Follow-on after #7811. Remove the panicking test. It *does* currently
cause a panic if replacement contains scopes (e.g. `() => 123`) but
that's a bug. So we shouldn't have a test saying that it *should* panic.
- Closes https://github.com/oxc-project/oxc/issues/7797
Removing span seems to work, but I'm not sure traversing all the time is
okay.
---------
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Some refactoring works to update
https://github.com/oxc-project/oxc/issues/5068#issuecomment-2507272735
table.
- Implement `array::is_consisely_printed_array()` and use it
- This improved compat-rate a bit ✌🏻
- Align exported function names to align prettier's
- Split `format/mod.rs` into `js`, `jsx` and `typescript`
- Move `format/*.rs` to `format/print/*.rs`
Ref #7456
- Creates a custom action in the language server that applies all
auto-fixes for a given file.
- Updates VS Code to use the custom action with a command to apply fixes
for the currently active text editor.
> zizmor is a static analysis tool for GitHub Actions. It can find many
common security issues in typical GitHub Actions CI/CD setups.
https://woodruffw.github.io/zizmor/
EDIT: what is the right PR-syntax for this?
---------
Co-authored-by: Boshen <boshenc@gmail.com>
Write indentation faster. Previously was writing indentation to buffer with a call to `memset`, which is relatively expensive. Where indentation `<= 16` (common case), write 16 tabs with a single 16-byte XMM write, which is faster.
Code in instance property initializers moves from class body into constructor, or a `_super` function. Update parent `ScopeId`s for first level scopes in initializers.
Refactor only. Does not change behavior. Determine where instance prop initializers will need to be inserted in a separate step before instance props are extracted from class.
This achieves nothing in itself, but is preparatory work for transforming initializers to correct their `ScopeId`s and deal with binding clashes.
This PR support for transforming `super.prop` to `babelHelpers.superPropGet(_B, "prop", _B)`
Input:
```js
class A {
static prop = 1;
}
class B extends A {
static prop = 2;
static propA = super.prop;
static getPropA = () => super.prop;
}
```
Output:
```js
var _B;
class A {}
babelHelpers.defineProperty(A, "prop", 1);
class B extends A {}
_B = B;
babelHelpers.defineProperty(B, "prop", 2);
babelHelpers.defineProperty(B, "propA", babelHelpers.superPropGet(_B, "prop", _B));
babelHelpers.defineProperty(B, "getPropA", () => babelHelpers.superPropGet(_B, "prop", _B));
```
We should move the handling of `<CWD>` to the test runner because this is just only used in testing, and it causes us always get a path by `self.ctx.source_path` like `<CWD>/xxx/xxx.js`, we should get a real path for this.
Fixes: #7809
`ExportNamedDecalration` and `ExportDefaultDeclaration` can reference both type binding and value, so we need to make sure the `ReferenceFlags` is `Read | Type`
Add `Visit::visit_span` and `VisitMut::visit_span` methods, to facilitate #7811.
Both are no-ops by default, and marked `#[inline]`, so this produces no performance impact.
The `--override` flag used to write the output which is generated by the transformer to the `overrides` folder according to the test path. The acting is similar to the previous `takeover` mode
This PR does the following things.
1. Move the override output of the `snapshots` folder to the `overrides` folder.
2. Support `override` mode to replace `takeover` mode
3. The `update_fixtures.js` no longer uses `overrides`'s `output.js` to replace Babel's `output.js`.
### How does `override` mode work?
When running each test, it checks whether an output file for that test exists in the `overrides` directory. If it does, the output file will be used to compare with the transformed code.
In "takeover" mode, transformer conformance test runner was using `HelperLoaderMode::Runtime`. Switch this to `HelperLoaderMode::External` to match standard test runner mode.
Follow-on after #7798. Pure refactor, just shorten code.
I *think* `expect` has same behavior as `unreachable!` in terms of making the compiler backtrack to try to prove the check can be elided. But in this case, I doubt it can prove it either way.
The root cause is due to transform wrongly a PrivateFieldExpression that doesn't contain any optional expression, so call `to_member_expression_mut` causes unwrap to fail. I have fixed the incorrect transform and changed `to_member_expression_mut` to `as_member_expression_mut`.
