app.py

import gradio as gr
from transformers import pipeline
import pandas as pd
import os
import re

# Load models
english_model = pipeline(
    "sentiment-analysis",
    model="siebert/sentiment-roberta-large-english"
)

urdu_model = pipeline(
    "sentiment-analysis",
    model="mrgmd01/SA_Model_bert-base-multilingual-uncased"
)

roman_urdu_model = pipeline(
    "sentiment-analysis",
    model="mrgmd01/SA_Model_bert-base-multilingual-uncased"  # replace with roman urdu model if available
)

# File to store only sentences
SAVE_FILE = "sentences.csv"

# Initialize file if not exists
if not os.path.exists(SAVE_FILE):
    df = pd.DataFrame(columns=["Sentence"])
    df.to_csv(SAVE_FILE, index=False)

# Detect language: English / Urdu / Roman Urdu
def detect_language(text):
    urdu_chars = set("ابتثجحخدذرزسشصضطظعغفقکلمنوہیءآؤئۀ")
    if any(ch in urdu_chars for ch in text):
        return "Urdu"
    roman_urdu_pattern = r"\b(hai|kia|kyun|nahi|bohot|acha|galat|sahi|parhai|ustad)\b"
    if re.search(roman_urdu_pattern, text.lower()):
        return "Roman Urdu"
    return "English"

# Normalize labels
def normalize_label(label):
    label = label.lower()
    if "positive" in label:
        return "Positive"
    elif "negative" in label:
        return "Negative"
    else:
        return "Neutral"

# Add emojis for polarity
def add_emoji(label):
    mapping = {
        "Positive": "😊",
        "Negative": "😞",
        "Neutral": "😐"
    }
    return f"{label} {mapping.get(label, '')}"

# Prediction + Save sentence
def analyze_single(text, lang_hint):
    if not text.strip():
        return "Please enter a sentence.", "", "", SAVE_FILE

    # Auto detect if user keeps default "English"
    lang = detect_language(text) if lang_hint == "English" else lang_hint

    if lang == "English":
        result = english_model(text)[0]
    elif lang == "Urdu":
        result = urdu_model(text)[0]
    else:
        result = roman_urdu_model(text)[0]

    sentiment = normalize_label(result["label"])
    sentiment_with_emoji = add_emoji(sentiment)
    score = round(result["score"], 3)

    # Save only the sentence
    df = pd.read_csv(SAVE_FILE)
    new_row = pd.DataFrame([[text]], columns=["Sentence"])
    df = pd.concat([df, new_row], ignore_index=True)
    df.to_csv(SAVE_FILE, index=False)

    return sentiment, str(score), sentiment_with_emoji, SAVE_FILE

# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown(
        "## 🌍 Multilingual Sentiment Analysis (Positive • Neutral • Negative)\n"
        "**Languages:** English, Urdu, Roman Urdu  \n"
        "📌 Models:  \n"
        "- `siebert/sentiment-roberta-large-english (English)`  \n"
        "- `mrgmd01/SA_Model_bert-base-multilingual-uncased (Urdu & Roman Urdu)`"
    )

    with gr.Tab("Sentiment Analyzer"):
        user_text = gr.Textbox(label="Enter text", placeholder="Type in English, Urdu, or Roman Urdu...")
        lang_dropdown = gr.Dropdown(["English", "Urdu", "Roman Urdu"], label="Language Hint", value="English")
        btn = gr.Button("🔍 Analyze")

        out_sent = gr.Textbox(label="Sentiment")
        out_conf = gr.Textbox(label="Confidence (0–1)")
        out_pol  = gr.Textbox(label="Polarity")
        out_file = gr.File(label="⬇️ Download Sentences (.csv)", type="filepath")

        btn.click(analyze_single, inputs=[user_text, lang_dropdown],
                  outputs=[out_sent, out_conf, out_pol, out_file])

if __name__ == "__main__":
    demo.launch()