Connect `Project` to `RemoteRepository`

reconnect_repository.py

# Script to connect `Project` to `RemoteRepository` without the user intervention.
#
# We will only connect them if:
#  - the `Project` is not connected to a `RemoteRepository` already
#  - we are able to find the final URL based on `Project.repo`
#  - a `RemoteRepository` object exists in our database with `html_url` or `ssh_url` matching the final URL
#  - at least one project's admin (on Read the Docs) is admin of the VCS repository
#

import requests

from django.conf import settings
from django.db.models import Sum, Q
from readthedocs.projects.models import Project
from readthedocs.core.permissions import AdminPermission
from readthedocs.oauth.models import RemoteRepository

NUMBER_PROJECTS = 100

projects = Project.objects.annotate(spam_score=Sum("spam_rules__value")).filter(
    spam_score__lt=settings.RTD_SPAM_THRESHOLD_DENY_ON_ROBOTS,
    remote_repository__isnull=True,
)[:NUMBER_PROJECTS]

print(f'Projects to reconnect: {projects.count()}')


def get_final_url(project):
    final_url = None
    url = project.repo
    if "@" in url:
        # NOTE: Do we need to convert this? After checking with
        # community/commercial data it seems it's not required and it will
        # match with `RemoteRepository.ssh_url`
        #
        # url = convert_to_https(url)
        final_url = url
    if url.startswith("http"):
        try:
            response = requests.head(project.repo, allow_redirects=True)
            final_url = response.url
        except requests.exceptions.TooManyRedirects:
            print('Too many redirects')
    return final_url


def get_remote_repository(project, url):
    return RemoteRepository.objects.filter(
        Q(html_url=url) | Q(ssh_url=url),
        remote_repository_relations__user__in=AdminPermission.admins(project),
        remote_repository_relations__admin=True,
    ).first()


def connect_remote_repository(project):
    url = get_final_url(project)
    remote_repository = get_remote_repository(project, url)
    if not url:
        print('URL not found for this project.')
        return

    if not remote_repository:
        print(f'RemoteRepository not found for this project. url={url}')
        return

    if project.remote_repository:
        print(f'WARNING: project already connected to RemoteRepository: {project.remote_repository.pk}')

    print(f'  slug: {project.slug}')
    print(f'  remoterepository:')
    print(f'    {remote_repository.full_name}')
    print(f'    {remote_repository.pk}')
    print(f'  url: {url}')

    # TODO: save the connection when we are ready


for project in projects:
    print(f'Connecting: {project.slug}')
    connect_remote_repository(project)