Preserving line breaks across math formats

Suleman Uzair Ronald Tse on 10 Apr 2025

More than a change of syntax

A mathematical expression rarely stays in the format in which it was written. An author might enter LaTeX, while the published document needs MathML for the web and OMML for Word. This is the sort of publishing problem that led us to create Plurimath.

Converting the symbols is only part of that job. The layout also tells the reader how to follow the math. A line break, for example, can separate one step of a calculation from the next:

$${\displaystyle x+1=2}$$ $${\displaystyle x=1}$$

There is nothing complicated about this calculation. That is what makes it useful here: even with every symbol converted correctly, running the two steps together would make the result harder to read.

So how should a converter carry that separation into another math language?

One formula, or separate equations?

One approach is to keep the calculation in a single equation and mark the point where a new line should begin. This is Plurimath’s default MathML output for the example above: one <math> element containing a <mo linebreak="newline"/> marker between the steps.

That leaves the viewer to act on the marker. If it does, the calculation can appear on two lines. If it ignores the marker, preserving it in the converted markup has not preserved the layout the reader sees.

Another approach is to make each step an equation in its own right. This gives the application publishing the result separate equation blocks to place in the document, rather than an internal break for the math viewer to interpret.

Plurimath supports this second approach through split_on_linebreak, an export option for MathML and OMML. The distinction is small in the API, but useful when a publishing workflow needs to handle the steps separately.

From LaTeX to MathML

Here is the calculation in LaTeX, with \\ marking the break between its two steps. We parse it once, then ask for separate equations on export:

require "plurimath"

input = <<~'LATEX'
  x + 1 = 2 \\
  x = 1
LATEX

formula = Plurimath::Math.parse(input, :latex)
puts formula.to_mathml(split_on_linebreak: true)

The result contains two complete <math> elements, one for each step. The following outline leaves out the expression markup to show the change in structure:

<math xmlns="http://www.w3.org/1998/Math/MathML" display="block">
  <!-- MathML for x + 1 = 2 -->
</math>
<math xmlns="http://www.w3.org/1998/Math/MathML" display="block">
  <!-- MathML for x = 1 -->
</math>

The actual output includes the symbols and their <mstyle> wrappers. There is no need to split the LaTeX string and parse each line yourself; Plurimath uses the break already present in the parsed formula.

The option belongs to the export call. We can therefore keep the same formula and choose the output structure to suit the application consuming it. Calling formula.to_mathml without the option retains the single <math> element and its line-break marker.

The same choice for Word

For a Word publishing workflow, we can export that formula as OMML instead:

puts formula.to_omml(split_on_linebreak: true)

Here, splitting produces two <m:oMath> equation elements inside one <m:oMathPara> paragraph. The surrounding document and the application rendering it still determine their placement.

OMML has a manual line-break mechanism of its own. The purpose of this option is to offer separate equation elements when that is the structure the publishing application needs.

Where splitting helps

There is a boundary between preserving the author’s breaks and typesetting the calculation. split_on_linebreak handles the former: it uses explicit breaks in the input. It does not decide where a long expression should wrap, or align the equals signs in a derivation. Those remain layout decisions for the document or math renderer.

For a workflow that already displays internal breaks as intended, the default output may be all that is needed. When the application needs each step as a separate equation, split_on_linebreak: true provides that structure without changing the source calculation.

The option was introduced in Plurimath 0.7.0 and is available on both to_mathml and to_omml. See the usage guide for installing Plurimath and converting your own formulas.