import pandas as pd
from sklearn.model_selection import train_test_split
import tensorflow as tf
import sys
from Status.Status import Status
sys.path.append('/home/skjena/cnnCancerTherapy/modules/NeuralNet/core/classifiers/dnnClassifier')
sys.path.append('/home/skjena/cnnCancerTherapy/modules/NeuralNet/core/regressors/dnnRegressor')

from NeuralNet.core.classifiers.dnnClassifier import DNNClassifierModel
#from NeuralNet.core.regressors.dnnRegressor import DNNRegressorModel
from NeuralNet.core.classifiers.dnnClassifier import dataProcessor

class predictManager():
	def __init__(self, predict_x, expected):
		self.predict_x = predict_x
		self.expected = expected

	def run(self, dnnModel, problemType):
		self.status.message(1, "run(self, model)")
		if(problemType == "0"):
			predictions = dnnModel.predict(
			input_fn=lambda:dataProcessor.predict_input_fn(self.predict_x,
			labels=None, batch_size=DNNClassifierModel.getBatchSize()))
		else:
			predictions = dnnModel.predict(
			input_fn=lambda:dataProcessor.predict_input_fn(self.predict_x,
			labels=None, batch_size=DNNRegressorModel.getBatchSize()))

		for pred_dict, expec in zip(predictions, expected):
			template = ('\nPrediction is "{}" ({:.1f}%), expected "{}"')
			class_id = pred_dict['class_ids'][0]
			probability = pred_dict['probabilities'][class_id]
			# print the correct answer's label, it's probability scaled into a percentage, and the expected class from the list
		print(template.format(dataProcessor.TUMOR[class_id], 100 * probability, expec))
		self.status.message(0, "run(self, model)")
		return