bollian / tree-sitter-openscad Goto Github PK

List comprehensions can contain many sub-clauses:

if_for_ifelse = [for (x = 0) if (x < 0) for (y = 2) if (y == 2) y else x];
// effectively equivalent to
if_for_ifelse = [for (x = 0) if (x < 0) for (y = 2) (if (y == 2) y else x)];

This means subclauses need to be groupable, and this parser currently doesn't support that. Example:

if_for_ifelse = [for (x = 0) if (x < 0) for (y = 2) (if (y == 2) y) else x];

I noticed that this project does not seem to use semantic versioning. I think it would be valuable if it tagged actual releases so that it's clear what version projects based on this (e.g., the VS Code plugin) can use this project at specific releases.

The releated issue.

Is python required in order to install this? I didn't see anything about this in the docs or issues.

When installing this package as a dependency with npm (npm i tree-sitter-openscad) I am gettting:
npm ERR! gyp ERR! find Python Python is not set from command line or npm configuration

Thanks!

It's surprisingly difficult to publish a clean version of this library because both cargo and npm are very liberal with which parts of the repository they include in the uploaded tarball. The most recent 0.5 and 0.5.1 releases, for example, both include the upstream openscad submodule, including a litany of pngs and other files that users of this library don't actually need. It's also easy for local tooling files to wind up in the archive, which has happened before.

I need to create an action for publishing to both crates.io and NPM that ensures a proper, clean upload.

The CI added by the recent merge of #9 is currently failing. This needs to be fixed.

I've already updated the versions of the used GitHub actions in order to fix some warnings about old node versions, but there's apparently something else wrong with the added tests.

This grammer has semi-colon listed as optional at the end of use/include. It isn't optional it simply isn't necessary at the end of use/include.

include_statement: $ => prec.left(seq(choice('use', 'include'), $.include_path, optional(';'))),

A semi-colon at the end of use/include does compile but that is because ; is a legal statement all by itself. So use <whatever.scad>; is actually two statements. The use <whatever.scad> is one, and then the ; is another statement that just happens to be empty.

There are several test cases in the upstream OpenScad repository which this project currently fails to parse. This can be seen in the expected_failed_parses list in the "test.js" file. These failed parses are all marked as "todo" in the test file so that the CI continues to pass, but these are valid OpenScad files and thus should successfully parse. The ignored_examples list contains files that shouldn't successfully parse, and doesn't represent any issues for this project.

This issue exists to track all the component causes of these "todo" failing parses. Each root-cause issue will be broken out into its own ticket.

Hey @bollian, I have a few questions I think would come under an FAQ.

• Why did you create a grammar for OpenSCAD?
• Is this grammar complete against the grammar specification? I've had a quick look and it seems good.

I'm personally looking to use this library to create an LSP/formatter for vscode using Topiary because the current openscad plugins I use have really poor formatting. I assume it's possible with these tools together.

Thank you for putting the grammar together.

OpenSCAD allows numbers to be formatted using scientific notation with integer exponents. All of the following are allowed:

assert(5.2e1 == 52);
assert(5.2e-1 == 0.52);
assert(-5.2e0 == -5.2);

The parser should handle all of the above cases.