Incorrect data-latex attribute for multi-digit numeric subscripts

[gelprec]

Issue Summary

For multi-digit numeric subscripts, the data-latex attribute fails to match the subscript value correctly, causing the first digit to be ignored. This issue does not occur with symbolic subscripts. Tested on the latest Mathjax v4

Steps to Reproduce:

1. Open the MathJax live demo
2. Enter

   $a_{123}$
   

3. Analyze the corresponding HTML output in the browser's developer console. You will see something like the following:

   <mjx-texatom data-latex="{23}"...>...</mjx-texatom>
   

Expected Behavior

I expect something like the following:

<mjx-texatom data-latex="{123}"...>...</mjx-texatom>

Technical details:

• MathJax Version: 4.0
• Client OS: Windows 11
• Browser: Chrome

[dpvc]

Thanks for the report. There are a number of issues with the data-latex attribute, and I've looked into the problem and will be filing a PR to fix this and several other problems with it.

In the meantime, you can use the following configuration to get most of the improvements. It should take care of your subscript issue.

MathJax = {
  startup: {
    ready() {
      const NodeUtil = MathJax._.input.tex.NodeUtil.default;
      NodeUtil.copyAttributes = (oldNode, newNode) => {
        newNode.attributes = oldNode.attributes;
        for (const [prop, value] of Object.entries(oldNode.getAllProperties())) {
          newNode.setProperty(prop, value);
        }
      };
      const TexParser = MathJax._.input.tex.TexParser.default.prototype;
      const [LATEX, LATEXITEM] = ['data-latex', 'data-latex-item'];
      Object.assign(TexParser, {
        _Push_: TexParser.Push,
        Push(item) {
          if (item.kind === 'stop') {
            this.stack.global.stopping = true;
          }
          return this._Push_(item);
        },
        parse(kind, input) {
          const i = this.saveI;
          this.saveI = this.i - (kind === 'character' && input[1] !== '&' ? 1 : 0);
          const result = this.configuration.handlers.get(kind).parse(input);
          if (kind !== 'macro') {
            this.updateResult(input[1], i);
          }
          this.saveI = i;
          return result;
        },
        mml() {
          if (!this.stack.Top().isKind('mml')) {
            return null;
          }
          const node = this.stack.Top().First;
          this.configuration.popParser();
          const latex = this.trimTex(this.string);
          if (latex) {
            node.attributes.set(LATEX, latex);
          }
          return node;
        },
        trimTex(tex) {
          return tex.trim() + (tex.match(/(?:^|[^\\])(?:\\\\)*\\\s+$/) ? ' ' : '');
        },
        updateResult(input, old) {
          const node = this.stack.Prev(true);
          if (!node || this.stack.global.stopping) return;
          const latex = node.attributes.get(LATEX);
          const existing = node.attributes.get(LATEXITEM);
          if (existing !== undefined) {
            if (!latex) {
              if (input === '}' || existing === '}') {
                this.composeBraces(node);
              } else {
                node.attributes.set(LATEX, existing);
              }
            }
            return;
          }
          old = old < this.saveI ? this.saveI : old;
          const str = this.trimTex(old !== this.i ? this.string.slice(old, this.i) : input);
          if (!str || str === latex || (input === '\\' && str === '\\')) return;
          if (str === '_' || str === '^') {
            node.setProperty('sub-sup', str);
          } else {
            switch (node.getProperty('sub-sup')) {
             case '^':
              if (node.childNodes[2]) {
                if (str === '}') {
                  this.composeBraces(node.childNodes[2]);
                } else if (!node.childNodes[2].attributes.hasExplicit(LATEX)) {
                  node.childNodes[2].attributes.set(LATEX, str);
                }
              }
              if (node.childNodes[1]) {
                const sub = node.childNodes[1].attributes.get(LATEX);
                this.composeLatex(node, `_${sub}^`, 0, 2);
              } else {
                this.composeLatex(node, '^', 0, 2);
              }
              return;
             case '_':
              if (node.childNodes[1]) {
                if (str === '}') {
                  this.composeBraces(node.childNodes[1]);
                } else if (!node.childNodes[1].attributes.hasExplicit(LATEX)) {
                  node.childNodes[1].attributes.set(LATEX, str);
                }
              }
              if (node.childNodes[2]) {
                const sup = node.childNodes[2].attributes.get(LATEX);
                this.composeLatex(node, `^${sup}_`, 0, 1);
              } else {
                this.composeLatex(node, '_', 0, 1);
              }
              return;
            }
            if (str === '}') {
              this.composeBraces(node);
              return;
            }
          }
          node.attributes.set(LATEX, str);
        }
      });
      MathJax.startup.defaultReady();
    }
  }
};

It's longer than I like, but the fixes are at a pretty low-level that is not easy to patch without replacing entire functions, as I do here. One function that is changed is the constructor for the TexParser object, which can't be easily replaced, so there are some extra changes to work around that.

[gelprec]

Yes, everything looks good now with the configuration you provided above. Thanks a lot for your help!

[dpvc]

Thanks for the confirmation.