How to avoid incorrect regex interpretations in visual mode?

[YaoYYoung]

I have some practical snippets which will allow me to switch the callout style easily by

• click the callout, which means to select the callout texts
• enter the corresponding key（like important is linked to 1）
• then the original note/tip/... callout style will switch to important style.

But when the context contains LaTeX characters which starts with NUMBER these snippets will behave incorrectly.

I find the following part causing the issue:

{trigger: "7", replacement: (sel) => {
    return sel; 
}, 
options: "vtA"},

// or 

{trigger: "7", replacement: (sel) => {
    sel = sel.replaceAll(/\$(\d+)/g,'\$xxx$1');
    sel = sel.replaceAll(/(\$)xxx(\d+)/g,'$1$2');
    return sel;
}, 
options: "vtA"},

• 🟢this works fine (and does nothing) when you select normal characters and press 7;
• 🔴however, when it encounters characters like $2+8$, it treats $2 as a placeholder instead of normal characters, and the output becomes +8$.

I believe the reason of this bug is that the replacement => {} will implicitly perform a regex substitution at the end.

So, how can I avoid this incorrect regex interpretation when the callout contents contain LaTeX formulas which starts with a number?

[YaoYYoung]

I just found out an eclectic method to solve this problem:

When a "dollar-number" combination prefix appears in a selected math formula, I will distinctively enclose the entire math formula in ${{}}$ instead of using $$.

For example, $3.12 \times 10^{18}$ will be changed into ${{3.12 \times 10^{18}}}$ which remains completely unchanged for $\LaTeX$ compiler, avoiding an incorrect result like .12 \times 10^{18}$.

Therefore, every vtA mode snippets will add the following code at the beginning of replacement function to avoid this incorrect regex interpretation in visual mode.

const sensitive_math = /(\s*)\$(\d[^$]*)\$(\s*)/g; 
if (sel.match(sensitive_math)) 
	sel = sel.replaceAll(sensitive_math, "$1\${{$2}}\$$$3"); 

But the weakness of this method is apparent. Now I have 16 vtA mode snippet so I have to place this "subsippet" code into these 16 snippets respectively, which is quite superfluous.