quay/buildtrigger/gitlabhandler.py

import logging
import os

from calendar import timegm
from functools import wraps

import dateutil.parser

from app import app, gitlab_trigger

from jsonschema import validate
from buildtrigger.triggerutil import (RepositoryReadException, TriggerActivationException,
                                      TriggerDeactivationException, TriggerStartException,
                                      SkipRequestException, InvalidPayloadException,
                                      determine_build_ref, raise_if_skipped_build,
                                      find_matching_branches)

from buildtrigger.basehandler import BuildTriggerHandler

from util.security.ssh import generate_ssh_keypair
from util.dict_wrappers import JSONPathDict, SafeDictSetter
from endpoints.exception import ExternalServiceError

import gitlab
import requests

logger = logging.getLogger(__name__)

GITLAB_WEBHOOK_PAYLOAD_SCHEMA = {
  'type': 'object',
  'properties': {
    'ref': {
      'type': 'string',
    },
    'checkout_sha': {
      'type': ['string', 'null'],
    },
    'repository': {
      'type': 'object',
      'properties': {
        'git_ssh_url': {
          'type': 'string',
        },
      },
      'required': ['git_ssh_url'],
    },
    'commits': {
      'type': 'array',
      'items': {
        'type': 'object',
        'properties': {
          'id': {
            'type': 'string',
          },
          'url': {
            'type': 'string',
          },
          'message': {
            'type': 'string',
          },
          'timestamp': {
            'type': 'string',
          },
          'author': {
            'type': 'object',
            'properties': {
              'email': {
                'type': 'string',
              },
            },
            'required': ['email'],
          },
        },
        'required': ['id', 'url', 'message', 'timestamp'],
      },
    },
  },
  'required': ['ref', 'checkout_sha', 'repository'],
}

_ACCESS_LEVEL_MAP = {
  50: ("owner", True),
  40: ("master", True),
  30: ("developer", False),
  20: ("reporter", False),
  10: ("guest", False),
}

_PER_PAGE_COUNT = 20


def _catch_timeouts(func):
  @wraps(func)
  def wrapper(*args, **kwargs):
    try:
      return func(*args, **kwargs)
    except requests.exceptions.Timeout:
      msg = 'Request to the GitLab API timed out'
      logger.exception(msg)
      raise ExternalServiceError(msg)
  return wrapper


def _paginated_iterator(func, exc):
  """ Returns an iterator over invocations of the given function, automatically handling
      pagination.
  """
  page = 0
  while True:
    result = func(page=page, per_page=_PER_PAGE_COUNT)
    if result is False:
      raise exc

    counter = 0
    for item in result:
      yield item
      counter = counter + 1

    if counter < _PER_PAGE_COUNT:
      break

    page = page + 1


def get_transformed_webhook_payload(gl_payload, default_branch=None, lookup_user=None,
                                    lookup_commit=None):
  """ Returns the Gitlab webhook JSON payload transformed into our own payload
      format. If the gl_payload is not valid, returns None.
  """
  try:
    validate(gl_payload, GITLAB_WEBHOOK_PAYLOAD_SCHEMA)
  except Exception as exc:
    raise InvalidPayloadException(exc.message)

  payload = JSONPathDict(gl_payload)

  if payload['object_kind'] != 'push' and payload['object_kind'] != 'tag_push':
    # Unknown kind of webhook.
    raise SkipRequestException

  # Check for empty commits. The commits list will be empty if the branch is deleted.
  commits = payload['commits']
  if payload['object_kind'] == 'push' and not commits:
    raise SkipRequestException

  config = SafeDictSetter()
  config['commit'] = payload['checkout_sha']
  config['ref'] = payload['ref']
  config['default_branch'] = default_branch
  config['git_url'] = payload['repository.git_ssh_url']

  found_commit = JSONPathDict({})
  if payload['object_kind'] == 'push':
    # Find the commit associated with the checkout_sha. Gitlab doesn't (necessary) send this in
    # any order, so we cannot simply index into the commits list.
    found_commit = None
    for commit in commits:
      if commit['id'] == payload['checkout_sha']:
        found_commit = JSONPathDict(commit)
        break

    if found_commit is None:
      raise SkipRequestException

  elif payload['object_kind'] == 'tag_push':
    # Gitlab doesn't send commit information for tag pushes (WHY?!), so we need to lookup the
    # commit SHA directly.
    if lookup_commit:
      found_commit_info = lookup_commit(payload['project_id'], payload['checkout_sha'])
      found_commit = JSONPathDict(found_commit_info or {})

  config['commit_info.url'] = found_commit['url']
  config['commit_info.message'] = found_commit['message']
  config['commit_info.date'] = found_commit['timestamp']

  # Note: Gitlab does not send full user information with the payload, so we have to
  # (optionally) look it up.
  author_email = found_commit['author.email'] or found_commit['author_email']
  if lookup_user and author_email:
    author_info = lookup_user(author_email)
    if author_info:
      config['commit_info.author.username'] = author_info['username']
      config['commit_info.author.url'] = author_info['html_url']
      config['commit_info.author.avatar_url'] = author_info['avatar_url']

  return config.dict_value()


class GitLabBuildTrigger(BuildTriggerHandler):
  """
  BuildTrigger for GitLab.
  """
  @classmethod
  def service_name(cls):
    return 'gitlab'

  def _get_authorized_client(self):
    auth_token = self.auth_token or 'invalid'
    return gitlab.Gitlab(gitlab_trigger.api_endpoint(), oauth_token=auth_token, timeout=5)

  def is_active(self):
    return 'hook_id' in self.config

  @_catch_timeouts
  def activate(self, standard_webhook_url):
    config = self.config
    new_build_source = config['build_source']
    gl_client = self._get_authorized_client()

    # Find the GitLab repository.
    repository = gl_client.getproject(new_build_source)
    if repository is False:
      msg = 'Unable to find GitLab repository for source: %s' % new_build_source
      raise TriggerActivationException(msg)

    # Add a deploy key to the repository.
    public_key, private_key = generate_ssh_keypair()
    config['credentials'] = [
      {
        'name': 'SSH Public Key',
        'value': public_key,
      },
    ]
    key = gl_client.adddeploykey(repository['id'], '%s Builder' % app.config['REGISTRY_TITLE'],
                                 public_key)
    if key is False:
      msg = 'Unable to add deploy key to repository: %s' % new_build_source
      raise TriggerActivationException(msg)
    config['key_id'] = key['id']

    # Add the webhook to the GitLab repository.
    hook = gl_client.addprojecthook(repository['id'], standard_webhook_url, push=True)
    if hook is False:
      msg = 'Unable to create webhook on repository: %s' % new_build_source
      raise TriggerActivationException(msg)

    config['hook_id'] = hook['id']
    self.config = config
    return config, {'private_key': private_key}

  def deactivate(self):
    config = self.config
    gl_client = self._get_authorized_client()

    # Find the GitLab repository.
    repository = gl_client.getproject(config['build_source'])
    if repository is False:
      msg = 'Unable to find GitLab repository for source: %s' % config['build_source']
      raise TriggerDeactivationException(msg)

    # Remove the webhook.
    success = gl_client.deleteprojecthook(repository['id'], config['hook_id'])
    if success is False:
      msg = 'Unable to remove hook: %s' % config['hook_id']
      raise TriggerDeactivationException(msg)
    config.pop('hook_id', None)

    # Remove the key
    success = gl_client.deletedeploykey(repository['id'], config['key_id'])
    if success is False:
      msg = 'Unable to remove deploy key: %s' % config['key_id']
      raise TriggerDeactivationException(msg)
    config.pop('key_id', None)

    self.config = config
    return config

  @_catch_timeouts
  def list_build_source_namespaces(self):
    gl_client = self._get_authorized_client()
    current_user = gl_client.currentuser()
    if current_user is False:
      raise RepositoryReadException('Unable to get current user')

    namespaces = {}
    repositories = _paginated_iterator(gl_client.getprojects, RepositoryReadException)
    for repo in repositories:
      namespace = repo.get('namespace') or {}
      if not namespace:
        continue

      namespace_id = namespace['id']

      avatar_url = ''
      if 'avatar' in namespace:
        avatar_data = namespace.get('avatar') or {}
        avatar_url = avatar_data.get('url')
      elif 'owner' in repo:
        owner_data = repo.get('owner') or {}
        avatar_url = owner_data.get('avatar_url')

      if namespace_id in namespaces:
        namespaces[namespace_id]['score'] = namespaces[namespace_id]['score'] + 1
      else:
        owner = namespace['name']
        namespaces[namespace_id] = {
          'personal': owner == current_user['username'],
          'id': namespace['path'],
          'title': namespace['name'],
          'avatar_url': avatar_url,
          'score': 1,
          'url': gl_client.host + '/' + namespace['path'],
        }

    return BuildTriggerHandler.build_namespaces_response(namespaces)

  @_catch_timeouts
  def list_build_sources_for_namespace(self, namespace):
    def repo_view(repo):
      # Because *anything* can be None in GitLab API!
      permissions = repo.get('permissions') or {}
      group_access = permissions.get('group_access') or {}
      project_access = permissions.get('project_access') or {}

      missing_group_access = permissions.get('group_access') is None
      missing_project_access = permissions.get('project_access') is None

      access_level = max(group_access.get('access_level') or 0,
                         project_access.get('access_level') or 0)

      has_admin_permission = _ACCESS_LEVEL_MAP.get(access_level, ("", False))[1]
      if missing_group_access or missing_project_access:
        # Default to has permission if we cannot check the permissions. This will allow our users
        # to select the repository and then GitLab's own checks will ensure that the webhook is
        # added only if allowed.
        # TODO: Do we want to display this differently in the UI?
        has_admin_permission = True

      view = {
        'name': repo['path'],
        'full_name': repo['path_with_namespace'],
        'description': repo.get('description') or '',
        'url': repo.get('web_url'),
        'has_admin_permissions': has_admin_permission,
        'private': repo.get('public', False) is False,
      }

      if repo.get('last_activity_at'):
        try:
          last_modified = dateutil.parser.parse(repo['last_activity_at'])
          view['last_updated'] = timegm(last_modified.utctimetuple())
        except ValueError:
          logger.exception('Gitlab gave us an invalid last_activity_at: %s', last_modified)

      return view

    gl_client = self._get_authorized_client()
    repositories = _paginated_iterator(gl_client.getprojects, RepositoryReadException)
    repos = [repo_view(repo) for repo in repositories if repo['namespace']['path'] == namespace]
    return BuildTriggerHandler.build_sources_response(repos)

  @_catch_timeouts
  def list_build_subdirs(self):
    config = self.config
    gl_client = self._get_authorized_client()
    new_build_source = config['build_source']

    repository = gl_client.getproject(new_build_source)
    if repository is False:
      msg = 'Unable to find GitLab repository for source: %s' % new_build_source
      raise RepositoryReadException(msg)

    repo_branches = gl_client.getbranches(repository['id'])
    if repo_branches is False:
      msg = 'Unable to find GitLab branches for source: %s' % new_build_source
      raise RepositoryReadException(msg)

    branches = [branch['name'] for branch in repo_branches]
    branches = find_matching_branches(config, branches)
    branches = branches or [repository['default_branch'] or 'master']

    repo_tree = gl_client.getrepositorytree(repository['id'], ref_name=branches[0])
    if repo_tree is False:
      msg = 'Unable to find GitLab repository tree for source: %s' % new_build_source
      raise RepositoryReadException(msg)

    return ["/" + node['name'] for node in repo_tree if self.filename_is_dockerfile(node['name'])]

  @_catch_timeouts
  def load_dockerfile_contents(self):
    gl_client = self._get_authorized_client()
    path = self.get_dockerfile_path()

    repository = gl_client.getproject(self.config['build_source'])
    if repository is False:
      return None

    branches = self.list_field_values('branch_name')
    branches = find_matching_branches(self.config, branches)
    if branches == []:
      return None

    branch_name = branches[0]
    if repository['default_branch'] in branches:
      branch_name = repository['default_branch']

    contents = gl_client.getrawfile(repository['id'], branch_name, path)
    if contents is False:
      return None

    return contents

  @_catch_timeouts
  def list_field_values(self, field_name, limit=None):
    if field_name == 'refs':
      branches = self.list_field_values('branch_name')
      tags = self.list_field_values('tag_name')

      return ([{'kind': 'branch', 'name': b} for b in branches] +
              [{'kind': 'tag', 'name': t} for t in tags])

    gl_client = self._get_authorized_client()
    repo = gl_client.getproject(self.config['build_source'])
    if repo is False:
      return []

    if field_name == 'tag_name':
      tags = gl_client.getrepositorytags(repo['id'])
      if tags is False:
        return []

      if limit:
        tags = tags[0:limit]

      return [tag['name'] for tag in tags]

    if field_name == 'branch_name':
      branches = gl_client.getbranches(repo['id'])
      if branches is False:
        return []

      if limit:
        branches = branches[0:limit]

      return [branch['name'] for branch in branches]

    return None

  def get_repository_url(self):
    return gitlab_trigger.get_public_url(self.config['build_source'])

  @_catch_timeouts
  def lookup_commit(self, repo_id, commit_sha):
    if repo_id is None:
      return None

    gl_client = self._get_authorized_client()
    commit = gl_client.getrepositorycommit(repo_id, commit_sha)
    if commit is False:
      return None

    return commit

  @_catch_timeouts
  def lookup_user(self, email):
    gl_client = self._get_authorized_client()
    try:
      result = gl_client.getusers(search=email)
      if result is False:
        return None

      [user] = result
      return {
        'username': user['username'],
        'html_url': gl_client.host + '/' + user['username'],
        'avatar_url': user['avatar_url']
      }
    except ValueError:
      return None

  @_catch_timeouts
  def get_metadata_for_commit(self, commit_sha, ref, repo):
    gl_client = self._get_authorized_client()
    commit = gl_client.getrepositorycommit(repo['id'], commit_sha)

    metadata = {
      'commit': commit['id'],
      'ref': ref,
      'default_branch': repo['default_branch'],
      'git_url': repo['ssh_url_to_repo'],
      'commit_info': {
        'url': gl_client.host + '/' + repo['path_with_namespace'] + '/commit/' + commit['id'],
        'message': commit['message'],
        'date': commit['committed_date'],
      },
    }

    committer = None
    if 'committer_email' in commit:
      committer = self.lookup_user(commit['committer_email'])

    author = None
    if 'author_email' in commit:
      author = self.lookup_user(commit['author_email'])

    if committer is not None:
      metadata['commit_info']['committer'] = {
        'username': committer['username'],
        'avatar_url': committer['avatar_url'],
        'url': gl_client.host + '/' + committer['username'],
      }

    if author is not None:
      metadata['commit_info']['author'] = {
        'username': author['username'],
        'avatar_url': author['avatar_url'],
        'url': gl_client.host + '/' + author['username']
      }

    return metadata

  @_catch_timeouts
  def manual_start(self, run_parameters=None):
    gl_client = self._get_authorized_client()

    repo = gl_client.getproject(self.config['build_source'])
    if repo is False:
      raise TriggerStartException('Could not find repository')

    def get_tag_sha(tag_name):
      tags = gl_client.getrepositorytags(repo['id'])
      if tags is False:
        raise TriggerStartException('Could not find tag in repository')

      for tag in tags:
        if tag['name'] == tag_name:
          return tag['commit']['id']

      raise TriggerStartException('Could not find tag in repository')

    def get_branch_sha(branch_name):
      branch = gl_client.getbranch(repo['id'], branch_name)
      if branch is False:
        raise TriggerStartException('Could not find branch in repository')

      return branch['commit']['id']

    # Find the branch or tag to build.
    (commit_sha, ref) = determine_build_ref(run_parameters, get_branch_sha, get_tag_sha,
                                            repo['default_branch'])

    metadata = self.get_metadata_for_commit(commit_sha, ref, repo)
    return self.prepare_build(metadata, is_manual=True)

  @_catch_timeouts
  def handle_trigger_request(self, request):
    payload = request.get_json()
    if not payload:
      raise InvalidPayloadException()

    logger.debug('GitLab trigger payload %s', payload)

    # Lookup the default branch.
    gl_client = self._get_authorized_client()
    repo = gl_client.getproject(self.config['build_source'])
    if repo is False:
      logger.debug('Skipping GitLab build; project %s not found', self.config['build_source'])
      raise InvalidPayloadException()

    default_branch = repo['default_branch']
    metadata = get_transformed_webhook_payload(payload, default_branch=default_branch,
                                               lookup_user=self.lookup_user,
                                               lookup_commit=self.lookup_commit)
    prepared = self.prepare_build(metadata)

    # Check if we should skip this build.
    raise_if_skipped_build(prepared, self.config)
    return prepared