9 from htsworkflow.util import mount
13 from pyinotify import EventsCodes
15 from benderjab import rpc
17 def is_runfolder(name):
21 >>> print is_runfolder('090630_HWUSI-EAS999_0006_30LNFAAXX')
23 >>> print is_runfolder('hello')
26 if re.match("[0-9]{6}_.*", name):
31 def get_top_dir(root, path):
33 Return the directory in path that is a subdirectory of root.
36 >>> print get_top_dir('/a/b/c', '/a/b/c/d/e/f')
38 >>> print get_top_dir('/a/b/c/', '/a/b/c/d/e/f')
40 >>> print get_top_dir('/a/b/c', '/g/e/f')
42 >>> print get_top_dir('/a/b/c', '/a/b/c')
45 if path.startswith(root):
46 subpath = path[len(root):]
47 if subpath.startswith('/'):
49 return subpath.split(os.path.sep)[0]
53 class WatcherEvent(object):
55 Track information about a file event
57 Currently its time, and if it was an indication we've completed the run.
59 def __init__(self, event_root=None):
60 self.time = time.time()
61 self.event_root = event_root
64 def __unicode__(self):
66 complete = "(completed)"
69 return u"<WatchEvent: %s %s %s>" % (time.ctime(self.time), self.event_root, complete)
71 class Handler(pyinotify.ProcessEvent):
72 def __init__(self, watchmanager, bot, completion_file=None):
74 Completion file contains current "completion" filename
77 self.watchmanager = watchmanager
79 if completion_file is not None:
80 completion_file = completion_file.lower()
81 self.completion_file = completion_file
83 def process_IN_CREATE(self, event):
84 for wdd in self.bot.wdds:
85 for watch_path in self.bot.watchdirs:
86 run_already_complete = False
87 # I only care about things created inside the watch directory, so
88 # the event path needs to be longer than the watch path in addition to
89 # starting with the watch_path
90 if len(event.path) > len(watch_path) and event.path.startswith(watch_path):
91 # compute name of the top level directory that had an event
92 # in the current watch path
93 target = get_top_dir(watch_path, event.path)
94 runfolder = os.path.join(watch_path, target)
96 if not is_runfolder(target):
97 logging.debug("Skipping %s, not a runfolder" % (target,))
100 # grab the previous events for this watch path
101 watch_path_events = self.last_event.setdefault(watch_path, {})
103 # if we've already seen an event in this directory (AKA runfolder)
104 # keep track if its already hit the "completed" flag
105 if watch_path_events.has_key(target):
106 run_already_complete = watch_path_events[target].complete
108 watch_path_events[target] = WatcherEvent(target)
109 #self.last_event.setdefault(watch_path, {})[target] = WatcherEvent(target)
111 msg = "Create: %s %s %s %s" % (watch_path, target, event.path, event.name)
113 # the ReadPrep step uses some of the same file completion flags as the
114 # main analysis, which means this completion code might get tripped because of it
115 # so we need to make sure we're getting the completion file in the root of the
117 if (self.completion_file == event.name.lower() and event.path == runfolder) \
118 or run_already_complete:
119 self.last_event[watch_path][target].complete = True
124 def process_IN_DELETE(self, event):
125 logging.debug("Remove: %s" % os.path.join(event.path, event.name))
128 def process_IN_UNMOUNT(self, event):
129 pathname = os.path.join(event.path, event.name)
130 logging.debug("IN_UNMOUNT: %s" % (pathname,))
131 self.bot.unmount_watch(event.path)
133 class SpoolWatcher(rpc.XmlRpcBot):
135 Watch a directory and send a message when another process is done writing.
137 This monitors a directory tree using inotify (linux specific) and
138 after some files having been written will send a message after <timeout>
139 seconds of no file writing.
141 (Basically when the solexa machine finishes dumping a round of data
142 this'll hopefully send out a message saying hey look theres data available
145 # these params need to be in the config file
146 # I wonder where I should put the documentation
148 # `watchdirs` - list of directories to monitor for modifications
149 # `profile` - specify which .htsworkflow profile to use
150 # `write_timeout` - how many seconds to wait for writes to finish to
152 # `notify_timeout` - how often to timeout from notify
153 # `completion_file` - what file indicates we've finished sequencing
154 # defaults to: netcopy_complete.txt
156 def __init__(self, section=None, configfile=None):
157 #if configfile is None:
158 # self.configfile = "~/.htsworkflow"
159 super(SpoolWatcher, self).__init__(section, configfile)
161 self.cfg['watchdirs'] = None
162 self.cfg['write_timeout'] = 10
163 self.cfg['notify_users'] = None
164 self.cfg['notify_runner'] = None
165 self.cfg['completion_file'] = 'netcopy_complete.txt'
168 self.watchdir_url_map = {}
169 self.notify_timeout = 0.001
172 self.notify_users = None
173 self.notify_runner = None
176 # keep track if the specified mount point is currently mounted
177 self.mounted_points = {}
178 # keep track of which mount points tie to which watch directories
179 # so maybe we can remount them.
180 self.mounts_to_watches = {}
182 self.eventTasks.append(self.process_notify)
184 def read_config(self, section=None, configfile=None):
185 # Don't give in to the temptation to use logging functions here,
186 # need to wait until after we detach in start
187 super(SpoolWatcher, self).read_config(section, configfile)
189 self.watchdirs = shlex.split(self._check_required_option('watchdirs'))
190 # see if there's an alternate url that should be used for the watchdir
191 for watchdir in self.watchdirs:
192 self.watchdir_url_map[watchdir] = self.cfg.get(watchdir, watchdir)
194 self.write_timeout = int(self.cfg['write_timeout'])
195 self.completion_file = self.cfg['completion_file']
197 self.notify_users = self._parse_user_list(self.cfg['notify_users'])
199 self.notify_runner = \
200 self._parse_user_list(self.cfg['notify_runner'],
201 require_resource=True)
202 except bot.JIDMissingResource:
203 msg = 'need a full jabber ID + resource for xml-rpc destinations'
204 raise bot.JIDMissingResource(msg)
209 def add_watch(self, watchdirs=None):
211 start watching watchdir or self.watchdir
212 we're currently limited to watching one directory tree.
214 # create the watch managers if we need them
216 self.wm = pyinotify.WatchManager()
217 self.handler = Handler(self.wm, self, self.completion_file)
218 self.notifier = pyinotify.Notifier(self.wm, self.handler)
220 # the one tree limit is mostly because self.wdd is a single item
221 # but managing it as a list might be a bit more annoying
222 if watchdirs is None:
223 watchdirs = self.watchdirs
225 mask = EventsCodes.IN_CREATE | EventsCodes.IN_UNMOUNT
226 # rec traverses the tree and adds all the directories that are there
228 # auto_add will add in new directories as they are created
230 mount_location = mount.find_mount_point_for(w)
231 self.mounted_points[mount_location] = True
232 mounts = self.mounts_to_watches.get(mount_location, [])
235 self.mounts_to_watches[mount_location] = mounts
237 logging.info(u"Watching:"+unicode(w))
238 self.wdds.append(self.wm.add_watch(w, mask, rec=True, auto_add=True))
240 def unmount_watch(self, event_path):
241 # remove backwards so we don't get weirdness from
242 # the list getting shorter
243 for i in range(len(self.wdds),0, -1):
245 logging.info(u'unmounting: '+unicode(wdd.items()))
246 self.wm.rm_watch(wdd.values())
250 def make_copy_url(self, watchdir, list_event_dir):
251 root_copy_url = self.watchdir_url_map[watchdir]
252 if root_copy_url[-1] != '/':
254 copy_url = root_copy_url + list_event_dir
255 logging.debug('Copy url: %s' % (copy_url,))
258 def process_notify(self, *args):
259 if self.notifier is None:
262 # process the queue of events as explained above
263 self.notifier.process_events()
264 #check events waits timeout
265 if self.notifier.check_events(self.notify_timeout):
266 # read notified events and enqeue them
267 self.notifier.read_events()
268 # should we do something?
269 # has something happened?
270 for watchdir, last_events in self.handler.last_event.items():
271 for last_event_dir, last_event_detail in last_events.items():
272 time_delta = time.time() - last_event_detail.time
273 if time_delta > self.write_timeout:
274 print "timeout", unicode(last_event_detail)
275 copy_url = self.make_copy_url(watchdir, last_event_dir)
276 self.startCopy(copy_url)
277 if last_event_detail.complete:
278 self.sequencingFinished(last_event_detail.event_root)
280 self.handler.last_event[watchdir] = {}
281 # handle unmounted filesystems
282 for mount_point, was_mounted in self.mounted_points.items():
283 if not was_mounted and mount.is_mounted(mount_point):
284 # we've been remounted. Huzzah!
286 for watch in self.mounts_to_watches[mount_point]:
287 self.add_watch(watch)
289 "%s was remounted, restarting watch" % \
292 self.mounted_points[mount_point] = True
294 def _parser(self, msg, who):
296 Parse xmpp chat messages
298 help = u"I can send [copy] message, or squencer [finished]"
299 if re.match(u"help", msg):
301 elif re.match("copy", msg):
303 reply = u"sent copy message"
304 elif re.match(u"finished", msg):
307 self.sequencingFinished(words[1])
308 reply = u"sending sequencing finished for %s" % (words[1])
310 reply = u"need runfolder name"
312 reply = u"I didn't understand '%s'" %(msg)
319 # we have to configure pyinotify after BenderJab.start is called
320 # as weird things happen to pyinotify if the stdio is closed
321 # after it's initialized.
323 super(SpoolWatcher, self).run()
329 # destroy the inotify's instance on this interrupt (stop monitoring)
330 if self.notifier is not None:
332 super(SpoolWatcher, self).stop()
334 def startCopy(self, copy_url=None):
335 logging.debug("writes seem to have stopped")
336 if self.notify_runner is not None:
337 for r in self.notify_runner:
338 self.rpc_send(r, tuple([copy_url]), 'startCopy')
339 if self.notify_users is not None:
340 for u in self.notify_users:
341 self.send(u, 'startCopy %s.' % (copy_url,))
343 def sequencingFinished(self, run_dir):
344 # need to strip off self.watchdirs from rundir I suspect.
345 logging.info("run.completed in " + str(run_dir))
346 for watch in self.watchdirs:
347 if not run_dir.startswith(watch):
348 print "%s didn't start with %s" % (run_dir, watch)
350 if watch[-1] != os.path.sep:
352 stripped_run_dir = re.sub(watch, "", run_dir)
354 stripped_run_dir = run_dir
356 logging.debug("stripped to " + stripped_run_dir)
357 if self.notify_users is not None:
358 for u in self.notify_users:
359 self.send(u, 'Sequencing run %s finished' % \
361 if self.notify_runner is not None:
362 for r in self.notify_runner:
363 self.rpc_send(r, (stripped_run_dir,), 'sequencingFinished')
367 return bot.main(args)
369 if __name__ == "__main__":
370 ret = main(sys.argv[1:])
374 # send messages to copier specifying which mount to copy