.net)

Limitation on both, no pagination support in either, so only returns the first page of repos (as of Feb 26th this is 100).

Shell-script

Requires jq and curl.

#!/usr/bin/env bash

if [ -z "${GITLAB_TOKEN}" ]; then
    echo 'Please provide the environment variable $GITLAB_TOKEN'
    exit 1
fi

if [ $# -lt 2 ]; then
    echo "Usage: $0 <gitlab_host> <gitlab_namespace> [</path/to/target.config>]"
    exit 1
fi

prefix="$(pwd)"
gitlab_host="${1}"
namespace="${2}"
config_file="${3:-./vcspull.yaml}"

current_namespace_path=""

curl --silent --show-error --header "Authorization: Bearer ${GITLAB_TOKEN}" "https://${gitlab_host}/api/v4/groups/${namespace}/projects?include_subgroups=true&per_page=100" \
    | jq -r '.[]|.namespace.full_path + " " + .path' \
    | LC_ALL=C sort \
    | while read namespace_path reponame; do
        if [ "${current_namespace_path}" != "${namespace_path}" ]; then
            current_namespace_path="${namespace_path}"

            echo "${prefix}/${current_namespace_path}:"
        fi

        # simplified config not working - https://github.com/vcs-python/vcspull/issues/332
        #echo "  ${reponame}: 'git+ssh://git@${gitlab_host}/${current_namespace_path}/${reponame}.git'"

        echo "  ${reponame}:"
        echo "    url: 'git+ssh://git@${gitlab_host}/${current_namespace_path}/${reponame}.git'"
        echo "    remotes:"
        echo "      origin: 'ssh://git@${gitlab_host}/${current_namespace_path}/${reponame}.git'"
    done \
   | tee "${config_file}"

$ env GITLAB_TOKEN=mySecretToken \
  /path/to/generate_gitlab.sh gitlab.mycompany.com desired_namespace

To be executed from the path where the repos should later be stored. It will use the current working directory as a “prefix” for the path used in the new config file.

Optional: Set config file output path as additional argument (will overwrite)

$ env GITLAB_TOKEN=mySecretToken \
  /path/to/generate_gitlab.sh gitlab.mycompany.com desired_namespace /path/to/config.yaml

Demonstration

Assume current directory of /home/user/workspace/ and script at /home/user/workspace/scripts/generate_gitlab.sh:

$ ./scripts/generate_gitlab.sh gitlab.com vcs-python

New file vcspull.yaml:

/my/workspace/:
  g:
    url: "git+ssh://[email protected]/vcs-python/g.git"
    remotes:
      origin: "ssh://[email protected]/vcs-python/g.git"
  libvcs:
    url: "git+ssh://[email protected]/vcs-python/libvcs.git"
    remotes:
      origin: "ssh://[email protected]/vcs-python/libvcs.git"
  vcspull:
    url: "git+ssh://[email protected]/vcs-python/vcspull.git"
    remotes:
      origin: "ssh://[email protected]/vcs-python/vcspull.git"

Python

Requires requests and pyyaml.

This confirms file overwrite, if already exists. It also requires passing the protocol/schema of the gitlab mirror, e.g. https://gitlab.com instead of gitlab.com.

#!/usr/bin/env python
"""Example script for export gitlab organization to vcspull config file."""

import argparse
import logging
import os
import pathlib
import sys

import requests
import yaml
from libvcs.sync.git import GitRemote

from vcspull.cli.sync import CouldNotGuessVCSFromURL, guess_vcs
from vcspull.types import RawConfig

log = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO, format="%(message)s")

try:
    gitlab_token = os.environ["GITLAB_TOKEN"]
except KeyError:
    log.info("Please provide the environment variable GITLAB_TOKEN")
    sys.exit(1)

parser = argparse.ArgumentParser(
    description="Script to generate vcsconfig for all repositories \
    under the given namespace (needs Gitlab >= 10.3)",
)
parser.add_argument("gitlab_host", type=str, help="url to the gitlab instance")
parser.add_argument(
    "gitlab_namespace",
    type=str,
    help="namespace/group in gitlab to generate vcsconfig for",
)
parser.add_argument(
    "-c",
    type=str,
    help="path to the target config file (default: ./vcspull.yaml)",
    dest="config_file_name",
    required=False,
    default="./vcspull.yaml",
)

args = vars(parser.parse_args())
gitlab_host = args["gitlab_host"]
gitlab_namespace = args["gitlab_namespace"]
config_filename = pathlib.Path(args["config_file_name"])

try:
    if config_filename.is_file():
        result = input(
            f"The target config file ({config_filename}) already exists, \
            do you want to overwrite it? [y/N] ",
        )

        if result != "y":
            log.info(
                f"Aborting per user request as existing config file ({config_filename})"
                + " should not be overwritten!",
            )
            sys.exit(0)

    config_file = config_filename.open(mode="w")
except OSError:
    log.info(f"File {config_filename} not accessible")
    sys.exit(1)

response = requests.get(
    f"{gitlab_host}/api/v4/groups/{gitlab_namespace}/projects",
    params={"include_subgroups": "true", "per_page": "100"},
    headers={"Authorization": f"Bearer {gitlab_token}"},
)

if response.status_code != 200:
    log.info(f"Error: {response}")
    sys.exit(1)

path_prefix = pathlib.Path().cwd()
config: RawConfig = {}


for group in response.json():
    url_to_repo = group["ssh_url_to_repo"].replace(":", "/")
    namespace_path = group["namespace"]["full_path"]
    reponame = group["path"]

    path = f"{path_prefix}/{namespace_path}"

    if path not in config:
        config[path] = {}

    # simplified config not working - https://github.com/vcs-python/vcspull/issues/332
    # config[path][reponame] = 'git+ssh://%s' % (url_to_repo)

    vcs = guess_vcs(url_to_repo)
    if vcs is None:
        raise CouldNotGuessVCSFromURL(url_to_repo)

    config[path][reponame] = {
        "name": reponame,
        "path": path / reponame,
        "url": f"git+ssh://{url_to_repo}",
        "remotes": {
            "origin": GitRemote(
                name="origin",
                fetch_url=f"ssh://{url_to_repo}",
                push_url=f"ssh://{url_to_repo}",
            ),
        },
        "vcs": vcs,
    }

config_yaml = yaml.dump(config)

log.info(config_yaml)

config_file.write(config_yaml)
config_file.close()

Demonstration

Assume current directory of /home/user/workspace/ and script at /home/user/workspace/scripts/generate_gitlab.sh:

$ ./scripts/generate_gitlab.py https://gitlab.com vcs-python

/my/workspace/vcs-python:
  g:
    remotes:
      origin: ssh://[email protected]/vcs-python/g.git
    url: git+ssh://[email protected]/vcs-python/g.git
  libvcs:
    remotes:
      origin: ssh://[email protected]/vcs-python/libvcs.git
    url: git+ssh://[email protected]/vcs-python/libvcs.git
  vcspull:
    remotes:
      origin: ssh://[email protected]/vcs-python/vcspull.git
    url: git+ssh://[email protected]/vcs-python/vcspull.git

Contribute your own

Post yours on https://github.com/vcs-python/vcspull/discussions or create a PR to add yours to scripts/ and be featured here