import difflib
import re


def normalize_question(question):
    if question is None:
        return None

    # Trim whitespace and remove all trailing punctuation
    # Remove trailing punctuation (one or more of '.', ',', '!', '?')
    question = re.sub(r"[.,!?]+$", "", question.strip())

    # Convert to lowercase
    return question.lower()


def extract_question(qa_content):
    # Use regular expression to find the text after "User question:"
    match = re.search(r"User question:\s*(.*)", qa_content, re.IGNORECASE)
    if match:
        # Extract the matched question and process it
        question = match.group(1).strip()  # Trim leading and trailing whitespace
        return normalize_question(question)
    return None


class SimpleCache:
    def __init__(self, items: list = []):
        self.cache = []
        self.questions_in_cache = []
        for item in items:
            self.add(item["question"], item["job_id"])

    def add(self, question, job_id):
        """
        Adds a question and job_id to the cache.
        """
        question = normalize_question(question)
        if question is None or question in self.questions_in_cache:
            return
        self.cache.append({"question": normalize_question(question), "job_id": job_id})
        self.questions_in_cache.append(normalize_question(question))

    def get(self, question):
        """
        Retrieves the job_id for a given question using simple string matching.
        """
        question = normalize_question(question)
        if question is None:
            return None
        closest_match = difflib.get_close_matches(
            question, self.questions_in_cache, n=1, cutoff=0.90
        )

        if closest_match:
            for item in self.cache:
                if item["question"] == closest_match[0]:
                    return item["job_id"]
        return None


# It's a workaround solution to reduce duplicated job validation
common_questions = [
    {
        "question": "Does TiDB support serverless?",
        "job_id": "b936c233-5731-43cc-b483-281623d648a5",
    },
    {
        "question": "Does TiDB support vector?",
        "job_id": "98c50963-7485-475e-9dc9-45568d9dafce",
    },
    {
        "question": "Does TiDB support vector search?",
        "job_id": "98c50963-7485-475e-9dc9-45568d9dafce",
    },
    {
        "question": "Does TiDB support FOREIGN KEY?",
        "job_id": "b0567683-257e-4006-ab34-ef3238fcc538",
    },
    {"question": "What is TiDB?", "job_id": "5185084f-6dca-4afd-8df5-cd98cf975098"},
    {"question": "what's tidb?", "job_id": "5185084f-6dca-4afd-8df5-cd98cf975098"},
]
simple_semantic_cache = SimpleCache(common_questions)