pubify-mpl 1.0.4

Publication-oriented Matplotlib figure export and LaTeX layout helpers

pubify-mpl

pubify-mpl is a small Python tool for exporting Matplotlib figures at sizes that match your LaTeX document, so the figures drop cleanly into papers, theses, and proceedings without trial-and-error resizing.

It combines two parts of a publication workflow:

• Python helpers for exporting Matplotlib figures using document-aware sizes
• a LaTeX package, pubify.sty, for arranging exported figures into common panel layouts

This package is meant for researchers who already use Matplotlib and LaTeX and want a cleaner path from Python plots to publication-ready figures. It is not a replacement for Matplotlib. It sits at the export and layout stage.

See CHANGELOG.md for release history and user-visible changes, and CONTRIBUTING.md for contributor and release workflow guidance.

Requirements

• Python 3.10+
• Matplotlib and NumPy
• a working LaTeX installation

save_fig() uses text.usetex=True, so LaTeX must be available when you export figures from Python.

On macOS, a common choice is MacTeX. If you want a smaller install, BasicTeX can also work, but you may need to add missing packages yourself.

LaTeX package requirements:

• pubify.sty depends on graphicx, subcaption, and etoolbox
• the staged gallery/debug workspace also uses lipsum, so minimal TeX installs may need that package separately

How It Works

Technically, save_fig(...) does not modify your original Matplotlib figure in place. Instead it:

1. makes a copy of the figure
2. applies publication styling and any requested cleanup only to that copy
3. resizes the copy to match the requested width or named LaTeX layout
4. saves the copy to PDF

That means your original interactive figure stays unchanged in Python. This is important if you want to keep using the same figure object in a notebook or script after export.

On the export side, save_fig(...) can:

• remove titles, labels, annotations, grids, or colorbars from the exported copy
• scales your copied figure to match the named layout
• use LaTeX text rendering during export so fonts and math match the document more closely

Quick Start

Install the package:

pip install pubify-mpl

Export a figure into a LaTeX project:

import matplotlib.pyplot as plt

from pubify_mpl import prepare, save_fig

paper_dir = "~/my-paper"
figures_dir = "~/my-paper/figures"

template = {
    # Values for the default LaTeX article class example.
    "textwidth_in": 5.39643,
    "textheight_in": 7.58960,
    "base_fontsize_pt": 12.0,
    "caption_lineheight_pt": 13.6,
    "subcaption_lineheight_pt": 13.6,
}

fig, ax = plt.subplots()
ax.plot([0, 1], [0, 1])
ax.set_xlabel("x")
ax.set_ylabel("y")

prepare(paper_dir, template=template)
save_fig(fig, "onewide", f"{figures_dir}/plot.pdf", template=template)

Then in LaTeX:

\usepackage{pubify}

Note that prepare(...) generates pubify-template.tex, and it contains specs that pubify.sty uses on the LaTeX side. Keep pubify-template.tex next to pubify.sty. Your paper directory will usually look like:

main.tex
pubify.sty
pubify-template.tex
figures/plot.pdf

If you do not already know your document dimensions and typography settings, first run prepare(...) so pubify.sty is available in the LaTeX project. Then add \figprintlayoutspec to the LaTeX document once and compile it. That prints the current textwidth, textheight, base font size, and caption/subcaption line heights, which you can then copy into the Python template dictionary.

pubify-mpl is fully compatible with Overleaf. If your LaTeX project is synced locally, the same folder layout works directly. If you do not have local sync set up, copy pubify.sty, pubify-template.tex, and the exported figure PDFs into the Overleaf project manually.

Examples

• examples/quickstart.ipynb: notebook version of the minimal quick-start example, writing into the tracked example project at examples/tex/main.tex
• gallery/layout-gallery-examples.tex: sample LaTeX source showing the supported panel layouts and macro usage

Layouts

See layout options in gallery/layout-gallery.pdf including:

• "one": one large panel
• "onewide": one short wide panel
• "two": two stacked panels
• "twowide": two side-by-side panels
• "three": three stacked panels
• "threewide": three side-by-side panels
• "four": 2x2 grid
• "six": 2x3 grid
• "sixwide": 3x2 grid
• "nine": 3x3 grid
• "twelve": 3x4 grid
• "twelvewide": 4x3 grid
• "fifteen": 3x5 grid
• "sixteen": 4x4 grid
• "twenty": 4x5 grid

Typical Workflow

1. Run prepare(...) to install pubify.sty and write an initial pubify-template.tex.
2. Measure your real LaTeX document using \figprintlayoutspec.
3. Update the template with the values printed by \figprintlayoutspec.
4. Export your figures with save_fig(...).
5. Include them in LaTeX with \figfloat, \fig, and the layout macros.

Advanced Python Usage

This is an alternative way to use pubify-mpl when you export many figures for the same document or keep several document templates in Python.

Define a common set of templates:

PUBIFY_TEMPLATES = {
    "article": {
        "textwidth_in": 5.39643,
        "textheight_in": 7.58960,
        "base_fontsize_pt": 12.0,
        "caption_lineheight_pt": 13.6,
        "subcaption_lineheight_pt": 13.6,
    },
    "thesis": {
        "textwidth_in": 6.5,
        "textheight_in": 8.5,
        "base_fontsize_pt": 11.0,
        "caption_lineheight_pt": 13.0,
        "subcaption_lineheight_pt": 13.0,
    },
}

Then when you export many figures for the same document, use_template(...) lets you set a default template inside a with block:

import matplotlib.pyplot as plt
from pubify_mpl import prepare, save_fig, use_template

paper_dir = "~/mythesis"
figures_dir = "~/mythesis/figures"

# do this once
prepare(paper_dir, template=PUBIFY_TEMPLATES["thesis"])

# create your figures

with use_template(PUBIFY_TEMPLATES["thesis"]):
    save_fig(fig1, "one", f"{figures_dir}/plot-1.pdf")
    save_fig(fig2, "twowide", f"{figures_dir}/plot-2.pdf")

For figure construction that needs publication styling at creation time, pubify_rc_context(...) exposes the construction-time publication rc subset used for font and text defaults:

from pubify_mpl import pubify_rc_context

with pubify_rc_context(PUBIFY_TEMPLATES["thesis"]):
    fig = build_custom_plot()

The intended styling flow is:

• build under pubify_rc_context(...) when artist creation depends on Matplotlib rc defaults
• let save_fig(...) run its full export-time TeX setup plus normal generic cleanup and normalization afterward
• use prepare_export(...) only for figure-specific artists that pubify still cannot reach generically

Lower-level figure adjustment helpers and prepare_export(...) are available in the Python API reference for those final figure-specific cases.

For figure-specific styling that pubify cannot discover generically, prepare_export may accept a second style argument:

def prepare_export(fig_export, style):
    for text in iter_custom_tick_labels(fig_export.axes[0]):
        text.set_fontfamily(style.font_family)
        text.set_fontsize(style.tick_labelsize_pt)

One-argument callbacks still work. The optional style payload carries the resolved publication styling values for text, lines, ticks, and spines.

Template Keys

The template dictionary tells pubify-mpl how much space is available in your LaTeX document and what spacing pubify.sty should use around figure rows, subcaptions, and captions. Exported figures use the standard LaTeX serif face; the template controls typography sizes, not font family.

Key	Meaning	Default
textwidth_in	document \textwidth in inches	required
textheight_in	document \textheight in inches	required
caption_lineheight_pt	measured line height used to estimate main caption height	13.6pt
subcaption_lineheight_pt	measured line height used to estimate subcaption height	13.6pt
base_fontsize_pt	base font size used when styling the exported figure copy	12pt
axes_labelsize_pt	axis-label font size; when unset, defaults to base_fontsize_pt, and -1 leaves existing axis-label sizes unchanged	base_fontsize_pt
tick_labelsize_pt	tick-label font size; when unset, defaults to base_fontsize_pt - 1, and -1 leaves existing tick-label sizes unchanged	base_fontsize_pt - 1
legend_fontsize_pt	legend font size; when unset, defaults to base_fontsize_pt - 1, and -1 leaves existing legend sizes unchanged	base_fontsize_pt - 1
title_fontsize_pt	axes-title font size; when unset, defaults to base_fontsize_pt + 1, and -1 leaves existing title sizes unchanged	base_fontsize_pt + 1
line_width_pt	stroke width for plotted lines and compatible collections; -1 leaves existing line widths unchanged	-1pt
axes_line_width_pt	stroke width for axes spines and tick marks; -1 leaves existing axes and tick stroke widths unchanged	0.8pt
tick_length_pt	tick length; -1 leaves existing tick lengths unchanged	3.0pt
caption_allowance_in	extra buffer added beyond the estimated main caption text height	0.08in
subcaption_allowance_in	extra buffer added beyond the estimated subcaption text height	0.08in
single_row_layout_max_height_in	maximum total layout height budget used for single-row layouts such as "onewide" or "twowide"	textheight_in / 3
subcaption_skip_in	vertical space between a panel and its subcaption	0.08in
row_skip_in	vertical space between rows in stacked layouts	0.11in
caption_skip_in	vertical space between the figure body and the main caption	0.11in
post_caption_skip_in	additional vertical space after the main caption	0in
col_gap_in	horizontal space between columns	0.02 * textwidth_in

For the best match to a real document, copy textwidth_in, textheight_in, base_fontsize_pt, caption_lineheight_pt, and subcaption_lineheight_pt from \figprintlayoutspec.

Most Common save_fig() Options

Some save_fig() options control how the exported figure fits into the LaTeX layout:

• caption_lines=...: estimate how many lines the main caption will use
• subcaption_lines=...: estimate how many lines each subcaption will use
• force_width=...: force a smaller export width, as long as it still fits inside the selected layout
• force_aspect=...: force a specific aspect ratio for the exported copy

Other options let you simplify the exported figure content without changing the original figure in Python:

• hide_labels=True: remove axis labels from the exported copy
• hide_grid=True: disable the grid on the exported copy
• hide_cbar=True: remove a colorbar when exporting a single-axes panel
• hide_annotations=True: remove ax.text(...) annotations from the exported copy

LaTeX Macros

pubify.sty separates LaTeX figure layout into three pieces:

• \figfloat[placement]{body}[caption][label] creates the outer floating figure environment and adds the main caption and label.
• the layout macros such as \figonewide, \figtwowide, and \figfour arrange one or more \fig{...} panels into a specific layout.
• \fig{file}[subcaption][label] describes one exported panel. The optional subcaption and label are for subcaptions.

In normal use, you place one layout macro inside \figfloat. For example:

\figfloat[b!]
  {\figonewide{\fig{figures/plot.pdf}}}
  [A simple exported plot.]
  [fig:plot]

For small multi-panel figures, you can use a direct panel-by-panel form:

\figfloat[t]
  {
    \figtwowide
      {\fig{figures/left.pdf}[Left panel][fig:left]}
      {\fig{figures/right.pdf}[Right panel][fig:right]}
  }
  [Two-panel figure.]
  [fig:two-panel]

For larger grids, use the row-grouped form:

\figfloat[t]
  {
    \figsix
    {{\fig{figures/a.pdf}}{\fig{figures/b.pdf}}}
    {{\fig{figures/c.pdf}}{\fig{figures/d.pdf}}}
    {{\fig{figures/e.pdf}}{\fig{figures/f.pdf}}}
  }
  [Six-panel figure.]
  [fig:six-panel]

The supported layout macros are:

• \figone, \figonewide
• \figtwo, \figtwowide
• \figthree, \figthreewide
• \figfour, \figsix, \figsixwide
• \fignine, \figtwelve, \figtwelvewide
• \figfifteen, \figsixteen, \figtwenty

Troubleshooting

• If Python export fails with a LaTeX error, check that your TeX installation is available from the command line and includes the required packages.
• If exported figure sizes do not match your document, print \figprintlayoutspec from the real document and update your Python template values from that output.
• If LaTeX reports that a figure float is too large for the page, the usual fixes are to shorten the caption, choose a less tall layout, or adjust the template spacing and allowance values.

Python API Reference

Detailed Python API documentation is at nvnunes.github.io/pubify-mpl/api/.

LaTeX Package Reference

Common forms:

\usepackage{pubify}
\usepackage[template=path/pubify-template.tex]{pubify}
\usepackage[debug]{pubify}
\usepackage[template=path/pubify-template.tex,debug]{pubify}

Package options:

• template=...: load an explicit template file
• debug: add figure borders and print layout diagnostics to the LaTeX log

Template resolution:

• if template=... is given, pubify.sty loads that file
• otherwise, pubify.sty loads pubify-template.tex if it is present next to pubify.sty
• otherwise, pubify.sty falls back to its built-in default lengths

Development Approach

The implementation is intentionally pragmatic. Priority was given to producing a useful, validated tool rather than to maximizing internal elegance or generality. Parts of the implementation were developed with AI-assisted workflows. Development effort was focused on documented behavior, intended performance, and validation rather than on highly refined internal structure.

License

MIT