from abc import ABC, abstractmethod

class ClusteringPipeline(ABC):

    

    @abstractmethod
    def preprocessor(self,execution_traces_agilkia_format,**kwargs):
        pass
    
    @abstractmethod
    def fit_predict(self,preprocessed_execution_traces,k):
        pass

from abc import ABC, abstractmethod
class SampleHeuristic(ABC):


    @abstractmethod
    def tests_extraction(self,execution_traces_agilkia_format,cluster_labels):
        pass

from .clustering_pipeline import ClusteringPipeline
from .utils_preprocessing import traceset_to_textset
from sklearn.cluster import KMeans


import gensim
import numpy as np

class KmeansW2v(ClusteringPipeline):
    
    def __init__(self,dataset_name,filepath,**kwargs):
        self.dataset_name=dataset_name
        self.filepath=filepath
    def preprocessor(self, execution_traces_agilkia_format,**kwargs):
        listset=traceset_to_textset(execution_traces_agilkia_format,format='lst')
        model = gensim.models.Word2Vec(sentences=listset,vector_size=10,window=5,min_count=1)
        means=[]
        for seq in listset:
            vecs=[model.wv[elt] for elt in seq]
            means.append(np.mean(vecs,axis=0))
        X=np.array(means)
        
        return X

    def fit_predict(self, preprocessed_execution_traces,k):
        model=KMeans(n_clusters=k)
        y=model.fit_predict(preprocessed_execution_traces)
        self.model=model
        return  y

pipelines+=[("BOW|Kmeans|RS",KmeansPipeline,Sampling,vae_dict)]
pipelines+=[("BOW|Kmeans|BUC",KmeansPipeline,BUC,vae_dict)]
pipelines+=[("BOW+POH|Kmeans|BUC",BowPlus,BUC,vae_dict)]
pipelines+=[("BOW+POH|Kmeans|RS",BowPlus,Sampling,vae_dict)]
pipelines+=[("TermFreq|Kmeans|RS",TfIdf,Sampling,vae_dict)]
pipelines+=[("TermFreq|Kmeans|BUC",TfIdf,BUC,vae_dict)]