1021: Wasm dependencies r=matklad a=detrumi
As a first step towards running RA on WASM (see #1007), this tweaks the dependencies somewhat so that projects built using `wasm-pack` can use `ra_ide_api` as a dependency.
There were two problems:
- use of undeclared type or module `MmapInner`
This error occurred because of the `memmap` crate, as a dependency of `fst`
Solution: specify `default-features = false` for the `fst` package (see https://github.com/BurntSushi/fst/issues/70)
- use of undeclared type or module `imp`
This happened in the `wait-timeout` crate ([which uses `Command` under the hood](https://github.com/alexcrichton/wait-timeout/issues/18)), a dependency of `rusty-fork` which is a dependency of `proptest`.
Solution: move `proptest` to dev-dependencies and add `#[cfg(test)]` to the `test_utils` crate.
**Edit:** Oh, that causes trouble with resolving the import when running the tests. Hmm...
Co-authored-by: Wilco Kusee <wilcokusee@gmail.com>
1017: line_index and line_index_utils moved to ra_ide_api r=matklad a=pasa
line_index and line_index_utils for issue #1009
Co-authored-by: Sergey Parilin <sergey.parilin@fxdd.com>
1013: Fuzz reparsing and fix found bugs r=matklad a=pcpthm
Add fuzz test for reparsing which:
- Checks reparsing doesn't panic and validate result syntax tree.
- Checks that incremental reparsing produces the same syntax tree as full reparse.
- Check for that errors are the same as full reparsing is disabled because errors are less important than syntax tree and produce failures which I couldn't figure out how to fix immediately (FIXME comment).
I guess the current input generation is inefficient but still found several bugs:
- Arithmetic overflow (negative result on an unsigned type). I changed the signature of `SyntaxError::add_offset` to solve this problem.
- When reparsing a leaf, the token of the leaf can be joined to the next characters. Such case was not considered.
- UNDERSCORE token was not produced when text length is exactly 1 (not a reparsing bug).
- When reparsing a block, *inner* curly braces should be balanced. i.e. `{}{}` is invalid.
- Effects of deleting newlines were not considered.
Co-authored-by: pcpthm <pcpthm@gmail.com>
997: Improve filtering of file roots r=matklad a=vipentti
`ProjectWorkspace::to_roots` now returns a new `ProjectRoot` which contains
information regarding whether or not the given path is part of the current
workspace or an external dependency. This information can then be used in
`ra_batch` and `ra_lsp_server` to implement more advanced filtering. This allows
us to filter some unnecessary folders from external dependencies such as tests,
examples and benches.
Relates to discussion in #869
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
1003: make Name::new private r=flodiebold a=matklad
This maybe is overengineering, but it seems cool to keep names completely opaque.
r? @flodiebold
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
`ProjectWorkspace::to_roots` now returns a new `ProjectRoot` which contains
information regarding whether or not the given path is part of the current
workspace or an external dependency. This information can then be used in
`ra_batch` and `ra_lsp_server` to implement more advanced filtering. This allows
us to filter some unnecessary folders from external dependencies such as tests,
examples and benches.
996: Allow attributes on top level expressions r=matklad a=pcpthm
This PR modifies parser to allow outer attributes on top level expression. Here, top level expression means either
- Expression statement e.g. `foo();`
- Last expression in a block without semicolon `bar()` in `{ foo(); bar() }`.
Except for binary operation expressions and `if` expressions, which are errors (feature gated) in rustc.
Attributes on inner expressions like `foo(#[a] 1)` are not implemented.
I first tried to implement this by passing `Maker` to expression parsers. However, this implementation couldn't parse `#[attr] foo()` correctly as `CallExpr(Attr(..), PathExpr(..), ArgList(..))` and instead parsed incorrectly as `CallExpr(PathExpr(Attr(..), ..), ArgList(..))` due to the way left recursion is handled.
In the end, I introduce `undo_completion` method. Which is not the suggested approach, but it seems not very bad.
Fix#759.
Co-authored-by: pcpthm <pcpthm@gmail.com>
993: Fix installing vscode extension on MacOS r=matklad a=funkill
VSCode often installed in MacOS as `Visual Studio Code.app` package and `code` binary located at `Contents/Resources/app/bin` in package. This path not exists in `$PATH` variable and we can't run `code`.
In previous version of `do_run` function all before space was command and all after - arguments. If path or command has spaces, extracting command breaks. To fix this i extracted command to separated argument of function.
All packages can be placed in system app dir (`/Applications`) or user app dir (`~/Applications`). I created helper function for find app in this directories.
Co-authored-by: funkill2 <funkill2@gmail.com>
991: Use Marker argument for item parsers r=matklad a=pcpthm
Before doing this for expressions, I found that the pattern (Marker argument) should be applied to the item parsers because visiblity and modifiers are parsed in a separate function.
Fixed some parser bugs:
- Fix pub_expr: `pub 42;` was allowed.
- Fix incorrect parsing of crate::path: incorrectly parsed as `crate` as a visibility.
Co-authored-by: pcpthm <pcpthm@gmail.com>
989: Implement naive version of fill_struct_fields assist r=matklad a=yanchith
Fixes#964
This implements the `fill_struct_fields` assist. Currently only works for named struct fields, but not for tuple structs, because we seem to be missing a `TupleStructLit` (akin to `StructLit`, but for tuple structs). I am happy to implement `TupleStructLit` parsing given some guidance (provided it's really missing) and make the assist work for tuple structs as well. Could do so either in this PR, or another one 🙂
Sorry if I missed something important, this is my first PR for Rust Analyzer.
Btw is there any way to run the assists in emacs?
UPDATE: I just realized that parsing `TupleStructLit` would be quite difficult as it it really similar, if not identical to a function call...
Co-authored-by: yanchith <yanchi.toth@gmail.com>
987: Refactor maybe_item to use Marker argument r=pcpthm a=pcpthm
As suggested at <https://github.com/rust-analyzer/rust-analyzer/pull/980#issuecomment-473659745>.
For expression paring functions, changing signature
- from `fn(&mut Parser) -> Option<CompletedMarker>` to `fn(&mut Parser, Marker) -> Result<CompletedMarker, Marker>`
- from `fn(&mut Parser) -> CompletedMarker` to `fn(&mut Parser, Marker) -> CompletedMarker`
is my plan.
Co-authored-by: pcpthm <pcpthm@gmail.com>
983: support remainder assignment operator r=matklad a=JeanMertz
`%=` was returning errors for me, turns out it wasn't added as a valid assignment operation.
I'm not sure what the best location would be to add a test for this. Please let me know and I'll add one.
Co-authored-by: Jean Mertz <jean@mertz.fm>
Asymptotically computing a set difference is faster but in the average
case we won't have more than ~10 functions. Also prefer not using hash
sets as these may yield nondeterministic results.
Implement `BindingMode` for pattern matching, so that types can be
correctly inferred using match ergonomics. The binding mode defaults to
`Move` (referred to as 'BindingMode::BindByValue` in rustc), and is
updated by automatic dereferencing of the value being matched.
If the client doesn't specify this explicitly, that very likely means it doesn't
know about it and so we shouldn't send decorations. In particular, the recent
change to this default caused decorations to be sent to emacs, resulting in a
lot of warning spam.
948: Fix test_missing_module_code_action_in_json_project on Windows r=matklad a=vipentti
The test would fail on Windows due to the paths not being properly escaped for
JSON.
In addition adds extra braces around the fn main to actually introduce braces in
the file.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
The test would fail on Windows due to the paths not being properly escaped for
JSON.
In addition adds extra braces around the fn main to actually introduce braces in
the file.
939: Initial implementation of project-lock.json. r=davidtwco a=davidtwco
Fixes#792.
This PR adds a initial implementation of project-lock.json, a build
system agnostic method of specifying the crate graph and roots.
Co-authored-by: David Wood <david@davidtw.co>
933: Check installed extension r=matklad a=c410-f3r
Fixes#918.
Edit: Windows encoding for Unicode is UTF-16 so String::from_utf8 will probably fail unless `Vec<u8>` is already UTF-8 somehow, which I don't know for sure.
Co-authored-by: Caio <c410.f3r@gmail.com>
This allows users to control whether or not they want to see the "workspace
loaded" notification.
This is done on the server side using InitializationOptions which are provided
by the client. By default show_workspace_loaded is true, meaning the
notification is sent.
This now allows us to send a notification that can be shown in the UI when the
workspace has been loaded.
Additionally this removes the need for internal_mode flag.
912: Make goto definition/hover work for associated items r=matklad a=kjeremy
Just functions so far. Looking for comments.
Fixes#911
Towards #832
Co-authored-by: kjeremy <kjeremy@gmail.com>
Co-authored-by: Jeremy Kolb <kjeremy@gmail.com>
We simply remove all the CUSTOM_MARKERS before attempting to parse the file.
This allows for the syntax selection to work with most of the test strings.
This allows us to select a string or portions of it and try parsing it as rust
syntax. This is mostly helpful when developing tests where the test
itself contains some rust syntax as a string.
908: Enable markup for hover on expressions which resolve using type_of r=matklad a=vipentti
This adds highlighting when hovering over items which are resolved using
`type_of`.
This adds basic highlighting, discussed in #904.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
915: Bring BodySyntaxMapping in line with other source-map instances r=flodiebold a=matklad
* rename to SourceMap
* don't store the actual body inline, just return a pair
r? @flodiebold
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
906: Add support for goto definition and hover for struct fields r=matklad a=vipentti
This works partially towards fixing #512
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
900: Add new trait ast::TypeAscriptionOwner r=vipentti a=vipentti
This trait should be implemented for nodes which have an ascribed type,
e.g. thing : Type. Such as let, const, static, param, named struct fields.
In addition, we update some places where previously we used node + node.type_ref() with `TypeAscriptionOwner` in the trait bounds.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
901: Add basic support for showing fn signature when hovering r=matklad a=vipentti
This adds basic support for displaying function signature when hovering over a usage of a function.
Additionally refactored `hover` to return `HoverResult` to ease with testing and in general to be more robust.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
vscode would report "A request has failed" when it got "Content modified"
message and this would cause a pop-up to appear. This works around the issue by
returning an "empty" response that vscode can ignore.
This replaces places where we would use node + node.type_ref() with things that
have an ascribed type, with using the TypeAscriptionOwner as the trait bound so
we can simply pass the node.
897: Add basic const/static type inference r=flodiebold a=vipentti
This adds basic const/static type inference discussed in #887.
Currently the inference does not work for const/static declared inside a block. In addition the inference does not work inside the bodies of const/static.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
885: Parse token trees directy r=matklad a=matklad
This takes advantage of the recent macro refactoring to directly parse token stream into a syntax tree.
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
It was just getting too big. We now have:
- ty: the `Ty` enum and helpers
- ty::infer: actual type inference
- ty::lower: lowering from HIR to `Ty`
- ty::op: helpers for binary operations, currently
879: Fixes to goto definition r=vipentti a=vipentti
Previously goto definition would fail when the cursor was over the name of the definition. Now we should properly resolve to a `NavigationTarget` when on top of the name of a definition.
In addition this adds `name_range` field to `FileSymbol`, this further fixes goto_definition and symbol based navigation by allowing the `NavigationTarget` to actually have a `focus_range`, meaning instead of focusing on the start of the `full_range`, we can have the cursor focus on the name.
e.g. goto definition
```rust
fn bar() {
fn foo() { }
foo<|>();
}
```
Previously this would put the cursor at the start of the FN_DEF:
```rust
fn bar() {
<|>fn foo() { }
foo();
}
```
Now when using the symbol based resolving, we'll have a proper focus range and instead put the cursor at the start of the name.
```rust
fn bar() {
fn <|>foo() { }
foo();
}
```
This fixes#877 but doesn't contain the refactoring of the return type for `goto_definition`
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
I think it'll be better to make the path resolution the number of unresolved
segments, not the first unresolved index; then this error could simply not have
happened. But I'll do that separately.
This contains the syntax range of the name itself, allowing NavigationTarget to
properly set the focus_range. This should make it so that when using symbol
based navigation, we should always focus on the name, instead of the full range.
866: Implement basic support for Associated Methods r=flodiebold a=vipentti
This is my attempt at learning to understand how the type inference works by adding basic support for associated methods. Currently it does not resolve associated types or constants.
The basic idea is that `Resolver::resolve_path` returns a new `PathResult` type, which has two variants, `FullyResolved` and `PartiallyResolved`, fully resolved matches the previous behavior, where as `PartiallyResolved` contains the `PerNs<Resolution` in addition to a `segment_index` which contains the index of the segment which we failed to resolve. This index can then be used to continue inference in `infer_path_expr` using the `Type` we managed to resolve.
This changes some of the previous apis, so looking for feedback and suggestions.
This should enable fixing #832
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
This also adds new pub(crate) resolve_path_segments which returns the
PathResult, which may or may not be fully resolved. PathResult is also now
pub(crate) since it is an implementation detail.
864: Fix handling of generics in tuple variants and refactor a bit r=matklad a=flodiebold
(The problem was that we created separate substitutions for the return value, so we lost the connection between the type arguments in the constructor call and the type arguments of the result.)
Also make them display a tiny bit nicer.
Fixes#860.
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
Co-authored-by: Florian Diebold <florian.diebold@freiheit.com>
876: Fix join_lines not adding a comma after join_single_expr_block with match arm r=matklad a=vipentti
Fixes#868
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
We will also remove optional whitespace between the expression and the comma.
e.g.
```rust
fn foo() {
let x = (<|>{
4
} ,); // NOTE: whitespace
}
```
becomes
```rust
fn foo() {
let x = (<|>4,);
}
```
This is done in `infer_path_expr`. When `Resolver::resolve_path` returns
`PartiallyResolved`, we use the returned `Resolution` together with the given
`segment_index` to check if we can find something matching the segment at
segment_index in the impls for that particular type.
Benchmarks show no difference. This is probably because we are
bottlenecked on memory allocations, and we should fix that, but we are
not optimizing for performance just yet.
changes. Lines starting # with '#' will be ignored, and an empty
message aborts the commit. # # On branch token-source # Changes to be
committed: # modified: crates/ra_syntax/src/parsing/parser_api.rs #
modified: crates/ra_syntax/src/parsing/parser_impl.rs #
846: WIP: Enable parsing of attributes inside a match block r=matklad a=vipentti
We allow invalid inner attributes to be parsed, e.g. inner attributes that are
not directly after the opening brace of the match block.
Instead we run validation on `MatchArmList` to allow better reporting of errors.
This fixes#845 and works towards #759
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
847: Refactor vfs r=matklad a=matklad
Some slight refctorings of VFS, in preparation for moving it to a separate repo
Co-authored-by: Aleksey Kladov <aleksey.kladov@gmail.com>
We allow invalid inner attributes to be parsed, e.g. inner attributes that are
not directly after the opening brace of the match block.
Instead we run validation on `MatchArmList` to allow better reporting of errors.
844: Refactor find_all_refs to return ReferenceSearchResult r=vipentti a=vipentti
This refactors `find_all_refs` to return a new `ReferenceSearchResult` based on feedback in #839.
There are few questions/notes regarding the refactor:
1. Introducing `NavigationTarget::from_bind_pat` this simply forwards the call to `NavigationTarget::from_named`, could we just expose `from_named` directly as `pub(crate)` ?
2. Added an utility method `NavigationTarget::range` since there were few places where you would use `self.focus_range.unwrap_or(self.full_range)`
3. Implementing `IntoIterator` for `ReferenceSearchResult`. This turns `ReferenceSearchResult` into an iterator over `FileRanges` and allows previous code to mostly stay as it was based on the order that `find_all_refs` previously had (declaration first and then the references). I'm not sure if there is a way of doing the conversion to `IntoIter` without the allocation of a new vector
4. Is it possible to have a binding without a name? I'm not sure if the `NavigationTarget::from_bind_pat` can cause some edge-cases that previously were ok
This fixes#835.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
This makes it more like the other code model types.
Also make Module::definition_source/declaration_source return HirFileIds, to
make them more like the other source functions.
836: auto_import: fix bug when the common path was shorter than both the c… r=matklad a=eulerdisk
Fix bug when the common path was shorter than both the current use and target path.
Wrong 1 (unnecessary self)
```
use std::fmt::nested::Debug;
std::fmt::Display<|>
```
--->
```
use std::fmt::{ self, Display, nested::Debug};
Display<|>
```
Wrong 2 (unnecessary, Debug disappear!!)
```
use std::fmt::Debug;
std::fmt::nested::Display
```
-->
```
use std::fmt::Debug::{ self, nested::Display, };
Display<|>
```
Co-authored-by: Andrea Pretto <eulerdisk@gmail.com>
Namely, the allowance for up to 7 events, and thus requiring anyone
with fewer than 7 events to wait for the 3 second timeout, is only
relevant to fsevents (i.e. Mac OS X)
830: Fix test_vfs_works failing on Windows due to extra Write events r=pnkfelix a=vipentti
On Windows `notify` generates extra `Write` events for folders, which caused
`process_tasks` to not handle all tasks generated on Windows.
This fixes#827
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
On Windows `notify` generates extra `Write` events for folders, which caused
`process_tasks` to not handle all tasks generated on Windows.
This fixes#827
Instead of using canonicalize, we now join the given path to
`std::env::current_dir()`, which either replaces the path, if the given path is
absolute, or joins the paths.
This fixes#821.
818: In `RootConfig::contains`, check against canonicalized version of root path r=matklad a=pnkfelix
In `RootConfig::contains`, check against canonicalized version of root path since OS may hand us data that uses the canonical form rather than the root as specified by the user.
This is a step towards a resolution of issue #734 but does not completely fix the problem there.
Co-authored-by: Felix S. Klock II <pnkfelix@pnkfx.org>
813: Add support for container_name in workspace/symbol query r=matklad a=vipentti
Currently this does not fill in the container_info if a type is defined on the top level in a file.
e.g. `foo.rs`
```rust
enum Foo { }
```
`Foo` will have None as the container_name, however
```rust
mod foo_mod {
enum Foo { }
}
```
`Foo` has `foo_mod` as the container_name.
This closes#559
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
814: auto_import: import in enclosing module by default r=matklad a=eulerdisk
Simpler version of #795
Co-authored-by: Andrea Pretto <eulerdisk@gmail.com>
781: Refactor to allow for multiple assists r=matklad a=eulerdisk
This is necessary to allow assist "providers" (which currently are simple free function) to produce multiple assists. I'm not sure this is the best possible refactoring tough.
Co-authored-by: Andrea Pretto <eulerdisk@gmail.com>
762: "Dumb" auto import assist r=matklad a=eulerdisk
This adds a new assist to "add xxx::yyy to the current file" when the cursor is on a PATH. It manages correctly nested imports,`self` keyword and creates new nested imports if necessary. [See the tests]
It doesn't use name resolution so in that sense is 'dumb', but I have plans to do that. That in the future will be useful to auto import trait names in autocompletion for example.
It can easily be extended to provide multiple actions to select in which scope to import. That's another thing I plan to do.
@matklad I copied some indentation code from `ide_light`, I don't know at the moment if/how you want to refactor that code. This assist was meant to be in `ide_light`.
Co-authored-by: Andrea Pretto <eulerdisk@gmail.com>
770: Fix introduce var duplicating newlines r=matklad a=vipentti
This fixes#713.
If the block before the statement we want to use introduce var on, had empty
lines these empty lines would also be added between the let-statement and
the current line where the new variable is used.
This fixes that by trimming excess newlines from the start of the indent chunk
and simply adding a single newline (when the chunk had newlines) between the
let-statement and the current statement. If there were no newlines this
matches the previous behaviour.
Co-authored-by: Ville Penttinen <villem.penttinen@gmail.com>
This fixes#713.
If the block before the statement we want to use introduce var on, had empty
lines these empty lines would also be added between the let-statement and
the current line where the new variable is used.
This fixes that by trimming excess newlines from the start of the indent chunk
and simply adding a single newline (when the chunk had newlines) between the
let-statement and the current statement. If there were no newlines this
matches the previous behaviour.
This fixes#758.
Currently we try to maintain the cursor position relative to the statement under
cursor, if the cursor is inside the dbg! macro call.
Meaning:
let foo = dbg!(some.complex<|>().expression());
Should turn into:
let foo = some.complex<|>().expression();
With the cursor staying in place.
742: Extern crate r=matklad a=flodiebold
This implements `extern crate` declarations by lowering them to (absolute) imports, and adds support for absolute paths. It also extracts the extern prelude from the per-module item map, and handles the special case of extern crates in the crate root adding to the extern prelude.
This means we finally resolve `Arc`, so it fixes#523😄
Co-authored-by: Florian Diebold <flodiebold@gmail.com>
727: Fix macro_rules separator parsing. r=matklad a=jrmuizel
macro_rules rules are separated by ';' including an optional ';' at the end
Co-authored-by: Jeff Muizelaar <jrmuizel@gmail.com>
