quay/endpoints/api/logs.py

""" Access usage logs for organizations or repositories. """

import json

from datetime import datetime, timedelta
from dateutil.relativedelta import relativedelta

from endpoints.api import (resource, nickname, ApiResource, query_param, parse_args,
                           RepositoryParamResource, require_repo_admin, related_user_resource,
                           format_date, require_user_admin, path_param, require_scope, page_support)
from endpoints.exception import Unauthorized, NotFound
from auth.permissions import AdministerOrganizationPermission
from auth.auth_context import get_authenticated_user
from data import model, database
from auth import scopes
from app import avatar
from tzlocal import get_localzone

LOGS_PER_PAGE = 20
SERVICE_LEVEL_LOG_KINDS = set(['service_key_create', 'service_key_approve', 'service_key_delete',
                               'service_key_modify', 'service_key_extend', 'service_key_rotate'])

def log_view(log, kinds, include_namespace):
  view = {
    'kind': kinds[log.kind_id],
    'metadata': json.loads(log.metadata_json),
    'ip': log.ip,
    'datetime': format_date(log.datetime),
  }

  if log.performer and log.performer.username:
    view['performer'] = {
      'kind': 'user',
      'name': log.performer.username,
      'is_robot': log.performer.robot,
      'avatar': avatar.get_data_for_user(log.performer),
    }

  if include_namespace:
    if log.account and log.account.username:
      if log.account.organization:
        view['namespace'] = {
          'kind': 'org',
          'name': log.account.username,
          'avatar': avatar.get_data_for_org(log.account),
        }
      else:
        view['namespace'] = {
          'kind': 'user',
          'name': log.account.username,
          'avatar': avatar.get_data_for_user(log.account),
        }

  return view

def aggregated_log_view(log, kinds, start_time):
  # Because we aggregate based on the day of the month in SQL, we only have that information.
  # Therefore, create a synthetic date based on the day and the month of the start time.
  # Logs are allowed for a maximum period of one week, so this calculation should always work.
  synthetic_date = datetime(start_time.year, start_time.month, int(log.day), tzinfo=get_localzone())
  if synthetic_date.day < start_time.day:
    synthetic_date = synthetic_date + relativedelta(months=1)

  view = {
    'kind': kinds[log.kind_id],
    'count': log.count,
    'datetime': format_date(synthetic_date),
  }

  return view

def _validate_logs_arguments(start_time, end_time, performer_name):
  performer = None
  if performer_name:
    performer = model.user.get_user(performer_name)

  if start_time:
    try:
      start_time = datetime.strptime(start_time + ' UTC', '%m/%d/%Y %Z')
    except ValueError:
      start_time = None

  if not start_time:
    start_time = datetime.today() - timedelta(7) # One week

  if end_time:
    try:
      end_time = datetime.strptime(end_time + ' UTC', '%m/%d/%Y %Z')
      end_time = end_time + timedelta(days=1)
    except ValueError:
      end_time = None

  if not end_time:
    end_time = datetime.today()

  return (start_time, end_time, performer)


def get_logs(start_time, end_time, performer_name=None, repository=None, namespace=None,
             page_token=None, ignore=None):
  (start_time, end_time, performer) = _validate_logs_arguments(start_time, end_time, performer_name)
  kinds = model.log.get_log_entry_kinds()
  logs_query = model.log.get_logs_query(start_time, end_time, performer=performer,
                                        repository=repository, namespace=namespace,
                                        ignore=ignore)

  logs, next_page_token = model.modelutil.paginate(logs_query, database.LogEntry, descending=True,
                                                   page_token=page_token, limit=LOGS_PER_PAGE)

  include_namespace = namespace is None and repository is None
  return {
    'start_time': format_date(start_time),
    'end_time': format_date(end_time),
    'logs': [log_view(log, kinds, include_namespace) for log in logs],
  }, next_page_token


def get_aggregate_logs(start_time, end_time, performer_name=None, repository=None, namespace=None,
                       ignore=None):
  (start_time, end_time, performer) = _validate_logs_arguments(start_time, end_time, performer_name)

  kinds = model.log.get_log_entry_kinds()
  aggregated_logs = model.log.get_aggregated_logs(start_time, end_time, performer=performer,
                                                  repository=repository, namespace=namespace,
                                                  ignore=ignore)

  return {
    'aggregated': [aggregated_log_view(log, kinds, start_time) for log in aggregated_logs]
  }


@resource('/v1/repository/<apirepopath:repository>/logs')
@path_param('repository', 'The full path of the repository. e.g. namespace/name')
class RepositoryLogs(RepositoryParamResource):
  """ Resource for fetching logs for the specific repository. """
  @require_repo_admin
  @nickname('listRepoLogs')
  @parse_args()
  @query_param('starttime', 'Earliest time from which to get logs (%m/%d/%Y %Z)', type=str)
  @query_param('endtime', 'Latest time to which to get logs (%m/%d/%Y %Z)', type=str)
  @query_param('page', 'The page number for the logs', type=int, default=1)
  @page_support()
  def get(self, namespace, repository, page_token, parsed_args):
    """ List the logs for the specified repository. """
    repo = model.repository.get_repository(namespace, repository)
    if not repo:
      raise NotFound()

    start_time = parsed_args['starttime']
    end_time = parsed_args['endtime']
    return get_logs(start_time, end_time, repository=repo, page_token=page_token,
                    ignore=SERVICE_LEVEL_LOG_KINDS)


@resource('/v1/user/logs')
class UserLogs(ApiResource):
  """ Resource for fetching logs for the current user. """
  @require_user_admin
  @nickname('listUserLogs')
  @parse_args()
  @query_param('starttime', 'Earliest time from which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('endtime', 'Latest time to which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('performer', 'Username for which to filter logs.', type=str)
  @page_support()
  def get(self, parsed_args, page_token):
    """ List the logs for the current user. """
    performer_name = parsed_args['performer']
    start_time = parsed_args['starttime']
    end_time = parsed_args['endtime']

    user = get_authenticated_user()
    return get_logs(start_time, end_time, performer_name=performer_name, namespace=user.username,
                    page_token=page_token, ignore=SERVICE_LEVEL_LOG_KINDS)


@resource('/v1/organization/<orgname>/logs')
@path_param('orgname', 'The name of the organization')
@related_user_resource(UserLogs)
class OrgLogs(ApiResource):
  """ Resource for fetching logs for the entire organization. """
  @nickname('listOrgLogs')
  @parse_args()
  @query_param('starttime', 'Earliest time from which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('endtime', 'Latest time to which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('performer', 'Username for which to filter logs.', type=str)
  @query_param('page', 'The page number for the logs', type=int, default=1)
  @page_support()
  @require_scope(scopes.ORG_ADMIN)
  def get(self, orgname, page_token, parsed_args):
    """ List the logs for the specified organization. """
    permission = AdministerOrganizationPermission(orgname)
    if permission.can():
      performer_name = parsed_args['performer']
      start_time = parsed_args['starttime']
      end_time = parsed_args['endtime']

      return get_logs(start_time, end_time, namespace=orgname, performer_name=performer_name,
                      page_token=page_token, ignore=SERVICE_LEVEL_LOG_KINDS)

    raise Unauthorized()


@resource('/v1/repository/<apirepopath:repository>/aggregatelogs')
@path_param('repository', 'The full path of the repository. e.g. namespace/name')
class RepositoryAggregateLogs(RepositoryParamResource):
  """ Resource for fetching aggregated logs for the specific repository. """
  @require_repo_admin
  @nickname('getAggregateRepoLogs')
  @parse_args()
  @query_param('starttime', 'Earliest time from which to get logs (%m/%d/%Y %Z)', type=str)
  @query_param('endtime', 'Latest time to which to get logs (%m/%d/%Y %Z)', type=str)
  def get(self, namespace, repository, parsed_args):
    """ Returns the aggregated logs for the specified repository. """
    repo = model.repository.get_repository(namespace, repository)
    if not repo:
      raise NotFound()

    start_time = parsed_args['starttime']
    end_time = parsed_args['endtime']
    return get_aggregate_logs(start_time, end_time, repository=repo,
                              ignore=SERVICE_LEVEL_LOG_KINDS)


@resource('/v1/user/aggregatelogs')
class UserAggregateLogs(ApiResource):
  """ Resource for fetching aggregated logs for the current user. """
  @require_user_admin
  @nickname('getAggregateUserLogs')
  @parse_args()
  @query_param('starttime', 'Earliest time from which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('endtime', 'Latest time to which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('performer', 'Username for which to filter logs.', type=str)
  def get(self, parsed_args):
    """ Returns the aggregated logs for the current user. """
    performer_name = parsed_args['performer']
    start_time = parsed_args['starttime']
    end_time = parsed_args['endtime']

    user = get_authenticated_user()
    return get_aggregate_logs(start_time, end_time, performer_name=performer_name,
                              namespace=user.username, ignore=SERVICE_LEVEL_LOG_KINDS)


@resource('/v1/organization/<orgname>/aggregatelogs')
@path_param('orgname', 'The name of the organization')
@related_user_resource(UserLogs)
class OrgAggregateLogs(ApiResource):
  """ Resource for fetching aggregate logs for the entire organization. """
  @nickname('getAggregateOrgLogs')
  @parse_args()
  @query_param('starttime', 'Earliest time from which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('endtime', 'Latest time to which to get logs. (%m/%d/%Y %Z)', type=str)
  @query_param('performer', 'Username for which to filter logs.', type=str)
  @require_scope(scopes.ORG_ADMIN)
  def get(self, orgname, parsed_args):
    """ Gets the aggregated logs for the specified organization. """
    permission = AdministerOrganizationPermission(orgname)
    if permission.can():
      performer_name = parsed_args['performer']
      start_time = parsed_args['starttime']
      end_time = parsed_args['endtime']

      return get_aggregate_logs(start_time, end_time, namespace=orgname,
                                performer_name=performer_name, ignore=SERVICE_LEVEL_LOG_KINDS)

    raise Unauthorized()