app.py

import re
from flask import Flask, render_template, request
from asgiref.wsgi import WsgiToAsgi

# Create Flask app
app = Flask(__name__)

def process_ssml_text(ssml_text):
    """
    Process SSML text to extract content between SSML tags and format it properly.
    """
    # Regular expression to find all textnorm tags and their content
    result = []
    
    # Process each line separately
    lines = ssml_text.split('\n')
    
    for line in lines:
        # Create a new line by removing all SSML tags and just using the clean text
        if "<textnorm" in line:
            
            clean_line = re.sub(r'<textnorm[^>]*>([^<]*)</textnorm>', r'\1', line)
        
            # Replace specific symbols to match the expected output
            clean_line = clean_line.replace('A☼M', 'A※M').replace('P☼M', 'P※M')
        
            result.append(clean_line)
    
    return '\n'.join(result)


def process_old_ssml_text(ssml_text):
    """
    Process old SSML text format to extract and format content.
    """
    result = []
    
    # Process each line separately
    lines = ssml_text.split('\n')
    
    for line in lines:
        # Remove DECIMAL and CURRENCY tags, keeping the spoken text and any trailing content
        if "<" in line:
            clean_line = re.sub(r'<(?:DECIMAL|DATE02|NUMERIC_RANGE|CURRENCY|DATE|TIME|DURATION|TELEPHONE|EMAIL|URL|SYMBOL|ORDINAL|CARDINAL|MEASUREMENT|ROMAN|FRACTION|MATH_EXPRESSION|MATH_EQUATION|BANK_NUMBER|ADDRESS|ALPHA_NUMERIC|DIGIT|LONG_NUMBER|LATIN|PHONE|CONTACT|WEB|NUMERIC|ALPHABETIC|SPECIAL_CHAR|PUNCTUATION|SPACE|TAG_REPLACEMENT|LANGUAGE|LOCALE):\s*([^>]+)>\s*', r'\1 ', line)
        
            # Remove textnorm tags
            clean_line = re.sub(r'<textnorm[^>]*>([^<]*)</textnorm>', r'\1', clean_line)
        
            # Replace specific symbols to match the expected output
            clean_line = clean_line.replace('A☼M', 'A※M').replace('P☼M', 'P※M')
        
            # Strip extra whitespace
            clean_line = clean_line.strip()
        
            result.append(clean_line)
    
    return '\n'.join(result) 


@app.route('/', methods=['GET', 'POST'])
def index():
    processed_text = ""
    input_text = ""
    current_endpoint = "/"
    
    if request.method == 'POST':
        input_text = request.form.get('input_text', '')
        processed_text = process_ssml_text(input_text)
    
    return render_template('index.html', processed_text=processed_text, input_text=input_text, current_endpoint=current_endpoint)

@app.route('/old', methods=['GET', 'POST'])
def old_index():
    processed_text = ""
    input_text = ""
    current_endpoint = "/old"
    
    if request.method == 'POST':
        input_text = request.form.get('input_text', '')
        processed_text = process_old_ssml_text(input_text)
    
    return render_template('index.html', processed_text=processed_text, input_text=input_text, current_endpoint=current_endpoint)

if __name__ == '__main__':
    app.run(debug=True)

# Wrap Flask app with ASGI adapter for Uvicorn/Hugging Face Spaces compatibility
asgi_app = WsgiToAsgi(app)