streamlit-lexical-extended

A rich-text Markdown editor for Streamlit, powered by Lexical. It supports formatting, lists, quotes, code blocks, tables, undo/redo, programmatic updates, and multiple independent editors.

Version 0.3 uses package-based Streamlit Custom Components v2. The editor renders directly in Streamlit's DOM without an iframe.

Requirements

• Python 3.10+
• Streamlit 1.51+
• Node.js 24+ for development

Installation

pip install streamlit-lexical-extended

Quick start

The component returns a native Components v2 ComponentResult. Markdown is available as result.value. When key is set, Streamlit stores the same result object in st.session_state[key] before on_change runs.

import streamlit as st
from streamlit_lexical_extended import streamlit_lexical_extended


def editor_changed() -> None:
    markdown = st.session_state["editor"].value
    st.toast(f"Document updated: {len(markdown)} characters")


result = streamlit_lexical_extended(
    value="# Hello",
    key="editor",
    height=360,
    on_change=editor_changed,
)

st.markdown(result.value)
st.code(st.session_state["editor"].value, language="markdown")

The function accepts value, placeholder, height, min_height, debounce, key, on_change, width, and toolbar.

• value=None leaves the current editor content untouched; value="" clears it. Any other string replaces the current document.
• height=int uses a fixed height with internal scrolling. height=None grows with content and respects min_height.
• width accepts "stretch", "content", or a positive pixel integer.
• on_change receives no arguments.

Toolbar configuration

By default, toolbar=None displays every tool. Pass a list to display only the controls needed by a particular editor:

result = streamlit_lexical_extended(
    value="# Compact editor",
    key="compact_editor",
    toolbar=["undo", "redo", "block_type", "bold", "italic"],
)

An empty list hides the toolbar:

result = streamlit_lexical_extended(
    value="Keyboard-first editor",
    toolbar=[],
)

Available tool names:

Selected controls retain the editor's standard toolbar order.

Right-to-left applications

The editor automatically reads the computed direction of its Components v2 Shadow host. No component argument is required. Parent application styles such as the following switch the editor, toolbar, lists, quotes, tables, and table menus to RTL:

st.markdown(
    """
<style>
body, html {
    direction: RTL;
    unicode-bidi: bidi-override;
    text-align: right;
}
p, div, input, label, h1, h2, h3, h4, h5, h6 {
    direction: RTL;
    unicode-bidi: bidi-override;
    text-align: right;
}
</style>
""",
    unsafe_allow_html=True,
)

Use a unique key for each editor:

st.session_state["editor"].value  # Markdown string

Development

Install uv and Node.js 24, then run:

uv sync --extra dev
npm install --prefix streamlit_lexical_extended/frontend
npm run build --prefix streamlit_lexical_extended/frontend
uv run streamlit run streamlit_lexical_extended/example.py

For a clean, lockfile-based frontend install use npm ci. Frontend watch mode uses stable development asset names:

npm run dev --prefix streamlit_lexical_extended/frontend

Refresh the Streamlit page after a successful rebuild.

VS Code

Open the repository and select Streamlit: run example in Run and Debug. It creates/synchronizes .venv, installs frontend dependencies, builds the component, starts Streamlit, and opens the app.

Use Streamlit: develop frontend for frontend watch mode. Cmd/Ctrl+Shift+B runs the default quick verification. Additional tasks cover unit tests, E2E tests, package building, and package metadata checks.

Tests

uv run pytest tests -q
uv run playwright install chromium
uv run pytest streamlit_lexical_extended/tests/e2e -q

The E2E suite covers direct-DOM rendering, callbacks, focus, external updates, multiple editors, dimensions, formatting, themes, tables, and Markdown round-trips.

Publishing

1. Update the version in the root pyproject.toml, component pyproject.toml, and frontend package.json, then update CHANGELOG.md.
2. Run the VS Code tasks Project: verify, Tests: e2e, and Package: check.
3. Configure a PyPI Trusted Publisher for this repository, workflow publish.yml, and environment pypi.
4. Create and publish a GitHub release whose tag matches the version, such as v0.3.0.

The release workflow builds the frontend and Python distributions, validates the metadata, and publishes to PyPI using short-lived OIDC credentials.

See CONTRIBUTING.md for contributor setup and CHANGELOG.md for release history.

License

MIT