qplot

qplot is a command-line tool for quickly inspecting and plotting scientific data from CSV, TSV, TXT, DAT, NumPy, and Excel files. It detects tabular structure, chooses useful datasets and x/y columns, explains its inference, and can show a plot interactively or save it without a GUI.

Alpha status: qplot is usable, but its interface and automatic inference may change before the first stable release.

Installation

Python 3.10 or newer is required. The distribution is named qplot-cli, while the command it installs is qplot.

The recommended installation for command-line use is pipx:

pipx install qplot-cli

Alternatively, install it into the current Python environment:

python -m pip install qplot-cli

Confirm the command is available:

qplot --version
qplot --help

If Windows cannot find qplot after installation, use python -m qplot or ensure the Python Scripts directory reported by python -m site --user-base is on PATH.

Quick start

Inspect a file without plotting:

qplot data.csv --columns
qplot data.csv --info

Infer columns and open an interactive plot:

qplot data.csv
qplot experiment.npy
qplot results.npz
qplot workbook.xlsx
qplot coordinates.csv

Save a plot on a desktop, server, or CI runner without opening a window:

qplot data.csv --no-gui --save plot.png

Select columns by exact case-insensitive name or zero-based index:

qplot measurements.tsv -x time -y voltage -y temperature
qplot measurements.tsv -x 0 -y 1 -y 2
qplot coordinates.csv -x horizontal -y vertical -z depth

Apply plotting and loading overrides:

qplot data.csv --scatter --logx --logy --title "Field sweep"
qplot data.txt --delimiter ";" --header 2 --rows 100:500
qplot headerless.dat --no-header --all
qplot legacy-export.csv --encoding cp1252 --info
qplot data.csv --missing-y error
qplot instrument.csv --invalid-y error

--rows START:STOP uses zero-based, stop-exclusive slicing. --all plots all numeric y columns remaining after x selection. Run qplot --help for every option.

Text input defaults to utf-8-sig, which supports ordinary UTF-8 and UTF-8 files with a byte-order mark. Use --encoding for files produced with another encoding, such as cp1252 or latin-1.

Empty cells and the case-insensitive markers NA, N/A, NaN, and null are interpreted as missing values. By default, missing y values produce visible gaps in line plots and omitted markers in scatter plots, and qplot prints a warning for every affected y column. Use --missing-y drop to connect across missing points or --missing-y error to reject them. Missing x values are always rejected because they affect every plotted series. Unrecognised tokens such as ERROR are not silently treated as missing data. A mixed column is eligible for automatic y inference when at least half of its non-missing values are numeric; it is never considered as an automatic x column. Invalid y values produce gaps by default, with warnings that identify their row positions and tokens. Line plots containing gaps show point markers so isolated valid measurements remain visible at their true x positions. Use --invalid-y drop to remove those points or --invalid-y error for strict validation. Mixed columns below the 50% threshold remain excluded and are explained by --info; they can still be selected explicitly with -y.

Supported files and automatic behaviour

• CSV, TSV, TXT, and DAT files use automatic delimiter and header detection.
• One- and two-dimensional .npy arrays become tabular datasets.
• Each usable array in an .npz archive becomes a dataset.
• Each non-empty worksheet in an .xlsx workbook becomes a dataset.

For sources containing multiple datasets, qplot skips unplottable candidates, then prefers an inferred x column, more inferred y columns, and more rows. Ties preserve the source order. Empty or unsupported archive members and worksheets are skipped with visible warnings when usable data remains.

Inference deterministically scores x candidates using recognised names, monotonicity, uniqueness, and spacing. Y candidates use recognised response names and variation. Automatic plots include at most three y traces; --all removes that limit.

Exactly three numeric columns named x, y, and z (case-insensitively) are treated as strong evidence for an automatic 3D scatter plot. Ambiguous three-column data remains a normal 2D plot. A user can explicitly request a 3D scatter for differently named columns with -x X -y Y -z Z; names and zero-based indices are both accepted.

Platform notes

Headless plotting uses Matplotlib's Agg backend and does not require a display server. Interactive plotting uses a GUI backend available on the host system. Windows Python installations commonly include Tk support; some Linux systems need an additional package such as python3-tk. If interactive plotting is not available, use --no-gui --save OUTPUT.png.

Windows and Ubuntu are tested in continuous integration on Python 3.10, 3.12, and 3.14. Other Python versions from 3.10 onward are expected to work but are not included in the current CI matrix.

Development

Clone the repository and install it in editable mode with development tools:

python -m pip install -e ".[dev]"
python -m pytest

CI runs the test suite on Windows and Ubuntu and builds both a wheel and source distribution. See CONTRIBUTING.md before proposing a larger change.

Current limitations

• One input file is supported per invocation.
• Files are loaded eagerly into memory; text encoding defaults to utf-8-sig and can be overridden explicitly.
• Detection supports ordinary delimited tables and leading blank or comment lines; arbitrary instrument metadata and concatenated tables are not yet interpreted.
• NumPy arrays with three or more dimensions are reported and skipped. Current 3D scatter support applies to tabular x/y/z point data, not volume arrays, surfaces, or coordinate grids.
• Legacy .xls, HDF5, and Parquet files are not yet supported.
• Interactive plotting depends on a working local Matplotlib GUI backend.
• Qt, fitting, smoothing, presets, arbitrary expressions, and plugins are not currently included.

Licence

qplot is released under the MIT License.