Krishna086's picture
Update app.py
6c6d6f8 verified
raw
history blame
7.95 kB
import streamlit as st
import importlib
from io import BytesIO
import docx
from PyPDF2 import PdfReader
st.set_page_config(page_title="Multilingual Translator", page_icon="🌐", layout="centered")
# Import LANGUAGES from translation.py
try:
from translation import LANGUAGES
except ImportError as e:
st.error(f"Failed to import translation module: {e}")
st.stop()
# Function to extract text from uploaded files (PDF, DOCX, TXT)
def extract_text_from_file(uploaded_file):
try:
if uploaded_file.type == "application/pdf":
pdf_reader = PdfReader(uploaded_file)
text = ""
for page in pdf_reader.pages:
text += page.extract_text() or ""
return text.encode().decode('utf-8', errors='ignore').strip()
elif uploaded_file.type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
doc = docx.Document(uploaded_file)
text = "\n".join([para.text for para in doc.paragraphs])
return text.encode().decode('utf-8', errors='ignore').strip()
elif uploaded_file.type == "text/plain":
return uploaded_file.read().decode('utf-8', errors='ignore').strip()
return ""
except Exception:
return ""
# Callback to update input text when file is uploaded
def on_file_upload():
uploaded_file = st.session_state.file_input
if uploaded_file and uploaded_file.size < 1024*1024:
st.session_state.user_input_text = extract_text_from_file(uploaded_file)
elif uploaded_file and uploaded_file.size >= 1024*1024:
st.error("File size must be less than 1 MB")
# Main application function
def main():
try:
translation_module = importlib.import_module("translation")
language_detector = importlib.import_module("lang_detect")
audio_processor_module = importlib.import_module("audio_processor")
# Header
st.markdown("<h1 style='text-align: center; color: #4285F4;'>Multilingual Translator</h1>", unsafe_allow_html=True)
st.markdown("<p style='text-align: center; color: #666;'>Effortless Multilingual Translation</p>", unsafe_allow_html=True)
# Custom CSS to hide size limit and show only file types
st.markdown(
"""
<style>
.stFileUploader > div > div > div[role="button"] {
display: none !important;
}
.stFileUploader label {
display: block !important;
color: #666 !important;
font-size: 0.8em !important;
}
.stFileUploader label::after {
content: 'TXT, DOCX, PDF' !important;
}
</style>
""",
unsafe_allow_html=True
)
# Language and Input/Output Layout with symmetric columns
left_col, right_col = st.columns([1, 1]) # Equal width for symmetric layout
with left_col:
detected_options = language_detector.detect_language(st.session_state.get("user_input_text", "")) if st.session_state.get("user_input_text", "").strip() else [("Auto-detect", 1.0, "Auto-detect")]
source_language = detected_options[0][2] if detected_options[0][0] != "Auto-detect" else "Auto-detect"
source_lang_code = next((k for k, v in LANGUAGES.items() if v[1] == source_language), "hi") if source_language != "Auto-detect" else "auto"
source_options = ["Auto-detect"] + [f"{v[0]} ({v[1]})" for v in LANGUAGES.values()]
st.selectbox("Source Language", options=source_options, index=0 if source_language == "Auto-detect" else source_options.index(f"{LANGUAGES[source_lang_code][0]} ({source_language})"), key="source_lang")
user_input_text = st.text_area("Input Text", height=400, key="user_input_text", placeholder="Enter text here", label_visibility="hidden")
input_type = st.radio("Input Type", ["Text", "File"], horizontal=True, label_visibility="hidden", key="input_type")
if input_type == "File":
st.file_uploader("Upload File", type=["txt", "docx", "pdf"], key="file_input", on_change=on_file_upload, label_visibility="hidden")
if st.session_state.get("file_input") and st.session_state.get("file_input").size >= 1024*1024:
st.error("File size must be less than 1 MB")
st.button("Translate", key="translate_btn", on_click=trigger_translation, args=(translation_module, language_detector, audio_processor_module))
with right_col:
source_lang_display = st.session_state.source_lang.split(" (")[0] if " (" in st.session_state.source_lang else st.session_state.source_lang
target_options = [f"{v[0]} ({v[1]})" for v in LANGUAGES.values() if v[0] != source_lang_display and v[1] != source_lang_display]
st.selectbox("Target Language", options=target_options, index=target_options.index(f"{LANGUAGES['en'][0]} ({LANGUAGES['en'][1]})") if "English" not in source_lang_display else 0, key="target_lang")
if "translated_text" in st.session_state:
st.text_area("Output Text", value=st.session_state.translated_text, height=400, key="output_text", disabled=True, label_visibility="hidden")
# Play audio button and playback below output
if st.button("πŸ”Š", key="audio_btn", on_click=play_audio, args=(audio_processor_module,), help="Play audio", use_container_width=False):
pass
# Footer
if "translated_text" in st.session_state:
st.markdown("""
<p style="font-size: small; color: grey; text-align: center;">
Developed By: Krishna Prakash
<a href="https://www.linkedin.com/in/krishnaprakash-profile/" target="_blank">
<img src="https://img.icons8.com/ios-filled/30/0077b5/linkedin.png" alt="LinkedIn" style="vertical-align: middle; margin: 0 5px;"/>
</a>
</p>
""", unsafe_allow_html=True)
except Exception as e:
st.error(f"App error: {e}")
# Function to trigger translation process with progress indicator
def trigger_translation(translation_module, language_detector, audio_processor_module):
user_input_text = st.session_state.get("user_input_text", "").strip()
if user_input_text:
with st.spinner("Translating..."):
source_lang = st.session_state.source_lang.split(" (")[0] if " (" in st.session_state.source_lang else st.session_state.source_lang
target_lang = st.session_state.target_lang.split(" (")[0] if " (" in st.session_state.target_lang else st.session_state.target_lang
if source_lang == "Auto-detect":
detected_options = language_detector.detect_language(user_input_text)
source_lang_code = next((k for k, v in LANGUAGES.items() if v[1] == detected_options[0][0]), "hi")
else:
source_lang_code = next((k for k, v in LANGUAGES.items() if v[0] == source_lang), "hi")
target_lang_code = next((k for k, v in LANGUAGES.items() if v[0] == target_lang), "en")
translated_text = translation_module.translate(user_input_text, source_lang_code, target_lang_code)
st.session_state.translated_text = translated_text or user_input_text
# Function to handle audio playback
def play_audio(audio_processor_module):
if "translated_text" in st.session_state and st.session_state.translated_text:
target_lang = next((k for k, v in LANGUAGES.items() if v[0] == st.session_state.target_lang.split(" (")[0]), "en")
audio_data = audio_processor_module.text_to_speech(st.session_state.translated_text, target_lang)
if audio_data and audio_data.getbuffer().nbytes > 0:
st.audio(audio_data, format="audio/mp3")
if __name__ == "__main__":
main()