Coverage for lib/ansible/plugins/connection/__init__.py : 60%

Hot-keys on this page
r m x p toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
# (c) 2012-2014, Michael DeHaan <michael.dehaan@gmail.com> # (c) 2015 Toshio Kuratomi <tkuratomi@ansible.com> # (c) 2017, Peter Sprygada <psprygad@redhat.com> # (c) 2017 Ansible Project
except ImportError: from ansible.utils.display import Display display = Display()
def wrapped(self, *args, **kwargs):
''' A base class for connections to contain common code. '''
# When running over this connection type, prefer modules written in a certain language # as discovered by the specified file extension. An empty string as the # language means any language.
# the following control whether or not the connection supports the # persistent connection framework or not
# All these hasattrs allow subclasses to override these parameters # Backwards compat: self._display isn't really needed, just import the global display and use that.
self._shell = shell
# load the shell plugin for this action/connection shell_type = play_context.shell shell_type = getattr(self, '_shell_type') else:
raise AnsibleError("Invalid shell type specified (%s), or the plugin for that shell type is missing." % shell_type)
def connected(self): '''Read-only property holding whether the connection to the remote host is active or closed.'''
def socket_path(self): '''Read-only property holding the connection socket path for this remote host'''
''' Checks if the current class supports this privilege escalation method '''
if self._play_context.become_method in self.become_methods: return True
raise AnsibleError("Internal Error: this connection module does not support running commands via %s" % self._play_context.become_method)
def _split_ssh_args(argstring): """ Takes a string like '-o Foo=1 -o Bar="foo bar"' and returns a list ['-o', 'Foo=1', '-o', 'Bar=foo bar'] that can be added to the argument list. The list will not contain any empty elements. """ # Python 2.6.x shlex doesn't handle unicode type so we have to # convert args to byte string for that case. More efficient to # try without conversion first but python2.6 doesn't throw an # exception, it merely mangles the output: # >>> shlex.split(u't e') # ['t\x00\x00\x00', '\x00\x00\x00e\x00\x00\x00'] except AttributeError: # In Python3, shlex.split doesn't work on a byte string. return [to_text(x.strip()) for x in shlex.split(argstring) if x.strip()]
def transport(self): """String used to identify this Connection class from other classes""" pass
def _connect(self): """Connect to the host we've been initialized with"""
# Check if PE is supported if self._play_context.become: self._become_method_supported()
"""Run a command on the remote host.
:arg cmd: byte string containing the command :kwarg in_data: If set, this data is passed to the command's stdin. This is used to implement pipelining. Currently not all connection plugins implement pipelining. :kwarg sudoable: Tell the connection plugin if we're executing a command via a privilege escalation mechanism. This may affect how the connection plugin returns data. Note that not all connections can handle privilege escalation. :returns: a tuple of (return code, stdout, stderr) The return code is an int while stdout and stderr are both byte strings.
When a command is executed, it goes through multiple commands to get there. It looks approximately like this::
[LocalShell] ConnectionCommand [UsersLoginShell (*)] ANSIBLE_SHELL_EXECUTABLE [(BecomeCommand ANSIBLE_SHELL_EXECUTABLE)] Command :LocalShell: Is optional. It is run locally to invoke the ``Connection Command``. In most instances, the ``ConnectionCommand`` can be invoked directly instead. The ssh connection plugin which can have values that need expanding locally specified via ssh_args is the sole known exception to this. Shell metacharacters in the command itself should be processed on the remote machine, not on the local machine so no shell is needed on the local machine. (Example, ``/bin/sh``) :ConnectionCommand: This is the command that connects us to the remote machine to run the rest of the command. ``ansible_ssh_user``, ``ansible_ssh_host`` and so forth are fed to this piece of the command to connect to the correct host (Examples ``ssh``, ``chroot``) :UsersLoginShell: This shell may or may not be created depending on the ConnectionCommand used by the connection plugin. This is the shell that the ``ansible_ssh_user`` has configured as their login shell. In traditional UNIX parlance, this is the last field of a user's ``/etc/passwd`` entry We do not specifically try to run the ``UsersLoginShell`` when we connect. Instead it is implicit in the actions that the ``ConnectionCommand`` takes when it connects to a remote machine. ``ansible_shell_type`` may be set to inform ansible of differences in how the ``UsersLoginShell`` handles things like quoting if a shell has different semantics than the Bourne shell. :ANSIBLE_SHELL_EXECUTABLE: This is the shell set via the inventory var ``ansible_shell_executable`` or via ``constants.DEFAULT_EXECUTABLE`` if the inventory var is not set. We explicitly invoke this shell so that we have predictable quoting rules at this point. ``ANSIBLE_SHELL_EXECUTABLE`` is only settable by the user because some sudo setups may only allow invoking a specific shell. (For instance, ``/bin/bash`` may be allowed but ``/bin/sh``, our default, may not). We invoke this twice, once after the ``ConnectionCommand`` and once after the ``BecomeCommand``. After the ConnectionCommand, this is run by the ``UsersLoginShell``. After the ``BecomeCommand`` we specify that the ``ANSIBLE_SHELL_EXECUTABLE`` is being invoked directly. :BecomeComand ANSIBLE_SHELL_EXECUTABLE: Is the command that performs privilege escalation. Setting this up is performed by the action plugin prior to running ``exec_command``. So we just get passed :param:`cmd` which has the BecomeCommand already added. (Examples: sudo, su) If we have a BecomeCommand then we will invoke a ANSIBLE_SHELL_EXECUTABLE shell inside of it so that we have a consistent view of quoting. :Command: Is the command we're actually trying to run remotely. (Examples: mkdir -p $HOME/.ansible, python $HOME/.ansible/tmp-script-file) """
def put_file(self, in_path, out_path): """Transfer a file from local to remote"""
def fetch_file(self, in_path, out_path): """Fetch a file from remote to local""" pass
def close(self): """Terminate the connection""" pass
b_success_key = to_bytes(self._play_context.success_key) for b_line in b_output.splitlines(True): if b_success_key == b_line.rstrip(): return True return False
if self._play_context.prompt is None: return False elif isinstance(self._play_context.prompt, string_types): b_prompt = to_bytes(self._play_context.prompt).strip() b_lines = b_output.splitlines() return any(l.strip().startswith(b_prompt) for l in b_lines) else: return self._play_context.prompt(b_output)
b_incorrect_password = to_bytes(gettext.dgettext(self._play_context.become_method, C.BECOME_ERROR_STRINGS[self._play_context.become_method])) return b_incorrect_password and b_incorrect_password in b_output
b_missing_password = to_bytes(gettext.dgettext(self._play_context.become_method, C.BECOME_MISSING_STRINGS[self._play_context.become_method])) return b_missing_password and b_missing_password in b_output
f = self._play_context.connection_lockfd display.vvvv('CONNECTION: pid %d waiting for lock on %d' % (os.getpid(), f), host=self._play_context.remote_addr) fcntl.lockf(f, fcntl.LOCK_EX) display.vvvv('CONNECTION: pid %d acquired lock on %d' % (os.getpid(), f), host=self._play_context.remote_addr)
f = self._play_context.connection_lockfd fcntl.lockf(f, fcntl.LOCK_UN) display.vvvv('CONNECTION: pid %d released lock on %d' % (os.getpid(), f), host=self._play_context.remote_addr)
display.warning("Reset is not implemented for this connection") |