Extract vault related bits of DataLoader._get_file_contents to DataLoader._decrypt_if_vault_data
When loading vault password files, detect if they are vault encrypted, and if so, try to decrypt with any already known vault secrets.
This implements the 'Allow vault password files to be vault encrypted' (#31002) feature card from
the 2.5.0 project at https://github.com/ansible/ansible/projects/9Fixes#31002
Shell is implemented via the command module. There was a special case
in mod_args to do that. Make shell into an action plugin to handle that
instead.
Also move the special case for the command nanny into a command module
action plugin. This is more appropriate as we then do not have to send
a parameter that is only for the command module to every single module.
* Better handling of malformed vault data envelope
If an embedded vaulted variable ('!vault' in yaml)
had an invalid format, it would eventually cause
an error for seemingly unrelated reasons.
"Invalid" meaning not valid hexlify (extra chars,
non-hex chars, etc).
For ex, if a host_vars file had invalid vault format
variables, on py2, it would cause an error like:
'ansible.vars.hostvars.HostVars object' has no
attribute u'broken.example.com'
Depending on where the invalid vault is, it could
also cause "VARIABLE IS NOT DEFINED!". The behavior
can also change if ansible-playbook is py2 or py3.
Root cause is errors from binascii.unhexlify() not
being handled consistently.
Fix is to add a AnsibleVaultFormatError exception and
raise it on any unhexlify() errors and to handle it
properly elsewhere.
Add a _unhexlify() that try/excepts around a binascii.unhexlify()
and raises an AnsibleVaultFormatError on invalid vault data.
This is so the same exception type is always raised for this
case. Previous it was different between py2 and py3.
binascii.unhexlify() raises a binascii.Error if the hexlified
blobs in a vault data blob are invalid.
On py2, binascii.Error is a subclass of Exception.
On py3, binascii.Error is a subclass of TypeError
When decrypting content of vault encrypted variables,
if a binascii.Error is raised it propagates up to
playbook.base.Base.post_validate(). post_validate()
handles exceptions for TypeErrors but not for
base Exception subclasses (like py2 binascii.Error).
* Add a display.warning on vault format errors
* Unit tests for _unhexlify, parse_vaulttext*
* Add intg test cases for invalid vault formats
Fixes#28038
This adds a new type of vault-password script (a 'client') that takes advantage of and enhances the
multiple vault password support.
If a vault password script basename ends with the name '-client', consider it a vault password script client.
A vault password script 'client' just means that the script will take a '--vault-id' command line arg.
The previous vault password script (as invoked by --vault-password-file pointing to an executable) takes
no args and returns the password on stdout. But it doesnt know anything about --vault-id or multiple vault
passwords.
The new 'protocol' of the vault password script takes a cli arg ('--vault-id') so that it can lookup that specific
vault-id and return it's password.
Since existing vault password scripts don't know the new 'protocol', a way to distinguish password scripts
that do understand the protocol was needed. The convention now is to consider password scripts that are
named like 'something-client.py' (and executable) to be vault password client scripts.
The new client scripts get invoked with the '--vault-id' they were requested for. An example:
ansible-playbook --vault-id my_vault_id@contrib/vault/vault-keyring-client.py some_playbook.yml
That will cause the 'contrib/vault/vault-keyring-client.py' script to be invoked as:
contrib/vault/vault-keyring-client.py --vault-id my_vault_id
The previous vault-keyring.py password script was extended to become vault-keyring-client.py. It uses
the python 'keyring' module to request secrets from various backends. The plain 'vault-keyring.py' script
would determine which key id and keyring name to use based on values that had to be set in ansible.cfg.
So it was also limited to one keyring name.
The new vault-keyring-client.py will request the secret for the vault id provided via the '--vault-id' option.
The script can be used without config and can be used for multiple keyring ids (and keyrings).
On success, a vault password client script will print the password to stdout and exit with a return code of 0.
If the 'client' script can't find a secret for the --vault-id, the script will exit with return code of 2 and print an error to stderr.
* Use vault_id when encrypted via vault-edit
On the encryption stage of
'ansible-vault edit --vault-id=someid@passfile somefile',
the vault id was not being passed to encrypt() so the files were
always saved with the default vault id in the 1.1 version format.
When trying to edit that file a second time, also with a --vault-id,
the file would be decrypted with the secret associated with the
provided vault-id, but since the encrypted file had no vault id
in the envelope there would be no match for 'default' secrets.
(Only the --vault-id was included in the potential matches, so
the vault id actually used to decrypt was not).
If that list was empty, there would be an IndexError when trying
to encrypted the changed file. This would result in the displayed
error:
ERROR! Unexpected Exception, this is probably a bug: list index out of range
Fix is two parts:
1) use the vault id when encrypting from edit
2) when matching the secret to use for encrypting after edit,
include the vault id that was used for decryption and not just
the vault id (or lack of vault id) from the envelope.
add unit tests for #30575 and intg tests for 'ansible-vault edit'
Fixes#30575
* This commit includes a unit test to exercise the _is_role
function and make sure it doesn't break in any Python version.
* Import os.path and other minor fixups
* Add network value to support_by field.
* New support_by value, certified
* Deprecate curated in favor of certified
* Add conversion from 1.0 to 1.1 to metadata-tool
* Add supported by Red Hat field to ansible-doc output
* Better handling of empty/invalid passwords
empty password files are global error and cause an
exit. A warning is also emitted with more detail.
ie, if any of the password/secret sources provide
a bogus password (ie, empty) or fail (exception,
ctrl-d, EOFError), we stop at the first error and exit.
This makes behavior when entering empty password at
prompt match 2.3 (ie, an error)
* rm unneeded parens following assert
* rm unused parse_vaulttext_envelope from yaml.constructor
* No longer need index/enumerate over vault_ids
* rm unnecessary else
* rm unused VaultCli.secrets
* rm unused vault_id arg on VaultAES.decrypt()
pylint: Unused argument 'vault_id'
pylint: Unused parse_vaulttext_envelope imported from ansible.parsing.vault
pylint: Unused variable 'index'
pylint: Unnecessary parens after 'assert' keyword
pylint: Unnecessary "else" after "return" (no-else-return)
pylint: Attribute 'editor' defined outside __init__
* use 'dummy' for unused variables instead of _
Based on pylint unused variable warnings.
Existing code use '_' for this, but that is old
and busted. The hot new thing is 'dummy'. It
is so fetch.
Except for where we get warnings for reusing
the 'dummy' var name inside of a list comprehension.
* Add super().__init__ call to PromptVaultSecret.__init__
pylint: __init__ method from base class 'VaultSecret' is not called (super-init-not-called)
* Make FileVaultSecret.read_file reg method again
The base class read_file() doesnt need self but
the sub classes do.
Rm now unneeded loader arg to read_file()
* Fix err msg string literal that had no effect
pylint: String statement has no effect
The indent on the continuation of the msg_format was wrong
so the second half was dropped.
There was also no need to join() filename (copy/paste from
original with a command list I assume...)
* Use local cipher_name in VaultEditor.edit_file not instance
pylint: Unused variable 'cipher_name'
pylint: Unused variable 'b_ciphertext'
Use the local cipher_name returned from parse_vaulttext_envelope()
instead of the instance self.cipher_name var.
Since there is only one valid cipher_name either way, it was
equilivent, but it will not be with more valid cipher_names
* Rm unused b_salt arg on VaultAES256._encrypt*
pylint: Unused argument 'b_salt'
Previously the methods computed the keys and iv themselves
so needed to be passed in the salt, but now the key/iv
are built before and passed in so b_salt arg is not used
anymore.
* rm redundant import of call from subprocess
pylint: Imports from package subprocess are not grouped
use via subprocess module now instead of direct
import.
* self._bytes is set in super init now, rm dup
* Make FileVaultSecret.read_file() -> _read_file()
_read_file() is details of the implementation of
load(), so now 'private'.
When parsing a vaulttext blob, use .splitlines()
instead of split(b'\n') to handle \n newlines and
windows style \r\n (CRLF) new lines.
The vaulttext enevelope at this point is just the header line
and a hexlify()'ed blob, so CRLF is a valid newline here.
Fixes#22914
Fixes#13243
** Add --vault-id to name/identify multiple vault passwords
Use --vault-id to indicate id and path/type
--vault-id=prompt # prompt for default vault id password
--vault-id=myorg@prompt # prompt for a vault_id named 'myorg'
--vault-id=a_password_file # load ./a_password_file for default id
--vault-id=myorg@a_password_file # load file for 'myorg' vault id
vault_id's are created implicitly for existing --vault-password-file
and --ask-vault-pass options.
Vault ids are just for UX purposes and bookkeeping. Only the vault
payload and the password bytestring is needed to decrypt a
vault blob.
Replace passing password around everywhere with
a VaultSecrets object.
If we specify a vault_id, mention that in password prompts
Specifying multiple -vault-password-files will
now try each until one works
** Rev vault format in a backwards compatible way
The 1.2 vault format adds the vault_id to the header line
of the vault text. This is backwards compatible with older
versions of ansible. Old versions will just ignore it and
treat it as the default (and only) vault id.
Note: only 2.4+ supports multiple vault passwords, so while
earlier ansible versions can read the vault-1.2 format, it
does not make them magically support multiple vault passwords.
use 1.1 format for 'default' vault_id
Vaulted items that need to include a vault_id will be
written in 1.2 format.
If we set a new DEFAULT_VAULT_IDENTITY, then the default will
use version 1.2
vault will only use a vault_id if one is specified. So if none
is specified and C.DEFAULT_VAULT_IDENTITY is 'default'
we use the old format.
** Changes/refactors needed to implement multiple vault passwords
raise exceptions on decrypt fail, check vault id early
split out parsing the vault plaintext envelope (with the
sha/original plaintext) to _split_plaintext_envelope()
some cli fixups for specifying multiple paths in
the unfrack_paths optparse callback
fix py3 dict.keys() 'dict_keys object is not indexable' error
pluralize cli.options.vault_password_file -> vault_password_files
pluralize cli.options.new_vault_password_file -> new_vault_password_files
pluralize cli.options.vault_id -> cli.options.vault_ids
** Add a config option (vault_id_match) to force vault id matching.
With 'vault_id_match=True' and an ansible
vault that provides a vault_id, then decryption will require
that a matching vault_id is required. (via
--vault-id=my_vault_id@password_file, for ex).
In other words, if the config option is true, then only
the vault secrets with matching vault ids are candidates for
decrypting a vault. If option is false (the default), then
all of the provided vault secrets will be selected.
If a user doesn't want all vault secrets to be tried to
decrypt any vault content, they can enable this option.
Note: The vault id used for the match is not encrypted or
cryptographically signed. It is just a label/id/nickname used
for referencing a specific vault secret.
* add unit test: nested dynamic includes
* nested dynamic includes: avoid AnsibleFileNotFound error
Error was:
Unable to retrieve file contents
Could not find or access 'include2.yml'
Before 8f758204cf, at the end of
'path_dwim_relative' method, the 'search' variable contained amongst
others paths:
'/tmp/roles/testrole/tasks/tasks/included.yml' and
'/tmp/roles/testrole/tasks/included.yml'.
The commit mentioned before removed the last one despite the method
docstrings specify 'with or without explicitly named dirname subdirs'.
* add integration test: nested includes
* Fix ansible-doc traceback when a plugin doesn't parse correctly
* Change extract_metadata ivocation to take either an ast or source
code. When given source code, it can find file offsets for the start
and end of dict. When given the ast, it is quicker as it doesn't have
to reparse the source. Requires changing the call to the function to
use a keyword arg.
* Fix reading of metadata to find the last occurrence of
ANSIBLE_METADATA instead of the first.
* Add some more unittests to get closer to complete coverage
* Unittests for extracting metadata from plugins
* Port plugin_docs to use the generic extract_metadata function
* Make the helper functions seek_end_of{string,dict} private
Make pyca/cryptography the preferred backend for cryptographic needs (mainly vault) falling back to pycrypto
pyca/cryptography is already implicitly a dependency in many cases
through paramiko (2.0+) as well as the new openssl_publickey module,
which requires pyOpenSSL 16.0+. Additionally, pyca/cryptography is
an optional dep for better performance with vault already.
This commit leverages cryptography's padding, constant time comparisons,
and CBC/CTR modes to reduce the amount of code ansible needs to
maintain.
* Handle wrong password given for VaultAES format
* Do not display deprecation warning for cryptography on python-2.6
* Namespace all of the pycrypto imports and always import them
Makes unittests better and the code less likely to get stupid mistakes
(like using HMAC from cryptogrpahy when the one from pycrypto is needed)
* Add back in atfork since we need pycrypto to reinitialize its RNG just in case we're being used with old paramiko
* contrib/inventory/gce: Remove spurious require on pycrypto
(cherry picked from commit 9e16b9db275263b3ea8d1b124966fdebfc9ab271)
* Add cryptography to ec2_win_password module requirements
* Fix python3 bug which would pass text strings to a function which
requires byte strings.
* Attempt to add pycrypto version to setup deps
* Change hacking README for dual pycrypto/cryptography
* update dependencies for various CI scripts
* additional CI dockerfile/script updates
* add paramiko to the windows and sanity requirement set
This is needed because ansible lists it as a requirement. Previously
the missing dep wasn't enforced, but cryptography imports pkg_resources
so you can't ignore a requirement any more
* Add integration test cases for old vault and for wrong passwords
* helper script for manual testing of pycrypto/cryptography
* Skip the pycrypto tests so that users without it installed can still run the unittests
* Run unittests for vault with both cryptography and pycrypto backend
* test/: PEP8 compliancy
- Make PEP8 compliant
* Python3 chokes on casting int to bytes (#24952)
But if we tell the formatter that the var is a number, it works
* Fix vault reading from stdin (avoid realpath() on non-links)
os.path.realpath() is used to find the target of file paths that
are symlinks so vault operations happen directly on the target.
However, in addition to resolving symlinks, realpath() also returns
a full path. when reading from stdin, vault cli uses '-' as a special
file path so VaultEditor() will replace with stdin.
realpath() was expanding '-' with the CWD to something like
'/home/user/playbooks/-' causing errors like:
ERROR! [Errno 2] No such file or directory: u'/home/user/ansible/-'
Fix is to specialcase '-' to not use realpath()
Fixes#23567
* to_text decrypt output when writing to stdout
* Update module_utils.six to latest
We've been held back on the version of six we could use on the module
side to 1.4.x because of python-2.4 compatibility. Now that our minimum
is Python-2.6, we can update to the latest version of six in
module_utils and get rid of the second copy in lib/ansible/compat.
Since vault edit attempts to unlink
edited files before creating a new file
with the same name and writing to it, if
the file was a symlink, the symlink would
be replaced with a regular file.
VaultEditor file ops now check if files
it is changing are symlinks and instead
works directly on the target, so that
os.rename() and shutils do the right thing.
Add unit tests cases for this case and
assorted VaultEditor test cases.
Fixes#20264
* added docs for vault and made trigger shorter: !vault
* added single var valuting
* Update playbooks_vault.rst
Edit pass for spelling and grammar. Ship it!
* Update playbooks_vault.rst
Typo fixes.
- Replace nose usage with pytest.
- Remove legacy Shippable integration.sh.
- Update Makefile to use pytest and ansible-test.
- Convert most yield unit tests to pytest parametrize.
* Add a encode() to AnsibleVaultEncryptedUnicode
Without it, calling encode() on it results in a bytestring
of the encrypted !vault-encrypted string.
ssh connection plugin triggers this if ansible_password
is from a var using !vault-encrypted. That path ends up
calling .encode() instead of using the __str__.
Fixes#19795
* Fix str.encode() errors on py2.6
py2.6 str.encode() does not take keyword arguments.
* Fix bug (#18355) where encrypted inventories fail
This is first part of fix for #18355
* Make DataLoader._get_file_contents return bytes
The issue #18355 is caused by a change to inventory to
stop using _get_file_contents so that it can handle text
encoding itself to better protect against harmless text
encoding errors in ini files (invalid unicode text in
comment fields).
So this makes _get_file_contents return bytes so it and other
callers can handle the to_text().
The data returned by _get_file_contents() is now a bytes object
instead of a text object. The callers of _get_file_contents() have
been updated to call to_text() themselves on the results.
Previously, the ini parser attempted to work around
ini files that potentially include non-vailid unicode
in comment lines. To do this, it stopped using
DataLoader._get_file_contents() which does the decryption of
files if vault encrypted. It didn't use that because _get_file_contents
previously did to_text() on the read data itself.
_get_file_contents() returns a bytestring now, so ini.py
can call it and still special case ini file comments when
converting to_text(). That also means encrypted inventory files
are decrypted first.
Fixes#18355
- Remove shebangs from:
- ini files
- unit tests
- module_utils
- plugins
- module_docs_fragments
- non-executable Makefiles
- Change non-modules from '/usr/bin/python' to '/usr/bin/env python'.
- Change '/bin/env' to '/usr/bin/env'.
Also removed main functions from unit tests (since they no longer
have a shebang) and fixed a python 3 compatibility issue with
update_bundled.py so it does not need to specify a python 2 shebang.
A script was added to check for unexpected shebangs in files.
This script is run during CI on Shippable.
* Make is_encrypted_file handle both files opened in text and binary mode
On python3, by default files are opened in text mode. Since we know
the encoding of vault files (and especially the header which is the
first set of bytes) we can decide whether the file is an encrypted
vault file in either case.
* Fix is_encrypted_file not resetting the file position
* Update is_encrypted_file to check that all the data in the file is ascii
* For is_encrypted_file(), add start_pos and count parameters
This allows callers to specify reading vaulttext from the middle of
a file if necessary.
* Combine VaultLib.encrypt() and VaultLib.encrypt_bytestring()
* Change vault's is_encrypted() to take either text or byte strings and to return False if any part of the data is non-ascii.
* Remove unnecessary use of six.b
* Vault Cipher: mark a few methods as private.
* VaultAES256._is_equal throws a TypeError if given non byte strings
* Make VaultAES256 methods that don't need self staticmethods and classmethods
* Mark VaultAES and is_encrypted as deprecated
* Get rid of VaultFile (unused and feature implemented in a different way)
* Normalize variable and parameter names on plaintext, ciphertext, vaulttext
* Normalize variable and parameter names on "b_" prefix when dealing with bytes
* Test changes:
* Remove redundant tests( both checking the same byte string)
* Fix use of format string without format operator
* Enable vault editor tests on python3
* Initialize the vault_cipher for VaultAES256 testing in setUp()
* Make assertTrue and assertFalse take the actual method calls for
better error messages.
* Test that non-ascii byte strings compare correctly.
* Test that unicode strings and ints raise TypeError
* Test-specific:
* Removed test_methods_exist(). We only have one VaultLib so the
implementation is the assurance that the methods exist. (Can use an abc for
this if it changes).
* Add tests for both byte string and text string input where the API takes either.
* Convert "assert" to unittest assert functions or add a custom message where
that will make failures easier to debug.
* Move instantiating the VaultLib into setUp().
We couldn't copy to_unicode, to_bytes, to_str into module_utils because
of licensing. So once created it we had two sets of functions that did
the same things but had different implementations. To remedy that, this
change removes the ansible.utils.unicode versions of those functions.
Make some python3 fixes to make the unittests pass:
* galaxy imports
* dictionary iteration in role requirements
* swap_stdout helper for unittests
* Normalize to text string in a facts.py function
Make !vault-encrypted create a AnsibleVaultUnicode
yaml object that can be used as a regular string object.
This allows a playbook to include a encrypted vault
blob for the value of a yaml variable. A 'secret_password'
variable can have it's value encrypted instead of having
to vault encrypt an entire vars file.
Add __ENCRYPTED__ to the vault yaml types so
template.Template can treat it similar
to __UNSAFE__ flags.
vault.VaultLib api changes:
- Split VaultLib.encrypt to encrypt and encrypt_bytestring
- VaultLib.encrypt() previously accepted the plaintext data
as either a byte string or a unicode string.
Doing the right thing based on the input type would fail
on py3 if given a arg of type 'bytes'. To simplify the
API, vaultlib.encrypt() now assumes input plaintext is a
py2 unicode or py3 str. It will encode to utf-8 then call
the new encrypt_bytestring(). The new methods are less
ambiguous.
- moved VaultLib.is_encrypted logic to vault module scope
and split to is_encrypted() and is_encrypted_file().
Add a test/unit/mock/yaml_helper.py
It has some helpers for testing parsing/yaml
Integration tests added as roles test_vault and test_vault_embedded
* Changed parse_addresses to throw exceptions instead of passing None
* Switched callers to trap and pass through the original values.
* Added very verbose notice
* Look at deprecating this and possibly validate at plugin instead
fixes#13608
Labels must start with an alphanumeric character, may contain
alphanumeric characters or hyphens, but must not end with a hyphen.
We enforce those rules, but allow underscores wherever hyphens are
accepted, and allow alphanumeric ranges anywhere.
We relax the definition of "alphanumeric" to include Unicode characters
even though such inventory hostnames cannot be used in practice unless
an ansible_ssh_host is set for each of them.
We still don't enforce length restrictions—the fact that we have to
accept ranges makes it more complex, and it doesn't seem especially
worthwhile.
This adds a parse_address(pattern) utility function that returns
(host,port), and uses it wherever where we accept IPv4 and IPv6
addresses and hostnames (or host patterns): the inventory parser
the the add_host action plugin.
It also introduces a more extensive set of unit tests that supersedes
the old add_host unit tests (which didn't actually test add_host, but
only the parsing function).
Required some rewiring in inventory code to make sure we're using
the DataLoader class for some data file operations, which makes mocking
them much easier.
Also identified two corner cases not currently handled by the code, related
to inventory variable sources and which one "wins". Also noticed we weren't
properly merging variables from multiple group/host_var file locations
(inventory directory vs. playbook directory locations) so fixed as well.
Note that this test was broken in devel because it was really just
duplicating the AES256 test because setting v.cipher_name to 'AES'
no longer selected AES after it was de-write-whitelisted.
Now that we've removed the VaultAES encryption code, we embed static
output from an earlier version and test that we can decrypt it.
Adrian Likins