import requests
import urllib.parse
import gradio as gr

API_URL = "https://api-inference.huggingface.co/models/ProsusAI/finbert"
headers = {"Authorization": "Bearer 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"}

# def query(payload):
# 	response = requests.post(API_URL, headers=headers, json=payload)
#     sentiment_prediction = response.json()

#     pos, neg, neutr = sentiment_prediction[:, 0].item(), sentiment_prediction[:, 1].item(), sentiment_prediction[:, 2].item()
#     sentiment_label = "Positive" if pos > neg and pos > neutr else "Negative" if neg > pos and neg > neutr else "Neutral"

# 	return sentiment_label
	
# output = query({
# 	"inputs": "I like you. I love you",
# })


def predict_sentiment(payload):
    # Sentiment Analysis
    response = requests.post(API_URL, headers=headers, json=payload)
    sentiment_prediction = response.json()
    
    # with torch.no_grad():
    #     sentiment_outputs = requests.post(API_URL, headers=headers, json=payload)
    #     sentiment_prediction = torch.nn.functional.softmax(sentiment_outputs.logits, dim=-1)

    for sentiment_prediction in response_json:
        if sentiment_prediction['label'] == 'positive':
            pos = sentiment_prediction['score']
        elif sentiment_prediction['label'] == 'neutral':
            neutr = sentiment_prediction['score']
        elif sentiment_prediction['label'] == 'negative':
            neg = sentiment_prediction['score']

    # pos, neg, neutr = sentiment_prediction[:, 0].item(), sentiment_prediction[:, 1].item(), sentiment_prediction[:, 2].item()
    sentiment_label = "Positive" if pos > neg and pos > neutr else "Negative" if neg > pos and neg > neutr else "Neutral"

    return sentiment_label



# Gradio Interface
iface = gr.Interface(
    fn=predict_sentiment,
    inputs=[gr.Textbox(lines=2, label="Financial Statement")],
    outputs=[
        gr.Textbox(label="Sentiment"),
    ],
    live=True,
    title="Financial Content Sentiment Analysis",
    description="Enter a financial statement to analyze its sentiment."
)

iface.launch()