test.py

from modules.factcheckexplorer import FactCheckLib
from modules import adapter
import environ

env = environ.Env()
environ.Env.read_env()
ad = adapter.Adapter(env)

fact_check = FactCheckLib(language="en", num_results=50)

claim = "climage change is fake"

extract_prompt = "Extract the keywords from the following text: {claim}. These keywords will be used to search for information in a database. Only return the key words. Do not include any other text."
validate_prompt = """Validate the following claim: {claim} based on the following information: {report}. 
Answer the claim, if the claim is not a question, but keywords, then review the data and determine if the claim subject is true or false."""
reduce_prompt = "The following keywords: '{key_words}' are too broad. The most important keywords are typically nouns. Remove the least relevant keyword. Return the reduced keywords only."

def clean_keywords(key_words):
    if ":" in key_words:
        key_words = key_words.split(":")[1].strip()
    key_words = key_words.replace(",", "").replace("and", "").replace("or", "").replace('*', "")
    print(f"Extracted key words: {key_words}")
    return key_words

key_words = clean_keywords(ad.llm_chat.invoke(extract_prompt.format(claim=claim)).content)




while True:
    report = fact_check.process(key_words)
    if len( report ) == 0:
        if len(key_words) == 0:
            print("No keywords extracted. Exiting.")
            break
        else:
            key_words =  clean_keywords(ad.llm_chat.invoke(reduce_prompt.format(key_words=key_words)).content)
            print( f"Reduced key words: {key_words}")
    else:    
        check = ad.llm_chat.invoke(validate_prompt.format(claim=claim, report=report)).content
        print(f"Validation result: {check}")
        break


# print(ad.llm_chat.invoke(reduce_prompt.format(key_words=key_words)).content)