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 outline;
15pub mod proto;
16mod syntax_map;
17mod task_context;
18mod toolchain;
19
20#[cfg(test)]
21pub mod buffer_tests;
22pub mod markdown;
23
24use crate::language_settings::SoftWrap;
25use anyhow::{anyhow, Context, Result};
26use async_trait::async_trait;
27use collections::{HashMap, HashSet};
28use futures::Future;
29use gpui::{AppContext, AsyncAppContext, Model, SharedString, Task};
30pub use highlight_map::HighlightMap;
31use http_client::HttpClient;
32pub use language_registry::{LanguageName, LoadedLanguage};
33use lsp::{
34 CodeActionKind, InitializeParams, LanguageServerBinary, LanguageServerBinaryOptions,
35 LanguageServerName,
36};
37use parking_lot::Mutex;
38use regex::Regex;
39use schemars::{
40 gen::SchemaGenerator,
41 schema::{InstanceType, Schema, SchemaObject},
42 JsonSchema,
43};
44use serde::{de, Deserialize, Deserializer, Serialize, Serializer};
45use serde_json::Value;
46use settings::WorktreeId;
47use smol::future::FutureExt as _;
48use std::{
49 any::Any,
50 ffi::OsStr,
51 fmt::Debug,
52 hash::Hash,
53 mem,
54 ops::{DerefMut, Range},
55 path::{Path, PathBuf},
56 pin::Pin,
57 str,
58 sync::{
59 atomic::{AtomicU64, AtomicUsize, Ordering::SeqCst},
60 Arc, LazyLock,
61 },
62};
63use std::{num::NonZeroU32, sync::OnceLock};
64use syntax_map::{QueryCursorHandle, SyntaxSnapshot};
65use task::RunnableTag;
66pub use task_context::{ContextProvider, RunnableRange};
67use theme::SyntaxTheme;
68pub use toolchain::{LanguageToolchainStore, Toolchain, ToolchainList, ToolchainLister};
69use tree_sitter::{self, wasmtime, Query, QueryCursor, WasmStore};
70use util::serde::default_true;
71
72pub use buffer::Operation;
73pub use buffer::*;
74pub use diagnostic_set::DiagnosticEntry;
75pub use language_registry::{
76 AvailableLanguage, LanguageNotFound, LanguageQueries, LanguageRegistry,
77 LanguageServerBinaryStatus, QUERY_FILENAME_PREFIXES,
78};
79pub use lsp::LanguageServerId;
80pub use outline::*;
81pub use syntax_map::{OwnedSyntaxLayer, SyntaxLayer, ToTreeSitterPoint, TreeSitterOptions};
82pub use text::{AnchorRangeExt, LineEnding};
83pub use tree_sitter::{Node, Parser, Tree, TreeCursor};
84
85/// Initializes the `language` crate.
86///
87/// This should be called before making use of items from the create.
88pub fn init(cx: &mut AppContext) {
89 language_settings::init(cx);
90}
91
92static QUERY_CURSORS: Mutex<Vec<QueryCursor>> = Mutex::new(vec![]);
93static PARSERS: Mutex<Vec<Parser>> = Mutex::new(vec![]);
94
95pub fn with_parser<F, R>(func: F) -> R
96where
97 F: FnOnce(&mut Parser) -> R,
98{
99 let mut parser = PARSERS.lock().pop().unwrap_or_else(|| {
100 let mut parser = Parser::new();
101 parser
102 .set_wasm_store(WasmStore::new(&WASM_ENGINE).unwrap())
103 .unwrap();
104 parser
105 });
106 parser.set_included_ranges(&[]).unwrap();
107 let result = func(&mut parser);
108 PARSERS.lock().push(parser);
109 result
110}
111
112pub fn with_query_cursor<F, R>(func: F) -> R
113where
114 F: FnOnce(&mut QueryCursor) -> R,
115{
116 let mut cursor = QueryCursorHandle::new();
117 func(cursor.deref_mut())
118}
119
120static NEXT_LANGUAGE_ID: LazyLock<AtomicUsize> = LazyLock::new(Default::default);
121static NEXT_GRAMMAR_ID: LazyLock<AtomicUsize> = LazyLock::new(Default::default);
122static WASM_ENGINE: LazyLock<wasmtime::Engine> = LazyLock::new(|| {
123 wasmtime::Engine::new(&wasmtime::Config::new()).expect("Failed to create Wasmtime engine")
124});
125
126/// A shared grammar for plain text, exposed for reuse by downstream crates.
127pub static PLAIN_TEXT: LazyLock<Arc<Language>> = LazyLock::new(|| {
128 Arc::new(Language::new(
129 LanguageConfig {
130 name: "Plain Text".into(),
131 soft_wrap: Some(SoftWrap::EditorWidth),
132 matcher: LanguageMatcher {
133 path_suffixes: vec!["txt".to_owned()],
134 first_line_pattern: None,
135 },
136 ..Default::default()
137 },
138 None,
139 ))
140});
141
142/// Types that represent a position in a buffer, and can be converted into
143/// an LSP position, to send to a language server.
144pub trait ToLspPosition {
145 /// Converts the value into an LSP position.
146 fn to_lsp_position(self) -> lsp::Position;
147}
148
149#[derive(Debug, Clone, PartialEq, Eq, Hash)]
150pub struct Location {
151 pub buffer: Model<Buffer>,
152 pub range: Range<Anchor>,
153}
154
155/// Represents a Language Server, with certain cached sync properties.
156/// Uses [`LspAdapter`] under the hood, but calls all 'static' methods
157/// once at startup, and caches the results.
158pub struct CachedLspAdapter {
159 pub name: LanguageServerName,
160 pub disk_based_diagnostic_sources: Vec<String>,
161 pub disk_based_diagnostics_progress_token: Option<String>,
162 language_ids: HashMap<String, String>,
163 pub adapter: Arc<dyn LspAdapter>,
164 pub reinstall_attempt_count: AtomicU64,
165 cached_binary: futures::lock::Mutex<Option<LanguageServerBinary>>,
166 attach_kind: OnceLock<Attach>,
167}
168
169impl Debug for CachedLspAdapter {
170 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
171 f.debug_struct("CachedLspAdapter")
172 .field("name", &self.name)
173 .field(
174 "disk_based_diagnostic_sources",
175 &self.disk_based_diagnostic_sources,
176 )
177 .field(
178 "disk_based_diagnostics_progress_token",
179 &self.disk_based_diagnostics_progress_token,
180 )
181 .field("language_ids", &self.language_ids)
182 .field("reinstall_attempt_count", &self.reinstall_attempt_count)
183 .finish_non_exhaustive()
184 }
185}
186
187impl CachedLspAdapter {
188 pub fn new(adapter: Arc<dyn LspAdapter>) -> Arc<Self> {
189 let name = adapter.name();
190 let disk_based_diagnostic_sources = adapter.disk_based_diagnostic_sources();
191 let disk_based_diagnostics_progress_token = adapter.disk_based_diagnostics_progress_token();
192 let language_ids = adapter.language_ids();
193
194 Arc::new(CachedLspAdapter {
195 name,
196 disk_based_diagnostic_sources,
197 disk_based_diagnostics_progress_token,
198 language_ids,
199 adapter,
200 cached_binary: Default::default(),
201 reinstall_attempt_count: AtomicU64::new(0),
202 attach_kind: Default::default(),
203 })
204 }
205
206 pub fn name(&self) -> LanguageServerName {
207 self.adapter.name().clone()
208 }
209
210 pub async fn get_language_server_command(
211 self: Arc<Self>,
212 delegate: Arc<dyn LspAdapterDelegate>,
213 toolchains: Arc<dyn LanguageToolchainStore>,
214 binary_options: LanguageServerBinaryOptions,
215 cx: &mut AsyncAppContext,
216 ) -> Result<LanguageServerBinary> {
217 let cached_binary = self.cached_binary.lock().await;
218 self.adapter
219 .clone()
220 .get_language_server_command(delegate, toolchains, binary_options, cached_binary, cx)
221 .await
222 }
223
224 pub fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
225 self.adapter.code_action_kinds()
226 }
227
228 pub fn process_diagnostics(&self, params: &mut lsp::PublishDiagnosticsParams) {
229 self.adapter.process_diagnostics(params)
230 }
231
232 pub async fn process_completions(&self, completion_items: &mut [lsp::CompletionItem]) {
233 self.adapter.process_completions(completion_items).await
234 }
235
236 pub async fn labels_for_completions(
237 &self,
238 completion_items: &[lsp::CompletionItem],
239 language: &Arc<Language>,
240 ) -> Result<Vec<Option<CodeLabel>>> {
241 self.adapter
242 .clone()
243 .labels_for_completions(completion_items, language)
244 .await
245 }
246
247 pub async fn labels_for_symbols(
248 &self,
249 symbols: &[(String, lsp::SymbolKind)],
250 language: &Arc<Language>,
251 ) -> Result<Vec<Option<CodeLabel>>> {
252 self.adapter
253 .clone()
254 .labels_for_symbols(symbols, language)
255 .await
256 }
257
258 pub fn language_id(&self, language_name: &LanguageName) -> String {
259 self.language_ids
260 .get(language_name.0.as_ref())
261 .cloned()
262 .unwrap_or_else(|| language_name.lsp_id())
263 }
264 pub fn find_project_root(
265 &self,
266 path: &Path,
267 ancestor_depth: usize,
268 delegate: &Arc<dyn LspAdapterDelegate>,
269 ) -> Option<Arc<Path>> {
270 self.adapter
271 .find_project_root(path, ancestor_depth, delegate)
272 }
273 pub fn attach_kind(&self) -> Attach {
274 *self.attach_kind.get_or_init(|| self.adapter.attach_kind())
275 }
276}
277
278#[derive(Clone, Copy, Debug, PartialEq)]
279pub enum Attach {
280 /// Create a single language server instance per subproject root.
281 InstancePerRoot,
282 /// Use one shared language server instance for all subprojects within a project.
283 Shared,
284}
285
286impl Attach {
287 pub fn root_path(
288 &self,
289 root_subproject_path: (WorktreeId, Arc<Path>),
290 ) -> (WorktreeId, Arc<Path>) {
291 match self {
292 Attach::InstancePerRoot => root_subproject_path,
293 Attach::Shared => (root_subproject_path.0, Arc::from(Path::new(""))),
294 }
295 }
296}
297
298/// [`LspAdapterDelegate`] allows [`LspAdapter]` implementations to interface with the application
299// e.g. to display a notification or fetch data from the web.
300#[async_trait]
301pub trait LspAdapterDelegate: Send + Sync {
302 fn show_notification(&self, message: &str, cx: &mut AppContext);
303 fn http_client(&self) -> Arc<dyn HttpClient>;
304 fn worktree_id(&self) -> WorktreeId;
305 fn worktree_root_path(&self) -> &Path;
306 fn update_status(&self, language: LanguageServerName, status: LanguageServerBinaryStatus);
307 async fn language_server_download_dir(&self, name: &LanguageServerName) -> Option<Arc<Path>>;
308
309 async fn npm_package_installed_version(
310 &self,
311 package_name: &str,
312 ) -> Result<Option<(PathBuf, String)>>;
313 async fn which(&self, command: &OsStr) -> Option<PathBuf>;
314 async fn shell_env(&self) -> HashMap<String, String>;
315 async fn read_text_file(&self, path: PathBuf) -> Result<String>;
316 async fn try_exec(&self, binary: LanguageServerBinary) -> Result<()>;
317}
318
319#[async_trait(?Send)]
320pub trait LspAdapter: 'static + Send + Sync {
321 fn name(&self) -> LanguageServerName;
322
323 fn get_language_server_command<'a>(
324 self: Arc<Self>,
325 delegate: Arc<dyn LspAdapterDelegate>,
326 toolchains: Arc<dyn LanguageToolchainStore>,
327 binary_options: LanguageServerBinaryOptions,
328 mut cached_binary: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
329 cx: &'a mut AsyncAppContext,
330 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
331 async move {
332 // First we check whether the adapter can give us a user-installed binary.
333 // If so, we do *not* want to cache that, because each worktree might give us a different
334 // binary:
335 //
336 // worktree 1: user-installed at `.bin/gopls`
337 // worktree 2: user-installed at `~/bin/gopls`
338 // worktree 3: no gopls found in PATH -> fallback to Zed installation
339 //
340 // We only want to cache when we fall back to the global one,
341 // because we don't want to download and overwrite our global one
342 // for each worktree we might have open.
343 if binary_options.allow_path_lookup {
344 if let Some(binary) = self.check_if_user_installed(delegate.as_ref(), toolchains, cx).await {
345 log::info!(
346 "found user-installed language server for {}. path: {:?}, arguments: {:?}",
347 self.name().0,
348 binary.path,
349 binary.arguments
350 );
351 return Ok(binary);
352 }
353 }
354
355 if !binary_options.allow_binary_download {
356 return Err(anyhow!("downloading language servers disabled"));
357 }
358
359 if let Some(cached_binary) = cached_binary.as_ref() {
360 return Ok(cached_binary.clone());
361 }
362
363 let Some(container_dir) = delegate.language_server_download_dir(&self.name()).await else {
364 anyhow::bail!("no language server download dir defined")
365 };
366
367 let mut binary = try_fetch_server_binary(self.as_ref(), &delegate, container_dir.to_path_buf(), cx).await;
368
369 if let Err(error) = binary.as_ref() {
370 if let Some(prev_downloaded_binary) = self
371 .cached_server_binary(container_dir.to_path_buf(), delegate.as_ref())
372 .await
373 {
374 log::info!(
375 "failed to fetch newest version of language server {:?}. error: {:?}, falling back to using {:?}",
376 self.name(),
377 error,
378 prev_downloaded_binary.path
379 );
380 binary = Ok(prev_downloaded_binary);
381 } else {
382 delegate.update_status(
383 self.name(),
384 LanguageServerBinaryStatus::Failed {
385 error: format!("{error:?}"),
386 },
387 );
388 }
389 }
390
391 if let Ok(binary) = &binary {
392 *cached_binary = Some(binary.clone());
393 }
394
395 binary
396 }
397 .boxed_local()
398 }
399
400 async fn check_if_user_installed(
401 &self,
402 _: &dyn LspAdapterDelegate,
403 _: Arc<dyn LanguageToolchainStore>,
404 _: &AsyncAppContext,
405 ) -> Option<LanguageServerBinary> {
406 None
407 }
408
409 async fn fetch_latest_server_version(
410 &self,
411 delegate: &dyn LspAdapterDelegate,
412 ) -> Result<Box<dyn 'static + Send + Any>>;
413
414 fn will_fetch_server(
415 &self,
416 _: &Arc<dyn LspAdapterDelegate>,
417 _: &mut AsyncAppContext,
418 ) -> Option<Task<Result<()>>> {
419 None
420 }
421
422 async fn check_if_version_installed(
423 &self,
424 _version: &(dyn 'static + Send + Any),
425 _container_dir: &PathBuf,
426 _delegate: &dyn LspAdapterDelegate,
427 ) -> Option<LanguageServerBinary> {
428 None
429 }
430
431 async fn fetch_server_binary(
432 &self,
433 latest_version: Box<dyn 'static + Send + Any>,
434 container_dir: PathBuf,
435 delegate: &dyn LspAdapterDelegate,
436 ) -> Result<LanguageServerBinary>;
437
438 async fn cached_server_binary(
439 &self,
440 container_dir: PathBuf,
441 delegate: &dyn LspAdapterDelegate,
442 ) -> Option<LanguageServerBinary>;
443
444 fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
445
446 /// Post-processes completions provided by the language server.
447 async fn process_completions(&self, _: &mut [lsp::CompletionItem]) {}
448
449 async fn labels_for_completions(
450 self: Arc<Self>,
451 completions: &[lsp::CompletionItem],
452 language: &Arc<Language>,
453 ) -> Result<Vec<Option<CodeLabel>>> {
454 let mut labels = Vec::new();
455 for (ix, completion) in completions.iter().enumerate() {
456 let label = self.label_for_completion(completion, language).await;
457 if let Some(label) = label {
458 labels.resize(ix + 1, None);
459 *labels.last_mut().unwrap() = Some(label);
460 }
461 }
462 Ok(labels)
463 }
464
465 async fn label_for_completion(
466 &self,
467 _: &lsp::CompletionItem,
468 _: &Arc<Language>,
469 ) -> Option<CodeLabel> {
470 None
471 }
472
473 async fn labels_for_symbols(
474 self: Arc<Self>,
475 symbols: &[(String, lsp::SymbolKind)],
476 language: &Arc<Language>,
477 ) -> Result<Vec<Option<CodeLabel>>> {
478 let mut labels = Vec::new();
479 for (ix, (name, kind)) in symbols.iter().enumerate() {
480 let label = self.label_for_symbol(name, *kind, language).await;
481 if let Some(label) = label {
482 labels.resize(ix + 1, None);
483 *labels.last_mut().unwrap() = Some(label);
484 }
485 }
486 Ok(labels)
487 }
488
489 async fn label_for_symbol(
490 &self,
491 _: &str,
492 _: lsp::SymbolKind,
493 _: &Arc<Language>,
494 ) -> Option<CodeLabel> {
495 None
496 }
497
498 /// Returns initialization options that are going to be sent to a LSP server as a part of [`lsp::InitializeParams`]
499 async fn initialization_options(
500 self: Arc<Self>,
501 _: &Arc<dyn LspAdapterDelegate>,
502 ) -> Result<Option<Value>> {
503 Ok(None)
504 }
505
506 async fn workspace_configuration(
507 self: Arc<Self>,
508 _: &Arc<dyn LspAdapterDelegate>,
509 _: Arc<dyn LanguageToolchainStore>,
510 _cx: &mut AsyncAppContext,
511 ) -> Result<Value> {
512 Ok(serde_json::json!({}))
513 }
514
515 /// Returns a list of code actions supported by a given LspAdapter
516 fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
517 Some(vec![
518 CodeActionKind::EMPTY,
519 CodeActionKind::QUICKFIX,
520 CodeActionKind::REFACTOR,
521 CodeActionKind::REFACTOR_EXTRACT,
522 CodeActionKind::SOURCE,
523 ])
524 }
525
526 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
527 Default::default()
528 }
529
530 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
531 None
532 }
533
534 fn language_ids(&self) -> HashMap<String, String> {
535 Default::default()
536 }
537
538 /// Support custom initialize params.
539 fn prepare_initialize_params(&self, original: InitializeParams) -> Result<InitializeParams> {
540 Ok(original)
541 }
542 fn attach_kind(&self) -> Attach {
543 Attach::Shared
544 }
545 fn find_project_root(
546 &self,
547
548 _path: &Path,
549 _ancestor_depth: usize,
550 _: &Arc<dyn LspAdapterDelegate>,
551 ) -> Option<Arc<Path>> {
552 // By default all language servers are rooted at the root of the worktree.
553 Some(Arc::from("".as_ref()))
554 }
555}
556
557async fn try_fetch_server_binary<L: LspAdapter + 'static + Send + Sync + ?Sized>(
558 adapter: &L,
559 delegate: &Arc<dyn LspAdapterDelegate>,
560 container_dir: PathBuf,
561 cx: &mut AsyncAppContext,
562) -> Result<LanguageServerBinary> {
563 if let Some(task) = adapter.will_fetch_server(delegate, cx) {
564 task.await?;
565 }
566
567 let name = adapter.name();
568 log::info!("fetching latest version of language server {:?}", name.0);
569 delegate.update_status(name.clone(), LanguageServerBinaryStatus::CheckingForUpdate);
570
571 let latest_version = adapter
572 .fetch_latest_server_version(delegate.as_ref())
573 .await?;
574
575 if let Some(binary) = adapter
576 .check_if_version_installed(latest_version.as_ref(), &container_dir, delegate.as_ref())
577 .await
578 {
579 log::info!("language server {:?} is already installed", name.0);
580 delegate.update_status(name.clone(), LanguageServerBinaryStatus::None);
581 Ok(binary)
582 } else {
583 log::info!("downloading language server {:?}", name.0);
584 delegate.update_status(adapter.name(), LanguageServerBinaryStatus::Downloading);
585 let binary = adapter
586 .fetch_server_binary(latest_version, container_dir, delegate.as_ref())
587 .await;
588
589 delegate.update_status(name.clone(), LanguageServerBinaryStatus::None);
590 binary
591 }
592}
593
594#[derive(Clone, Debug, Default, PartialEq, Eq)]
595pub struct CodeLabel {
596 /// The text to display.
597 pub text: String,
598 /// Syntax highlighting runs.
599 pub runs: Vec<(Range<usize>, HighlightId)>,
600 /// The portion of the text that should be used in fuzzy filtering.
601 pub filter_range: Range<usize>,
602}
603
604#[derive(Clone, Deserialize, JsonSchema)]
605pub struct LanguageConfig {
606 /// Human-readable name of the language.
607 pub name: LanguageName,
608 /// The name of this language for a Markdown code fence block
609 pub code_fence_block_name: Option<Arc<str>>,
610 // The name of the grammar in a WASM bundle (experimental).
611 pub grammar: Option<Arc<str>>,
612 /// The criteria for matching this language to a given file.
613 #[serde(flatten)]
614 pub matcher: LanguageMatcher,
615 /// List of bracket types in a language.
616 #[serde(default)]
617 #[schemars(schema_with = "bracket_pair_config_json_schema")]
618 pub brackets: BracketPairConfig,
619 /// If set to true, auto indentation uses last non empty line to determine
620 /// the indentation level for a new line.
621 #[serde(default = "auto_indent_using_last_non_empty_line_default")]
622 pub auto_indent_using_last_non_empty_line: bool,
623 // Whether indentation of pasted content should be adjusted based on the context.
624 #[serde(default)]
625 pub auto_indent_on_paste: Option<bool>,
626 /// A regex that is used to determine whether the indentation level should be
627 /// increased in the following line.
628 #[serde(default, deserialize_with = "deserialize_regex")]
629 #[schemars(schema_with = "regex_json_schema")]
630 pub increase_indent_pattern: Option<Regex>,
631 /// A regex that is used to determine whether the indentation level should be
632 /// decreased in the following line.
633 #[serde(default, deserialize_with = "deserialize_regex")]
634 #[schemars(schema_with = "regex_json_schema")]
635 pub decrease_indent_pattern: Option<Regex>,
636 /// A list of characters that trigger the automatic insertion of a closing
637 /// bracket when they immediately precede the point where an opening
638 /// bracket is inserted.
639 #[serde(default)]
640 pub autoclose_before: String,
641 /// A placeholder used internally by Semantic Index.
642 #[serde(default)]
643 pub collapsed_placeholder: String,
644 /// A line comment string that is inserted in e.g. `toggle comments` action.
645 /// A language can have multiple flavours of line comments. All of the provided line comments are
646 /// used for comment continuations on the next line, but only the first one is used for Editor::ToggleComments.
647 #[serde(default)]
648 pub line_comments: Vec<Arc<str>>,
649 /// Starting and closing characters of a block comment.
650 #[serde(default)]
651 pub block_comment: Option<(Arc<str>, Arc<str>)>,
652 /// A list of language servers that are allowed to run on subranges of a given language.
653 #[serde(default)]
654 pub scope_opt_in_language_servers: Vec<LanguageServerName>,
655 #[serde(default)]
656 pub overrides: HashMap<String, LanguageConfigOverride>,
657 /// A list of characters that Zed should treat as word characters for the
658 /// purpose of features that operate on word boundaries, like 'move to next word end'
659 /// or a whole-word search in buffer search.
660 #[serde(default)]
661 pub word_characters: HashSet<char>,
662 /// Whether to indent lines using tab characters, as opposed to multiple
663 /// spaces.
664 #[serde(default)]
665 pub hard_tabs: Option<bool>,
666 /// How many columns a tab should occupy.
667 #[serde(default)]
668 pub tab_size: Option<NonZeroU32>,
669 /// How to soft-wrap long lines of text.
670 #[serde(default)]
671 pub soft_wrap: Option<SoftWrap>,
672 /// The name of a Prettier parser that will be used for this language when no file path is available.
673 /// If there's a parser name in the language settings, that will be used instead.
674 #[serde(default)]
675 pub prettier_parser_name: Option<String>,
676 /// If true, this language is only for syntax highlighting via an injection into other
677 /// languages, but should not appear to the user as a distinct language.
678 #[serde(default)]
679 pub hidden: bool,
680}
681
682#[derive(Clone, Debug, Serialize, Deserialize, Default, JsonSchema)]
683pub struct LanguageMatcher {
684 /// 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`.
685 #[serde(default)]
686 pub path_suffixes: Vec<String>,
687 /// A regex pattern that determines whether the language should be assigned to a file or not.
688 #[serde(
689 default,
690 serialize_with = "serialize_regex",
691 deserialize_with = "deserialize_regex"
692 )]
693 #[schemars(schema_with = "regex_json_schema")]
694 pub first_line_pattern: Option<Regex>,
695}
696
697/// Represents a language for the given range. Some languages (e.g. HTML)
698/// interleave several languages together, thus a single buffer might actually contain
699/// several nested scopes.
700#[derive(Clone, Debug)]
701pub struct LanguageScope {
702 language: Arc<Language>,
703 override_id: Option<u32>,
704}
705
706#[derive(Clone, Deserialize, Default, Debug, JsonSchema)]
707pub struct LanguageConfigOverride {
708 #[serde(default)]
709 pub line_comments: Override<Vec<Arc<str>>>,
710 #[serde(default)]
711 pub block_comment: Override<(Arc<str>, Arc<str>)>,
712 #[serde(skip)]
713 pub disabled_bracket_ixs: Vec<u16>,
714 #[serde(default)]
715 pub word_characters: Override<HashSet<char>>,
716 #[serde(default)]
717 pub opt_into_language_servers: Vec<LanguageServerName>,
718}
719
720#[derive(Clone, Deserialize, Debug, Serialize, JsonSchema)]
721#[serde(untagged)]
722pub enum Override<T> {
723 Remove { remove: bool },
724 Set(T),
725}
726
727impl<T> Default for Override<T> {
728 fn default() -> Self {
729 Override::Remove { remove: false }
730 }
731}
732
733impl<T> Override<T> {
734 fn as_option<'a>(this: Option<&'a Self>, original: Option<&'a T>) -> Option<&'a T> {
735 match this {
736 Some(Self::Set(value)) => Some(value),
737 Some(Self::Remove { remove: true }) => None,
738 Some(Self::Remove { remove: false }) | None => original,
739 }
740 }
741}
742
743impl Default for LanguageConfig {
744 fn default() -> Self {
745 Self {
746 name: LanguageName::new(""),
747 code_fence_block_name: None,
748 grammar: None,
749 matcher: LanguageMatcher::default(),
750 brackets: Default::default(),
751 auto_indent_using_last_non_empty_line: auto_indent_using_last_non_empty_line_default(),
752 auto_indent_on_paste: None,
753 increase_indent_pattern: Default::default(),
754 decrease_indent_pattern: Default::default(),
755 autoclose_before: Default::default(),
756 line_comments: Default::default(),
757 block_comment: Default::default(),
758 scope_opt_in_language_servers: Default::default(),
759 overrides: Default::default(),
760 word_characters: Default::default(),
761 collapsed_placeholder: Default::default(),
762 hard_tabs: None,
763 tab_size: None,
764 soft_wrap: None,
765 prettier_parser_name: None,
766 hidden: false,
767 }
768 }
769}
770
771fn auto_indent_using_last_non_empty_line_default() -> bool {
772 true
773}
774
775fn deserialize_regex<'de, D: Deserializer<'de>>(d: D) -> Result<Option<Regex>, D::Error> {
776 let source = Option::<String>::deserialize(d)?;
777 if let Some(source) = source {
778 Ok(Some(regex::Regex::new(&source).map_err(de::Error::custom)?))
779 } else {
780 Ok(None)
781 }
782}
783
784fn regex_json_schema(_: &mut SchemaGenerator) -> Schema {
785 Schema::Object(SchemaObject {
786 instance_type: Some(InstanceType::String.into()),
787 ..Default::default()
788 })
789}
790
791fn serialize_regex<S>(regex: &Option<Regex>, serializer: S) -> Result<S::Ok, S::Error>
792where
793 S: Serializer,
794{
795 match regex {
796 Some(regex) => serializer.serialize_str(regex.as_str()),
797 None => serializer.serialize_none(),
798 }
799}
800
801#[doc(hidden)]
802#[cfg(any(test, feature = "test-support"))]
803pub struct FakeLspAdapter {
804 pub name: &'static str,
805 pub initialization_options: Option<Value>,
806 pub prettier_plugins: Vec<&'static str>,
807 pub disk_based_diagnostics_progress_token: Option<String>,
808 pub disk_based_diagnostics_sources: Vec<String>,
809 pub language_server_binary: LanguageServerBinary,
810
811 pub capabilities: lsp::ServerCapabilities,
812 pub initializer: Option<Box<dyn 'static + Send + Sync + Fn(&mut lsp::FakeLanguageServer)>>,
813 pub label_for_completion: Option<
814 Box<
815 dyn 'static
816 + Send
817 + Sync
818 + Fn(&lsp::CompletionItem, &Arc<Language>) -> Option<CodeLabel>,
819 >,
820 >,
821}
822
823/// Configuration of handling bracket pairs for a given language.
824///
825/// This struct includes settings for defining which pairs of characters are considered brackets and
826/// also specifies any language-specific scopes where these pairs should be ignored for bracket matching purposes.
827#[derive(Clone, Debug, Default, JsonSchema)]
828pub struct BracketPairConfig {
829 /// A list of character pairs that should be treated as brackets in the context of a given language.
830 pub pairs: Vec<BracketPair>,
831 /// A list of tree-sitter scopes for which a given bracket should not be active.
832 /// N-th entry in `[Self::disabled_scopes_by_bracket_ix]` contains a list of disabled scopes for an n-th entry in `[Self::pairs]`
833 #[serde(skip)]
834 pub disabled_scopes_by_bracket_ix: Vec<Vec<String>>,
835}
836
837fn bracket_pair_config_json_schema(gen: &mut SchemaGenerator) -> Schema {
838 Option::<Vec<BracketPairContent>>::json_schema(gen)
839}
840
841#[derive(Deserialize, JsonSchema)]
842pub struct BracketPairContent {
843 #[serde(flatten)]
844 pub bracket_pair: BracketPair,
845 #[serde(default)]
846 pub not_in: Vec<String>,
847}
848
849impl<'de> Deserialize<'de> for BracketPairConfig {
850 fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
851 where
852 D: Deserializer<'de>,
853 {
854 let result = Vec::<BracketPairContent>::deserialize(deserializer)?;
855 let mut brackets = Vec::with_capacity(result.len());
856 let mut disabled_scopes_by_bracket_ix = Vec::with_capacity(result.len());
857 for entry in result {
858 brackets.push(entry.bracket_pair);
859 disabled_scopes_by_bracket_ix.push(entry.not_in);
860 }
861
862 Ok(BracketPairConfig {
863 pairs: brackets,
864 disabled_scopes_by_bracket_ix,
865 })
866 }
867}
868
869/// Describes a single bracket pair and how an editor should react to e.g. inserting
870/// an opening bracket or to a newline character insertion in between `start` and `end` characters.
871#[derive(Clone, Debug, Default, Deserialize, PartialEq, JsonSchema)]
872pub struct BracketPair {
873 /// Starting substring for a bracket.
874 pub start: String,
875 /// Ending substring for a bracket.
876 pub end: String,
877 /// True if `end` should be automatically inserted right after `start` characters.
878 pub close: bool,
879 /// True if selected text should be surrounded by `start` and `end` characters.
880 #[serde(default = "default_true")]
881 pub surround: bool,
882 /// True if an extra newline should be inserted while the cursor is in the middle
883 /// of that bracket pair.
884 pub newline: bool,
885}
886
887#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
888pub(crate) struct LanguageId(usize);
889
890impl LanguageId {
891 pub(crate) fn new() -> Self {
892 Self(NEXT_LANGUAGE_ID.fetch_add(1, SeqCst))
893 }
894}
895
896pub struct Language {
897 pub(crate) id: LanguageId,
898 pub(crate) config: LanguageConfig,
899 pub(crate) grammar: Option<Arc<Grammar>>,
900 pub(crate) context_provider: Option<Arc<dyn ContextProvider>>,
901 pub(crate) toolchain: Option<Arc<dyn ToolchainLister>>,
902}
903
904#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
905pub struct GrammarId(pub usize);
906
907impl GrammarId {
908 pub(crate) fn new() -> Self {
909 Self(NEXT_GRAMMAR_ID.fetch_add(1, SeqCst))
910 }
911}
912
913pub struct Grammar {
914 id: GrammarId,
915 pub ts_language: tree_sitter::Language,
916 pub(crate) error_query: Query,
917 pub(crate) highlights_query: Option<Query>,
918 pub(crate) brackets_config: Option<BracketConfig>,
919 pub(crate) redactions_config: Option<RedactionConfig>,
920 pub(crate) runnable_config: Option<RunnableConfig>,
921 pub(crate) indents_config: Option<IndentConfig>,
922 pub outline_config: Option<OutlineConfig>,
923 pub text_object_config: Option<TextObjectConfig>,
924 pub embedding_config: Option<EmbeddingConfig>,
925 pub(crate) injection_config: Option<InjectionConfig>,
926 pub(crate) override_config: Option<OverrideConfig>,
927 pub(crate) highlight_map: Mutex<HighlightMap>,
928}
929
930struct IndentConfig {
931 query: Query,
932 indent_capture_ix: u32,
933 start_capture_ix: Option<u32>,
934 end_capture_ix: Option<u32>,
935 outdent_capture_ix: Option<u32>,
936}
937
938pub struct OutlineConfig {
939 pub query: Query,
940 pub item_capture_ix: u32,
941 pub name_capture_ix: u32,
942 pub context_capture_ix: Option<u32>,
943 pub extra_context_capture_ix: Option<u32>,
944 pub open_capture_ix: Option<u32>,
945 pub close_capture_ix: Option<u32>,
946 pub annotation_capture_ix: Option<u32>,
947}
948
949#[derive(Debug, Clone, Copy, PartialEq)]
950pub enum TextObject {
951 InsideFunction,
952 AroundFunction,
953 InsideClass,
954 AroundClass,
955 InsideComment,
956 AroundComment,
957}
958
959impl TextObject {
960 pub fn from_capture_name(name: &str) -> Option<TextObject> {
961 match name {
962 "function.inside" => Some(TextObject::InsideFunction),
963 "function.around" => Some(TextObject::AroundFunction),
964 "class.inside" => Some(TextObject::InsideClass),
965 "class.around" => Some(TextObject::AroundClass),
966 "comment.inside" => Some(TextObject::InsideComment),
967 "comment.around" => Some(TextObject::AroundComment),
968 _ => None,
969 }
970 }
971
972 pub fn around(&self) -> Option<Self> {
973 match self {
974 TextObject::InsideFunction => Some(TextObject::AroundFunction),
975 TextObject::InsideClass => Some(TextObject::AroundClass),
976 TextObject::InsideComment => Some(TextObject::AroundComment),
977 _ => None,
978 }
979 }
980}
981
982pub struct TextObjectConfig {
983 pub query: Query,
984 pub text_objects_by_capture_ix: Vec<(u32, TextObject)>,
985}
986
987#[derive(Debug)]
988pub struct EmbeddingConfig {
989 pub query: Query,
990 pub item_capture_ix: u32,
991 pub name_capture_ix: Option<u32>,
992 pub context_capture_ix: Option<u32>,
993 pub collapse_capture_ix: Option<u32>,
994 pub keep_capture_ix: Option<u32>,
995}
996
997struct InjectionConfig {
998 query: Query,
999 content_capture_ix: u32,
1000 language_capture_ix: Option<u32>,
1001 patterns: Vec<InjectionPatternConfig>,
1002}
1003
1004struct RedactionConfig {
1005 pub query: Query,
1006 pub redaction_capture_ix: u32,
1007}
1008
1009#[derive(Clone, Debug, PartialEq)]
1010enum RunnableCapture {
1011 Named(SharedString),
1012 Run,
1013}
1014
1015struct RunnableConfig {
1016 pub query: Query,
1017 /// A mapping from capture indice to capture kind
1018 pub extra_captures: Vec<RunnableCapture>,
1019}
1020
1021struct OverrideConfig {
1022 query: Query,
1023 values: HashMap<u32, OverrideEntry>,
1024}
1025
1026#[derive(Debug)]
1027struct OverrideEntry {
1028 name: String,
1029 range_is_inclusive: bool,
1030 value: LanguageConfigOverride,
1031}
1032
1033#[derive(Default, Clone)]
1034struct InjectionPatternConfig {
1035 language: Option<Box<str>>,
1036 combined: bool,
1037}
1038
1039struct BracketConfig {
1040 query: Query,
1041 open_capture_ix: u32,
1042 close_capture_ix: u32,
1043}
1044
1045impl Language {
1046 pub fn new(config: LanguageConfig, ts_language: Option<tree_sitter::Language>) -> Self {
1047 Self::new_with_id(LanguageId::new(), config, ts_language)
1048 }
1049
1050 fn new_with_id(
1051 id: LanguageId,
1052 config: LanguageConfig,
1053 ts_language: Option<tree_sitter::Language>,
1054 ) -> Self {
1055 Self {
1056 id,
1057 config,
1058 grammar: ts_language.map(|ts_language| {
1059 Arc::new(Grammar {
1060 id: GrammarId::new(),
1061 highlights_query: None,
1062 brackets_config: None,
1063 outline_config: None,
1064 text_object_config: None,
1065 embedding_config: None,
1066 indents_config: None,
1067 injection_config: None,
1068 override_config: None,
1069 redactions_config: None,
1070 runnable_config: None,
1071 error_query: Query::new(&ts_language, "(ERROR) @error").unwrap(),
1072 ts_language,
1073 highlight_map: Default::default(),
1074 })
1075 }),
1076 context_provider: None,
1077 toolchain: None,
1078 }
1079 }
1080
1081 pub fn with_context_provider(mut self, provider: Option<Arc<dyn ContextProvider>>) -> Self {
1082 self.context_provider = provider;
1083 self
1084 }
1085
1086 pub fn with_toolchain_lister(mut self, provider: Option<Arc<dyn ToolchainLister>>) -> Self {
1087 self.toolchain = provider;
1088 self
1089 }
1090
1091 pub fn with_queries(mut self, queries: LanguageQueries) -> Result<Self> {
1092 if let Some(query) = queries.highlights {
1093 self = self
1094 .with_highlights_query(query.as_ref())
1095 .context("Error loading highlights query")?;
1096 }
1097 if let Some(query) = queries.brackets {
1098 self = self
1099 .with_brackets_query(query.as_ref())
1100 .context("Error loading brackets query")?;
1101 }
1102 if let Some(query) = queries.indents {
1103 self = self
1104 .with_indents_query(query.as_ref())
1105 .context("Error loading indents query")?;
1106 }
1107 if let Some(query) = queries.outline {
1108 self = self
1109 .with_outline_query(query.as_ref())
1110 .context("Error loading outline query")?;
1111 }
1112 if let Some(query) = queries.embedding {
1113 self = self
1114 .with_embedding_query(query.as_ref())
1115 .context("Error loading embedding query")?;
1116 }
1117 if let Some(query) = queries.injections {
1118 self = self
1119 .with_injection_query(query.as_ref())
1120 .context("Error loading injection query")?;
1121 }
1122 if let Some(query) = queries.overrides {
1123 self = self
1124 .with_override_query(query.as_ref())
1125 .context("Error loading override query")?;
1126 }
1127 if let Some(query) = queries.redactions {
1128 self = self
1129 .with_redaction_query(query.as_ref())
1130 .context("Error loading redaction query")?;
1131 }
1132 if let Some(query) = queries.runnables {
1133 self = self
1134 .with_runnable_query(query.as_ref())
1135 .context("Error loading runnables query")?;
1136 }
1137 if let Some(query) = queries.text_objects {
1138 self = self
1139 .with_text_object_query(query.as_ref())
1140 .context("Error loading textobject query")?;
1141 }
1142 Ok(self)
1143 }
1144
1145 pub fn with_highlights_query(mut self, source: &str) -> Result<Self> {
1146 let grammar = self
1147 .grammar_mut()
1148 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1149 grammar.highlights_query = Some(Query::new(&grammar.ts_language, source)?);
1150 Ok(self)
1151 }
1152
1153 pub fn with_runnable_query(mut self, source: &str) -> Result<Self> {
1154 let grammar = self
1155 .grammar_mut()
1156 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1157
1158 let query = Query::new(&grammar.ts_language, source)?;
1159 let mut extra_captures = Vec::with_capacity(query.capture_names().len());
1160
1161 for name in query.capture_names().iter() {
1162 let kind = if *name == "run" {
1163 RunnableCapture::Run
1164 } else {
1165 RunnableCapture::Named(name.to_string().into())
1166 };
1167 extra_captures.push(kind);
1168 }
1169
1170 grammar.runnable_config = Some(RunnableConfig {
1171 extra_captures,
1172 query,
1173 });
1174
1175 Ok(self)
1176 }
1177
1178 pub fn with_outline_query(mut self, source: &str) -> Result<Self> {
1179 let grammar = self
1180 .grammar_mut()
1181 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1182 let query = Query::new(&grammar.ts_language, source)?;
1183 let mut item_capture_ix = None;
1184 let mut name_capture_ix = None;
1185 let mut context_capture_ix = None;
1186 let mut extra_context_capture_ix = None;
1187 let mut open_capture_ix = None;
1188 let mut close_capture_ix = None;
1189 let mut annotation_capture_ix = None;
1190 get_capture_indices(
1191 &query,
1192 &mut [
1193 ("item", &mut item_capture_ix),
1194 ("name", &mut name_capture_ix),
1195 ("context", &mut context_capture_ix),
1196 ("context.extra", &mut extra_context_capture_ix),
1197 ("open", &mut open_capture_ix),
1198 ("close", &mut close_capture_ix),
1199 ("annotation", &mut annotation_capture_ix),
1200 ],
1201 );
1202 if let Some((item_capture_ix, name_capture_ix)) = item_capture_ix.zip(name_capture_ix) {
1203 grammar.outline_config = Some(OutlineConfig {
1204 query,
1205 item_capture_ix,
1206 name_capture_ix,
1207 context_capture_ix,
1208 extra_context_capture_ix,
1209 open_capture_ix,
1210 close_capture_ix,
1211 annotation_capture_ix,
1212 });
1213 }
1214 Ok(self)
1215 }
1216
1217 pub fn with_text_object_query(mut self, source: &str) -> Result<Self> {
1218 let grammar = self
1219 .grammar_mut()
1220 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1221 let query = Query::new(&grammar.ts_language, source)?;
1222
1223 let mut text_objects_by_capture_ix = Vec::new();
1224 for (ix, name) in query.capture_names().iter().enumerate() {
1225 if let Some(text_object) = TextObject::from_capture_name(name) {
1226 text_objects_by_capture_ix.push((ix as u32, text_object));
1227 }
1228 }
1229
1230 grammar.text_object_config = Some(TextObjectConfig {
1231 query,
1232 text_objects_by_capture_ix,
1233 });
1234 Ok(self)
1235 }
1236
1237 pub fn with_embedding_query(mut self, source: &str) -> Result<Self> {
1238 let grammar = self
1239 .grammar_mut()
1240 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1241 let query = Query::new(&grammar.ts_language, source)?;
1242 let mut item_capture_ix = None;
1243 let mut name_capture_ix = None;
1244 let mut context_capture_ix = None;
1245 let mut collapse_capture_ix = None;
1246 let mut keep_capture_ix = None;
1247 get_capture_indices(
1248 &query,
1249 &mut [
1250 ("item", &mut item_capture_ix),
1251 ("name", &mut name_capture_ix),
1252 ("context", &mut context_capture_ix),
1253 ("keep", &mut keep_capture_ix),
1254 ("collapse", &mut collapse_capture_ix),
1255 ],
1256 );
1257 if let Some(item_capture_ix) = item_capture_ix {
1258 grammar.embedding_config = Some(EmbeddingConfig {
1259 query,
1260 item_capture_ix,
1261 name_capture_ix,
1262 context_capture_ix,
1263 collapse_capture_ix,
1264 keep_capture_ix,
1265 });
1266 }
1267 Ok(self)
1268 }
1269
1270 pub fn with_brackets_query(mut self, source: &str) -> Result<Self> {
1271 let grammar = self
1272 .grammar_mut()
1273 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1274 let query = Query::new(&grammar.ts_language, source)?;
1275 let mut open_capture_ix = None;
1276 let mut close_capture_ix = None;
1277 get_capture_indices(
1278 &query,
1279 &mut [
1280 ("open", &mut open_capture_ix),
1281 ("close", &mut close_capture_ix),
1282 ],
1283 );
1284 if let Some((open_capture_ix, close_capture_ix)) = open_capture_ix.zip(close_capture_ix) {
1285 grammar.brackets_config = Some(BracketConfig {
1286 query,
1287 open_capture_ix,
1288 close_capture_ix,
1289 });
1290 }
1291 Ok(self)
1292 }
1293
1294 pub fn with_indents_query(mut self, source: &str) -> Result<Self> {
1295 let grammar = self
1296 .grammar_mut()
1297 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1298 let query = Query::new(&grammar.ts_language, source)?;
1299 let mut indent_capture_ix = None;
1300 let mut start_capture_ix = None;
1301 let mut end_capture_ix = None;
1302 let mut outdent_capture_ix = None;
1303 get_capture_indices(
1304 &query,
1305 &mut [
1306 ("indent", &mut indent_capture_ix),
1307 ("start", &mut start_capture_ix),
1308 ("end", &mut end_capture_ix),
1309 ("outdent", &mut outdent_capture_ix),
1310 ],
1311 );
1312 if let Some(indent_capture_ix) = indent_capture_ix {
1313 grammar.indents_config = Some(IndentConfig {
1314 query,
1315 indent_capture_ix,
1316 start_capture_ix,
1317 end_capture_ix,
1318 outdent_capture_ix,
1319 });
1320 }
1321 Ok(self)
1322 }
1323
1324 pub fn with_injection_query(mut self, source: &str) -> Result<Self> {
1325 let grammar = self
1326 .grammar_mut()
1327 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1328 let query = Query::new(&grammar.ts_language, source)?;
1329 let mut language_capture_ix = None;
1330 let mut injection_language_capture_ix = None;
1331 let mut content_capture_ix = None;
1332 let mut injection_content_capture_ix = None;
1333 get_capture_indices(
1334 &query,
1335 &mut [
1336 ("language", &mut language_capture_ix),
1337 ("injection.language", &mut injection_language_capture_ix),
1338 ("content", &mut content_capture_ix),
1339 ("injection.content", &mut injection_content_capture_ix),
1340 ],
1341 );
1342 language_capture_ix = match (language_capture_ix, injection_language_capture_ix) {
1343 (None, Some(ix)) => Some(ix),
1344 (Some(_), Some(_)) => {
1345 return Err(anyhow!(
1346 "both language and injection.language captures are present"
1347 ));
1348 }
1349 _ => language_capture_ix,
1350 };
1351 content_capture_ix = match (content_capture_ix, injection_content_capture_ix) {
1352 (None, Some(ix)) => Some(ix),
1353 (Some(_), Some(_)) => {
1354 return Err(anyhow!(
1355 "both content and injection.content captures are present"
1356 ));
1357 }
1358 _ => content_capture_ix,
1359 };
1360 let patterns = (0..query.pattern_count())
1361 .map(|ix| {
1362 let mut config = InjectionPatternConfig::default();
1363 for setting in query.property_settings(ix) {
1364 match setting.key.as_ref() {
1365 "language" | "injection.language" => {
1366 config.language.clone_from(&setting.value);
1367 }
1368 "combined" | "injection.combined" => {
1369 config.combined = true;
1370 }
1371 _ => {}
1372 }
1373 }
1374 config
1375 })
1376 .collect();
1377 if let Some(content_capture_ix) = content_capture_ix {
1378 grammar.injection_config = Some(InjectionConfig {
1379 query,
1380 language_capture_ix,
1381 content_capture_ix,
1382 patterns,
1383 });
1384 }
1385 Ok(self)
1386 }
1387
1388 pub fn with_override_query(mut self, source: &str) -> anyhow::Result<Self> {
1389 let query = {
1390 let grammar = self
1391 .grammar
1392 .as_ref()
1393 .ok_or_else(|| anyhow!("no grammar for language"))?;
1394 Query::new(&grammar.ts_language, source)?
1395 };
1396
1397 let mut override_configs_by_id = HashMap::default();
1398 for (ix, mut name) in query.capture_names().iter().copied().enumerate() {
1399 let mut range_is_inclusive = false;
1400 if name.starts_with('_') {
1401 continue;
1402 }
1403 if let Some(prefix) = name.strip_suffix(".inclusive") {
1404 name = prefix;
1405 range_is_inclusive = true;
1406 }
1407
1408 let value = self.config.overrides.get(name).cloned().unwrap_or_default();
1409 for server_name in &value.opt_into_language_servers {
1410 if !self
1411 .config
1412 .scope_opt_in_language_servers
1413 .contains(server_name)
1414 {
1415 util::debug_panic!("Server {server_name:?} has been opted-in by scope {name:?} but has not been marked as an opt-in server");
1416 }
1417 }
1418
1419 override_configs_by_id.insert(
1420 ix as u32,
1421 OverrideEntry {
1422 name: name.to_string(),
1423 range_is_inclusive,
1424 value,
1425 },
1426 );
1427 }
1428
1429 let referenced_override_names = self.config.overrides.keys().chain(
1430 self.config
1431 .brackets
1432 .disabled_scopes_by_bracket_ix
1433 .iter()
1434 .flatten(),
1435 );
1436
1437 for referenced_name in referenced_override_names {
1438 if !override_configs_by_id
1439 .values()
1440 .any(|entry| entry.name == *referenced_name)
1441 {
1442 Err(anyhow!(
1443 "language {:?} has overrides in config not in query: {referenced_name:?}",
1444 self.config.name
1445 ))?;
1446 }
1447 }
1448
1449 for entry in override_configs_by_id.values_mut() {
1450 entry.value.disabled_bracket_ixs = self
1451 .config
1452 .brackets
1453 .disabled_scopes_by_bracket_ix
1454 .iter()
1455 .enumerate()
1456 .filter_map(|(ix, disabled_scope_names)| {
1457 if disabled_scope_names.contains(&entry.name) {
1458 Some(ix as u16)
1459 } else {
1460 None
1461 }
1462 })
1463 .collect();
1464 }
1465
1466 self.config.brackets.disabled_scopes_by_bracket_ix.clear();
1467
1468 let grammar = self
1469 .grammar_mut()
1470 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1471 grammar.override_config = Some(OverrideConfig {
1472 query,
1473 values: override_configs_by_id,
1474 });
1475 Ok(self)
1476 }
1477
1478 pub fn with_redaction_query(mut self, source: &str) -> anyhow::Result<Self> {
1479 let grammar = self
1480 .grammar_mut()
1481 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1482
1483 let query = Query::new(&grammar.ts_language, source)?;
1484 let mut redaction_capture_ix = None;
1485 get_capture_indices(&query, &mut [("redact", &mut redaction_capture_ix)]);
1486
1487 if let Some(redaction_capture_ix) = redaction_capture_ix {
1488 grammar.redactions_config = Some(RedactionConfig {
1489 query,
1490 redaction_capture_ix,
1491 });
1492 }
1493
1494 Ok(self)
1495 }
1496
1497 fn grammar_mut(&mut self) -> Option<&mut Grammar> {
1498 Arc::get_mut(self.grammar.as_mut()?)
1499 }
1500
1501 pub fn name(&self) -> LanguageName {
1502 self.config.name.clone()
1503 }
1504
1505 pub fn code_fence_block_name(&self) -> Arc<str> {
1506 self.config
1507 .code_fence_block_name
1508 .clone()
1509 .unwrap_or_else(|| self.config.name.0.to_lowercase().into())
1510 }
1511
1512 pub fn context_provider(&self) -> Option<Arc<dyn ContextProvider>> {
1513 self.context_provider.clone()
1514 }
1515
1516 pub fn toolchain_lister(&self) -> Option<Arc<dyn ToolchainLister>> {
1517 self.toolchain.clone()
1518 }
1519
1520 pub fn highlight_text<'a>(
1521 self: &'a Arc<Self>,
1522 text: &'a Rope,
1523 range: Range<usize>,
1524 ) -> Vec<(Range<usize>, HighlightId)> {
1525 let mut result = Vec::new();
1526 if let Some(grammar) = &self.grammar {
1527 let tree = grammar.parse_text(text, None);
1528 let captures =
1529 SyntaxSnapshot::single_tree_captures(range.clone(), text, &tree, self, |grammar| {
1530 grammar.highlights_query.as_ref()
1531 });
1532 let highlight_maps = vec![grammar.highlight_map()];
1533 let mut offset = 0;
1534 for chunk in
1535 BufferChunks::new(text, range, Some((captures, highlight_maps)), false, None)
1536 {
1537 let end_offset = offset + chunk.text.len();
1538 if let Some(highlight_id) = chunk.syntax_highlight_id {
1539 if !highlight_id.is_default() {
1540 result.push((offset..end_offset, highlight_id));
1541 }
1542 }
1543 offset = end_offset;
1544 }
1545 }
1546 result
1547 }
1548
1549 pub fn path_suffixes(&self) -> &[String] {
1550 &self.config.matcher.path_suffixes
1551 }
1552
1553 pub fn should_autoclose_before(&self, c: char) -> bool {
1554 c.is_whitespace() || self.config.autoclose_before.contains(c)
1555 }
1556
1557 pub fn set_theme(&self, theme: &SyntaxTheme) {
1558 if let Some(grammar) = self.grammar.as_ref() {
1559 if let Some(highlights_query) = &grammar.highlights_query {
1560 *grammar.highlight_map.lock() =
1561 HighlightMap::new(highlights_query.capture_names(), theme);
1562 }
1563 }
1564 }
1565
1566 pub fn grammar(&self) -> Option<&Arc<Grammar>> {
1567 self.grammar.as_ref()
1568 }
1569
1570 pub fn default_scope(self: &Arc<Self>) -> LanguageScope {
1571 LanguageScope {
1572 language: self.clone(),
1573 override_id: None,
1574 }
1575 }
1576
1577 pub fn lsp_id(&self) -> String {
1578 self.config.name.lsp_id()
1579 }
1580
1581 pub fn prettier_parser_name(&self) -> Option<&str> {
1582 self.config.prettier_parser_name.as_deref()
1583 }
1584
1585 pub fn config(&self) -> &LanguageConfig {
1586 &self.config
1587 }
1588}
1589
1590impl LanguageScope {
1591 pub fn path_suffixes(&self) -> &[String] {
1592 &self.language.path_suffixes()
1593 }
1594
1595 pub fn language_name(&self) -> LanguageName {
1596 self.language.config.name.clone()
1597 }
1598
1599 pub fn collapsed_placeholder(&self) -> &str {
1600 self.language.config.collapsed_placeholder.as_ref()
1601 }
1602
1603 /// Returns line prefix that is inserted in e.g. line continuations or
1604 /// in `toggle comments` action.
1605 pub fn line_comment_prefixes(&self) -> &[Arc<str>] {
1606 Override::as_option(
1607 self.config_override().map(|o| &o.line_comments),
1608 Some(&self.language.config.line_comments),
1609 )
1610 .map_or(&[] as &[_], |e| e.as_slice())
1611 }
1612
1613 pub fn block_comment_delimiters(&self) -> Option<(&Arc<str>, &Arc<str>)> {
1614 Override::as_option(
1615 self.config_override().map(|o| &o.block_comment),
1616 self.language.config.block_comment.as_ref(),
1617 )
1618 .map(|e| (&e.0, &e.1))
1619 }
1620
1621 /// Returns a list of language-specific word characters.
1622 ///
1623 /// By default, Zed treats alphanumeric characters (and '_') as word characters for
1624 /// the purpose of actions like 'move to next word end` or whole-word search.
1625 /// It additionally accounts for language's additional word characters.
1626 pub fn word_characters(&self) -> Option<&HashSet<char>> {
1627 Override::as_option(
1628 self.config_override().map(|o| &o.word_characters),
1629 Some(&self.language.config.word_characters),
1630 )
1631 }
1632
1633 /// Returns a list of bracket pairs for a given language with an additional
1634 /// piece of information about whether the particular bracket pair is currently active for a given language.
1635 pub fn brackets(&self) -> impl Iterator<Item = (&BracketPair, bool)> {
1636 let mut disabled_ids = self
1637 .config_override()
1638 .map_or(&[] as _, |o| o.disabled_bracket_ixs.as_slice());
1639 self.language
1640 .config
1641 .brackets
1642 .pairs
1643 .iter()
1644 .enumerate()
1645 .map(move |(ix, bracket)| {
1646 let mut is_enabled = true;
1647 if let Some(next_disabled_ix) = disabled_ids.first() {
1648 if ix == *next_disabled_ix as usize {
1649 disabled_ids = &disabled_ids[1..];
1650 is_enabled = false;
1651 }
1652 }
1653 (bracket, is_enabled)
1654 })
1655 }
1656
1657 pub fn should_autoclose_before(&self, c: char) -> bool {
1658 c.is_whitespace() || self.language.config.autoclose_before.contains(c)
1659 }
1660
1661 pub fn language_allowed(&self, name: &LanguageServerName) -> bool {
1662 let config = &self.language.config;
1663 let opt_in_servers = &config.scope_opt_in_language_servers;
1664 if opt_in_servers.iter().any(|o| *o == *name) {
1665 if let Some(over) = self.config_override() {
1666 over.opt_into_language_servers.iter().any(|o| *o == *name)
1667 } else {
1668 false
1669 }
1670 } else {
1671 true
1672 }
1673 }
1674
1675 pub fn override_name(&self) -> Option<&str> {
1676 let id = self.override_id?;
1677 let grammar = self.language.grammar.as_ref()?;
1678 let override_config = grammar.override_config.as_ref()?;
1679 override_config.values.get(&id).map(|e| e.name.as_str())
1680 }
1681
1682 fn config_override(&self) -> Option<&LanguageConfigOverride> {
1683 let id = self.override_id?;
1684 let grammar = self.language.grammar.as_ref()?;
1685 let override_config = grammar.override_config.as_ref()?;
1686 override_config.values.get(&id).map(|e| &e.value)
1687 }
1688}
1689
1690impl Hash for Language {
1691 fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
1692 self.id.hash(state)
1693 }
1694}
1695
1696impl PartialEq for Language {
1697 fn eq(&self, other: &Self) -> bool {
1698 self.id.eq(&other.id)
1699 }
1700}
1701
1702impl Eq for Language {}
1703
1704impl Debug for Language {
1705 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
1706 f.debug_struct("Language")
1707 .field("name", &self.config.name)
1708 .finish()
1709 }
1710}
1711
1712impl Grammar {
1713 pub fn id(&self) -> GrammarId {
1714 self.id
1715 }
1716
1717 fn parse_text(&self, text: &Rope, old_tree: Option<Tree>) -> Tree {
1718 with_parser(|parser| {
1719 parser
1720 .set_language(&self.ts_language)
1721 .expect("incompatible grammar");
1722 let mut chunks = text.chunks_in_range(0..text.len());
1723 parser
1724 .parse_with(
1725 &mut move |offset, _| {
1726 chunks.seek(offset);
1727 chunks.next().unwrap_or("").as_bytes()
1728 },
1729 old_tree.as_ref(),
1730 )
1731 .unwrap()
1732 })
1733 }
1734
1735 pub fn highlight_map(&self) -> HighlightMap {
1736 self.highlight_map.lock().clone()
1737 }
1738
1739 pub fn highlight_id_for_name(&self, name: &str) -> Option<HighlightId> {
1740 let capture_id = self
1741 .highlights_query
1742 .as_ref()?
1743 .capture_index_for_name(name)?;
1744 Some(self.highlight_map.lock().get(capture_id))
1745 }
1746}
1747
1748impl CodeLabel {
1749 pub fn plain(text: String, filter_text: Option<&str>) -> Self {
1750 let mut result = Self {
1751 runs: Vec::new(),
1752 filter_range: 0..text.len(),
1753 text,
1754 };
1755 if let Some(filter_text) = filter_text {
1756 if let Some(ix) = result.text.find(filter_text) {
1757 result.filter_range = ix..ix + filter_text.len();
1758 }
1759 }
1760 result
1761 }
1762
1763 pub fn push_str(&mut self, text: &str, highlight: Option<HighlightId>) {
1764 let start_ix = self.text.len();
1765 self.text.push_str(text);
1766 let end_ix = self.text.len();
1767 if let Some(highlight) = highlight {
1768 self.runs.push((start_ix..end_ix, highlight));
1769 }
1770 }
1771
1772 pub fn text(&self) -> &str {
1773 self.text.as_str()
1774 }
1775
1776 pub fn filter_text(&self) -> &str {
1777 &self.text[self.filter_range.clone()]
1778 }
1779}
1780
1781impl From<String> for CodeLabel {
1782 fn from(value: String) -> Self {
1783 Self::plain(value, None)
1784 }
1785}
1786
1787impl From<&str> for CodeLabel {
1788 fn from(value: &str) -> Self {
1789 Self::plain(value.to_string(), None)
1790 }
1791}
1792
1793impl Ord for LanguageMatcher {
1794 fn cmp(&self, other: &Self) -> std::cmp::Ordering {
1795 self.path_suffixes.cmp(&other.path_suffixes).then_with(|| {
1796 self.first_line_pattern
1797 .as_ref()
1798 .map(Regex::as_str)
1799 .cmp(&other.first_line_pattern.as_ref().map(Regex::as_str))
1800 })
1801 }
1802}
1803
1804impl PartialOrd for LanguageMatcher {
1805 fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1806 Some(self.cmp(other))
1807 }
1808}
1809
1810impl Eq for LanguageMatcher {}
1811
1812impl PartialEq for LanguageMatcher {
1813 fn eq(&self, other: &Self) -> bool {
1814 self.path_suffixes == other.path_suffixes
1815 && self.first_line_pattern.as_ref().map(Regex::as_str)
1816 == other.first_line_pattern.as_ref().map(Regex::as_str)
1817 }
1818}
1819
1820#[cfg(any(test, feature = "test-support"))]
1821impl Default for FakeLspAdapter {
1822 fn default() -> Self {
1823 Self {
1824 name: "the-fake-language-server",
1825 capabilities: lsp::LanguageServer::full_capabilities(),
1826 initializer: None,
1827 disk_based_diagnostics_progress_token: None,
1828 initialization_options: None,
1829 disk_based_diagnostics_sources: Vec::new(),
1830 prettier_plugins: Vec::new(),
1831 language_server_binary: LanguageServerBinary {
1832 path: "/the/fake/lsp/path".into(),
1833 arguments: vec![],
1834 env: Default::default(),
1835 },
1836 label_for_completion: None,
1837 }
1838 }
1839}
1840
1841#[cfg(any(test, feature = "test-support"))]
1842#[async_trait(?Send)]
1843impl LspAdapter for FakeLspAdapter {
1844 fn name(&self) -> LanguageServerName {
1845 LanguageServerName(self.name.into())
1846 }
1847
1848 async fn check_if_user_installed(
1849 &self,
1850 _: &dyn LspAdapterDelegate,
1851 _: Arc<dyn LanguageToolchainStore>,
1852 _: &AsyncAppContext,
1853 ) -> Option<LanguageServerBinary> {
1854 Some(self.language_server_binary.clone())
1855 }
1856
1857 fn get_language_server_command<'a>(
1858 self: Arc<Self>,
1859 _: Arc<dyn LspAdapterDelegate>,
1860 _: Arc<dyn LanguageToolchainStore>,
1861 _: LanguageServerBinaryOptions,
1862 _: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
1863 _: &'a mut AsyncAppContext,
1864 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
1865 async move { Ok(self.language_server_binary.clone()) }.boxed_local()
1866 }
1867
1868 async fn fetch_latest_server_version(
1869 &self,
1870 _: &dyn LspAdapterDelegate,
1871 ) -> Result<Box<dyn 'static + Send + Any>> {
1872 unreachable!();
1873 }
1874
1875 async fn fetch_server_binary(
1876 &self,
1877 _: Box<dyn 'static + Send + Any>,
1878 _: PathBuf,
1879 _: &dyn LspAdapterDelegate,
1880 ) -> Result<LanguageServerBinary> {
1881 unreachable!();
1882 }
1883
1884 async fn cached_server_binary(
1885 &self,
1886 _: PathBuf,
1887 _: &dyn LspAdapterDelegate,
1888 ) -> Option<LanguageServerBinary> {
1889 unreachable!();
1890 }
1891
1892 fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
1893
1894 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
1895 self.disk_based_diagnostics_sources.clone()
1896 }
1897
1898 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
1899 self.disk_based_diagnostics_progress_token.clone()
1900 }
1901
1902 async fn initialization_options(
1903 self: Arc<Self>,
1904 _: &Arc<dyn LspAdapterDelegate>,
1905 ) -> Result<Option<Value>> {
1906 Ok(self.initialization_options.clone())
1907 }
1908
1909 async fn label_for_completion(
1910 &self,
1911 item: &lsp::CompletionItem,
1912 language: &Arc<Language>,
1913 ) -> Option<CodeLabel> {
1914 let label_for_completion = self.label_for_completion.as_ref()?;
1915 label_for_completion(item, language)
1916 }
1917}
1918
1919fn get_capture_indices(query: &Query, captures: &mut [(&str, &mut Option<u32>)]) {
1920 for (ix, name) in query.capture_names().iter().enumerate() {
1921 for (capture_name, index) in captures.iter_mut() {
1922 if capture_name == name {
1923 **index = Some(ix as u32);
1924 break;
1925 }
1926 }
1927 }
1928}
1929
1930pub fn point_to_lsp(point: PointUtf16) -> lsp::Position {
1931 lsp::Position::new(point.row, point.column)
1932}
1933
1934pub fn point_from_lsp(point: lsp::Position) -> Unclipped<PointUtf16> {
1935 Unclipped(PointUtf16::new(point.line, point.character))
1936}
1937
1938pub fn range_to_lsp(range: Range<PointUtf16>) -> Result<lsp::Range> {
1939 if range.start > range.end {
1940 Err(anyhow!(
1941 "Inverted range provided to an LSP request: {:?}-{:?}",
1942 range.start,
1943 range.end
1944 ))
1945 } else {
1946 Ok(lsp::Range {
1947 start: point_to_lsp(range.start),
1948 end: point_to_lsp(range.end),
1949 })
1950 }
1951}
1952
1953pub fn range_from_lsp(range: lsp::Range) -> Range<Unclipped<PointUtf16>> {
1954 let mut start = point_from_lsp(range.start);
1955 let mut end = point_from_lsp(range.end);
1956 if start > end {
1957 log::warn!("range_from_lsp called with inverted range {start:?}-{end:?}");
1958 mem::swap(&mut start, &mut end);
1959 }
1960 start..end
1961}
1962
1963#[cfg(test)]
1964mod tests {
1965 use super::*;
1966 use gpui::TestAppContext;
1967
1968 #[gpui::test(iterations = 10)]
1969 async fn test_language_loading(cx: &mut TestAppContext) {
1970 let languages = LanguageRegistry::test(cx.executor());
1971 let languages = Arc::new(languages);
1972 languages.register_native_grammars([
1973 ("json", tree_sitter_json::LANGUAGE),
1974 ("rust", tree_sitter_rust::LANGUAGE),
1975 ]);
1976 languages.register_test_language(LanguageConfig {
1977 name: "JSON".into(),
1978 grammar: Some("json".into()),
1979 matcher: LanguageMatcher {
1980 path_suffixes: vec!["json".into()],
1981 ..Default::default()
1982 },
1983 ..Default::default()
1984 });
1985 languages.register_test_language(LanguageConfig {
1986 name: "Rust".into(),
1987 grammar: Some("rust".into()),
1988 matcher: LanguageMatcher {
1989 path_suffixes: vec!["rs".into()],
1990 ..Default::default()
1991 },
1992 ..Default::default()
1993 });
1994 assert_eq!(
1995 languages.language_names(),
1996 &[
1997 "JSON".to_string(),
1998 "Plain Text".to_string(),
1999 "Rust".to_string(),
2000 ]
2001 );
2002
2003 let rust1 = languages.language_for_name("Rust");
2004 let rust2 = languages.language_for_name("Rust");
2005
2006 // Ensure language is still listed even if it's being loaded.
2007 assert_eq!(
2008 languages.language_names(),
2009 &[
2010 "JSON".to_string(),
2011 "Plain Text".to_string(),
2012 "Rust".to_string(),
2013 ]
2014 );
2015
2016 let (rust1, rust2) = futures::join!(rust1, rust2);
2017 assert!(Arc::ptr_eq(&rust1.unwrap(), &rust2.unwrap()));
2018
2019 // Ensure language is still listed even after loading it.
2020 assert_eq!(
2021 languages.language_names(),
2022 &[
2023 "JSON".to_string(),
2024 "Plain Text".to_string(),
2025 "Rust".to_string(),
2026 ]
2027 );
2028
2029 // Loading an unknown language returns an error.
2030 assert!(languages.language_for_name("Unknown").await.is_err());
2031 }
2032}