---

Overview

dash-nle-timeline brings a purpose-built video-editor timeline to Plotly Dash. A calendar/scheduler widget thinks in dates; a video editor needs an integer frame grid, a real draggable playhead, snapping to clip edges / markers / grid, audio waveforms, and edits expressed as data. This library is exactly that — written in TypeScript on React 18 and compiled with dash-component-boilerplate, so everything is driven from pure Python.

• Frame-accurate timeline — integer frames are the canonical unit; seconds are derived from fps
• Full editing surface — drag to move (cross-track), drag edges to trim/resize, split at the playhead, delete; every edit reported as a single lastEdit event
• Audio waveforms — painted from precomputed peaks (minmax / magnitude); no audio decoding in the browser
• Snapping — to clip edges, the playhead, markers, and the frame grid, with a pixel-space threshold for a constant feel at any zoom
• Playback & transport — rAF playback loop, built-in transport bar with timecode, keyboard shortcuts, and an imperative command prop
• Scene compositor — a hybrid DOM + Canvas preview (DashNleScene) that renders any Dash component as its background and overlays image/video/text layers with per-layer visibility windows
• Control plane / media plane split — only small JSON crosses Dash callbacks; your media stays in a <video>, an audio engine, or a server render, driven off the timeline's outputs

The package ships the compiled JS bundle — a normal pip install needs no Node.

Installation

pip install dash-nle-timeline

Quick Start

import dash_nle_timeline as nle
from dash import Dash, Input, Output, callback, html

app = Dash(__name__)

app.layout = html.Div([
    nle.DashNleTimeline(
        id="tl", fps=30, duration=300,
        tracks=[
            {"id": "v1", "kind": "video", "label": "V1"},
            {"id": "a1", "kind": "audio", "label": "A1"},
        ],
        clips=[
            {"id": "c1", "trackId": "v1", "start": 0, "duration": 90, "label": "intro.mp4"},
            {"id": "c2", "trackId": "a1", "start": 0, "duration": 220, "label": "music.wav",
             "peaks": [...], "peaksFormat": "minmax"},
        ],
    ),
    html.Pre(id="out"),
])

@callback(Output("out", "children"), Input("tl", "playhead"))
def show(frame):
    return f"frame {frame}"

if __name__ == "__main__":
    app.run(debug=True)

Edits (move / trim / resize / split / delete) persist on their own — the component applies them to its optimistic clips state and reports each one as a single lastEdit event. Write clips back from a callback only when you want to validate or override.

Documentation

Full documentation, with live examples, lives at the open-source documentation index maintained by Pip Install Python LLC:

📚 pip-install-python.com

You can also run the bundled demo locally:

pip install -r requirements.txt
pip install -e .
python usage.py           # open http://127.0.0.1:8060

usage.py exercises the full surface: scrubbing, selection, move/trim/resize/split/delete, waveforms from precomputed peaks, the imperative command prop, playback, and the scene preview synced to the playhead.

Components

Component	What it is
DashNleTimeline	Multi-track, frame-accurate NLE timeline: playhead, playback, editing, snapping, waveforms, markers, keyboard shortcuts, and an imperative command channel.
DashNleScene	Hybrid DOM + Canvas preview compositor: renders a background (any Dash component and/or a native XYZ tileset) plus overlay image/video/text layers at the current playhead, with a screen-space camera.

Both load from a single bundled JS file — no extra external_scripts, CSS, or React setup.

Wire the two together with a plain callback:

@callback(Output("scene", "playhead"), Input("tl", "playhead"))
def sync(frame):
    return frame

The data boundary

• Frames are canonical. fps and duration (in frames) define the grid; playhead is an integer frame that round-trips both ways (scrub in the browser, or set it from Python).
• tracks is a list of dicts — {"id", "kind": "video" | "audio" | "camera", "label", "color"}.
• clips is a list of dicts — id, trackId, start, duration, plus optional inPoint, label, color, thumbnails, locked, and (audio) peaks / peaksFormat. It is both an input and optimistic round-trip state: user edits persist without any callback.
• lastEdit (output) reports each user edit as one event — move, trim, resize, split, delete — so Python can validate, override, or mirror it into your own model.
• selectedClipIds / selectedClipId round-trip; drive selection from Python too.
• command (input) is the imperative channel — play, pause, seek, splitAtPlayhead, copy, paste, duplicate, selectAll, zoomToFit, … — de-duplicated by id.

DashNleTimeline in depth

• Time — fps, duration (frames), playhead (round-trips), playing, loop; rAF playback loop with a throttled playhead round-trip.
• View — pixelsPerSecond (zoom), scrollX (pan); Ctrl/⌘ + wheel zooms anchored at the cursor.
• Editing — drag to move (cross-track), drag edges to trim/resize, S to split at the playhead, Delete to delete; snapping via snapping / snapThreshold.
• Waveforms — precomputed peaks painted on a per-lane canvas; formats minmax (interleaved [min0, max0, …]) and magnitude.
• Markers & keyframes — markers ({"id", "frame", "label", "color"}) on the ruler; camera-keyframe diamonds on camera tracks via cameraKeyframes.
• Keyboard — the keyboard prop (default on) toggles all shortcuts: Space, arrows (Shift = 1s), S split, Delete, ⌘/Ctrl + C/V/D copy/paste/duplicate, ⌘/Ctrl + A select all, Home/End, + / - zoom.
• Transport bar — the controls prop (default on): to-start / step / play-pause / step + timecode above the ruler.
• Theming — theme="dark" / "light", plus style / className passthrough.

API reference

DashNleTimeline props (selected)

Prop	Type	Description
fps / duration	int	Frame rate and total length in frames — the canonical time grid
playhead	int	Current frame — input and output
playing / loop	bool	Playback state (rAF loop inside the component)
tracks	list	Track dicts (video / audio / camera)
clips	list	Clip dicts — input and optimistic round-trip state
lastEdit	output	The most recent user edit as a single event
selectedClipIds / selectedClipId	list / str	Selection — round-trips both ways
markers	list	Ruler markers {"id", "frame", "label", "color"}
cameraKeyframes	list	Keyframe diamonds rendered on camera tracks
snapping / snapThreshold	bool / int	Snap to edges / playhead / markers / grid; pixel-space threshold
pixelsPerSecond / scrollX	number	Zoom and horizontal pan
command	dict	Imperative channel (seek, togglePlay, zoomToFit, …), de-duped by id
keyboard / controls	bool	Toggle shortcuts / the built-in transport bar
theme / style / className	—	Appearance and sizing

DashNleScene props (selected)

Prop	Type	Description
playhead	int	Frame to render — drive it from the timeline
resolution	list	Design-space [width, height] of the scene
children	Dash components	Background slot — a map, a plot, a video feed, anything
tileset	dict	Native XYZ tile background
layers	list	Overlay layers (image / video / text) with [start, end) visibility windows
camera	dict	Screen-space camera (pan / zoom over the scene)

The full prop tables are generated into dash_nle_timeline/DashNleTimeline.py and dash_nle_timeline/DashNleScene.py docstrings by dash-generate-components.

Development

# Install dependencies
npm install                  # TypeScript + webpack toolchain
pip install -r requirements.txt

# Build the JS bundle + regenerate the Python wrappers
npm run build                # webpack UMD bundle + dash-generate-components → dash_nle_timeline/*.py
npm run watch                # JS-only rebuild loop (when prop signatures don't change)

# Standalone smoke test
python usage.py              # http://127.0.0.1:8060

# Build a distribution
python -m build

• TypeScript source of truth: src/ts/ — components/ holds the two public components, internal/ the subcomponents, and time.ts / snap.ts / waveform.ts / edits.ts / tiles.ts are pure helpers.
• The built bundle and generated wrappers are committed, so pip install -e . works without npm.
• dash_nle_timeline/__init__.py is hand-maintained (it registers the UMD bundle) — dash-generate-components does not regenerate it.
• Keep the version in package.json and pyproject.toml in sync when cutting a release.

Requirements

• Python >= 3.8
• Dash >= 4.1
• Node.js >= 16 (for development / rebuilding the JS bundle only)

Community & support

Come build with us:

• 💬 Discord — discord.gg/WEnZR35mrK
• ▶️ YouTube — @2plotai
• 🐛 Issues — github.com/pip-install-python/dash-nle-timeline/issues

More from Pip Install Python LLC

dash-nle-timeline is one of several tools built and maintained by Pip Install Python LLC:

Project	What it is
📚 Pip Install Python	Open-source documentation index for the Python & Dash ecosystem
🔀 PiratesBargain.com	E-commerce / digital commerce
🧠 ai-agent.buzz	Infinite AI canvas
🎬 2plot.media	Videography application

License

MIT — see LICENSE. Built by Pip Install Python LLC to bring a real video-editing timeline into the Dash framework.