gaworkflow/runner.py

   1 #!/usr/bin/env python
   2 import logging
   3 import os
   4 import re
   5 import sys
   6 import time
   7 import threading
   8
   9 from benderjab import rpc
  10
  11 from gaworkflow.pipeline.configure_run import *
  12 from gaworkflow.pipeline.monitors import startCmdLineStatusMonitor
  13
  14 #s_fc = re.compile('FC[0-9]+')
  15 s_fc = re.compile('_[0-9a-zA-Z]*$')
  16
  17
  18 def _get_flowcell_from_rundir(run_dir):
  19     """
  20     Returns flowcell string based on run_dir.
  21     Returns None and logs error if flowcell can't be found.
  22     """
  23     junk, dirname = os.path.split(run_dir)
  24     mo = s_fc.search(dirname)
  25     if not mo:
  26         logging.error('RunDir 2 FlowCell error: %s' % (run_dir))
  27         return None
  28
  29     return dirname[mo.start()+1:]
  30
  31
  32
  33 class Runner(rpc.XmlRpcBot):
  34     """
  35     Manage running pipeline jobs.
  36     """
  37     def __init__(self, section=None, configfile=None):
  38         #if configfile is None:
  39         #    self.configfile = "~/.gaworkflow"
  40         super(Runner, self).__init__(section, configfile)
  41
  42         self.cfg['notify_users'] = None
  43         self.cfg['genome_dir'] = None
  44         self.cfg['base_analysis_dir'] = None
  45
  46         self.cfg['notify_users'] = None
  47         self.cfg['notify_postanalysis'] = None
  48
  49         self.conf_info_dict = {}
  50
  51         self.register_function(self.sequencingFinished)
  52         #self.eventTasks.append(self.update)
  53
  54
  55     def read_config(self, section=None, configfile=None):
  56         super(Runner, self).read_config(section, configfile)
  57
  58         self.genome_dir = self._check_required_option('genome_dir')
  59         self.base_analysis_dir = self._check_required_option('base_analysis_dir')
  60
  61         self.notify_users = self._parse_user_list(self.cfg['notify_users'])
  62         #FIXME: process notify_postpipeline cfg
  63
  64
  65     def _parser(self, msg, who):
  66         """
  67         Parse xmpp chat messages
  68         """
  69         help = u"I can send [start] a run, or report [status]"
  70         if re.match(u"help", msg):
  71             reply = help
  72         elif re.match("status", msg):
  73             reply = u"not implemented"
  74         elif re.match(u"start", msg):
  75             words = msg.split()
  76             if len(words) == 2:
  77                 self.sequencingFinished(words[1])
  78                 reply = u"starting run for %s" % (words[1])
  79             else:
  80                 reply = u"need runfolder name"
  81         else:
  82             reply = u"I didn't understand '%s'" %(msg)
  83
  84         logging.debug("reply: " + str(reply))
  85         return reply
  86
  87
  88     def start(self, daemonize):
  89         """
  90         Start application
  91         """
  92         super(Runner, self).start(daemonize)
  93
  94
  95     def stop(self):
  96         """
  97         shutdown application
  98         """
  99         super(Runner, self).stop()
 100
 101
 102     def sequencingFinished(self, run_dir):
 103         """
 104         Sequenceing (and copying) is finished, time to start pipeline
 105         """
 106         logging.debug("received sequencing finished message")
 107
 108         # Setup config info object
 109         ci = ConfigInfo()
 110         ci.base_analysis_dir = self.base_analysis_dir
 111         ci.analysis_dir = os.path.join(self.base_analysis_dir, run_dir)
 112
 113         # get flowcell from run_dir name
 114         flowcell = _get_flowcell_from_rundir(run_dir)
 115
 116         # Store ci object in dictionary
 117         self.conf_info_dict[flowcell] = ci
 118
 119
 120         # Launch the job in it's own thread and turn.
 121         self.launchJob(run_dir, flowcell, ci)
 122
 123
 124     def pipelineFinished(self, run_dir):
 125         # need to strip off self.watch_dir from rundir I suspect.
 126         logging.info("pipeline finished in" + str(run_dir))
 127         #pattern = self.watch_dir
 128         #if pattern[-1] != os.path.sep:
 129         #    pattern += os.path.sep
 130         #stripped_run_dir = re.sub(pattern, "", run_dir)
 131         #logging.debug("stripped to " + stripped_run_dir)
 132
 133         # Notify each user that the run has finished.
 134         if self.notify_users is not None:
 135             for u in self.notify_users:
 136                 self.send(u, 'Pipeline run %s finished' % (run_dir))
 137
 138         #if self.notify_runner is not None:
 139         #    for r in self.notify_runner:
 140         #        self.rpc_send(r, (stripped_run_dir,), 'sequencingFinished')
 141
 142     def reportMsg(self, msg):
 143
 144         if self.notify_users is not None:
 145             for u in self.notify_users:
 146                 self.send(u, msg)
 147
 148
 149     def _runner(self, run_dir, flowcell, conf_info):
 150
 151         # retrieve config step
 152         cfg_filepath = os.path.join(conf_info.analysis_dir,
 153                                     'config32auto.txt')
 154         status_retrieve_cfg = retrieve_config(conf_info,
 155                                           flowcell,
 156                                           cfg_filepath,
 157                                           self.genome_dir)
 158         if status_retrieve_cfg:
 159             logging.info("Runner: Retrieve config: success")
 160             self.reportMsg("Retrieve config (%s): success" % (run_dir))
 161         else:
 162             logging.error("Runner: Retrieve config: failed")
 163             self.reportMsg("Retrieve config (%s): FAILED" % (run_dir))
 164
 165
 166         # configure step
 167         if status_retrieve_cfg:
 168             status = configure(conf_info)
 169             if status:
 170                 logging.info("Runner: Configure: success")
 171                 self.reportMsg("Configure (%s): success" % (run_dir))
 172             else:
 173                 logging.error("Runner: Configure: failed")
 174                 self.reportMsg("Configure (%s): FAILED" % (run_dir))
 175
 176             #if successful, continue
 177             if status:
 178                 # Setup status cmdline status monitor
 179                 #startCmdLineStatusMonitor(ci)
 180
 181                 # running step
 182                 print 'Running pipeline now!'
 183                 run_status = run_pipeline(conf_info)
 184                 if run_status is True:
 185                     logging.info('Runner: Pipeline: success')
 186                     self.piplineFinished(run_dir)
 187                 else:
 188                     logging.info('Runner: Pipeline: failed')
 189                     self.reportMsg("Pipeline run (%s): FAILED" % (run_dir))
 190
 191
 192     def launchJob(self, run_dir, flowcell, conf_info):
 193         """
 194         Starts up a thread for running the pipeline
 195         """
 196         t = threading.Thread(target=self._runner,
 197                         args=[run_dir, flowcell, conf_info])
 198         t.setDaemon(True)
 199         t.start()
 200
 201
 202
 203 def main(args=None):
 204     bot = Runner('demobot')
 205     bot.cfg['loglevel'] = 'DEBUG'
 206     return bot.main(args)
 207
 208 if __name__ == "__main__":
 209     sys.exit(main(sys.argv[1:]))
 210