Patch Parsec's comment parsing

[cprecioso]

Part of #2949

Problem

Parsec...

When creating a parser with Parsec, you can create a base lexer over which to build up your high-level parser. This lexer accepts options for describing how comments work in the language, and will then handle them transparently as part of the whiteSpace parser.

However, the comment options only accept Strings. Parsec will immediately turn those strings into Parser Strings by running Text.Parsec.string commentLine, but it doesn't accept parsers itself.

... and Prisma

The Prisma Schema Language (PSL), has two kinds of comments: regular comments that can be ignored (start with //), and documentation comments that have to become part of the AST (start with ///).

If we just tell Parsec that comments start with //, it will ignore the documentation comments too. But if we don't let Parsec handle the comments automatically, we will have to remember to put optional comment handling in all of our parsers, spamming it semi-randomly through the codebase, and risking forgetting it some place, or not creating a test for a specific comment placements.

Solution

We need to be able to tell Parsec that the comment definition is

commentLine :: Parser ()
commentLine = string "//" >> notFollowedBy (char '/')

The patch inside of Parsec to accept that is just a couple of lines. But to do it I've needed to copy the Text.Parsec.Language and Text.Parsec.Token modules into our codebase, and modify them so they accept a Parser for comment symbols instead of only a string.

[cloudflare-workers-and-pages]

Deploying wasp-docs-on-main with    Cloudflare Pages

Latest commit:	d47e94a
Status:	✅  Deploy successful!
Preview URL:	https://c0ab840a.wasp-docs-on-main.pages.dev
Branch Preview URL:	https://push-tmkmvmvnruvk.wasp-docs-on-main.pages.dev

View logs

[cprecioso]

This is the main change, this used to be

-        _ <- try (string (commentLine languageDef))
+        _ <- try (commentLine languageDef)

[cprecioso]

If we merge this solution, I will create an issue and link to here so that we know that we must remove this workaround