reflex-rosencharts

A port of the rosencharts charting library (43 React/TSX components built with D3.js + Tailwind) to a custom component for Reflex, so the charts can be used from pure Python.

Status: scaffold + SDD plan. The charts are implemented in phases according to specs/plans/implementation-plan.md.

Why?

rosencharts is not an npm package: it's a set of .tsx components to copy and paste, with hardcoded data and Tailwind styling. This project wraps them as local Reflex components (rx.asset + library="$/public..."), parametrizes them to receive data from rx.State, and exposes them as Python functions:

import reflex as rx
import reflex_rosencharts as rxc

class State(rx.State):
    sales: list[dict] = [{"date": "2023-05-01", "value": 6}, {"date": "2023-05-02", "value": 8}]

def index() -> rx.Component:
    return rxc.line_chart(data=State.sales)

Charts (43 across 8 families)

Full catalog and props: specs/api/component-api-v1.md.

Axis options (scatter and line charts)

The scatter charts build both axes from the extent of the data, label them with regular ticks that line up with the grid lines, and reserve room for the dots so none is clipped at the edges — the data does not have to arrive sorted:

rxc.scatter_chart(
    data=State.repos,      # [{"revenue": x, "value": y, "company": label}, ...]
    x_scale="log",         # "linear" (default), "log" or "symlog"
    margin_left="46px",    # optional; by default it fits the longest y label
)

"log" needs strictly positive values and falls back to "symlog" when any value is <= 0. See CHANGELOG.md for what changed in 0.2.2.

Installation

pip install reflex-rosencharts
# or, with uv:
uv add reflex-rosencharts

Requires Python ≥ 3.13 and Reflex ≥ 0.10.0 (since 0.3.0; use reflex-rosencharts<0.3 on Reflex 0.9).

Project layout

This repo is a Reflex custom component:

custom_components/
└── reflex_rosencharts/          # the published package (import name: reflex_rosencharts)
    ├── __init__.py              # public API surface
    ├── reflex_rosencharts.py    # shared base wrapper (RosenChart)
    └── components/<family>/     # chart wrappers, ported in phases
reflex_rosencharts_demo/         # demo app (reflex run) showcasing the component
pyproject.toml                   # packaging metadata (where = ["custom_components"])

Development

Managed with uv:

uv sync --extra dev --group test   # install runtime + publishing + test tooling
uv run reflex run                  # start the demo app / gallery

Axis regression check for the scatter and line charts (server-renders them with clustered, unsorted and extreme-valued data and asserts the tick, margin and clipping guarantees). It uses the frontend dependencies Reflex installs on the first reflex run, plus esbuild (Reflex 0.10 builds with Vite 8 / rolldown and no longer installs it):

npm install --no-save --prefix scripts esbuild
node scripts/check_scatter_axes.mjs

Build & publish (custom component)

Reflex 0.10 removed reflex component build, so the build is two steps:

# 1. Generate the .pyi stubs (PYTHONPATH=. lets the generator resolve the
#    package under custom_components/).
PYTHONPATH=. uv run python -m reflex_base.utils.pyi_generator custom_components/reflex_rosencharts

# 2. Build the sdist + wheel into dist/.
uv build

# 3. Sanity-check the artifacts.
uv run twine check --strict dist/*

# 4. Publish to PyPI with uv (needs a PyPI API token):
uv publish --token "$PYPI_TOKEN"
# …or to TestPyPI first:
uv publish --publish-url https://test.pypi.org/legacy/ --token "$TEST_PYPI_TOKEN"

Automated release: bump version in pyproject.toml and __version__, add the CHANGELOG.md section, merge to main and push a vX.Y.Z tag. The Release workflow re-runs CI on the tag, builds, publishes to PyPI (secret PYPI_API_TOKEN in the pypi environment, or Trusted Publishing) and creates the GitHub release. An existing tag can be released from Actions → Release → Run workflow.

See Reflex's publishing prerequisites.

Specifications (SDD)

The project follows Spec-Driven Design. The specs are the primary artifact:

• PRD — the what and the who-for
• API Spec — public Python API (functions, props)
• Technical Design — wrapping pattern and architecture
• Data Model — per-chart data schemas
• Implementation Plan — phases by family

The original code (reference, read-only) lives in reference/rosencharts/.

License and attribution

MIT. This project is a port of rosencharts by Filsommer (MIT). See LICENSE and NOTICE.