Work in progress: Docker -> ACI conversion

formats/aci.py

@@ -0,0 +1,185 @@
+from app import app
+from util.streamlayerformat import StreamLayerMerger
+from formats.tarimageformatter import TarImageFormatter
+
+import json
+
+class ACIImage(TarImageFormatter):
+  """ Image formatter which produces an ACI-compatible TAR.
+  """
+
+  def stream_generator(self, namespace, repository, tag, synthetic_image_id,
+                       layer_json, get_image_iterator, get_layer_iterator):
+    # ACI Format (.tar):
+    #   manifest - The JSON manifest
+    #   rootfs - The root file system
+
+    # Yield the manifest.
+    yield self.tar_file('manifest', self._build_manifest(namespace, repository, tag, layer_json,
+                                                         synthetic_image_id))
+
+    # Yield the merged layer dtaa.
+    yield self.tar_folder('rootfs')
+
+    layer_merger = StreamLayerMerger(get_layer_iterator, path_prefix='rootfs/')
+    for entry in layer_merger.get_generator():
+       yield entry
+
+  def _build_isolators(self, docker_config):
+    """ Builds ACI isolator config from the docker config. """
+
+    def _isolate_memory(memory):
+      return {
+        "name": "memory/limit",
+        "value": str(memory) + 'B'
+      }
+
+    def _isolate_swap(memory):
+      return {
+        "name": "memory/swap",
+        "value": str(memory) + 'B'
+      }
+
+    def _isolate_cpu(cpu):
+      return {
+        "name": "cpu/shares",
+        "value": str(cpu)
+      }
+
+    def _isolate_capabilities(capabilities_set):
+      return {
+        "name": "capabilities/bounding-set",
+        "value": str(capabilities_set)
+      }
+
+    mappers = {
+      'Memory': _isolate_memory,
+      'MemorySwap': _isolate_swap,
+      'CpuShares': _isolate_cpu,
+      'Cpuset': _isolate_capabilities
+    }
+
+    isolators = []
+
+    for config_key in mappers:
+      value = docker_config.get(config_key)
+      if value:
+        isolators.append(mappers[config_key](value))
+
+    return isolators
+
+  def _build_ports(self, docker_config):
+    """ Builds the ports definitions for the ACI. """
+    ports = []
+
+    for docker_port_definition in docker_config.get('ports', {}):
+      # Formats:
+      # port/tcp
+      # port/udp
+      # port
+
+      protocol = 'tcp'
+      port_number = -1
+
+      if '/' in docker_port_definition:
+        (port_number, protocol) = docker_port_definition.split('/')
+      else:
+        port_number = docker_port_definition
+
+      try:
+        port_number = int(port_number)
+        ports.append({
+          "name": "port-%s" % port_number,
+          "port": port_number,
+          "protocol": protocol
+        })
+      except ValueError:
+        pass
+
+    return ports
+
+  def _build_volumes(self, docker_config):
+    """ Builds the volumes definitions for the ACI. """
+    volumes = []
+    names = set()
+
+    def get_name(docker_volume_path):
+      parts = docker_volume_path.split('/')
+      name = ''
+
+      while True:
+        name = name + parts[-1]
+        parts = parts[0:-1]
+        if names.add(name):
+          break
+
+        name = '/' + name
+
+      return name
+
+    for docker_volume_path in docker_config.get('volumes', {}):
+      volumes.append({
+        "name": get_name(docker_volume_path),
+        "path": docker_volume_path,
+        "readOnly": False
+      })
+    return volumes
+
+
+  def _build_manifest(self, namespace, repository, tag, docker_layer_data, synthetic_image_id):
+    """ Builds an ACI manifest from the docker layer data. """
+
+    config = docker_layer_data.get('config', {})
+    config.update(docker_layer_data.get('container_config', {}))
+
+    source_url = "%s://%s/%s/%s:%s" % (app.config['PREFERRED_URL_SCHEME'],
+                                       app.config['SERVER_HOSTNAME'],
+                                       namespace, repository, tag)
+
+    exec_path = config.get('Cmd', [])
+    if exec_path:
+      if not exec_path[0].startswith('/'):
+        exec_path[0] = '/bin/' + exec_path[0]
+
+    # TODO: ACI doesn't support : in the name, so remove any ports.
+    hostname = app.config['SERVER_HOSTNAME']
+    hostname = hostname.split(':', 1)[0]
+
+    manifest = {
+      "acKind": "ImageManifest",
+      "acVersion": "0.1.1",
+      "name": '%s/%s/%s/%s' % (hostname, namespace, repository, tag),
+      "labels": [
+          {
+              "name": "version",
+              "value": "1.0.0"
+          },
+          {
+              "name": "arch",
+              "value": docker_layer_data.get('architecture', 'amd64')
+          },
+          {
+              "name": "os",
+              "value": docker_layer_data.get('os', 'linux')
+          }
+      ],
+      "app": {
+        "exec": exec_path,
+        "user": config.get('User', '') or 'root',
+        "group": config.get('Group', '') or 'root',
+        "eventHandlers": [],
+        "workingDirectory": config.get('WorkingDir', ''),
+        "environment": {key:value for (key, value) in [e.split('=') for e in config.get('Env')]},
+        "isolators": self._build_isolators(config),
+        "mountPoints": self._build_volumes(config),
+        "ports": self._build_ports(config),
+        "annotations": [
+          {"name": "created", "value": docker_layer_data.get('created', '')},
+          {"name": "homepage", "value": source_url},
+          {"name": "quay.io/derived_image", "value": synthetic_image_id},
+        ]
+      },
+    }
+
+    return json.dumps(manifest)
+

formats/squashed.py

@@ -0,0 +1,102 @@
+from app import app
+from util.gzipwrap import GZIP_BUFFER_SIZE
+from util.streamlayerformat import StreamLayerMerger
+from formats.tarimageformatter import TarImageFormatter
+
+import copy
+import json
+import tarfile
+
+class FileEstimationException(Exception):
+  """ Exception raised by build_docker_load_stream if the estimated size of the layer TAR
+      was lower than the actual size. This means the sent TAR header is wrong, and we have
+      to fail.
+  """
+  pass
+
+
+class SquashedDockerImage(TarImageFormatter):
+  """ Image formatter which produces a squashed image compatible with the `docker load`
+      command.
+  """
+
+  def stream_generator(self, namespace, repository, tag, synthetic_image_id,
+                       layer_json, get_image_iterator, get_layer_iterator):
+    # Docker import V1 Format (.tar):
+    #  repositories - JSON file containing a repo -> tag -> image map
+    #  {image ID folder}:
+    #     json - The layer JSON
+    #     layer.tar - The TARed contents of the layer
+    #     VERSION - The docker import version: '1.0'
+    layer_merger = StreamLayerMerger(get_layer_iterator)
+
+    # Yield the repositories file:
+    synthetic_layer_info = {}
+    synthetic_layer_info[tag + '.squash'] = synthetic_image_id
+
+    hostname = app.config['SERVER_HOSTNAME']
+    repositories = {}
+    repositories[hostname + '/' + namespace + '/' + repository] = synthetic_layer_info
+
+    yield self.tar_file('repositories', json.dumps(repositories))
+
+    # Yield the image ID folder.
+    yield self.tar_folder(synthetic_image_id)
+
+    # Yield the JSON layer data.
+    layer_json = self._build_layer_json(layer_json, synthetic_image_id)
+    yield self.tar_file(synthetic_image_id + '/json', json.dumps(layer_json))
+
+    # Yield the VERSION file.
+    yield self.tar_file(synthetic_image_id + '/VERSION', '1.0')
+
+    # Yield the merged layer data's header.
+    estimated_file_size = 0
+    for image in get_image_iterator():
+      estimated_file_size += image.storage.uncompressed_size
+
+    yield self.tar_file_header(synthetic_image_id + '/layer.tar', estimated_file_size)
+
+    # Yield the contents of the merged layer.
+    yielded_size = 0
+    for entry in layer_merger.get_generator():
+      yield entry
+      yielded_size += len(entry)
+
+    # If the yielded size is more than the estimated size (which is unlikely but possible), then
+    # raise an exception since the tar header will be wrong.
+    if yielded_size > estimated_file_size:
+      raise FileEstimationException()
+
+    # If the yielded size is less than the estimated size (which is likely), fill the rest with
+    # zeros.
+    if yielded_size < estimated_file_size:
+      to_yield = estimated_file_size - yielded_size
+      while to_yield > 0:
+        yielded = min(to_yield, GZIP_BUFFER_SIZE)
+        yield '\0' * yielded
+        to_yield -= yielded
+
+    # Yield any file padding to 512 bytes that is necessary.
+    yield self.tar_file_padding(estimated_file_size)
+
+    # Last two records are empty in TAR spec.
+    yield '\0' * 512
+    yield '\0' * 512
+
+
+  def _build_layer_json(self, layer_json, synthetic_image_id):
+    updated_json = copy.deepcopy(layer_json)
+    updated_json['id'] = synthetic_image_id
+
+    if 'parent' in updated_json:
+      del updated_json['parent']
+
+    if 'config' in updated_json and 'Image' in updated_json['config']:
+      updated_json['config']['Image'] = synthetic_image_id
+
+    if 'container_config' in updated_json and 'Image' in updated_json['container_config']:
+      updated_json['container_config']['Image'] = synthetic_image_id
+
+    return updated_json
+

formats/tarimageformatter.py

@@ -0,0 +1,46 @@
+import tarfile
+from util.gzipwrap import GzipWrap
+
+class TarImageFormatter(object):
+  """ Base class for classes which produce a TAR containing image and layer data. """
+
+  def build_stream(self, namespace, repository, tag, synthetic_image_id, layer_json,
+                   get_image_iterator, get_layer_iterator):
+    """ Builds and streams a synthetic .tar.gz that represents the formatted TAR created by this
+        class's implementation.
+    """
+    return GzipWrap(self.stream_generator(namespace, repository, tag,
+                                             synthetic_image_id, layer_json,
+                                             get_image_iterator, get_layer_iterator))
+
+  def stream_generator(self, namespace, repository, tag, synthetic_image_id,
+                       layer_json, get_image_iterator, get_layer_iterator):
+    raise NotImplementedError
+
+  def tar_file(self, name, contents):
+    """ Returns the TAR binary representation for a file with the given name and file contents. """
+    length = len(contents)
+    tar_data = self.tar_file_header(name, length)
+    tar_data += contents
+    tar_data += self.tar_file_padding(length)
+    return tar_data
+
+  def tar_file_padding(self, length):
+    """ Returns TAR file padding for file data of the given length. """
+    if length % 512 != 0:
+      return '\0' * (512 - (length % 512))
+
+    return ''
+
+  def tar_file_header(self, name, file_size):
+    """ Returns TAR file header data for a file with the given name and size. """
+    info = tarfile.TarInfo(name=name)
+    info.type = tarfile.REGTYPE
+    info.size = file_size
+    return info.tobuf()
+
+  def tar_folder(self, name):
+    """ Returns TAR file header data for a folder with the given name. """
+    info = tarfile.TarInfo(name=name)
+    info.type = tarfile.DIRTYPE
+    return info.tobuf()