Some struct fields are reserved names e.g. `type`. They are written in source as `r#type`. Trim off the `r#` prefix in `FieldDef::name` and add it back when generating output.
Main motivation is to have the unescaped field names in JSON schema.
I noticed that in JSON schema the `docs` property contains e.g. `" The name of the identifier being referenced."` (with an excess space on the start). Trim that off.
Avoid the `insert!` macro in AST codegen. Use doc comments starting with special symbol `@` instead.
* Before: `insert!("// plain comment");`
* After: `///@ plain comment`
* Or: `//!@ plain comment`
Either `///@` or `//!@` is converted to plain `//` in output.
`//!@` is legal in top-of-file position, which allows us to inline `#![allow(...)]` attributes, which in my opinion makes the generators a bit easier to read.
#4773 makes types like `Span` importable from `oxc_ast::ast`, so remove the imports from other crates in generated code.
I'm not sure why clippy's `wildcard_imports` rule was not being triggered for `use crate::ast::*;`, but add `#[allow(clippy::wildcard_imports)]` on these statements just to make sure.
follow-up to #4735
I was accidentally using the old code to filter the viable target types to derive. It means before this PR we were still using the `#[ast(visit)]` for this purpose.
Remove `Generator::name` and `Pass::name` methods. All impls for these methods return a string identical to the struct name, so can set return value of `Runner::name` in `define_generator!` and `define_pass!` macros instead.
Simplify `derive_get_span` generator that was introduced in #4735. No change to functionality, just aiming for greater readability.
In particular:
* Move defining idents/tokens which are specific to `GetSpan` / `GetSpanMut` into those specific generators, rather than branching on `MUT` later on.
* Remove `const MUT` param.
* Remove the confusing pairs of closures and functions both called `derive_enum` / `derive_struct`.
* Inline function which generates the impls - prioritizing readability over DRY code.
This PR introduces `EarlyCtx` and `LateCtx` in place of the old `CodegenCtx`, Early passes operate at the AST level while generators and other late passes operate on the schema.
It will also replace the confusing `RType` name with something more idiomatic ~~(open for suggestions, I haven't found a good name yet)~~ I've named it `AstType` and dropped the `R` prefix for `REnum` and `RStruct`.
There are some qualities of life improvements too, Things like `to_type_elide` can be used to simplify the code.
Related to #4442 (and can potentially mark it "close as fixed").
Detect types such as `Cell<Option<ScopeId>>` and mark them as such! We didn't used to use this method for these options but now we have to check all types to calculate their layouts which means we need to process them correctly(instead of falling to their inner value).
Closes#4606.
Introduce `GetSpanMut` trait and implement it on all AST node types.
This has to be a separate trait, rather than adding `span_mut` method to `GetSpan` because `AstKind` implements `GetSpan`, and it only has an immutable `&` ref to AST node it contains.
In `AstBuilder`'s `alloc_*` methods, use `Box::new_in` instead of `.into_in`. This is more explicit, so I feel easier to understand. It may also make life easier for compiler by not requiring it to perform type coercion.
Mark everything mentioned in https://github.com/oxc-project/oxc/pull/3815#issuecomment-2186736258 as AST.
We now error on the occurrence of non-ast items in the source of truth. It doesn't make sure that all fields and variants are `#[ast]` and therefore `repr_stable` but there are only a handful of non-AST types used here(mainly Atom and Span). Since we don't have access to the external types we can't make sure of it unless we find a way to const assert it.
The best we can do until then is to check all field/variant types to be either `#[ast]` or in a white list. I can add this check to the codegen in an upcoming PR.
Add `#[inline]` to empty default implementations of `enter_node` etc. Hopefully compiler will inline them automatically within Oxc even cross-crate because we compile with LTO, but maybe not for external consumers who don't use LTO.
