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 as _, Result};
26use async_trait::async_trait;
27use collections::{HashMap, HashSet};
28use fs::Fs;
29use futures::Future;
30use gpui::{App, AsyncApp, Entity, SharedString, Task};
31pub use highlight_map::HighlightMap;
32use http_client::HttpClient;
33pub use language_registry::{LanguageName, LoadedLanguage};
34use lsp::{
35 CodeActionKind, InitializeParams, LanguageServerBinary, LanguageServerBinaryOptions,
36 LanguageServerName,
37};
38use parking_lot::Mutex;
39use regex::Regex;
40use schemars::{
41 gen::SchemaGenerator,
42 schema::{InstanceType, Schema, SchemaObject},
43 JsonSchema,
44};
45use serde::{de, Deserialize, Deserializer, Serialize, Serializer};
46use serde_json::Value;
47use settings::WorktreeId;
48use smol::future::FutureExt as _;
49use std::num::NonZeroU32;
50use std::{
51 any::Any,
52 ffi::OsStr,
53 fmt::Debug,
54 hash::Hash,
55 mem,
56 ops::{DerefMut, Range},
57 path::{Path, PathBuf},
58 pin::Pin,
59 str,
60 sync::{
61 atomic::{AtomicU64, AtomicUsize, Ordering::SeqCst},
62 Arc, LazyLock,
63 },
64};
65use syntax_map::{QueryCursorHandle, SyntaxSnapshot};
66use task::RunnableTag;
67pub use task_context::{ContextProvider, RunnableRange};
68use theme::SyntaxTheme;
69pub use toolchain::{LanguageToolchainStore, Toolchain, ToolchainList, ToolchainLister};
70use tree_sitter::{self, wasmtime, Query, QueryCursor, WasmStore};
71use util::serde::default_true;
72
73pub use buffer::Operation;
74pub use buffer::*;
75pub use diagnostic_set::DiagnosticEntry;
76pub use language_registry::{
77 AvailableLanguage, LanguageNotFound, LanguageQueries, LanguageRegistry,
78 LanguageServerBinaryStatus, QUERY_FILENAME_PREFIXES,
79};
80pub use lsp::LanguageServerId;
81pub use outline::*;
82pub use syntax_map::{OwnedSyntaxLayer, SyntaxLayer, ToTreeSitterPoint, TreeSitterOptions};
83pub use text::{AnchorRangeExt, LineEnding};
84pub use tree_sitter::{Node, Parser, Tree, TreeCursor};
85
86/// Initializes the `language` crate.
87///
88/// This should be called before making use of items from the create.
89pub fn init(cx: &mut App) {
90 language_settings::init(cx);
91}
92
93static QUERY_CURSORS: Mutex<Vec<QueryCursor>> = Mutex::new(vec![]);
94static PARSERS: Mutex<Vec<Parser>> = Mutex::new(vec![]);
95
96pub fn with_parser<F, R>(func: F) -> R
97where
98 F: FnOnce(&mut Parser) -> R,
99{
100 let mut parser = PARSERS.lock().pop().unwrap_or_else(|| {
101 let mut parser = Parser::new();
102 parser
103 .set_wasm_store(WasmStore::new(&WASM_ENGINE).unwrap())
104 .unwrap();
105 parser
106 });
107 parser.set_included_ranges(&[]).unwrap();
108 let result = func(&mut parser);
109 PARSERS.lock().push(parser);
110 result
111}
112
113pub fn with_query_cursor<F, R>(func: F) -> R
114where
115 F: FnOnce(&mut QueryCursor) -> R,
116{
117 let mut cursor = QueryCursorHandle::new();
118 func(cursor.deref_mut())
119}
120
121static NEXT_LANGUAGE_ID: LazyLock<AtomicUsize> = LazyLock::new(Default::default);
122static NEXT_GRAMMAR_ID: LazyLock<AtomicUsize> = LazyLock::new(Default::default);
123static WASM_ENGINE: LazyLock<wasmtime::Engine> = LazyLock::new(|| {
124 wasmtime::Engine::new(&wasmtime::Config::new()).expect("Failed to create Wasmtime engine")
125});
126
127/// A shared grammar for plain text, exposed for reuse by downstream crates.
128pub static PLAIN_TEXT: LazyLock<Arc<Language>> = LazyLock::new(|| {
129 Arc::new(Language::new(
130 LanguageConfig {
131 name: "Plain Text".into(),
132 soft_wrap: Some(SoftWrap::EditorWidth),
133 matcher: LanguageMatcher {
134 path_suffixes: vec!["txt".to_owned()],
135 first_line_pattern: None,
136 },
137 ..Default::default()
138 },
139 None,
140 ))
141});
142
143/// Types that represent a position in a buffer, and can be converted into
144/// an LSP position, to send to a language server.
145pub trait ToLspPosition {
146 /// Converts the value into an LSP position.
147 fn to_lsp_position(self) -> lsp::Position;
148}
149
150#[derive(Debug, Clone, PartialEq, Eq, Hash)]
151pub struct Location {
152 pub buffer: Entity<Buffer>,
153 pub range: Range<Anchor>,
154}
155
156/// Represents a Language Server, with certain cached sync properties.
157/// Uses [`LspAdapter`] under the hood, but calls all 'static' methods
158/// once at startup, and caches the results.
159pub struct CachedLspAdapter {
160 pub name: LanguageServerName,
161 pub disk_based_diagnostic_sources: Vec<String>,
162 pub disk_based_diagnostics_progress_token: Option<String>,
163 language_ids: HashMap<String, String>,
164 pub adapter: Arc<dyn LspAdapter>,
165 pub reinstall_attempt_count: AtomicU64,
166 cached_binary: futures::lock::Mutex<Option<LanguageServerBinary>>,
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 })
203 }
204
205 pub fn name(&self) -> LanguageServerName {
206 self.adapter.name().clone()
207 }
208
209 pub async fn get_language_server_command(
210 self: Arc<Self>,
211 delegate: Arc<dyn LspAdapterDelegate>,
212 toolchains: Arc<dyn LanguageToolchainStore>,
213 binary_options: LanguageServerBinaryOptions,
214 cx: &mut AsyncApp,
215 ) -> Result<LanguageServerBinary> {
216 let cached_binary = self.cached_binary.lock().await;
217 self.adapter
218 .clone()
219 .get_language_server_command(delegate, toolchains, binary_options, cached_binary, cx)
220 .await
221 }
222
223 pub fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
224 self.adapter.code_action_kinds()
225 }
226
227 pub fn process_diagnostics(&self, params: &mut lsp::PublishDiagnosticsParams) {
228 self.adapter.process_diagnostics(params)
229 }
230
231 pub async fn process_completions(&self, completion_items: &mut [lsp::CompletionItem]) {
232 self.adapter.process_completions(completion_items).await
233 }
234
235 pub async fn labels_for_completions(
236 &self,
237 completion_items: &[lsp::CompletionItem],
238 language: &Arc<Language>,
239 ) -> Result<Vec<Option<CodeLabel>>> {
240 self.adapter
241 .clone()
242 .labels_for_completions(completion_items, language)
243 .await
244 }
245
246 pub async fn labels_for_symbols(
247 &self,
248 symbols: &[(String, lsp::SymbolKind)],
249 language: &Arc<Language>,
250 ) -> Result<Vec<Option<CodeLabel>>> {
251 self.adapter
252 .clone()
253 .labels_for_symbols(symbols, language)
254 .await
255 }
256
257 pub fn language_id(&self, language_name: &LanguageName) -> String {
258 self.language_ids
259 .get(language_name.as_ref())
260 .cloned()
261 .unwrap_or_else(|| language_name.lsp_id())
262 }
263}
264
265/// [`LspAdapterDelegate`] allows [`LspAdapter]` implementations to interface with the application
266// e.g. to display a notification or fetch data from the web.
267#[async_trait]
268pub trait LspAdapterDelegate: Send + Sync {
269 fn show_notification(&self, message: &str, cx: &mut App);
270 fn http_client(&self) -> Arc<dyn HttpClient>;
271 fn worktree_id(&self) -> WorktreeId;
272 fn worktree_root_path(&self) -> &Path;
273 fn update_status(&self, language: LanguageServerName, status: LanguageServerBinaryStatus);
274 async fn language_server_download_dir(&self, name: &LanguageServerName) -> Option<Arc<Path>>;
275
276 async fn npm_package_installed_version(
277 &self,
278 package_name: &str,
279 ) -> Result<Option<(PathBuf, String)>>;
280 async fn which(&self, command: &OsStr) -> Option<PathBuf>;
281 async fn shell_env(&self) -> HashMap<String, String>;
282 async fn read_text_file(&self, path: PathBuf) -> Result<String>;
283 async fn try_exec(&self, binary: LanguageServerBinary) -> Result<()>;
284}
285
286#[async_trait(?Send)]
287pub trait LspAdapter: 'static + Send + Sync {
288 fn name(&self) -> LanguageServerName;
289
290 fn get_language_server_command<'a>(
291 self: Arc<Self>,
292 delegate: Arc<dyn LspAdapterDelegate>,
293 toolchains: Arc<dyn LanguageToolchainStore>,
294 binary_options: LanguageServerBinaryOptions,
295 mut cached_binary: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
296 cx: &'a mut AsyncApp,
297 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
298 async move {
299 // First we check whether the adapter can give us a user-installed binary.
300 // If so, we do *not* want to cache that, because each worktree might give us a different
301 // binary:
302 //
303 // worktree 1: user-installed at `.bin/gopls`
304 // worktree 2: user-installed at `~/bin/gopls`
305 // worktree 3: no gopls found in PATH -> fallback to Zed installation
306 //
307 // We only want to cache when we fall back to the global one,
308 // because we don't want to download and overwrite our global one
309 // for each worktree we might have open.
310 if binary_options.allow_path_lookup {
311 if let Some(binary) = self.check_if_user_installed(delegate.as_ref(), toolchains, cx).await {
312 log::info!(
313 "found user-installed language server for {}. path: {:?}, arguments: {:?}",
314 self.name().0,
315 binary.path,
316 binary.arguments
317 );
318 return Ok(binary);
319 }
320 }
321
322 if !binary_options.allow_binary_download {
323 return Err(anyhow!("downloading language servers disabled"));
324 }
325
326 if let Some(cached_binary) = cached_binary.as_ref() {
327 return Ok(cached_binary.clone());
328 }
329
330 let Some(container_dir) = delegate.language_server_download_dir(&self.name()).await else {
331 anyhow::bail!("no language server download dir defined")
332 };
333
334 let mut binary = try_fetch_server_binary(self.as_ref(), &delegate, container_dir.to_path_buf(), cx).await;
335
336 if let Err(error) = binary.as_ref() {
337 if let Some(prev_downloaded_binary) = self
338 .cached_server_binary(container_dir.to_path_buf(), delegate.as_ref())
339 .await
340 {
341 log::info!(
342 "failed to fetch newest version of language server {:?}. error: {:?}, falling back to using {:?}",
343 self.name(),
344 error,
345 prev_downloaded_binary.path
346 );
347 binary = Ok(prev_downloaded_binary);
348 } else {
349 delegate.update_status(
350 self.name(),
351 LanguageServerBinaryStatus::Failed {
352 error: format!("{error:?}"),
353 },
354 );
355 }
356 }
357
358 if let Ok(binary) = &binary {
359 *cached_binary = Some(binary.clone());
360 }
361
362 binary
363 }
364 .boxed_local()
365 }
366
367 async fn check_if_user_installed(
368 &self,
369 _: &dyn LspAdapterDelegate,
370 _: Arc<dyn LanguageToolchainStore>,
371 _: &AsyncApp,
372 ) -> Option<LanguageServerBinary> {
373 None
374 }
375
376 async fn fetch_latest_server_version(
377 &self,
378 delegate: &dyn LspAdapterDelegate,
379 ) -> Result<Box<dyn 'static + Send + Any>>;
380
381 fn will_fetch_server(
382 &self,
383 _: &Arc<dyn LspAdapterDelegate>,
384 _: &mut AsyncApp,
385 ) -> Option<Task<Result<()>>> {
386 None
387 }
388
389 async fn check_if_version_installed(
390 &self,
391 _version: &(dyn 'static + Send + Any),
392 _container_dir: &PathBuf,
393 _delegate: &dyn LspAdapterDelegate,
394 ) -> Option<LanguageServerBinary> {
395 None
396 }
397
398 async fn fetch_server_binary(
399 &self,
400 latest_version: Box<dyn 'static + Send + Any>,
401 container_dir: PathBuf,
402 delegate: &dyn LspAdapterDelegate,
403 ) -> Result<LanguageServerBinary>;
404
405 async fn cached_server_binary(
406 &self,
407 container_dir: PathBuf,
408 delegate: &dyn LspAdapterDelegate,
409 ) -> Option<LanguageServerBinary>;
410
411 fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
412
413 /// Post-processes completions provided by the language server.
414 async fn process_completions(&self, _: &mut [lsp::CompletionItem]) {}
415
416 async fn labels_for_completions(
417 self: Arc<Self>,
418 completions: &[lsp::CompletionItem],
419 language: &Arc<Language>,
420 ) -> Result<Vec<Option<CodeLabel>>> {
421 let mut labels = Vec::new();
422 for (ix, completion) in completions.iter().enumerate() {
423 let label = self.label_for_completion(completion, language).await;
424 if let Some(label) = label {
425 labels.resize(ix + 1, None);
426 *labels.last_mut().unwrap() = Some(label);
427 }
428 }
429 Ok(labels)
430 }
431
432 async fn label_for_completion(
433 &self,
434 _: &lsp::CompletionItem,
435 _: &Arc<Language>,
436 ) -> Option<CodeLabel> {
437 None
438 }
439
440 async fn labels_for_symbols(
441 self: Arc<Self>,
442 symbols: &[(String, lsp::SymbolKind)],
443 language: &Arc<Language>,
444 ) -> Result<Vec<Option<CodeLabel>>> {
445 let mut labels = Vec::new();
446 for (ix, (name, kind)) in symbols.iter().enumerate() {
447 let label = self.label_for_symbol(name, *kind, language).await;
448 if let Some(label) = label {
449 labels.resize(ix + 1, None);
450 *labels.last_mut().unwrap() = Some(label);
451 }
452 }
453 Ok(labels)
454 }
455
456 async fn label_for_symbol(
457 &self,
458 _: &str,
459 _: lsp::SymbolKind,
460 _: &Arc<Language>,
461 ) -> Option<CodeLabel> {
462 None
463 }
464
465 /// Returns initialization options that are going to be sent to a LSP server as a part of [`lsp::InitializeParams`]
466 async fn initialization_options(
467 self: Arc<Self>,
468 _: &dyn Fs,
469 _: &Arc<dyn LspAdapterDelegate>,
470 ) -> Result<Option<Value>> {
471 Ok(None)
472 }
473
474 async fn workspace_configuration(
475 self: Arc<Self>,
476 _: &dyn Fs,
477 _: &Arc<dyn LspAdapterDelegate>,
478 _: Arc<dyn LanguageToolchainStore>,
479 _cx: &mut AsyncApp,
480 ) -> Result<Value> {
481 Ok(serde_json::json!({}))
482 }
483
484 /// Returns a list of code actions supported by a given LspAdapter
485 fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
486 Some(vec![
487 CodeActionKind::EMPTY,
488 CodeActionKind::QUICKFIX,
489 CodeActionKind::REFACTOR,
490 CodeActionKind::REFACTOR_EXTRACT,
491 CodeActionKind::SOURCE,
492 ])
493 }
494
495 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
496 Default::default()
497 }
498
499 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
500 None
501 }
502
503 fn language_ids(&self) -> HashMap<String, String> {
504 Default::default()
505 }
506
507 /// Support custom initialize params.
508 fn prepare_initialize_params(&self, original: InitializeParams) -> Result<InitializeParams> {
509 Ok(original)
510 }
511}
512
513async fn try_fetch_server_binary<L: LspAdapter + 'static + Send + Sync + ?Sized>(
514 adapter: &L,
515 delegate: &Arc<dyn LspAdapterDelegate>,
516 container_dir: PathBuf,
517 cx: &mut AsyncApp,
518) -> Result<LanguageServerBinary> {
519 if let Some(task) = adapter.will_fetch_server(delegate, cx) {
520 task.await?;
521 }
522
523 let name = adapter.name();
524 log::info!("fetching latest version of language server {:?}", name.0);
525 delegate.update_status(name.clone(), LanguageServerBinaryStatus::CheckingForUpdate);
526
527 let latest_version = adapter
528 .fetch_latest_server_version(delegate.as_ref())
529 .await?;
530
531 if let Some(binary) = adapter
532 .check_if_version_installed(latest_version.as_ref(), &container_dir, delegate.as_ref())
533 .await
534 {
535 log::info!("language server {:?} is already installed", name.0);
536 delegate.update_status(name.clone(), LanguageServerBinaryStatus::None);
537 Ok(binary)
538 } else {
539 log::info!("downloading language server {:?}", name.0);
540 delegate.update_status(adapter.name(), LanguageServerBinaryStatus::Downloading);
541 let binary = adapter
542 .fetch_server_binary(latest_version, container_dir, delegate.as_ref())
543 .await;
544
545 delegate.update_status(name.clone(), LanguageServerBinaryStatus::None);
546 binary
547 }
548}
549
550#[derive(Clone, Debug, Default, PartialEq, Eq)]
551pub struct CodeLabel {
552 /// The text to display.
553 pub text: String,
554 /// Syntax highlighting runs.
555 pub runs: Vec<(Range<usize>, HighlightId)>,
556 /// The portion of the text that should be used in fuzzy filtering.
557 pub filter_range: Range<usize>,
558}
559
560#[derive(Clone, Deserialize, JsonSchema)]
561pub struct LanguageConfig {
562 /// Human-readable name of the language.
563 pub name: LanguageName,
564 /// The name of this language for a Markdown code fence block
565 pub code_fence_block_name: Option<Arc<str>>,
566 // The name of the grammar in a WASM bundle (experimental).
567 pub grammar: Option<Arc<str>>,
568 /// The criteria for matching this language to a given file.
569 #[serde(flatten)]
570 pub matcher: LanguageMatcher,
571 /// List of bracket types in a language.
572 #[serde(default)]
573 #[schemars(schema_with = "bracket_pair_config_json_schema")]
574 pub brackets: BracketPairConfig,
575 /// If set to true, auto indentation uses last non empty line to determine
576 /// the indentation level for a new line.
577 #[serde(default = "auto_indent_using_last_non_empty_line_default")]
578 pub auto_indent_using_last_non_empty_line: bool,
579 // Whether indentation of pasted content should be adjusted based on the context.
580 #[serde(default)]
581 pub auto_indent_on_paste: Option<bool>,
582 /// A regex that is used to determine whether the indentation level should be
583 /// increased in the following line.
584 #[serde(default, deserialize_with = "deserialize_regex")]
585 #[schemars(schema_with = "regex_json_schema")]
586 pub increase_indent_pattern: Option<Regex>,
587 /// A regex that is used to determine whether the indentation level should be
588 /// decreased in the following line.
589 #[serde(default, deserialize_with = "deserialize_regex")]
590 #[schemars(schema_with = "regex_json_schema")]
591 pub decrease_indent_pattern: Option<Regex>,
592 /// A list of characters that trigger the automatic insertion of a closing
593 /// bracket when they immediately precede the point where an opening
594 /// bracket is inserted.
595 #[serde(default)]
596 pub autoclose_before: String,
597 /// A placeholder used internally by Semantic Index.
598 #[serde(default)]
599 pub collapsed_placeholder: String,
600 /// A line comment string that is inserted in e.g. `toggle comments` action.
601 /// A language can have multiple flavours of line comments. All of the provided line comments are
602 /// used for comment continuations on the next line, but only the first one is used for Editor::ToggleComments.
603 #[serde(default)]
604 pub line_comments: Vec<Arc<str>>,
605 /// Starting and closing characters of a block comment.
606 #[serde(default)]
607 pub block_comment: Option<(Arc<str>, Arc<str>)>,
608 /// A list of language servers that are allowed to run on subranges of a given language.
609 #[serde(default)]
610 pub scope_opt_in_language_servers: Vec<LanguageServerName>,
611 #[serde(default)]
612 pub overrides: HashMap<String, LanguageConfigOverride>,
613 /// A list of characters that Zed should treat as word characters for the
614 /// purpose of features that operate on word boundaries, like 'move to next word end'
615 /// or a whole-word search in buffer search.
616 #[serde(default)]
617 pub word_characters: HashSet<char>,
618 /// Whether to indent lines using tab characters, as opposed to multiple
619 /// spaces.
620 #[serde(default)]
621 pub hard_tabs: Option<bool>,
622 /// How many columns a tab should occupy.
623 #[serde(default)]
624 pub tab_size: Option<NonZeroU32>,
625 /// How to soft-wrap long lines of text.
626 #[serde(default)]
627 pub soft_wrap: Option<SoftWrap>,
628 /// The name of a Prettier parser that will be used for this language when no file path is available.
629 /// If there's a parser name in the language settings, that will be used instead.
630 #[serde(default)]
631 pub prettier_parser_name: Option<String>,
632 /// If true, this language is only for syntax highlighting via an injection into other
633 /// languages, but should not appear to the user as a distinct language.
634 #[serde(default)]
635 pub hidden: bool,
636}
637
638#[derive(Clone, Debug, Serialize, Deserialize, Default, JsonSchema)]
639pub struct LanguageMatcher {
640 /// 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`.
641 #[serde(default)]
642 pub path_suffixes: Vec<String>,
643 /// A regex pattern that determines whether the language should be assigned to a file or not.
644 #[serde(
645 default,
646 serialize_with = "serialize_regex",
647 deserialize_with = "deserialize_regex"
648 )]
649 #[schemars(schema_with = "regex_json_schema")]
650 pub first_line_pattern: Option<Regex>,
651}
652
653/// Represents a language for the given range. Some languages (e.g. HTML)
654/// interleave several languages together, thus a single buffer might actually contain
655/// several nested scopes.
656#[derive(Clone, Debug)]
657pub struct LanguageScope {
658 language: Arc<Language>,
659 override_id: Option<u32>,
660}
661
662#[derive(Clone, Deserialize, Default, Debug, JsonSchema)]
663pub struct LanguageConfigOverride {
664 #[serde(default)]
665 pub line_comments: Override<Vec<Arc<str>>>,
666 #[serde(default)]
667 pub block_comment: Override<(Arc<str>, Arc<str>)>,
668 #[serde(skip)]
669 pub disabled_bracket_ixs: Vec<u16>,
670 #[serde(default)]
671 pub word_characters: Override<HashSet<char>>,
672 #[serde(default)]
673 pub opt_into_language_servers: Vec<LanguageServerName>,
674}
675
676#[derive(Clone, Deserialize, Debug, Serialize, JsonSchema)]
677#[serde(untagged)]
678pub enum Override<T> {
679 Remove { remove: bool },
680 Set(T),
681}
682
683impl<T> Default for Override<T> {
684 fn default() -> Self {
685 Override::Remove { remove: false }
686 }
687}
688
689impl<T> Override<T> {
690 fn as_option<'a>(this: Option<&'a Self>, original: Option<&'a T>) -> Option<&'a T> {
691 match this {
692 Some(Self::Set(value)) => Some(value),
693 Some(Self::Remove { remove: true }) => None,
694 Some(Self::Remove { remove: false }) | None => original,
695 }
696 }
697}
698
699impl Default for LanguageConfig {
700 fn default() -> Self {
701 Self {
702 name: LanguageName::new(""),
703 code_fence_block_name: None,
704 grammar: None,
705 matcher: LanguageMatcher::default(),
706 brackets: Default::default(),
707 auto_indent_using_last_non_empty_line: auto_indent_using_last_non_empty_line_default(),
708 auto_indent_on_paste: None,
709 increase_indent_pattern: Default::default(),
710 decrease_indent_pattern: Default::default(),
711 autoclose_before: Default::default(),
712 line_comments: Default::default(),
713 block_comment: Default::default(),
714 scope_opt_in_language_servers: Default::default(),
715 overrides: Default::default(),
716 word_characters: Default::default(),
717 collapsed_placeholder: Default::default(),
718 hard_tabs: None,
719 tab_size: None,
720 soft_wrap: None,
721 prettier_parser_name: None,
722 hidden: false,
723 }
724 }
725}
726
727fn auto_indent_using_last_non_empty_line_default() -> bool {
728 true
729}
730
731fn deserialize_regex<'de, D: Deserializer<'de>>(d: D) -> Result<Option<Regex>, D::Error> {
732 let source = Option::<String>::deserialize(d)?;
733 if let Some(source) = source {
734 Ok(Some(regex::Regex::new(&source).map_err(de::Error::custom)?))
735 } else {
736 Ok(None)
737 }
738}
739
740fn regex_json_schema(_: &mut SchemaGenerator) -> Schema {
741 Schema::Object(SchemaObject {
742 instance_type: Some(InstanceType::String.into()),
743 ..Default::default()
744 })
745}
746
747fn serialize_regex<S>(regex: &Option<Regex>, serializer: S) -> Result<S::Ok, S::Error>
748where
749 S: Serializer,
750{
751 match regex {
752 Some(regex) => serializer.serialize_str(regex.as_str()),
753 None => serializer.serialize_none(),
754 }
755}
756
757#[doc(hidden)]
758#[cfg(any(test, feature = "test-support"))]
759pub struct FakeLspAdapter {
760 pub name: &'static str,
761 pub initialization_options: Option<Value>,
762 pub prettier_plugins: Vec<&'static str>,
763 pub disk_based_diagnostics_progress_token: Option<String>,
764 pub disk_based_diagnostics_sources: Vec<String>,
765 pub language_server_binary: LanguageServerBinary,
766
767 pub capabilities: lsp::ServerCapabilities,
768 pub initializer: Option<Box<dyn 'static + Send + Sync + Fn(&mut lsp::FakeLanguageServer)>>,
769 pub label_for_completion: Option<
770 Box<
771 dyn 'static
772 + Send
773 + Sync
774 + Fn(&lsp::CompletionItem, &Arc<Language>) -> Option<CodeLabel>,
775 >,
776 >,
777}
778
779/// Configuration of handling bracket pairs for a given language.
780///
781/// This struct includes settings for defining which pairs of characters are considered brackets and
782/// also specifies any language-specific scopes where these pairs should be ignored for bracket matching purposes.
783#[derive(Clone, Debug, Default, JsonSchema)]
784pub struct BracketPairConfig {
785 /// A list of character pairs that should be treated as brackets in the context of a given language.
786 pub pairs: Vec<BracketPair>,
787 /// A list of tree-sitter scopes for which a given bracket should not be active.
788 /// N-th entry in `[Self::disabled_scopes_by_bracket_ix]` contains a list of disabled scopes for an n-th entry in `[Self::pairs]`
789 #[serde(skip)]
790 pub disabled_scopes_by_bracket_ix: Vec<Vec<String>>,
791}
792
793fn bracket_pair_config_json_schema(gen: &mut SchemaGenerator) -> Schema {
794 Option::<Vec<BracketPairContent>>::json_schema(gen)
795}
796
797#[derive(Deserialize, JsonSchema)]
798pub struct BracketPairContent {
799 #[serde(flatten)]
800 pub bracket_pair: BracketPair,
801 #[serde(default)]
802 pub not_in: Vec<String>,
803}
804
805impl<'de> Deserialize<'de> for BracketPairConfig {
806 fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
807 where
808 D: Deserializer<'de>,
809 {
810 let result = Vec::<BracketPairContent>::deserialize(deserializer)?;
811 let mut brackets = Vec::with_capacity(result.len());
812 let mut disabled_scopes_by_bracket_ix = Vec::with_capacity(result.len());
813 for entry in result {
814 brackets.push(entry.bracket_pair);
815 disabled_scopes_by_bracket_ix.push(entry.not_in);
816 }
817
818 Ok(BracketPairConfig {
819 pairs: brackets,
820 disabled_scopes_by_bracket_ix,
821 })
822 }
823}
824
825/// Describes a single bracket pair and how an editor should react to e.g. inserting
826/// an opening bracket or to a newline character insertion in between `start` and `end` characters.
827#[derive(Clone, Debug, Default, Deserialize, PartialEq, JsonSchema)]
828pub struct BracketPair {
829 /// Starting substring for a bracket.
830 pub start: String,
831 /// Ending substring for a bracket.
832 pub end: String,
833 /// True if `end` should be automatically inserted right after `start` characters.
834 pub close: bool,
835 /// True if selected text should be surrounded by `start` and `end` characters.
836 #[serde(default = "default_true")]
837 pub surround: bool,
838 /// True if an extra newline should be inserted while the cursor is in the middle
839 /// of that bracket pair.
840 pub newline: bool,
841}
842
843#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
844pub(crate) struct LanguageId(usize);
845
846impl LanguageId {
847 pub(crate) fn new() -> Self {
848 Self(NEXT_LANGUAGE_ID.fetch_add(1, SeqCst))
849 }
850}
851
852pub struct Language {
853 pub(crate) id: LanguageId,
854 pub(crate) config: LanguageConfig,
855 pub(crate) grammar: Option<Arc<Grammar>>,
856 pub(crate) context_provider: Option<Arc<dyn ContextProvider>>,
857 pub(crate) toolchain: Option<Arc<dyn ToolchainLister>>,
858}
859
860#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
861pub struct GrammarId(pub usize);
862
863impl GrammarId {
864 pub(crate) fn new() -> Self {
865 Self(NEXT_GRAMMAR_ID.fetch_add(1, SeqCst))
866 }
867}
868
869pub struct Grammar {
870 id: GrammarId,
871 pub ts_language: tree_sitter::Language,
872 pub(crate) error_query: Query,
873 pub(crate) highlights_query: Option<Query>,
874 pub(crate) brackets_config: Option<BracketConfig>,
875 pub(crate) redactions_config: Option<RedactionConfig>,
876 pub(crate) runnable_config: Option<RunnableConfig>,
877 pub(crate) indents_config: Option<IndentConfig>,
878 pub outline_config: Option<OutlineConfig>,
879 pub text_object_config: Option<TextObjectConfig>,
880 pub embedding_config: Option<EmbeddingConfig>,
881 pub(crate) injection_config: Option<InjectionConfig>,
882 pub(crate) override_config: Option<OverrideConfig>,
883 pub(crate) highlight_map: Mutex<HighlightMap>,
884}
885
886struct IndentConfig {
887 query: Query,
888 indent_capture_ix: u32,
889 start_capture_ix: Option<u32>,
890 end_capture_ix: Option<u32>,
891 outdent_capture_ix: Option<u32>,
892}
893
894pub struct OutlineConfig {
895 pub query: Query,
896 pub item_capture_ix: u32,
897 pub name_capture_ix: u32,
898 pub context_capture_ix: Option<u32>,
899 pub extra_context_capture_ix: Option<u32>,
900 pub open_capture_ix: Option<u32>,
901 pub close_capture_ix: Option<u32>,
902 pub annotation_capture_ix: Option<u32>,
903}
904
905#[derive(Debug, Clone, Copy, PartialEq)]
906pub enum TextObject {
907 InsideFunction,
908 AroundFunction,
909 InsideClass,
910 AroundClass,
911 InsideComment,
912 AroundComment,
913}
914
915impl TextObject {
916 pub fn from_capture_name(name: &str) -> Option<TextObject> {
917 match name {
918 "function.inside" => Some(TextObject::InsideFunction),
919 "function.around" => Some(TextObject::AroundFunction),
920 "class.inside" => Some(TextObject::InsideClass),
921 "class.around" => Some(TextObject::AroundClass),
922 "comment.inside" => Some(TextObject::InsideComment),
923 "comment.around" => Some(TextObject::AroundComment),
924 _ => None,
925 }
926 }
927
928 pub fn around(&self) -> Option<Self> {
929 match self {
930 TextObject::InsideFunction => Some(TextObject::AroundFunction),
931 TextObject::InsideClass => Some(TextObject::AroundClass),
932 TextObject::InsideComment => Some(TextObject::AroundComment),
933 _ => None,
934 }
935 }
936}
937
938pub struct TextObjectConfig {
939 pub query: Query,
940 pub text_objects_by_capture_ix: Vec<(u32, TextObject)>,
941}
942
943#[derive(Debug)]
944pub struct EmbeddingConfig {
945 pub query: Query,
946 pub item_capture_ix: u32,
947 pub name_capture_ix: Option<u32>,
948 pub context_capture_ix: Option<u32>,
949 pub collapse_capture_ix: Option<u32>,
950 pub keep_capture_ix: Option<u32>,
951}
952
953struct InjectionConfig {
954 query: Query,
955 content_capture_ix: u32,
956 language_capture_ix: Option<u32>,
957 patterns: Vec<InjectionPatternConfig>,
958}
959
960struct RedactionConfig {
961 pub query: Query,
962 pub redaction_capture_ix: u32,
963}
964
965#[derive(Clone, Debug, PartialEq)]
966enum RunnableCapture {
967 Named(SharedString),
968 Run,
969}
970
971struct RunnableConfig {
972 pub query: Query,
973 /// A mapping from capture indice to capture kind
974 pub extra_captures: Vec<RunnableCapture>,
975}
976
977struct OverrideConfig {
978 query: Query,
979 values: HashMap<u32, OverrideEntry>,
980}
981
982#[derive(Debug)]
983struct OverrideEntry {
984 name: String,
985 range_is_inclusive: bool,
986 value: LanguageConfigOverride,
987}
988
989#[derive(Default, Clone)]
990struct InjectionPatternConfig {
991 language: Option<Box<str>>,
992 combined: bool,
993}
994
995struct BracketConfig {
996 query: Query,
997 open_capture_ix: u32,
998 close_capture_ix: u32,
999}
1000
1001impl Language {
1002 pub fn new(config: LanguageConfig, ts_language: Option<tree_sitter::Language>) -> Self {
1003 Self::new_with_id(LanguageId::new(), config, ts_language)
1004 }
1005
1006 fn new_with_id(
1007 id: LanguageId,
1008 config: LanguageConfig,
1009 ts_language: Option<tree_sitter::Language>,
1010 ) -> Self {
1011 Self {
1012 id,
1013 config,
1014 grammar: ts_language.map(|ts_language| {
1015 Arc::new(Grammar {
1016 id: GrammarId::new(),
1017 highlights_query: None,
1018 brackets_config: None,
1019 outline_config: None,
1020 text_object_config: None,
1021 embedding_config: None,
1022 indents_config: None,
1023 injection_config: None,
1024 override_config: None,
1025 redactions_config: None,
1026 runnable_config: None,
1027 error_query: Query::new(&ts_language, "(ERROR) @error").unwrap(),
1028 ts_language,
1029 highlight_map: Default::default(),
1030 })
1031 }),
1032 context_provider: None,
1033 toolchain: None,
1034 }
1035 }
1036
1037 pub fn with_context_provider(mut self, provider: Option<Arc<dyn ContextProvider>>) -> Self {
1038 self.context_provider = provider;
1039 self
1040 }
1041
1042 pub fn with_toolchain_lister(mut self, provider: Option<Arc<dyn ToolchainLister>>) -> Self {
1043 self.toolchain = provider;
1044 self
1045 }
1046
1047 pub fn with_queries(mut self, queries: LanguageQueries) -> Result<Self> {
1048 if let Some(query) = queries.highlights {
1049 self = self
1050 .with_highlights_query(query.as_ref())
1051 .context("Error loading highlights query")?;
1052 }
1053 if let Some(query) = queries.brackets {
1054 self = self
1055 .with_brackets_query(query.as_ref())
1056 .context("Error loading brackets query")?;
1057 }
1058 if let Some(query) = queries.indents {
1059 self = self
1060 .with_indents_query(query.as_ref())
1061 .context("Error loading indents query")?;
1062 }
1063 if let Some(query) = queries.outline {
1064 self = self
1065 .with_outline_query(query.as_ref())
1066 .context("Error loading outline query")?;
1067 }
1068 if let Some(query) = queries.embedding {
1069 self = self
1070 .with_embedding_query(query.as_ref())
1071 .context("Error loading embedding query")?;
1072 }
1073 if let Some(query) = queries.injections {
1074 self = self
1075 .with_injection_query(query.as_ref())
1076 .context("Error loading injection query")?;
1077 }
1078 if let Some(query) = queries.overrides {
1079 self = self
1080 .with_override_query(query.as_ref())
1081 .context("Error loading override query")?;
1082 }
1083 if let Some(query) = queries.redactions {
1084 self = self
1085 .with_redaction_query(query.as_ref())
1086 .context("Error loading redaction query")?;
1087 }
1088 if let Some(query) = queries.runnables {
1089 self = self
1090 .with_runnable_query(query.as_ref())
1091 .context("Error loading runnables query")?;
1092 }
1093 if let Some(query) = queries.text_objects {
1094 self = self
1095 .with_text_object_query(query.as_ref())
1096 .context("Error loading textobject query")?;
1097 }
1098 Ok(self)
1099 }
1100
1101 pub fn with_highlights_query(mut self, source: &str) -> Result<Self> {
1102 let grammar = self
1103 .grammar_mut()
1104 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1105 grammar.highlights_query = Some(Query::new(&grammar.ts_language, source)?);
1106 Ok(self)
1107 }
1108
1109 pub fn with_runnable_query(mut self, source: &str) -> Result<Self> {
1110 let grammar = self
1111 .grammar_mut()
1112 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1113
1114 let query = Query::new(&grammar.ts_language, source)?;
1115 let mut extra_captures = Vec::with_capacity(query.capture_names().len());
1116
1117 for name in query.capture_names().iter() {
1118 let kind = if *name == "run" {
1119 RunnableCapture::Run
1120 } else {
1121 RunnableCapture::Named(name.to_string().into())
1122 };
1123 extra_captures.push(kind);
1124 }
1125
1126 grammar.runnable_config = Some(RunnableConfig {
1127 extra_captures,
1128 query,
1129 });
1130
1131 Ok(self)
1132 }
1133
1134 pub fn with_outline_query(mut self, source: &str) -> Result<Self> {
1135 let grammar = self
1136 .grammar_mut()
1137 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1138 let query = Query::new(&grammar.ts_language, source)?;
1139 let mut item_capture_ix = None;
1140 let mut name_capture_ix = None;
1141 let mut context_capture_ix = None;
1142 let mut extra_context_capture_ix = None;
1143 let mut open_capture_ix = None;
1144 let mut close_capture_ix = None;
1145 let mut annotation_capture_ix = None;
1146 get_capture_indices(
1147 &query,
1148 &mut [
1149 ("item", &mut item_capture_ix),
1150 ("name", &mut name_capture_ix),
1151 ("context", &mut context_capture_ix),
1152 ("context.extra", &mut extra_context_capture_ix),
1153 ("open", &mut open_capture_ix),
1154 ("close", &mut close_capture_ix),
1155 ("annotation", &mut annotation_capture_ix),
1156 ],
1157 );
1158 if let Some((item_capture_ix, name_capture_ix)) = item_capture_ix.zip(name_capture_ix) {
1159 grammar.outline_config = Some(OutlineConfig {
1160 query,
1161 item_capture_ix,
1162 name_capture_ix,
1163 context_capture_ix,
1164 extra_context_capture_ix,
1165 open_capture_ix,
1166 close_capture_ix,
1167 annotation_capture_ix,
1168 });
1169 }
1170 Ok(self)
1171 }
1172
1173 pub fn with_text_object_query(mut self, source: &str) -> Result<Self> {
1174 let grammar = self
1175 .grammar_mut()
1176 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1177 let query = Query::new(&grammar.ts_language, source)?;
1178
1179 let mut text_objects_by_capture_ix = Vec::new();
1180 for (ix, name) in query.capture_names().iter().enumerate() {
1181 if let Some(text_object) = TextObject::from_capture_name(name) {
1182 text_objects_by_capture_ix.push((ix as u32, text_object));
1183 }
1184 }
1185
1186 grammar.text_object_config = Some(TextObjectConfig {
1187 query,
1188 text_objects_by_capture_ix,
1189 });
1190 Ok(self)
1191 }
1192
1193 pub fn with_embedding_query(mut self, source: &str) -> Result<Self> {
1194 let grammar = self
1195 .grammar_mut()
1196 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1197 let query = Query::new(&grammar.ts_language, source)?;
1198 let mut item_capture_ix = None;
1199 let mut name_capture_ix = None;
1200 let mut context_capture_ix = None;
1201 let mut collapse_capture_ix = None;
1202 let mut keep_capture_ix = None;
1203 get_capture_indices(
1204 &query,
1205 &mut [
1206 ("item", &mut item_capture_ix),
1207 ("name", &mut name_capture_ix),
1208 ("context", &mut context_capture_ix),
1209 ("keep", &mut keep_capture_ix),
1210 ("collapse", &mut collapse_capture_ix),
1211 ],
1212 );
1213 if let Some(item_capture_ix) = item_capture_ix {
1214 grammar.embedding_config = Some(EmbeddingConfig {
1215 query,
1216 item_capture_ix,
1217 name_capture_ix,
1218 context_capture_ix,
1219 collapse_capture_ix,
1220 keep_capture_ix,
1221 });
1222 }
1223 Ok(self)
1224 }
1225
1226 pub fn with_brackets_query(mut self, source: &str) -> Result<Self> {
1227 let grammar = self
1228 .grammar_mut()
1229 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1230 let query = Query::new(&grammar.ts_language, source)?;
1231 let mut open_capture_ix = None;
1232 let mut close_capture_ix = None;
1233 get_capture_indices(
1234 &query,
1235 &mut [
1236 ("open", &mut open_capture_ix),
1237 ("close", &mut close_capture_ix),
1238 ],
1239 );
1240 if let Some((open_capture_ix, close_capture_ix)) = open_capture_ix.zip(close_capture_ix) {
1241 grammar.brackets_config = Some(BracketConfig {
1242 query,
1243 open_capture_ix,
1244 close_capture_ix,
1245 });
1246 }
1247 Ok(self)
1248 }
1249
1250 pub fn with_indents_query(mut self, source: &str) -> Result<Self> {
1251 let grammar = self
1252 .grammar_mut()
1253 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1254 let query = Query::new(&grammar.ts_language, source)?;
1255 let mut indent_capture_ix = None;
1256 let mut start_capture_ix = None;
1257 let mut end_capture_ix = None;
1258 let mut outdent_capture_ix = None;
1259 get_capture_indices(
1260 &query,
1261 &mut [
1262 ("indent", &mut indent_capture_ix),
1263 ("start", &mut start_capture_ix),
1264 ("end", &mut end_capture_ix),
1265 ("outdent", &mut outdent_capture_ix),
1266 ],
1267 );
1268 if let Some(indent_capture_ix) = indent_capture_ix {
1269 grammar.indents_config = Some(IndentConfig {
1270 query,
1271 indent_capture_ix,
1272 start_capture_ix,
1273 end_capture_ix,
1274 outdent_capture_ix,
1275 });
1276 }
1277 Ok(self)
1278 }
1279
1280 pub fn with_injection_query(mut self, source: &str) -> Result<Self> {
1281 let grammar = self
1282 .grammar_mut()
1283 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1284 let query = Query::new(&grammar.ts_language, source)?;
1285 let mut language_capture_ix = None;
1286 let mut injection_language_capture_ix = None;
1287 let mut content_capture_ix = None;
1288 let mut injection_content_capture_ix = None;
1289 get_capture_indices(
1290 &query,
1291 &mut [
1292 ("language", &mut language_capture_ix),
1293 ("injection.language", &mut injection_language_capture_ix),
1294 ("content", &mut content_capture_ix),
1295 ("injection.content", &mut injection_content_capture_ix),
1296 ],
1297 );
1298 language_capture_ix = match (language_capture_ix, injection_language_capture_ix) {
1299 (None, Some(ix)) => Some(ix),
1300 (Some(_), Some(_)) => {
1301 return Err(anyhow!(
1302 "both language and injection.language captures are present"
1303 ));
1304 }
1305 _ => language_capture_ix,
1306 };
1307 content_capture_ix = match (content_capture_ix, injection_content_capture_ix) {
1308 (None, Some(ix)) => Some(ix),
1309 (Some(_), Some(_)) => {
1310 return Err(anyhow!(
1311 "both content and injection.content captures are present"
1312 ));
1313 }
1314 _ => content_capture_ix,
1315 };
1316 let patterns = (0..query.pattern_count())
1317 .map(|ix| {
1318 let mut config = InjectionPatternConfig::default();
1319 for setting in query.property_settings(ix) {
1320 match setting.key.as_ref() {
1321 "language" | "injection.language" => {
1322 config.language.clone_from(&setting.value);
1323 }
1324 "combined" | "injection.combined" => {
1325 config.combined = true;
1326 }
1327 _ => {}
1328 }
1329 }
1330 config
1331 })
1332 .collect();
1333 if let Some(content_capture_ix) = content_capture_ix {
1334 grammar.injection_config = Some(InjectionConfig {
1335 query,
1336 language_capture_ix,
1337 content_capture_ix,
1338 patterns,
1339 });
1340 }
1341 Ok(self)
1342 }
1343
1344 pub fn with_override_query(mut self, source: &str) -> anyhow::Result<Self> {
1345 let query = {
1346 let grammar = self
1347 .grammar
1348 .as_ref()
1349 .ok_or_else(|| anyhow!("no grammar for language"))?;
1350 Query::new(&grammar.ts_language, source)?
1351 };
1352
1353 let mut override_configs_by_id = HashMap::default();
1354 for (ix, mut name) in query.capture_names().iter().copied().enumerate() {
1355 let mut range_is_inclusive = false;
1356 if name.starts_with('_') {
1357 continue;
1358 }
1359 if let Some(prefix) = name.strip_suffix(".inclusive") {
1360 name = prefix;
1361 range_is_inclusive = true;
1362 }
1363
1364 let value = self.config.overrides.get(name).cloned().unwrap_or_default();
1365 for server_name in &value.opt_into_language_servers {
1366 if !self
1367 .config
1368 .scope_opt_in_language_servers
1369 .contains(server_name)
1370 {
1371 util::debug_panic!("Server {server_name:?} has been opted-in by scope {name:?} but has not been marked as an opt-in server");
1372 }
1373 }
1374
1375 override_configs_by_id.insert(
1376 ix as u32,
1377 OverrideEntry {
1378 name: name.to_string(),
1379 range_is_inclusive,
1380 value,
1381 },
1382 );
1383 }
1384
1385 let referenced_override_names = self.config.overrides.keys().chain(
1386 self.config
1387 .brackets
1388 .disabled_scopes_by_bracket_ix
1389 .iter()
1390 .flatten(),
1391 );
1392
1393 for referenced_name in referenced_override_names {
1394 if !override_configs_by_id
1395 .values()
1396 .any(|entry| entry.name == *referenced_name)
1397 {
1398 Err(anyhow!(
1399 "language {:?} has overrides in config not in query: {referenced_name:?}",
1400 self.config.name
1401 ))?;
1402 }
1403 }
1404
1405 for entry in override_configs_by_id.values_mut() {
1406 entry.value.disabled_bracket_ixs = self
1407 .config
1408 .brackets
1409 .disabled_scopes_by_bracket_ix
1410 .iter()
1411 .enumerate()
1412 .filter_map(|(ix, disabled_scope_names)| {
1413 if disabled_scope_names.contains(&entry.name) {
1414 Some(ix as u16)
1415 } else {
1416 None
1417 }
1418 })
1419 .collect();
1420 }
1421
1422 self.config.brackets.disabled_scopes_by_bracket_ix.clear();
1423
1424 let grammar = self
1425 .grammar_mut()
1426 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1427 grammar.override_config = Some(OverrideConfig {
1428 query,
1429 values: override_configs_by_id,
1430 });
1431 Ok(self)
1432 }
1433
1434 pub fn with_redaction_query(mut self, source: &str) -> anyhow::Result<Self> {
1435 let grammar = self
1436 .grammar_mut()
1437 .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1438
1439 let query = Query::new(&grammar.ts_language, source)?;
1440 let mut redaction_capture_ix = None;
1441 get_capture_indices(&query, &mut [("redact", &mut redaction_capture_ix)]);
1442
1443 if let Some(redaction_capture_ix) = redaction_capture_ix {
1444 grammar.redactions_config = Some(RedactionConfig {
1445 query,
1446 redaction_capture_ix,
1447 });
1448 }
1449
1450 Ok(self)
1451 }
1452
1453 fn grammar_mut(&mut self) -> Option<&mut Grammar> {
1454 Arc::get_mut(self.grammar.as_mut()?)
1455 }
1456
1457 pub fn name(&self) -> LanguageName {
1458 self.config.name.clone()
1459 }
1460
1461 pub fn code_fence_block_name(&self) -> Arc<str> {
1462 self.config
1463 .code_fence_block_name
1464 .clone()
1465 .unwrap_or_else(|| self.config.name.as_ref().to_lowercase().into())
1466 }
1467
1468 pub fn context_provider(&self) -> Option<Arc<dyn ContextProvider>> {
1469 self.context_provider.clone()
1470 }
1471
1472 pub fn toolchain_lister(&self) -> Option<Arc<dyn ToolchainLister>> {
1473 self.toolchain.clone()
1474 }
1475
1476 pub fn highlight_text<'a>(
1477 self: &'a Arc<Self>,
1478 text: &'a Rope,
1479 range: Range<usize>,
1480 ) -> Vec<(Range<usize>, HighlightId)> {
1481 let mut result = Vec::new();
1482 if let Some(grammar) = &self.grammar {
1483 let tree = grammar.parse_text(text, None);
1484 let captures =
1485 SyntaxSnapshot::single_tree_captures(range.clone(), text, &tree, self, |grammar| {
1486 grammar.highlights_query.as_ref()
1487 });
1488 let highlight_maps = vec![grammar.highlight_map()];
1489 let mut offset = 0;
1490 for chunk in
1491 BufferChunks::new(text, range, Some((captures, highlight_maps)), false, None)
1492 {
1493 let end_offset = offset + chunk.text.len();
1494 if let Some(highlight_id) = chunk.syntax_highlight_id {
1495 if !highlight_id.is_default() {
1496 result.push((offset..end_offset, highlight_id));
1497 }
1498 }
1499 offset = end_offset;
1500 }
1501 }
1502 result
1503 }
1504
1505 pub fn path_suffixes(&self) -> &[String] {
1506 &self.config.matcher.path_suffixes
1507 }
1508
1509 pub fn should_autoclose_before(&self, c: char) -> bool {
1510 c.is_whitespace() || self.config.autoclose_before.contains(c)
1511 }
1512
1513 pub fn set_theme(&self, theme: &SyntaxTheme) {
1514 if let Some(grammar) = self.grammar.as_ref() {
1515 if let Some(highlights_query) = &grammar.highlights_query {
1516 *grammar.highlight_map.lock() =
1517 HighlightMap::new(highlights_query.capture_names(), theme);
1518 }
1519 }
1520 }
1521
1522 pub fn grammar(&self) -> Option<&Arc<Grammar>> {
1523 self.grammar.as_ref()
1524 }
1525
1526 pub fn default_scope(self: &Arc<Self>) -> LanguageScope {
1527 LanguageScope {
1528 language: self.clone(),
1529 override_id: None,
1530 }
1531 }
1532
1533 pub fn lsp_id(&self) -> String {
1534 self.config.name.lsp_id()
1535 }
1536
1537 pub fn prettier_parser_name(&self) -> Option<&str> {
1538 self.config.prettier_parser_name.as_deref()
1539 }
1540
1541 pub fn config(&self) -> &LanguageConfig {
1542 &self.config
1543 }
1544}
1545
1546impl LanguageScope {
1547 pub fn path_suffixes(&self) -> &[String] {
1548 &self.language.path_suffixes()
1549 }
1550
1551 pub fn language_name(&self) -> LanguageName {
1552 self.language.config.name.clone()
1553 }
1554
1555 pub fn collapsed_placeholder(&self) -> &str {
1556 self.language.config.collapsed_placeholder.as_ref()
1557 }
1558
1559 /// Returns line prefix that is inserted in e.g. line continuations or
1560 /// in `toggle comments` action.
1561 pub fn line_comment_prefixes(&self) -> &[Arc<str>] {
1562 Override::as_option(
1563 self.config_override().map(|o| &o.line_comments),
1564 Some(&self.language.config.line_comments),
1565 )
1566 .map_or([].as_slice(), |e| e.as_slice())
1567 }
1568
1569 pub fn block_comment_delimiters(&self) -> Option<(&Arc<str>, &Arc<str>)> {
1570 Override::as_option(
1571 self.config_override().map(|o| &o.block_comment),
1572 self.language.config.block_comment.as_ref(),
1573 )
1574 .map(|e| (&e.0, &e.1))
1575 }
1576
1577 /// Returns a list of language-specific word characters.
1578 ///
1579 /// By default, Zed treats alphanumeric characters (and '_') as word characters for
1580 /// the purpose of actions like 'move to next word end` or whole-word search.
1581 /// It additionally accounts for language's additional word characters.
1582 pub fn word_characters(&self) -> Option<&HashSet<char>> {
1583 Override::as_option(
1584 self.config_override().map(|o| &o.word_characters),
1585 Some(&self.language.config.word_characters),
1586 )
1587 }
1588
1589 /// Returns a list of bracket pairs for a given language with an additional
1590 /// piece of information about whether the particular bracket pair is currently active for a given language.
1591 pub fn brackets(&self) -> impl Iterator<Item = (&BracketPair, bool)> {
1592 let mut disabled_ids = self
1593 .config_override()
1594 .map_or(&[] as _, |o| o.disabled_bracket_ixs.as_slice());
1595 self.language
1596 .config
1597 .brackets
1598 .pairs
1599 .iter()
1600 .enumerate()
1601 .map(move |(ix, bracket)| {
1602 let mut is_enabled = true;
1603 if let Some(next_disabled_ix) = disabled_ids.first() {
1604 if ix == *next_disabled_ix as usize {
1605 disabled_ids = &disabled_ids[1..];
1606 is_enabled = false;
1607 }
1608 }
1609 (bracket, is_enabled)
1610 })
1611 }
1612
1613 pub fn should_autoclose_before(&self, c: char) -> bool {
1614 c.is_whitespace() || self.language.config.autoclose_before.contains(c)
1615 }
1616
1617 pub fn language_allowed(&self, name: &LanguageServerName) -> bool {
1618 let config = &self.language.config;
1619 let opt_in_servers = &config.scope_opt_in_language_servers;
1620 if opt_in_servers.iter().any(|o| *o == *name) {
1621 if let Some(over) = self.config_override() {
1622 over.opt_into_language_servers.iter().any(|o| *o == *name)
1623 } else {
1624 false
1625 }
1626 } else {
1627 true
1628 }
1629 }
1630
1631 pub fn override_name(&self) -> Option<&str> {
1632 let id = self.override_id?;
1633 let grammar = self.language.grammar.as_ref()?;
1634 let override_config = grammar.override_config.as_ref()?;
1635 override_config.values.get(&id).map(|e| e.name.as_str())
1636 }
1637
1638 fn config_override(&self) -> Option<&LanguageConfigOverride> {
1639 let id = self.override_id?;
1640 let grammar = self.language.grammar.as_ref()?;
1641 let override_config = grammar.override_config.as_ref()?;
1642 override_config.values.get(&id).map(|e| &e.value)
1643 }
1644}
1645
1646impl Hash for Language {
1647 fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
1648 self.id.hash(state)
1649 }
1650}
1651
1652impl PartialEq for Language {
1653 fn eq(&self, other: &Self) -> bool {
1654 self.id.eq(&other.id)
1655 }
1656}
1657
1658impl Eq for Language {}
1659
1660impl Debug for Language {
1661 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
1662 f.debug_struct("Language")
1663 .field("name", &self.config.name)
1664 .finish()
1665 }
1666}
1667
1668impl Grammar {
1669 pub fn id(&self) -> GrammarId {
1670 self.id
1671 }
1672
1673 fn parse_text(&self, text: &Rope, old_tree: Option<Tree>) -> Tree {
1674 with_parser(|parser| {
1675 parser
1676 .set_language(&self.ts_language)
1677 .expect("incompatible grammar");
1678 let mut chunks = text.chunks_in_range(0..text.len());
1679 parser
1680 .parse_with(
1681 &mut move |offset, _| {
1682 chunks.seek(offset);
1683 chunks.next().unwrap_or("").as_bytes()
1684 },
1685 old_tree.as_ref(),
1686 )
1687 .unwrap()
1688 })
1689 }
1690
1691 pub fn highlight_map(&self) -> HighlightMap {
1692 self.highlight_map.lock().clone()
1693 }
1694
1695 pub fn highlight_id_for_name(&self, name: &str) -> Option<HighlightId> {
1696 let capture_id = self
1697 .highlights_query
1698 .as_ref()?
1699 .capture_index_for_name(name)?;
1700 Some(self.highlight_map.lock().get(capture_id))
1701 }
1702}
1703
1704impl CodeLabel {
1705 pub fn fallback_for_completion(
1706 item: &lsp::CompletionItem,
1707 language: Option<&Language>,
1708 ) -> Self {
1709 let highlight_id = item.kind.and_then(|kind| {
1710 let grammar = language?.grammar()?;
1711 use lsp::CompletionItemKind as Kind;
1712 match kind {
1713 Kind::CLASS => grammar.highlight_id_for_name("type"),
1714 Kind::CONSTANT => grammar.highlight_id_for_name("constant"),
1715 Kind::CONSTRUCTOR => grammar.highlight_id_for_name("constructor"),
1716 Kind::ENUM => grammar
1717 .highlight_id_for_name("enum")
1718 .or_else(|| grammar.highlight_id_for_name("type")),
1719 Kind::FIELD => grammar.highlight_id_for_name("property"),
1720 Kind::FUNCTION => grammar.highlight_id_for_name("function"),
1721 Kind::INTERFACE => grammar.highlight_id_for_name("type"),
1722 Kind::METHOD => grammar
1723 .highlight_id_for_name("function.method")
1724 .or_else(|| grammar.highlight_id_for_name("function")),
1725 Kind::OPERATOR => grammar.highlight_id_for_name("operator"),
1726 Kind::PROPERTY => grammar.highlight_id_for_name("property"),
1727 Kind::STRUCT => grammar.highlight_id_for_name("type"),
1728 Kind::VARIABLE => grammar.highlight_id_for_name("variable"),
1729 Kind::KEYWORD => grammar.highlight_id_for_name("keyword"),
1730 _ => None,
1731 }
1732 });
1733
1734 let label = &item.label;
1735 let label_length = label.len();
1736 let runs = highlight_id
1737 .map(|highlight_id| vec![(0..label_length, highlight_id)])
1738 .unwrap_or_default();
1739 let text = if let Some(detail) = &item.detail {
1740 format!("{label} {detail}")
1741 } else if let Some(description) = item
1742 .label_details
1743 .as_ref()
1744 .and_then(|label_details| label_details.description.as_ref())
1745 {
1746 format!("{label} {description}")
1747 } else {
1748 label.clone()
1749 };
1750 Self {
1751 text,
1752 runs,
1753 filter_range: 0..label_length,
1754 }
1755 }
1756
1757 pub fn plain(text: String, filter_text: Option<&str>) -> Self {
1758 let mut result = Self {
1759 runs: Vec::new(),
1760 filter_range: 0..text.len(),
1761 text,
1762 };
1763 if let Some(filter_text) = filter_text {
1764 if let Some(ix) = result.text.find(filter_text) {
1765 result.filter_range = ix..ix + filter_text.len();
1766 }
1767 }
1768 result
1769 }
1770
1771 pub fn push_str(&mut self, text: &str, highlight: Option<HighlightId>) {
1772 let start_ix = self.text.len();
1773 self.text.push_str(text);
1774 let end_ix = self.text.len();
1775 if let Some(highlight) = highlight {
1776 self.runs.push((start_ix..end_ix, highlight));
1777 }
1778 }
1779
1780 pub fn text(&self) -> &str {
1781 self.text.as_str()
1782 }
1783
1784 pub fn filter_text(&self) -> &str {
1785 &self.text[self.filter_range.clone()]
1786 }
1787}
1788
1789impl From<String> for CodeLabel {
1790 fn from(value: String) -> Self {
1791 Self::plain(value, None)
1792 }
1793}
1794
1795impl From<&str> for CodeLabel {
1796 fn from(value: &str) -> Self {
1797 Self::plain(value.to_string(), None)
1798 }
1799}
1800
1801impl Ord for LanguageMatcher {
1802 fn cmp(&self, other: &Self) -> std::cmp::Ordering {
1803 self.path_suffixes.cmp(&other.path_suffixes).then_with(|| {
1804 self.first_line_pattern
1805 .as_ref()
1806 .map(Regex::as_str)
1807 .cmp(&other.first_line_pattern.as_ref().map(Regex::as_str))
1808 })
1809 }
1810}
1811
1812impl PartialOrd for LanguageMatcher {
1813 fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1814 Some(self.cmp(other))
1815 }
1816}
1817
1818impl Eq for LanguageMatcher {}
1819
1820impl PartialEq for LanguageMatcher {
1821 fn eq(&self, other: &Self) -> bool {
1822 self.path_suffixes == other.path_suffixes
1823 && self.first_line_pattern.as_ref().map(Regex::as_str)
1824 == other.first_line_pattern.as_ref().map(Regex::as_str)
1825 }
1826}
1827
1828#[cfg(any(test, feature = "test-support"))]
1829impl Default for FakeLspAdapter {
1830 fn default() -> Self {
1831 Self {
1832 name: "the-fake-language-server",
1833 capabilities: lsp::LanguageServer::full_capabilities(),
1834 initializer: None,
1835 disk_based_diagnostics_progress_token: None,
1836 initialization_options: None,
1837 disk_based_diagnostics_sources: Vec::new(),
1838 prettier_plugins: Vec::new(),
1839 language_server_binary: LanguageServerBinary {
1840 path: "/the/fake/lsp/path".into(),
1841 arguments: vec![],
1842 env: Default::default(),
1843 },
1844 label_for_completion: None,
1845 }
1846 }
1847}
1848
1849#[cfg(any(test, feature = "test-support"))]
1850#[async_trait(?Send)]
1851impl LspAdapter for FakeLspAdapter {
1852 fn name(&self) -> LanguageServerName {
1853 LanguageServerName(self.name.into())
1854 }
1855
1856 async fn check_if_user_installed(
1857 &self,
1858 _: &dyn LspAdapterDelegate,
1859 _: Arc<dyn LanguageToolchainStore>,
1860 _: &AsyncApp,
1861 ) -> Option<LanguageServerBinary> {
1862 Some(self.language_server_binary.clone())
1863 }
1864
1865 fn get_language_server_command<'a>(
1866 self: Arc<Self>,
1867 _: Arc<dyn LspAdapterDelegate>,
1868 _: Arc<dyn LanguageToolchainStore>,
1869 _: LanguageServerBinaryOptions,
1870 _: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
1871 _: &'a mut AsyncApp,
1872 ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
1873 async move { Ok(self.language_server_binary.clone()) }.boxed_local()
1874 }
1875
1876 async fn fetch_latest_server_version(
1877 &self,
1878 _: &dyn LspAdapterDelegate,
1879 ) -> Result<Box<dyn 'static + Send + Any>> {
1880 unreachable!();
1881 }
1882
1883 async fn fetch_server_binary(
1884 &self,
1885 _: Box<dyn 'static + Send + Any>,
1886 _: PathBuf,
1887 _: &dyn LspAdapterDelegate,
1888 ) -> Result<LanguageServerBinary> {
1889 unreachable!();
1890 }
1891
1892 async fn cached_server_binary(
1893 &self,
1894 _: PathBuf,
1895 _: &dyn LspAdapterDelegate,
1896 ) -> Option<LanguageServerBinary> {
1897 unreachable!();
1898 }
1899
1900 fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
1901
1902 fn disk_based_diagnostic_sources(&self) -> Vec<String> {
1903 self.disk_based_diagnostics_sources.clone()
1904 }
1905
1906 fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
1907 self.disk_based_diagnostics_progress_token.clone()
1908 }
1909
1910 async fn initialization_options(
1911 self: Arc<Self>,
1912 _: &dyn Fs,
1913 _: &Arc<dyn LspAdapterDelegate>,
1914 ) -> Result<Option<Value>> {
1915 Ok(self.initialization_options.clone())
1916 }
1917
1918 async fn label_for_completion(
1919 &self,
1920 item: &lsp::CompletionItem,
1921 language: &Arc<Language>,
1922 ) -> Option<CodeLabel> {
1923 let label_for_completion = self.label_for_completion.as_ref()?;
1924 label_for_completion(item, language)
1925 }
1926}
1927
1928fn get_capture_indices(query: &Query, captures: &mut [(&str, &mut Option<u32>)]) {
1929 for (ix, name) in query.capture_names().iter().enumerate() {
1930 for (capture_name, index) in captures.iter_mut() {
1931 if capture_name == name {
1932 **index = Some(ix as u32);
1933 break;
1934 }
1935 }
1936 }
1937}
1938
1939pub fn point_to_lsp(point: PointUtf16) -> lsp::Position {
1940 lsp::Position::new(point.row, point.column)
1941}
1942
1943pub fn point_from_lsp(point: lsp::Position) -> Unclipped<PointUtf16> {
1944 Unclipped(PointUtf16::new(point.line, point.character))
1945}
1946
1947pub fn range_to_lsp(range: Range<PointUtf16>) -> Result<lsp::Range> {
1948 if range.start > range.end {
1949 Err(anyhow!(
1950 "Inverted range provided to an LSP request: {:?}-{:?}",
1951 range.start,
1952 range.end
1953 ))
1954 } else {
1955 Ok(lsp::Range {
1956 start: point_to_lsp(range.start),
1957 end: point_to_lsp(range.end),
1958 })
1959 }
1960}
1961
1962pub fn range_from_lsp(range: lsp::Range) -> Range<Unclipped<PointUtf16>> {
1963 let mut start = point_from_lsp(range.start);
1964 let mut end = point_from_lsp(range.end);
1965 if start > end {
1966 log::warn!("range_from_lsp called with inverted range {start:?}-{end:?}");
1967 mem::swap(&mut start, &mut end);
1968 }
1969 start..end
1970}
1971
1972#[cfg(test)]
1973mod tests {
1974 use super::*;
1975 use gpui::TestAppContext;
1976
1977 #[gpui::test(iterations = 10)]
1978 async fn test_language_loading(cx: &mut TestAppContext) {
1979 let languages = LanguageRegistry::test(cx.executor());
1980 let languages = Arc::new(languages);
1981 languages.register_native_grammars([
1982 ("json", tree_sitter_json::LANGUAGE),
1983 ("rust", tree_sitter_rust::LANGUAGE),
1984 ]);
1985 languages.register_test_language(LanguageConfig {
1986 name: "JSON".into(),
1987 grammar: Some("json".into()),
1988 matcher: LanguageMatcher {
1989 path_suffixes: vec!["json".into()],
1990 ..Default::default()
1991 },
1992 ..Default::default()
1993 });
1994 languages.register_test_language(LanguageConfig {
1995 name: "Rust".into(),
1996 grammar: Some("rust".into()),
1997 matcher: LanguageMatcher {
1998 path_suffixes: vec!["rs".into()],
1999 ..Default::default()
2000 },
2001 ..Default::default()
2002 });
2003 assert_eq!(
2004 languages.language_names(),
2005 &[
2006 "JSON".to_string(),
2007 "Plain Text".to_string(),
2008 "Rust".to_string(),
2009 ]
2010 );
2011
2012 let rust1 = languages.language_for_name("Rust");
2013 let rust2 = languages.language_for_name("Rust");
2014
2015 // Ensure language is still listed even if it's being loaded.
2016 assert_eq!(
2017 languages.language_names(),
2018 &[
2019 "JSON".to_string(),
2020 "Plain Text".to_string(),
2021 "Rust".to_string(),
2022 ]
2023 );
2024
2025 let (rust1, rust2) = futures::join!(rust1, rust2);
2026 assert!(Arc::ptr_eq(&rust1.unwrap(), &rust2.unwrap()));
2027
2028 // Ensure language is still listed even after loading it.
2029 assert_eq!(
2030 languages.language_names(),
2031 &[
2032 "JSON".to_string(),
2033 "Plain Text".to_string(),
2034 "Rust".to_string(),
2035 ]
2036 );
2037
2038 // Loading an unknown language returns an error.
2039 assert!(languages.language_for_name("Unknown").await.is_err());
2040 }
2041}