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