app.py

import streamlit as st
import pysrt
import io
import re

# Bold specified language chunks in a given subtitle line
def bold_chunks_in_text(text, chunks):
    for phrase in chunks:
        # Use case-insensitive and exact matching
        pattern = re.compile(re.escape(phrase), re.IGNORECASE)
        text = pattern.sub(lambda m: f"<b>{m.group(0)}</b>", text)
    return text

# Main processing function
def process_subtitles(srt_content, german_lines, bold_chunks):
    subs = pysrt.from_string(srt_content.decode('utf-8'))
    
    if len(subs) != len(german_lines):
        st.error(f"Line mismatch: {len(subs)} subtitle entries vs {len(german_lines)} German lines.")
        return None

    for i, sub in enumerate(subs):
        original = sub.text.strip()
        german = german_lines[i].strip()

        original_bold = bold_chunks_in_text(original, bold_chunks)
        german_bold = bold_chunks_in_text(german, bold_chunks)

        sub.text = f"{original_bold}\n{german_bold}"

    output = io.StringIO()
    subs.write_into(output)
    return output.getvalue()

# ---------- Streamlit UI ----------
st.set_page_config(page_title="Subtitle Formatter", layout="centered")
st.title("🎬 Bilingual Subtitle Formatter for Language Learning")

st.markdown("Upload a `.srt` file, paste German translations, and highlight key phrases.")

uploaded_srt = st.file_uploader("📄 Upload Original Subtitle (.srt)", type=["srt"])
german_input = st.text_area("🇩🇪 German Translations (1 per subtitle line)", height=200)
chunks_input = st.text_input("🔍 Highlight Phrases (comma-separated)", placeholder="comment ça va, wie geht es dir")

if st.button("✨ Format Subtitles"):
    if not uploaded_srt or not german_input:
        st.warning("Please upload a subtitle file and enter German translations.")
    else:
        german_lines = [line.strip() for line in german_input.strip().splitlines()]
        bold_chunks = [chunk.strip() for chunk in chunks_input.split(",") if chunk.strip()]

        result_srt = process_subtitles(uploaded_srt.read(), german_lines, bold_chunks)
        
        if result_srt:
            st.success("✅ Subtitle processing complete!")
            st.download_button("📥 Download Formatted Subtitle", result_srt, "formatted_bilingual.srt", "text/plain")