Math support (Simple)

[bennyboer]

We need support for a #MATH# Thingy allowing to enter LaTeX formatted formulas that will be properly renderered in the document.

Maybe we want to convert LaTeX first to MathML using SnuggleTeX and then rather typeset MathML.

[bennyboer]

Branch is math

[bennyboer]

The considered SnuggleTex and JEuclid seem to be pretty old which will most certainly cause problems with current Java versions with module support. Thus we'll write our own LaTeX math parser + converter to MathML + MathML typesetter

[bennyboer]

We are restricting the supported MathML element types (See https://developer.mozilla.org/de/docs/Web/MathML/Element) to a specific subset, that most common math expressions will work with v0.1.

Element types to support for v0.1:

• math (Partial support - no attributes currently supported)
  • Basic rendering
• mrow
  • Basic rendering
• msup
  • Basic rendering
  • Attribute superscriptshift
• msub
  • Basic rendering
  • Attribute subscriptshift
• msubsup
  • Basic rendering
  • Attribute subscriptshift
  • Attribute superscriptshift
• mroot
  • Basic rendering
• msqrt
  • Basic rendering
• mi
  • Basic rendering
  • Attribute mathsize
  • Attribute mathvariant
• mn
  • Basic rendering
  • Attribute mathsize
  • Attribute mathvariant
• mtext
  • Basic rendering
  • Attribute mathsize
  • Attribute mathvariant
• mfrac
  • Basic rendering
  • Attribute bevelled
  • Attributes numalign and denomalign
  • Attribute linethickness
• mover
  • Basic rendering
  • Attribute align
• munder
  • Basic rendering
  • Attribute align
• munderover
  • Basic rendering
  • Attribute align
• mspace
  • Basic rendering
  • Attributes depth, height and width
• mpadded
  • Basic rendering
• mo
  • Basic rendering
  • Attribute lspace, rspace
  • Attribute mathsize
  • Attribute mathvariant
  • Attribute stretchy

Other Tasks

• Create Thaw project that shows off the MathML torture test (https://mdn.mozillademos.org/en-US/docs/Mozilla/MathML_Project/MathML_Torture_Test$samples/MathML_Torture_Test?revision=1506691)
• Check if we can use the STIX 2 font as the math font. No real need to let the user configure the math font in my opinion. We need to check the license of the font and how to embed it. (https://github.com/stipub/stixfonts)

That way we will allow most use-cases while some special cases may remain to be done in later releases.

[bennyboer]

Thingy ideas:

Thingy should be allowed both inline and as its own paragraph.
Note that we won't adjust line heights depending on the size of the math expression.
If the expression is too big to be displayed inline, you will have to create its own paragraph for it.
When displayed inline the math expression should be fit to the line height. So if you have an extremely verbose expression it should be scaled to fit the line height.

Name: #MATH#
Arguments: Either MathML or TeX style math expressions (Should be determined automatically)
Options:

• caption: Caption for the math (only for math thingies in their own paragraph)
• number: Whether to number the expression
• label: Internal document reference label (Is already working for all thingies)

[bennyboer]

First end-to-end process draft is finished supporting MathML with math, mi, mn and mo nodes.
For example the following is already working:

#MATH,
<math>
	<mi>a</mi>
	<mo>+</mo>
	<mn>1</mn>
</math>
#

[bennyboer]

Now supporting mfrac and mrow as well as the attribute mathvariant for mi, mo, mn to display for example set identifiers (N, R, C, ...)

[bennyboer]

Currently already possible is something like:

Compared to Firefox:

<math>
	<msubsup>
		<mo>&#x222B;</mo>
		<mn>0</mn>
		<mn>1</mn>
	</msubsup>
	<mroot>
		<mfrac>
			<mrow>
				<mi>sin</mi>
				<mo>(</mo>
				<mi>&#x3C0;</mi>
				<mo>)</mo>
			</mrow>
			<mn>2</mn>
		</mfrac>
		<mn>3</mn>
	</mroot>
	<mroot>
		<mi>x</mi>
		<mn>3</mn>
	</mroot>
	<mo>+</mo>
	<mn>42</mn>
	<msqrt>
		<mrow>
			<msubsup>
				<mo>&#x222B;</mo>
				<mn>0</mn>
				<mn>1</mn>
			</msubsup>
			<mroot>
				<mfrac>
					<mrow>
						<mi>sin</mi>
						<mo>(</mo>
						<mi>&#x3C0;</mi>
						<mo>)</mo>
					</mrow>
					<mn>2</mn>
				</mfrac>
				<mn>3</mn>
			</mroot>
			<mroot>
				<mi>x</mi>
				<mn>3</mn>
			</mroot>
			<mo>+</mo>
			<mn>42</mn>
		</mrow>
	</msqrt>
</math>

[bennyboer]

Support for the bevelled and linethickness attributes of the <mfrac> node:

#MATH, '
<math>
	<mfrac bevelled="true" linethickness="1.0">
		<mrow>
			<mfrac>
				<msqrt>
					<mn>2</mn>
				</msqrt>
				<mn>42</mn>
			</mfrac>
			<mo>-</mo>
			<mi>y</mi>
		</mrow>
		<mfrac>
			<mroot>
				<mrow>
					<mn>42</mn>
					<mo>*</mo>
					<mi>x</mi>
				</mrow>
				<mn>3</mn>
			</mroot>
			<mrow>
				<mi>sin</mi>
				<mo>(</mo>
				<mn>2.43</mn>
				<mo>)</mo>
			</mrow>
		</mfrac>
	</mfrac>
	<mo>+</mo>
	<mn>1</mn>
</math>
'#

[bennyboer]

#MATH# thingy now also working in-line with text.

[bennyboer]

Just checked whether we can use the STIX 2.0 font for math. License seems to be OK, needs to be documented in the third party notice file in the root of the repository.

Also checked whether Apache PDFBox does support OpenType fonts and it works just fine by using something like:

OTFParser otfParser = new OTFParser();
OpenTypeFont otf = otfParser.parse(new File("C:/Users/beder/Downloads/code/CODE Light.otf"));

PDFont font = PDType0Font.load(document, otf, false);

When calling the PDType0Font.load method we have to set the third parameter to false as PDFBox does currently not support OTF font subsetting (See https://issues.apache.org/jira/browse/PDFBOX-3758).

[bennyboer]

LaTeX to MathML converter will not be done in the context of this issue, instead see issue #23.

[bennyboer]

There is already an issue about the extended math support where we will implement more sophisticated and less usefull MathML features: #22

[bennyboer]

STIX 2 font added.

[bennyboer]

math branch has been merged to master