Remove unncessary check for macro call
Since `macro_rules` is a contextual keyword, it is an `IDENT` token and thus `is_path_start` already identifies it correctly. You can tell the previous check is unnecessary because the relevant tests still pass.
internal: Improve readability of the parser code
The code is basically equivalent to the previous version, but it improves the readability by making it much more simpler and concise.
internal: Remove unused keyword from visibility recovery
We removed support `crate` visibility keyword, but forgot to remove it from the recovery token list.
fix: Preserve $ and \ in postfix format completions
`parse_format_exprs` doesn't escape these two as of #16781, so they have to be escaped as a separate step.
fix: Don't invalid body query results when generating desugared names
The hack remains until we get hygiene, but with this the generated names are stable across bodies
fix: Remove accidental dependency between `parse_macro_expansion` and `parse`
Turns out my idea from https://github.com/rust-lang/rust-analyzer/pull/15251 causes all builtin derive expansions to obviously rely on the main parse, meaning the entire `macro_arg` layer becomes kind of pointless. So this reverts that PR again.
internal: Implement parent-child relation for `SourceRoot`s
This commit adds the said relation by keeping a map of type `FxHashMap<SourceRootId,Option<SourceRootId>>` inside the `GlobalState`. Its primary use case is reading `rust-analyzer.toml`(#13529) files that can be placed in every local source root. As a config will be found by traversing this "tree" we need the parent information for every local source root. This commit omits defining this relation for library source roots entirely.
This commit adds the said relation by keeping a map of type `FxHashMap<SourceRootId,Option<SourceRootId>>`
inside the `GlobalState`. Its primary use case is reading the rust-analyzer.toml files that can be
placed under every local source root. As a config will be found by traversing this "tree" we need the parent information
for every local source root. This commit omits defining this relation for library source roots entirely.
fix: panic when using float numbers without dots in chain calls
Fix#16278.
This PR fixes the panic caused by using floating-point numbers without a dot (such as `1e2`) in chain calls.
-------------
Although this syntax is very odd 🤣, r-a should not panic.
fix: keep attributes in assist 'generate_delegate_trait'
fix#15198.
This PR address the issue that `impl` generated by `generate_delegate_trait` doesn't keep attributes.
Add test explorer
This PR implements the vscode testing api similar to #14589, this time using a set of lsp extensions in order to make it useful for clients other than vscode, and make the vscode client side logic simpler (its now around ~100 line of TS code)
Fix#3601
Add `fn index()` and `fn parent_fn()` accessors for `hir::Param`/`hir::SelfParam`
(the PR is motivated by an outside use of the `ra_ap_hir` crate that would benefit from being able to access a `hir::Param`'s/`hir::SelfParam`'s index and parent function)
fix: goto-definition for constants inside range pattern
Fix#15653.
This PR addresses the issue where it was not possible to navigate to constants in range patterns, specifically including two major changes:
1. Previously, both the `start` and `end` fields in `Pat::Range` were of type `LiteralOrConst`. When performing `goto-definition` on constants inside range patterns, r-a would use `resolve_bind_pat_to_const` to find their definitions. However, because the content of a `Const` is not `Pat` but `Path`, it was not stored in the `source_map`, so `resolve_bind_pat_to_const` would returns `None`. This PR changes them to `Const(PatId)`, so that during the lowering process, they are considered as a `pat`, allowing their definitions to be found later through `resolve_bind_pat_to_const`.
2. The process related to range patterns in MIR-lowering has been modified to correctly handle the above changes.
