import tensorflow as tf

import numpy as np

from autoencoder import autoencoder

from optparse import OptionParser

from tools import get_data_from_minst

parser = OptionParser()

parser.add_option("-t","--remove_mean",dest="rem_mean",default=True)

parser.add_option("-u", "--hidden", dest="hidden",

help="number of hidden units",default=10)

parser.add_option("-a", "--activation",

dest="activation", default='sigmoid',

help="activation function")

parser.add_option("-b","--batch",dest="batch",default=False,

help="Using batch learning")

parser.add_option("-n","--number_batch",dest="n_batch",default=5,

help="Number of batches for learning")

parser.add_option("-g","--gradient",dest="gradient",default='gradient',

help="Optimization metodh")

parser.add_option("-l","--learning_rate",dest="learn_rate",default=0.01,

help="Learning rate")

parser.add_option("-i","--iters",dest="iters",default=1000,

help="Number of iterations")

parser.add_option("-m","--model_name",dest="model_name",default="./model.ckpt",

help="Filename for model file")

parser.add_option("-s","--symm",dest="symm",default=False,help="Symmetric autoencoder")

parser.add_option("-r","--reg_w",dest="reg_w",default=False,help="Cost function with regularized weights")

parser.add_option("-w","--w_file",dest="w_file",default="./weight.txt",help="File for storing norm weights")

parser.add_option("-c","--class_label",dest="class_label",default=0,help="Class label to use")

parser.add_option("-p","--pre_train_learning_rate",dest="pre_learn_rate",default=0.001,

help="Learning rate for RBM pre-training")

parser.add_option("-e","--pre_train",dest="pre_train",default="rbm",help="Select different pre-train or none(default RBM)")

parser.add_option("-d","--reg_lambda",dest="reg_lambda",default=0.05,help="regularization weight (lambda)")

parser.add_option("-o","--drop_out",dest="drop_out",default=False,help="using dropout")

parser.add_option("-k","--keep_prob",dest="keep_prob",default=0.5,help="probability for dropout")

parser.add_option("-z","--euris",dest="euris",default=True,help="using euristic for weight initialization")

(options, args) = parser.parse_args()

units = [784,int(options.hidden)]

action = [options.activation]

k = float(options.keep_prob)

l_rate = options.learn_rate

grad = options.gradient

arr,lab = get_data_from_minst.get_data_from_minst()

data = arr

print options

auto = autoencoder(units,action)

auto.generate_encoder(euris=options.euris)

auto.generate_decoder(symmetric=options.symm)

#auto.pre_train_rbm(data,n_iters=10,learning_rate=float(options.pre_learn_rate),adapt_learn=True)

if(not options.batch):

bat = None

else:

from tools.data_manipulation.batch import seq_batch

bat = seq_batch(data,int(options.n_batch))

auto.train(data,n_iters=int(options.iters),record_weight=True,w_file=options.w_file,use_dropout=True,keep_prob=k,reg_weight=False,reg_lambda=0.0,model_name=options.model_name,batch=bat,display=False,noise=True,gradient=options.gradient,learning_rate=float(options.learn_rate))