from huggingface_hub import InferenceClient
import nltk
import re 
import requests
import os 

api_key = os.getenv("HF_KEY")

nltk.download('punkt')
nltk.download('punkt_tab')
nltk.download('averaged_perceptron_tagger')
nltk.download('averaged_perceptron_tagger_eng')


client = InferenceClient(api_key=api_key)


def extract_product_info(text):
    # Initialize result dictionary
    result = {"brand": None, "model": None, "description": None, "price": None}

    try:

        # Improved regex to prioritize currency-related patterns
        price_match = re.search(
            r'(\$\s?\d{1,3}(?:,\d{3})*(?:\.\d{2})?|(?:\d{1,3}(?:,\d{3})*(?:\.\d{2})?\s?(?:USD|usd|dollars|DOLLARS)))',
            text
        )
        if price_match:
            price = price_match.group().strip()
            # Clean up the price format
            if "$" in price or "USD" in price or "usd" in price:
                result["price"] = re.sub(r'[^\d.]', '', price)  # Keep only digits and decimals
            else:
                result["price"] = price
            # Remove the price part from the text to prevent it from being included in the brand/model extraction
            text = text.replace(price_match.group(), "").strip()

        try:
            tokens = nltk.word_tokenize(text)
            print(f"Tokens: {tokens}")
        except Exception as e:
            print(f"Error during tokenization: {e}")
            # Fall back to a simple split if tokenization fails
            tokens = text.split()
            print(f"Fallback tokens: {tokens}")

        try:
            pos_tags = nltk.pos_tag(tokens)
            print(f"POS Tags: {pos_tags}")
        except Exception as e:
            print(f"Error during POS tagging: {e}")
            # If POS tagging fails, create dummy tags
            pos_tags = [(word, "NN") for word in tokens]
            print(f"Fallback POS Tags: {pos_tags}")

        # Extract brand, model, and description
        brand_parts = []
        model_parts = []
        description_parts = []
        
        for word, tag in pos_tags:
            if tag == 'NNP' or re.match(r'[A-Za-z0-9-]+', word):
                if len(brand_parts) == 0:  # Assume the first proper noun is the brand
                    brand_parts.append(word)
                else:  # Model number tends to follow the brand
                    model_parts.append(word)
            else:
                description_parts.append(word)
        
        # Assign values to the result dictionary
        if brand_parts:
            result["brand"] = " ".join(brand_parts)
        if model_parts:
            result["model"] = " ".join(model_parts)
        if description_parts:
            result["description"] = " ".join(description_parts)
        
        print(f"Extract function returned: {result}")
        
    except Exception as e:
        print(f"Unexpected error: {e}")
        # Return a fallback result in case of a critical error
        result["description"] = text
        print(f"Fallback result: {result}")

    return result



def extract_info(text):
    API_URL = "https://api-inference.huggingface.co/models/google/flan-t5-large"
    headers = {"Authorization": f"Bearer {api_key}"}
    payload = {"inputs": f"From the given text, extract brand name, model number, description about it, and its average price in today's market. Give me back a python dictionary with keys as brand_name, model_number, desc, price. The text is {text}.",}
    response = requests.post(API_URL, headers=headers, json=payload)
    print('GOOGLEE LLM OUTPUTTTTTTT\n\n',response )
    output = response.json()
    print(output)


def get_name(url, object):
    messages = [
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": f"Is this a {object}?. Can you guess what it is and give me the closest brand it resembles to? or a model number? And give me its average price in today's market in USD. In output, give me its normal name, model name, model number and price. separated by commas. No description is needed." 
                },
                {
                    "type": "image_url",
                    "image_url": {
                        "url": url
                    }
                }
            ]
        }
    ]

    completion = client.chat.completions.create(
        model="meta-llama/Llama-3.2-11B-Vision-Instruct", 
        messages=messages, 
        max_tokens=500
    )

    print(f'\n\nNow output of LLM:\n')
    llm_result = completion.choices[0].message['content']
    print(llm_result)

    # print(f'\n\nThat is the output')
    print(f"Extracting from the output now, function calling")
    result = extract_product_info(llm_result)
    print(f'\n\nResult brand and price:{result}')
    print(f'\n\nThat is the output')

    # result2 = extract_info(llm_result)
    # print(f'\n\nFrom Google llm:{result2}')

    return result

# url = "https://i.ibb.co/mNYvqDL/crop_39.jpg"
# object="fridge"

# get_name(url, object)