app.py

import gradio as gr
import numpy as np
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity

# Load the new model and tokenizer
context_model = SentenceTransformer("all-MiniLM-L6-v2")

# Define your labels
labels = [
    "aerospace", "anatomy", "anthropology", "art", 
    "automotive", "blockchain", "biology", "chemistry", 
    "cryptocurrency", "data science", "design", "e-commerce",
    "education", "engineering", "entertainment", "environment",
    "fashion", "finance", "food commerce", "general",
    "gaming", "healthcare", "history", "html",
    "information technology", "IT", "keywords", "legal",
    "literature", "machine learning", "marketing", "medicine",
    "music", "personal development", "philosophy", "physics",
    "politics", "poetry", "programming", "real estate", "retail", 
    "robotics", "slang", "social media", "speech", "sports",
    "sustained", "technical", "theater", "tourism", "travel"
]

# Pre-compute label embeddings
label_embeddings = context_model.encode(labels)

def detect_context(input_text, top_n=3, score_threshold=0.05):
    # Encode input text
    input_embedding = context_model.encode([input_text])
    
    # Compute cosine similarity with labels
    similarities = cosine_similarity(input_embedding, label_embeddings)[0]
    
    # Pair labels with scores
    label_scores = [(label, score) for label, score in zip(labels, similarities)]
    
    # Sort by score and filter by threshold
    sorted_labels = sorted(label_scores, key=lambda x: x[1], reverse=True)
    filtered_labels = [label for label, score in sorted_labels if score > score_threshold]
    
    # Return top N contexts
    return filtered_labels[:top_n] if filtered_labels else ["general"]

# Translation client for space_3
from gradio_client import Client
translation_client = Client("Frenchizer/space_3")  # Replace with your Space name

def translate_text(input_text):
    # Call the translation model
    result = translation_client.predict(input_text)
    return result

def process_request(input_text):
    # Detect context
    context = detect_context(input_text)
    print(f"Detected context: {context}")
    
    # Translate text
    translation = translate_text(input_text)
    return translation

# Create a Gradio interface
interface = gr.Interface(
    fn=process_request,
    inputs="text",
    outputs="text",
    title="Frenchizer",
    description="Translate text from English to French with context detection."
)

# Launch the Gradio app
interface.launch()