initexit.py 26.4 KB
Newer Older
1
2
3
4
#!/usr/bin/env python
# da_initexit.py

"""
Peters, Wouter's avatar
Peters, Wouter committed
5
6
.. module:: initexit
.. moduleauthor:: Wouter Peters 
7
8
9
10

Revision History:
File created on 13 May 2009.

Peters, Wouter's avatar
Peters, Wouter committed
11
The CycleControl class is found in the module :mod:`initexit`. It is derived from the standard python :class:`dictionary` object. It is the only core object of CTDAS that is automatically created in the pipeline, the user (normally) does not need to modify or extend it. The class is created based on options and arguments passes on the command line when submitting your main CTDAS job. 
12
13
14

Valid options are defined in 

15
.. autofunction:: da.tools.initexit.parse_options
16
17
18
19
20
21

With the name of a valid ``rc-file``, the CycleControl object is instantiated and validated. An example rc-file looks
like this:::

    ! Info on the data assimilation cycle

22
23
24
25
26
27
    time.restart        : False                     ! Restart from an existing run T/F
    time.start          : 2000-01-01 00:00:00       ! Start time of first cycle
    time.finish         : 2000-01-08 00:00:00       ! End time of last cycle
    time.cycle          : 7                         ! length of each cycle, 7 means one week
    time.nlag           : 5                         ! number of cycles in one smoother window
    dir.da_run          : ${HOME}/tmp/test_da       ! the run directory for you project
28
29
30

    ! Info on the DA system used

31
32
    da.system           : CarbonTracker             ! an identifier for your inversion system
    da.system.rc        : da/rc/carbontracker.rc    ! the settings needed in your inversion system
33
34
35

    ! Info on the forward model to be used

36
37
38
    da.obsoperator         : TM5                                ! an identifier for your observation operator
    da.obsoperator.rc      : ${HOME}/Modeling/TM5/tm5-ctdas.rc  ! the rc-file needed to run youobservation operator
    da.optimizer.nmembers  : 30                                 ! the number of ensemble members desired in the optimization
39
40
41
42

The most important method of the CycleControl object are listed below:

.. autoclass:: da.tools.initexit.CycleControl 
43
   :members: setup, finalize,  collect_restart_data, move_restart_data, 
44
             submit_next_cycle, setup_file_structure, recover_run, random_seed
45
46
47
48
49
50
51

Two important attributes of the CycleControl object are:
    (1) DaSystem, an instance of a :ref:`dasystem`
    (2) DaPlatForm, an instance of a :ref:`platform`

Other functions in the module initexit that are related to the control of a DA cycle are:

52
53
.. autofunction:: da.tools.initexit.start_logger 
.. autofunction:: da.tools.initexit.validate_opts_args 
54
55


56
"""
karolina's avatar
karolina committed
57
58
59
60
import logging
import os
import sys
import shutil
karolina's avatar
karolina committed
61
import copy
62
63
64
import getopt
import cPickle
import numpy as np
karolina's avatar
karolina committed
65
from string import join
66

karolina's avatar
karolina committed
67
import da.tools.rc as rc
68
from da.tools.general import create_dirs, to_datetime, advance_time
69

karolina's avatar
karolina committed
70
needed_da_items = [
71
72
73
74
75
    'time.start',
    'time.finish',
    'time.nlag',
    'time.cycle',
    'dir.da_run',
76
77
78
    'da.resources.ncycles_per_job',
    'da.resources.ntasks',
    'da.resources.ntime',
79
80
81
82
83
    'da.system',
    'da.system.rc',
    'da.obsoperator',
    'da.obsoperator.rc',
    'da.optimizer.nmembers']
84
85
86
87
88
89
90

# only needed in an earlier implemented where each substep was a separate job
# validprocesses = ['start','done','samplestate','advance','invert']


class CycleControl(dict):
    """
91
    This object controls the CTDAS system flow and functionality.
92
    """
93
        
karolina's avatar
karolina committed
94
    def __init__(self, opts=[], args={}):
95
        """
Peters, Wouter's avatar
Peters, Wouter committed
96
97
        The CycleControl object is instantiated with a set of options and arguments.
        The list of arguments must contain the name of an existing ``rc-file``. 
98
99
        This rc-file is loaded by method :meth:`~da.tools.initexit.CycleControl.load_rc` and validated
        by :meth:`~da.tools.initexit.CycleControl.validate_rc`
100

Peters, Wouter's avatar
Peters, Wouter committed
101
        Options for the CycleControl consist of accepted command line flags or arguments 
102
        in :func:`~da.tools.initexit.CycleControl.parse_options`
103

104
        """
karolina's avatar
karolina committed
105
106
        rcfile = args['rc']
        self.load_rc(rcfile)
107
        self.validate_rc()
108
109
110
111
        self.opts = opts

        # Add some useful variables to the rc-file dictionary

karolina's avatar
karolina committed
112
        self['jobrcfilename'] = rcfile
karolina's avatar
karolina committed
113
        self['dir.da_submit'] = os.getcwd()
114
        self['da.crash.recover'] = '-r' in opts
115
        self['transition'] = '-t' in opts
karolina's avatar
karolina committed
116
        self['verbose'] = '-v' in opts
117
118
119
        self.dasystem = None # to be filled later
        self.restart_filelist = [] # List of files needed for restart, to be extended later
        self.output_filelist = [] # List of files needed for output, to be extended later
120
121


karolina's avatar
karolina committed
122
    def load_rc(self, rcfilename):
123
124
125
126
        """ 
        This method loads a DA Cycle rc-file with settings for this simulation 
        """

karolina's avatar
karolina committed
127
        rcdata = rc.read(rcfilename)
karolina's avatar
karolina committed
128
        for k, v in rcdata.iteritems():
129
130
            self[k] = v

karolina's avatar
karolina committed
131
132
        logging.info('DA Cycle rc-file (%s) loaded successfully' % rcfilename)
        
133

134
    def validate_rc(self):
135
        """ 
Peters, Wouter's avatar
Peters, Wouter committed
136
137
        Validate the contents of the rc-file given a dictionary of required keys. 
        Currently required keys are :attr:`~da.tools.initexit.needed_da_items`
138
139
        """

karolina's avatar
karolina committed
140
141
142
143
144
        for k, v in self.iteritems():
            if v in ['True', 'true', 't', 'T', 'y', 'yes']:
                self[k] = True
            if v in ['False', 'false', 'f', 'F', 'n', 'no']:
                self[k] = False
145
            if 'date' in k : 
146
                self[k] = to_datetime(v)
147
            if k in ['time.start', 'time.end', 'time.finish', 'da.restart.tstamp']:
148
                self[k] = to_datetime(v)
149
150
        for key in needed_da_items:
            if not self.has_key(key):
karolina's avatar
karolina committed
151
152
153
154
155
156
157
158
                msg = 'Missing a required value in rc-file : %s' % key
                logging.error(msg)
                logging.error('!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ')
                logging.error('Please note the update on Dec 02 2011 where rc-file names for DaSystem and ')
                logging.error('are from now on specified in the main rc-file (see da/rc/da.rc for example)')
                logging.error('!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ')
                raise IOError, msg
        logging.debug('DA Cycle settings have been validated succesfully')
159

160
    def parse_times(self):
161
162
163
164
165
166
167
168
        """ 
        Parse time related parameters into datetime objects for later use 
        """

        startdate = self['time.start']
        finaldate = self['time.finish']                  

        if finaldate <= startdate:
karolina's avatar
karolina committed
169
            logging.error('The start date (%s) is not greater than the end date (%s), please revise' % (startdate.strftime('%Y%m%d'), finaldate.strftime('%Y%m%d'))) 
170
171
172
173
174
175
176
177
            raise ValueError
        cyclelength = self['time.cycle']                 # get time step

# Determine end date

        if cyclelength == 'infinite':
            enddate = finaldate
        else:
178
            enddate = advance_time(startdate, cyclelength)
179

karolina's avatar
karolina committed
180
        dt = enddate - startdate
181

karolina's avatar
karolina committed
182
        if enddate > finaldate:  # do not run beyond finaldate
183
184
            enddate = finaldate

karolina's avatar
karolina committed
185
186
187
188
        self['time.start'] = startdate
        self['time.end'] = enddate
        self['time.finish'] = finaldate
        self['cyclelength'] = dt
189

karolina's avatar
karolina committed
190
191
        logging.info("===============================================================")
        logging.info("DA Cycle start date is %s" % startdate.strftime('%Y-%m-%d %H:%M'))
192
        logging.info("DA Cycle end date is %s" % enddate.strftime('%Y-%m-%d %H:%M'))
karolina's avatar
karolina committed
193
194
195
196
        logging.info("DA Cycle final date is %s" % finaldate.strftime('%Y-%m-%d %H:%M'))  
        logging.info("DA Cycle cycle length is %s" % cyclelength)
        logging.info("DA Cycle restart is %s" % str(self['time.restart']))
        logging.info("===============================================================")
197
198


199
    def set_sample_times(self, lag):
200
201
202
203
204
205
        """
        Set the times over which a sampling interval will loop, depending on 
        the lag. Note that lag falls in the interval [0,nlag-1]
        """

        # Start from cycle times 
karolina's avatar
karolina committed
206
207
        self['time.sample.start'] = copy.deepcopy(self['time.start'])
        self['time.sample.end'] = copy.deepcopy(self['time.end'])
208
209
210
211

        # Now advance depending on lag

        for l in range(lag):
212
            self.advance_sample_times()
213
214


215
    def advance_sample_times(self):
216
217
218
219
        """ 
        Advance sampling start and end time by one cycle interval
        """

220
        days = self['cyclelength'].days                
221

222
223
        self['time.sample.start'] = advance_time(self['time.sample.start'], days)
        self['time.sample.end'] = advance_time(self['time.sample.end'], days)
224
225
    

226
    def advance_cycle_times(self):
227
228
229
        """ 
        Advance cycle start and end time by one cycle interval
        """
230
231
              
        days = self['cyclelength'].days                  
Peters, Wouter's avatar
Peters, Wouter committed
232

233
234
        startdate = advance_time(self['time.start'], days)
        enddate = advance_time(self['time.end'], days)
235

karolina's avatar
karolina committed
236
237
        filtertime = startdate.strftime('%Y%m%d')
        self['dir.output'] = os.path.join(self['dir.da_run'], 'output', filtertime)
238

karolina's avatar
karolina committed
239
240
        self['time.start'] = startdate
        self['time.end'] = enddate
241
242


243
244
245
246
247
248
    def write_random_seed(self):
        filename = os.path.join(self['dir.restart'], 'randomseed_%s.pickle' % self['time.start'].strftime('%Y%m%d'))
        f = open(filename, 'wb')
        seed = np.random.get_state()
        cPickle.dump(seed, f, -1)
        f.close()
249

250
        logging.info("Saved the random seed generator values to file")
251
252


253
254
255
256
257
258
259
    def read_random_seed(self, first=False):
        if first:
            filename = self.dasystem['random.seed.init']
            logging.info("Initialised random seed from: %s"%filename)
        else: 
            filename = os.path.join(self['dir.restart'], 'randomseed_%s.pickle' % self['da.restart.tstamp'].strftime('%Y%m%d'))
            logging.info("Retrieved the random seed generator values of last cycle from file")
260
261
262
263
        f = open(filename, 'rb')
        seed = cPickle.load(f)
        np.random.set_state(seed)
        f.close()
264

265

266
    def setup(self):
267
268
269
        """ 
        This method determines how to proceed with the cycle. Three options are implemented:

Peters, Wouter's avatar
Peters, Wouter committed
270
271
272
            1. *Fresh start*  : set up the required file structure for this simulation and start
            2. *Restart*      : use latest da_runtime variables from the exec dir and restart
            3. *Recover*      : restart after crash by getting data from restart/one-ago folder
273
274
275

        The choice that gets executed depends on the presence of 

Peters, Wouter's avatar
Peters, Wouter committed
276
277
            # the ``-r`` option on the command line, this triggers a recover
            # the ``time.restart : True`` option in the da.rc file
278
279

        The latter is automatically set if the filter submits the next cycle at the end of the current one, 
280
        through method :meth:`~da.tools.initexit.CycleControl.submit_next_cycle`.
Peters, Wouter's avatar
Peters, Wouter committed
281

Peters, Wouter's avatar
Peters, Wouter committed
282
        The specific call tree under each scenario is: 
Peters, Wouter's avatar
Peters, Wouter committed
283

Peters, Wouter's avatar
Peters, Wouter committed
284
            1. *Fresh Start*
285
                *  :meth:`~da.tools.initexit.CycleControl.setup_file_structure()`  <- Create directory tree
Peters, Wouter's avatar
Peters, Wouter committed
286
            2. *Restart*
287
288
                *  :meth:`~da.tools.initexit.CycleControl.setup_file_structure()`
                *  :meth:`~da.tools.initexit.CycleControl.random_seed`    <- Read the random seed from file
Peters, Wouter's avatar
Peters, Wouter committed
289
            3. *Recover*
290
291
292
                *  :meth:`~da.tools.initexit.CycleControl.setup_file_structure()`
                *  :meth:`~da.tools.initexit.CycleControl.recover_run()`          <- Recover files from restart/one-ago dir, reset ``time.start``
                *  :meth:`~da.tools.initexit.CycleControl.random_seed` 
Peters, Wouter's avatar
Peters, Wouter committed
293
294
295

        And is always followed by a call to

296
            * parse_times()
Peters, Wouter's avatar
Peters, Wouter committed
297
            * WriteRc('jobfilename')
298
        """        
299
300
301
302
303
304
305
306
        if self['transition']:
            logging.info("Transition of filter from previous step with od meteo from 25 to 34 levels")
            self.setup_file_structure()
            strippedname = os.path.split(self['jobrcfilename'])[-1]
            self['jobrcfilename'] = os.path.join(self['dir.exec'], strippedname)
            self.read_random_seed(False)

        elif self['time.restart']:
karolina's avatar
karolina committed
307
            logging.info("Restarting filter from previous step")
308
            self.setup_file_structure()
309
            strippedname = os.path.split(self['jobrcfilename'])[-1]
karolina's avatar
karolina committed
310
            self['jobrcfilename'] = os.path.join(self['dir.exec'], strippedname)
311
312
            self.read_random_seed(False)

313
        else: #assume that it is a fresh start, change this condition to more specific if crash recover added
karolina's avatar
karolina committed
314
            logging.info("First time step in filter sequence")
315
            self.setup_file_structure()
316
317

            # expand jobrcfilename to include exec dir from now on.
318
319
320
            # First strip current leading path from filename

            strippedname = os.path.split(self['jobrcfilename'])[-1]
karolina's avatar
karolina committed
321
            self['jobrcfilename'] = os.path.join(self['dir.exec'], strippedname)
322
            if self.has_key('random.seed.init'):
323
                self.read_random_seed(True)
324

325
        self.parse_times()
326
        self.write_rc(self['jobrcfilename'])
327

328
    def setup_file_structure(self):
329
330
        """ 
        Create file structure needed for data assimilation system.
Peters, Wouter's avatar
Peters, Wouter committed
331
332
333
334
335
336
337
338
        In principle this looks like:

            * ``${da_rundir}``
            * ``${da_rundir}/input``
            * ``${da_rundir}/output``
            * ``${da_rundir}/exec``
            * ``${da_rundir}/analysis``
            * ``${da_rundir}/jobs``
339
340
            * ``${da_rundir}/restart/current``
            * ``${da_rundir}/restart/one-ago``
Peters, Wouter's avatar
Peters, Wouter committed
341

342
        .. note:: The exec dir will actually be a simlink to the directory where
343
344
345
                 the observation operator executable lives. This directory is passed through
                 the ``da.rc`` file. 

346
        .. note:: The observation input files will be placed in the exec dir,
347
                 and the resulting simulated values will be retrieved from there as well.
Peters, Wouter's avatar
Peters, Wouter committed
348

349
350
351
352
        """

# Create the run directory for this DA job, including I/O structure

karolina's avatar
karolina committed
353
        filtertime = self['time.start'].strftime('%Y%m%d')
354

karolina's avatar
karolina committed
355
356
357
358
359
360
        self['dir.exec'] = os.path.join(self['dir.da_run'], 'exec')
        self['dir.input'] = os.path.join(self['dir.da_run'], 'input')
        self['dir.output'] = os.path.join(self['dir.da_run'], 'output', filtertime)
        self['dir.analysis'] = os.path.join(self['dir.da_run'], 'analysis')
        self['dir.jobs'] = os.path.join(self['dir.da_run'], 'jobs')
        self['dir.restart'] = os.path.join(self['dir.da_run'], 'restart')
361

362
363
364
365
366
367
368
        create_dirs(self['dir.da_run'])
        create_dirs(os.path.join(self['dir.exec']))
        create_dirs(os.path.join(self['dir.input']))
        create_dirs(os.path.join(self['dir.output']))
        create_dirs(os.path.join(self['dir.analysis']))
        create_dirs(os.path.join(self['dir.jobs']))
        create_dirs(os.path.join(self['dir.restart']))
369

karolina's avatar
karolina committed
370
        logging.info('Succesfully created the file structure for the assimilation job')
371

372

373
    def finalize(self):
374
        """
375
        finalize the da cycle, this means writing the save data and rc-files for the next run. 
376
377
378
        The following sequence of actions occur:

            * Write the randomseed to file for reuse in next cycle
Peters, Wouter's avatar
Peters, Wouter committed
379
            * Write a new ``rc-file`` with ``time.restart : True``, and new ``time.start`` and ``time.end``
380
381
382
383
            * Collect all needed data needed for check-pointing (restart from current system state)
            * Move the previous check pointing data out of the way, and replace with current
            * Submit the next cycle

384
        """
385
386
        self.write_random_seed()                              
        self.write_new_rc_file()                              
387
        
388
389
390
        self.collect_restart_data()  # Collect restart data for next cycle into a clean restart/current folder
        self.collect_output()  # Collect restart data for next cycle into a clean restart/current folder
        self.submit_next_cycle()
391

392
    def collect_output(self):
393
        """ Collect files that are part of the requested output for this cycle. This function allows users to add files 
394
395
            to a list, and then the system will copy these to the current cycle's output directory.
            The list of files included is read from the 
396
            attribute "output_filelist" which is a simple list of files that can be appended by other objects/methods that
397
398
399
400
            require output data to be saved.


        """
karolina's avatar
karolina committed
401
        targetdir = os.path.join(self['dir.output'])
402
        create_dirs(targetdir)
403

karolina's avatar
karolina committed
404
405
        logging.info("Collecting the required output data") 
        logging.debug("           to   directory: %s " % targetdir)
406

407
        for file in set(self.output_filelist):
408
409
            if os.path.isdir(file): # skip dirs
                continue
410
            if not os.path.exists(file): # skip dirs
karolina's avatar
karolina committed
411
                logging.warning("           [not found] .... %s " % file)
412
                continue
413

karolina's avatar
karolina committed
414
415
            logging.debug("           [copy] .... %s " % file)
            shutil.copy(file, file.replace(os.path.split(file)[0], targetdir))
416
417
418



419
    def collect_restart_data(self):
420
        """ Collect files needed for the restart of this cycle in case of a crash, or for the continuation of the next cycle. 
421
            All files needed are written to the restart/current directory. The list of files included is read from the 
422
            attribute "restart_filelist" which is a simple list of files that can be appended by other objects/methods that
423
424
            require restart data to be saved.

425
            .. note:: Before collecting the files in the ``restart_filelist``, the restart/current directory will be emptied and
426
                     recreated. This prevents files from accumulating in the restart/current and restart/one-ago folders. It 
427
                     also means that if a file is missing from the ``restart_filelist``, it will not be available for check-pointing
428
429
430
                     if your run crashes or dies!

            Currently, the following files are included:
431

432
                * The ``da_runtime.rc`` file
433
                * The ``randomseed.pickle`` file
434
                * The savestate.nc file
435
                * The files in the ``ObservationOperator.restart_filelist``, i.e., restart data for the transport model
436

437

Peters, Wouter's avatar
Peters, Wouter committed
438
            .. note:: We assume that the restart files for the :ref:`ObservationOperator` 
439
                      reside in a separate folder, i.e, the ObservationOperator does *not* write directly to the CTDAS restart dir!
440
441
442

        """

443
        targetdir = os.path.join(self['dir.restart'])
444

445
        #logging.info("Purging the current restart directory before collecting new data")
446

447
        #create_dirs(targetdir, forceclean=True)
448

karolina's avatar
karolina committed
449
450
        logging.info("Collecting the required restart data")
        logging.debug("           to   directory: %s " % targetdir)
451

452
        for file in set(self.restart_filelist):
453
454
            if os.path.isdir(file): # skip dirs
                continue
455
            if not os.path.exists(file): 
karolina's avatar
karolina committed
456
                logging.warning("           [not found] .... %s " % file)
457
            else:
karolina's avatar
karolina committed
458
459
                logging.debug("           [copy] .... %s " % file)
                shutil.copy(file, file.replace(os.path.split(file)[0], targetdir))
460
461
462
463



#
464
    def write_new_rc_file(self):
465
466
467
468
469
470
        """ Write the rc-file for the next DA cycle. 

            .. note:: The start time for the next cycle is the end time of this one, while 
                      the end time for the next cycle is the current end time + one cycle length. 
                      
            The resulting rc-file is written to the ``dir.exec`` so that it can be used when resubmitting the next cycle
471
472
            
        """
karolina's avatar
karolina committed
473
        
474
        # We make a copy of the current dacycle object, and modify the start + end dates and restart value
475

karolina's avatar
karolina committed
476
477
478
479
        new_dacycle = copy.deepcopy(self)
        new_dacycle['da.restart.tstamp'] = self['time.start']
        new_dacycle.advance_cycle_times()
        new_dacycle['time.restart'] = True
480
        
481
        # Create the name of the rc-file that will hold this new input, and write it
482

483
484
        #fname = os.path.join(self['dir.exec'], 'da_runtime.rc')  # current exec dir holds next rc file
        
karolina's avatar
karolina committed
485
        fname = os.path.join(self['dir.restart'], 'da_runtime_%s.rc' % new_dacycle['time.start'].strftime('%Y%m%d'))#advanced time
486
        
karolina's avatar
karolina committed
487
        rc.write(fname, new_dacycle)
488
        logging.debug('Wrote new da_runtime.rc (%s) to restart dir' % fname)
489

490
        # The rest is info needed for a system restart, so it modifies the current dacycle object (self)
491

karolina's avatar
karolina committed
492
        self['da.restart.fname'] = fname    # needed for next job template
493
494
        #self.restart_filelist.append(fname)  # not that needed since it is already written to the restart dir...
        #logging.debug('Added da_runtime.rc to the restart_filelist for later collection')
495
496


497
    def write_rc(self, fname):
498
499
        """ Write RC file after each process to reflect updated info """

karolina's avatar
karolina committed
500
        rc.write(fname, self)
karolina's avatar
karolina committed
501
        logging.debug('Wrote expanded rc-file (%s)' % fname)
karolina's avatar
karolina committed
502
        
503

504
    def submit_next_cycle(self):
505
        """ 
Peters, Wouter's avatar
Peters, Wouter committed
506
        Submit the next job of a DA cycle, this consists of 
507
508
509
            * Changing to the working directory from which the job was started initially
            * create a line to start the master script again with a newly created rc-file
            * Submitting the jobfile 
Peters, Wouter's avatar
Peters, Wouter committed
510
511
512

        If the end of the cycle series is reached, no new job is submitted.

513
        """
karolina's avatar
karolina committed
514
        
515

516
        if self['time.end'] < self['time.finish']:
517

Peters, Wouter's avatar
test    
Peters, Wouter committed
518
            # file ID and names
karolina's avatar
karolina committed
519
520
521
            jobid = self['time.end'].strftime('%Y%m%d') 
            targetdir = os.path.join(self['dir.exec'])
            jobfile = os.path.join(targetdir, 'jb.%s.jb' % jobid)
522
            logfile = os.path.join(targetdir, 'jb.%s.log' % jobid)
Peters, Wouter's avatar
test    
Peters, Wouter committed
523
            # Template and commands for job
524
            jobparams = {'jobname':"j.%s" % jobid, 'jobnodes':self['da.resources.ntasks'], 'jobtime': self['da.resources.ntime'], 'logfile': logfile, 'errfile': logfile}
525
            template = self.daplatform.get_job_template(jobparams)
karolina's avatar
karolina committed
526
            execcommand = os.path.join(self['dir.da_submit'], sys.argv[0]) 
527
528
            if '-t' in self.opts:
                (self.opts).remove('-t') 
529

530
531
532
533
534
            if not os.environ.has_key('icycle_in_job'):
                logging.info('Environment variable icycle_in_job not found, resubmitting after this cycle')
                os.environ['icycle_in_job'] = self['da.resources.ncycles_per_job']  # assume that if no cycle number is set, we should submit the next job by default
            else:
                logging.info('Environment variable icycle_in_job was found, processing cycle %s of %s in this job'%(os.environ['icycle_in_job'],self['da.resources.ncycles_per_job']) )
535

536
            ncycles = int(self['da.resources.ncycles_per_job'])
537
538
539
540
            for cycle in range(ncycles): 
                nextjobid = '%s'% ( (self['time.end']+cycle*self['cyclelength']).strftime('%Y%m%d'),)
                nextrestartfilename = self['da.restart.fname'].replace(jobid,nextjobid)
                nextlogfilename = logfile.replace(jobid,nextjobid)
541
            	template += '\nexport icycle_in_job=%d\npython %s rc=%s %s >&%s\n' % (cycle+1,execcommand, nextrestartfilename, join(self.opts, ''), nextlogfilename,) 
Peters, Wouter's avatar
test    
Peters, Wouter committed
542
543

            # write and submit 
544
            self.daplatform.write_job(jobfile, template, jobid)
545
	    if 'da.resources.ncycles_per_job' in self:
546
		do_submit = (int(os.environ['icycle_in_job']) >= int(self['da.resources.ncycles_per_job']))
547
	    else:
548
                dosubmit = False
549
550
551
552
          
            if do_submit:
                jobid = self.daplatform.submit_job(jobfile, joblog=logfile)

553
554
555
556
        else:
            logging.info('Final date reached, no new cycle started')


557
def start_logger(level=logging.INFO):
558
559
560
561
    """ start the logging of messages to screen"""

# start the logging basic configuration by setting up a log file

karolina's avatar
karolina committed
562
563
564
    logging.basicConfig(level=level,
                        format=' [%(levelname)-7s] (%(asctime)s) py-%(module)-20s : %(message)s',
                        datefmt='%Y-%m-%d %H:%M:%S')
565

566
def parse_options():
567
568
569
570
571
572
573
574
575
576
    """ 
    Function parses options from the command line and returns the arguments as a dictionary.
    Accepted command line arguments are:

    ========  =======
    Argument  Meaning
    ========  =======
    -v        verbose output in log files
    -h        display help
    -r        start a simulation by recovering from a previous crash
577
    -t        start a simulation by transitioning from 25 to 34 layers in December 2005 (od meteo)
578
579
580
    ========  =======

    """
581
582
583

# Parse keywords, the only option accepted so far is the "-h" flag for help

karolina's avatar
karolina committed
584
585
    opts = []
    args = []
586
    try:                                
587
        opts, args = getopt.gnu_getopt(sys.argv[1:], "-rvt")
588
    except getopt.GetoptError, msg:           
karolina's avatar
karolina committed
589
        logging.error('%s' % msg)
590
591
592
        sys.exit(2)      

    for options in opts:
karolina's avatar
karolina committed
593
        options = options[0].lower()
594
595
        if options == '-r':
            logging.info('-r flag specified on command line: recovering from crash')
596
597
        if options == '-t':
            logging.info('-t flag specified on command line: transition with od from December 2005')    
598
599
        if options == '-v':
            logging.info('-v flag specified on command line: extra verbose output')
karolina's avatar
karolina committed
600
            logging.root.setLevel(logging.DEBUG)
601

602
    if opts: 
603
        optslist = [item[0] for item in opts]
604
    else:
karolina's avatar
karolina committed
605
        optslist = []
606

607
608
# Parse arguments and return as dictionary

karolina's avatar
karolina committed
609
    arguments = {}
610
611
612
613
614
615
616
617
    for item in args:
        #item=item.lower()

# Catch arguments that are passed not in "key=value" format

        if '=' in item:
            key, arg = item.split('=')
        else:
karolina's avatar
karolina committed
618
619
            logging.error('%s' % 'Argument passed without description (%s)' % item)
            raise getopt.GetoptError, arg
620

karolina's avatar
karolina committed
621
        arguments[key] = arg
622
623


624
    return optslist, arguments
625

626
def validate_opts_args(opts, args):
627
628
629
630
631
    """ 
 Validate the options and arguments passed from the command line before starting the cycle. The validation consists of checking for the presence of an argument "rc", and the existence of
 the specified rc-file.  
 
    """
632
    if not args.has_key("rc"):
karolina's avatar
karolina committed
633
634
635
        msg = "There is no rc-file specified on the command line. Please use rc=yourfile.rc"
        logging.error(msg)
        raise IOError, msg
636
    elif not os.path.exists(args['rc']):
karolina's avatar
karolina committed
637
638
639
        msg = "The specified rc-file (%s) does not exist " % args['rc'] 
        logging.error(msg)
        raise IOError, msg
640
641
642
643
644
645
646
647
648

    # WP not needed anymore
    #if not args.has_key('process'):
    #    msg = "There is no process specified on the command line, assuming process=Start"   ; logging.info(msg)
    #    args['process'] = 'start'
    #if args['process'].lower() not in validprocesses:
    #    msg = "The specified process (%s) is not valid"%args['process']   ; logging.error(msg)
    #    raise IOError,msg

karolina's avatar
karolina committed
649
    return opts, args
650
651
652


if __name__ == "__main__":
653
    pass
654