openml_search.py

import openml
from langchain_core.prompts import PromptTemplate
from langchain_folder.llm_helper import llm
from dotenv import load_dotenv
import os
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.chrome.service import Service
import requests
import openai_openml as oo


load_dotenv()
url_list = []
api_key = os.getenv('openml_api')
openml.config.apikey = api_key

def extract_keywords(query):
    prompt = PromptTemplate.from_template("""
        You are an assistant whose job is to extract the keywords from the query and return it:
        Query = "{query}"
        For example, if the query is Generate a list of links to datasets related to house price prediction
        your response should be -> "house price".
        Note that the query might not always be related to house price predictions, it can be related to other things as well.
        return only the keywords do not return anything else
    """)
    rendered_prompt = prompt.format(query=query)
    response = llm.invoke(rendered_prompt)
    return response.content

def fetch_dataset_urls(query, limit=4):
    print(f"Searching for datasets related to: {query}")
    # datasets = openml.datasets.list_datasets(output_format="dataframe")
    # matching_datasets = datasets[datasets['name'].str.contains(query, case=False, na=False)]
    # if matching_datasets.empty:
    #     keywords = query.lower().split()
    #     mask = datasets['name'].apply(lambda name: all(kw in str(name).lower() for kw in keywords))
    #     matching_datasets = datasets[mask]

    # if matching_datasets.empty:
    #     print("No datasets found for the query.")
    # else:
    #     matching_datasets = matching_datasets.head(limit)
    #     for index, row in matching_datasets.iterrows():
    #         print(f"📌 Dataset: {row['name']}")
    #         dataset_url = f"https://www.openml.org/d/{row['did']}"
    #         url_list.append(dataset_url)
    #         print(f"🔗 URL: https://www.openml.org/d/{row['did']}\n")
    global url_list
    url_list=oo.openDataset(query)


def openDataset(user_prompt):
    # user_prompt = input("Enter user prompt: ")
    extracted_keywords = extract_keywords(user_prompt)
    print(extracted_keywords)
    fetch_dataset_urls(extracted_keywords)

    download_folder = "./input_folder/"+user_prompt
    if not os.path.exists(download_folder):
        os.makedirs(download_folder)

    chrome_options = Options()
    chrome_options.add_argument("--headless")  # Uncomment to run headless (no UI)
    chrome_options.add_experimental_option("prefs", {
        "download.default_directory": download_folder,  # Set the custom download folder
        "download.prompt_for_download": False,  # Don't ask for confirmation to download
        "download.directory_upgrade": True,  # Allow downloading into the custom folder
        "safebrowsing.enabled": True  # Enable safe browsing (to avoid warnings during download)
    })
    driver = webdriver.Chrome(options=chrome_options)
    for url in url_list:
        driver.get(url)
        try:
            download_button = WebDriverWait(driver, 10).until(
            EC.presence_of_element_located((By.CSS_SELECTOR, "a[aria-label='Download dataset']"))
        )
            actual_download_url = download_button.get_attribute("href")
            filename = actual_download_url.split("/")[-2] + "_" + actual_download_url.split("/")[-1]
            file_path = os.path.join(download_folder, filename)
        
            print(f"⬇️ Downloading from {actual_download_url}")
            response = requests.get(actual_download_url)
            with open(file_path, "wb") as f:
                f.write(response.content)
            print(f"✅ Saved to {file_path}\n")
    
        except Exception as e:
            print(f"❌ Failed to fetch or download from {url}: {e}")

# openDataset("stock market predictions")