WebScout_Deep_DDC.py

from webscout import DeepWEBS

def perform_web_search(query, result_num=10, safe=True, extract_webpage=True, overwrite_query_html=False, overwrite_webpage_html=False):
    
    # Initialize the DeepWEBS class
    deep_search = DeepWEBS()
    
    # Set up the search parameters
    search_params = deep_search.DeepSearch(
        queries=[query],  # Query to search
        result_num=result_num,  # Number of search results
        safe=safe,  # Enable SafeSearch
        types=["web"],  # Search type: web
        extract_webpage=extract_webpage,  # True for extracting webpages
        overwrite_query_html=overwrite_query_html,
        overwrite_webpage_html=overwrite_webpage_html,
    )

    
    # Execute the search and retrieve results
    results = deep_search.queries_to_search_results(search_params)
    
    return results

def attributes_ext(raw_result):
    titles_list = []
    urls_list = []
    text_list = []

    for values in raw_result:
        for title in values['query_results']:
            titles_list.append(title['title'])

        for url in values['query_results']:
            urls_list.append(url['url'])

        for body in values['query_results']:
            text_list.append(body['text'])

    return titles_list, urls_list, text_list

def main(query, result_num=10, safe=True, extract_webpage=True, overwrite_query_html=False, overwrite_webpage_html=False):
    
    raw_output = perform_web_search(query, result_num=result_num, safe=safe, extract_webpage=extract_webpage, overwrite_query_html=overwrite_query_html, overwrite_webpage_html=overwrite_webpage_html)

    titles_list, urls_list, text_list = attributes_ext(raw_output)

    return titles_list, urls_list, text_list

if __name__ == "__main__":
    # Prompt the user for a search query
    query = "ipl 2p24"
        
    # Perform the web search
    results = perform_web_search(query)