python-tuf/tuf/formats.py

"""
<Program Name>
  formats.py

<Author>
  Geremy Condra
  Vladimir Diaz <vladimir.v.diaz@gmail.com>

<Started>
  Refactored April 30, 2012. -Vlad

<Copyright>
  See LICENSE for licensing information.

<Purpose>
  A central location for all format-related checking of TUF objects.
  Note: 'formats.py' depends heavily on 'schema.py', so the 'schema.py'
  module should be read and understood before tackling this module.

  'formats.py' can be broken down into three sections.  (1) Schemas and object
  matching.  (2) Classes that represent Role Metadata and help produce correctly
  formatted files.  (3) Functions that help produce or verify TUF objects.

  The first section deals with schemas and object matching based on format.
  There are two ways of checking the format of objects.  The first method
  raises a 'tuf.FormatError' exception if the match fails and the other
  returns a Boolean result.

  tuf.formats.<SCHEMA>.check_match(object)
  tuf.formats.<SCHEMA>.matches(object)

  Example:
  rsa_key = {'keytype': 'rsa'
             'keyid': 34892fc465ac76bc3232fab 
             'keyval': {'public': 'public_key',
                        'private': 'private_key'}

  tuf.formats.RSAKEY_SCHEMA.check_match(rsa_key)
  tuf.formats.RSAKEY_SCHEMA.matches(rsa_key)

  In this example, if a dict key or dict value is missing or incorrect,
  the match fails.  There are numerous variations of object checking
  provided by 'formats.py' and 'schema.py'.

  The second section deals with the role metadata classes.  There are
  multiple top-level roles, each with differing metadata formats.
  Example:
  
  root_object = tuf.formats.RootFile.from_metadata(root_metadata_file)
  targets_metadata = tuf.formats.TargetsFile.make_metadata(...)

  The input and output of these classes are checked against their respective
  schema to ensure correctly formatted metadata.

  The last section contains miscellaneous functions related to the format of
  TUF objects.
  Example: 
  
  signable_object = make_signable(unsigned_object)

"""


import binascii
import calendar
import re
import string
import time

import tuf.schema

SCHEMA = tuf.schema


# Note that in the schema definitions below, the 'SCHEMA.Object' types allow
# additional keys which are not defined. Thus, any additions to them will be
# easily backwards compatible with clients that are already deployed.

# A date in 'YYYY-MM-DD HH:MM:SS' format.
TIME_SCHEMA = SCHEMA.RegularExpression(r'\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}')

# A hexadecimal value in '23432df87ab..' format.
HASH_SCHEMA = SCHEMA.RegularExpression(r'[a-fA-F0-9]+')

# A dict in {'sha256': '23432df87ab..', 'sha512': '34324abc34df..', ...} format.
HASHDICT_SCHEMA = SCHEMA.DictOf(
  key_schema=SCHEMA.AnyString(),
  value_schema=HASH_SCHEMA)

# A hexadecimal value in '23432df87ab..' format.
HEX_SCHEMA = SCHEMA.RegularExpression(r'[a-fA-F0-9]+')

# A key identifier (e.g., a hexadecimal value identifying an RSA key).
KEYID_SCHEMA = HASH_SCHEMA
KEYIDS_SCHEMA = SCHEMA.ListOf(KEYID_SCHEMA)

# The method used for a generated signature (e.g., 'evp').
SIG_METHOD_SCHEMA = SCHEMA.AnyString()

# A relative file path (e.g., 'metadata/root/').
RELPATH_SCHEMA = SCHEMA.AnyString()
RELPATHS_SCHEMA = SCHEMA.ListOf(RELPATH_SCHEMA)

# An absolute path.
PATH_SCHEMA = SCHEMA.AnyString()
PATHS_SCHEMA = SCHEMA.ListOf(PATH_SCHEMA)

# Uniform Resource Locator identifier (e.g., 'https://www.updateframework.com/').
URL_SCHEMA = SCHEMA.AnyString()

# A dictionary holding version information.
VERSION_SCHEMA = SCHEMA.Object(
  object_name='version',
  major=SCHEMA.Integer(lo=0),
  minor=SCHEMA.Integer(lo=0),
  fix=SCHEMA.Integer(lo=0))

# An integer representing length.  Must be 0 and greater.
LENGTH_SCHEMA = SCHEMA.Integer(lo=0)

# A string representing a named object.
NAME_SCHEMA = SCHEMA.AnyString()

# A value that is either True or False, on or off, etc.
TOGGLE_SCHEMA = SCHEMA.Boolean()

# A role's threshold value (i.e., the minimum number
# of signatures required to sign a metadata file).
# Must be 1 and greater.
THRESHOLD_SCHEMA = SCHEMA.Integer(lo=1)

# A string representing a role's name. 
ROLENAME_SCHEMA = SCHEMA.AnyString()

# The minimum number of bits for an RSA key.  Must be 2048 bits and greater.
RSAKEYBITS_SCHEMA = SCHEMA.Integer(lo=2048)

# A string representing a password.
PASSWORD_SCHEMA = SCHEMA.AnyString()

# A list of passwords.
PASSWORDS_SCHEMA = SCHEMA.ListOf(PASSWORD_SCHEMA)

# The actual values of a key, as opposed to meta data such as a key type and
# key identifier ('rsa', 233df889cb).  For RSA keys, the key value is a pair of
# public and private keys in PEM Format stored as strings.
KEYVAL_SCHEMA = SCHEMA.Object(
  object_name='keyval',
  public=SCHEMA.AnyString(),
  private=SCHEMA.AnyString())

# A generic key.  All TUF keys should be saved to metadata files in this format.
KEY_SCHEMA = SCHEMA.Object(
  object_name='key',
  keytype=SCHEMA.AnyString(),
  keyval=KEYVAL_SCHEMA)

# An RSA key.
RSAKEY_SCHEMA = SCHEMA.Object(
  object_name='rsakey',
  keytype=SCHEMA.String('rsa'),
  keyid=KEYID_SCHEMA,
  keyval=KEYVAL_SCHEMA)

# Info that describes both metadata and target files.
# This schema allows the storage of multiple hashes for the same file
# (e.g., sha256 and sha512 may be computed for the same file and stored).
FILEINFO_SCHEMA = SCHEMA.Object(
  object_name='fileinfo',
  length=LENGTH_SCHEMA,
  hashes=HASHDICT_SCHEMA,
  custom=SCHEMA.Optional(SCHEMA.Object()))

# A dict holding the information for a particular file.  The keys hold the
# relative file path and the values the relevant file information.
FILEDICT_SCHEMA = SCHEMA.DictOf(
  key_schema=RELPATH_SCHEMA,
  value_schema=FILEINFO_SCHEMA)

# A dict holding a target file.
TARGETFILE_SCHEMA = SCHEMA.Object(
  object_name='targetfile',
  filepath=RELPATH_SCHEMA,
  fileinfo=FILEINFO_SCHEMA)
TARGETFILES_SCHEMA = SCHEMA.ListOf(TARGETFILE_SCHEMA)

# A single signature of an object.  Indicates the signature, the id of the
# signing key, and the signing method.
# I debated making the signature schema not contain the key id and instead have
# the signatures of a file be a dictionary with the key being the keyid and the
# value being the signature schema without the keyid. That would be under
# the argument that a key should only be able to sign a file once. However,
# one can imagine that maybe a key wants to sign multiple times with different
# signature methods.
SIGNATURE_SCHEMA = SCHEMA.Object(
  object_name='signature',
  keyid=KEYID_SCHEMA,
  method=SIG_METHOD_SCHEMA,
  sig=HEX_SCHEMA)

# A schema holding the result of checking the signatures of a particular
# 'SIGNABLE_SCHEMA' role.
# For example, how many of the signatures for the 'Target' role are
# valid?  This SCHEMA holds this information.  See 'sig.py' for
# more information.
SIGNATURESTATUS_SCHEMA = SCHEMA.Object(
  object_name='signaturestatus',
  threshold=SCHEMA.Integer(),
  good_sigs=SCHEMA.ListOf(KEYID_SCHEMA),
  bad_sigs=SCHEMA.ListOf(KEYID_SCHEMA),
  unknown_sigs=SCHEMA.ListOf(KEYID_SCHEMA),
  untrusted_sigs=SCHEMA.ListOf(KEYID_SCHEMA),
  unknown_method_sigs=SCHEMA.ListOf(KEYID_SCHEMA))

# A signable object.  Holds the signing role and its associated signatures.
SIGNABLE_SCHEMA = SCHEMA.Object(
  object_name='signable',
  signed=SCHEMA.Any(),
  signatures=SCHEMA.ListOf(SIGNATURE_SCHEMA))

# A dict where the dict keys hold a keyid and the dict values a key object.
KEYDICT_SCHEMA = SCHEMA.DictOf(
  key_schema=KEYID_SCHEMA,
  value_schema=KEY_SCHEMA)

# The format used by the key database to store keys.  The dict keys hold a key
# identifier and the dict values any object.  The key database should store
# key objects in the values (e.g., 'RSAKEY_SCHEMA', 'DSAKEY_SCHEMA').
KEYDB_SCHEMA = SCHEMA.DictOf(
  key_schema=KEYID_SCHEMA,
  value_schema=SCHEMA.Any())

# The format of the resulting "scp config dict" after extraction from the
# push configuration file (i.e., push.cfg).  In the case of a config file
# utilizing the scp transfer module, it must contain the 'general' and 'scp'
# sections, where 'general' must contain a 'transfer_module' and
# 'metadata_path' entry, and 'scp' the 'host', 'user', 'identity_file', and
# 'remote_directory' entries.  See 'tuf/pushtools/pushtoolslib.py' and
# 'tuf/pushtools/push.py'.
SCPCONFIG_SCHEMA = SCHEMA.Object(
  object_name='scp_config',
  general=SCHEMA.Object(
    object_name='[general]',
    transfer_module=SCHEMA.String('scp'),
    metadata_path=PATH_SCHEMA,
    targets_directory=PATH_SCHEMA),
  scp=SCHEMA.Object(
    object_name='[scp]',
    host=URL_SCHEMA,
    user=NAME_SCHEMA,
    identity_file=PATH_SCHEMA,
    remote_directory=PATH_SCHEMA))

# The format of the resulting "receive config dict" after extraction from the
# receive configuration file (i.e., receive.cfg).  The receive config file
# must contain a 'general' section, and this section the 'pushroots',
# 'repository_directory', 'metadata_directory', 'targets_directory', and
# 'backup_directory' entries.
# see 'tuf/pushtools/pushtoolslib.py' and 'tuf/pushtools/receive/receive.py'
RECEIVECONFIG_SCHEMA = SCHEMA.Object(
  object_name='receive_config',
  general=SCHEMA.Object(
    object_name='[general]',
    pushroots=SCHEMA.ListOf(PATH_SCHEMA),
    repository_directory=PATH_SCHEMA,
    metadata_directory=PATH_SCHEMA,
    targets_directory=PATH_SCHEMA,
    backup_directory=PATH_SCHEMA)) 

# Role object in {'keyids': [keydids..], 'name': 'ABC', 'threshold': 1,
# 'paths':[filepaths..]} # format.
ROLE_SCHEMA = SCHEMA.Object(
  object_name='role',
  keyids=SCHEMA.ListOf(KEYID_SCHEMA),
  name=SCHEMA.Optional(ROLENAME_SCHEMA),
  threshold=THRESHOLD_SCHEMA,
  paths=SCHEMA.Optional(SCHEMA.ListOf(RELPATH_SCHEMA)))

# A dict of roles where the dict keys are role names and the dict values holding 
# the role data/information.
ROLEDICT_SCHEMA = SCHEMA.DictOf(
  key_schema=ROLENAME_SCHEMA,
  value_schema=ROLE_SCHEMA)

# Like ROLEDICT_SCHEMA, except that ROLE_SCHEMA instances are stored in order.
ROLELIST_SCHEMA = SCHEMA.ListOf(ROLE_SCHEMA)

# The root: indicates root keys and top-level roles.
ROOT_SCHEMA = SCHEMA.Object(
  object_name='root',
  _type=SCHEMA.String('Root'),
  ts=TIME_SCHEMA,
  expires=TIME_SCHEMA,
  keys=KEYDICT_SCHEMA,
  roles=ROLEDICT_SCHEMA)

# Targets. Indicates targets and delegates target paths to other roles.
TARGETS_SCHEMA = SCHEMA.Object(
  object_name='targets',
  _type=SCHEMA.String('Targets'),
  ts=TIME_SCHEMA,
  expires=TIME_SCHEMA,
  targets=FILEDICT_SCHEMA,
  delegations=SCHEMA.Optional(SCHEMA.Object(
    keys=KEYDICT_SCHEMA,
    roles=ROLELIST_SCHEMA)))

# A Release: indicates the latest versions of all metadata (except timestamp).
RELEASE_SCHEMA = SCHEMA.Object(
  object_name='release',
  _type=SCHEMA.String('Release'),
  ts=TIME_SCHEMA,
  expires=TIME_SCHEMA,
  meta=FILEDICT_SCHEMA)

# A Timestamp: indicates the latest version of the release file.
TIMESTAMP_SCHEMA = SCHEMA.Object(
  object_name='timestamp',
  _type=SCHEMA.String('Timestamp'),
  ts=TIME_SCHEMA,
  expires=TIME_SCHEMA,
  meta=FILEDICT_SCHEMA)

# A schema containing information a repository mirror may require,
# such as a url, the path of the directory metadata files, etc.
MIRROR_SCHEMA = SCHEMA.Object(
  object_name='mirror',
  url_prefix=URL_SCHEMA,
  metadata_path=RELPATH_SCHEMA,
  targets_path=RELPATH_SCHEMA,
  confined_target_dirs=SCHEMA.ListOf(RELPATH_SCHEMA),
  custom=SCHEMA.Optional(SCHEMA.Object()))

# A dictionary of mirrors where the dict keys hold the mirror's name and
# and the dict values the mirror's data (i.e., 'MIRROR_SCHEMA').
# The repository class of 'updater.py' accepts dictionaries
# of this type provided by the TUF client.
MIRRORDICT_SCHEMA = SCHEMA.DictOf(
  key_schema=SCHEMA.AnyString(),
  value_schema=MIRROR_SCHEMA)

# A Mirrorlist: indicates all the live mirrors, and what documents they
# serve.
MIRRORLIST_SCHEMA = SCHEMA.Object(
  object_name='mirrorlist',
  _type=SCHEMA.String('Mirrors'),
  ts=TIME_SCHEMA,
  expires=TIME_SCHEMA,
  mirrors=SCHEMA.ListOf(MIRROR_SCHEMA))





class MetaFile(object):
  """
  <Purpose>
    Base class for all metadata file classes.
    Classes representing metadata files such as RootFile
    and ReleaseFile all inherit from MetaFile.  The
    __eq__, __ne__, perform 'equal' and 'not equal' comparisons
    between Metadata File objects.

  """

  info = None

  def __eq__(self, other):
    return isinstance(other, MetaFile) and self.info == other.info


  def __ne__(self, other):
    return not self.__eq__(other)


  def __getattr__(self, name):
    """
      Allow all metafile objects to have their interesting attributes
      referred to directly without the info dict. The info dict is just
      to be able to do the __eq__ comparison generically.
    
    """
    
    if name in self.info:
      return self.info[name]
    raise AttributeError, name





class TimestampFile(MetaFile):
  def __init__(self, timestamp, expires, filedict):
    self.info = {}
    self.info['ts'] = timestamp
    self.info['expires'] = expires
    self.info['meta'] = filedict


  @staticmethod
  def from_metadata(object):
    # Is 'object' a Timestamp metadata file?
    # Raise tuf.FormatError if not.
    TIMESTAMP_SCHEMA.check_match(object) 

    timestamp = parse_time(object['ts'])
    expires = parse_time(object['expires'])
    filedict = object['meta']
    return TimestampFile(timestamp, expires, filedict)
    
    
  @staticmethod
  def make_metadata(filedict):
    result = {'_type' : 'Timestamp'}
    # TODO: set the expires time another way.
    result['ts'] = format_time(time.time())
    result['expires'] = format_time(time.time() + 3600 * 24 * 365)
    result['meta'] = filedict

    # Is 'result' a Timestamp metadata file?
    # Raise 'tuf.FormatError' if not.
    TIMESTAMP_SCHEMA.check_match(result)

    return result





class RootFile(MetaFile):
  def __init__(self, timestamp, expires, keys, roles):
    self.info = {}
    self.info['ts'] = timestamp
    self.info['expires'] = expires
    self.info['keys'] = keys
    self.info['roles'] = roles


  @staticmethod
  def from_metadata(object):
    # Is 'object' a Root metadata file?
    # Raise tuf.FormatError if not.
    ROOT_SCHEMA.check_match(object) 
    
    timestamp = parse_time(object['ts'])
    expires = parse_time(object['expires'])
    keys = object['keys']
    roles = object['roles']
    
    return RootFile(timestamp, expires, keys, roles)


  @staticmethod
  def make_metadata(keydict, roledict, expiration_seconds):
    # Is 'expiration_seconds' properly formatted?
    # Raise 'tuf.FormatError' if not.
    LENGTH_SCHEMA.check_match(expiration_seconds)
    
    result = {'_type' : 'Root'}
    # TODO: set the expires time another way.
    result['ts'] = format_time(time.time())
    result['expires'] = format_time(time.time() + expiration_seconds)
    result['keys'] = keydict
    result['roles'] = roledict
    
    # Is 'result' a Root metadata file?
    # Raise 'tuf.FormatError' if not.
    ROOT_SCHEMA.check_match(result)
    
    return result





class ReleaseFile(MetaFile):
  def __init__(self, timestamp, expires, filedict):
    self.info = {}
    self.info['ts'] = timestamp
    self.info['expires'] = expires
    self.info['meta'] = filedict


  @staticmethod
  def from_metadata(object):
    # Is 'object' a Release metadata file?
    # Raise tuf.FormatError if not.
    RELEASE_SCHEMA.check_match(object)
    
    timestamp = parse_time(object['ts'])
    expires = parse_time(object['expires'])
    filedict = object['meta']
    
    return ReleaseFile(timestamp, expires, filedict)


  @staticmethod
  def make_metadata(filedict):
    result = {'_type' : 'Release'}
    # TODO: set the expires time another way.
    result['ts'] = format_time(time.time())
    result['expires'] = format_time(time.time() + 3600 * 24 * 365)
    result['meta'] = filedict

    # Is 'result' a Release metadata file?
    # Raise 'tuf.FormatError' if not.
    RELEASE_SCHEMA.check_match(result)
    
    return result





class TargetsFile(MetaFile):
  def __init__(self, timestamp, expires, filedict=None, delegations=None):
    if filedict is None:
      filedict = {}
    if delegations is None:
      delegations = {}
    self.info = {}
    self.info['ts'] = timestamp
    self.info['expires'] = expires
    self.info['targets'] = filedict
    self.info['delegations'] = delegations


  @staticmethod
  def from_metadata(object):
    # Is 'object' a Targets metadata file?
    # Raise tuf.FormatError if not.
    TARGETS_SCHEMA.check_match(object)
    
    timestamp = parse_time(object['ts'])
    expires = parse_time(object['expires'])
    filedict = object.get('targets')
    delegations = object.get('delegations')
    
    return TargetsFile(timestamp, expires, filedict, delegations)


  @staticmethod
  def make_metadata(filedict=None, delegations=None):
    if filedict is None and delegations is None:
      raise tuf.Error('We don\'t allow completely empty targets metadata.')

    result = {'_type' : 'Targets'}
    # TODO: set the expires time another way.
    result['ts'] = format_time(time.time())
    result['expires'] = format_time(time.time() + 3600 * 24 * 365)
    if filedict is not None:
      result['targets'] = filedict
    if delegations is not None:
      result['delegations'] = delegations

    # Is 'result' a Targets metadata file?
    # Raise 'tuf.FormatError' if not.
    TARGETS_SCHEMA.check_match(result)
    
    return result





class MirrorsFile(MetaFile):
  def __init__(self, timestamp, expires):
    self.info = {}
    self.info['ts'] = timestamp
    self.info['expires'] = expires


  @staticmethod
  def from_metadata(object):
    raise NotImplementedError


  @staticmethod
  def make_metadata():
    raise NotImplementedError





# A dict holding the recognized schemas for the top-level roles.
SCHEMAS_BY_TYPE = {
  'Root' : ROOT_SCHEMA,
  'Targets' : TARGETS_SCHEMA,
  'Release' : RELEASE_SCHEMA,
  'Timestamp' : TIMESTAMP_SCHEMA,
  'Mirrors' : MIRRORLIST_SCHEMA}

# A dict holding the recognized class names for the top-level roles.
# That is, the role classes listed in this module (e.g., class TargetsFile()).
ROLE_CLASSES_BY_TYPE = {
  'Root' : RootFile,
  'Targets' : TargetsFile,
  'Release' : ReleaseFile,
  'Timestamp' : TimestampFile,
  'Mirrors' : MirrorsFile}





def format_time(timestamp):
  """
  <Purpose>
    Encode 'timestamp' in YYYY-MM-DD HH:MM:SS format.
    'timestamp' is a Unix timestamp value.  For example, it is the time
    format returned by calendar.timegm(). 

    >>> format_time(499137720)
    '1985-10-26 01:22:00'

  <Arguments>
    timestamp:
      The time to format.

  <Exceptions>
    tuf.Error, if 'timestamp' is invalid.

  <Side Effects>
    None.

  <Returns>
    A string in 'YYYY-MM-DD HH:MM:SS' format.

  """
   
  try:
    return time.strftime('%Y-%m-%d %H:%M:%S', time.gmtime(timestamp))
  except (ValueError, TypeError):
    raise tuf.FormatError('Invalid argument value')




def parse_time(string):
  """
  <Purpose>
    Parse 'string' in YYYY-MM-DD HH:MM:SS format.

  <Arguments>
    string:
      A string representing the time (e.g., '1985-10-26 01:20:00').

  <Exceptions>
    tuf.FormatError, if parsing 'string' fails.

  <Side Effects>
    None.

  <Returns>
    A timestamp (e.g., 499137660).

  """
  
  # Is 'string' properly formatted?
  # Raise 'tuf.FormatError' if there is a mismatch.
  TIME_SCHEMA.check_match(string)
  
  try:
    return calendar.timegm(time.strptime(string, '%Y-%m-%d %H:%M:%S'))
  except ValueError:
    raise tuf.FormatError('Malformed time '+repr(string))





def format_base64(data):
  """
  <Purpose>
    Return the base64 encoding of 'data' with whitespace
    and '=' signs omitted.

  <Arguments>
    data:
      A string or buffer of data to convert.

  <Exceptions>
    tuf.FormatError, if the base64 encoding fails or the argument
    is invalid.

  <Side Effects>
    None.

  <Returns>
    A base64-encoded string.

  """
  
  try:
    return binascii.b2a_base64(data).rstrip('=\n ')
  except (TypeError, binascii.Error), e:
    raise tuf.FormatError('Invalid base64 encoding: '+str(e))





def parse_base64(base64_string):
  """
  <Purpose>
    Parse a base64 encoding with whitespace and '=' signs omitted.
  
  <Arguments>
    base64_string:
      A string holding a base64 value.

  <Exceptions>
    tuf.FormatError, if 'base64_string' cannot be parsed due to
    an invalid base64 encoding.

  <Side Effects>
    None.

  <Returns>
    A byte string representing the parsed based64 encoding of
    'base64_string'.

  """

  if not isinstance(base64_string, basestring):
    message = 'Invalid argument: '+repr(base64_string)
    raise tuf.FormatError(message)

  extra = len(base64_string) % 4
  if extra:
    padding = '=' * (4 - extra)
    base64_string = base64_string + padding

  try:
    return binascii.a2b_base64(base64_string)
  except (TypeError, binascii.Error), e:
    raise tuf.FormatError('Invalid base64 encoding: '+str(e))





def make_signable(object):
  """
  <Purpose>
    Return the role metadata 'object' in 'SIGNABLE_SCHEMA' format.
    'object' is added to the 'signed' key, and an empty list
    initialized to the 'signatures' key.  The caller adds signatures
    to this second field.
    Note: check_signable_object_format() should be called after
    make_signable() and signatures added to ensure the final
    signable object has a valid format (i.e., a signable containing
    a supported role metadata).

  <Arguments>
    object:
      A role schema dict (e.g., 'ROOT_SCHEMA', 'RELEASE_SCHEMA'). 

  <Exceptions>
    None.

  <Side Effects>
    None.

  <Returns>
    A dict in 'SIGNABLE_SCHEMA' format.

  """

  if not isinstance(object, dict) or 'signed' not in object:
    return { 'signed' : object, 'signatures' : [] }
  else:
    return object





def make_fileinfo(length, hashes, custom=None):
  """
  <Purpose>
    Create a dictionary conformant to 'FILEINFO_SCHEMA'.
    This dict describes both metadata and target files.

  <Arguments>
    length:
      An integer representing the size of the file.

    hashes:
      A dict of hashes in 'HASHDICT_SCHEMA' format, which has the form:
       {'sha256': 123df8a9b12, 'sha512': 324324dfc121, ...}

    custom:
      An optional object providing additional information about the file.

  <Exceptions>
    tuf.FormatError, if the 'FILEINFO_SCHEMA' to be returned
    does not have the correct format.

  <Side Effects>
    If any of the arguments are incorrectly formatted, the dict
    returned will be checked for formatting errors, and if found,
    will raise a 'tuf.FormatError' exception.

  <Returns>
    A dictionary conformant to 'FILEINFO_SCHEMA', representing the file
    information of a metadata or target file.

  """

  fileinfo = {'length' : length, 'hashes' : hashes}
  if custom is not None:
    fileinfo['custom'] = custom

  # Raise 'tuf.FormatError' if the check fails.
  FILEINFO_SCHEMA.check_match(fileinfo)

  return fileinfo





def make_role_metadata(keyids, threshold, name=None, paths=None):
  """
  <Purpose>
    Create a dictionary conforming to 'tuf.formats.ROLE_SCHEMA',
    representing the role with 'keyids', 'threshold', and 'paths'
    as field values.  'paths' is optional (i.e., used only by the
    'Target' role).

  <Arguments>
    keyids: a list of key ids.

    threshold:
      An integer denoting the number of required keys
      for the signing role.

    name:
      A string that is the name of this role.

    paths:
      The 'Target' role stores the paths of target files
      in its metadata file.  'paths' is a list of
      file paths.
  
  <Exceptions>
    tuf.FormatError, if the returned role meta is
    formatted incorrectly.

  <Side Effects>
    If any of the arguments do not have a proper format, a 
    tuf.formats exception is raised when the 'ROLE_SCHEMA' dict
    is created.

  <Returns>
    A properly formatted role meta dict, conforming to
    'ROLE_SCHEMA'.
    
  """

  role_meta = {}
  role_meta['keyids'] = keyids
  role_meta['threshold'] = threshold

  if name is not None:
    role_meta['name'] = name

  if paths is not None:
    role_meta['paths'] = paths

  # Does 'role_meta' have the correct type?
  # This check ensures 'role_meta' conforms to
  # tuf.formats.ROLE_SCHEMA.
  ROLE_SCHEMA.check_match(role_meta)

  return role_meta





def get_role_class(expected_rolename):
  """
  <Purpose>
    Return the role class corresponding to
    'expected_rolename'.  The role name returned
    by expected_meta_rolename() should be the name
    passed as an argument to this function.  If
    'expected_rolename' is 'Root', the class
    RootFile is returned.

  <Arguments>
    expected_rolename:
      The role name used to determine which role class
      to return.

  <Exceptions>
    tuf.FormatError, if 'expected_rolename' is not a
    supported role.

  <Side Effects>
    None.

  <Returns>
    The class corresponding to 'expected_rolename'.
    E.g., 'Release' as an argument to this function causes
    'ReleaseFile' to be returned. 

  """
 
  # Does 'expected_rolename' have the correct type?
  # This check ensures 'expected_rolename' conforms to
  # 'tuf.formats.NAME_SCHEMA'.
  # Raise 'tuf.FormatError' if there is a mismatch.
  NAME_SCHEMA.check_match(expected_rolename)
  
  try:
    role_class = ROLE_CLASSES_BY_TYPE[expected_rolename]
  except KeyError:
    raise tuf.FormatError(repr(expected_rolename)+' not supported.')
  else:
    return role_class





def expected_meta_rolename(meta_rolename):
  """
  <Purpose>
    Ensure 'meta_rolename' is properly formatted.
    'targets' is returned as 'Targets'.
    'targets role1' is returned as 'Targets Role1'.

    The words in the string (i.e., separated by whitespace)
    are capitalized.

  <Arguments>
    meta_rolename:
      A string representing the rolename.
      E.g., 'root', 'targets'.

  <Exceptions>
    tuf.FormatError, if 'meta_rolename' is improperly formatted.

  <Side Effects>
    None.

  <Returns>
    A string (e.g., 'Root', 'Targets').
    
  """
   
  # Does 'meta_rolename' have the correct type?
  # This check ensures 'meta_rolename' conforms to
  # 'tuf.formats.NAME_SCHEMA'.
  # Raise 'tuf.FormatError' if there is a mismatch.
  NAME_SCHEMA.check_match(meta_rolename)
  
  return string.capwords(meta_rolename)





def check_signable_object_format(object):
  """
  <Purpose>
    Ensure 'object' is properly formatted, conformant to
    'tuf.formats.SIGNABLE_SCHEMA'.  Return the signing role on success.
    Note: The 'signed' field of a 'SIGNABLE_SCHEMA' is checked against
    tuf.schema.Any().  The 'signed' field, however, should actually
    hold one of the supported role schemas (e.g., 'ROOT_SCHEMA',
    'TARGETS_SCHEMA').  The role schemas all differ in their format, so this
    function determines exactly which schema is listed in the 'signed'
    field.

  <Arguments>
    object:
     The object compare against 'SIGNABLE.SCHEMA'. 

  <Exceptions>
    tuf.FormatError, if 'object' does not have the correct format.

  <Side Effects>
    None.

  <Returns>
    A string representing the signing role (e.g., 'root', 'targets').
    The role string is returned with characters all lower case.

  """
  
  # Does 'object' have the correct type?
  # This check ensures 'object' conforms to
  # 'tuf.formats.SIGNABLE_SCHEMA'.
  SIGNABLE_SCHEMA.check_match(object)

  try:
    role_type = object['signed']['_type']
  except (KeyError, TypeError):
    raise tuf.FormatError('Untyped object')
  try:
    schema = SCHEMAS_BY_TYPE[role_type]
  except KeyError:
    raise tuf.FormatError('Unrecognized type '+repr(role_type))
  
  # 'tuf.FormatError' raised if 'object' does not have a properly
  # formatted role schema.
  schema.check_match(object['signed'])

  return role_type.lower()





def _canonical_string_encoder(string):
  """
  <Purpose>
    Encode 'string' to canonical string format.
    
  <Arguments>
    string:
      The string to encode.

  <Exceptions>
    None.

  <Side Effects>
    None.

  <Returns>
    A string with the canonical-encoded 'string' embedded.

  """

  string = '"%s"' % re.sub(r'(["\\])', r'\\\1', string)
  if isinstance(string, unicode):
    return string.encode('utf-8')
  else:
    return string





def _encode_canonical(object, output_function):
  # Helper for encode_canonical.  Older versions of json.encoder don't
  # even let us replace the separators.

  if isinstance(object, basestring):
    output_function(_canonical_string_encoder(object))
  elif object is True:
    output_function("true")
  elif object is False:
    output_function("false")
  elif object is None:
    output_function("null")
  elif isinstance(object, (int, long)):
    output_function(str(object))
  elif isinstance(object, (tuple, list)):
    output_function("[")
    if len(object):
      for item in object[:-1]:
        _encode_canonical(item, output_function)
        output_function(",")
      _encode_canonical(object[-1], output_function)
    output_function("]")
  elif isinstance(object, dict):
    output_function("{")
    if len(object):
      items = object.items()
      items.sort()
      for key, value in items[:-1]:
        output_function(_canonical_string_encoder(key))
        output_function(":")
        _encode_canonical(value, output_function)
        output_function(",")
      key, value = items[-1]
      output_function(_canonical_string_encoder(key))
      output_function(":")
      _encode_canonical(value, output_function)
    output_function("}")
  else:
    raise tuf.FormatError('I cannot encode '+repr(object))





def encode_canonical(object, output_function=None):
  """
  <Purpose>
    Encode 'object' in canonical JSON form, as specified at
    http://wiki.laptop.org/go/Canonical_JSON .  It's a restricted
    dialect of JSON in which keys are always lexically sorted,
    there is no whitespace, floats aren't allowed, and only quote
    and backslash get escaped.  The result is encoded in UTF-8,
    and the resulting bits are passed to output_function (if provided),
    or joined into a string and returned.

    Note: This function should be called prior to computing the hash or
    signature of a JSON object in TUF.  For example, generating a signature
    of a signing role object such as 'ROOT_SCHEMA' is required to ensure
    repeatable hashes are generated across different json module versions
    and platforms.  Code elsewhere is free to dump JSON objects in any format
    they wish (e.g., utilizing indentation and single quotes around object
    keys).  These objects are only required to be in "canonical JSON" format
    when their hashes or signatures are needed.

    >>> encode_canonical("")
    '""'
    >>> encode_canonical([1, 2, 3])
    '[1,2,3]'
    >>> encode_canonical([])
    '[]'
    >>> encode_canonical({"A": [99]})
    '{"A":[99]}'
    >>> encode_canonical({"x" : 3, "y" : 2})
    '{"x":3,"y":2}'
  
  <Arguments>
    object:
      The object to be encoded.

    output_function:
      The result will be passed as arguments to 'output_function'
      (e.g., output_function('result')).

  <Exceptions>
    tuf.FormatError, if 'object' cannot be encoded or 'output_function'
    is not callable.

  <Side Effects>
    The results are fed to 'output_function()' if 'output_function' is set.  

  <Returns>
    A string representing the 'object' encoded in canonical JSON form.

  """

  result = None
  # If 'output_function' is unset, treat it as
  # appending to a list.
  if output_function is None:
    result = []
    output_function = result.append

  try:
    _encode_canonical(object, output_function)
  except TypeError, e:
    message = 'Could not encode '+repr(object)+': '+str(e)
    raise tuf.FormatError(message)

  # Return the encoded 'object' as a string.
  # Note: Implies 'output_function' is None,
  # otherwise results are sent to 'output_function'.
  if result is not None:
    return ''.join(result)





if __name__ == '__main__':
  # The interactive sessions of the documentation strings can
  # be tested by running formats.py as a standalone module.
  # python -B formats.py
  import doctest
  doctest.testmod()