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