annotate galaxy-tools/biobank/utils/drop_flowcell_related_items.py @ 12:46f08bb8dd68 draft default tip

Uploaded
author ric
date Wed, 28 Sep 2016 04:59:02 -0400
parents 43be74e62bfe
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
43be74e62bfe Uploaded
ric
parents:
diff changeset
1 import argparse, sys
43be74e62bfe Uploaded
ric
parents:
diff changeset
2 from collections import Counter
43be74e62bfe Uploaded
ric
parents:
diff changeset
3
43be74e62bfe Uploaded
ric
parents:
diff changeset
4 from bl.vl.kb import KnowledgeBase as KB
43be74e62bfe Uploaded
ric
parents:
diff changeset
5 import bl.vl.utils.ome_utils as vlu
43be74e62bfe Uploaded
ric
parents:
diff changeset
6 from bl.vl.utils import get_logger, LOG_LEVELS
43be74e62bfe Uploaded
ric
parents:
diff changeset
7 from bl.vl.kb.drivers.omero.sequencing import SeqDataSample, SequencerOutput
43be74e62bfe Uploaded
ric
parents:
diff changeset
8
43be74e62bfe Uploaded
ric
parents:
diff changeset
9
43be74e62bfe Uploaded
ric
parents:
diff changeset
10 def make_parser():
43be74e62bfe Uploaded
ric
parents:
diff changeset
11 parser = argparse.ArgumentParser(description='delete all items related to the given flowcell')
43be74e62bfe Uploaded
ric
parents:
diff changeset
12 parser.add_argument('--logfile', type=str, help='log file (default=stderr)')
43be74e62bfe Uploaded
ric
parents:
diff changeset
13 parser.add_argument('--loglevel', type=str, choices=LOG_LEVELS,
43be74e62bfe Uploaded
ric
parents:
diff changeset
14 help='logging level (default=INFO)', default='INFO')
43be74e62bfe Uploaded
ric
parents:
diff changeset
15 parser.add_argument('-H', '--host', type=str, help='omero hostname')
43be74e62bfe Uploaded
ric
parents:
diff changeset
16 parser.add_argument('-U', '--user', type=str, help='omero user')
43be74e62bfe Uploaded
ric
parents:
diff changeset
17 parser.add_argument('-P', '--passwd', type=str, help='omero password')
43be74e62bfe Uploaded
ric
parents:
diff changeset
18 parser.add_argument('--flowcell-label', type=str, required=True,
43be74e62bfe Uploaded
ric
parents:
diff changeset
19 help='flowcell label')
43be74e62bfe Uploaded
ric
parents:
diff changeset
20 parser.add_argument('--datasets-only', action='store_true',
43be74e62bfe Uploaded
ric
parents:
diff changeset
21 help='delete only data samples and data objects related to the given flowcell')
43be74e62bfe Uploaded
ric
parents:
diff changeset
22 return parser
43be74e62bfe Uploaded
ric
parents:
diff changeset
23
43be74e62bfe Uploaded
ric
parents:
diff changeset
24
43be74e62bfe Uploaded
ric
parents:
diff changeset
25 def get_sources(objs):
43be74e62bfe Uploaded
ric
parents:
diff changeset
26 sources = set()
43be74e62bfe Uploaded
ric
parents:
diff changeset
27 for x in objs:
43be74e62bfe Uploaded
ric
parents:
diff changeset
28 try:
43be74e62bfe Uploaded
ric
parents:
diff changeset
29 sources.add(x.action.target)
43be74e62bfe Uploaded
ric
parents:
diff changeset
30 except AttributeError:
43be74e62bfe Uploaded
ric
parents:
diff changeset
31 # Action has no "target" attribute, no source for item x
43be74e62bfe Uploaded
ric
parents:
diff changeset
32 pass
43be74e62bfe Uploaded
ric
parents:
diff changeset
33 return sources
43be74e62bfe Uploaded
ric
parents:
diff changeset
34
43be74e62bfe Uploaded
ric
parents:
diff changeset
35
43be74e62bfe Uploaded
ric
parents:
diff changeset
36 def delete_objects(objs, kb, logger, max_retries = 3):
43be74e62bfe Uploaded
ric
parents:
diff changeset
37 retry_ct = Counter()
43be74e62bfe Uploaded
ric
parents:
diff changeset
38 while len(objs) > 0:
43be74e62bfe Uploaded
ric
parents:
diff changeset
39 o = objs.pop(0)
43be74e62bfe Uploaded
ric
parents:
diff changeset
40 if type(o) in [SeqDataSample, SequencerOutput]:
43be74e62bfe Uploaded
ric
parents:
diff changeset
41 logger.info('Loading DataObjects for %s:%s' % (o.__class__.__name__,
43be74e62bfe Uploaded
ric
parents:
diff changeset
42 o.label))
43be74e62bfe Uploaded
ric
parents:
diff changeset
43 dobjs = kb.get_data_objects(o)
43be74e62bfe Uploaded
ric
parents:
diff changeset
44 logger.info('%d DataObjects loaded' % len(dobjs))
43be74e62bfe Uploaded
ric
parents:
diff changeset
45 for d in dobjs:
43be74e62bfe Uploaded
ric
parents:
diff changeset
46 logger.info('Deleting %s:%s' % (d.__class__.__name__,
43be74e62bfe Uploaded
ric
parents:
diff changeset
47 d.path))
43be74e62bfe Uploaded
ric
parents:
diff changeset
48 kb.delete(d)
43be74e62bfe Uploaded
ric
parents:
diff changeset
49 try:
43be74e62bfe Uploaded
ric
parents:
diff changeset
50 logger.info('Deleting %s:%s' % (o.__class__.__name__,
43be74e62bfe Uploaded
ric
parents:
diff changeset
51 o.id))
43be74e62bfe Uploaded
ric
parents:
diff changeset
52 act = o.action
43be74e62bfe Uploaded
ric
parents:
diff changeset
53 kb.delete(o)
43be74e62bfe Uploaded
ric
parents:
diff changeset
54 try:
43be74e62bfe Uploaded
ric
parents:
diff changeset
55 logger.info('Deleting source action %s:%s' % (act.__class__.__name__,
43be74e62bfe Uploaded
ric
parents:
diff changeset
56 act.id))
43be74e62bfe Uploaded
ric
parents:
diff changeset
57 kb.delete(act)
43be74e62bfe Uploaded
ric
parents:
diff changeset
58 except:
43be74e62bfe Uploaded
ric
parents:
diff changeset
59 logger.info('Can\'t delete action')
43be74e62bfe Uploaded
ric
parents:
diff changeset
60 except:
43be74e62bfe Uploaded
ric
parents:
diff changeset
61 logger.info('Can\'t delete, putting back into objects list')
43be74e62bfe Uploaded
ric
parents:
diff changeset
62 if retry_ct['%s:%s' % (type(o), o.id)] < max_retries:
43be74e62bfe Uploaded
ric
parents:
diff changeset
63 objs.append(o)
43be74e62bfe Uploaded
ric
parents:
diff changeset
64 retry_ct['%s:%s' % (type(o), o.id)] += 1
43be74e62bfe Uploaded
ric
parents:
diff changeset
65 else:
43be74e62bfe Uploaded
ric
parents:
diff changeset
66 logger.info('Reached maximum retry limit for the object, skipping')
43be74e62bfe Uploaded
ric
parents:
diff changeset
67
43be74e62bfe Uploaded
ric
parents:
diff changeset
68 def main(argv):
43be74e62bfe Uploaded
ric
parents:
diff changeset
69 parser = make_parser()
43be74e62bfe Uploaded
ric
parents:
diff changeset
70 args = parser.parse_args(argv)
43be74e62bfe Uploaded
ric
parents:
diff changeset
71
43be74e62bfe Uploaded
ric
parents:
diff changeset
72 logger = get_logger('drop_flowcell_related_items', level=args.loglevel,
43be74e62bfe Uploaded
ric
parents:
diff changeset
73 filename=args.logfile)
43be74e62bfe Uploaded
ric
parents:
diff changeset
74
43be74e62bfe Uploaded
ric
parents:
diff changeset
75 try:
43be74e62bfe Uploaded
ric
parents:
diff changeset
76 host = args.host or vlu.ome_host()
43be74e62bfe Uploaded
ric
parents:
diff changeset
77 user = args.user or vlu.ome_user()
43be74e62bfe Uploaded
ric
parents:
diff changeset
78 passwd = args.passwd or vlu.ome_passwd()
43be74e62bfe Uploaded
ric
parents:
diff changeset
79 except ValueError, ve:
43be74e62bfe Uploaded
ric
parents:
diff changeset
80 logger.critical(ve)
43be74e62bfe Uploaded
ric
parents:
diff changeset
81 sys.exit(ve)
43be74e62bfe Uploaded
ric
parents:
diff changeset
82
43be74e62bfe Uploaded
ric
parents:
diff changeset
83 kb = KB(driver='omero')(host, user, passwd)
43be74e62bfe Uploaded
ric
parents:
diff changeset
84
43be74e62bfe Uploaded
ric
parents:
diff changeset
85 logger.info('Retrieving flowcell with label %s' % args.flowcell_label)
43be74e62bfe Uploaded
ric
parents:
diff changeset
86 query = 'SELECT fc FROM FlowCell fc WHERE fc.label = :fc_label'
43be74e62bfe Uploaded
ric
parents:
diff changeset
87 results = kb.find_all_by_query(query, {'fc_label' : args.flowcell_label})
43be74e62bfe Uploaded
ric
parents:
diff changeset
88 if len(results) == 0:
43be74e62bfe Uploaded
ric
parents:
diff changeset
89 logger.info('No FlowCell with label %s in the database' % args.flowcell_label)
43be74e62bfe Uploaded
ric
parents:
diff changeset
90 sys.exit(0)
43be74e62bfe Uploaded
ric
parents:
diff changeset
91 fc = results[0]
43be74e62bfe Uploaded
ric
parents:
diff changeset
92 logger.info('Loaded FlowCell with ID %s' % fc.id)
43be74e62bfe Uploaded
ric
parents:
diff changeset
93 logger.info('Loading related Lanes')
43be74e62bfe Uploaded
ric
parents:
diff changeset
94 lanes = list(kb.get_lanes_by_flowcell(fc))
43be74e62bfe Uploaded
ric
parents:
diff changeset
95 logger.info('%d Lanes loaded' % len(lanes))
43be74e62bfe Uploaded
ric
parents:
diff changeset
96 logger.info('Loading related LaneSlots')
43be74e62bfe Uploaded
ric
parents:
diff changeset
97 lane_slots = []
43be74e62bfe Uploaded
ric
parents:
diff changeset
98 for l in lanes:
43be74e62bfe Uploaded
ric
parents:
diff changeset
99 lane_slots += list(kb.get_laneslots_by_lane(l))
43be74e62bfe Uploaded
ric
parents:
diff changeset
100 logger.info('%d LaneSlots loaded' % len(lane_slots))
43be74e62bfe Uploaded
ric
parents:
diff changeset
101 logger.info('Loading related Tubes')
43be74e62bfe Uploaded
ric
parents:
diff changeset
102 sub_samples = get_sources(lane_slots)
43be74e62bfe Uploaded
ric
parents:
diff changeset
103 samples = get_sources(sub_samples)
43be74e62bfe Uploaded
ric
parents:
diff changeset
104 logger.info('%d Tubes loaded' % (len(sub_samples) + len(samples)))
43be74e62bfe Uploaded
ric
parents:
diff changeset
105 logger.info('Loading related SequencerOutputs')
43be74e62bfe Uploaded
ric
parents:
diff changeset
106 seq_out = kb.dt.get_connected(fc, aklass = kb.SequencerOutput, direction = kb.dt.DIRECTION_OUTGOING)
43be74e62bfe Uploaded
ric
parents:
diff changeset
107 logger.info('%d SequencerOutputs loaded' % len(seq_out))
43be74e62bfe Uploaded
ric
parents:
diff changeset
108 logger.info('Loading related SeqDataSamples')
43be74e62bfe Uploaded
ric
parents:
diff changeset
109 seq_dsamples = kb.dt.get_connected(fc, aklass = kb.SeqDataSample, direction = kb.dt.DIRECTION_OUTGOING)
43be74e62bfe Uploaded
ric
parents:
diff changeset
110 logger.info('%d SeqDataSamples loaded' % len(seq_dsamples))
43be74e62bfe Uploaded
ric
parents:
diff changeset
111
43be74e62bfe Uploaded
ric
parents:
diff changeset
112 if args.datasets_only:
43be74e62bfe Uploaded
ric
parents:
diff changeset
113 delete_items = [seq_dsamples]
43be74e62bfe Uploaded
ric
parents:
diff changeset
114 else:
43be74e62bfe Uploaded
ric
parents:
diff changeset
115 delete_items = [seq_dsamples, seq_out, lane_slots, lanes,
43be74e62bfe Uploaded
ric
parents:
diff changeset
116 [fc], list(sub_samples), list(samples)]
43be74e62bfe Uploaded
ric
parents:
diff changeset
117 for items in delete_items:
43be74e62bfe Uploaded
ric
parents:
diff changeset
118 delete_objects(items, kb, logger)
43be74e62bfe Uploaded
ric
parents:
diff changeset
119
43be74e62bfe Uploaded
ric
parents:
diff changeset
120 logger.info('Job completed')
43be74e62bfe Uploaded
ric
parents:
diff changeset
121
43be74e62bfe Uploaded
ric
parents:
diff changeset
122
43be74e62bfe Uploaded
ric
parents:
diff changeset
123 if __name__ == '__main__':
43be74e62bfe Uploaded
ric
parents:
diff changeset
124 main(sys.argv[1:])