1//! The `language` crate provides a large chunk of Zed's language-related
2//! features (the other big contributors being project and lsp crates that revolve around LSP features).
3//! Namely, this crate:
4//! - Provides [`Language`], [`Grammar`] and [`LanguageRegistry`] types that
5//! use Tree-sitter to provide syntax highlighting to the editor; note though that `language` doesn't perform the highlighting by itself. It only maps ranges in a buffer to colors. Treesitter is also used for buffer outlines (lists of symbols in a buffer)
6//! - Exposes [`LanguageConfig`] that describes how constructs (like brackets or line comments) should be handled by the editor for a source file of a particular language.
7//!
8//! Notably we do *not* assign a single language to a single file; in real world a single file can consist of multiple programming languages - HTML is a good example of that - and `language` crate tends to reflect that status quo in its API.
9mod buffer;
10mod diagnostic_set;
11mod highlight_map;
12mod language_registry;
13pub mod language_settings;
14mod manifest;
15mod outline;
16pub mod proto;
17mod syntax_map;
18mod task_context;
19mod text_diff;
20mod toolchain;
21
22#[cfg(test)]
23pub mod buffer_tests;
24
25pub use crate::language_settings::EditPredictionsMode;
26use crate::language_settings::SoftWrap;
27use anyhow::{Context as _, Result};
28use async_trait::async_trait;
29use collections::{HashMap, HashSet, IndexSet};
30use futures::Future;
31use gpui::{App, AsyncApp, Entity, SharedString};
32pub use highlight_map::HighlightMap;
33use http_client::HttpClient;
34pub use language_registry::{
35 LanguageName, LanguageServerStatusUpdate, LoadedLanguage, ServerHealth,
36};
37use lsp::{CodeActionKind, InitializeParams, LanguageServerBinary, LanguageServerBinaryOptions};
38pub use manifest::{ManifestDelegate, ManifestName, ManifestProvider, ManifestQuery};
39use parking_lot::Mutex;
40use regex::Regex;
41use schemars::{JsonSchema, SchemaGenerator, json_schema};
42use serde::{Deserialize, Deserializer, Serialize, Serializer, de};
43use serde_json::Value;
44use settings::WorktreeId;
45use smol::future::FutureExt as _;
46use std::num::NonZeroU32;
47use std::{
48 ffi::OsStr,
49 fmt::Debug,
50 hash::Hash,
51 mem,
52 ops::{DerefMut, Range},
53 path::{Path, PathBuf},
54 pin::Pin,
55 str,
56 sync::{
57 Arc, LazyLock,
58 atomic::{AtomicU64, AtomicUsize, Ordering::SeqCst},
59 },
60};
61use syntax_map::{QueryCursorHandle, SyntaxSnapshot};
62use task::RunnableTag;
63pub use task_context::{ContextLocation, ContextProvider, RunnableRange};
64pub use text_diff::{
65 DiffOptions, apply_diff_patch, line_diff, text_diff, text_diff_with_options, unified_diff,
66};
67use theme::SyntaxTheme;
68pub use toolchain::{
69 LanguageToolchainStore, LocalLanguageToolchainStore, Toolchain, ToolchainList, ToolchainLister,
70 ToolchainMetadata, ToolchainScope,
71};
72use tree_sitter::{self, Query, QueryCursor, WasmStore, wasmtime};
73use util::serde::default_true;
74
75pub use buffer::Operation;
76pub use buffer::*;
77pub use diagnostic_set::{DiagnosticEntry, DiagnosticGroup};
78pub use language_registry::{
79 AvailableLanguage, BinaryStatus, LanguageNotFound, LanguageQueries, LanguageRegistry,
80 QUERY_FILENAME_PREFIXES,
81};
82pub use lsp::{LanguageServerId, LanguageServerName};
83pub use outline::*;
84pub use syntax_map::{OwnedSyntaxLayer, SyntaxLayer, ToTreeSitterPoint, TreeSitterOptions};
85pub use text::{AnchorRangeExt, LineEnding};
86pub use tree_sitter::{Node, Parser, Tree, TreeCursor};
87
88/// Initializes the `language` crate.
89///
90/// This should be called before making use of items from the create.
91pub fn init(cx: &mut App) {
92 language_settings::init(cx);
93}
94
95static QUERY_CURSORS: Mutex<Vec<QueryCursor>> = Mutex::new(vec![]);
96static PARSERS: Mutex<Vec<Parser>> = Mutex::new(vec![]);
97
98pub fn with_parser<F, R>(func: F) -> R
99where
100 F: FnOnce(&mut Parser) -> R,
101{
102 let mut parser = PARSERS.lock().pop().unwrap_or_else(|| {
103 let mut parser = Parser::new();
104 parser
105 .set_wasm_store(WasmStore::new(&WASM_ENGINE).unwrap())
106 .unwrap();
107 parser
108 });
109 parser.set_included_ranges(&[]).unwrap();
110 let result = func(&mut parser);
111 PARSERS.lock().push(parser);
112 result
113}
114
115pub fn with_query_cursor<F, R>(func: F) -> R
116where
117 F: FnOnce(&mut QueryCursor) -> R,
118{
119 let mut cursor = QueryCursorHandle::new();
120 func(cursor.deref_mut())
121}
122
123static NEXT_LANGUAGE_ID: AtomicUsize = AtomicUsize::new(0);
124static NEXT_GRAMMAR_ID: AtomicUsize = AtomicUsize::new(0);
125static WASM_ENGINE: LazyLock<wasmtime::Engine> = LazyLock::new(|| {
126 wasmtime::Engine::new(&wasmtime::Config::new()).expect("Failed to create Wasmtime engine")
127});
128
129/// A shared grammar for plain text, exposed for reuse by downstream crates.
130pub static PLAIN_TEXT: LazyLock<Arc<Language>> = LazyLock::new(|| {
131 Arc::new(Language::new(
132 LanguageConfig {
133 name: "Plain Text".into(),
134 soft_wrap: Some(SoftWrap::EditorWidth),
135 matcher: LanguageMatcher {
136 path_suffixes: vec!["txt".to_owned()],
137 first_line_pattern: None,
138 },
139 ..Default::default()
140 },
141 None,
142 ))
143});
144
145/// Types that represent a position in a buffer, and can be converted into
146/// an LSP position, to send to a language server.
147pub trait ToLspPosition {
148 /// Converts the value into an LSP position.
149 fn to_lsp_position(self) -> lsp::Position;
150}
151
152#[derive(Debug, Clone, PartialEq, Eq, Hash)]
153pub struct Location {
154 pub buffer: Entity<Buffer>,
155 pub range: Range<Anchor>,
156}
157
158type ServerBinaryCache = futures::lock::Mutex<Option<(bool, LanguageServerBinary)>>;
159
160/// Represents a Language Server, with certain cached sync properties.
161/// Uses [`LspAdapter`] under the hood, but calls all 'static' methods
162/// once at startup, and caches the results.
163pub struct CachedLspAdapter {
164 pub name: LanguageServerName,
165 pub disk_based_diagnostic_sources: Vec<String>,
166 pub disk_based_diagnostics_progress_token: Option<String>,
167 language_ids: HashMap<LanguageName, String>,
168 pub adapter: Arc<dyn LspAdapter>,
169 pub reinstall_attempt_count: AtomicU64,
170 cached_binary: ServerBinaryCache,
171}
172
173impl Debug for CachedLspAdapter {
174 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
175 f.debug_struct("CachedLspAdapter")
176 .field("name", &self.name)
177 .field(
178 "disk_based_diagnostic_sources",
179 &self.disk_based_diagnostic_sources,
180 )
181 .field(
182 "disk_based_diagnostics_progress_token",
183 &self.disk_based_diagnostics_progress_token,
184 )
185 .field("language_ids", &self.language_ids)
186 .field("reinstall_attempt_count", &self.reinstall_attempt_count)
187 .finish_non_exhaustive()
188 }
189}
190
191impl CachedLspAdapter {
192 pub fn new(adapter: Arc<dyn LspAdapter>) -> Arc<Self> {
193 let name = adapter.name();
194 let disk_based_diagnostic_sources = adapter.disk_based_diagnostic_sources();
195 let disk_based_diagnostics_progress_token = adapter.disk_based_diagnostics_progress_token();
196 let language_ids = adapter.language_ids();
197
198 Arc::new(CachedLspAdapter {
199 name,
200 disk_based_diagnostic_sources,
201 disk_based_diagnostics_progress_token,
202 language_ids,
203 adapter,
204 cached_binary: Default::default(),
205 reinstall_attempt_count: AtomicU64::new(0),
206 })
207 }
208
209 pub fn name(&self) -> LanguageServerName {
210 self.adapter.name()
211 }
212
213 pub async fn get_language_server_command(
214 self: Arc<Self>,
215 delegate: Arc<dyn LspAdapterDelegate>,
216 toolchains: Option<Toolchain>,
217 binary_options: LanguageServerBinaryOptions,
218 cx: &mut AsyncApp,
219 ) -> Result<LanguageServerBinary> {
220 let mut cached_binary = self.cached_binary.lock().await;
221 self.adapter
222 .clone()
223 .get_language_server_command(
224 delegate,
225 toolchains,
226 binary_options,
227 &mut cached_binary,
228 cx,
229 )
230 .await
231 }
232
233 pub fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
234 self.adapter.code_action_kinds()
235 }
236
237 pub fn process_diagnostics(
238 &self,
239 params: &mut lsp::PublishDiagnosticsParams,
240 server_id: LanguageServerId,
241 existing_diagnostics: Option<&'_ Buffer>,
242 ) {
243 self.adapter
244 .process_diagnostics(params, server_id, existing_diagnostics)
245 }
246
247 pub fn retain_old_diagnostic(&self, previous_diagnostic: &Diagnostic, cx: &App) -> bool {
248 self.adapter.retain_old_diagnostic(previous_diagnostic, cx)
249 }
250
251 pub fn underline_diagnostic(&self, diagnostic: &lsp::Diagnostic) -> bool {
252 self.adapter.underline_diagnostic(diagnostic)
253 }
254
255 pub fn diagnostic_message_to_markdown(&self, message: &str) -> Option<String> {
256 self.adapter.diagnostic_message_to_markdown(message)
257 }
258
259 pub async fn process_completions(&self, completion_items: &mut [lsp::CompletionItem]) {
260 self.adapter.process_completions(completion_items).await
261 }
262
263 pub async fn labels_for_completions(
264 &self,
265 completion_items: &[lsp::CompletionItem],
266 language: &Arc<Language>,
267 ) -> Result<Vec<Option<CodeLabel>>> {
268 self.adapter
269 .clone()
270 .labels_for_completions(completion_items, language)
271 .await
272 }
273
274 pub async fn labels_for_symbols(
275 &self,
276 symbols: &[(String, lsp::SymbolKind)],
277 language: &Arc<Language>,
278 ) -> Result<Vec<Option<CodeLabel>>> {
279 self.adapter
280 .clone()
281 .labels_for_symbols(symbols, language)
282 .await
283 }
284
285 pub fn language_id(&self, language_name: &LanguageName) -> String {
286 let candidate = self
287 .language_ids
288 .get(language_name)
289 .cloned()
290 .unwrap_or_else(|| language_name.lsp_id());
291 if candidate == "vue.js" {
292 "vue".into()
293 } else {
294 candidate
295 }
296 }
297}
298
299/// [`LspAdapterDelegate`] allows [`LspAdapter]` implementations to interface with the application
300// e.g. to display a notification or fetch data from the web.
301#[async_trait]
302pub trait LspAdapterDelegate: Send + Sync {
303 fn show_notification(&self, message: &str, cx: &mut App);
304 fn http_client(&self) -> Arc<dyn HttpClient>;
305 fn worktree_id(&self) -> WorktreeId;
306 fn worktree_root_path(&self) -> &Path;
307 fn update_status(&self, language: LanguageServerName, status: BinaryStatus);
308 fn registered_lsp_adapters(&self) -> Vec<Arc<dyn LspAdapter>>;
309 async fn language_server_download_dir(&self, name: &LanguageServerName) -> Option<Arc<Path>>;
310
311 async fn npm_package_installed_version(
312 &self,
313 package_name: &str,
314 ) -> Result<Option<(PathBuf, String)>>;
315 async fn which(&self, command: &OsStr) -> Option<PathBuf>;
316 async fn shell_env(&self) -> HashMap<String, String>;
317 async fn read_text_file(&self, path: PathBuf) -> Result<String>;
318 async fn try_exec(&self, binary: LanguageServerBinary) -> Result<()>;
319}
320
321#[async_trait(?Send)]
322pub trait LspAdapter: 'static + Send + Sync + DynLspInstaller {
323 fn name(&self) -> LanguageServerName;
324
325 fn process_diagnostics(
326 &self,
327 _: &mut lsp::PublishDiagnosticsParams,
328 _: LanguageServerId,
329 _: Option<&'_ Buffer>,
330 ) {
331 }
332
333 /// When processing new `lsp::PublishDiagnosticsParams` diagnostics, whether to retain previous one(s) or not.
334 fn retain_old_diagnostic(&self, _previous_diagnostic: &Diagnostic, _cx: &App) -> bool {
335 false
336 }
337
338 /// Whether to underline a given diagnostic or not, when rendering in the editor.
339 ///
340 /// https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#diagnosticTag
341 /// states that
342 /// > Clients are allowed to render diagnostics with this tag faded out instead of having an error squiggle.
343 /// for the unnecessary diagnostics, so do not underline them.
344 fn underline_diagnostic(&self, _diagnostic: &lsp::Diagnostic) -> bool {
345 true
346 }
347
348 /// Post-processes completions provided by the language server.
349 async fn process_completions(&self, _: &mut [lsp::CompletionItem]) {}
350
351 fn diagnostic_message_to_markdown(&self, _message: &str) -> Option<String> {
352 None
353 }
354
355 async fn labels_for_completions(
356 self: Arc<Self>,
357 completions: &[lsp::CompletionItem],
358 language: &Arc<Language>,
359 ) -> Result<Vec<Option<CodeLabel>>> {
360 let mut labels = Vec::new();
361 for (ix, completion) in completions.iter().enumerate() {
362 let label = self.label_for_completion(completion, language).await;
363 if let Some(label) = label {
364 labels.resize(ix + 1, None);
365 *labels.last_mut().unwrap() = Some(label);
366 }
367 }
368 Ok(labels)
369 }
370
371 async fn label_for_completion(
372 &self,
373 _: &lsp::CompletionItem,
374 _: &Arc<Language>,
375 ) -> Option<CodeLabel> {
376 None
377 }
378
379 async fn labels_for_symbols(
380 self: Arc<Self>,
381 symbols: &[(String, lsp::SymbolKind)],
382 language: &Arc<Language>,
383 ) -> Result<Vec<Option<CodeLabel>>> {
384 let mut labels = Vec::new();
385 for (ix, (name, kind)) in symbols.iter().enumerate() {
386 let label = self.label_for_symbol(name, *kind, language).await;
387 if let Some(label) = label {
388 labels.resize(ix + 1, None);
389 *labels.last_mut().unwrap() = Some(label);
390 }
391 }
392 Ok(labels)
393 }
394
395 async fn label_for_symbol(
396 &self,
397 _: &str,
398 _: lsp::SymbolKind,
399 _: &Arc<Language>,
400 ) -> Option<CodeLabel> {
401 None
402 }
403
404 /// Returns initialization options that are going to be sent to a LSP server as a part of [`lsp::InitializeParams`]
405 async fn initialization_options(
406 self: Arc<Self>,
407 _: &Arc<dyn LspAdapterDelegate>,
408 ) -> Result<Option<Value>> {
409 Ok(None)
410 }
411
412 async fn workspace_configuration(
413 self: Arc<Self>,
414 _: &Arc<dyn LspAdapterDelegate>,
415 _: Option<Toolchain>,
416 _cx: &mut AsyncApp,
417 ) -> Result<Value> {
418 Ok(serde_json::json!({}))
419 }
420
421 async fn additional_initialization_options(
422 self: Arc<Self>,
423 _target_language_server_id: LanguageServerName,
424 _: &Arc<dyn LspAdapterDelegate>,
425 ) -> Result<Option<Value>> {
426 Ok(None)
427 }
428
429 async fn additional_workspace_configuration(
430 self: Arc<Self>,
431 _target_language_server_id: LanguageServerName,
432 _: &Arc<dyn LspAdapterDelegate>,
433 _cx: &mut AsyncApp,
434 ) -> Result<Option<Value>> {
435 Ok(None)
436 }
437
438 /// Returns a list of code actions supported by a given LspAdapter
439 fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
440 None
441 }
442
443 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
444 Default::default()
445 }
446
447 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
448 None
449 }
450
451 fn language_ids(&self) -> HashMap<LanguageName, String> {
452 HashMap::default()
453 }
454
455 /// Support custom initialize params.
456 fn prepare_initialize_params(
457 &self,
458 original: InitializeParams,
459 _: &App,
460 ) -> Result<InitializeParams> {
461 Ok(original)
462 }
463
464 /// Method only implemented by the default JSON language server adapter.
465 /// Used to provide dynamic reloading of the JSON schemas used to
466 /// provide autocompletion and diagnostics in Zed setting and keybind
467 /// files
468 fn is_primary_zed_json_schema_adapter(&self) -> bool {
469 false
470 }
471
472 /// Method only implemented by the default JSON language server adapter.
473 /// Used to clear the cache of JSON schemas that are used to provide
474 /// autocompletion and diagnostics in Zed settings and keybinds files.
475 /// Should not be called unless the callee is sure that
476 /// `Self::is_primary_zed_json_schema_adapter` returns `true`
477 async fn clear_zed_json_schema_cache(&self) {
478 unreachable!(
479 "Not implemented for this adapter. This method should only be called on the default JSON language server adapter"
480 );
481 }
482
483 /// True for the extension adapter and false otherwise.
484 fn is_extension(&self) -> bool {
485 false
486 }
487}
488
489pub trait LspInstaller {
490 type BinaryVersion;
491 fn check_if_user_installed(
492 &self,
493 _: &dyn LspAdapterDelegate,
494 _: Option<Toolchain>,
495 _: &AsyncApp,
496 ) -> impl Future<Output = Option<LanguageServerBinary>> {
497 async { None }
498 }
499
500 fn fetch_latest_server_version(
501 &self,
502 delegate: &dyn LspAdapterDelegate,
503 pre_release: bool,
504 cx: &mut AsyncApp,
505 ) -> impl Future<Output = Result<Self::BinaryVersion>>;
506
507 fn check_if_version_installed(
508 &self,
509 _version: &Self::BinaryVersion,
510 _container_dir: &PathBuf,
511 _delegate: &dyn LspAdapterDelegate,
512 ) -> impl Future<Output = Option<LanguageServerBinary>> {
513 async { None }
514 }
515
516 fn fetch_server_binary(
517 &self,
518 latest_version: Self::BinaryVersion,
519 container_dir: PathBuf,
520 delegate: &dyn LspAdapterDelegate,
521 ) -> impl Future<Output = Result<LanguageServerBinary>>;
522
523 fn cached_server_binary(
524 &self,
525 container_dir: PathBuf,
526 delegate: &dyn LspAdapterDelegate,
527 ) -> impl Future<Output = Option<LanguageServerBinary>>;
528}
529
530#[async_trait(?Send)]
531pub trait DynLspInstaller {
532 async fn try_fetch_server_binary(
533 &self,
534 delegate: &Arc<dyn LspAdapterDelegate>,
535 container_dir: PathBuf,
536 pre_release: bool,
537 cx: &mut AsyncApp,
538 ) -> Result<LanguageServerBinary>;
539 fn get_language_server_command<'a>(
540 self: Arc<Self>,
541 delegate: Arc<dyn LspAdapterDelegate>,
542 toolchains: Option<Toolchain>,
543 binary_options: LanguageServerBinaryOptions,
544 cached_binary: &'a mut Option<(bool, LanguageServerBinary)>,
545 cx: &'a mut AsyncApp,
546 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>>;
547}
548
549#[async_trait(?Send)]
550impl<LI, BinaryVersion> DynLspInstaller for LI
551where
552 LI: LspInstaller<BinaryVersion = BinaryVersion> + LspAdapter,
553{
554 async fn try_fetch_server_binary(
555 &self,
556 delegate: &Arc<dyn LspAdapterDelegate>,
557 container_dir: PathBuf,
558 pre_release: bool,
559 cx: &mut AsyncApp,
560 ) -> Result<LanguageServerBinary> {
561 let name = self.name();
562
563 log::debug!("fetching latest version of language server {:?}", name.0);
564 delegate.update_status(name.clone(), BinaryStatus::CheckingForUpdate);
565
566 let latest_version = self
567 .fetch_latest_server_version(delegate.as_ref(), pre_release, cx)
568 .await?;
569
570 if let Some(binary) = self
571 .check_if_version_installed(&latest_version, &container_dir, delegate.as_ref())
572 .await
573 {
574 log::debug!("language server {:?} is already installed", name.0);
575 delegate.update_status(name.clone(), BinaryStatus::None);
576 Ok(binary)
577 } else {
578 log::debug!("downloading language server {:?}", name.0);
579 delegate.update_status(name.clone(), BinaryStatus::Downloading);
580 let binary = self
581 .fetch_server_binary(latest_version, container_dir, delegate.as_ref())
582 .await;
583
584 delegate.update_status(name.clone(), BinaryStatus::None);
585 binary
586 }
587 }
588 fn get_language_server_command<'a>(
589 self: Arc<Self>,
590 delegate: Arc<dyn LspAdapterDelegate>,
591 toolchain: Option<Toolchain>,
592 binary_options: LanguageServerBinaryOptions,
593 cached_binary: &'a mut Option<(bool, LanguageServerBinary)>,
594 cx: &'a mut AsyncApp,
595 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
596 async move {
597 // First we check whether the adapter can give us a user-installed binary.
598 // If so, we do *not* want to cache that, because each worktree might give us a different
599 // binary:
600 //
601 // worktree 1: user-installed at `.bin/gopls`
602 // worktree 2: user-installed at `~/bin/gopls`
603 // worktree 3: no gopls found in PATH -> fallback to Zed installation
604 //
605 // We only want to cache when we fall back to the global one,
606 // because we don't want to download and overwrite our global one
607 // for each worktree we might have open.
608 if binary_options.allow_path_lookup
609 && let Some(binary) = self
610 .check_if_user_installed(delegate.as_ref(), toolchain, cx)
611 .await
612 {
613 log::info!(
614 "found user-installed language server for {}. path: {:?}, arguments: {:?}",
615 self.name().0,
616 binary.path,
617 binary.arguments
618 );
619 return Ok(binary);
620 }
621
622 anyhow::ensure!(
623 binary_options.allow_binary_download,
624 "downloading language servers disabled"
625 );
626
627 if let Some((pre_release, cached_binary)) = cached_binary
628 && *pre_release == binary_options.pre_release
629 {
630 return Ok(cached_binary.clone());
631 }
632
633 let Some(container_dir) = delegate.language_server_download_dir(&self.name()).await
634 else {
635 anyhow::bail!("no language server download dir defined")
636 };
637
638 let mut binary = self
639 .try_fetch_server_binary(
640 &delegate,
641 container_dir.to_path_buf(),
642 binary_options.pre_release,
643 cx,
644 )
645 .await;
646
647 if let Err(error) = binary.as_ref() {
648 if let Some(prev_downloaded_binary) = self
649 .cached_server_binary(container_dir.to_path_buf(), delegate.as_ref())
650 .await
651 {
652 log::info!(
653 "failed to fetch newest version of language server {:?}. \
654 error: {:?}, falling back to using {:?}",
655 self.name(),
656 error,
657 prev_downloaded_binary.path
658 );
659 binary = Ok(prev_downloaded_binary);
660 } else {
661 delegate.update_status(
662 self.name(),
663 BinaryStatus::Failed {
664 error: format!("{error:?}"),
665 },
666 );
667 }
668 }
669
670 if let Ok(binary) = &binary {
671 *cached_binary = Some((binary_options.pre_release, binary.clone()));
672 }
673
674 binary
675 }
676 .boxed_local()
677 }
678}
679
680#[derive(Clone, Debug, Default, PartialEq, Eq)]
681pub struct CodeLabel {
682 /// The text to display.
683 pub text: String,
684 /// Syntax highlighting runs.
685 pub runs: Vec<(Range<usize>, HighlightId)>,
686 /// The portion of the text that should be used in fuzzy filtering.
687 pub filter_range: Range<usize>,
688}
689
690#[derive(Clone, Deserialize, JsonSchema)]
691pub struct LanguageConfig {
692 /// Human-readable name of the language.
693 pub name: LanguageName,
694 /// The name of this language for a Markdown code fence block
695 pub code_fence_block_name: Option<Arc<str>>,
696 // The name of the grammar in a WASM bundle (experimental).
697 pub grammar: Option<Arc<str>>,
698 /// The criteria for matching this language to a given file.
699 #[serde(flatten)]
700 pub matcher: LanguageMatcher,
701 /// List of bracket types in a language.
702 #[serde(default)]
703 pub brackets: BracketPairConfig,
704 /// If set to true, auto indentation uses last non empty line to determine
705 /// the indentation level for a new line.
706 #[serde(default = "auto_indent_using_last_non_empty_line_default")]
707 pub auto_indent_using_last_non_empty_line: bool,
708 // Whether indentation of pasted content should be adjusted based on the context.
709 #[serde(default)]
710 pub auto_indent_on_paste: Option<bool>,
711 /// A regex that is used to determine whether the indentation level should be
712 /// increased in the following line.
713 #[serde(default, deserialize_with = "deserialize_regex")]
714 #[schemars(schema_with = "regex_json_schema")]
715 pub increase_indent_pattern: Option<Regex>,
716 /// A regex that is used to determine whether the indentation level should be
717 /// decreased in the following line.
718 #[serde(default, deserialize_with = "deserialize_regex")]
719 #[schemars(schema_with = "regex_json_schema")]
720 pub decrease_indent_pattern: Option<Regex>,
721 /// A list of rules for decreasing indentation. Each rule pairs a regex with a set of valid
722 /// "block-starting" tokens. When a line matches a pattern, its indentation is aligned with
723 /// the most recent line that began with a corresponding token. This enables context-aware
724 /// outdenting, like aligning an `else` with its `if`.
725 #[serde(default)]
726 pub decrease_indent_patterns: Vec<DecreaseIndentConfig>,
727 /// A list of characters that trigger the automatic insertion of a closing
728 /// bracket when they immediately precede the point where an opening
729 /// bracket is inserted.
730 #[serde(default)]
731 pub autoclose_before: String,
732 /// A placeholder used internally by Semantic Index.
733 #[serde(default)]
734 pub collapsed_placeholder: String,
735 /// A line comment string that is inserted in e.g. `toggle comments` action.
736 /// A language can have multiple flavours of line comments. All of the provided line comments are
737 /// used for comment continuations on the next line, but only the first one is used for Editor::ToggleComments.
738 #[serde(default)]
739 pub line_comments: Vec<Arc<str>>,
740 /// Delimiters and configuration for recognizing and formatting block comments.
741 #[serde(default)]
742 pub block_comment: Option<BlockCommentConfig>,
743 /// Delimiters and configuration for recognizing and formatting documentation comments.
744 #[serde(default, alias = "documentation")]
745 pub documentation_comment: Option<BlockCommentConfig>,
746 /// A list of additional regex patterns that should be treated as prefixes
747 /// for creating boundaries during rewrapping, ensuring content from one
748 /// prefixed section doesn't merge with another (e.g., markdown list items).
749 /// By default, Zed treats as paragraph and comment prefixes as boundaries.
750 #[serde(default, deserialize_with = "deserialize_regex_vec")]
751 #[schemars(schema_with = "regex_vec_json_schema")]
752 pub rewrap_prefixes: Vec<Regex>,
753 /// A list of language servers that are allowed to run on subranges of a given language.
754 #[serde(default)]
755 pub scope_opt_in_language_servers: Vec<LanguageServerName>,
756 #[serde(default)]
757 pub overrides: HashMap<String, LanguageConfigOverride>,
758 /// A list of characters that Zed should treat as word characters for the
759 /// purpose of features that operate on word boundaries, like 'move to next word end'
760 /// or a whole-word search in buffer search.
761 #[serde(default)]
762 pub word_characters: HashSet<char>,
763 /// Whether to indent lines using tab characters, as opposed to multiple
764 /// spaces.
765 #[serde(default)]
766 pub hard_tabs: Option<bool>,
767 /// How many columns a tab should occupy.
768 #[serde(default)]
769 pub tab_size: Option<NonZeroU32>,
770 /// How to soft-wrap long lines of text.
771 #[serde(default)]
772 pub soft_wrap: Option<SoftWrap>,
773 /// When set, selections can be wrapped using prefix/suffix pairs on both sides.
774 #[serde(default)]
775 pub wrap_characters: Option<WrapCharactersConfig>,
776 /// The name of a Prettier parser that will be used for this language when no file path is available.
777 /// If there's a parser name in the language settings, that will be used instead.
778 #[serde(default)]
779 pub prettier_parser_name: Option<String>,
780 /// If true, this language is only for syntax highlighting via an injection into other
781 /// languages, but should not appear to the user as a distinct language.
782 #[serde(default)]
783 pub hidden: bool,
784 /// If configured, this language contains JSX style tags, and should support auto-closing of those tags.
785 #[serde(default)]
786 pub jsx_tag_auto_close: Option<JsxTagAutoCloseConfig>,
787 /// A list of characters that Zed should treat as word characters for completion queries.
788 #[serde(default)]
789 pub completion_query_characters: HashSet<char>,
790 /// A list of preferred debuggers for this language.
791 #[serde(default)]
792 pub debuggers: IndexSet<SharedString>,
793}
794
795#[derive(Clone, Debug, Deserialize, Default, JsonSchema)]
796pub struct DecreaseIndentConfig {
797 #[serde(default, deserialize_with = "deserialize_regex")]
798 #[schemars(schema_with = "regex_json_schema")]
799 pub pattern: Option<Regex>,
800 #[serde(default)]
801 pub valid_after: Vec<String>,
802}
803
804#[derive(Clone, Debug, Serialize, Deserialize, Default, JsonSchema)]
805pub struct LanguageMatcher {
806 /// Given a list of `LanguageConfig`'s, the language of a file can be determined based on the path extension matching any of the `path_suffixes`.
807 #[serde(default)]
808 pub path_suffixes: Vec<String>,
809 /// A regex pattern that determines whether the language should be assigned to a file or not.
810 #[serde(
811 default,
812 serialize_with = "serialize_regex",
813 deserialize_with = "deserialize_regex"
814 )]
815 #[schemars(schema_with = "regex_json_schema")]
816 pub first_line_pattern: Option<Regex>,
817}
818
819/// The configuration for JSX tag auto-closing.
820#[derive(Clone, Deserialize, JsonSchema)]
821pub struct JsxTagAutoCloseConfig {
822 /// The name of the node for a opening tag
823 pub open_tag_node_name: String,
824 /// The name of the node for an closing tag
825 pub close_tag_node_name: String,
826 /// The name of the node for a complete element with children for open and close tags
827 pub jsx_element_node_name: String,
828 /// The name of the node found within both opening and closing
829 /// tags that describes the tag name
830 pub tag_name_node_name: String,
831 /// Alternate Node names for tag names.
832 /// Specifically needed as TSX represents the name in `<Foo.Bar>`
833 /// as `member_expression` rather than `identifier` as usual
834 #[serde(default)]
835 pub tag_name_node_name_alternates: Vec<String>,
836 /// Some grammars are smart enough to detect a closing tag
837 /// that is not valid i.e. doesn't match it's corresponding
838 /// opening tag or does not have a corresponding opening tag
839 /// This should be set to the name of the node for invalid
840 /// closing tags if the grammar contains such a node, otherwise
841 /// detecting already closed tags will not work properly
842 #[serde(default)]
843 pub erroneous_close_tag_node_name: Option<String>,
844 /// See above for erroneous_close_tag_node_name for details
845 /// This should be set if the node used for the tag name
846 /// within erroneous closing tags is different from the
847 /// normal tag name node name
848 #[serde(default)]
849 pub erroneous_close_tag_name_node_name: Option<String>,
850}
851
852/// The configuration for block comments for this language.
853#[derive(Clone, Debug, JsonSchema, PartialEq)]
854pub struct BlockCommentConfig {
855 /// A start tag of block comment.
856 pub start: Arc<str>,
857 /// A end tag of block comment.
858 pub end: Arc<str>,
859 /// A character to add as a prefix when a new line is added to a block comment.
860 pub prefix: Arc<str>,
861 /// A indent to add for prefix and end line upon new line.
862 pub tab_size: u32,
863}
864
865impl<'de> Deserialize<'de> for BlockCommentConfig {
866 fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
867 where
868 D: Deserializer<'de>,
869 {
870 #[derive(Deserialize)]
871 #[serde(untagged)]
872 enum BlockCommentConfigHelper {
873 New {
874 start: Arc<str>,
875 end: Arc<str>,
876 prefix: Arc<str>,
877 tab_size: u32,
878 },
879 Old([Arc<str>; 2]),
880 }
881
882 match BlockCommentConfigHelper::deserialize(deserializer)? {
883 BlockCommentConfigHelper::New {
884 start,
885 end,
886 prefix,
887 tab_size,
888 } => Ok(BlockCommentConfig {
889 start,
890 end,
891 prefix,
892 tab_size,
893 }),
894 BlockCommentConfigHelper::Old([start, end]) => Ok(BlockCommentConfig {
895 start,
896 end,
897 prefix: "".into(),
898 tab_size: 0,
899 }),
900 }
901 }
902}
903
904/// Represents a language for the given range. Some languages (e.g. HTML)
905/// interleave several languages together, thus a single buffer might actually contain
906/// several nested scopes.
907#[derive(Clone, Debug)]
908pub struct LanguageScope {
909 language: Arc<Language>,
910 override_id: Option<u32>,
911}
912
913#[derive(Clone, Deserialize, Default, Debug, JsonSchema)]
914pub struct LanguageConfigOverride {
915 #[serde(default)]
916 pub line_comments: Override<Vec<Arc<str>>>,
917 #[serde(default)]
918 pub block_comment: Override<BlockCommentConfig>,
919 #[serde(skip)]
920 pub disabled_bracket_ixs: Vec<u16>,
921 #[serde(default)]
922 pub word_characters: Override<HashSet<char>>,
923 #[serde(default)]
924 pub completion_query_characters: Override<HashSet<char>>,
925 #[serde(default)]
926 pub opt_into_language_servers: Vec<LanguageServerName>,
927 #[serde(default)]
928 pub prefer_label_for_snippet: Option<bool>,
929}
930
931#[derive(Clone, Deserialize, Debug, Serialize, JsonSchema)]
932#[serde(untagged)]
933pub enum Override<T> {
934 Remove { remove: bool },
935 Set(T),
936}
937
938impl<T> Default for Override<T> {
939 fn default() -> Self {
940 Override::Remove { remove: false }
941 }
942}
943
944impl<T> Override<T> {
945 fn as_option<'a>(this: Option<&'a Self>, original: Option<&'a T>) -> Option<&'a T> {
946 match this {
947 Some(Self::Set(value)) => Some(value),
948 Some(Self::Remove { remove: true }) => None,
949 Some(Self::Remove { remove: false }) | None => original,
950 }
951 }
952}
953
954impl Default for LanguageConfig {
955 fn default() -> Self {
956 Self {
957 name: LanguageName::new(""),
958 code_fence_block_name: None,
959 grammar: None,
960 matcher: LanguageMatcher::default(),
961 brackets: Default::default(),
962 auto_indent_using_last_non_empty_line: auto_indent_using_last_non_empty_line_default(),
963 auto_indent_on_paste: None,
964 increase_indent_pattern: Default::default(),
965 decrease_indent_pattern: Default::default(),
966 decrease_indent_patterns: Default::default(),
967 autoclose_before: Default::default(),
968 line_comments: Default::default(),
969 block_comment: Default::default(),
970 documentation_comment: Default::default(),
971 rewrap_prefixes: Default::default(),
972 scope_opt_in_language_servers: Default::default(),
973 overrides: Default::default(),
974 word_characters: Default::default(),
975 collapsed_placeholder: Default::default(),
976 hard_tabs: None,
977 tab_size: None,
978 soft_wrap: None,
979 wrap_characters: None,
980 prettier_parser_name: None,
981 hidden: false,
982 jsx_tag_auto_close: None,
983 completion_query_characters: Default::default(),
984 debuggers: Default::default(),
985 }
986 }
987}
988
989#[derive(Clone, Debug, Deserialize, JsonSchema)]
990pub struct WrapCharactersConfig {
991 /// Opening token split into a prefix and suffix. The first caret goes
992 /// after the prefix (i.e., between prefix and suffix).
993 pub start_prefix: String,
994 pub start_suffix: String,
995 /// Closing token split into a prefix and suffix. The second caret goes
996 /// after the prefix (i.e., between prefix and suffix).
997 pub end_prefix: String,
998 pub end_suffix: String,
999}
1000
1001fn auto_indent_using_last_non_empty_line_default() -> bool {
1002 true
1003}
1004
1005fn deserialize_regex<'de, D: Deserializer<'de>>(d: D) -> Result<Option<Regex>, D::Error> {
1006 let source = Option::<String>::deserialize(d)?;
1007 if let Some(source) = source {
1008 Ok(Some(regex::Regex::new(&source).map_err(de::Error::custom)?))
1009 } else {
1010 Ok(None)
1011 }
1012}
1013
1014fn regex_json_schema(_: &mut schemars::SchemaGenerator) -> schemars::Schema {
1015 json_schema!({
1016 "type": "string"
1017 })
1018}
1019
1020fn serialize_regex<S>(regex: &Option<Regex>, serializer: S) -> Result<S::Ok, S::Error>
1021where
1022 S: Serializer,
1023{
1024 match regex {
1025 Some(regex) => serializer.serialize_str(regex.as_str()),
1026 None => serializer.serialize_none(),
1027 }
1028}
1029
1030fn deserialize_regex_vec<'de, D: Deserializer<'de>>(d: D) -> Result<Vec<Regex>, D::Error> {
1031 let sources = Vec::<String>::deserialize(d)?;
1032 sources
1033 .into_iter()
1034 .map(|source| regex::Regex::new(&source))
1035 .collect::<Result<_, _>>()
1036 .map_err(de::Error::custom)
1037}
1038
1039fn regex_vec_json_schema(_: &mut SchemaGenerator) -> schemars::Schema {
1040 json_schema!({
1041 "type": "array",
1042 "items": { "type": "string" }
1043 })
1044}
1045
1046#[doc(hidden)]
1047#[cfg(any(test, feature = "test-support"))]
1048pub struct FakeLspAdapter {
1049 pub name: &'static str,
1050 pub initialization_options: Option<Value>,
1051 pub prettier_plugins: Vec<&'static str>,
1052 pub disk_based_diagnostics_progress_token: Option<String>,
1053 pub disk_based_diagnostics_sources: Vec<String>,
1054 pub language_server_binary: LanguageServerBinary,
1055
1056 pub capabilities: lsp::ServerCapabilities,
1057 pub initializer: Option<Box<dyn 'static + Send + Sync + Fn(&mut lsp::FakeLanguageServer)>>,
1058 pub label_for_completion: Option<
1059 Box<
1060 dyn 'static
1061 + Send
1062 + Sync
1063 + Fn(&lsp::CompletionItem, &Arc<Language>) -> Option<CodeLabel>,
1064 >,
1065 >,
1066}
1067
1068/// Configuration of handling bracket pairs for a given language.
1069///
1070/// This struct includes settings for defining which pairs of characters are considered brackets and
1071/// also specifies any language-specific scopes where these pairs should be ignored for bracket matching purposes.
1072#[derive(Clone, Debug, Default, JsonSchema)]
1073#[schemars(with = "Vec::<BracketPairContent>")]
1074pub struct BracketPairConfig {
1075 /// A list of character pairs that should be treated as brackets in the context of a given language.
1076 pub pairs: Vec<BracketPair>,
1077 /// A list of tree-sitter scopes for which a given bracket should not be active.
1078 /// N-th entry in `[Self::disabled_scopes_by_bracket_ix]` contains a list of disabled scopes for an n-th entry in `[Self::pairs]`
1079 pub disabled_scopes_by_bracket_ix: Vec<Vec<String>>,
1080}
1081
1082impl BracketPairConfig {
1083 pub fn is_closing_brace(&self, c: char) -> bool {
1084 self.pairs.iter().any(|pair| pair.end.starts_with(c))
1085 }
1086}
1087
1088#[derive(Deserialize, JsonSchema)]
1089pub struct BracketPairContent {
1090 #[serde(flatten)]
1091 pub bracket_pair: BracketPair,
1092 #[serde(default)]
1093 pub not_in: Vec<String>,
1094}
1095
1096impl<'de> Deserialize<'de> for BracketPairConfig {
1097 fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
1098 where
1099 D: Deserializer<'de>,
1100 {
1101 let result = Vec::<BracketPairContent>::deserialize(deserializer)?;
1102 let (brackets, disabled_scopes_by_bracket_ix) = result
1103 .into_iter()
1104 .map(|entry| (entry.bracket_pair, entry.not_in))
1105 .unzip();
1106
1107 Ok(BracketPairConfig {
1108 pairs: brackets,
1109 disabled_scopes_by_bracket_ix,
1110 })
1111 }
1112}
1113
1114/// Describes a single bracket pair and how an editor should react to e.g. inserting
1115/// an opening bracket or to a newline character insertion in between `start` and `end` characters.
1116#[derive(Clone, Debug, Default, Deserialize, PartialEq, JsonSchema)]
1117pub struct BracketPair {
1118 /// Starting substring for a bracket.
1119 pub start: String,
1120 /// Ending substring for a bracket.
1121 pub end: String,
1122 /// True if `end` should be automatically inserted right after `start` characters.
1123 pub close: bool,
1124 /// True if selected text should be surrounded by `start` and `end` characters.
1125 #[serde(default = "default_true")]
1126 pub surround: bool,
1127 /// True if an extra newline should be inserted while the cursor is in the middle
1128 /// of that bracket pair.
1129 pub newline: bool,
1130}
1131
1132#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
1133pub struct LanguageId(usize);
1134
1135impl LanguageId {
1136 pub(crate) fn new() -> Self {
1137 Self(NEXT_LANGUAGE_ID.fetch_add(1, SeqCst))
1138 }
1139}
1140
1141pub struct Language {
1142 pub(crate) id: LanguageId,
1143 pub(crate) config: LanguageConfig,
1144 pub(crate) grammar: Option<Arc<Grammar>>,
1145 pub(crate) context_provider: Option<Arc<dyn ContextProvider>>,
1146 pub(crate) toolchain: Option<Arc<dyn ToolchainLister>>,
1147 pub(crate) manifest_name: Option<ManifestName>,
1148}
1149
1150#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
1151pub struct GrammarId(pub usize);
1152
1153impl GrammarId {
1154 pub(crate) fn new() -> Self {
1155 Self(NEXT_GRAMMAR_ID.fetch_add(1, SeqCst))
1156 }
1157}
1158
1159pub struct Grammar {
1160 id: GrammarId,
1161 pub ts_language: tree_sitter::Language,
1162 pub(crate) error_query: Option<Query>,
1163 pub(crate) highlights_query: Option<Query>,
1164 pub(crate) brackets_config: Option<BracketsConfig>,
1165 pub(crate) redactions_config: Option<RedactionConfig>,
1166 pub(crate) runnable_config: Option<RunnableConfig>,
1167 pub(crate) indents_config: Option<IndentConfig>,
1168 pub outline_config: Option<OutlineConfig>,
1169 pub text_object_config: Option<TextObjectConfig>,
1170 pub embedding_config: Option<EmbeddingConfig>,
1171 pub(crate) injection_config: Option<InjectionConfig>,
1172 pub(crate) override_config: Option<OverrideConfig>,
1173 pub(crate) debug_variables_config: Option<DebugVariablesConfig>,
1174 pub(crate) highlight_map: Mutex<HighlightMap>,
1175}
1176
1177struct IndentConfig {
1178 query: Query,
1179 indent_capture_ix: u32,
1180 start_capture_ix: Option<u32>,
1181 end_capture_ix: Option<u32>,
1182 outdent_capture_ix: Option<u32>,
1183 suffixed_start_captures: HashMap<u32, SharedString>,
1184}
1185
1186pub struct OutlineConfig {
1187 pub query: Query,
1188 pub item_capture_ix: u32,
1189 pub name_capture_ix: u32,
1190 pub context_capture_ix: Option<u32>,
1191 pub extra_context_capture_ix: Option<u32>,
1192 pub open_capture_ix: Option<u32>,
1193 pub close_capture_ix: Option<u32>,
1194 pub annotation_capture_ix: Option<u32>,
1195}
1196
1197#[derive(Debug, Clone, Copy, PartialEq)]
1198pub enum DebuggerTextObject {
1199 Variable,
1200 Scope,
1201}
1202
1203impl DebuggerTextObject {
1204 pub fn from_capture_name(name: &str) -> Option<DebuggerTextObject> {
1205 match name {
1206 "debug-variable" => Some(DebuggerTextObject::Variable),
1207 "debug-scope" => Some(DebuggerTextObject::Scope),
1208 _ => None,
1209 }
1210 }
1211}
1212
1213#[derive(Debug, Clone, Copy, PartialEq)]
1214pub enum TextObject {
1215 InsideFunction,
1216 AroundFunction,
1217 InsideClass,
1218 AroundClass,
1219 InsideComment,
1220 AroundComment,
1221}
1222
1223impl TextObject {
1224 pub fn from_capture_name(name: &str) -> Option<TextObject> {
1225 match name {
1226 "function.inside" => Some(TextObject::InsideFunction),
1227 "function.around" => Some(TextObject::AroundFunction),
1228 "class.inside" => Some(TextObject::InsideClass),
1229 "class.around" => Some(TextObject::AroundClass),
1230 "comment.inside" => Some(TextObject::InsideComment),
1231 "comment.around" => Some(TextObject::AroundComment),
1232 _ => None,
1233 }
1234 }
1235
1236 pub fn around(&self) -> Option<Self> {
1237 match self {
1238 TextObject::InsideFunction => Some(TextObject::AroundFunction),
1239 TextObject::InsideClass => Some(TextObject::AroundClass),
1240 TextObject::InsideComment => Some(TextObject::AroundComment),
1241 _ => None,
1242 }
1243 }
1244}
1245
1246pub struct TextObjectConfig {
1247 pub query: Query,
1248 pub text_objects_by_capture_ix: Vec<(u32, TextObject)>,
1249}
1250
1251#[derive(Debug)]
1252pub struct EmbeddingConfig {
1253 pub query: Query,
1254 pub item_capture_ix: u32,
1255 pub name_capture_ix: Option<u32>,
1256 pub context_capture_ix: Option<u32>,
1257 pub collapse_capture_ix: Option<u32>,
1258 pub keep_capture_ix: Option<u32>,
1259}
1260
1261struct InjectionConfig {
1262 query: Query,
1263 content_capture_ix: u32,
1264 language_capture_ix: Option<u32>,
1265 patterns: Vec<InjectionPatternConfig>,
1266}
1267
1268struct RedactionConfig {
1269 pub query: Query,
1270 pub redaction_capture_ix: u32,
1271}
1272
1273#[derive(Clone, Debug, PartialEq)]
1274enum RunnableCapture {
1275 Named(SharedString),
1276 Run,
1277}
1278
1279struct RunnableConfig {
1280 pub query: Query,
1281 /// A mapping from capture indice to capture kind
1282 pub extra_captures: Vec<RunnableCapture>,
1283}
1284
1285struct OverrideConfig {
1286 query: Query,
1287 values: HashMap<u32, OverrideEntry>,
1288}
1289
1290#[derive(Debug)]
1291struct OverrideEntry {
1292 name: String,
1293 range_is_inclusive: bool,
1294 value: LanguageConfigOverride,
1295}
1296
1297#[derive(Default, Clone)]
1298struct InjectionPatternConfig {
1299 language: Option<Box<str>>,
1300 combined: bool,
1301}
1302
1303#[derive(Debug)]
1304struct BracketsConfig {
1305 query: Query,
1306 open_capture_ix: u32,
1307 close_capture_ix: u32,
1308 patterns: Vec<BracketsPatternConfig>,
1309}
1310
1311#[derive(Clone, Debug, Default)]
1312struct BracketsPatternConfig {
1313 newline_only: bool,
1314}
1315
1316pub struct DebugVariablesConfig {
1317 pub query: Query,
1318 pub objects_by_capture_ix: Vec<(u32, DebuggerTextObject)>,
1319}
1320
1321impl Language {
1322 pub fn new(config: LanguageConfig, ts_language: Option<tree_sitter::Language>) -> Self {
1323 Self::new_with_id(LanguageId::new(), config, ts_language)
1324 }
1325
1326 pub fn id(&self) -> LanguageId {
1327 self.id
1328 }
1329
1330 fn new_with_id(
1331 id: LanguageId,
1332 config: LanguageConfig,
1333 ts_language: Option<tree_sitter::Language>,
1334 ) -> Self {
1335 Self {
1336 id,
1337 config,
1338 grammar: ts_language.map(|ts_language| {
1339 Arc::new(Grammar {
1340 id: GrammarId::new(),
1341 highlights_query: None,
1342 brackets_config: None,
1343 outline_config: None,
1344 text_object_config: None,
1345 embedding_config: None,
1346 indents_config: None,
1347 injection_config: None,
1348 override_config: None,
1349 redactions_config: None,
1350 runnable_config: None,
1351 error_query: Query::new(&ts_language, "(ERROR) @error").ok(),
1352 debug_variables_config: None,
1353 ts_language,
1354 highlight_map: Default::default(),
1355 })
1356 }),
1357 context_provider: None,
1358 toolchain: None,
1359 manifest_name: None,
1360 }
1361 }
1362
1363 pub fn with_context_provider(mut self, provider: Option<Arc<dyn ContextProvider>>) -> Self {
1364 self.context_provider = provider;
1365 self
1366 }
1367
1368 pub fn with_toolchain_lister(mut self, provider: Option<Arc<dyn ToolchainLister>>) -> Self {
1369 self.toolchain = provider;
1370 self
1371 }
1372
1373 pub fn with_manifest(mut self, name: Option<ManifestName>) -> Self {
1374 self.manifest_name = name;
1375 self
1376 }
1377
1378 pub fn with_queries(mut self, queries: LanguageQueries) -> Result<Self> {
1379 if let Some(query) = queries.highlights {
1380 self = self
1381 .with_highlights_query(query.as_ref())
1382 .context("Error loading highlights query")?;
1383 }
1384 if let Some(query) = queries.brackets {
1385 self = self
1386 .with_brackets_query(query.as_ref())
1387 .context("Error loading brackets query")?;
1388 }
1389 if let Some(query) = queries.indents {
1390 self = self
1391 .with_indents_query(query.as_ref())
1392 .context("Error loading indents query")?;
1393 }
1394 if let Some(query) = queries.outline {
1395 self = self
1396 .with_outline_query(query.as_ref())
1397 .context("Error loading outline query")?;
1398 }
1399 if let Some(query) = queries.embedding {
1400 self = self
1401 .with_embedding_query(query.as_ref())
1402 .context("Error loading embedding query")?;
1403 }
1404 if let Some(query) = queries.injections {
1405 self = self
1406 .with_injection_query(query.as_ref())
1407 .context("Error loading injection query")?;
1408 }
1409 if let Some(query) = queries.overrides {
1410 self = self
1411 .with_override_query(query.as_ref())
1412 .context("Error loading override query")?;
1413 }
1414 if let Some(query) = queries.redactions {
1415 self = self
1416 .with_redaction_query(query.as_ref())
1417 .context("Error loading redaction query")?;
1418 }
1419 if let Some(query) = queries.runnables {
1420 self = self
1421 .with_runnable_query(query.as_ref())
1422 .context("Error loading runnables query")?;
1423 }
1424 if let Some(query) = queries.text_objects {
1425 self = self
1426 .with_text_object_query(query.as_ref())
1427 .context("Error loading textobject query")?;
1428 }
1429 if let Some(query) = queries.debugger {
1430 self = self
1431 .with_debug_variables_query(query.as_ref())
1432 .context("Error loading debug variables query")?;
1433 }
1434 Ok(self)
1435 }
1436
1437 pub fn with_highlights_query(mut self, source: &str) -> Result<Self> {
1438 let grammar = self.grammar_mut()?;
1439 grammar.highlights_query = Some(Query::new(&grammar.ts_language, source)?);
1440 Ok(self)
1441 }
1442
1443 pub fn with_runnable_query(mut self, source: &str) -> Result<Self> {
1444 let grammar = self.grammar_mut()?;
1445
1446 let query = Query::new(&grammar.ts_language, source)?;
1447 let extra_captures: Vec<_> = query
1448 .capture_names()
1449 .iter()
1450 .map(|&name| match name {
1451 "run" => RunnableCapture::Run,
1452 name => RunnableCapture::Named(name.to_string().into()),
1453 })
1454 .collect();
1455
1456 grammar.runnable_config = Some(RunnableConfig {
1457 extra_captures,
1458 query,
1459 });
1460
1461 Ok(self)
1462 }
1463
1464 pub fn with_outline_query(mut self, source: &str) -> Result<Self> {
1465 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1466 let mut item_capture_ix = 0;
1467 let mut name_capture_ix = 0;
1468 let mut context_capture_ix = None;
1469 let mut extra_context_capture_ix = None;
1470 let mut open_capture_ix = None;
1471 let mut close_capture_ix = None;
1472 let mut annotation_capture_ix = None;
1473 if populate_capture_indices(
1474 &query,
1475 &self.config.name,
1476 "outline",
1477 &[],
1478 &mut [
1479 Capture::Required("item", &mut item_capture_ix),
1480 Capture::Required("name", &mut name_capture_ix),
1481 Capture::Optional("context", &mut context_capture_ix),
1482 Capture::Optional("context.extra", &mut extra_context_capture_ix),
1483 Capture::Optional("open", &mut open_capture_ix),
1484 Capture::Optional("close", &mut close_capture_ix),
1485 Capture::Optional("annotation", &mut annotation_capture_ix),
1486 ],
1487 ) {
1488 self.grammar_mut()?.outline_config = Some(OutlineConfig {
1489 query,
1490 item_capture_ix,
1491 name_capture_ix,
1492 context_capture_ix,
1493 extra_context_capture_ix,
1494 open_capture_ix,
1495 close_capture_ix,
1496 annotation_capture_ix,
1497 });
1498 }
1499 Ok(self)
1500 }
1501
1502 pub fn with_text_object_query(mut self, source: &str) -> Result<Self> {
1503 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1504
1505 let mut text_objects_by_capture_ix = Vec::new();
1506 for (ix, name) in query.capture_names().iter().enumerate() {
1507 if let Some(text_object) = TextObject::from_capture_name(name) {
1508 text_objects_by_capture_ix.push((ix as u32, text_object));
1509 } else {
1510 log::warn!(
1511 "unrecognized capture name '{}' in {} textobjects TreeSitter query",
1512 name,
1513 self.config.name,
1514 );
1515 }
1516 }
1517
1518 self.grammar_mut()?.text_object_config = Some(TextObjectConfig {
1519 query,
1520 text_objects_by_capture_ix,
1521 });
1522 Ok(self)
1523 }
1524
1525 pub fn with_embedding_query(mut self, source: &str) -> Result<Self> {
1526 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1527 let mut item_capture_ix = 0;
1528 let mut name_capture_ix = None;
1529 let mut context_capture_ix = None;
1530 let mut collapse_capture_ix = None;
1531 let mut keep_capture_ix = None;
1532 if populate_capture_indices(
1533 &query,
1534 &self.config.name,
1535 "embedding",
1536 &[],
1537 &mut [
1538 Capture::Required("item", &mut item_capture_ix),
1539 Capture::Optional("name", &mut name_capture_ix),
1540 Capture::Optional("context", &mut context_capture_ix),
1541 Capture::Optional("keep", &mut keep_capture_ix),
1542 Capture::Optional("collapse", &mut collapse_capture_ix),
1543 ],
1544 ) {
1545 self.grammar_mut()?.embedding_config = Some(EmbeddingConfig {
1546 query,
1547 item_capture_ix,
1548 name_capture_ix,
1549 context_capture_ix,
1550 collapse_capture_ix,
1551 keep_capture_ix,
1552 });
1553 }
1554 Ok(self)
1555 }
1556
1557 pub fn with_debug_variables_query(mut self, source: &str) -> Result<Self> {
1558 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1559
1560 let mut objects_by_capture_ix = Vec::new();
1561 for (ix, name) in query.capture_names().iter().enumerate() {
1562 if let Some(text_object) = DebuggerTextObject::from_capture_name(name) {
1563 objects_by_capture_ix.push((ix as u32, text_object));
1564 } else {
1565 log::warn!(
1566 "unrecognized capture name '{}' in {} debugger TreeSitter query",
1567 name,
1568 self.config.name,
1569 );
1570 }
1571 }
1572
1573 self.grammar_mut()?.debug_variables_config = Some(DebugVariablesConfig {
1574 query,
1575 objects_by_capture_ix,
1576 });
1577 Ok(self)
1578 }
1579
1580 pub fn with_brackets_query(mut self, source: &str) -> Result<Self> {
1581 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1582 let mut open_capture_ix = 0;
1583 let mut close_capture_ix = 0;
1584 if populate_capture_indices(
1585 &query,
1586 &self.config.name,
1587 "brackets",
1588 &[],
1589 &mut [
1590 Capture::Required("open", &mut open_capture_ix),
1591 Capture::Required("close", &mut close_capture_ix),
1592 ],
1593 ) {
1594 let patterns = (0..query.pattern_count())
1595 .map(|ix| {
1596 let mut config = BracketsPatternConfig::default();
1597 for setting in query.property_settings(ix) {
1598 if setting.key.as_ref() == "newline.only" {
1599 config.newline_only = true
1600 }
1601 }
1602 config
1603 })
1604 .collect();
1605 self.grammar_mut()?.brackets_config = Some(BracketsConfig {
1606 query,
1607 open_capture_ix,
1608 close_capture_ix,
1609 patterns,
1610 });
1611 }
1612 Ok(self)
1613 }
1614
1615 pub fn with_indents_query(mut self, source: &str) -> Result<Self> {
1616 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1617 let mut indent_capture_ix = 0;
1618 let mut start_capture_ix = None;
1619 let mut end_capture_ix = None;
1620 let mut outdent_capture_ix = None;
1621 if populate_capture_indices(
1622 &query,
1623 &self.config.name,
1624 "indents",
1625 &["start."],
1626 &mut [
1627 Capture::Required("indent", &mut indent_capture_ix),
1628 Capture::Optional("start", &mut start_capture_ix),
1629 Capture::Optional("end", &mut end_capture_ix),
1630 Capture::Optional("outdent", &mut outdent_capture_ix),
1631 ],
1632 ) {
1633 let mut suffixed_start_captures = HashMap::default();
1634 for (ix, name) in query.capture_names().iter().enumerate() {
1635 if let Some(suffix) = name.strip_prefix("start.") {
1636 suffixed_start_captures.insert(ix as u32, suffix.to_owned().into());
1637 }
1638 }
1639
1640 self.grammar_mut()?.indents_config = Some(IndentConfig {
1641 query,
1642 indent_capture_ix,
1643 start_capture_ix,
1644 end_capture_ix,
1645 outdent_capture_ix,
1646 suffixed_start_captures,
1647 });
1648 }
1649 Ok(self)
1650 }
1651
1652 pub fn with_injection_query(mut self, source: &str) -> Result<Self> {
1653 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1654 let mut language_capture_ix = None;
1655 let mut injection_language_capture_ix = None;
1656 let mut content_capture_ix = None;
1657 let mut injection_content_capture_ix = None;
1658 if populate_capture_indices(
1659 &query,
1660 &self.config.name,
1661 "injections",
1662 &[],
1663 &mut [
1664 Capture::Optional("language", &mut language_capture_ix),
1665 Capture::Optional("injection.language", &mut injection_language_capture_ix),
1666 Capture::Optional("content", &mut content_capture_ix),
1667 Capture::Optional("injection.content", &mut injection_content_capture_ix),
1668 ],
1669 ) {
1670 language_capture_ix = match (language_capture_ix, injection_language_capture_ix) {
1671 (None, Some(ix)) => Some(ix),
1672 (Some(_), Some(_)) => {
1673 anyhow::bail!("both language and injection.language captures are present");
1674 }
1675 _ => language_capture_ix,
1676 };
1677 content_capture_ix = match (content_capture_ix, injection_content_capture_ix) {
1678 (None, Some(ix)) => Some(ix),
1679 (Some(_), Some(_)) => {
1680 anyhow::bail!("both content and injection.content captures are present")
1681 }
1682 _ => content_capture_ix,
1683 };
1684 let patterns = (0..query.pattern_count())
1685 .map(|ix| {
1686 let mut config = InjectionPatternConfig::default();
1687 for setting in query.property_settings(ix) {
1688 match setting.key.as_ref() {
1689 "language" | "injection.language" => {
1690 config.language.clone_from(&setting.value);
1691 }
1692 "combined" | "injection.combined" => {
1693 config.combined = true;
1694 }
1695 _ => {}
1696 }
1697 }
1698 config
1699 })
1700 .collect();
1701 if let Some(content_capture_ix) = content_capture_ix {
1702 self.grammar_mut()?.injection_config = Some(InjectionConfig {
1703 query,
1704 language_capture_ix,
1705 content_capture_ix,
1706 patterns,
1707 });
1708 } else {
1709 log::error!(
1710 "missing required capture in injections {} TreeSitter query: \
1711 content or injection.content",
1712 &self.config.name,
1713 );
1714 }
1715 }
1716 Ok(self)
1717 }
1718
1719 pub fn with_override_query(mut self, source: &str) -> anyhow::Result<Self> {
1720 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1721
1722 let mut override_configs_by_id = HashMap::default();
1723 for (ix, mut name) in query.capture_names().iter().copied().enumerate() {
1724 let mut range_is_inclusive = false;
1725 if name.starts_with('_') {
1726 continue;
1727 }
1728 if let Some(prefix) = name.strip_suffix(".inclusive") {
1729 name = prefix;
1730 range_is_inclusive = true;
1731 }
1732
1733 let value = self.config.overrides.get(name).cloned().unwrap_or_default();
1734 for server_name in &value.opt_into_language_servers {
1735 if !self
1736 .config
1737 .scope_opt_in_language_servers
1738 .contains(server_name)
1739 {
1740 util::debug_panic!(
1741 "Server {server_name:?} has been opted-in by scope {name:?} but has not been marked as an opt-in server"
1742 );
1743 }
1744 }
1745
1746 override_configs_by_id.insert(
1747 ix as u32,
1748 OverrideEntry {
1749 name: name.to_string(),
1750 range_is_inclusive,
1751 value,
1752 },
1753 );
1754 }
1755
1756 let referenced_override_names = self.config.overrides.keys().chain(
1757 self.config
1758 .brackets
1759 .disabled_scopes_by_bracket_ix
1760 .iter()
1761 .flatten(),
1762 );
1763
1764 for referenced_name in referenced_override_names {
1765 if !override_configs_by_id
1766 .values()
1767 .any(|entry| entry.name == *referenced_name)
1768 {
1769 anyhow::bail!(
1770 "language {:?} has overrides in config not in query: {referenced_name:?}",
1771 self.config.name
1772 );
1773 }
1774 }
1775
1776 for entry in override_configs_by_id.values_mut() {
1777 entry.value.disabled_bracket_ixs = self
1778 .config
1779 .brackets
1780 .disabled_scopes_by_bracket_ix
1781 .iter()
1782 .enumerate()
1783 .filter_map(|(ix, disabled_scope_names)| {
1784 if disabled_scope_names.contains(&entry.name) {
1785 Some(ix as u16)
1786 } else {
1787 None
1788 }
1789 })
1790 .collect();
1791 }
1792
1793 self.config.brackets.disabled_scopes_by_bracket_ix.clear();
1794
1795 let grammar = self.grammar_mut()?;
1796 grammar.override_config = Some(OverrideConfig {
1797 query,
1798 values: override_configs_by_id,
1799 });
1800 Ok(self)
1801 }
1802
1803 pub fn with_redaction_query(mut self, source: &str) -> anyhow::Result<Self> {
1804 let query = Query::new(&self.expect_grammar()?.ts_language, source)?;
1805 let mut redaction_capture_ix = 0;
1806 if populate_capture_indices(
1807 &query,
1808 &self.config.name,
1809 "redactions",
1810 &[],
1811 &mut [Capture::Required("redact", &mut redaction_capture_ix)],
1812 ) {
1813 self.grammar_mut()?.redactions_config = Some(RedactionConfig {
1814 query,
1815 redaction_capture_ix,
1816 });
1817 }
1818 Ok(self)
1819 }
1820
1821 fn expect_grammar(&self) -> Result<&Grammar> {
1822 self.grammar
1823 .as_ref()
1824 .map(|grammar| grammar.as_ref())
1825 .context("no grammar for language")
1826 }
1827
1828 fn grammar_mut(&mut self) -> Result<&mut Grammar> {
1829 Arc::get_mut(self.grammar.as_mut().context("no grammar for language")?)
1830 .context("cannot mutate grammar")
1831 }
1832
1833 pub fn name(&self) -> LanguageName {
1834 self.config.name.clone()
1835 }
1836 pub fn manifest(&self) -> Option<&ManifestName> {
1837 self.manifest_name.as_ref()
1838 }
1839
1840 pub fn code_fence_block_name(&self) -> Arc<str> {
1841 self.config
1842 .code_fence_block_name
1843 .clone()
1844 .unwrap_or_else(|| self.config.name.as_ref().to_lowercase().into())
1845 }
1846
1847 pub fn context_provider(&self) -> Option<Arc<dyn ContextProvider>> {
1848 self.context_provider.clone()
1849 }
1850
1851 pub fn toolchain_lister(&self) -> Option<Arc<dyn ToolchainLister>> {
1852 self.toolchain.clone()
1853 }
1854
1855 pub fn highlight_text<'a>(
1856 self: &'a Arc<Self>,
1857 text: &'a Rope,
1858 range: Range<usize>,
1859 ) -> Vec<(Range<usize>, HighlightId)> {
1860 let mut result = Vec::new();
1861 if let Some(grammar) = &self.grammar {
1862 let tree = grammar.parse_text(text, None);
1863 let captures =
1864 SyntaxSnapshot::single_tree_captures(range.clone(), text, &tree, self, |grammar| {
1865 grammar.highlights_query.as_ref()
1866 });
1867 let highlight_maps = vec![grammar.highlight_map()];
1868 let mut offset = 0;
1869 for chunk in
1870 BufferChunks::new(text, range, Some((captures, highlight_maps)), false, None)
1871 {
1872 let end_offset = offset + chunk.text.len();
1873 if let Some(highlight_id) = chunk.syntax_highlight_id
1874 && !highlight_id.is_default()
1875 {
1876 result.push((offset..end_offset, highlight_id));
1877 }
1878 offset = end_offset;
1879 }
1880 }
1881 result
1882 }
1883
1884 pub fn path_suffixes(&self) -> &[String] {
1885 &self.config.matcher.path_suffixes
1886 }
1887
1888 pub fn should_autoclose_before(&self, c: char) -> bool {
1889 c.is_whitespace() || self.config.autoclose_before.contains(c)
1890 }
1891
1892 pub fn set_theme(&self, theme: &SyntaxTheme) {
1893 if let Some(grammar) = self.grammar.as_ref()
1894 && let Some(highlights_query) = &grammar.highlights_query
1895 {
1896 *grammar.highlight_map.lock() =
1897 HighlightMap::new(highlights_query.capture_names(), theme);
1898 }
1899 }
1900
1901 pub fn grammar(&self) -> Option<&Arc<Grammar>> {
1902 self.grammar.as_ref()
1903 }
1904
1905 pub fn default_scope(self: &Arc<Self>) -> LanguageScope {
1906 LanguageScope {
1907 language: self.clone(),
1908 override_id: None,
1909 }
1910 }
1911
1912 pub fn lsp_id(&self) -> String {
1913 self.config.name.lsp_id()
1914 }
1915
1916 pub fn prettier_parser_name(&self) -> Option<&str> {
1917 self.config.prettier_parser_name.as_deref()
1918 }
1919
1920 pub fn config(&self) -> &LanguageConfig {
1921 &self.config
1922 }
1923}
1924
1925impl LanguageScope {
1926 pub fn path_suffixes(&self) -> &[String] {
1927 self.language.path_suffixes()
1928 }
1929
1930 pub fn language_name(&self) -> LanguageName {
1931 self.language.config.name.clone()
1932 }
1933
1934 pub fn collapsed_placeholder(&self) -> &str {
1935 self.language.config.collapsed_placeholder.as_ref()
1936 }
1937
1938 /// Returns line prefix that is inserted in e.g. line continuations or
1939 /// in `toggle comments` action.
1940 pub fn line_comment_prefixes(&self) -> &[Arc<str>] {
1941 Override::as_option(
1942 self.config_override().map(|o| &o.line_comments),
1943 Some(&self.language.config.line_comments),
1944 )
1945 .map_or([].as_slice(), |e| e.as_slice())
1946 }
1947
1948 /// Config for block comments for this language.
1949 pub fn block_comment(&self) -> Option<&BlockCommentConfig> {
1950 Override::as_option(
1951 self.config_override().map(|o| &o.block_comment),
1952 self.language.config.block_comment.as_ref(),
1953 )
1954 }
1955
1956 /// Config for documentation-style block comments for this language.
1957 pub fn documentation_comment(&self) -> Option<&BlockCommentConfig> {
1958 self.language.config.documentation_comment.as_ref()
1959 }
1960
1961 /// Returns additional regex patterns that act as prefix markers for creating
1962 /// boundaries during rewrapping.
1963 ///
1964 /// By default, Zed treats as paragraph and comment prefixes as boundaries.
1965 pub fn rewrap_prefixes(&self) -> &[Regex] {
1966 &self.language.config.rewrap_prefixes
1967 }
1968
1969 /// Returns a list of language-specific word characters.
1970 ///
1971 /// By default, Zed treats alphanumeric characters (and '_') as word characters for
1972 /// the purpose of actions like 'move to next word end` or whole-word search.
1973 /// It additionally accounts for language's additional word characters.
1974 pub fn word_characters(&self) -> Option<&HashSet<char>> {
1975 Override::as_option(
1976 self.config_override().map(|o| &o.word_characters),
1977 Some(&self.language.config.word_characters),
1978 )
1979 }
1980
1981 /// Returns a list of language-specific characters that are considered part of
1982 /// a completion query.
1983 pub fn completion_query_characters(&self) -> Option<&HashSet<char>> {
1984 Override::as_option(
1985 self.config_override()
1986 .map(|o| &o.completion_query_characters),
1987 Some(&self.language.config.completion_query_characters),
1988 )
1989 }
1990
1991 /// Returns whether to prefer snippet `label` over `new_text` to replace text when
1992 /// completion is accepted.
1993 ///
1994 /// In cases like when cursor is in string or renaming existing function,
1995 /// you don't want to expand function signature instead just want function name
1996 /// to replace existing one.
1997 pub fn prefers_label_for_snippet_in_completion(&self) -> bool {
1998 self.config_override()
1999 .and_then(|o| o.prefer_label_for_snippet)
2000 .unwrap_or(false)
2001 }
2002
2003 /// Returns a list of bracket pairs for a given language with an additional
2004 /// piece of information about whether the particular bracket pair is currently active for a given language.
2005 pub fn brackets(&self) -> impl Iterator<Item = (&BracketPair, bool)> {
2006 let mut disabled_ids = self
2007 .config_override()
2008 .map_or(&[] as _, |o| o.disabled_bracket_ixs.as_slice());
2009 self.language
2010 .config
2011 .brackets
2012 .pairs
2013 .iter()
2014 .enumerate()
2015 .map(move |(ix, bracket)| {
2016 let mut is_enabled = true;
2017 if let Some(next_disabled_ix) = disabled_ids.first()
2018 && ix == *next_disabled_ix as usize
2019 {
2020 disabled_ids = &disabled_ids[1..];
2021 is_enabled = false;
2022 }
2023 (bracket, is_enabled)
2024 })
2025 }
2026
2027 pub fn should_autoclose_before(&self, c: char) -> bool {
2028 c.is_whitespace() || self.language.config.autoclose_before.contains(c)
2029 }
2030
2031 pub fn language_allowed(&self, name: &LanguageServerName) -> bool {
2032 let config = &self.language.config;
2033 let opt_in_servers = &config.scope_opt_in_language_servers;
2034 if opt_in_servers.contains(name) {
2035 if let Some(over) = self.config_override() {
2036 over.opt_into_language_servers.contains(name)
2037 } else {
2038 false
2039 }
2040 } else {
2041 true
2042 }
2043 }
2044
2045 pub fn override_name(&self) -> Option<&str> {
2046 let id = self.override_id?;
2047 let grammar = self.language.grammar.as_ref()?;
2048 let override_config = grammar.override_config.as_ref()?;
2049 override_config.values.get(&id).map(|e| e.name.as_str())
2050 }
2051
2052 fn config_override(&self) -> Option<&LanguageConfigOverride> {
2053 let id = self.override_id?;
2054 let grammar = self.language.grammar.as_ref()?;
2055 let override_config = grammar.override_config.as_ref()?;
2056 override_config.values.get(&id).map(|e| &e.value)
2057 }
2058}
2059
2060impl Hash for Language {
2061 fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
2062 self.id.hash(state)
2063 }
2064}
2065
2066impl PartialEq for Language {
2067 fn eq(&self, other: &Self) -> bool {
2068 self.id.eq(&other.id)
2069 }
2070}
2071
2072impl Eq for Language {}
2073
2074impl Debug for Language {
2075 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
2076 f.debug_struct("Language")
2077 .field("name", &self.config.name)
2078 .finish()
2079 }
2080}
2081
2082impl Grammar {
2083 pub fn id(&self) -> GrammarId {
2084 self.id
2085 }
2086
2087 fn parse_text(&self, text: &Rope, old_tree: Option<Tree>) -> Tree {
2088 with_parser(|parser| {
2089 parser
2090 .set_language(&self.ts_language)
2091 .expect("incompatible grammar");
2092 let mut chunks = text.chunks_in_range(0..text.len());
2093 parser
2094 .parse_with_options(
2095 &mut move |offset, _| {
2096 chunks.seek(offset);
2097 chunks.next().unwrap_or("").as_bytes()
2098 },
2099 old_tree.as_ref(),
2100 None,
2101 )
2102 .unwrap()
2103 })
2104 }
2105
2106 pub fn highlight_map(&self) -> HighlightMap {
2107 self.highlight_map.lock().clone()
2108 }
2109
2110 pub fn highlight_id_for_name(&self, name: &str) -> Option<HighlightId> {
2111 let capture_id = self
2112 .highlights_query
2113 .as_ref()?
2114 .capture_index_for_name(name)?;
2115 Some(self.highlight_map.lock().get(capture_id))
2116 }
2117
2118 pub fn debug_variables_config(&self) -> Option<&DebugVariablesConfig> {
2119 self.debug_variables_config.as_ref()
2120 }
2121}
2122
2123impl CodeLabel {
2124 pub fn fallback_for_completion(
2125 item: &lsp::CompletionItem,
2126 language: Option<&Language>,
2127 ) -> Self {
2128 let highlight_id = item.kind.and_then(|kind| {
2129 let grammar = language?.grammar()?;
2130 use lsp::CompletionItemKind as Kind;
2131 match kind {
2132 Kind::CLASS => grammar.highlight_id_for_name("type"),
2133 Kind::CONSTANT => grammar.highlight_id_for_name("constant"),
2134 Kind::CONSTRUCTOR => grammar.highlight_id_for_name("constructor"),
2135 Kind::ENUM => grammar
2136 .highlight_id_for_name("enum")
2137 .or_else(|| grammar.highlight_id_for_name("type")),
2138 Kind::ENUM_MEMBER => grammar
2139 .highlight_id_for_name("variant")
2140 .or_else(|| grammar.highlight_id_for_name("property")),
2141 Kind::FIELD => grammar.highlight_id_for_name("property"),
2142 Kind::FUNCTION => grammar.highlight_id_for_name("function"),
2143 Kind::INTERFACE => grammar.highlight_id_for_name("type"),
2144 Kind::METHOD => grammar
2145 .highlight_id_for_name("function.method")
2146 .or_else(|| grammar.highlight_id_for_name("function")),
2147 Kind::OPERATOR => grammar.highlight_id_for_name("operator"),
2148 Kind::PROPERTY => grammar.highlight_id_for_name("property"),
2149 Kind::STRUCT => grammar.highlight_id_for_name("type"),
2150 Kind::VARIABLE => grammar.highlight_id_for_name("variable"),
2151 Kind::KEYWORD => grammar.highlight_id_for_name("keyword"),
2152 _ => None,
2153 }
2154 });
2155
2156 let label = &item.label;
2157 let label_length = label.len();
2158 let runs = highlight_id
2159 .map(|highlight_id| vec![(0..label_length, highlight_id)])
2160 .unwrap_or_default();
2161 let text = if let Some(detail) = item.detail.as_deref().filter(|detail| detail != label) {
2162 format!("{label} {detail}")
2163 } else if let Some(description) = item
2164 .label_details
2165 .as_ref()
2166 .and_then(|label_details| label_details.description.as_deref())
2167 .filter(|description| description != label)
2168 {
2169 format!("{label} {description}")
2170 } else {
2171 label.clone()
2172 };
2173 let filter_range = item
2174 .filter_text
2175 .as_deref()
2176 .and_then(|filter| text.find(filter).map(|ix| ix..ix + filter.len()))
2177 .unwrap_or(0..label_length);
2178 Self {
2179 text,
2180 runs,
2181 filter_range,
2182 }
2183 }
2184
2185 pub fn plain(text: String, filter_text: Option<&str>) -> Self {
2186 let filter_range = filter_text
2187 .and_then(|filter| text.find(filter).map(|ix| ix..ix + filter.len()))
2188 .unwrap_or(0..text.len());
2189 Self {
2190 runs: Vec::new(),
2191 filter_range,
2192 text,
2193 }
2194 }
2195
2196 pub fn push_str(&mut self, text: &str, highlight: Option<HighlightId>) {
2197 let start_ix = self.text.len();
2198 self.text.push_str(text);
2199 let end_ix = self.text.len();
2200 if let Some(highlight) = highlight {
2201 self.runs.push((start_ix..end_ix, highlight));
2202 }
2203 }
2204
2205 pub fn text(&self) -> &str {
2206 self.text.as_str()
2207 }
2208
2209 pub fn filter_text(&self) -> &str {
2210 &self.text[self.filter_range.clone()]
2211 }
2212}
2213
2214impl From<String> for CodeLabel {
2215 fn from(value: String) -> Self {
2216 Self::plain(value, None)
2217 }
2218}
2219
2220impl From<&str> for CodeLabel {
2221 fn from(value: &str) -> Self {
2222 Self::plain(value.to_string(), None)
2223 }
2224}
2225
2226impl Ord for LanguageMatcher {
2227 fn cmp(&self, other: &Self) -> std::cmp::Ordering {
2228 self.path_suffixes.cmp(&other.path_suffixes).then_with(|| {
2229 self.first_line_pattern
2230 .as_ref()
2231 .map(Regex::as_str)
2232 .cmp(&other.first_line_pattern.as_ref().map(Regex::as_str))
2233 })
2234 }
2235}
2236
2237impl PartialOrd for LanguageMatcher {
2238 fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
2239 Some(self.cmp(other))
2240 }
2241}
2242
2243impl Eq for LanguageMatcher {}
2244
2245impl PartialEq for LanguageMatcher {
2246 fn eq(&self, other: &Self) -> bool {
2247 self.path_suffixes == other.path_suffixes
2248 && self.first_line_pattern.as_ref().map(Regex::as_str)
2249 == other.first_line_pattern.as_ref().map(Regex::as_str)
2250 }
2251}
2252
2253#[cfg(any(test, feature = "test-support"))]
2254impl Default for FakeLspAdapter {
2255 fn default() -> Self {
2256 Self {
2257 name: "the-fake-language-server",
2258 capabilities: lsp::LanguageServer::full_capabilities(),
2259 initializer: None,
2260 disk_based_diagnostics_progress_token: None,
2261 initialization_options: None,
2262 disk_based_diagnostics_sources: Vec::new(),
2263 prettier_plugins: Vec::new(),
2264 language_server_binary: LanguageServerBinary {
2265 path: "/the/fake/lsp/path".into(),
2266 arguments: vec![],
2267 env: Default::default(),
2268 },
2269 label_for_completion: None,
2270 }
2271 }
2272}
2273
2274#[cfg(any(test, feature = "test-support"))]
2275impl LspInstaller for FakeLspAdapter {
2276 type BinaryVersion = ();
2277
2278 async fn fetch_latest_server_version(
2279 &self,
2280 _: &dyn LspAdapterDelegate,
2281 _: bool,
2282 _: &mut AsyncApp,
2283 ) -> Result<Self::BinaryVersion> {
2284 unreachable!()
2285 }
2286
2287 async fn check_if_user_installed(
2288 &self,
2289 _: &dyn LspAdapterDelegate,
2290 _: Option<Toolchain>,
2291 _: &AsyncApp,
2292 ) -> Option<LanguageServerBinary> {
2293 Some(self.language_server_binary.clone())
2294 }
2295
2296 async fn fetch_server_binary(
2297 &self,
2298 _: (),
2299 _: PathBuf,
2300 _: &dyn LspAdapterDelegate,
2301 ) -> Result<LanguageServerBinary> {
2302 unreachable!();
2303 }
2304
2305 async fn cached_server_binary(
2306 &self,
2307 _: PathBuf,
2308 _: &dyn LspAdapterDelegate,
2309 ) -> Option<LanguageServerBinary> {
2310 unreachable!();
2311 }
2312}
2313
2314#[cfg(any(test, feature = "test-support"))]
2315#[async_trait(?Send)]
2316impl LspAdapter for FakeLspAdapter {
2317 fn name(&self) -> LanguageServerName {
2318 LanguageServerName(self.name.into())
2319 }
2320
2321 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
2322 self.disk_based_diagnostics_sources.clone()
2323 }
2324
2325 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
2326 self.disk_based_diagnostics_progress_token.clone()
2327 }
2328
2329 async fn initialization_options(
2330 self: Arc<Self>,
2331 _: &Arc<dyn LspAdapterDelegate>,
2332 ) -> Result<Option<Value>> {
2333 Ok(self.initialization_options.clone())
2334 }
2335
2336 async fn label_for_completion(
2337 &self,
2338 item: &lsp::CompletionItem,
2339 language: &Arc<Language>,
2340 ) -> Option<CodeLabel> {
2341 let label_for_completion = self.label_for_completion.as_ref()?;
2342 label_for_completion(item, language)
2343 }
2344
2345 fn is_extension(&self) -> bool {
2346 false
2347 }
2348}
2349
2350enum Capture<'a> {
2351 Required(&'static str, &'a mut u32),
2352 Optional(&'static str, &'a mut Option<u32>),
2353}
2354
2355fn populate_capture_indices(
2356 query: &Query,
2357 language_name: &LanguageName,
2358 query_type: &str,
2359 expected_prefixes: &[&str],
2360 captures: &mut [Capture<'_>],
2361) -> bool {
2362 let mut found_required_indices = Vec::new();
2363 'outer: for (ix, name) in query.capture_names().iter().enumerate() {
2364 for (required_ix, capture) in captures.iter_mut().enumerate() {
2365 match capture {
2366 Capture::Required(capture_name, index) if capture_name == name => {
2367 **index = ix as u32;
2368 found_required_indices.push(required_ix);
2369 continue 'outer;
2370 }
2371 Capture::Optional(capture_name, index) if capture_name == name => {
2372 **index = Some(ix as u32);
2373 continue 'outer;
2374 }
2375 _ => {}
2376 }
2377 }
2378 if !name.starts_with("_")
2379 && !expected_prefixes
2380 .iter()
2381 .any(|&prefix| name.starts_with(prefix))
2382 {
2383 log::warn!(
2384 "unrecognized capture name '{}' in {} {} TreeSitter query \
2385 (suppress this warning by prefixing with '_')",
2386 name,
2387 language_name,
2388 query_type
2389 );
2390 }
2391 }
2392 let mut missing_required_captures = Vec::new();
2393 for (capture_ix, capture) in captures.iter().enumerate() {
2394 if let Capture::Required(capture_name, _) = capture
2395 && !found_required_indices.contains(&capture_ix)
2396 {
2397 missing_required_captures.push(*capture_name);
2398 }
2399 }
2400 let success = missing_required_captures.is_empty();
2401 if !success {
2402 log::error!(
2403 "missing required capture(s) in {} {} TreeSitter query: {}",
2404 language_name,
2405 query_type,
2406 missing_required_captures.join(", ")
2407 );
2408 }
2409 success
2410}
2411
2412pub fn point_to_lsp(point: PointUtf16) -> lsp::Position {
2413 lsp::Position::new(point.row, point.column)
2414}
2415
2416pub fn point_from_lsp(point: lsp::Position) -> Unclipped<PointUtf16> {
2417 Unclipped(PointUtf16::new(point.line, point.character))
2418}
2419
2420pub fn range_to_lsp(range: Range<PointUtf16>) -> Result<lsp::Range> {
2421 anyhow::ensure!(
2422 range.start <= range.end,
2423 "Inverted range provided to an LSP request: {:?}-{:?}",
2424 range.start,
2425 range.end
2426 );
2427 Ok(lsp::Range {
2428 start: point_to_lsp(range.start),
2429 end: point_to_lsp(range.end),
2430 })
2431}
2432
2433pub fn range_from_lsp(range: lsp::Range) -> Range<Unclipped<PointUtf16>> {
2434 let mut start = point_from_lsp(range.start);
2435 let mut end = point_from_lsp(range.end);
2436 if start > end {
2437 log::warn!("range_from_lsp called with inverted range {start:?}-{end:?}");
2438 mem::swap(&mut start, &mut end);
2439 }
2440 start..end
2441}
2442
2443#[cfg(test)]
2444mod tests {
2445 use super::*;
2446 use gpui::TestAppContext;
2447 use pretty_assertions::assert_matches;
2448
2449 #[gpui::test(iterations = 10)]
2450 async fn test_language_loading(cx: &mut TestAppContext) {
2451 let languages = LanguageRegistry::test(cx.executor());
2452 let languages = Arc::new(languages);
2453 languages.register_native_grammars([
2454 ("json", tree_sitter_json::LANGUAGE),
2455 ("rust", tree_sitter_rust::LANGUAGE),
2456 ]);
2457 languages.register_test_language(LanguageConfig {
2458 name: "JSON".into(),
2459 grammar: Some("json".into()),
2460 matcher: LanguageMatcher {
2461 path_suffixes: vec!["json".into()],
2462 ..Default::default()
2463 },
2464 ..Default::default()
2465 });
2466 languages.register_test_language(LanguageConfig {
2467 name: "Rust".into(),
2468 grammar: Some("rust".into()),
2469 matcher: LanguageMatcher {
2470 path_suffixes: vec!["rs".into()],
2471 ..Default::default()
2472 },
2473 ..Default::default()
2474 });
2475 assert_eq!(
2476 languages.language_names(),
2477 &[
2478 LanguageName::new("JSON"),
2479 LanguageName::new("Plain Text"),
2480 LanguageName::new("Rust"),
2481 ]
2482 );
2483
2484 let rust1 = languages.language_for_name("Rust");
2485 let rust2 = languages.language_for_name("Rust");
2486
2487 // Ensure language is still listed even if it's being loaded.
2488 assert_eq!(
2489 languages.language_names(),
2490 &[
2491 LanguageName::new("JSON"),
2492 LanguageName::new("Plain Text"),
2493 LanguageName::new("Rust"),
2494 ]
2495 );
2496
2497 let (rust1, rust2) = futures::join!(rust1, rust2);
2498 assert!(Arc::ptr_eq(&rust1.unwrap(), &rust2.unwrap()));
2499
2500 // Ensure language is still listed even after loading it.
2501 assert_eq!(
2502 languages.language_names(),
2503 &[
2504 LanguageName::new("JSON"),
2505 LanguageName::new("Plain Text"),
2506 LanguageName::new("Rust"),
2507 ]
2508 );
2509
2510 // Loading an unknown language returns an error.
2511 assert!(languages.language_for_name("Unknown").await.is_err());
2512 }
2513
2514 #[gpui::test]
2515 async fn test_completion_label_omits_duplicate_data() {
2516 let regular_completion_item_1 = lsp::CompletionItem {
2517 label: "regular1".to_string(),
2518 detail: Some("detail1".to_string()),
2519 label_details: Some(lsp::CompletionItemLabelDetails {
2520 detail: None,
2521 description: Some("description 1".to_string()),
2522 }),
2523 ..lsp::CompletionItem::default()
2524 };
2525
2526 let regular_completion_item_2 = lsp::CompletionItem {
2527 label: "regular2".to_string(),
2528 label_details: Some(lsp::CompletionItemLabelDetails {
2529 detail: None,
2530 description: Some("description 2".to_string()),
2531 }),
2532 ..lsp::CompletionItem::default()
2533 };
2534
2535 let completion_item_with_duplicate_detail_and_proper_description = lsp::CompletionItem {
2536 detail: Some(regular_completion_item_1.label.clone()),
2537 ..regular_completion_item_1.clone()
2538 };
2539
2540 let completion_item_with_duplicate_detail = lsp::CompletionItem {
2541 detail: Some(regular_completion_item_1.label.clone()),
2542 label_details: None,
2543 ..regular_completion_item_1.clone()
2544 };
2545
2546 let completion_item_with_duplicate_description = lsp::CompletionItem {
2547 label_details: Some(lsp::CompletionItemLabelDetails {
2548 detail: None,
2549 description: Some(regular_completion_item_2.label.clone()),
2550 }),
2551 ..regular_completion_item_2.clone()
2552 };
2553
2554 assert_eq!(
2555 CodeLabel::fallback_for_completion(®ular_completion_item_1, None).text,
2556 format!(
2557 "{} {}",
2558 regular_completion_item_1.label,
2559 regular_completion_item_1.detail.unwrap()
2560 ),
2561 "LSP completion items with both detail and label_details.description should prefer detail"
2562 );
2563 assert_eq!(
2564 CodeLabel::fallback_for_completion(®ular_completion_item_2, None).text,
2565 format!(
2566 "{} {}",
2567 regular_completion_item_2.label,
2568 regular_completion_item_2
2569 .label_details
2570 .as_ref()
2571 .unwrap()
2572 .description
2573 .as_ref()
2574 .unwrap()
2575 ),
2576 "LSP completion items without detail but with label_details.description should use that"
2577 );
2578 assert_eq!(
2579 CodeLabel::fallback_for_completion(
2580 &completion_item_with_duplicate_detail_and_proper_description,
2581 None
2582 )
2583 .text,
2584 format!(
2585 "{} {}",
2586 regular_completion_item_1.label,
2587 regular_completion_item_1
2588 .label_details
2589 .as_ref()
2590 .unwrap()
2591 .description
2592 .as_ref()
2593 .unwrap()
2594 ),
2595 "LSP completion items with both detail and label_details.description should prefer description only if the detail duplicates the completion label"
2596 );
2597 assert_eq!(
2598 CodeLabel::fallback_for_completion(&completion_item_with_duplicate_detail, None).text,
2599 regular_completion_item_1.label,
2600 "LSP completion items with duplicate label and detail, should omit the detail"
2601 );
2602 assert_eq!(
2603 CodeLabel::fallback_for_completion(&completion_item_with_duplicate_description, None)
2604 .text,
2605 regular_completion_item_2.label,
2606 "LSP completion items with duplicate label and detail, should omit the detail"
2607 );
2608 }
2609
2610 #[test]
2611 fn test_deserializing_comments_backwards_compat() {
2612 // current version of `block_comment` and `documentation_comment` work
2613 {
2614 let config: LanguageConfig = ::toml::from_str(
2615 r#"
2616 name = "Foo"
2617 block_comment = { start = "a", end = "b", prefix = "c", tab_size = 1 }
2618 documentation_comment = { start = "d", end = "e", prefix = "f", tab_size = 2 }
2619 "#,
2620 )
2621 .unwrap();
2622 assert_matches!(config.block_comment, Some(BlockCommentConfig { .. }));
2623 assert_matches!(
2624 config.documentation_comment,
2625 Some(BlockCommentConfig { .. })
2626 );
2627
2628 let block_config = config.block_comment.unwrap();
2629 assert_eq!(block_config.start.as_ref(), "a");
2630 assert_eq!(block_config.end.as_ref(), "b");
2631 assert_eq!(block_config.prefix.as_ref(), "c");
2632 assert_eq!(block_config.tab_size, 1);
2633
2634 let doc_config = config.documentation_comment.unwrap();
2635 assert_eq!(doc_config.start.as_ref(), "d");
2636 assert_eq!(doc_config.end.as_ref(), "e");
2637 assert_eq!(doc_config.prefix.as_ref(), "f");
2638 assert_eq!(doc_config.tab_size, 2);
2639 }
2640
2641 // former `documentation` setting is read into `documentation_comment`
2642 {
2643 let config: LanguageConfig = ::toml::from_str(
2644 r#"
2645 name = "Foo"
2646 documentation = { start = "a", end = "b", prefix = "c", tab_size = 1}
2647 "#,
2648 )
2649 .unwrap();
2650 assert_matches!(
2651 config.documentation_comment,
2652 Some(BlockCommentConfig { .. })
2653 );
2654
2655 let config = config.documentation_comment.unwrap();
2656 assert_eq!(config.start.as_ref(), "a");
2657 assert_eq!(config.end.as_ref(), "b");
2658 assert_eq!(config.prefix.as_ref(), "c");
2659 assert_eq!(config.tab_size, 1);
2660 }
2661
2662 // old block_comment format is read into BlockCommentConfig
2663 {
2664 let config: LanguageConfig = ::toml::from_str(
2665 r#"
2666 name = "Foo"
2667 block_comment = ["a", "b"]
2668 "#,
2669 )
2670 .unwrap();
2671 assert_matches!(config.block_comment, Some(BlockCommentConfig { .. }));
2672
2673 let config = config.block_comment.unwrap();
2674 assert_eq!(config.start.as_ref(), "a");
2675 assert_eq!(config.end.as_ref(), "b");
2676 assert_eq!(config.prefix.as_ref(), "");
2677 assert_eq!(config.tab_size, 0);
2678 }
2679 }
2680}