app.py

import gradio as gr
from gtts import gTTS
import pdfplumber
from pdf2image import convert_from_path
import pytesseract
import re
from pydub import AudioSegment

# 🔹 ভাষা detect
def detect_lang(text):
    if re.search(r'[\u0980-\u09FF]', text):
        return "bn"
    else:
        return "en"

# 🔹 Mixed text to speech
def mixed_text_to_speech(text):
    if not text or not text.strip():
        return None

    # বাংলা + ইংরেজি ব্লকে ভাগ
    chunks = re.findall(r'[\u0980-\u09FF\s,.!?]+|[a-zA-Z0-9\s,.!?]+', text)

    if not chunks:
        return None

    combined = AudioSegment.silent(duration=500)
    for i, chunk in enumerate(chunks):
        chunk = chunk.strip()
        if not chunk:
            continue
        lang = detect_lang(chunk)
        try:
            tts = gTTS(text=chunk, lang=lang)
            filename = f"chunk_{i}.mp3"
            tts.save(filename)
            audio = AudioSegment.from_file(filename, format="mp3")
            combined += audio + AudioSegment.silent(duration=300)
        except Exception as e:
            print(f"Skipping chunk due to error: {chunk} | {e}")

    output_path = "output.mp3"
    combined.export(output_path, format="mp3")
    return output_path

# 🔹 PDF থেকে text বের করা (OCR fallback)
def pdf_to_text(pdf_file):
    text = ""
    # প্রথমে pdfplumber দিয়ে চেষ্টা
    try:
        with pdfplumber.open(pdf_file.name) as pdf:
            for page in pdf.pages:
                page_text = page.extract_text()
                if page_text:
                    text += page_text + "\n"
    except:
        pass

    # যদি text ফাঁকা থাকে, OCR ব্যবহার কর
    if not text.strip():
        images = convert_from_path(pdf_file.name)
        for img in images:
            text += pytesseract.image_to_string(img, lang='ben+eng') + "\n"

    return text

def pdf_to_speech(pdf_file):
    if pdf_file is None:
        return None

    text = pdf_to_text(pdf_file)
    if not text.strip():
        return None

    return mixed_text_to_speech(text)

# 🔹 Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("## 🗣️ Mixed Language TTS (Bangla + English + PDF)")

    with gr.Tab("🔤 Text to Speech"):
        text_input = gr.Textbox(label="Enter Bangla + English text")
        text_output = gr.Audio(label="Generated Voice", type="filepath")
        text_btn = gr.Button("🎙️ Convert")
        text_btn.click(fn=mixed_text_to_speech, inputs=text_input, outputs=text_output)

    with gr.Tab("📘 PDF to Speech"):
        pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
        pdf_output = gr.Audio(label="Generated Voice", type="filepath")
        pdf_btn = gr.Button("📖 Convert PDF")
        pdf_btn.click(fn=pdf_to_speech, inputs=pdf_input, outputs=pdf_output)

demo.launch(show_error=True)