app.py

import gradio as gr
from transformers import pipeline
import spacy
from textblob import TextBlob
from gradio_client import Client
import re

# Initialize models
nlp = spacy.load("en_core_web_sm")
spell_checker = pipeline("text2text-generation", model="oliverguhr/spelling-correction-english-base")

def preprocess_capitalization(text: str) -> str:
    """Preprocess input text to handle capitalization rules."""
    words = text.split(" ")
    processed_words = []
    
    for word in words:
        # Check if the word is an acronym (all uppercase letters)
        if re.match(r"^[A-Z]+$", word):
            processed_words.append(word)  # Leave acronyms unchanged
        # Check if the word has mixed capitalization (e.g., "HEllo")
        elif re.search(r"[A-Z]", word) and re.search(r"[a-z]", word):
            processed_words.append(word[0].upper() + word[1:].lower())  # Correct capitalization
        else:
            processed_words.append(word)  # Leave other words unchanged
    
    return " ".join(processed_words)

def preprocess_text(text: str):
    """Process text and return corrections with position information."""
    result = {
        "spell_suggestions": [],
        "entities": [],
        "tags": []
    }

    # Apply capitalization preprocessing
    capitalized_text = preprocess_capitalization(text)
    if capitalized_text != text:
        result["spell_suggestions"].append({
            "original": text,
            "corrected": capitalized_text
        })
        text = capitalized_text  # Update text for further processing

    # Find and record positions of corrections
    doc = nlp(text)
    
    # TextBlob spell check with position tracking
    blob = TextBlob(text)
    corrected = str(blob.correct())
    if corrected != text:
        result["spell_suggestions"].append({
            "original": text,
            "corrected": corrected
        })
    
    # Transformer spell check
    spell_checked = spell_checker(text, max_length=512)[0]['generated_text']
    if spell_checked != text and spell_checked != corrected:
        result["spell_suggestions"].append({
            "original": text,
            "corrected": spell_checked
        })

    # Add entities and tags
    result["entities"] = [{"text": ent.text, "label": ent.label_} for ent in doc.ents]
    result["tags"] = [token.text for token in doc if token.text.startswith(('#', '@'))]

    return text, result

def preprocess_and_forward(text: str):
    """Process text and forward to translation service."""
    original_text, preprocessing_result = preprocess_text(text)
    
    # Forward original text to translation service
    client = Client("Frenchizer/space_17")
    try:
        translation = client.predict(original_text)
        return translation, preprocessing_result
    except Exception as e:
        return f"Error: {str(e)}", preprocessing_result

# Gradio interface
with gr.Blocks() as demo:
    input_text = gr.Textbox(label="Input Text")
    output_text = gr.Textbox(label="Output Text")
    preprocess_button = gr.Button("Process")
    preprocess_button.click(fn=preprocess_and_forward, inputs=[input_text], outputs=[output_text])

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