def _test():
from testdata import pseudo_shingakunet
model = transition_model(data = pseudo_shingakunet(),
goal = 'https://shingakunet.com/net2/shiryoSeikyu/entry/complete')
return model
elif netloc_ratio >= upper_criteria_url and path_ratio >= upper_criteria_url: #クエリが長過ぎて一致条件に当てはまらないものを救う
if query_ratio >= lower_criteria_url * 0.50:
states[state].append(url)
adopted_flag = True
accumulated += uu
break
#以前に採用したstateとマージされなかった場合は新規stateとして採用する
if not adopted_flag:
states.setdefault(url, [])
states[url].append(url)
accumulated += uu
#条件を満たしたらマージをやめる
if len(states) >= max_state + accumulated_states or accumulated >= minimum_volume * total_uu:
print 'states:', len(states) - accumulated_states, 'volume:', float(accumulated) / float(total_uu)
break
for state, url in states.items():
states[state] = list(set(states[state]))
#uu_volumeが十分にとれなかった場合には何らかの方法で救済する?
return states
if __name__ == '__main__':
from get_model import make_uu_base_data
from testdata import pseudo_shingakunet
states = summarize(make_uu_base_data(pseudo_shingakunet()),
'https://shingakunet.com/net2/shiryoSeikyu/entry/complete')
for state, sub_states in states.items():
print 'state:', state, 'sub-state:', sub_states
