oliverxchen / sqlean Goto Github PK

A safe mode CLI option should test whether the parse tree of the original query is equivalent to the parse tree of the styled query.

Hello Oliver! I'm the maintainer of the sqlean project, which started in 2021 and is being actively developed till nowadays. The project reached the point where I'd like to publish it as a PyPI package, but found the sqlean name taken by your project.

As your project seems abandoned and not used by anyone (sorry if I'm wrong) — would you, by any chance, agree to transfer the sqlean package name to my project?

Sorry if I'm asking too much.

The jinja parts are just parsed as string blobs without structure. These should be properly parsed.

sqlean is currently using lark v0.12.0. Parsing breaks when upgrading to lark v1.0.0. Lark release are here: https://github.com/lark-parser/lark/releases

The current way of considering comments is not scalable for the grammar. From the lark docs:

Using the %ignore directive results in a cleaner grammar.

It’s especially important for the LALR(1) algorithm, because adding whitespace (or comments, or other extraneous elements) explicitly in the grammar, harms its predictive abilities, which are based on a lookahead of 1.

Currently, comments are explicitly included in the grammar and then styled appropriately. A potential alternative strategy is to:

1. Use the %ignore directive for comments when parsing with lark.
2. Do a regex pass to find the starting location of all comments
3. Add the comments to the parse tree.