tree-sitter-sql VS taplo

Which grammar are you using? https://github.com/DerekStride/tree-sitter-sql supports backticks. This is the one that is used by nvim-treesitter.

A complete grammar of the SQL language is rather large, as can be seen by this grammar https://github.com/DerekStride/tree-sitter-sql/blob/main/grammar.js that is 2000 lines of code.

Alternatively, you could use a tree-sitter parser, such as https://github.com/DerekStride/tree-sitter-sql , which would provide fast incremental parses (but may result in a less detailed AST).

You might find useful this: https://github.com/tamasfe/taplo

Some CLI tools have man pages and in such situations like writing completion spec, people usually use them as a source to learn about the tool and its subcommands options, etc. (Usually they have the best reputation for being the most accurate source of information). Though, the tool I chose to write the completion spec, taplo, did not have any manual entry pages. Even better! I found their website and documentation which was really helpful.

Because I targeted WASM, I needed to roll my own. It simply needs async functions as handlers and a world state that is passed along with the requests to the handlers. (example here). You can use it as an example, the code is not the prettiest, but I'll happily answer questions.