# (c) 2014, James Tanner # # Ansible is free software: you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation, either version 3 of the License, or # (at your option) any later version. # # Ansible is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with Ansible. If not, see . # Make coding more python3-ish from __future__ import (absolute_import, division, print_function) __metaclass__ = type import os import shlex import shutil import sys import tempfile import random from io import BytesIO from subprocess import call from ansible.errors import AnsibleError from hashlib import sha256 from binascii import hexlify from binascii import unhexlify # Note: Only used for loading obsolete VaultAES files. All files are written # using the newer VaultAES256 which does not require md5 from hashlib import md5 try: from Crypto.Hash import SHA256, HMAC HAS_HASH = True except ImportError: HAS_HASH = False # Counter import fails for 2.0.1, requires >= 2.6.1 from pip try: from Crypto.Util import Counter HAS_COUNTER = True except ImportError: HAS_COUNTER = False # KDF import fails for 2.0.1, requires >= 2.6.1 from pip try: from Crypto.Protocol.KDF import PBKDF2 HAS_PBKDF2 = True except ImportError: HAS_PBKDF2 = False # AES IMPORTS try: from Crypto.Cipher import AES as AES HAS_AES = True except ImportError: HAS_AES = False # OpenSSL pbkdf2_hmac HAS_PBKDF2HMAC = False try: from cryptography.hazmat.primitives.hashes import SHA256 as c_SHA256 from cryptography.hazmat.primitives.kdf.pbkdf2 import PBKDF2HMAC from cryptography.hazmat.backends import default_backend HAS_PBKDF2HMAC = True except ImportError: pass from ansible.compat.six import PY3 from ansible.utils.unicode import to_unicode, to_bytes HAS_ANY_PBKDF2HMAC = HAS_PBKDF2 or HAS_PBKDF2HMAC CRYPTO_UPGRADE = "ansible-vault requires a newer version of pycrypto than the one installed on your platform. You may fix this with OS-specific commands such as: yum install python-devel; rpm -e --nodeps python-crypto; pip install pycrypto" b_HEADER = b'$ANSIBLE_VAULT' CIPHER_WHITELIST = frozenset((u'AES', u'AES256')) CIPHER_WRITE_WHITELIST=frozenset((u'AES256',)) # See also CIPHER_MAPPING at the bottom of the file which maps cipher strings # (used in VaultFile header) to a cipher class def check_prereqs(): if not HAS_AES or not HAS_COUNTER or not HAS_ANY_PBKDF2HMAC or not HAS_HASH: raise AnsibleError(CRYPTO_UPGRADE) class VaultLib: def __init__(self, password): self.b_password = to_bytes(password, errors='strict', encoding='utf-8') self.cipher_name = None self.b_version = b'1.1' def is_encrypted(self, data): """ Test if this is vault encrypted data :arg data: a byte str or unicode string to test whether it is recognized as vault encrypted data :returns: True if it is recognized. Otherwise, False. """ if to_bytes(data, errors='strict', encoding='utf-8').startswith(b_HEADER): return True return False def encrypt(self, data): """Vault encrypt a piece of data. :arg data: a utf-8 byte str or unicode string to encrypt. :returns: a utf-8 encoded byte str of encrypted data. The string contains a header identifying this as vault encrypted data and formatted to newline terminated lines of 80 characters. This is suitable for dumping as is to a vault file. """ b_data = to_bytes(data, errors='strict', encoding='utf-8') if self.is_encrypted(b_data): raise AnsibleError("input is already encrypted") if not self.cipher_name or self.cipher_name not in CIPHER_WRITE_WHITELIST: self.cipher_name = u"AES256" try: Cipher = CIPHER_MAPPING[self.cipher_name] except KeyError: raise AnsibleError(u"{0} cipher could not be found".format(self.cipher_name)) this_cipher = Cipher() # encrypt data b_enc_data = this_cipher.encrypt(b_data, self.b_password) # format the data for output to the file b_tmp_data = self._format_output(b_enc_data) return b_tmp_data def decrypt(self, data): """Decrypt a piece of vault encrypted data. :arg data: a string to decrypt. Since vault encrypted data is an ascii text format this can be either a byte str or unicode string. :returns: a byte string containing the decrypted data """ b_data = to_bytes(data, errors='strict', encoding='utf-8') if self.b_password is None: raise AnsibleError("A vault password must be specified to decrypt data") if not self.is_encrypted(b_data): raise AnsibleError("input is not encrypted") # clean out header b_data = self._split_header(b_data) # create the cipher object cipher_class_name = u'Vault{0}'.format(self.cipher_name) if cipher_class_name in globals() and self.cipher_name in CIPHER_WHITELIST: Cipher = globals()[cipher_class_name] this_cipher = Cipher() else: raise AnsibleError("{0} cipher could not be found".format(self.cipher_name)) # try to unencrypt data b_data = this_cipher.decrypt(b_data, self.b_password) if b_data is None: raise AnsibleError("Decryption failed") return b_data def _format_output(self, b_data): """ Add header and format to 80 columns :arg b_data: the encrypted and hexlified data as a byte string :returns: a byte str that should be dumped into a file. It's formatted to 80 char columns and has the header prepended """ if not self.cipher_name: raise AnsibleError("the cipher must be set before adding a header") header = b';'.join([b_HEADER, self.b_version, to_bytes(self.cipher_name, errors='strict', encoding='utf-8')]) tmpdata = [header] tmpdata += [b_data[i:i+80] for i in range(0, len(b_data), 80)] tmpdata += [b''] tmpdata = b'\n'.join(tmpdata) return tmpdata def _split_header(self, b_data): """Retrieve information about the Vault and clean the data When data is saved, it has a header prepended and is formatted into 80 character lines. This method extracts the information from the header and then removes the header and the inserted newlines. The string returned is suitable for processing by the Cipher classes. :arg b_data: byte str containing the data from a save file :returns: a byte str suitable for passing to a Cipher class's decrypt() function. """ # used by decrypt tmpdata = b_data.split(b'\n') tmpheader = tmpdata[0].strip().split(b';') self.b_version = tmpheader[1].strip() self.cipher_name = to_unicode(tmpheader[2].strip()) clean_data = b''.join(tmpdata[1:]) return clean_data class VaultEditor: def __init__(self, password): self.vault = VaultLib(password) def _shred_file_custom(self, tmp_path): """"Destroy a file, when shred (core-utils) is not available Unix `shred' destroys files "so that they can be recovered only with great difficulty with specialised hardware, if at all". It is based on the method from the paper "Secure Deletion of Data from Magnetic and Solid-State Memory", Proceedings of the Sixth USENIX Security Symposium (San Jose, California, July 22-25, 1996). We do not go to that length to re-implement shred in Python; instead, overwriting with a block of random data should suffice. See https://github.com/ansible/ansible/pull/13700 . """ file_len = os.path.getsize(tmp_path) if file_len > 0: # avoid work when file was empty max_chunk_len = min(1024*1024*2, file_len) passes = 3 with open(tmp_path, "wb") as fh: for _ in range(passes): fh.seek(0, 0) # get a random chunk of data, each pass with other length chunk_len = random.randint(max_chunk_len//2, max_chunk_len) data = os.urandom(chunk_len) for _ in range(0, file_len // chunk_len): fh.write(data) fh.write(data[:file_len % chunk_len]) assert(fh.tell() == file_len) # FIXME remove this assert once we have unittests to check its accuracy os.fsync(fh) def _shred_file(self, tmp_path): """Securely destroy a decrypted file Note standard limitations of GNU shred apply (For flash, overwriting would have no effect due to wear leveling; for other storage systems, the async kernel->filesystem->disk calls never guarantee data hits the disk; etc). Furthermore, if your tmp dirs is on tmpfs (ramdisks), it is a non-issue. Nevertheless, some form of overwriting the data (instead of just removing the fs index entry) is a good idea. If shred is not available (e.g. on windows, or no core-utils installed), fall back on a custom shredding method. """ if not os.path.isfile(tmp_path): # file is already gone return try: r = call(['shred', tmp_path]) except OSError: # shred is not available on this system, or some other error occured. r = 1 if r != 0: # we could not successfully execute unix shred; therefore, do custom shred. self._shred_file_custom(tmp_path) os.remove(tmp_path) def _edit_file_helper(self, filename, existing_data=None, force_save=False): # Create a tempfile _, tmp_path = tempfile.mkstemp() if existing_data: self.write_data(existing_data, tmp_path, shred=False) # drop the user into an editor on the tmp file try: call(self._editor_shell_command(tmp_path)) except: # whatever happens, destroy the decrypted file self._shred_file(tmp_path) raise tmpdata = self.read_data(tmp_path) # Do nothing if the content has not changed if existing_data == tmpdata and not force_save: self._shred_file(tmp_path) return # encrypt new data and write out to tmp enc_data = self.vault.encrypt(tmpdata) self.write_data(enc_data, tmp_path) # shuffle tmp file into place self.shuffle_files(tmp_path, filename) def encrypt_file(self, filename, output_file=None): check_prereqs() plaintext = self.read_data(filename) ciphertext = self.vault.encrypt(plaintext) self.write_data(ciphertext, output_file or filename) def decrypt_file(self, filename, output_file=None): check_prereqs() ciphertext = self.read_data(filename) plaintext = self.vault.decrypt(ciphertext) self.write_data(plaintext, output_file or filename, shred=False) def create_file(self, filename): """ create a new encrypted file """ check_prereqs() # FIXME: If we can raise an error here, we can probably just make it # behave like edit instead. if os.path.isfile(filename): raise AnsibleError("%s exists, please use 'edit' instead" % filename) self._edit_file_helper(filename) def edit_file(self, filename): check_prereqs() ciphertext = self.read_data(filename) plaintext = self.vault.decrypt(ciphertext) if self.vault.cipher_name not in CIPHER_WRITE_WHITELIST: # we want to get rid of files encrypted with the AES cipher self._edit_file_helper(filename, existing_data=plaintext, force_save=True) else: self._edit_file_helper(filename, existing_data=plaintext, force_save=False) def plaintext(self, filename): check_prereqs() ciphertext = self.read_data(filename) plaintext = self.vault.decrypt(ciphertext) return plaintext def rekey_file(self, filename, new_password): check_prereqs() prev = os.stat(filename) ciphertext = self.read_data(filename) plaintext = self.vault.decrypt(ciphertext) new_vault = VaultLib(new_password) new_ciphertext = new_vault.encrypt(plaintext) self.write_data(new_ciphertext, filename) # preserve permissions os.chmod(filename, prev.st_mode) os.chown(filename, prev.st_uid, prev.st_gid) def read_data(self, filename): try: if filename == '-': data = sys.stdin.read() else: with open(filename, "rb") as fh: data = fh.read() except Exception as e: raise AnsibleError(str(e)) return data def write_data(self, data, filename, shred=True): """write data to given path if shred==True, make sure that the original data is first shredded so that is cannot be recovered """ bytes = to_bytes(data, errors='strict') if filename == '-': sys.stdout.write(bytes) else: if os.path.isfile(filename): if shred: self._shred_file(filename) else: os.remove(filename) with open(filename, "wb") as fh: fh.write(bytes) def shuffle_files(self, src, dest): prev = None # overwrite dest with src if os.path.isfile(dest): prev = os.stat(dest) # old file 'dest' was encrypted, no need to _shred_file os.remove(dest) shutil.move(src, dest) # reset permissions if needed if prev is not None: #TODO: selinux, ACLs, xattr? os.chmod(dest, prev.st_mode) os.chown(dest, prev.st_uid, prev.st_gid) def _editor_shell_command(self, filename): EDITOR = os.environ.get('EDITOR','vim') editor = shlex.split(EDITOR) editor.append(filename) return editor class VaultFile(object): def __init__(self, password, filename): self.password = password self.filename = filename if not os.path.isfile(self.filename): raise AnsibleError("%s does not exist" % self.filename) try: self.filehandle = open(filename, "rb") except Exception as e: raise AnsibleError("Could not open %s: %s" % (self.filename, str(e))) _, self.tmpfile = tempfile.mkstemp() ### TODO: # __del__ can be problematic in python... For this use case, make # VaultFile a context manager instead (implement __enter__ and __exit__) def __del__(self): self.filehandle.close() os.unlink(self.tmplfile) def is_encrypted(self): peak = self.filehandle.readline() if peak.startswith(b_HEADER): return True else: return False def get_decrypted(self): check_prereqs() if self.is_encrypted(): tmpdata = self.filehandle.read() this_vault = VaultLib(self.password) dec_data = this_vault.decrypt(tmpdata) if dec_data is None: raise AnsibleError("Decryption failed") else: self.tmpfile.write(dec_data) return self.tmpfile else: return self.filename ######################################## # CIPHERS # ######################################## class VaultAES: # this version has been obsoleted by the VaultAES256 class # which uses encrypt-then-mac (fixing order) and also improving the KDF used # code remains for upgrade purposes only # http://stackoverflow.com/a/16761459 # Note: strings in this class should be byte strings by default. def __init__(self): if not HAS_AES: raise AnsibleError(CRYPTO_UPGRADE) def aes_derive_key_and_iv(self, password, salt, key_length, iv_length): """ Create a key and an initialization vector """ d = d_i = b'' while len(d) < key_length + iv_length: text = b''.join([d_i, password, salt]) d_i = to_bytes(md5(text).digest(), errors='strict') d += d_i key = d[:key_length] iv = d[key_length:key_length+iv_length] return key, iv def encrypt(self, data, password, key_length=32): """ Read plaintext data from in_file and write encrypted to out_file """ raise AnsibleError("Encryption disabled for deprecated VaultAES class") def decrypt(self, data, password, key_length=32): """ Read encrypted data from in_file and write decrypted to out_file """ # http://stackoverflow.com/a/14989032 data = unhexlify(data) in_file = BytesIO(data) in_file.seek(0) out_file = BytesIO() bs = AES.block_size tmpsalt = in_file.read(bs) salt = tmpsalt[len(b'Salted__'):] key, iv = self.aes_derive_key_and_iv(password, salt, key_length, bs) cipher = AES.new(key, AES.MODE_CBC, iv) next_chunk = b'' finished = False while not finished: chunk, next_chunk = next_chunk, cipher.decrypt(in_file.read(1024 * bs)) if len(next_chunk) == 0: if PY3: padding_length = chunk[-1] else: padding_length = ord(chunk[-1]) chunk = chunk[:-padding_length] finished = True out_file.write(chunk) out_file.flush() # reset the stream pointer to the beginning out_file.seek(0) out_data = out_file.read() out_file.close() # split out sha and verify decryption split_data = out_data.split(b"\n", 1) this_sha = split_data[0] this_data = split_data[1] test_sha = to_bytes(sha256(this_data).hexdigest()) if this_sha != test_sha: raise AnsibleError("Decryption failed") return this_data class VaultAES256: """ Vault implementation using AES-CTR with an HMAC-SHA256 authentication code. Keys are derived using PBKDF2 """ # http://www.daemonology.net/blog/2009-06-11-cryptographic-right-answers.html # Note: strings in this class should be byte strings by default. def __init__(self): check_prereqs() def create_key(self, password, salt, keylength, ivlength): hash_function = SHA256 # make two keys and one iv pbkdf2_prf = lambda p, s: HMAC.new(p, s, hash_function).digest() derivedkey = PBKDF2(password, salt, dkLen=(2 * keylength) + ivlength, count=10000, prf=pbkdf2_prf) return derivedkey def gen_key_initctr(self, password, salt): # 16 for AES 128, 32 for AES256 keylength = 32 # match the size used for counter.new to avoid extra work ivlength = 16 if HAS_PBKDF2HMAC: backend = default_backend() kdf = PBKDF2HMAC( algorithm=c_SHA256(), length=2 * keylength + ivlength, salt=salt, iterations=10000, backend=backend) derivedkey = kdf.derive(password) else: derivedkey = self.create_key(password, salt, keylength, ivlength) key1 = derivedkey[:keylength] key2 = derivedkey[keylength:(keylength * 2)] iv = derivedkey[(keylength * 2):(keylength * 2) + ivlength] return key1, key2, hexlify(iv) def encrypt(self, data, password): salt = os.urandom(32) key1, key2, iv = self.gen_key_initctr(password, salt) # PKCS#7 PAD DATA http://tools.ietf.org/html/rfc5652#section-6.3 bs = AES.block_size padding_length = (bs - len(data) % bs) or bs data += to_bytes(padding_length * chr(padding_length), encoding='ascii', errors='strict') # COUNTER.new PARAMETERS # 1) nbits (integer) - Length of the counter, in bits. # 2) initial_value (integer) - initial value of the counter. "iv" from gen_key_initctr ctr = Counter.new(128, initial_value=int(iv, 16)) # AES.new PARAMETERS # 1) AES key, must be either 16, 24, or 32 bytes long -- "key" from gen_key_initctr # 2) MODE_CTR, is the recommended mode # 3) counter= cipher = AES.new(key1, AES.MODE_CTR, counter=ctr) # ENCRYPT PADDED DATA cryptedData = cipher.encrypt(data) # COMBINE SALT, DIGEST AND DATA hmac = HMAC.new(key2, cryptedData, SHA256) message = b'\n'.join([hexlify(salt), to_bytes(hmac.hexdigest()), hexlify(cryptedData)]) message = hexlify(message) return message def decrypt(self, data, password): # SPLIT SALT, DIGEST, AND DATA data = unhexlify(data) salt, cryptedHmac, cryptedData = data.split(b"\n", 2) salt = unhexlify(salt) cryptedData = unhexlify(cryptedData) key1, key2, iv = self.gen_key_initctr(password, salt) # EXIT EARLY IF DIGEST DOESN'T MATCH hmacDecrypt = HMAC.new(key2, cryptedData, SHA256) if not self.is_equal(cryptedHmac, to_bytes(hmacDecrypt.hexdigest())): return None # SET THE COUNTER AND THE CIPHER ctr = Counter.new(128, initial_value=int(iv, 16)) cipher = AES.new(key1, AES.MODE_CTR, counter=ctr) # DECRYPT PADDED DATA decryptedData = cipher.decrypt(cryptedData) # UNPAD DATA try: padding_length = ord(decryptedData[-1]) except TypeError: padding_length = decryptedData[-1] decryptedData = decryptedData[:-padding_length] return decryptedData def is_equal(self, a, b): """ Comparing 2 byte arrrays in constant time to avoid timing attacks. It would be nice if there was a library for this but hey. """ # http://codahale.com/a-lesson-in-timing-attacks/ if len(a) != len(b): return False result = 0 for x, y in zip(a, b): if PY3: result |= x ^ y else: result |= ord(x) ^ ord(y) return result == 0 # Keys could be made bytes later if the code that gets the data is more # naturally byte-oriented CIPHER_MAPPING = { u'AES': VaultAES, u'AES256': VaultAES256, }