datadash

DataDash

A high-fidelity, interactive data visualization tool for your terminal.

DataDash turns streams or files of numbers into beautiful, interactive charts — no GUI, no browser, no cloud. Pipe in CSV, TSV, JSON, or hit an HTTP endpoint, and watch your data come to life with sub-cell braille resolution, per-series panels, live statistics, and full keyboard navigation.

✨ What's New in v0.5

Quick Start

# From source
go install github.com/keithknott26/datadash/cmd@latest

# Or download a binary from Releases:
# https://github.com/keithknott26/datadash/releases/latest

# Try it — no data needed
datadash demo

Installation

Pre-built binaries

Download from Releases — Linux, macOS, and Windows binaries for amd64 and arm64 are attached to every release.

Homebrew

brew install keithknott26/tap/datadash

From source

git clone https://github.com/keithknott26/datadash.git
cd datadash
go build -o datadash ./cmd
./datadash demo

Chart Types

DataDash supports three chart types via --chart (or -c):

Demos

Built-in demos

The fastest way to see DataDash in action — no input required.

datadash demo            # cycles through all generators (~30s each)
datadash demo metrics    # simulated server metrics (cpu, mem, rps, errors)
datadash demo sine       # multiple sine waves
datadash demo walk       # gaussian random walk (looks like a stock chart)
datadash demo anomaly    # baseline signal with random spikes
datadash demo pulse      # periodic pulses with exponential decay

Each demo also works with all chart types:

datadash demo metrics --chart bar
datadash demo metrics --chart sparkline

Streaming data (line chart)

Scrolls right, keeping the most recent data visible:

seq 4000 | awk 'BEGIN{OFS="\t"; print "x"}{x=$1/10; print cos(x); system("sleep 0.01")}' \
  | datadash --label-mode time --scroll

Streaming data (bar chart)

seq 4000 | awk 'BEGIN{OFS="\t"; print "x","sin(x)","cos(x)","rand(x)","rand(x)"}{x=$1/10; print x,sin(x),cos(x),rand(x),rand(x); system("sleep 0.02")}' \
  | datadash --chart bar

Streaming data (sparklines)

seq 4000 | awk 'BEGIN{OFS="\t"; print "x","sin(x)","cos(x)","rand(x)","rand(x)"}{x=$1/10; print x,sin(x),cos(x),rand(x),rand(x); system("sleep 0.02")}' \
  | datadash --chart sparkline

Tabular data with rolling average

seq 4000 | awk 'BEGIN{OFS="\t"; print "x","sin(x)","cos(x)","rand(x)","rand(x)","rand(x)"}{x=$1/10; print x,sin(x),cos(x),rand(x),rand(x),rand(x); system("sleep 0.02")}' \
  | datadash -a

JSON / NDJSON streams

Auto-discovery of numeric fields:

echo '{"cpu":45.2,"mem":72.1}
{"cpu":51.8,"mem":68.4}
{"cpu":39.1,"mem":75.0}' | datadash

Explicit field extraction with dot paths:

echo '{"host":"web1","metrics":{"cpu":45.2,"mem":72.1}}
{"host":"web1","metrics":{"cpu":51.8,"mem":68.4}}' \
  | datadash -f json --field metrics.cpu --field metrics.mem

HTTP polling

Watch live crypto prices:

datadash \
  --url 'https://api.coingecko.com/api/v3/simple/price?ids=bitcoin,ethereum,solana&vs_currencies=usd' \
  --field bitcoin.usd \
  --field ethereum.usd \
  --field solana.usd \
  --poll 30s

Watch weather data:

datadash \
  --url 'https://api.open-meteo.com/v1/forecast?latitude=37.77&longitude=-122.42&current=temperature_2m,wind_speed_10m,relative_humidity_2m' \
  --field current.temperature_2m \
  --field current.wind_speed_10m \
  --poll 60s

With authentication:

datadash \
  --url https://api.example.com/metrics \
  --header "Authorization: Bearer $TOKEN" \
  --field requests.rate --field errors.rate \
  --poll 5s

Input Methods

DataDash accepts data from four sources, in priority order:

1. datadash demo — built-in synthetic data
2. --url <endpoint> — HTTP polling
3. File argument — datadash data.txt
4. Stdin — cat data.txt | datadash

Data Format

DataDash auto-detects the format from the first line — TSV, CSV, pipe-delimited, JSON, NDJSON, whitespace, or single-column. Header rows are auto-detected and used as series names. Force a format with -f.

Single column (1 graph)

50
60
70

Multi-column (first column as label)

time	RowLabel1	RowLabel2
00:00	50	100
00:01	60	90
00:02	70	80

JSON / NDJSON

{"cpu": 45.2, "mem": 72.1, "errors": 0}
{"cpu": 51.8, "mem": 68.4, "errors": 2}

Decorated values

Currency symbols, percent signs, and thousands separators are stripped automatically:

$1,234.56	99%	€42.00

becomes 1234.56, 99, 42.

⌨️ Keyboard Controls

CLI Reference

Usage:
  datadash [file] [flags]
  datadash [command]

Available Commands:
  demo        Run a built-in demo dataset
  completion  Generate shell completions
  help        Help about any command

Flags:
  -d, --delimiter string         Column delimiter (auto-detected if omitted)
  -f, --format string            Input format: auto, csv, tsv, json, space, single (default "auto")
  -m, --label-mode string        X-axis label source: first, time, none (default "first")
      --field strings            JSON field to extract (dot path, repeatable)

  -c, --chart string             Chart type: line, bar, sparkline (default "line")
      --downsample string        Downsampling: lttb, minmax, none (default "lttb")
      --high-fidelity            Maximize braille sub-cell resolution (default true)

  -s, --scroll                   Scroll mode — keep newest data on screen
  -a, --average-line             Show a rolling average overlay (line chart only)
  -z, --average-seek int         Rolling average window size (default 500)
  -r, --redraw-interval duration Screen redraw interval (default 100ms)
  -l, --seek-interval duration   Input line read interval (default 20ms)

      --url string               Poll an HTTP endpoint instead of reading stdin/file
      --poll duration            HTTP poll interval (default 5s)
      --header strings           HTTP header, 'Key: Value' (repeatable)

      --debug                    Enable debug logging to stderr
  -v, --version                  Print version and exit

Shell completions

# Bash
datadash completion bash | sudo tee /etc/bash_completion.d/datadash

# Zsh
datadash completion zsh > "${fpath[1]}/_datadash"

# Fish
datadash completion fish > ~/.config/fish/completions/datadash.fish

Why DataDash Looks Sharp

Braille canvas. termdash renders lines using Unicode braille characters, giving 2×4 sub-pixel resolution per character cell. A 200×40 terminal effectively has a 400×160 drawing surface.

LTTB downsampling. When you have 100,000 data points and only 200 cells of space, DataDash uses the Largest Triangle Three Buckets algorithm — the same one Grafana and Plotly use — to pick the points that preserve the curve's visual character.

Architecture

datadash/
├── ring.go           # atomic ring buffer with stats + percentiles
├── row.go            # input parsing & format auto-detection
├── uniq.go           # generic deduplication
├── downsample.go     # LTTB & MinMax decimation
├── http_source.go    # HTTP polling reader
├── cmd/
│   ├── datadash.go   # main app: layout, keybinds, refresh loop
│   └── demo.go       # demo subcommand & data generators
└── tools/sampledata/ # bundled sample datasets

The library code is consumable as a Go package:

import "github.com/keithknott26/datadash"

ring := datadash.NewRing(10000)
ring.Push(42.5)
stats := ring.Stats()
sampled := datadash.LTTB(rawData, 1000)

Testing

go test -v -race -coverprofile=coverage.out ./...
go tool cover -html=coverage.out
go test -bench=. -benchmem ./...

Similar Projects

• termeter — the original inspiration
• ttyplot — minimal real-time terminal plotting
• asciigraph — ASCII charts as a library
• grafterm — terminal-based Grafana

Acknowledgments

Built on termdash. CLI powered by cobra.

License

MIT — see LICENSE.