quay/storage/swift.py

""" Swift storage driver.

    Uses: http://docs.openstack.org/developer/swift/overview_large_objects.html
"""
import os.path
import copy
import hmac
import string
import logging
from _pyio import BufferedReader
from uuid import uuid4

from swiftclient.client import Connection, ClientException
from urlparse import urlparse
from random import SystemRandom
from hashlib import sha1
from time import time, sleep
from collections import namedtuple

from util.registry import filelike
from storage.basestorage import BaseStorage
from util.registry.generatorfile import GeneratorFile


logger = logging.getLogger(__name__)

_PartUploadMetadata = namedtuple('_PartUploadMetadata', ['path', 'offset', 'length'])
_SEGMENTS_KEY = 'segments'
_SEGMENT_DIRECTORY = 'segments'
_MAXIMUM_SEGMENT_SIZE = 200000000 # ~200 MB
_DEFAULT_SWIFT_CONNECT_TIMEOUT = 5 # seconds

class SwiftStorage(BaseStorage):
  def __init__(self, metric_queue, swift_container, storage_path, auth_url, swift_user,
               swift_password, auth_version=None, os_options=None, ca_cert_path=None,
               temp_url_key=None, simple_path_concat=False, connect_timeout=None,
               retry_count=None, retry_on_ratelimit=True):
    super(SwiftStorage, self).__init__()
    self._swift_container = swift_container

    self._storage_path = storage_path.lstrip('/')
    self._simple_path_concat = simple_path_concat

    self._auth_url = auth_url
    self._ca_cert_path = ca_cert_path

    self._swift_user = swift_user
    self._swift_password = swift_password

    self._temp_url_key = temp_url_key
    self._connect_timeout = connect_timeout
    self._retry_count = retry_count
    self._retry_on_ratelimit = retry_on_ratelimit

    try:
      self._auth_version = int(auth_version or '2')
    except ValueError:
      self._auth_version = 2

    self._os_options = os_options or {}

    self._initialized = False

  def _get_connection(self):
    return Connection(
      authurl=self._auth_url,
      cacert=self._ca_cert_path,

      user=self._swift_user,
      key=self._swift_password,

      auth_version=self._auth_version,
      os_options=self._os_options,

      retry_on_ratelimit=self._retry_on_ratelimit,
      timeout=self._connect_timeout or _DEFAULT_SWIFT_CONNECT_TIMEOUT,
      retries=self._retry_count or 5,
    )

  def _normalize_path(self, object_path):
    """ No matter what inputs we get, we are going to return a path without a leading or trailing
        '/'
    """
    if self._simple_path_concat:
      return (self._storage_path + object_path).rstrip('/')
    else:
      return os.path.join(self._storage_path, object_path).rstrip('/')

  def _get_object(self, path, chunk_size=None):
    path = self._normalize_path(path)
    try:
      _, obj = self._get_connection().get_object(self._swift_container, path,
                                                 resp_chunk_size=chunk_size)
      return obj
    except Exception as ex:
      logger.exception('Could not get object at path %s: %s', path, ex)
      raise IOError('Path %s not found' % path)

  def _put_object(self, path, content, chunk=None, content_type=None, content_encoding=None,
                  headers=None):
    path = self._normalize_path(path)
    headers = headers or {}

    if content_encoding is not None:
      headers['Content-Encoding'] = content_encoding

    try:
      self._get_connection().put_object(self._swift_container, path, content,
                                        chunk_size=chunk, content_type=content_type,
                                        headers=headers)
    except ClientException:
      # We re-raise client exception here so that validation of config during setup can see
      # the client exception messages.
      raise
    except Exception as ex:
      logger.exception('Could not put object at path %s: %s', path, ex)
      raise IOError("Could not put content: %s" % path)

  def _head_object(self, path):
    path = self._normalize_path(path)
    try:
      return self._get_connection().head_object(self._swift_container, path)
    except ClientException as ce:
      if ce.http_status != 404:
        logger.exception('Could not head object at path %s: %s', path, ex)

      return None
    except Exception as ex:
      logger.exception('Could not head object at path %s: %s', path, ex)
      return None

  def get_direct_download_url(self, object_path, expires_in=60, requires_cors=False, head=False):
    if requires_cors:
      return None

    # Reference: http://docs.openstack.org/juno/config-reference/content/object-storage-tempurl.html
    if not self._temp_url_key:
      return None

    # Retrieve the auth details for the connection.
    try:
      object_url_value, _ = self._get_connection().get_auth()
    except ClientException:
      logger.exception('Got client exception when trying to load Swift auth')
      return None

    object_url = urlparse(object_url_value)
    scheme = object_url.scheme
    path = object_url.path.rstrip('/')
    hostname = object_url.netloc

    object_path = self._normalize_path(object_path)

    # Generate the signed HMAC body.
    method = 'HEAD' if head else 'GET'
    expires = int(time() + expires_in)
    full_path = '%s/%s/%s' % (path, self._swift_container, object_path)

    hmac_body = '%s\n%s\n%s' % (method, expires, full_path)
    sig = hmac.new(self._temp_url_key.encode('utf-8'), hmac_body.encode('utf-8'), sha1).hexdigest()

    surl = '{scheme}://{host}{full_path}?temp_url_sig={sig}&temp_url_expires={expires}'
    return surl.format(scheme=scheme, host=hostname, full_path=full_path, sig=sig, expires=expires)

  def validate(self, client):
    super(SwiftStorage, self).validate(client)

    if self._temp_url_key:
      # Generate a direct download URL.
      dd_url = self.get_direct_download_url('_verify')

      if not dd_url:
        raise Exception('Could not validate direct download URL; the token may be invalid.')

      # Try to retrieve the direct download URL.
      response = client.get(dd_url, timeout=2)
      if response.status_code != 200:
        logger.debug('Direct download failure: %s => %s with body %s', dd_url,
                     response.status_code, response.text)

        msg = 'Direct download URL failed with status code %s. Please check your temp-url-key.'
        raise Exception(msg % response.status_code)

  def get_content(self, path):
    return self._get_object(path)

  def put_content(self, path, content):
    self._put_object(path, content)

  def stream_read(self, path):
    for data in self._get_object(path, self.buffer_size):
      yield data

  def stream_read_file(self, path):
    return GeneratorFile(self.stream_read(path))

  def stream_write(self, path, fp, content_type=None, content_encoding=None):
    self._put_object(path, fp, self.buffer_size, content_type=content_type,
                     content_encoding=content_encoding)

  def exists(self, path):
    return bool(self._head_object(path))

  def remove(self, path):
    path = self._normalize_path(path)
    try:
      self._get_connection().delete_object(self._swift_container, path)
    except Exception:
      logger.exception('Could not delete path %s', path)
      raise IOError('Cannot delete path: %s' % path)

  def _random_checksum(self, count):
    chars = string.ascii_uppercase + string.digits
    return ''.join(SystemRandom().choice(chars) for _ in range(count))

  def get_checksum(self, path):
    headers = self._head_object(path)
    if not headers:
      raise IOError('Cannot lookup path: %s' % path)

    return headers.get('etag', '')[1:-1][:7] or self._random_checksum(7)

  @staticmethod
  def _segment_list_from_metadata(storage_metadata):
    return [_PartUploadMetadata(*segment_args) for segment_args in storage_metadata[_SEGMENTS_KEY]]

  def initiate_chunked_upload(self):
    random_uuid = str(uuid4())

    metadata = {
      _SEGMENTS_KEY: [],
    }

    return random_uuid, metadata

  def stream_upload_chunk(self, uuid, offset, length, in_fp, storage_metadata, content_type=None):
    if length == 0:
      return 0, storage_metadata, None

    # Note: Swift limits segments to a maximum of 5GB, so we keep writing segments until we
    # are finished hitting the data limit.
    total_bytes_written = 0
    upload_error = None
    read_until_end = length == filelike.READ_UNTIL_END

    while True:
      try:
        bytes_written, storage_metadata = self._stream_upload_segment(uuid, offset, length, in_fp,
                                                                      storage_metadata,
                                                                      content_type)
      except IOError as ex:
        message = ('Error writing to stream in stream_upload_chunk for uuid %s (offset %s' +
                   ', length %s, metadata: %s): %s')
        logger.exception(message, uuid, offset, length, storage_metadata, ex)
        upload_error = ex
        break

      if not read_until_end:
        length = length - bytes_written

      offset = offset + bytes_written
      total_bytes_written = total_bytes_written + bytes_written
      if bytes_written == 0 or (not read_until_end and length <= 0):
        return total_bytes_written, storage_metadata, upload_error

    return total_bytes_written, storage_metadata, upload_error

  def _stream_upload_segment(self, uuid, offset, length, in_fp, storage_metadata, content_type):
    updated_metadata = copy.deepcopy(storage_metadata)
    segment_count = len(updated_metadata[_SEGMENTS_KEY])
    segment_path = '%s/%s/%s' % (_SEGMENT_DIRECTORY, uuid, segment_count)

    # Track the number of bytes read and if an explicit length is specified, limit the
    # file stream to that length.
    if length == filelike.READ_UNTIL_END:
      length = _MAXIMUM_SEGMENT_SIZE
    else:
      length = min(_MAXIMUM_SEGMENT_SIZE, length)

    limiting_fp = filelike.LimitingStream(in_fp, length)

    # If retries are requested, then we need to use a buffered reader to allow for calls to
    # seek() on retries from within the Swift client.
    if self._retry_count > 0:
      limiting_fp = BufferedReader(limiting_fp, buffer_size=length)

    # Write the segment to Swift.
    self.stream_write(segment_path, limiting_fp, content_type)

    # We are only going to track keys to which data was confirmed written.
    bytes_written = limiting_fp.tell()
    if bytes_written > 0:
      updated_metadata[_SEGMENTS_KEY].append(_PartUploadMetadata(segment_path, offset,
                                                                 bytes_written))
    else:
      # Try to delete the empty segment, as it is not needed. This will occasionally fail
      # due to Swift's eventual consistency, so we retry a few times and then just leave it be.
      for remaining_retries in range(2, -1, -1):
        try:
          self.remove(segment_path)
        except IOError:
          if remaining_retries:
            sleep(0.25)
            continue

          # Otherwise, ignore it.
          break

    return bytes_written, updated_metadata

  def complete_chunked_upload(self, uuid, final_path, storage_metadata):
    """ Complete the chunked upload and store the final results in the path indicated.
        Returns nothing.
    """
    # Finally, we write an empty file at the proper location with a X-Object-Manifest
    # header pointing to the prefix for the segments.
    segments_prefix_path = self._normalize_path('%s/%s' % (_SEGMENT_DIRECTORY, uuid))
    contained_segments_prefix_path = '%s/%s' % (self._swift_container, segments_prefix_path)

    self._put_object(final_path, '', headers={'X-Object-Manifest': contained_segments_prefix_path})

  def cancel_chunked_upload(self, uuid, storage_metadata):
    """ Cancel the chunked upload and clean up any outstanding partially uploaded data.
        Returns nothing.
    """
    # Delete all the uploaded segments.
    for segment in SwiftStorage._segment_list_from_metadata(storage_metadata):
      self.remove(segment.path)