quay/workers/securityworker.py

import logging
import requests
import features
import time
import os
import random

from sys import exc_info
from peewee import JOIN_LEFT_OUTER
from app import app, storage, OVERRIDE_CONFIG_DIRECTORY
from workers.worker import Worker
from data.database import Image, ImageStorage, ImageStorageLocation, ImageStoragePlacement, db_random_func, UseThenDisconnect

logger = logging.getLogger(__name__)

BATCH_SIZE = 20
INDEXING_INTERVAL = 10
API_METHOD_INSERT = '/layers'
API_METHOD_VERSION = '/versions/engine'

def _get_image_to_export(version):
  Parent = Image.alias()
  ParentImageStorage = ImageStorage.alias()
  rimages = []

  # Without parent
  candidates = (Image
    .select(Image.docker_image_id, ImageStorage.uuid, ImageStorage.checksum)
    .join(ImageStorage)
    .where(Image.security_indexed_engine < version, Image.parent >> None, ImageStorage.uploading == False, ImageStorage.checksum != '')
    .limit(BATCH_SIZE*10)
    .alias('candidates'))

  images = (Image
    .select(candidates.c.docker_image_id, candidates.c.uuid, candidates.c.checksum)
    .from_(candidates)
    .order_by(db_random_func())
    .tuples()
    .limit(BATCH_SIZE))

  for image in images:
    rimages.append({'docker_image_id': image[0], 'storage_uuid': image[1], 'storage_checksum': image[2], 'parent_docker_image_id': None, 'parent_storage_uuid': None})

  # With analyzed parent
  candidates = (Image
      .select(Image.docker_image_id, ImageStorage.uuid, ImageStorage.checksum, Parent.docker_image_id.alias('parent_docker_image_id'), ParentImageStorage.uuid.alias('parent_storage_uuid'))
      .join(Parent, on=(Image.parent == Parent.id))
      .join(ParentImageStorage, on=(ParentImageStorage.id == Parent.storage))
      .switch(Image)
      .join(ImageStorage)
      .where(Image.security_indexed_engine < version, Parent.security_indexed == True, Parent.security_indexed_engine >= version, ImageStorage.uploading == False, ImageStorage.checksum != '')
      .limit(BATCH_SIZE*10)
      .alias('candidates'))

  images = (Image
    .select(candidates.c.docker_image_id, candidates.c.uuid, candidates.c.checksum, candidates.c.parent_docker_image_id, candidates.c.parent_storage_uuid)
    .from_(candidates)
    .order_by(db_random_func())
    .tuples()
    .limit(BATCH_SIZE))

  for image in images:
    rimages.append({'docker_image_id': image[0], 'storage_uuid': image[1], 'storage_checksum': image[2], 'parent_docker_image_id': image[3], 'parent_storage_uuid': image[4]})

  # Re-shuffle, otherwise the images without parents will always be on the top
  random.shuffle(rimages)

  return rimages

def _get_storage_locations(uuid):
  query = (ImageStoragePlacement
      .select()
      .join(ImageStorageLocation)
      .switch(ImageStoragePlacement)
      .join(ImageStorage, JOIN_LEFT_OUTER)
      .where(ImageStorage.uuid == uuid))

  locations = list()
  for location in query:
    locations.append(location.location.name)

  return locations

def _update_image(image, indexed, version):
  query = (Image
      .select()
      .join(ImageStorage)
      .where(Image.docker_image_id == image['docker_image_id'], ImageStorage.uuid == image['storage_uuid']))

  updated_images = list()
  for image in query:
    updated_images.append(image.id)

  query = (Image
      .update(security_indexed=indexed, security_indexed_engine=version)
      .where(Image.id << updated_images))
  query.execute()

class SecurityWorker(Worker):
  def __init__(self):
    super(SecurityWorker, self).__init__()
    if self._load_configuration():
      self.add_operation(self._index_images, INDEXING_INTERVAL)

  def _load_configuration(self):
    # Load configuration
    config = app.config.get('SECURITY_SCANNER')

    if not config or not 'ENDPOINT' in config or not 'ENGINE_VERSION_TARGET' in config or not 'DISTRIBUTED_STORAGE_PREFERENCE' in app.config:
      logger.exception('No configuration found for the security worker')
      return False
    self._api = config['ENDPOINT']
    self._target_version = config['ENGINE_VERSION_TARGET']
    self._default_storage_locations = app.config['DISTRIBUTED_STORAGE_PREFERENCE']

    self._ca_verification = False
    self._cert = None
    if 'CA_CERTIFICATE_FILENAME' in config:
      self._ca_verification = os.path.join(OVERRIDE_CONFIG_DIRECTORY, config['CA_CERTIFICATE_FILENAME'])
      if not os.path.isfile(self._ca_verification):
        logger.exception('Could not find configured CA file')
        return False
    if 'PRIVATE_KEY_FILENAME' in config and 'PUBLIC_KEY_FILENAME' in config:
      self._cert = (
        os.path.join(OVERRIDE_CONFIG_DIRECTORY, config['PUBLIC_KEY_FILENAME']),
        os.path.join(OVERRIDE_CONFIG_DIRECTORY, config['PRIVATE_KEY_FILENAME']),
      )
      if not os.path.isfile(self._cert[0]) or not os.path.isfile(self._cert[1]):
        logger.exception('Could not find configured key pair files')
        return False

    return True

  def _index_images(self):
    with UseThenDisconnect(app.config):
      while True:
        # Get images to analyze
        try:
          images = _get_image_to_export(self._target_version)
        except Image.DoesNotExist:
          logger.debug('No more image to analyze')
          return

        for img in images:
          # Get layer storage URL
          path = storage.image_layer_path(img['storage_uuid'])
          locations = self._default_storage_locations
          if not storage.exists(locations, path):
            locations = _get_storage_locations(img['storage_uuid'])
          if not storage.exists(locations, path):
            logger.warning('Could not find a valid location to download layer %s', img['docker_image_id']+'.'+img['storage_uuid'])
            # Mark as analyzed because that error is most likely to occur during the pre-process, with the database copy
            # when images are actually removed on the real database (and therefore in S3)
            _update_image(img, False, self._target_version)
            continue
          uri = storage.get_direct_download_url(locations, path)
          if uri == None:
            # Local storage hack
            uri = path

          # Forge request
          request = {
            'ID': img['docker_image_id']+'.'+img['storage_uuid'],
            'TarSum': img['storage_checksum'],
            'Path': uri
          }
          if img['parent_docker_image_id'] is not None and img['parent_storage_uuid'] is not None:
            request['ParentID'] = img['parent_docker_image_id']+'.'+img['parent_storage_uuid']

          # Post request
          try:
            logger.info('Analyzing %s', request['ID'])
            # Using invalid certificates doesn't return proper errors because of
            # https://github.com/shazow/urllib3/issues/556
            httpResponse = requests.post(self._api + API_METHOD_INSERT, json=request, cert=self._cert, verify=self._ca_verification)
          except:
            logger.exception('An exception occurred when analyzing layer ID %s : %s', request['ID'], exc_info()[0])
            return
          try:
            jsonResponse = httpResponse.json()
          except:
            logger.exception('An exception occurred when analyzing layer ID %s : the response is not valid JSON (%s)', request['ID'], httpResponse.text)
            return

          if httpResponse.status_code == 201:
            # The layer has been successfully indexed
            api_version = jsonResponse['Version']
            if api_version < self._target_version:
              logger.warning('An engine runs on version %d but the target version is %d')
            _update_image(img, True, api_version)
            logger.info('Layer ID %s : analyzed successfully', request['ID'])
          else:
            if 'Message' in jsonResponse:
              if 'OS and/or package manager are not supported' in jsonResponse['Message']:
                # The current engine could not index this layer
                logger.warning('A warning event occurred when analyzing layer ID %s : %s', request['ID'], jsonResponse['Message'])
                # Hopefully, there is no version lower than the target one running
                _update_image(img, False, self._target_version)
              else:
                logger.exception('An exception occurred when analyzing layer ID %s : %d %s', request['ID'], httpResponse.status_code, jsonResponse['Message'])
                return
            else:
              logger.exception('An exception occurred when analyzing layer ID %s : %d', request['ID'], httpResponse.status_code)
              return

if __name__ == '__main__':
  logging.getLogger('requests').setLevel(logging.WARNING)
  logging.getLogger('apscheduler').setLevel(logging.CRITICAL)

  if not features.SECURITY_SCANNER:
    logger.debug('Security scanner disabled; skipping')
    while True:
      time.sleep(100000)

  worker = SecurityWorker()
  worker.start()