scroll root parser evalMacros should escape dollar signs in the replacement string

[btheado]

As mentioned in this pldb issue, the djot page has messed up output.

I think the issue is the pldb example contains

$`p = mv`

and this line is using that in the regexp replacement string

scroll/parsers/root.parsers

Line 835 in b855598

Object.keys(macroMap).forEach(key => (codeAfterMacroSubstitution = codeAfterMacroSubstitution.replace(new RegExp(key, "g"), macroMap[key])))

According to MDN, the dollar sign followed by backtick has special meaning in the replacment string.

I think prior to using the replacement string (i.e. macroMap[key]), any dollar sign should be replaced with a double dollar sign.

[breck7]

Nice work!

Here's a simple repro:
https://try.scroll.pub/#scroll%0A%20buildTxt%0A%20%0A%20replace%20foo%20bar%0A%20foo%20foo%0A%20%0A%20replace%20problem%20%24%60%0A%20plainText%20a%20problem

buildTxt

replace foo bar
foo foo

replace problem $`
plainText a problem

[btheado]

Nice simplification! I wish I knew it was easily reproducible in a bare scroll context. Working via the pldb codebase was a bit hairy. It took a lot of work to simplify it down from that direction.

In order to build only djot.scroll in isolation, it didn't work to move it to a separate directory and build it. Well, it worked, but it still ended up still using the djot.scroll file in the concepts directory. I had to leave it in the concepts directory and remove all the other scroll files. I also had to remove all but the djot entry from pldb.json. Then I started paring down code from Computer.js.

A bit off topic for general scroll...I just wanted to share my experience.

[breck7]

Thank you thank you thank you for that hard work. I could simplify it only because you identified the macro issue.

PLDB is still pretty hacky (the Computer.js I want to eventually remove). And annoyingly slow. I'm working toward a cleanup that delivers 400% better perf. Hopefully tomorrow.

I also think what might be helpful in Scroll is to not just be able to add Parsers to a context, but to be able to remove them as well.

Perhaps something like:

outOfScope [list of parsers]

And then could turn off macros, etc.

like regulatory genes