remove orphans & duplicates - but no removal yet.

removeOrphans.py

from progressbar import ProgressBar, ETA, Percentage, Bar
widgets = [Percentage(), Bar(), ETA()]
from sys import exit as sys_exit
from DmpWorkflow.core.models import JobInstance, Job
from DmpWorkflow.core import db


db.connect()
all_jobs = Job.objects.all()
n_registered_instances = sum([JobInstance.objects.filter(job=j).count() for j in all_jobs])
n_instances_total = JobInstance.objects.all().count()

instances = JobInstance.objects.all()

if n_registered_instances == n_instances_total:
	print 'no duplicates or orphans found, returning'
	sys_exit()

instances = JobInstance.objects.all()
print 'found %i instances'%n_instances_total
pb = ProgressBar(widgets=widgets,maxval=n_instances_total)

duplicates = []
orphans = []
good_instances = []
pb.start()
for i,inst in enumerate(instances):
	if not inst.job in all_jobs:
		orphans.append(inst)
		continue
	if inst.instanceId in good_instances:
		duplicates.append(inst)
		continue
	good_instances.append(inst.instanceId)
	pb.update(i+1)
pb.finish()

if len(duplicates):
	print 'found %i duplicates'%len(duplicates)
if len(orphans):
	print 'found %i orphans'%len(orphans)