def _next_erepo(self):
try:
erepo = ERepo.get(ERepo._elaborated == False)
except ERepo.DoesNotExist:
erepo = None
return erepo
"""A script to interactively examine the distribution of repos."""
import cPickle as pickle
import datetime
import numpy as np
import random
from config import feature_pickle_name
from models import ERepo
import utils
cutoff = datetime.datetime.now() - datetime.timedelta(days=6 * 30)
frepos = [r for r in ERepo.select().where(ERepo.created_at < cutoff)]
frepos = [r for r in frepos if
r.size > 0 and
r._stars > 2 and
r.fork == False and
r.master_branch == 'master' and
not (r.size > 30720) #and r._stars < 100) # avoid huge repos, but keep pop
]
print 'filtered:', len(frepos)
def divy(ranges):
res = []
for l, h in ranges:
res.append([x for x in frepos if l <= x._stars <= h])
return res