'''
Created on Jul 4, 2017

@author: David
'''

import cPickle as pickle
import numpy as np

#p  = "swellex-bigram-mlp-1000-adaptive-subs.pickle"
receptive_pickle = "siwoco/ml/svalex-r-mlp-100_0.001.pickle"
productive_pickle = "siwoco/ml/swellex-r-mlp-50_10_0.001.pickle" 

#p1 = "swellex-logreg-a1c1.pickle"
#p2 = "swellex-logreg-a2b2.pickle"

classifier_receptive = pickle.load(open(receptive_pickle, "rb"))
classifier_productive = pickle.load(open(productive_pickle, "rb"))

#a1c1p = pickle.load(open(p1, "rb"))
#a2b2p = pickle.load(open(p2, "rb"))

def predict_adj(data, rop):
    p_pro = classifier_productive.predict(data)[0]
    p_rec = classifier_receptive.predict(data)[0]
    if p_rec > p_pro:  # adjust productive level if necessary
        p_rec = p_pro
    if p_rec == "A1" and p_pro == "B2":  # TODO evaluate this
        p_pro = "B1"
    if p_rec == "A2" and p_pro == "C1":
        p_pro = "B2"
    if rop == "both":
        return (p_rec,p_pro)
    if rop == "receptive":
        return p_rec
    return p_pro

def predict(data, receptiveOrProductive):
    return predict_adj(data,receptiveOrProductive)
#    if receptiveOrProductive == "both":
#        return (classifier_receptive.predict(data)[0], classifier_productive.predict(data)[0])
#    if receptiveOrProductive == "receptive":
#        return classifier_receptive.predict(data)[0]
#    else:
#        return classifier_productive.predict(data)[0]

def predict_hybrid(data):
    a1c1 = a1c1p.predict(data)
    if a1c1[0] == "B1":
        a2b2 = a2b2p.predict(data)
        if a2b2[0] == "B2":
            return "B2"
        if a2b2[0] == "A2":
            return "A2"
        return "B1"
    return a1c1[0]