Page MenuHomePhabricator
Create Task
Maniphest T397585

Provide a macro for putting arbitrary Unicode content in <mo> tag
Open, Needs TriagePublicFeature
Actions

Description

Feature summary (what you would like to be able to do and where):

As of today, there are several LaTeX macros which allows to put (almost) arbitrary Unicode content in a <mtext> tag in the produced MathML : \text, \mbox, and more generally every macro with the box_functions tag. I would like to have a macro which do the same, but which put the content in a <mo> tag.

Use case(s) (list the steps that you performed to discover that problem, and describe the actual underlying problem which you want to solve. Do not describe only a solution):

A lot of symbols cannot be typed in math blocks today, which is quite limiting, especially since actual LaTeX implementations allow arbitrary unicode in formulas (with unicode-math). For example, I couldn't type ⅋ and I had to do use spacing tricks to type ⟦ and ⟧ (see https://phabricator.wikimedia.org/T391290). With a macro named for example \mobox, one could simply type \mobox{⅋} or \mobox{⟦}.

Benefits (why should this be implemented?):
Of course, one could write these characters inside a \text, but it doesn't respect the semantics of MathML. Moreover, for delimiters for example, one would like that they scale if they are combined with \big or \left, so it is not fully satisfying.

I don't really know if this extension is commited to use only a valid subset of LaTeX, because then adding a macro *which doesn't exists in any LaTeX package* would break this commitment. Maybe the source rendering mode should just delete the use of this macro if we want it to output valide LaTeX (e.g. it would transform \mobox{a} in a). Also, I'm not sure about which options pass to the mo tag. Some operators use stretchy="false" by default, other do not (and MathML itself has its own default), even parentheses/brackets/braces are not consistent about fence="false"), and operators next to \left or \right get fence="true", stretchy="true", symmetric="true". I think a good default would be stretchy="false", fence="false", but maybe there can be several macros, or a macro with optional arguments, to handle all the cases.

I have working a prototype for this, which works with both client MathJax rendering and MathML rendering, but I don't know enough the extension to know if it broke something I didn't see.

Details

Related Changes in Gerrit:
SubjectRepoBranchLines +/-
Add \mobox primitive in order to put arbitrary symbols in <mo> MathML tagmediawiki/extensions/Mathmaster+327 -165
Customize query in gerrit

Related Objects
Search...

StatusSubtypeAssignedTask
 OpenFeatureJeanCASPART397585 Provide a macro for putting arbitrary Unicode content in <mo> tag
 OpenPhysikerweltT271001 Transition to MathML rendering as default
 StalledNoneT324234 Remove external service defaults from Extension:Math
 ResolvedtaaviT327297 Recreate mathoid-beta.wmflabs.org
 ResolvedPhysikerweltT373703 Enable native mathml rendering by default on group0 and test wikis in production
 ResolvedPhysikerweltT373705 Set default configuration for Math extension to native mathml
 DuplicateBUG REPORTNoneT381070 Error: Math extension cannot connect to Restbase.
 OpenBUG REPORTNoneT388304 Ugly rendering of <math>n \choose k</math> and <math>{n \choose k}</math>
 ResolvedPhysikerweltT390294 Use MathML rendering in wikibase
 OpenBUG REPORTNoneT395488 Not enough space between functions and arguments in MathML mode
 OpenPhysikerweltT398653 Ensure Math extension output is fully MathML core compliant
 OpenBUG REPORTNoneT382267 Math block inside a {{blockquote}} render in display style in MathML and Client side MathJax
 OpenPhysikerweltT413973 Change wmf-config for Math extension from Mathoid to MathML+MatJax on all wikis that have Parsoid and no known <math> usage
 OpenBUG REPORTPhysikerweltT375238 Client-side MathJax rendering problems
 ResolvedBUG REPORTPhysikerweltT375241 <math>\text{A }</math> with a space inside in Client Side MathJax mode give math input error
 ResolvedBUG REPORTPhysikerweltT375244 \boldsymbol{\alpha} gives a Math Output error in Client Side MathJax mode
 ResolvedBUG REPORTPhysikerweltT375863 \operatorname{sn} rendered with no space before and after, and too much space between characters in Client Side MathJax mode
 ResolvedBUG REPORTPhysikerweltT375907 \sum_condition puts condition in the in wrong position in MathML and client-side MathJax modes
 ResolvedBUG REPORTPhysikerweltT375935 The factorial symbol ! should be upright, it is sloping in Clients-side mathjax mode.
 ResolvedBUG REPORTPhysikerweltT375932 \mathcal uses incorrect font in Client side MathJax modes
 ResolvedBUG REPORTPhysikerweltT376546 \| should render as a double pipe || rather than a single one |. In MathML and Client-side MathJax modes.
 DuplicateBUG REPORTPhysikerweltT378257 \limits is rendered verbatim in MathML mode and client-side MathJax
 ResolvedBUG REPORTPhysikerweltT381102 <math>\left(a\right)'</math> in MathML and MathJax renders with one prime symbol too much
 DuplicateBUG REPORTNoneT381305 MathML/MathJax next to floating element or <blockquote> not rendered inline
 ResolvedBUG REPORTPhysikerweltT382196 \overbrace rendered below (not above) in MathML and client-side Mathjax
 ResolvedBUG REPORTPhysikerweltT382215 TeX macro "\ " not inserting space in MathML and client-side MathJax
 ResolvedBUG REPORTPhysikerweltT384794 Prime symbol after \operatorname not rendered as superscript
 ResolvedBUG REPORTPhysikerweltT386071 <math>_1</math> not rendered as subscript
 ResolvedPhysikerweltT386318 <math>\sin \cos^2</math> not rendered as superscript
 ResolvedPhysikerweltT387952 <math>{_{1}^2}</math> not rendered as subscript
 ResolvedBUG REPORTPhysikerweltT386585 <math>\textrm{foobar}</math> results in "Math input error" with MathJax
 ResolvedBUG REPORTJeanCASPART386787 Trailing linebreak via `\\` in <math> matrix should be ignored
 ResolvedBUG REPORTPhysikerweltT387249 <math>\hat{}</math> generates invalid MathML
 ResolvedBUG REPORTPhysikerweltT390259 <math>\mathbf{\Lambda}</math> not bold
 ResolvedBUG REPORTPhysikerweltT390260 Too much spacing in <math>\operatorname{arg\,max}</math> with MathJax
 DuplicateBUG REPORTNoneT391889 <math>\textrm{foo}</math> displays as Math Input Error in Client side MathJax and MathML modes
 ResolvedBUG REPORTPhysikerweltT397120 \textbf{1} renders incorrectly in MathML and client-side MathJax modes
 ResolvedBUG REPORTPhysikerweltT401623 Bad kerning of decimal separator in MathML
 ResolvedPhysikerweltT405439 Investigate <math>1\definecolor{c}{RGB}{255,0,0}+2</math>
ResolvedPhysikerweltT401718 \textstyle not applied to \binom in MathML
 ResolvedPhysikerweltT401740 \Bigg\vert is not big in MathML
 ResolvedPhysikerweltT404292 Translate MathML core alignment to MathJax formatting
 ResolvedPhysikerweltT409110 Update to MathJax v4
 OpenPhysikerweltT409152 \hat and \widehat look the same
 OpenPhysikerweltT380654 vertical space between multline equations is ignored
 In ProgressBUG REPORTPhysikerweltT414124 MathJax rendering of small oint is wrong
 OpenBUG REPORTNoneT414130 MathJax rendering of Riemanian volume form is wrong
 OpenKrinkleT414055 Performance review of MathJax client-side mode
 ResolvedPhysikerweltT414994 Load all MathJax files from WMF servers

Event Timeline

JeanCASPAR created this task.Jun 22 2025, 5:18 PM
Restricted Application added a subscriber: Aklapper. · View Herald TranscriptJun 22 2025, 5:18 PM
gerritbot added a comment.Jun 25 2025, 7:13 PM

Change #1163862 had a related patch set uploaded (by JeanCASPAR; author: JeanCASPAR):

[mediawiki/extensions/Math@master] Add \mobox primitive in order to put arbitrary symbols in <mo> MathML tag

https://gerrit.wikimedia.org/r/1163862

gerritbot added a project: Patch-For-Review.Jun 25 2025, 7:13 PM
Physikerwelt subscribed.Jul 1 2025, 9:32 AM

This is generally a good idea. However, I think this should be coordinated with the Latex project and we should not invent a macro that does not render in normal tex. Note that the latest Latex version also supports MathML output (see https://www.latex-project.org/news/2025/06/02/issue41-of-latex2e-released/) so I wonder if we can find a macro name that is compatible with the regular LaTeX rendering.
If we would be able to extend the list of supported input, we might also allow them in normal math mode. So instead of writing \mobox{⟦} we could simply write ⟦ (at least for the operators listed in the operator dictionary, because they have defined spacing https://www.w3.org/TR/mathml-core/#operator-dictionary-human).

Unfortunately, this can not be started before we get rid of the old Restbase rendering mode.

Physikerwelt added a subtask: T271001: Transition to MathML rendering as default.Jul 1 2025, 9:33 AM
JeanCASPAR claimed this task.Jul 12 2025, 7:23 PM

So, I did some tests with the MathML output and it seems to work fine, even with unicode, but I had to use the package unicode-math, as the symbols provided by external packages did not work (I think that it should be fine with the ones of the ams* packages). From what I understood, \mathrel, \mathord, etc. determinate the spacing and the output type (mi or mo) of the symbol, and by default non-letter symbol are put in a <mo> tag. It doesn't exactly solve this issue, but I think it is good enough: for example, \mathbin{A} yields <mo lspace="0.222em" rspace="0.222em">A</mo> and \mathbin{AB} yields :

<mspace width="0.222em"></mspace>
<mrow>
<mi>𝐵</mi>
<mi>𝐶</mi>
</mrow>
<mspace width="0.222em"></mspace>

Also, this it doesn't respect exactly https://www.w3.org/TR/mathml-core/#operator-dictionary-human : sometimes it is because of TeX related reason (*= is treated as two separate characters, as TeX does), sometimes I don't know why (⨵ has 0.222em of spacing instead of 0.166em), but I think it is not a problem.

I don't know how experimental it is, but we could try to use directly the MathML output generated by the TeX compiler (when Restbase is dropped). But it will make it harder if we want later to allow arbitrary wikicode inside of \text{}, which would be cool, for example in order to have links in math blocks or nested math expressions (e.g. it would be useful to allow things like 0 \text{if $x \leq 0$} in case expressions). This should not be a problem from a MathML perspective, as arbitrary HTML can be nested inside a <mtext>, but it may be a bit difficult (we need to call the whole wikicode parser back but only up to the end of the enclosing brace).

I think that trying to extend the list of accepted inputs may be easier, and will allow us to keep more control on the whole process.

Physikerwelt added a comment.Oct 17 2025, 5:43 AM

@JeanCASPAR I think we currently have still too many open problems to switch to plain MathML directly, however, we can use MathJax for rendering MathML which gives quite nice results.

Content licensed under Creative Commons Attribution-ShareAlike (CC BY-SA) 4.0 unless otherwise noted; code licensed under GNU General Public License (GPL) 2.0 or later and other open source licenses. By using this site, you agree to the Terms of Use, Privacy Policy, and Code of Conduct. · Wikimedia Foundation · Privacy Policy · Code of Conduct · Terms of Use · Disclaimer · CC-BY-SA · GPL · Credits