So instead of a well known (i.e. POSIX) quoting semantics and existing tool support, you want to introduce your own ad-hoc format? No thanks.
A correct parser for the syntax I described can be written in less than a 100 lines of code — even in C. It’s a strict subset of the shell command language defined by POSIX, and it’s sufficiently expressive as to support specifying any argument array unambiguously.
To correctly escape arbitrary shell syntax, not only do you need to handle the full POSIX syntax (which is quite complex) …
https://pubs.opengroup.org/onlinepubs/9799919799/utilities/V...
… but you must also cover any bugs and undocumented/underspecified extensions implemented by the actual shell providing /bin/sh on every platform and platform version to which your code will be deployed.
That’s not just difficult — it’s impossible, and everyone that has tried has failed, repeatedly. Leading to security bugs, repeatedly.
https://gist.github.com/Zenexer/40d02da5e07f151adeaeeaa11af9...
There’s a reason why we use parameterized queries instead of escaping to prevent SQL injection, and SQL syntax and parsing behavior is far more rigorously specified than the shell.
You don't need to care about the whole quoting spec. E.g. what you need to escape without quotes becomes irrelevant as soon as you consistently add quotes and what you can do with $ expressions is irrelevant if you simply escape all $. This is far from rocket science.
But it also doesn't matter how complex the standard is, it is a standard and one with much preexisting support. Your "standard" will end up being different for each application.