from xml.parsers.expat import model
import ml_utiliy
import json
from sklearn.model_selection import train_test_split#训练集/测试机划分函数


class TrainModle(object):
    def __init__(self) -> None:
        self.models_={}
        self.ihd_=ml_utiliy.ihyperDB()
    def Train(self,param_json):
        ruleid=param_json["ruleid"]
        model_type=param_json["model_type"]
        stime=param_json["stime"]
        etime=param_json["etime"]
        features=param_json["features"]
        target=param_json["target"]
        [target_flags,target_data]=self.ihd_.select_raw_data(target,stime,etime)
        # print(target_data)
        y=ml_utiliy.np.array(ml_utiliy.copy.deepcopy(target_data))
        X_d=[]
        for k in features:
            # print(k)
            [feature_flag,feature_data]=self.ihd_.select_raw_data(k,stime,etime)
            X_d.append(ml_utiliy.np.array(ml_utiliy.copy.deepcopy(feature_data)))
        X=ml_utiliy.np.array(X_d).T
        print(X.shape)
        print(y.shape)
        x_train, x_test, y_train, y_test = train_test_split(X, y, test_size=0.2,stratify=y)
        scores=ml_utiliy.train(ruleid=ruleid,model_type=model_type, x_train=x_train,y_train=y_train)
        self.models_[ruleid+model_type]=ml_utiliy.read_mode(ruleid=ruleid,model_type=model_type)
        print(scores)
        print(ml_utiliy.predict_score(ruleid=ruleid,model_type=model_type, x_test=x_test,y_test=y_test))
        return scores
        
    def Predict(self,param_json):
        ruleid=param_json["ruleid"]
        model_type=param_json["model_type"]
        stime=param_json["stime"]
        etime=param_json["etime"]
        features=param_json["features"]
        target=param_json["target"]
        try:
            self.models_[ruleid+model_type]
        except:
            print("载入")
            self.models_[ruleid+model_type]=ml_utiliy.read_mode(ruleid=ruleid,model_type=model_type)
        [target_flags,target_data]=self.ihd_.select_raw_data(target,stime,etime)
        # print(target_data)
        y=ml_utiliy.np.array(ml_utiliy.copy.deepcopy(target_data))
        X_d=[]
        for k in features:
            # print(k)
            [feature_flag,feature_data]=self.ihd_.select_raw_data(k,stime,etime)
            X_d.append(ml_utiliy.np.array(ml_utiliy.copy.deepcopy(feature_data)))
        X=ml_utiliy.np.array(X_d).T
        print(X.shape)
        print(y.shape)
        if(self.models_[ruleid+model_type]):
            scores=self.models_[ruleid+model_type].score(X,y)
            print("scores:{}".format(scores))
            return scores
        print("ERROR")
        scores=ml_utiliy.predict_score(ruleid=ruleid,model_type=model_type,x_test=X,y_test=y)
        print("scores:{}".format(scores))
        return scores