add EvalFormatter for batch system (PBS) launcher templates

parallel docs now reflect use of Formatter instead of Itpl.

IPython/parallel/apps/launcher.py

@@ -1,1067 +1,1070 b''
 #!/usr/bin/env python
 # encoding: utf-8
 """
 Facilities for launching IPython processes asynchronously.
 """
 
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2008-2009  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 
 import copy
 import logging
 import os
 import re
 import stat
 
 # signal imports, handling various platforms, versions
 
 from signal import SIGINT, SIGTERM
 try:
     from signal import SIGKILL
 except ImportError:
     # Windows
     SIGKILL=SIGTERM
 
 try:
     # Windows >= 2.7, 3.2
     from signal import CTRL_C_EVENT as SIGINT
 except ImportError:
     pass
 
 from subprocess import Popen, PIPE, STDOUT
 try:
     from subprocess import check_output
 except ImportError:
     # pre-2.7, define check_output with Popen
     def check_output(*args, **kwargs):
         kwargs.update(dict(stdout=PIPE))
         p = Popen(*args, **kwargs)
         out,err = p.communicate()
         return out
 
 from zmq.eventloop import ioloop
 
 # from IPython.config.configurable import Configurable
+from IPython.utils.text import EvalFormatter
 from IPython.utils.traitlets import Any, Int, List, Unicode, Dict, Instance
 from IPython.utils.path import get_ipython_module_path
 from IPython.utils.process import find_cmd, pycmd2argv, FindCmdError
 
 from IPython.parallel.factory import LoggingFactory
 
 from .win32support import forward_read_events
 
 from .winhpcjob import IPControllerTask, IPEngineTask, IPControllerJob, IPEngineSetJob
 
 WINDOWS = os.name == 'nt'
 
 #-----------------------------------------------------------------------------
 # Paths to the kernel apps
 #-----------------------------------------------------------------------------
 
 
 ipcluster_cmd_argv = pycmd2argv(get_ipython_module_path(
     'IPython.parallel.apps.ipclusterapp'
 ))
 
 ipengine_cmd_argv = pycmd2argv(get_ipython_module_path(
     'IPython.parallel.apps.ipengineapp'
 ))
 
 ipcontroller_cmd_argv = pycmd2argv(get_ipython_module_path(
     'IPython.parallel.apps.ipcontrollerapp'
 ))
 
 #-----------------------------------------------------------------------------
 # Base launchers and errors
 #-----------------------------------------------------------------------------
 
 
 class LauncherError(Exception):
     pass
 
 
 class ProcessStateError(LauncherError):
     pass
 
 
 class UnknownStatus(LauncherError):
     pass
 
 
 class BaseLauncher(LoggingFactory):
     """An asbtraction for starting, stopping and signaling a process."""
 
     # In all of the launchers, the work_dir is where child processes will be
     # run. This will usually be the profile_dir, but may not be. any work_dir
     # passed into the __init__ method will override the config value.
     # This should not be used to set the work_dir for the actual engine
     # and controller. Instead, use their own config files or the
     # controller_args, engine_args attributes of the launchers to add
     # the work_dir option.
     work_dir = Unicode(u'.')
     loop = Instance('zmq.eventloop.ioloop.IOLoop')
     
     start_data = Any()
     stop_data = Any()
     
     def _loop_default(self):
         return ioloop.IOLoop.instance()
 
     def __init__(self, work_dir=u'.', config=None, **kwargs):
         super(BaseLauncher, self).__init__(work_dir=work_dir, config=config, **kwargs)
         self.state = 'before' # can be before, running, after
         self.stop_callbacks = []
         self.start_data = None
         self.stop_data = None
 
     @property
     def args(self):
         """A list of cmd and args that will be used to start the process.
 
         This is what is passed to :func:`spawnProcess` and the first element
         will be the process name.
         """
         return self.find_args()
 
     def find_args(self):
         """The ``.args`` property calls this to find the args list.
 
         Subcommand should implement this to construct the cmd and args.
         """
         raise NotImplementedError('find_args must be implemented in a subclass')
 
     @property
     def arg_str(self):
         """The string form of the program arguments."""
         return ' '.join(self.args)
 
     @property
     def running(self):
         """Am I running."""
         if self.state == 'running':
             return True
         else:
             return False
 
     def start(self):
         """Start the process.
 
         This must return a deferred that fires with information about the
         process starting (like a pid, job id, etc.).
         """
         raise NotImplementedError('start must be implemented in a subclass')
 
     def stop(self):
         """Stop the process and notify observers of stopping.
 
         This must return a deferred that fires with information about the
         processing stopping, like errors that occur while the process is
         attempting to be shut down. This deferred won't fire when the process
         actually stops. To observe the actual process stopping, see
         :func:`observe_stop`.
         """
         raise NotImplementedError('stop must be implemented in a subclass')
 
     def on_stop(self, f):
         """Get a deferred that will fire when the process stops.
 
         The deferred will fire with data that contains information about
         the exit status of the process.
         """
         if self.state=='after':
             return f(self.stop_data)
         else:
             self.stop_callbacks.append(f)
 
     def notify_start(self, data):
         """Call this to trigger startup actions.
 
         This logs the process startup and sets the state to 'running'.  It is
         a pass-through so it can be used as a callback.
         """
 
         self.log.info('Process %r started: %r' % (self.args[0], data))
         self.start_data = data
         self.state = 'running'
         return data
 
     def notify_stop(self, data):
         """Call this to trigger process stop actions.
 
         This logs the process stopping and sets the state to 'after'. Call
         this to trigger all the deferreds from :func:`observe_stop`."""
 
         self.log.info('Process %r stopped: %r' % (self.args[0], data))
         self.stop_data = data
         self.state = 'after'
         for i in range(len(self.stop_callbacks)):
             d = self.stop_callbacks.pop()
             d(data)
         return data
 
     def signal(self, sig):
         """Signal the process.
 
         Return a semi-meaningless deferred after signaling the process.
 
         Parameters
         ----------
         sig : str or int
             'KILL', 'INT', etc., or any signal number
         """
         raise NotImplementedError('signal must be implemented in a subclass')
 
 
 #-----------------------------------------------------------------------------
 # Local process launchers
 #-----------------------------------------------------------------------------
 
 
 class LocalProcessLauncher(BaseLauncher):
     """Start and stop an external process in an asynchronous manner.
 
     This will launch the external process with a working directory of
     ``self.work_dir``.
     """
 
     # This is used to to construct self.args, which is passed to 
     # spawnProcess.
     cmd_and_args = List([])
     poll_frequency = Int(100) # in ms
 
     def __init__(self, work_dir=u'.', config=None, **kwargs):
         super(LocalProcessLauncher, self).__init__(
             work_dir=work_dir, config=config, **kwargs
         )
         self.process = None
         self.start_deferred = None
         self.poller = None
 
     def find_args(self):
         return self.cmd_and_args
 
     def start(self):
         if self.state == 'before':
             self.process = Popen(self.args,
                 stdout=PIPE,stderr=PIPE,stdin=PIPE,
                 env=os.environ,
                 cwd=self.work_dir
             )
             if WINDOWS:
                 self.stdout = forward_read_events(self.process.stdout)
                 self.stderr = forward_read_events(self.process.stderr)
             else:
                 self.stdout = self.process.stdout.fileno()
                 self.stderr = self.process.stderr.fileno()
             self.loop.add_handler(self.stdout, self.handle_stdout, self.loop.READ)
             self.loop.add_handler(self.stderr, self.handle_stderr, self.loop.READ)
             self.poller = ioloop.PeriodicCallback(self.poll, self.poll_frequency, self.loop)
             self.poller.start()
             self.notify_start(self.process.pid)
         else:
             s = 'The process was already started and has state: %r' % self.state
             raise ProcessStateError(s)
 
     def stop(self):
         return self.interrupt_then_kill()
 
     def signal(self, sig):
         if self.state == 'running':
             if WINDOWS and sig != SIGINT:
                 # use Windows tree-kill for better child cleanup
                 check_output(['taskkill', '-pid', str(self.process.pid), '-t', '-f'])
             else:
                 self.process.send_signal(sig)
 
     def interrupt_then_kill(self, delay=2.0):
         """Send INT, wait a delay and then send KILL."""
         try:
             self.signal(SIGINT)
         except Exception:
             self.log.debug("interrupt failed")
             pass
         self.killer  = ioloop.DelayedCallback(lambda : self.signal(SIGKILL), delay*1000, self.loop)
         self.killer.start()
 
     # callbacks, etc:
     
     def handle_stdout(self, fd, events):
         if WINDOWS:
             line = self.stdout.recv()
         else:
             line = self.process.stdout.readline()
         # a stopped process will be readable but return empty strings
         if line:
             self.log.info(line[:-1])
         else:
             self.poll()
     
     def handle_stderr(self, fd, events):
         if WINDOWS:
             line = self.stderr.recv()
         else:
             line = self.process.stderr.readline()
         # a stopped process will be readable but return empty strings
         if line:
             self.log.error(line[:-1])
         else:
             self.poll()
     
     def poll(self):
         status = self.process.poll()
         if status is not None:
             self.poller.stop()
             self.loop.remove_handler(self.stdout)
             self.loop.remove_handler(self.stderr)
             self.notify_stop(dict(exit_code=status, pid=self.process.pid))
         return status
 
 class LocalControllerLauncher(LocalProcessLauncher):
     """Launch a controller as a regular external process."""
 
     controller_cmd = List(ipcontroller_cmd_argv, config=True,
         help="""Popen command to launch ipcontroller.""")
     # Command line arguments to ipcontroller.
     controller_args = List(['--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="""command-line args to pass to ipcontroller""")
 
     def find_args(self):
         return self.controller_cmd + self.controller_args
 
     def start(self, profile_dir):
         """Start the controller by profile_dir."""
         self.controller_args.extend(['profile_dir=%s'%profile_dir])
         self.profile_dir = unicode(profile_dir)
         self.log.info("Starting LocalControllerLauncher: %r" % self.args)
         return super(LocalControllerLauncher, self).start()
 
 
 class LocalEngineLauncher(LocalProcessLauncher):
     """Launch a single engine as a regular externall process."""
 
     engine_cmd = List(ipengine_cmd_argv, config=True,
         help="""command to launch the Engine.""")
     # Command line arguments for ipengine.
     engine_args = List(['--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="command-line arguments to pass to ipengine"
     )
 
     def find_args(self):
         return self.engine_cmd + self.engine_args
 
     def start(self, profile_dir):
         """Start the engine by profile_dir."""
         self.engine_args.extend(['profile_dir=%s'%profile_dir])
         self.profile_dir = unicode(profile_dir)
         return super(LocalEngineLauncher, self).start()
 
 
 class LocalEngineSetLauncher(BaseLauncher):
     """Launch a set of engines as regular external processes."""
 
     # Command line arguments for ipengine.
     engine_args = List(
         ['--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="command-line arguments to pass to ipengine"
     )
     # launcher class
     launcher_class = LocalEngineLauncher
     
     launchers = Dict()
     stop_data = Dict()
     
     def __init__(self, work_dir=u'.', config=None, **kwargs):
         super(LocalEngineSetLauncher, self).__init__(
             work_dir=work_dir, config=config, **kwargs
         )
         self.stop_data = {}
 
     def start(self, n, profile_dir):
         """Start n engines by profile or profile_dir."""
         self.profile_dir = unicode(profile_dir)
         dlist = []
         for i in range(n):
             el = self.launcher_class(work_dir=self.work_dir, config=self.config, logname=self.log.name)
             # Copy the engine args over to each engine launcher.
             el.engine_args = copy.deepcopy(self.engine_args)
             el.on_stop(self._notice_engine_stopped)
             d = el.start(profile_dir)
             if i==0:
                 self.log.info("Starting LocalEngineSetLauncher: %r" % el.args)
             self.launchers[i] = el
             dlist.append(d)
         self.notify_start(dlist)
         # The consumeErrors here could be dangerous
         # dfinal = gatherBoth(dlist, consumeErrors=True)
         # dfinal.addCallback(self.notify_start)
         return dlist
 
     def find_args(self):
         return ['engine set']
 
     def signal(self, sig):
         dlist = []
         for el in self.launchers.itervalues():
             d = el.signal(sig)
             dlist.append(d)
         # dfinal = gatherBoth(dlist, consumeErrors=True)
         return dlist
 
     def interrupt_then_kill(self, delay=1.0):
         dlist = []
         for el in self.launchers.itervalues():
             d = el.interrupt_then_kill(delay)
             dlist.append(d)
         # dfinal = gatherBoth(dlist, consumeErrors=True)
         return dlist
 
     def stop(self):
         return self.interrupt_then_kill()
     
     def _notice_engine_stopped(self, data):
         pid = data['pid']
         for idx,el in self.launchers.iteritems():
             if el.process.pid == pid:
                 break
         self.launchers.pop(idx)
         self.stop_data[idx] = data
         if not self.launchers:
             self.notify_stop(self.stop_data)
 
 
 #-----------------------------------------------------------------------------
 # MPIExec launchers
 #-----------------------------------------------------------------------------
 
 
 class MPIExecLauncher(LocalProcessLauncher):
     """Launch an external process using mpiexec."""
 
     mpi_cmd = List(['mpiexec'], config=True,
         help="The mpiexec command to use in starting the process."
     )
     mpi_args = List([], config=True,
         help="The command line arguments to pass to mpiexec."
     )
     program = List(['date'], config=True,
         help="The program to start via mpiexec.")
     program_args = List([], config=True,
         help="The command line argument to the program."
     )
     n = Int(1)
 
     def find_args(self):
         """Build self.args using all the fields."""
         return self.mpi_cmd + ['-n', str(self.n)] + self.mpi_args + \
                self.program + self.program_args
 
     def start(self, n):
         """Start n instances of the program using mpiexec."""
         self.n = n
         return super(MPIExecLauncher, self).start()
 
 
 class MPIExecControllerLauncher(MPIExecLauncher):
     """Launch a controller using mpiexec."""
 
     controller_cmd = List(ipcontroller_cmd_argv, config=True,
         help="Popen command to launch the Contropper"
     )
     controller_args = List(['--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="Command line arguments to pass to ipcontroller."
     )
     n = Int(1)
 
     def start(self, profile_dir):
         """Start the controller by profile_dir."""
         self.controller_args.extend(['profile_dir=%s'%profile_dir])
         self.profile_dir = unicode(profile_dir)
         self.log.info("Starting MPIExecControllerLauncher: %r" % self.args)
         return super(MPIExecControllerLauncher, self).start(1)
 
     def find_args(self):
         return self.mpi_cmd + ['-n', self.n] + self.mpi_args + \
                self.controller_cmd + self.controller_args
 
 
 class MPIExecEngineSetLauncher(MPIExecLauncher):
 
     program = List(ipengine_cmd_argv, config=True,
         help="Popen command for ipengine"
     )
     program_args = List(
         ['--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="Command line arguments for ipengine."
     )
     n = Int(1)
 
     def start(self, n, profile_dir):
         """Start n engines by profile or profile_dir."""
         self.program_args.extend(['profile_dir=%s'%profile_dir])
         self.profile_dir = unicode(profile_dir)
         self.n = n
         self.log.info('Starting MPIExecEngineSetLauncher: %r' % self.args)
         return super(MPIExecEngineSetLauncher, self).start(n)
 
 #-----------------------------------------------------------------------------
 # SSH launchers
 #-----------------------------------------------------------------------------
 
 # TODO: Get SSH Launcher working again.
 
 class SSHLauncher(LocalProcessLauncher):
     """A minimal launcher for ssh.
 
     To be useful this will probably have to be extended to use the ``sshx``
     idea for environment variables.  There could be other things this needs
     as well.
     """
 
     ssh_cmd = List(['ssh'], config=True,
         help="command for starting ssh")
     ssh_args = List(['-tt'], config=True,
         help="args to pass to ssh")
     program = List(['date'], config=True,
         help="Program to launch via ssh")
     program_args = List([], config=True,
         help="args to pass to remote program")
     hostname = Unicode('', config=True,
         help="hostname on which to launch the program")
     user = Unicode('', config=True,
         help="username for ssh")
     location = Unicode('', config=True,
         help="user@hostname location for ssh in one setting")
 
     def _hostname_changed(self, name, old, new):
         if self.user:
             self.location = u'%s@%s' % (self.user, new)
         else:
             self.location = new
 
     def _user_changed(self, name, old, new):
         self.location = u'%s@%s' % (new, self.hostname)
 
     def find_args(self):
         return self.ssh_cmd + self.ssh_args + [self.location] + \
                self.program + self.program_args
 
     def start(self, profile_dir, hostname=None, user=None):
         self.profile_dir = unicode(profile_dir)
         if hostname is not None:
             self.hostname = hostname
         if user is not None:
             self.user = user
         
         return super(SSHLauncher, self).start()
     
     def signal(self, sig):
         if self.state == 'running':
             # send escaped ssh connection-closer
             self.process.stdin.write('~.')
             self.process.stdin.flush()
         
 
 
 class SSHControllerLauncher(SSHLauncher):
 
     program = List(ipcontroller_cmd_argv, config=True,
         help="remote ipcontroller command.")
     program_args = List(['--reuse-files', '--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="Command line arguments to ipcontroller.")
 
 
 class SSHEngineLauncher(SSHLauncher):
     program = List(ipengine_cmd_argv, config=True,
         help="remote ipengine command.")
     # Command line arguments for ipengine.
     program_args = List(
         ['--log-to-file','log_level=%i'%logging.INFO], config=True,
         help="Command line arguments to ipengine."
     )
     
 class SSHEngineSetLauncher(LocalEngineSetLauncher):
     launcher_class = SSHEngineLauncher
     engines = Dict(config=True,
         help="""dict of engines to launch.  This is a dict by hostname of ints,
         corresponding to the number of engines to start on that host.""")
     
     def start(self, n, profile_dir):
         """Start engines by profile or profile_dir.
         `n` is ignored, and the `engines` config property is used instead.
         """
         
         self.profile_dir = unicode(profile_dir)
         dlist = []
         for host, n in self.engines.iteritems():
             if isinstance(n, (tuple, list)):
                 n, args = n
             else:
                 args = copy.deepcopy(self.engine_args)
             
             if '@' in host:
                 user,host = host.split('@',1)
             else:
                 user=None
             for i in range(n):
                 el = self.launcher_class(work_dir=self.work_dir, config=self.config, logname=self.log.name)
                 
                 # Copy the engine args over to each engine launcher.
                 i
                 el.program_args = args
                 el.on_stop(self._notice_engine_stopped)
                 d = el.start(profile_dir, user=user, hostname=host)
                 if i==0:
                     self.log.info("Starting SSHEngineSetLauncher: %r" % el.args)
                 self.launchers[host+str(i)] = el
                 dlist.append(d)
         self.notify_start(dlist)
         return dlist
     
 
 
 #-----------------------------------------------------------------------------
 # Windows HPC Server 2008 scheduler launchers
 #-----------------------------------------------------------------------------
 
 
 # This is only used on Windows.
 def find_job_cmd():
     if WINDOWS:
         try:
             return find_cmd('job')
         except (FindCmdError, ImportError):
             # ImportError will be raised if win32api is not installed
             return 'job'
     else:
         return 'job'
 
 
 class WindowsHPCLauncher(BaseLauncher):
 
     job_id_regexp = Unicode(r'\d+', config=True,
         help="""A regular expression used to get the job id from the output of the
         submit_command. """
         )
     job_file_name = Unicode(u'ipython_job.xml', config=True,
         help="The filename of the instantiated job script.")
     # The full path to the instantiated job script. This gets made dynamically
     # by combining the work_dir with the job_file_name.
     job_file = Unicode(u'')
     scheduler = Unicode('', config=True,
         help="The hostname of the scheduler to submit the job to.")
     job_cmd = Unicode(find_job_cmd(), config=True,
         help="The command for submitting jobs.")
 
     def __init__(self, work_dir=u'.', config=None, **kwargs):
         super(WindowsHPCLauncher, self).__init__(
             work_dir=work_dir, config=config, **kwargs
         )
 
     @property
     def job_file(self):
         return os.path.join(self.work_dir, self.job_file_name)
 
     def write_job_file(self, n):
         raise NotImplementedError("Implement write_job_file in a subclass.")
 
     def find_args(self):
         return [u'job.exe']
         
     def parse_job_id(self, output):
         """Take the output of the submit command and return the job id."""
         m = re.search(self.job_id_regexp, output)
         if m is not None:
             job_id = m.group()
         else:
             raise LauncherError("Job id couldn't be determined: %s" % output)
         self.job_id = job_id
         self.log.info('Job started with job id: %r' % job_id)
         return job_id
 
     def start(self, n):
         """Start n copies of the process using the Win HPC job scheduler."""
         self.write_job_file(n)
         args = [
             'submit',
             '/jobfile:%s' % self.job_file,
             '/scheduler:%s' % self.scheduler
         ]
         self.log.info("Starting Win HPC Job: %s" % (self.job_cmd + ' ' + ' '.join(args),))
         # Twisted will raise DeprecationWarnings if we try to pass unicode to this
         output = check_output([self.job_cmd]+args,
             env=os.environ,
             cwd=self.work_dir,
             stderr=STDOUT
         )
         job_id = self.parse_job_id(output)
         self.notify_start(job_id)
         return job_id
 
     def stop(self):
         args = [
             'cancel',
             self.job_id,
             '/scheduler:%s' % self.scheduler
         ]
         self.log.info("Stopping Win HPC Job: %s" % (self.job_cmd + ' ' + ' '.join(args),))
         try:
             output = check_output([self.job_cmd]+args,
                 env=os.environ,
                 cwd=self.work_dir,
                 stderr=STDOUT
             )
         except:
             output = 'The job already appears to be stoppped: %r' % self.job_id
         self.notify_stop(dict(job_id=self.job_id, output=output))  # Pass the output of the kill cmd
         return output
 
 
 class WindowsHPCControllerLauncher(WindowsHPCLauncher):
 
     job_file_name = Unicode(u'ipcontroller_job.xml', config=True,
         help="WinHPC xml job file.")
     extra_args = List([], config=False,
         help="extra args to pass to ipcontroller")
 
     def write_job_file(self, n):
         job = IPControllerJob(config=self.config)
 
         t = IPControllerTask(config=self.config)
         # The tasks work directory is *not* the actual work directory of 
         # the controller. It is used as the base path for the stdout/stderr
         # files that the scheduler redirects to.
         t.work_directory = self.profile_dir
         # Add the profile_dir and from self.start().
         t.controller_args.extend(self.extra_args)
         job.add_task(t)
 
         self.log.info("Writing job description file: %s" % self.job_file)
         job.write(self.job_file)
 
     @property
     def job_file(self):
         return os.path.join(self.profile_dir, self.job_file_name)
 
     def start(self, profile_dir):
         """Start the controller by profile_dir."""
         self.extra_args = ['profile_dir=%s'%profile_dir]
         self.profile_dir = unicode(profile_dir)
         return super(WindowsHPCControllerLauncher, self).start(1)
 
 
 class WindowsHPCEngineSetLauncher(WindowsHPCLauncher):
 
     job_file_name = Unicode(u'ipengineset_job.xml', config=True,
         help="jobfile for ipengines job")
     extra_args = List([], config=False,
         help="extra args to pas to ipengine")
 
     def write_job_file(self, n):
         job = IPEngineSetJob(config=self.config)
 
         for i in range(n):
             t = IPEngineTask(config=self.config)
             # The tasks work directory is *not* the actual work directory of 
             # the engine. It is used as the base path for the stdout/stderr
             # files that the scheduler redirects to.
             t.work_directory = self.profile_dir
             # Add the profile_dir and from self.start().
             t.engine_args.extend(self.extra_args)
             job.add_task(t)
 
         self.log.info("Writing job description file: %s" % self.job_file)
         job.write(self.job_file)
 
     @property
     def job_file(self):
         return os.path.join(self.profile_dir, self.job_file_name)
 
     def start(self, n, profile_dir):
         """Start the controller by profile_dir."""
         self.extra_args = ['profile_dir=%s'%profile_dir]
         self.profile_dir = unicode(profile_dir)
         return super(WindowsHPCEngineSetLauncher, self).start(n)
 
 
 #-----------------------------------------------------------------------------
 # Batch (PBS) system launchers
 #-----------------------------------------------------------------------------
 
 class BatchSystemLauncher(BaseLauncher):
     """Launch an external process using a batch system.
 
     This class is designed to work with UNIX batch systems like PBS, LSF,
     GridEngine, etc.  The overall model is that there are different commands
     like qsub, qdel, etc. that handle the starting and stopping of the process.
 
     This class also has the notion of a batch script. The ``batch_template``
     attribute can be set to a string that is a template for the batch script.
     This template is instantiated using string formatting. Thus the template can
     use {n} fot the number of instances. Subclasses can add additional variables
     to the template dict.
     """
 
     # Subclasses must fill these in.  See PBSEngineSet
     submit_command = List([''], config=True,
         help="The name of the command line program used to submit jobs.")
     delete_command = List([''], config=True,
         help="The name of the command line program used to delete jobs.")
     job_id_regexp = Unicode('', config=True,
         help="""A regular expression used to get the job id from the output of the
         submit_command.""")
     batch_template = Unicode('', config=True,
         help="The string that is the batch script template itself.")
     batch_template_file = Unicode(u'', config=True,
         help="The file that contains the batch template.")
     batch_file_name = Unicode(u'batch_script', config=True,
         help="The filename of the instantiated batch script.")
     queue = Unicode(u'', config=True,
         help="The PBS Queue.")
     
     # not configurable, override in subclasses
     # PBS Job Array regex
     job_array_regexp = Unicode('')
     job_array_template = Unicode('')
     # PBS Queue regex
     queue_regexp = Unicode('')
     queue_template = Unicode('')
     # The default batch template, override in subclasses
     default_template = Unicode('')
     # The full path to the instantiated batch script.
     batch_file = Unicode(u'')
     # the format dict used with batch_template:
     context = Dict()
+    # the Formatter instance for rendering the templates:
+    formatter = Instance(EvalFormatter, (), {})
 
     
     def find_args(self):
         return self.submit_command + [self.batch_file]
     
     def __init__(self, work_dir=u'.', config=None, **kwargs):
         super(BatchSystemLauncher, self).__init__(
             work_dir=work_dir, config=config, **kwargs
         )
         self.batch_file = os.path.join(self.work_dir, self.batch_file_name)
 
     def parse_job_id(self, output):
         """Take the output of the submit command and return the job id."""
         m = re.search(self.job_id_regexp, output)
         if m is not None:
             job_id = m.group()
         else:
             raise LauncherError("Job id couldn't be determined: %s" % output)
         self.job_id = job_id
         self.log.info('Job submitted with job id: %r' % job_id)
         return job_id
 
     def write_batch_script(self, n):
         """Instantiate and write the batch script to the work_dir."""
         self.context['n'] = n
         self.context['queue'] = self.queue
         # first priority is batch_template if set
         if self.batch_template_file and not self.batch_template:
             # second priority is batch_template_file
             with open(self.batch_template_file) as f:
                 self.batch_template = f.read()
         if not self.batch_template:
             # third (last) priority is default_template
             self.batch_template = self.default_template
         
         regex = re.compile(self.job_array_regexp)
         # print regex.search(self.batch_template)
         if not regex.search(self.batch_template):
             self.log.info("adding job array settings to batch script")
             firstline, rest = self.batch_template.split('\n',1)
             self.batch_template = u'\n'.join([firstline, self.job_array_template, rest])
         
         regex = re.compile(self.queue_regexp)
         # print regex.search(self.batch_template)
         if self.queue and not regex.search(self.batch_template):
             self.log.info("adding PBS queue settings to batch script")
             firstline, rest = self.batch_template.split('\n',1)
             self.batch_template = u'\n'.join([firstline, self.queue_template, rest])
             
-        script_as_string = self.batch_template.format(**self.context)
+        script_as_string = self.formatter.format(self.batch_template, **self.context)
         self.log.info('Writing instantiated batch script: %s' % self.batch_file)
     
         with open(self.batch_file, 'w') as f:
             f.write(script_as_string)
         os.chmod(self.batch_file, stat.S_IRUSR | stat.S_IWUSR | stat.S_IXUSR)
 
     def start(self, n, profile_dir):
         """Start n copies of the process using a batch system."""
         # Here we save profile_dir in the context so they
         # can be used in the batch script template as {profile_dir}
         self.context['profile_dir'] = profile_dir
         self.profile_dir = unicode(profile_dir)
         self.write_batch_script(n)
         output = check_output(self.args, env=os.environ)
         
         job_id = self.parse_job_id(output)
         self.notify_start(job_id)
         return job_id
 
     def stop(self):
         output = check_output(self.delete_command+[self.job_id], env=os.environ)
         self.notify_stop(dict(job_id=self.job_id, output=output)) # Pass the output of the kill cmd
         return output
 
 
 class PBSLauncher(BatchSystemLauncher):
     """A BatchSystemLauncher subclass for PBS."""
 
     submit_command = List(['qsub'], config=True,
         help="The PBS submit command ['qsub']")
     delete_command = List(['qdel'], config=True,
         help="The PBS delete command ['qsub']")
     job_id_regexp = Unicode(r'\d+', config=True,
         help="Regular expresion for identifying the job ID [r'\d+']")
     
     batch_file = Unicode(u'')
     job_array_regexp = Unicode('#PBS\W+-t\W+[\w\d\-\$]+')
     job_array_template = Unicode('#PBS -t 1-{n}')
     queue_regexp = Unicode('#PBS\W+-q\W+\$?\w+')
     queue_template = Unicode('#PBS -q {queue}')
 
 
 class PBSControllerLauncher(PBSLauncher):
     """Launch a controller using PBS."""
 
     batch_file_name = Unicode(u'pbs_controller', config=True,
         help="batch file name for the controller job.")
     default_template= Unicode("""#!/bin/sh
 #PBS -V
 #PBS -N ipcontroller
 %s --log-to-file profile_dir={profile_dir}
 """%(' '.join(ipcontroller_cmd_argv)))
 
     def start(self, profile_dir):
         """Start the controller by profile or profile_dir."""
         self.log.info("Starting PBSControllerLauncher: %r" % self.args)
         return super(PBSControllerLauncher, self).start(1, profile_dir)
 
 
 class PBSEngineSetLauncher(PBSLauncher):
     """Launch Engines using PBS"""
     batch_file_name = Unicode(u'pbs_engines', config=True,
         help="batch file name for the engine(s) job.")
     default_template= Unicode(u"""#!/bin/sh
 #PBS -V
 #PBS -N ipengine
 %s profile_dir={profile_dir}
 """%(' '.join(ipengine_cmd_argv)))
 
     def start(self, n, profile_dir):
         """Start n engines by profile or profile_dir."""
         self.log.info('Starting %i engines with PBSEngineSetLauncher: %r' % (n, self.args))
         return super(PBSEngineSetLauncher, self).start(n, profile_dir)
 
 #SGE is very similar to PBS
 
 class SGELauncher(PBSLauncher):
     """Sun GridEngine is a PBS clone with slightly different syntax"""
     job_array_regexp = Unicode('#\$\W+\-t')
     job_array_template = Unicode('#$ -t 1-{n}')
     queue_regexp = Unicode('#\$\W+-q\W+\$?\w+')
     queue_template = Unicode('#$ -q $queue')
 
 class SGEControllerLauncher(SGELauncher):
     """Launch a controller using SGE."""
 
     batch_file_name = Unicode(u'sge_controller', config=True,
         help="batch file name for the ipontroller job.")
     default_template= Unicode(u"""#$ -V
 #$ -S /bin/sh
 #$ -N ipcontroller
 %s --log-to-file profile_dir={profile_dir}
 """%(' '.join(ipcontroller_cmd_argv)))
 
     def start(self, profile_dir):
         """Start the controller by profile or profile_dir."""
         self.log.info("Starting PBSControllerLauncher: %r" % self.args)
         return super(SGEControllerLauncher, self).start(1, profile_dir)
 
 class SGEEngineSetLauncher(SGELauncher):
     """Launch Engines with SGE"""
     batch_file_name = Unicode(u'sge_engines', config=True,
         help="batch file name for the engine(s) job.")
     default_template = Unicode("""#$ -V
 #$ -S /bin/sh
 #$ -N ipengine
 %s profile_dir={profile_dir}
 """%(' '.join(ipengine_cmd_argv)))
 
     def start(self, n, profile_dir):
         """Start n engines by profile or profile_dir."""
         self.log.info('Starting %i engines with SGEEngineSetLauncher: %r' % (n, self.args))
         return super(SGEEngineSetLauncher, self).start(n, profile_dir)
 
 
 #-----------------------------------------------------------------------------
 # A launcher for ipcluster itself!
 #-----------------------------------------------------------------------------
 
 
 class IPClusterLauncher(LocalProcessLauncher):
     """Launch the ipcluster program in an external process."""
 
     ipcluster_cmd = List(ipcluster_cmd_argv, config=True,
         help="Popen command for ipcluster")
     ipcluster_args = List(
         ['--clean-logs', '--log-to-file', 'log_level=%i'%logging.INFO], config=True,
         help="Command line arguments to pass to ipcluster.")
     ipcluster_subcommand = Unicode('start')
     ipcluster_n = Int(2)
 
     def find_args(self):
         return self.ipcluster_cmd + ['--'+self.ipcluster_subcommand] + \
             ['n=%i'%self.ipcluster_n] + self.ipcluster_args
 
     def start(self):
         self.log.info("Starting ipcluster: %r" % self.args)
         return super(IPClusterLauncher, self).start()
 
 #-----------------------------------------------------------------------------
 # Collections of launchers
 #-----------------------------------------------------------------------------
 
 local_launchers = [
     LocalControllerLauncher,
     LocalEngineLauncher,
     LocalEngineSetLauncher,
 ]
 mpi_launchers = [
     MPIExecLauncher,
     MPIExecControllerLauncher,
     MPIExecEngineSetLauncher,
 ]
 ssh_launchers = [
     SSHLauncher,
     SSHControllerLauncher,
     SSHEngineLauncher,
     SSHEngineSetLauncher,
 ]
 winhpc_launchers = [
     WindowsHPCLauncher,
     WindowsHPCControllerLauncher,
     WindowsHPCEngineSetLauncher,
 ]
 pbs_launchers = [
     PBSLauncher,
     PBSControllerLauncher,
     PBSEngineSetLauncher,
 ]
 sge_launchers = [
     SGELauncher,
     SGEControllerLauncher,
     SGEEngineSetLauncher,
 ]
 all_launchers = local_launchers + mpi_launchers + ssh_launchers + winhpc_launchers\
                 + pbs_launchers + sge_launchers

IPython/utils/text.py

@@ -1,521 +1,560 b''
 # encoding: utf-8
 """
 Utilities for working with strings and text.
 """
 
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2008-2009  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 
 import __main__
 
 import os
 import re
 import shutil
+from string import Formatter
 
 from IPython.external.path import path
 
 from IPython.utils.io import nlprint
 from IPython.utils.data import flatten
 
 #-----------------------------------------------------------------------------
 # Code
 #-----------------------------------------------------------------------------
 
 
 def unquote_ends(istr):
     """Remove a single pair of quotes from the endpoints of a string."""
 
     if not istr:
         return istr
     if (istr[0]=="'" and istr[-1]=="'") or \
        (istr[0]=='"' and istr[-1]=='"'):
         return istr[1:-1]
     else:
         return istr
 
 
 class LSString(str):
     """String derivative with a special access attributes.
 
     These are normal strings, but with the special attributes:
 
         .l (or .list) : value as list (split on newlines).
         .n (or .nlstr): original value (the string itself).
         .s (or .spstr): value as whitespace-separated string.
         .p (or .paths): list of path objects
 
     Any values which require transformations are computed only once and
     cached.
 
     Such strings are very useful to efficiently interact with the shell, which
     typically only understands whitespace-separated options for commands."""
 
     def get_list(self):
         try:
             return self.__list
         except AttributeError:
             self.__list = self.split('\n')
             return self.__list
 
     l = list = property(get_list)
 
     def get_spstr(self):
         try:
             return self.__spstr
         except AttributeError:
             self.__spstr = self.replace('\n',' ')
             return self.__spstr
 
     s = spstr = property(get_spstr)
 
     def get_nlstr(self):
         return self
 
     n = nlstr = property(get_nlstr)
 
     def get_paths(self):
         try:
             return self.__paths
         except AttributeError:
             self.__paths = [path(p) for p in self.split('\n') if os.path.exists(p)]
             return self.__paths
 
     p = paths = property(get_paths)
 
 # FIXME: We need to reimplement type specific displayhook and then add this
 # back as a custom printer. This should also be moved outside utils into the
 # core.
 
 # def print_lsstring(arg):
 #     """ Prettier (non-repr-like) and more informative printer for LSString """
 #     print "LSString (.p, .n, .l, .s available). Value:"
 #     print arg
 # 
 # 
 # print_lsstring = result_display.when_type(LSString)(print_lsstring)
 
 
 class SList(list):
     """List derivative with a special access attributes.
 
     These are normal lists, but with the special attributes:
 
         .l (or .list) : value as list (the list itself).
         .n (or .nlstr): value as a string, joined on newlines.
         .s (or .spstr): value as a string, joined on spaces.
         .p (or .paths): list of path objects
 
     Any values which require transformations are computed only once and
     cached."""
 
     def get_list(self):
         return self
 
     l = list = property(get_list)
 
     def get_spstr(self):
         try:
             return self.__spstr
         except AttributeError:
             self.__spstr = ' '.join(self)
             return self.__spstr
 
     s = spstr = property(get_spstr)
 
     def get_nlstr(self):
         try:
             return self.__nlstr
         except AttributeError:
             self.__nlstr = '\n'.join(self)
             return self.__nlstr
 
     n = nlstr = property(get_nlstr)
 
     def get_paths(self):
         try:
             return self.__paths
         except AttributeError:
             self.__paths = [path(p) for p in self if os.path.exists(p)]
             return self.__paths
 
     p = paths = property(get_paths)
 
     def grep(self, pattern, prune = False, field = None):
         """ Return all strings matching 'pattern' (a regex or callable)
 
         This is case-insensitive. If prune is true, return all items
         NOT matching the pattern.
 
         If field is specified, the match must occur in the specified
         whitespace-separated field.
 
         Examples::
 
             a.grep( lambda x: x.startswith('C') )
             a.grep('Cha.*log', prune=1)
             a.grep('chm', field=-1)
         """
 
         def match_target(s):
             if field is None:
                 return s
             parts = s.split()
             try:
                 tgt = parts[field]
                 return tgt
             except IndexError:
                 return ""
 
         if isinstance(pattern, basestring):
             pred = lambda x : re.search(pattern, x, re.IGNORECASE)
         else:
             pred = pattern
         if not prune:
             return SList([el for el in self if pred(match_target(el))])
         else:
             return SList([el for el in self if not pred(match_target(el))])
 
     def fields(self, *fields):
         """ Collect whitespace-separated fields from string list
 
         Allows quick awk-like usage of string lists.
 
         Example data (in var a, created by 'a = !ls -l')::
             -rwxrwxrwx  1 ville None      18 Dec 14  2006 ChangeLog
             drwxrwxrwx+ 6 ville None       0 Oct 24 18:05 IPython
 
         a.fields(0) is ['-rwxrwxrwx', 'drwxrwxrwx+']
         a.fields(1,0) is ['1 -rwxrwxrwx', '6 drwxrwxrwx+']
         (note the joining by space).
         a.fields(-1) is ['ChangeLog', 'IPython']
 
         IndexErrors are ignored.
 
         Without args, fields() just split()'s the strings.
         """
         if len(fields) == 0:
             return [el.split() for el in self]
 
         res = SList()
         for el in [f.split() for f in self]:
             lineparts = []
 
             for fd in fields:
                 try:
                     lineparts.append(el[fd])
                 except IndexError:
                     pass
             if lineparts:
                 res.append(" ".join(lineparts))
 
         return res
 
     def sort(self,field= None,  nums = False):
         """ sort by specified fields (see fields())
 
         Example::
             a.sort(1, nums = True)
 
         Sorts a by second field, in numerical order (so that 21 > 3)
 
         """
 
         #decorate, sort, undecorate
         if field is not None:
             dsu = [[SList([line]).fields(field),  line] for line in self]
         else:
             dsu = [[line,  line] for line in self]
         if nums:
             for i in range(len(dsu)):
                 numstr = "".join([ch for ch in dsu[i][0] if ch.isdigit()])
                 try:
                     n = int(numstr)
                 except ValueError:
                     n = 0;
                 dsu[i][0] = n
 
 
         dsu.sort()
         return SList([t[1] for t in dsu])
 
 
 # FIXME: We need to reimplement type specific displayhook and then add this
 # back as a custom printer. This should also be moved outside utils into the
 # core.
 
 # def print_slist(arg):
 #     """ Prettier (non-repr-like) and more informative printer for SList """
 #     print "SList (.p, .n, .l, .s, .grep(), .fields(), sort() available):"
 #     if hasattr(arg,  'hideonce') and arg.hideonce:
 #         arg.hideonce = False
 #         return
 # 
 #     nlprint(arg)
 # 
 # print_slist = result_display.when_type(SList)(print_slist)
 
 
 def esc_quotes(strng):
     """Return the input string with single and double quotes escaped out"""
 
     return strng.replace('"','\\"').replace("'","\\'")
 
 
 def make_quoted_expr(s):
     """Return string s in appropriate quotes, using raw string if possible.
 
     XXX - example removed because it caused encoding errors in documentation
     generation.  We need a new example that doesn't contain invalid chars.
 
     Note the use of raw string and padding at the end to allow trailing
     backslash.
     """
 
     tail = ''
     tailpadding = ''
     raw  = ''
     ucode = 'u'
     if "\\" in s:
         raw = 'r'
         if s.endswith('\\'):
             tail = '[:-1]'
             tailpadding = '_'
     if '"' not in s:
         quote = '"'
     elif "'" not in s:
         quote = "'"
     elif '"""' not in s and not s.endswith('"'):
         quote = '"""'
     elif "'''" not in s and not s.endswith("'"):
         quote = "'''"
     else:
         # give up, backslash-escaped string will do
         return '"%s"' % esc_quotes(s)
     res = ucode + raw + quote + s + tailpadding + quote + tail
     return res
 
 
 def qw(words,flat=0,sep=None,maxsplit=-1):
     """Similar to Perl's qw() operator, but with some more options.
 
     qw(words,flat=0,sep=' ',maxsplit=-1) -> words.split(sep,maxsplit)
 
     words can also be a list itself, and with flat=1, the output will be
     recursively flattened.
 
     Examples:
 
     >>> qw('1 2')
     ['1', '2']
 
     >>> qw(['a b','1 2',['m n','p q']])
     [['a', 'b'], ['1', '2'], [['m', 'n'], ['p', 'q']]]
 
     >>> qw(['a b','1 2',['m n','p q']],flat=1)
     ['a', 'b', '1', '2', 'm', 'n', 'p', 'q']
     """
 
     if isinstance(words, basestring):
         return [word.strip() for word in words.split(sep,maxsplit)
                 if word and not word.isspace() ]
     if flat:
         return flatten(map(qw,words,[1]*len(words)))
     return map(qw,words)
 
 
 def qwflat(words,sep=None,maxsplit=-1):
     """Calls qw(words) in flat mode. It's just a convenient shorthand."""
     return qw(words,1,sep,maxsplit)
 
 
 def qw_lol(indata):
     """qw_lol('a b') -> [['a','b']],
     otherwise it's just a call to qw().
 
     We need this to make sure the modules_some keys *always* end up as a
     list of lists."""
 
     if isinstance(indata, basestring):
         return [qw(indata)]
     else:
         return qw(indata)
 
 
 def grep(pat,list,case=1):
     """Simple minded grep-like function.
     grep(pat,list) returns occurrences of pat in list, None on failure.
 
     It only does simple string matching, with no support for regexps. Use the
     option case=0 for case-insensitive matching."""
 
     # This is pretty crude. At least it should implement copying only references
     # to the original data in case it's big. Now it copies the data for output.
     out=[]
     if case:
         for term in list:
             if term.find(pat)>-1: out.append(term)
     else:
         lpat=pat.lower()
         for term in list:
             if term.lower().find(lpat)>-1: out.append(term)
 
     if len(out): return out
     else: return None
 
 
 def dgrep(pat,*opts):
     """Return grep() on dir()+dir(__builtins__).
 
     A very common use of grep() when working interactively."""
 
     return grep(pat,dir(__main__)+dir(__main__.__builtins__),*opts)
 
 
 def idgrep(pat):
     """Case-insensitive dgrep()"""
 
     return dgrep(pat,0)
 
 
 def igrep(pat,list):
     """Synonym for case-insensitive grep."""
 
     return grep(pat,list,case=0)
 
 
 def indent(instr,nspaces=4, ntabs=0, flatten=False):
     """Indent a string a given number of spaces or tabstops.
 
     indent(str,nspaces=4,ntabs=0) -> indent str by ntabs+nspaces.
     
     Parameters
     ----------
     
     instr : basestring
         The string to be indented.
     nspaces : int (default: 4)
         The number of spaces to be indented.
     ntabs : int (default: 0)
         The number of tabs to be indented.
     flatten : bool (default: False)
         Whether to scrub existing indentation.  If True, all lines will be
         aligned to the same indentation.  If False, existing indentation will
         be strictly increased.
     
     Returns
     -------
     
     str|unicode : string indented by ntabs and nspaces.
     
     """
     if instr is None:
         return
     ind = '\t'*ntabs+' '*nspaces
     if flatten:
         pat = re.compile(r'^\s*', re.MULTILINE)
     else:
         pat = re.compile(r'^', re.MULTILINE)
     outstr = re.sub(pat, ind, instr)
     if outstr.endswith(os.linesep+ind):
         return outstr[:-len(ind)]
     else:
         return outstr
 
 def native_line_ends(filename,backup=1):
     """Convert (in-place) a file to line-ends native to the current OS.
 
     If the optional backup argument is given as false, no backup of the
     original file is left.  """
 
     backup_suffixes = {'posix':'~','dos':'.bak','nt':'.bak','mac':'.bak'}
 
     bak_filename = filename + backup_suffixes[os.name]
 
     original = open(filename).read()
     shutil.copy2(filename,bak_filename)
     try:
         new = open(filename,'wb')
         new.write(os.linesep.join(original.splitlines()))
         new.write(os.linesep) # ALWAYS put an eol at the end of the file
         new.close()
     except:
         os.rename(bak_filename,filename)
     if not backup:
         try:
             os.remove(bak_filename)
         except:
             pass
 
 
 def list_strings(arg):
     """Always return a list of strings, given a string or list of strings
     as input.
 
     :Examples:
 
         In [7]: list_strings('A single string')
         Out[7]: ['A single string']
 
         In [8]: list_strings(['A single string in a list'])
         Out[8]: ['A single string in a list']
 
         In [9]: list_strings(['A','list','of','strings'])
         Out[9]: ['A', 'list', 'of', 'strings']
     """
 
     if isinstance(arg,basestring): return [arg]
     else: return arg
 
 
 def marquee(txt='',width=78,mark='*'):
     """Return the input string centered in a 'marquee'.
 
     :Examples:
 
         In [16]: marquee('A test',40)
         Out[16]: '**************** A test ****************'
 
         In [17]: marquee('A test',40,'-')
         Out[17]: '---------------- A test ----------------'
 
         In [18]: marquee('A test',40,' ')
         Out[18]: '                 A test                 '
 
     """
     if not txt:
         return (mark*width)[:width]
     nmark = (width-len(txt)-2)/len(mark)/2
     if nmark < 0: nmark =0
     marks = mark*nmark
     return '%s %s %s' % (marks,txt,marks)
 
 
 ini_spaces_re = re.compile(r'^(\s+)')
 
 def num_ini_spaces(strng):
     """Return the number of initial spaces in a string"""
 
     ini_spaces = ini_spaces_re.match(strng)
     if ini_spaces:
         return ini_spaces.end()
     else:
         return 0
 
 
 def format_screen(strng):
     """Format a string for screen printing.
 
     This removes some latex-type format codes."""
     # Paragraph continue
     par_re = re.compile(r'\\$',re.MULTILINE)
     strng = par_re.sub('',strng)
     return strng
 
+
+class EvalFormatter(Formatter):
+    """A String Formatter that allows evaluation of simple expressions.
+    
+    Any time a format key is not found in the kwargs,
+    it will be tried as an expression in the kwargs namespace.
+    
+    This is to be used in templating cases, such as the parallel batch
+    script templates, where simple arithmetic on arguments is useful.
+    
+    Examples
+    --------
+    
+    In [1]: f = EvalFormatter()
+    In [2]: f.format('{n/4}', n=8)
+    Out[2]: '2'
+    
+    In [3]: f.format('{range(3)}')
+    Out[3]: '[0, 1, 2]'
+
+    In [4]: f.format('{3*2}')
+    Out[4]: '6'
+    """
+    
+    def get_value(self, key, args, kwargs):
+        if isinstance(key, (int, long)):
+            return args[key]
+        elif key in kwargs:
+            return kwargs[key]
+        else:
+            # evaluate the expression using kwargs as namespace
+            try:
+                return eval(key, kwargs)
+            except Exception:
+                # classify all bad expressions as key errors
+                raise KeyError(key)
+
+

docs/source/parallel/parallel_process.txt

@@ -1,507 +1,504 b''
 .. _parallel_process:
 
 ===========================================
 Starting the IPython controller and engines
 ===========================================
 
 To use IPython for parallel computing, you need to start one instance of
 the controller and one or more instances of the engine. The controller
 and each engine can run on different machines or on the same machine.
 Because of this, there are many different possibilities.
 
 Broadly speaking, there are two ways of going about starting a controller and engines:
 
 * In an automated manner using the :command:`ipcluster` command.
 * In a more manual way using the :command:`ipcontroller` and 
   :command:`ipengine` commands.
 
 This document describes both of these methods. We recommend that new users
 start with the :command:`ipcluster` command as it simplifies many common usage
 cases.
 
 General considerations
 ======================
 
 Before delving into the details about how you can start a controller and
 engines using the various methods, we outline some of the general issues that
 come up when starting the controller and engines. These things come up no
 matter which method you use to start your IPython cluster.
 
 Let's say that you want to start the controller on ``host0`` and engines on
 hosts ``host1``-``hostn``. The following steps are then required:
 
 1. Start the controller on ``host0`` by running :command:`ipcontroller` on
    ``host0``.
 2. Move the JSON file (:file:`ipcontroller-engine.json`) created by the 
    controller from ``host0`` to hosts ``host1``-``hostn``.
 3. Start the engines on hosts ``host1``-``hostn`` by running 
    :command:`ipengine`.  This command has to be told where the JSON file 
    (:file:`ipcontroller-engine.json`) is located.
 
 At this point, the controller and engines will be connected. By default, the JSON files
 created by the controller are put into the :file:`~/.ipython/cluster_default/security`
 directory. If the engines share a filesystem with the controller, step 2 can be skipped as
 the engines will automatically look at that location.
 
 The final step required to actually use the running controller from a client is to move
 the JSON file :file:`ipcontroller-client.json` from ``host0`` to any host where clients
 will be run. If these file are put into the :file:`~/.ipython/cluster_default/security`
 directory of the client's host, they will be found automatically. Otherwise, the full path
 to them has to be passed to the client's constructor.
 
 Using :command:`ipcluster`
 ===========================
 
 The :command:`ipcluster` command provides a simple way of starting a
 controller and engines in the following situations:
 
 1. When the controller and engines are all run on localhost. This is useful
    for testing or running on a multicore computer.
 2. When engines are started using the :command:`mpiexec` command that comes 
    with most MPI [MPI]_ implementations
 3. When engines are started using the PBS [PBS]_ batch system 
    (or other `qsub` systems, such as SGE).
 4. When the controller is started on localhost and the engines are started on 
    remote nodes using :command:`ssh`.
 5. When engines are started using the Windows HPC Server batch system.
 
 .. note::
 
     Currently :command:`ipcluster` requires that the
     :file:`~/.ipython/cluster_<profile>/security` directory live on a shared filesystem that is
     seen by both the controller and engines. If you don't have a shared file
     system you will need to use :command:`ipcontroller` and
     :command:`ipengine` directly.
 
 Under the hood, :command:`ipcluster` just uses :command:`ipcontroller`
 and :command:`ipengine` to perform the steps described above.
 
 The simplest way to use ipcluster requires no configuration, and will
 launch a controller and a number of engines on the local machine. For instance, 
 to start one controller and 4 engines on localhost, just do::
 
     $ ipcluster start n=4
 
 To see other command line options, do::
 
     $ ipcluster -h
 
 
 Configuring an IPython cluster
 ==============================
 
 Cluster configurations are stored as `profiles`.  You can create a new profile with::
 
     $ ipcluster create profile=myprofile
 
 This will create the directory :file:`IPYTHONDIR/cluster_myprofile`, and populate it
 with the default configuration files for the three IPython cluster commands. Once
 you edit those files, you can continue to call ipcluster/ipcontroller/ipengine
 with no arguments beyond ``p=myprofile``, and any configuration will be maintained.
 
 There is no limit to the number of profiles you can have, so you can maintain a profile for each
 of your common use cases. The default profile will be used whenever the
 profile argument is not specified, so edit :file:`IPYTHONDIR/cluster_default/*_config.py` to
 represent your most common use case.
 
 The configuration files are loaded with commented-out settings and explanations,
 which should cover most of the available possibilities.
 
 Using various batch systems with :command:`ipcluster`
 ------------------------------------------------------
 
 :command:`ipcluster` has a notion of Launchers that can start controllers
 and engines with various remote execution schemes.  Currently supported
 models include :command:`ssh`, :command`mpiexec`, PBS-style (Torque, SGE),
 and Windows HPC Server.
 
 .. note::
 
     The Launchers and configuration are designed in such a way that advanced
     users can subclass and configure them to fit their own system that we
     have not yet supported (such as Condor)
 
 Using :command:`ipcluster` in mpiexec/mpirun mode
 --------------------------------------------------
 
 
 The mpiexec/mpirun mode is useful if you:
 
 1. Have MPI installed.
 2. Your systems are configured to use the :command:`mpiexec` or
    :command:`mpirun` commands to start MPI processes.
 
 If these are satisfied, you can create a new profile::
 
     $ ipcluster create profile=mpi
 
 and edit the file :file:`IPYTHONDIR/cluster_mpi/ipcluster_config.py`.
 
 There, instruct ipcluster to use the MPIExec launchers by adding the lines:
 
 .. sourcecode:: python
 
     c.IPClusterEnginesApp.engine_launcher = 'IPython.parallel.apps.launcher.MPIExecEngineSetLauncher'
 
 If the default MPI configuration is correct, then you can now start your cluster, with::
 
     $ ipcluster start n=4 profile=mpi
 
 This does the following:
 
 1. Starts the IPython controller on current host.
 2. Uses :command:`mpiexec` to start 4 engines.
 
 If you have a reason to also start the Controller with mpi, you can specify:
 
 .. sourcecode:: python
 
     c.IPClusterStartApp.controller_launcher = 'IPython.parallel.apps.launcher.MPIExecControllerLauncher'
 
 .. note::
 
     The Controller *will not* be in the same MPI universe as the engines, so there is not
     much reason to do this unless sysadmins demand it.
 
 On newer MPI implementations (such as OpenMPI), this will work even if you
 don't make any calls to MPI or call :func:`MPI_Init`. However, older MPI
 implementations actually require each process to call :func:`MPI_Init` upon
 starting. The easiest way of having this done is to install the mpi4py
 [mpi4py]_ package and then specify the ``c.MPI.use`` option in :file:`ipengine_config.py`:
 
 .. sourcecode:: python
 
     c.MPI.use = 'mpi4py'
 
 Unfortunately, even this won't work for some MPI implementations. If you are
 having problems with this, you will likely have to use a custom Python
 executable that itself calls :func:`MPI_Init` at the appropriate time.
 Fortunately, mpi4py comes with such a custom Python executable that is easy to
 install and use. However, this custom Python executable approach will not work
 with :command:`ipcluster` currently.
 
 More details on using MPI with IPython can be found :ref:`here <parallelmpi>`.
 
 
 Using :command:`ipcluster` in PBS mode
 ---------------------------------------
 
 The PBS mode uses the Portable Batch System [PBS]_ to start the engines. 
 
 As usual, we will start by creating a fresh profile::
 
     $ ipcluster create profile=pbs
 
 And in :file:`ipcluster_config.py`, we will select the PBS launchers for the controller
 and engines:
 
 .. sourcecode:: python
 
     c.Global.controller_launcher = 'IPython.parallel.apps.launcher.PBSControllerLauncher'
     c.Global.engine_launcher = 'IPython.parallel.apps.launcher.PBSEngineSetLauncher'
 
 IPython does provide simple default batch templates for PBS and SGE, but you may need
 to specify your own. Here is a sample PBS script template:
 
 .. sourcecode:: bash
 
     #PBS -N ipython
     #PBS -j oe
     #PBS -l walltime=00:10:00
-    #PBS -l nodes=${n/4}:ppn=4
-    #PBS -q $queue
+    #PBS -l nodes={n/4}:ppn=4
+    #PBS -q {queue}
 
-    cd $$PBS_O_WORKDIR
-    export PATH=$$HOME/usr/local/bin
-    export PYTHONPATH=$$HOME/usr/local/lib/python2.7/site-packages
-    /usr/local/bin/mpiexec -n ${n} ipengine cluster_dir=${cluster_dir}
+    cd $PBS_O_WORKDIR
+    export PATH=$HOME/usr/local/bin
+    export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages
+    /usr/local/bin/mpiexec -n {n} ipengine profile_dir={profile_dir}
 
 There are a few important points about this template:
 
-1. This template will be rendered at runtime using IPython's :mod:`Itpl` 
-   template engine.
+1. This template will be rendered at runtime using IPython's :class:`EvalFormatter`.
+   This is simply a subclass of :class:`string.Formatter` that allows simple expressions
+   on keys.
 
 2. Instead of putting in the actual number of engines, use the notation
-   ``${n}`` to indicate the number of engines to be started. You can also uses
-   expressions like ``${n/4}`` in the template to indicate the number of 
-   nodes. There will always be a ${n} and ${cluster_dir} variable passed to the template.
+   ``{n}`` to indicate the number of engines to be started. You can also use
+   expressions like ``{n/4}`` in the template to indicate the number of nodes.
+   There will always be ``{n}`` and ``{profile_dir}`` variables passed to the formatter.
    These allow the batch system to know how many engines, and where the configuration
-   files reside. The same is true for the batch queue, with the template variable ``$queue``.
+   files reside. The same is true for the batch queue, with the template variable
+   ``{queue}``.
 
-3. Because ``$`` is a special character used by the template engine, you must  
-   escape any ``$`` by using ``$$``.  This is important when referring to 
-   environment variables in the template, or in SGE, where the config lines start
-   with ``#$``, which will have to be ``#$$``.
-
-4. Any options to :command:`ipengine` can be given in the batch script 
+3. Any options to :command:`ipengine` can be given in the batch script
    template, or in :file:`ipengine_config.py`.
 
-5. Depending on the configuration of you system, you may have to set 
+4. Depending on the configuration of you system, you may have to set
    environment variables in the script template.
 
 The controller template should be similar, but simpler:
 
 .. sourcecode:: bash
 
     #PBS -N ipython
     #PBS -j oe
     #PBS -l walltime=00:10:00
     #PBS -l nodes=1:ppn=4
-    #PBS -q $queue
+    #PBS -q {queue}
 
-    cd $$PBS_O_WORKDIR
-    export PATH=$$HOME/usr/local/bin
-    export PYTHONPATH=$$HOME/usr/local/lib/python2.7/site-packages
-    ipcontroller cluster_dir=${cluster_dir}
+    cd $PBS_O_WORKDIR
+    export PATH=$HOME/usr/local/bin
+    export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages
+    ipcontroller profile_dir={profile_dir}
 
 
 Once you have created these scripts, save them with names like
 :file:`pbs.engine.template`. Now you can load them into the :file:`ipcluster_config` with:
 
 .. sourcecode:: python
 
     c.PBSEngineSetLauncher.batch_template_file = "pbs.engine.template"
     
     c.PBSControllerLauncher.batch_template_file = "pbs.controller.template"
         
 
 Alternately, you can just define the templates as strings inside :file:`ipcluster_config`.
 
 Whether you are using your own templates or our defaults, the extra configurables available are
-the number of engines to launch (``$n``, and the batch system queue to which the jobs are to be
-submitted (``$queue``)). These are configurables, and can be specified in
+the number of engines to launch (``{n}``, and the batch system queue to which the jobs are to be
+submitted (``{queue}``)). These are configurables, and can be specified in
 :file:`ipcluster_config`:
 
 .. sourcecode:: python
 
     c.PBSLauncher.queue = 'veryshort.q'
-    c.PBSEngineSetLauncher.n = 64
+    c.IPClusterEnginesApp.n = 64
 
 Note that assuming you are running PBS on a multi-node cluster, the Controller's default behavior
 of listening only on localhost is likely too restrictive.  In this case, also assuming the
 nodes are safely behind a firewall, you can simply instruct the Controller to listen for
 connections on all its interfaces, by adding in :file:`ipcontroller_config`:
 
 .. sourcecode:: python
 
     c.RegistrationFactory.ip = '*'
 
 You can now run the cluster with::
 
     $ ipcluster start profile=pbs n=128
 
 Additional configuration options can be found in the PBS section of :file:`ipcluster_config`.
 
 .. note::
 
     Due to the flexibility of configuration, the PBS launchers work with simple changes
     to the template for other :command:`qsub`-using systems, such as Sun Grid Engine,
     and with further configuration in similar batch systems like Condor.
 
 
 Using :command:`ipcluster` in SSH mode
 ---------------------------------------
 
 
 The SSH mode uses :command:`ssh` to execute :command:`ipengine` on remote
 nodes and :command:`ipcontroller` can be run remotely as well, or on localhost.
 
 .. note::
 
     When using this mode it highly recommended that you have set up SSH keys
     and are using ssh-agent [SSH]_ for password-less logins.
 
 As usual, we start by creating a clean profile::
 
-    $ ipcluster create profile= ssh
+    $ ipcluster create profile=ssh
 
 To use this mode, select the SSH launchers in :file:`ipcluster_config.py`:
 
 .. sourcecode:: python
 
     c.Global.engine_launcher = 'IPython.parallel.apps.launcher.SSHEngineSetLauncher'
     # and if the Controller is also to be remote:
     c.Global.controller_launcher = 'IPython.parallel.apps.launcher.SSHControllerLauncher'
     
 
 The controller's remote location and configuration can be specified:
 
 .. sourcecode:: python
 
     # Set the user and hostname for the controller
     # c.SSHControllerLauncher.hostname = 'controller.example.com'
     # c.SSHControllerLauncher.user = os.environ.get('USER','username')
 
     # Set the arguments to be passed to ipcontroller
     # note that remotely launched ipcontroller will not get the contents of 
     # the local ipcontroller_config.py unless it resides on the *remote host*
-    # in the location specified by the `cluster_dir` argument.
-    # c.SSHControllerLauncher.program_args = ['-r', '-ip', '0.0.0.0', '--cluster_dir', '/path/to/cd']
+    # in the location specified by the `profile_dir` argument.
+    # c.SSHControllerLauncher.program_args = ['--reuse', 'ip=0.0.0.0', 'profile_dir=/path/to/cd']
 
 .. note::
 
     SSH mode does not do any file movement, so you will need to distribute configuration
     files manually.  To aid in this, the `reuse_files` flag defaults to True for ssh-launched
     Controllers, so you will only need to do this once, unless you override this flag back
     to False.
 
 Engines are specified in a dictionary, by hostname and the number of engines to be run
 on that host.
 
 .. sourcecode:: python
 
     c.SSHEngineSetLauncher.engines = { 'host1.example.com' : 2,
                 'host2.example.com' : 5,
-                'host3.example.com' : (1, ['cluster_dir=/home/different/location']),
+                'host3.example.com' : (1, ['profile_dir=/home/different/location']),
                 'host4.example.com' : 8 }
 
 * The `engines` dict, where the keys are the host we want to run engines on and
   the value is the number of engines to run on that host.
 * on host3, the value is a tuple, where the number of engines is first, and the arguments
   to be passed to :command:`ipengine` are the second element.
 
 For engines without explicitly specified arguments, the default arguments are set in
 a single location:
 
 .. sourcecode:: python
 
-    c.SSHEngineSetLauncher.engine_args = ['--cluster_dir', '/path/to/cluster_ssh']
+    c.SSHEngineSetLauncher.engine_args = ['profile_dir=/path/to/cluster_ssh']
 
 Current limitations of the SSH mode of :command:`ipcluster` are:
 
 * Untested on Windows. Would require a working :command:`ssh` on Windows.
   Also, we are using shell scripts to setup and execute commands on remote
   hosts.
 * No file movement - 
 
 Using the :command:`ipcontroller` and :command:`ipengine` commands
 ====================================================================
 
 It is also possible to use the :command:`ipcontroller` and :command:`ipengine`
 commands to start your controller and engines. This approach gives you full
 control over all aspects of the startup process.
 
 Starting the controller and engine on your local machine
 --------------------------------------------------------
 
 To use :command:`ipcontroller` and :command:`ipengine` to start things on your
 local machine, do the following.
 
 First start the controller::
 
     $ ipcontroller
     
 Next, start however many instances of the engine you want using (repeatedly)
 the command::
 
     $ ipengine
 
 The engines should start and automatically connect to the controller using the
 JSON files in :file:`~/.ipython/cluster_default/security`. You are now ready to use the
 controller and engines from IPython.
 
 .. warning:: 
     
     The order of the above operations may be important.  You *must*
     start the controller before the engines, unless you are reusing connection
     information (via `-r`), in which case ordering is not important.
 
 .. note::
 
     On some platforms (OS X), to put the controller and engine into the
     background you may need to give these commands in the form ``(ipcontroller
     &)`` and ``(ipengine &)`` (with the parentheses) for them to work
     properly.
 
 Starting the controller and engines on different hosts
 ------------------------------------------------------
 
 When the controller and engines are running on different hosts, things are
 slightly more complicated, but the underlying ideas are the same:
 
 1. Start the controller on a host using :command:`ipcontroller`.
 2. Copy :file:`ipcontroller-engine.json` from :file:`~/.ipython/cluster_<profile>/security` on
    the controller's host to the host where the engines will run.
 3. Use :command:`ipengine` on the engine's hosts to start the engines.
 
 The only thing you have to be careful of is to tell :command:`ipengine` where
 the :file:`ipcontroller-engine.json` file is located. There are two ways you
 can do this:
 
 * Put :file:`ipcontroller-engine.json` in the :file:`~/.ipython/cluster_<profile>/security`
   directory on the engine's host, where it will be found automatically.
 * Call :command:`ipengine` with the ``--file=full_path_to_the_file``
   flag.
 
 The ``--file`` flag works like this::
 
     $ ipengine --file=/path/to/my/ipcontroller-engine.json
 
 .. note::
     
     If the controller's and engine's hosts all have a shared file system  
     (:file:`~/.ipython/cluster_<profile>/security` is the same on all of them), then things
     will just work!
 
 Make JSON files persistent
 --------------------------
 
 At fist glance it may seem that that managing the JSON files is a bit
 annoying. Going back to the house and key analogy, copying the JSON around
 each time you start the controller is like having to make a new key every time
 you want to unlock the door and enter your house. As with your house, you want
 to be able to create the key (or JSON file) once, and then simply use it at
 any point in the future.
 
 To do this, the only thing you have to do is specify the `--reuse` flag, so that
 the connection information in the JSON files remains accurate::
 
     $ ipcontroller --reuse
 
 Then, just copy the JSON files over the first time and you are set. You can
 start and stop the controller and engines any many times as you want in the
 future, just make sure to tell the controller to reuse the file.
 
 .. note::
 
     You may ask the question: what ports does the controller listen on if you
     don't tell is to use specific ones? The default is to use high random port
     numbers. We do this for two reasons: i) to increase security through
     obscurity and ii) to multiple controllers on a given host to start and
     automatically use different ports.
 
 Log files
 ---------
 
 All of the components of IPython have log files associated with them.
 These log files can be extremely useful in debugging problems with
 IPython and can be found in the directory :file:`~/.ipython/cluster_<profile>/log`.
 Sending the log files to us will often help us to debug any problems.
 
 
 Configuring `ipcontroller`
 ---------------------------
 
 Ports and addresses
 *******************
 
 
 Database Backend
 ****************
 
 
 .. seealso::
 
     
 
 Configuring `ipengine`
 -----------------------
 
 .. note::
 
     TODO
 
 
 
 .. [PBS] Portable Batch System.  http://www.openpbs.org/
 .. [SSH] SSH-Agent http://en.wikipedia.org/wiki/ssh-agent