app.py

import streamlit as st
from dolma.core.utils import split_paragraphs, split_sentences

# Title of the Streamlit app
st.title('Text Splitter: Paragraphs and Sentences')

# Text input from user
sample_text = st.text_area("Paste your text below", height=300)

if sample_text:
    # Split the text into paragraphs
    paragraphs = split_paragraphs(sample_text)

    # Split the text into sentences
    sentences = split_sentences(sample_text)

    # Show number of paragraphs and sentences
    st.write(f"Number of paragraphs: {len(paragraphs)}")
    st.write(f"Number of sentences: {len(sentences)}")

    # Create two columns for separate views
    col1, col2 = st.columns(2)

    # Display paragraphs in the left column
    with col1:
        st.header("Paragraphs")
        for i, paragraph in enumerate(paragraphs):
            st.subheader(f"Paragraph {i + 1}")
            st.write(paragraph.text)

    # Display sentences in the right column
    with col2:
        st.header("Sentences")
        for i, sentence in enumerate(sentences):
            st.subheader(f"Sentence {i + 1}")
            st.write(sentence.text)

else:
    st.write("Please paste your text to split it into paragraphs and sentences.")