fix: rewrite code_action `generate_delegate_trait`
I've made substantial enhancements to the "generate delegate trait" code action in rust-analyzer. Here's a summary of the changes:
#### Resolved the "Can’t find CONST_ARG@158..159 in AstIdMap" error
Fix#15804, fix#15968, fix#15108
The issue stemmed from an incorrect application of PathTransform in the original code. Previously, a new 'impl' was generated first and then transformed, causing PathTransform to fail in locating the correct AST node, resulting in an error. I rectified this by performing the transformation before generating the new 'impl' (using make::impl_trait), ensuring a step-by-step transformation of associated items.
#### Rectified generation of `Self` type
`generate_delegate_trait` is unable to properly handle trait with `Self` type.
Let's take the following code as an example:
```rust
trait Trait {
fn f() -> Self;
}
struct B {}
impl Trait for B {
fn f() -> B { B{} }
}
struct S {
b: B,
}
```
Here, if we implement `Trait` for `S`, the type of `f` should be `() -> Self`, i.e. `() -> S`. However we cannot automatically generate a function that constructs `S`.
To ensure that the code action doesn't generate delegate traits for traits with Self types, I add a function named `has_self_type` to handle it.
#### Extended support for generics in structs and fields within this code action
The former version of `generate_delegate_trait` cannot handle structs with generics properly. Here's an example:
```rust
struct B<T> {
a: T
}
trait Trait<T> {
fn f(a: T);
}
impl<T1, T2> Trait<T1> for B<T2> {
fn f(a: T1) -> T2 { self.a }
}
struct A {}
struct S {
b$0 : B<A>,
}
```
The former version will generates improper code:
```rust
impl<T1, T2> Trait<T1, T2> for S {
fn f(&self, a: T1) -> T1 {
<B as Trait<T1, T2>>::f( &self.b , a)
}
}
```
The rewritten version can handle generics properly:
```rust
impl<T1> Trait<T1> for S {
fn f(&self, a: T1) -> T1 {
<B<A> as Trait<T1>>::f(&self.b, a)
}
}
```
See more examples in added unit tests.
I enabled support for generic structs in `generate_delegate_trait` through the following steps (using the code example provided):
1. Initially, to prevent conflicts between the generic parameters in struct `S` and the ones in the impl of `B`, I renamed the generic parameters of `S`.
2. Then, since `B`'s parameters are instantiated within `S`, the original generic parameters of `B` needed removal within `S` (to avoid errors from redundant parameters). An important consideration here arises when Trait and B share parameters in `B`'s impl. In such cases, these shared generic parameters cannot be removed.
3. Next, I addressed the matching of types between `B`'s type in `S` and its type in the impl. Given that some generic parameters in the impl are instantiated in `B`, I replaced these parameters with their instantiated results using PathTransform. For instance, in the example provided, matching `B<A>` and `B<T2>`, where `T2` is instantiated as `A`, I replaced all occurrences of `T2` in the impl with `A` (i.e. apply the instantiated generic arguments to the params).
4. Finally, I performed transformations on each assoc item (also to prevent the initial issue) and handled redundant where clauses.
For a more detailed explanation, please refer to the code and comments. I welcome suggestions and any further questions!
fix: self type replacement in inline-function
Fix#16113, fix#16091
The problem described in this issue actually involves three bugs.
Firstly, when using `ted` to modify the syntax tree, the offset of nodes on the tree changes, which causes the syntax range information from `hir` to become invalid. Therefore, we need to edit the AST after the last usage for `usages_for_locals`.
The second issue is that when inserting nodes, it's necessary to use `clone_subtree` for duplication because the `ted::replace` operation essentially moves a node.
The third issue is that we should use `ancestors_with_macros` instead of `ancestors` to handle impl definition in macros.
I have fixed the three bugs mentioned above and added unit tests.
internal: Migrate assists to the structured snippet API, part 5
Continuing from #15874
Migrates the following assists:
- `extract_variable`
- `generate_function`
- `replace_is_some_with_if_let_some`
- `replace_is_ok_with_if_let_ok`
fix: no code action 'introduce_named_generic' for impl inside types
Fix#15734.
### Changes Made
- Find params in `ancestors` instead of just `parent`
- Added tests (`replace_impl_with_mut` and `replace_impl_inside`)
Fix incorrectly replacing references in macro invocation in "Convert to named struct" assist
Fixes#15630.
Complements #13647 (same assist but missed this one), #14920 (inverse action assist).
fix: resolve Self type references in delegate method assist
This PR makes the delegate method assist resolve any `Self` type references in the parameters or return type. It also works across macros such as the `uint_impl!` macro used for `saturating_mul` in the issue example.
Closes#14485
internal: Migrate assists to the structured snippet API, part 4
Continuing from #15260
Migrates the following assists:
- `add_turbo_fish`
- `add_type_ascription`
- `destructure_tuple_binding`
- `destructure_tuple_binding_in_subpattern`
I did this a while ago, but forgot to make a PR for the changes until now. 😅
Due to the way the current tree mutation api works, we need to collect
changes before we can apply them to the real syntax tree, and also can only
switch to a file once.
`destructure_tuple_binding_in_sub_pattern` also gets migrated even
though can't be used.
feat: implement tuple return type to tuple struct assist
This PR implements the `convert_tuple_return_type_to_struct` assist, for converting the return type of a function or method from a tuple to a tuple struct. Additionally, it moves the `to_camel_case` and `char_has_case` functions from `case_conv` to `stdx` so that they can be used similar to `to_lower_snake_case`.
[tuple_return_type_to_tuple_struct.webm](https://github.com/rust-lang/rust-analyzer/assets/52933714/2803ff58-fde3-4144-9495-7c7c7e139075)
Currently, the assist puts the struct definition above the function, or above the nearest `impl` or `trait` if applicable and only rewrites literal tuples that are returned in the body of the function. Additionally, it only attempts to rewrite simple tuple pattern usages with the corresponding tuple struct pattern but does so across files and modules.
I think that this is sufficient for the majority of use cases but I could be wrong. One thing I'm still not sure how to approach is handling `Self` and generics/lifetimes in the tuple type to be extracted. I was thinking of either manually figuring out what lifetimes and generics are in scope and using them (sort of similar to the `generate_function` assist) or maybe using `ctx.sema.resolve_type` and `generic_params` on `hir::Type` but this seems to not deal with lifetimes.
Closes#14293
fix: make bool_to_enum assist create enum at top-level
This pr makes the `bool_to_enum` assist create the `enum` at the next closest module block or at top-level, which fixes a few tricky cases such as with an associated `const` in a trait or module:
```rust
trait Foo {
const $0BOOL: bool;
}
impl Foo for usize {
const BOOL: bool = true;
}
fn main() {
if <usize as Foo>::BOOL {
println!("foo");
}
}
```
Which now properly produces:
```rust
#[derive(PartialEq, Eq)]
enum Bool { True, False }
trait Foo {
const BOOL: Bool;
}
impl Foo for usize {
const BOOL: Bool = Bool::True;
}
fn main() {
if <usize as Foo>::BOOL == Bool::True {
println!("foo");
}
}
```
I also think it's a bit nicer, especially for local variables, but didn't really know to do it in the first PR :)
fix: panic with wrapping/unwrapping result return type assists
With the `wrap_return_type_in_result` assist, the following code results in a panic (note the lack of a semicolon):
```rust
fn foo(num: i32) -> $0i32 {
return num
}
=>
thread 'handlers::wrap_return_type_in_result::tests::wrap_return_in_tail_position' panicked at crates/syntax/src/ted.rs:137:41:
called `Option::unwrap()` on a `None` value
```
I think this is because it first walks the body expression to change any `return` expressions and then walks all tail expressions, resulting in the `return num` being changed twice since it is both a `return` and in tail position. This can also happen when a `match` is in tail position and `return` is used in a branch for example. Not really sure how big of an issue this is in practice though since this seems to be the only case that is impacted and can be reduced to just `num` instead of `return num`.
This also occurs with the `unwrap_result_return_type` assist but panics with the following instead:
```
thread 'handlers::unwrap_result_return_type::tests::wrap_return_in_tail_position' panicked at /rustc/3223b0b5e8dadda3f76c3fd1a8d6c5addc09599e/library/alloc/src/string.rs:1766:29:
assertion failed: self.is_char_boundary(n)
```
minor : Deunwrap convert_comment_block and desugar_doc_comment
Closes subtask 13 of #15398 . I still don't know a more idiomatic way for the for loops I added, any suggestion would make me happy.
Although it doesn't panic now, further changes to how we recover from incomplete syntax
may cause this assist to panic. To mitigate this a test case has been added.
feat: Bool to enum assist
This adds the `bool_to_enum` assist, which converts the type of boolean local variables, fields, constants and statics to a new `enum` type, making it easier to distinguish the meaning of `true` and `false` by renaming the variants.
Closes#14779
Field shorthand overwritten in promote local to const assist
Currently, running `promote_local_to_const` on the following:
```rust
struct Foo {
bar: usize,
}
fn main() {
let $0bar = 0;
let foo = Foo { bar };
}
```
Results in:
```rust
struct Foo {
bar: usize,
}
fn main() {
const BAR: usize = 0;
let foo = Foo { BAR };
}
```
But instead should be something like:
```rust
struct Foo {
bar: usize,
}
fn main() {
const BAR: usize = 0;
let foo = Foo { bar: BAR };
}
```
Bind unused parameter assistant
This PR introduces a new **Bind unused parameter assistant**.
While we do have a QuickFix from `rustc` (prefixing the parameter with an underscore), it's sometimes more convenient to suppress the warning using the following approach:
```rust
fn some_function(unused: i32) {}
```
->
```rust
fn some_function(unused: i32) {
let _ = unused;
}
```
minor : Deunwrap generate_derive
#15398 subtask 1. Since the editing closure has arms, I did something *experimental* ( in this case just a clever term for bad code ) to bypass creating an `Option` but I am ready to change this.
the "add missing members" assists: implemented substitution of default values of const params
To achieve this, I've made `hir::ConstParamData` store the default values
internal : rewrite DeMorgan assist
fixes#15239 , #15240 . This PR is a rewrite of the DeMorgan assist that essentially rids of all the string manipulation and modifies syntax trees to apply demorgan on a binary expr. The main reason for the rewrite is that I wanted to use `Expr::needs_parens_in` method to see if the expr on which the assist is applied would still need the parens it had once the parent expression's operator had equal precedence with that of the expression. I used `.clone_(subtree|for_update)` left and right and probably more than I should have, so I would also be happy to hear how I could have prevented redundant cloning.
internal: Turn unresolved proc macro expansions into missing expressions
Reduces the amount of type related errors one gets when proc macro expansion is disabled.
Add ExternCrateDecl to HIR
Adding these doesn't really require much design effort as they represent a single import, unlike use trees which are one item that represent 0 or more imports.
We only resolve to this definition when actually resolving on the name or alias of an `extern crate name as alias` item, not usages yet as that requires far more changes that won't lead anywhere without giving it more thought. Nevertheless the changes slightly improve IDE things, an example being hover on the decl showing the merged doc comments for example.
cc https://github.com/rust-lang/rust-analyzer/issues/14079
Added remove unused imports assist
This resolves the most important part of #5131. I needed to make a couple of cosmetic changes to the search infrastructure to do this.
A few open questions:
* Should imports that don't resolve to anything be considered unused? I figured probably not, but it would be a trivial change to make if we want it.
* Is there a cleaner way to make the edits to the use list?
* Is there a cleaner way to get the list of uses that intersect the current selection?
* Is the performance acceptable? When testing this on itself, it takes a good couple seconds to perform the assist.
* Is there a way to hide the rustc diagnostics that overlap with this functionality?
internal: Defer structured snippet rendering to allow escaping snippet bits
Since we know exactly where snippets are, we can transparently escape snippet bits to the exact text edits that need it, and not have to do it for anything other text edits.
Also will eventually fix#11006 once all assists are migrated. This comes as a side-effect of text edits that don't have snippets get marked as having no insert formatting at all.
Don't provide `add_missing_match_arms` assist when upmapping match arm list failed
Fixes#15310
We shouldn't provide the assist when we fail to find the original match arm list.
Note that this PR will temporarily make the assist not applicable when attribute macro operates on the match expression in question, just like the case in #15310, for most of the current stable toolchain users. This is because the sysroot-abi proc-macro-srv on the current stable [discards] spans for `Group` delimiters in some code paths, which the popular `proc-macro2` crate almost always calls, and it makes the identity of match arm list's brackets lost, leading to the upmapping failure. This has been fixed by #14960, which will land in the next stable, 1.71.
[discards]: 8ede3aae28/src/tools/rust-analyzer/crates/proc-macro-srv/src/abis/abi_sysroot/ra_server.rs (L231)
bugfix : skip doc(hidden) default members
fixes #14957 . I have two questions :
1. I am definitely looking for a more idiomatic way for the things I added in `crates/ide-assists/src/utils.rs`. See `FIXME` in that file.
2. Would it be actually better to change `DefaultMethods` to something like
```rust
enum DefaultMethods {
Only( IgnoreHidden ( bool ) ) ,
None
}
```
instead of adding a boolean to every function that calls `crates/ide-assists/src/utils.rs::filter_assoc_items`
internal: Migrate assists to the structured snippet API, part 3
Continuing from #15231
Migrates the following assists:
- `add_missing_match_arms`
- `fix_visibility`
- `promote_local_to_const`
The `add_missing_match_arms` changes are best reviewed commit-by-commit since they're relatively big changes compared to the rest of the commits.
`clone_for_update` is relatively cheap in comparison, since making a
node require parsing an entire source text
Adds a test to make sure that it doesn't crash when multiple uses are
present.
internal: Migrate more assists to use the structured snippet API
Continuing from #14979
Migrates the following assists:
- `generate_derive`
- `wrap_return_type_in_result`
- `generate_delegate_methods`
As a bonus, `generate_delegate_methods` now generates the function and impl block at the correct indentation 🎉.
`does_not_fill_wildcard_with_wildcard`
and `does_not_fill_wildcard_with_partial_wildcard_and_wildcard`
both made no modifications to the code,
which is a problem for mutable ast porting as it generates a best-effort
minimal set of text edits,
and assists require at least one text edit.
# Overview
Extracting a match arm value that has type unit into a function, when a
comma already follows the match arm value, results in an invalid (syntax
error) semicolon added between the newly generated function's generated
call and the comma.
# Example
Running this extraction
```rust
fn main() {
match () {
_ => $0()$0,
};
}
```
would lead to
```rust
fn main() {
match () {
_ => fun_name();,
};
}
fn fun_name() {
}
```
# Issue / Fix details
This happens because when there is no comma, rust-analyzer would simply
add the comma and wouldn't even try to evaluate whether it needs to add
a semicolon. But when the comma is there, it proceeds to evaluate
whether it needs to add a semicolon and it looks like the evaluation
logic erroneously ignores the possibility that we're in a match arm.
IIUC it never makes sense to add a semicolon when we're extracting from
a match arm value, so I've adjusted the logic to always decide against
adding a semicolon when we're in a match arm
Can actually split out adding the functions from getting the impl to
update or create thanks to being able to refer to the impl ast node.
FIXME Context:
Unfortunately we can't adjust the indentation of the newly added function
inside of `ast::AssocItemList::add_item` since for some reason the `todo!()`
placeholder generated by `add_missing_impl_members` and
`replace_derive_with_manual_impl` gets indented weirdly.
Unify getter and setter assists
This PR combines what previously have been two different files into a single file. I want to talk about the reasons why I did this. The issue that prompted this PR ( and before I forget : this pr fixes#15080 ) mentions an interesting behavior. We combine these two assists into an assist group and the order in which the assists are listed in this group changes depending on the text range of the selected area. The reason for that is that VSCode prioritizes actions that have a bigger impact in a smaller area and until now generate setter assist was only possible to be invoked for a single field whereas you could generate multiple getters for the getter assist. So I used the latter's infra to make former applicable to multiple fields, hence the unification. So this PR solves in essence
1. Make `generate setter` applicable to multiple fields
2. Provide a consistent order of the said assists in listing.
assist : generate trait from impl
fixes#14987 . As the name suggests this assist is used to generate traits from inherent impls while adapting the original impl to fit to the newly generated trait. I made some decisions regarding when the assist should be applicable. These are surely open to discussion. I looking forward to any feedback.
internal: Format let-else
As nightly finally got support for it I went ahead and formatted r-a with the latest nightly, then with the latest stable (in case other stuff changed)
Clean up `ImportMap`
There are several things in `hir_def::import_map` that are never used. This PR removes them and restructures the code. Namely:
- Removes `Query::name_only`, because it's *always* true.
- Because of this, we never took advantage of storing items' full path. This PR removes `ImportPath` and changes `ImportInfo` to only store items' name, which should reduce the memory consumption to some extent.
- Removes `SearchMode::Contains` for `Query` because it's never used.
- Merges `Query::assoc_items_only` and `Query::exclude_import_kinds` into `Query::assoc_mode`, because the latter is never used besides filtering associated items out.
Best reviewed one commit at a time. I made sure each commit passes full test suite. I can squash the first three commits if needed.
feature : assist delegate impl
This PR ( fixes#14386 ) introduces a new IDE assist that generates a trait impl for a struct that delegates a field. This is a draft because the current `ide_db::path_transform::PathTransform` produces some unwanted results when it deals with extern crates, an example of which I attach as a GIF.
GIFs :
1. A general case
2. A case where `ide_db::path_transform::PathTransform` fails to correctly resolve a property ( take `Allocator` as an example ) to its full path, thus causing an error to occur. ( Not to even mention that resolving this causes another error `use of unstable library feature 'allocator_api'` to occur
Remove scope_for_def calls as the definition have been removed entirely.
As a result of this change the problem with false path resolutions has been solved.
Lower const params with a bad id
cc #7434
This PR adds an `InTypeConstId` which is a `DefWithBodyId` and lower const generic parameters into bodies using it, and evaluate them with the mir interpreter. I think this is the last unimplemented const generic feature relative to rustc stable.
But there is a problem: The id used in the `InTypeConstId` is the raw `FileAstId`, which changes frequently. So these ids and their bodies will be invalidated very frequently, which is bad for incremental analysis.
Due this problem, I disabled lowering for local crates (in library crate the id is stable since files won't be changed). This might be overreacting (const generic expressions are usually small, maybe it would be better enabled with bad performance than disabled) but it makes motivation for doing it in the correct way, and it splits the potential panic and breakages that usually comes with const generic PRs in two steps.
Other than the id, I think (at least I hope) other parts are in the right direction.
fix: implemeted lifetime transformation fot assits
A part of https://github.com/rust-lang/rust-analyzer/issues/13363
I expect to implement transformation of const params in a separate PR
Other assists and a completion affected:
- `generate_function` currently just ignores lifetimes and, consequently, is not affected
- `inline_call` and `replace_derive_with...` don't seem to need lifetime transformation
- `trait_impl` (a completion) is fixed and tested
internal: Migrate some assists to use the structured snippet API
Migrates the following assists:
- `add_missing_impl_members`
- `extract_type_alias`
As an additional requirement, these assists are also migrated to use the mutable AST API, since otherwise there would be overlapping `Indel` spans
The change updates the logic to determine if a function parameter is
valid for replacing the type param with the trait implementation.
First all usages are determined, to check if they are used outside the function
parameter list. If an outside reference is found, e.g. in body, return type or
where clause, the assist is skipped. All remaining usages only appear in the
function param list. For each usage the param type is checked to see if
it's valid.
**Please note** the logic currently follows a heuristic and may not cover
all existing parameter declarations.
* determine valid usage references by checking ancestors (on AST level)
* split test into separate ones
Fix edits for `convert_named_struct_to_tuple_struct`
Two fixes:
- When replacing syntax nodes, macro files weren't taken into account. Edits were simply made for `node.syntax().text_range()`, which would be wrong range when `node` is inside a macro file.
- We do ancestor node traversal for every struct name reference to find record expressions/patterns to edit, but we didn't verify that expressions/patterns do actually refer to the struct we're operating on.
Best reviewed one commit at a time.
Fixes#13780Fixes#14927
Previously we didn't verify that record expressions/patterns that were
found did actually point to the struct we're operating on. Moreover,
when that record expressions/patterns had missing child nodes, we would
continue traversing their ancestor nodes.
This removes an existing generic param from the `GenericParamList`. It
also considers to remove the extra colon & whitespace to the previous
sibling.
* change order to get all param types first and mark them as mutable
before the first edit happens
* add helper function to remove a generic parameter
* fix test output
This adds a new assist named "replace named generic with impl" to move
the generic param type from the generic param list into the function
signature.
```rust
fn new<T: ToString>(input: T) -> Self {}
```
becomes
```rust
fn new(input: impl ToString) -> Self {}
```
The first step is to determine if the assist can be applied, there has
to be a match between generic trait param & function paramter types.
* replace function parameter type(s) with impl
* add new `impl_trait_type` function to generate the new trait bounds with `impl` keyword for use in the
function signature
fix: assists no longer break indentation
Fixes https://github.com/rust-lang/rust-analyzer/issues/14674
These are _ad hoc_ patches for a number of assists that can produce incorrectly indented code, namely:
- generate_derive
- add_missing_impl_members
- add_missing_default_members
Some general solution is required in future, as the same problem arises in many other assists, e.g.
- replace_derive_with...
- generate_default_from_enum...
- generate_default_from_new
- generate_delegate_methods
(the list is incomplete)
Restrict "sort items" assist for traits & impls
This restricts the "sort items alphabetically" assist when the selection is inside a `Impl` or `Trait` node & intersects with one of the associated items.
It re-orders the conditional checks of AST nodes in the `sort_items` function to check for more specific nodes first before checking `Trait` or `Impl` nodes. The `AssistContext` is passed into the `add_sort_methods_assist` function to check if the selection intersects with any inner items, e.g. associated const or type alias, function. In this case the assist does not apply.
Fixes: #14516
This fixes the applicability of the "sort items alphabetically" assist
when the selection is inside a `Trait` or `Impl`. It's now tested if the
selection is inside or overlaps with an inner node, e.g. associated
const or type alias, function.
minor: Fix some simple FIXMEs
Each FIXME fix has been split into its own commit, since they're all pretty independent changes.
(Forgot to open a PR for this a few days ago, oops)
