import sys
import os
import getFiles.getKaggle as getKaggle
import getFiles.getGoogle as getGoogle
import getFiles.getGithub as getGithub
import getFiles.getHuggingFace as gh
import cleanDataset
import openml_search
import clean_openml

sys.path.append(os.path.abspath(os.path.join(os.getcwd(), 'langchain_folder')))

from langchain_folder import main as m
import json

def downloadDatasets():
    data=input("enter query : ")
    kag,git,hug=getGoogle.googleDatasets(data)
    print(kag)
    print("this is github : ")
    print(git)
    print(hug)
    if(len(kag)>0):
        for url in kag:
            getKaggle.kaggleDataset(url,data)
    if(len(git)>0):
        for url in git:
            getGithub.githubDataset(url,data)
    if(len(hug)>0):
        for url in hug:
            gh.huggingDataset(url,data)

    openml_search.openDataset(data)
    clean_openml.clean(data)


    cleanDataset.clean(data)
    
downloadDatasets()