MathJax Render Actions — MathJax 4.0 documentation (original) (raw)

A MathDocument instance contains a sequence of functions called renderActions that are used during the typesetting process to perform the actions needed to locate expressions in the page, compile them into the internal format, obtain font metrics for the surrounding text, typeset the mathematics, insert the typeset math back into the page, add menu actions, and so on. Loading some components will add new functions into this list. For example, the a11y/semantic-enrich component adds a render action that computes the semantic-enrichment for each expression. All the typesetting and conversion functions listed above use this collection of functions to perform their typesetting or conversion operations.

The renderActions configuration option provides a means of linking your own functions into this processing pipeline at any point, replacing existing steps with your own custom versions, or even removing them to trim down what MathJax does.

The renderActions configuration object consists of one or morename: value pairs (separated by commas), where the namegives an identifier for the action, and the value is an array consisting of a number followed by zero, one, or two functions, and an optional boolean, as described below.

The number gives the priority of the action in relation to the other actions that are already defined. The functions for the lowest numbers are performed first. The number may be negative, or have decimal places. The existing priorities are found in theSTATE variable from the ts/core/MathItem.ts file or theMathJax._.core.MathItem object (when using MathJax Components). These include the following:

Here, the name is the value’s key in the STATE object that has the given priority. The identifier is the key in therenderAction object for the action with the given priority. Some states are for reference only and do not have associated actions; e.g., CONVERT and RERENDER are only used to tell where amathDocument.convert() ormathDocument.rerender() action should start in the list of render actions, while UNPROCESSED and LAST mark the usual start and end of the processing list, though values outside that range are allowed.

Some extensions add new render actions. These include

You may use these values to determine the appropriate priorities to use for your own actions. Actions with the same priority will be performed in the order in which they were added to the render action list for the document by you and the components you have loaded. ThenewState() function from the ts/core/MathItem.ts file can be used to create a new named state value like the ones above. Otherwise, you can give priorities as numeric values, or as an existing STATE, possibly value plus (or minus) a given number.

Following the priority in the action’s definition array should be zero, one, or two functions. The first is called whenever the page is typeset via a call to MathJax.typeset(),mathDocument.render() ormathDocument.rerender(), or their promise-based versions, and is passed the MathDocument object in which it is running. Usually this function will loop through the MathItem objects stored in the mathDocument.math list and perform its action on each of those individually.

The second function is called whenever a math expression is updated individually via its rerender() method (e.g., when a subexpression is collapsed, or a toggle item is toggled, or the speech ruleset is changed), or during a mathDocument.convert()call is made, or any of the conversion functions based on it (e.g.,MathJax.tex2svg()), or any of the promise-based versions of this is called. The function for this action is called with two arguments: the MathItem on which it is to operate, and theMathDocument that holds that item.

If either of these functions is given as an empty string, then that action is not performed (that way, you can define a typeset action without a rerender action, for example), and if it is a non-empty string rather than a function, that is taken to be the name of a method of the MathDocument (for the first function) orMathItem (for the second function) to be called for that action instead. If the function is missing, then the identifier used for the action in the renderAction object is taken as the method name to use.

Finally, the optional boolean value tells whether the second function is to be use for both convert() and rerender() actions (when true, the default), or only for rerender() actions (when false).

To remove a render action, set its identifier to an empty array. For example,

MathJax = { option: { renderActions: { addMenu: [], getMenu: [], checkLoading: [], } } }

would disable all the MathJax menu actions. This would disable to the MathJax menu, though it is easier to set enableMenu tofalse to accomplish that.

To replace a render action, set its identifier to the definition you would like to use instead. For example, you could replace thetypeset action with one that generates MathML output in the page rather than the CHTML or SVG output from MathJax. TheMathML Support section for an example of how to do this.

A Render Action for Tooltips

Here is an example of defining your own render action that adds the original TeX notation as a tooltip on the MathJax output:

MathJax = { addTooltip(item) { const adaptor = MathJax.startup.adaptor; adaptor.setAttribute(math.typesetRoot, 'title', item.math); }, options: { renderActions: { addTooltip: [175, (doc) => { for (const item of doc.math) { MathJax.config.addTooltip(item); } }, (math, doc) => MathJax.config.addTootip(math), ] } }, }

This uses a function addTooltip() that is stored in the MathJax configuration (and otherwise ignored by MathJax), and specified a new addTooltip render action at priority 175 (after typesetting by before inserting into the page). The first function of the action, which is called as part of a typeset action, loops through the math items in the document and call theaddTooltip() function on each one. The second, called during a convert or rerender action, calls the addTooltip() function on the math item that it was passed.

A Render Action to Collapse Complex Subexpressions

This example loads the a11y/complexity component and automatically collapses any subexpressions with complexity greater than a given value (20 in this case).

MathJax = { options: { menuOptions: { settings: { collapsible: true } }, renderActions: { collapse: [50, (doc) => { for (const math of doc.math) { if (!math.root || math.state > 50) continue; math.root.walkTree((node) => { if (node.isKind('maction') && node.attributes.get('data-collapsible')) { if (node.childNodes[1].attributes.get('data-semantic-complexity') > 20) { node.attributes.set('selection', 1); } } }); math.state(50); } }, '', false ] } } };

Here we define an action that occurs after the math is compiled and its complexity values have been computed, but before the math is typeset. The action only operates for typeset actions (not convert or rerender calls), and it walks the MathML tree to find mactionelements inserted by the complexity component that have complexity greater than 20. These get their selections set to 1, so they select the collapsed version instead of the expanded ones. Then the state is set to 50 so that if the item is processed again (either because the typesetting had to be restarted due to an asynchronous action later in the render actions, or by a later typesetting call), this math item will not be checked again.

A Render Action to use Tags for Math Delimiters

This example replaces the usual find action with a custom one. Instead of looking for the usual \(...\), \[...\], and$$...$$ math delimiters, this render action will look for<tex>...</tex> and <dtex>...</dtex> elements for in-line and display style TeX expressions.

MathJax = { options: { renderActions: { find: [10, (doc) => { for (const node of document.querySelectorAll('tex, dtex')) { if (node.childNodes.length !== 1 || node.firstChild.nodeName !== '#text') continue; const display = node.nodeName.toLowerCase() === 'dtex'; const math = new doc.options.MathItem(node.textContent, doc.inputJax.tex, display); const text = document.createTextNode(''); node.parentNode.replaceChild(text, node); math.start = {node: text, delim: '', n: 0}; math.end = {node: text, delim: '', n: 0}; doc.math.push(math); } }, '', false ] } } };

Thius render action replaces the usual find action with one that looks for all the <tex> and <dtex> elements in the document, and checks that they have only one child that is a text element. If so, it creates a new MathItem object with the node’s text content, using the TeX input jax, and marked as a display expression if the tag was dtex. It then makes a new empty text node and replaces the original node with that, then hooks theMathItem to the empty text node. Finally, it pushes the newMathItem onto the document’s math list.

The reason this render action replaces the usual one is because its identifier is find. If you want to look for both the original delimiters and these tags then you should use a different identifier, like findTags, instead.