language.rs

   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;
  18
  19#[cfg(test)]
  20mod buffer_tests;
  21pub mod markdown;
  22
  23use crate::language_settings::SoftWrap;
  24use anyhow::{anyhow, Context, Result};
  25use async_trait::async_trait;
  26use collections::{HashMap, HashSet};
  27use futures::Future;
  28use gpui::{AppContext, AsyncAppContext, Model, SharedString, Task};
  29pub use highlight_map::HighlightMap;
  30use http_client::HttpClient;
  31use lazy_static::lazy_static;
  32use lsp::{CodeActionKind, LanguageServerBinary};
  33use parking_lot::Mutex;
  34use regex::Regex;
  35use schemars::{
  36    gen::SchemaGenerator,
  37    schema::{InstanceType, Schema, SchemaObject},
  38    JsonSchema,
  39};
  40use serde::{de, Deserialize, Deserializer, Serialize, Serializer};
  41use serde_json::Value;
  42use smol::future::FutureExt as _;
  43use std::num::NonZeroU32;
  44use std::{
  45    any::Any,
  46    ffi::OsStr,
  47    fmt::Debug,
  48    hash::Hash,
  49    mem,
  50    ops::{DerefMut, Range},
  51    path::{Path, PathBuf},
  52    pin::Pin,
  53    str,
  54    sync::{
  55        atomic::{AtomicU64, AtomicUsize, Ordering::SeqCst},
  56        Arc,
  57    },
  58};
  59use syntax_map::{QueryCursorHandle, SyntaxSnapshot};
  60use task::RunnableTag;
  61pub use task_context::{ContextProvider, RunnableRange};
  62use theme::SyntaxTheme;
  63use tree_sitter::{self, wasmtime, Query, QueryCursor, WasmStore};
  64use util::serde::default_true;
  65
  66pub use buffer::Operation;
  67pub use buffer::*;
  68pub use diagnostic_set::DiagnosticEntry;
  69pub use language_registry::{
  70    LanguageNotFound, LanguageQueries, LanguageRegistry, LanguageServerBinaryStatus,
  71    PendingLanguageServer, QUERY_FILENAME_PREFIXES,
  72};
  73pub use lsp::LanguageServerId;
  74pub use outline::{render_item, Outline, OutlineItem};
  75pub use syntax_map::{OwnedSyntaxLayer, SyntaxLayer};
  76pub use text::{AnchorRangeExt, LineEnding};
  77pub use tree_sitter::{Node, Parser, Tree, TreeCursor};
  78
  79/// Initializes the `language` crate.
  80///
  81/// This should be called before making use of items from the create.
  82pub fn init(cx: &mut AppContext) {
  83    language_settings::init(cx);
  84}
  85
  86static QUERY_CURSORS: Mutex<Vec<QueryCursor>> = Mutex::new(vec![]);
  87static PARSERS: Mutex<Vec<Parser>> = Mutex::new(vec![]);
  88
  89pub fn with_parser<F, R>(func: F) -> R
  90where
  91    F: FnOnce(&mut Parser) -> R,
  92{
  93    let mut parser = PARSERS.lock().pop().unwrap_or_else(|| {
  94        let mut parser = Parser::new();
  95        parser
  96            .set_wasm_store(WasmStore::new(WASM_ENGINE.clone()).unwrap())
  97            .unwrap();
  98        parser
  99    });
 100    parser.set_included_ranges(&[]).unwrap();
 101    let result = func(&mut parser);
 102    PARSERS.lock().push(parser);
 103    result
 104}
 105
 106pub fn with_query_cursor<F, R>(func: F) -> R
 107where
 108    F: FnOnce(&mut QueryCursor) -> R,
 109{
 110    let mut cursor = QueryCursorHandle::new();
 111    func(cursor.deref_mut())
 112}
 113
 114lazy_static! {
 115    static ref NEXT_LANGUAGE_ID: AtomicUsize = Default::default();
 116    static ref NEXT_GRAMMAR_ID: AtomicUsize = Default::default();
 117    static ref WASM_ENGINE: wasmtime::Engine = {
 118        wasmtime::Engine::new(&wasmtime::Config::new()).unwrap()
 119    };
 120
 121    /// A shared grammar for plain text, exposed for reuse by downstream crates.
 122    pub static ref PLAIN_TEXT: Arc<Language> = Arc::new(Language::new(
 123        LanguageConfig {
 124            name: "Plain Text".into(),
 125            soft_wrap: Some(SoftWrap::EditorWidth),
 126            ..Default::default()
 127        },
 128        None,
 129    ));
 130}
 131
 132/// Types that represent a position in a buffer, and can be converted into
 133/// an LSP position, to send to a language server.
 134pub trait ToLspPosition {
 135    /// Converts the value into an LSP position.
 136    fn to_lsp_position(self) -> lsp::Position;
 137}
 138
 139/// A name of a language server.
 140#[derive(Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Deserialize, Serialize)]
 141pub struct LanguageServerName(pub Arc<str>);
 142
 143#[derive(Debug, Clone, PartialEq, Eq, Hash)]
 144pub struct Location {
 145    pub buffer: Model<Buffer>,
 146    pub range: Range<Anchor>,
 147}
 148
 149/// Represents a Language Server, with certain cached sync properties.
 150/// Uses [`LspAdapter`] under the hood, but calls all 'static' methods
 151/// once at startup, and caches the results.
 152pub struct CachedLspAdapter {
 153    pub name: LanguageServerName,
 154    pub disk_based_diagnostic_sources: Vec<String>,
 155    pub disk_based_diagnostics_progress_token: Option<String>,
 156    language_ids: HashMap<String, String>,
 157    pub adapter: Arc<dyn LspAdapter>,
 158    pub reinstall_attempt_count: AtomicU64,
 159    /// Indicates whether this language server is the primary language server
 160    /// for a given language. Currently, most LSP-backed features only work
 161    /// with one language server, so one server needs to be primary.
 162    pub is_primary: bool,
 163    cached_binary: futures::lock::Mutex<Option<LanguageServerBinary>>,
 164}
 165
 166impl CachedLspAdapter {
 167    pub fn new(adapter: Arc<dyn LspAdapter>, is_primary: bool) -> Arc<Self> {
 168        let name = adapter.name();
 169        let disk_based_diagnostic_sources = adapter.disk_based_diagnostic_sources();
 170        let disk_based_diagnostics_progress_token = adapter.disk_based_diagnostics_progress_token();
 171        let language_ids = adapter.language_ids();
 172
 173        Arc::new(CachedLspAdapter {
 174            name,
 175            disk_based_diagnostic_sources,
 176            disk_based_diagnostics_progress_token,
 177            language_ids,
 178            adapter,
 179            is_primary,
 180            cached_binary: Default::default(),
 181            reinstall_attempt_count: AtomicU64::new(0),
 182        })
 183    }
 184
 185    pub async fn get_language_server_command(
 186        self: Arc<Self>,
 187        language: Arc<Language>,
 188        container_dir: Arc<Path>,
 189        delegate: Arc<dyn LspAdapterDelegate>,
 190        cx: &mut AsyncAppContext,
 191    ) -> Result<LanguageServerBinary> {
 192        let cached_binary = self.cached_binary.lock().await;
 193        self.adapter
 194            .clone()
 195            .get_language_server_command(language, container_dir, delegate, cached_binary, cx)
 196            .await
 197    }
 198
 199    pub fn will_start_server(
 200        &self,
 201        delegate: &Arc<dyn LspAdapterDelegate>,
 202        cx: &mut AsyncAppContext,
 203    ) -> Option<Task<Result<()>>> {
 204        self.adapter.will_start_server(delegate, cx)
 205    }
 206
 207    pub fn can_be_reinstalled(&self) -> bool {
 208        self.adapter.can_be_reinstalled()
 209    }
 210
 211    pub async fn installation_test_binary(
 212        &self,
 213        container_dir: PathBuf,
 214    ) -> Option<LanguageServerBinary> {
 215        self.adapter.installation_test_binary(container_dir).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: &Language) -> String {
 253        self.language_ids
 254            .get(language.name().as_ref())
 255            .cloned()
 256            .unwrap_or_else(|| language.lsp_id())
 257    }
 258
 259    #[cfg(any(test, feature = "test-support"))]
 260    fn as_fake(&self) -> Option<&FakeLspAdapter> {
 261        self.adapter.as_fake()
 262    }
 263}
 264
 265/// [`LspAdapterDelegate`] allows [`LspAdapter]` implementations to interface with the application
 266// e.g. to display a notification or fetch data from the web.
 267#[async_trait]
 268pub trait LspAdapterDelegate: Send + Sync {
 269    fn show_notification(&self, message: &str, cx: &mut AppContext);
 270    fn http_client(&self) -> Arc<dyn HttpClient>;
 271    fn worktree_id(&self) -> u64;
 272    fn worktree_root_path(&self) -> &Path;
 273    fn update_status(&self, language: LanguageServerName, status: LanguageServerBinaryStatus);
 274
 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}
 279
 280#[async_trait(?Send)]
 281pub trait LspAdapter: 'static + Send + Sync {
 282    fn name(&self) -> LanguageServerName;
 283
 284    fn get_language_server_command<'a>(
 285        self: Arc<Self>,
 286        language: Arc<Language>,
 287        container_dir: Arc<Path>,
 288        delegate: Arc<dyn LspAdapterDelegate>,
 289        mut cached_binary: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
 290        cx: &'a mut AsyncAppContext,
 291    ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
 292        async move {
 293            // First we check whether the adapter can give us a user-installed binary.
 294            // If so, we do *not* want to cache that, because each worktree might give us a different
 295            // binary:
 296            //
 297            //      worktree 1: user-installed at `.bin/gopls`
 298            //      worktree 2: user-installed at `~/bin/gopls`
 299            //      worktree 3: no gopls found in PATH -> fallback to Zed installation
 300            //
 301            // We only want to cache when we fall back to the global one,
 302            // because we don't want to download and overwrite our global one
 303            // for each worktree we might have open.
 304            if let Some(binary) = self.check_if_user_installed(delegate.as_ref(), cx).await {
 305                log::info!(
 306                    "found user-installed language server for {}. path: {:?}, arguments: {:?}",
 307                    language.name(),
 308                    binary.path,
 309                    binary.arguments
 310                );
 311                return Ok(binary);
 312            }
 313
 314            if let Some(cached_binary) = cached_binary.as_ref() {
 315                return Ok(cached_binary.clone());
 316            }
 317
 318            if !container_dir.exists() {
 319                smol::fs::create_dir_all(&container_dir)
 320                    .await
 321                    .context("failed to create container directory")?;
 322            }
 323
 324            let mut binary = try_fetch_server_binary(self.as_ref(), &delegate, container_dir.to_path_buf(), cx).await;
 325
 326            if let Err(error) = binary.as_ref() {
 327                if let Some(prev_downloaded_binary) = self
 328                    .cached_server_binary(container_dir.to_path_buf(), delegate.as_ref())
 329                    .await
 330                {
 331                    log::info!(
 332                        "failed to fetch newest version of language server {:?}. falling back to using {:?}",
 333                        self.name(),
 334                        prev_downloaded_binary.path
 335                    );
 336                    binary = Ok(prev_downloaded_binary);
 337                } else {
 338                    delegate.update_status(
 339                        self.name(),
 340                        LanguageServerBinaryStatus::Failed {
 341                            error: format!("{error:?}"),
 342                        },
 343                    );
 344                }
 345            }
 346
 347            if let Ok(binary) = &binary {
 348                *cached_binary = Some(binary.clone());
 349            }
 350
 351            binary
 352        }
 353        .boxed_local()
 354    }
 355
 356    async fn check_if_user_installed(
 357        &self,
 358        _: &dyn LspAdapterDelegate,
 359        _: &AsyncAppContext,
 360    ) -> Option<LanguageServerBinary> {
 361        None
 362    }
 363
 364    async fn fetch_latest_server_version(
 365        &self,
 366        delegate: &dyn LspAdapterDelegate,
 367    ) -> Result<Box<dyn 'static + Send + Any>>;
 368
 369    fn will_fetch_server(
 370        &self,
 371        _: &Arc<dyn LspAdapterDelegate>,
 372        _: &mut AsyncAppContext,
 373    ) -> Option<Task<Result<()>>> {
 374        None
 375    }
 376
 377    fn will_start_server(
 378        &self,
 379        _: &Arc<dyn LspAdapterDelegate>,
 380        _: &mut AsyncAppContext,
 381    ) -> Option<Task<Result<()>>> {
 382        None
 383    }
 384
 385    async fn fetch_server_binary(
 386        &self,
 387        latest_version: Box<dyn 'static + Send + Any>,
 388        container_dir: PathBuf,
 389        delegate: &dyn LspAdapterDelegate,
 390    ) -> Result<LanguageServerBinary>;
 391
 392    async fn cached_server_binary(
 393        &self,
 394        container_dir: PathBuf,
 395        delegate: &dyn LspAdapterDelegate,
 396    ) -> Option<LanguageServerBinary>;
 397
 398    /// Returns `true` if a language server can be reinstalled.
 399    ///
 400    /// If language server initialization fails, a reinstallation will be attempted unless the value returned from this method is `false`.
 401    ///
 402    /// Implementations that rely on software already installed on user's system
 403    /// should have [`can_be_reinstalled`](Self::can_be_reinstalled) return `false`.
 404    fn can_be_reinstalled(&self) -> bool {
 405        true
 406    }
 407
 408    async fn installation_test_binary(
 409        &self,
 410        container_dir: PathBuf,
 411    ) -> Option<LanguageServerBinary>;
 412
 413    fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
 414
 415    /// Post-processes completions provided by the language server.
 416    async fn process_completions(&self, _: &mut [lsp::CompletionItem]) {}
 417
 418    async fn labels_for_completions(
 419        self: Arc<Self>,
 420        completions: &[lsp::CompletionItem],
 421        language: &Arc<Language>,
 422    ) -> Result<Vec<Option<CodeLabel>>> {
 423        let mut labels = Vec::new();
 424        for (ix, completion) in completions.into_iter().enumerate() {
 425            let label = self.label_for_completion(completion, language).await;
 426            if let Some(label) = label {
 427                labels.resize(ix + 1, None);
 428                *labels.last_mut().unwrap() = Some(label);
 429            }
 430        }
 431        Ok(labels)
 432    }
 433
 434    async fn label_for_completion(
 435        &self,
 436        _: &lsp::CompletionItem,
 437        _: &Arc<Language>,
 438    ) -> Option<CodeLabel> {
 439        None
 440    }
 441
 442    async fn labels_for_symbols(
 443        self: Arc<Self>,
 444        symbols: &[(String, lsp::SymbolKind)],
 445        language: &Arc<Language>,
 446    ) -> Result<Vec<Option<CodeLabel>>> {
 447        let mut labels = Vec::new();
 448        for (ix, (name, kind)) in symbols.into_iter().enumerate() {
 449            let label = self.label_for_symbol(name, *kind, language).await;
 450            if let Some(label) = label {
 451                labels.resize(ix + 1, None);
 452                *labels.last_mut().unwrap() = Some(label);
 453            }
 454        }
 455        Ok(labels)
 456    }
 457
 458    async fn label_for_symbol(
 459        &self,
 460        _: &str,
 461        _: lsp::SymbolKind,
 462        _: &Arc<Language>,
 463    ) -> Option<CodeLabel> {
 464        None
 465    }
 466
 467    /// Returns initialization options that are going to be sent to a LSP server as a part of [`lsp::InitializeParams`]
 468    async fn initialization_options(
 469        self: Arc<Self>,
 470        _: &Arc<dyn LspAdapterDelegate>,
 471    ) -> Result<Option<Value>> {
 472        Ok(None)
 473    }
 474
 475    async fn workspace_configuration(
 476        self: Arc<Self>,
 477        _: &Arc<dyn LspAdapterDelegate>,
 478        _cx: &mut AsyncAppContext,
 479    ) -> Result<Value> {
 480        Ok(serde_json::json!({}))
 481    }
 482
 483    /// Returns a list of code actions supported by a given LspAdapter
 484    fn code_action_kinds(&self) -> Option<Vec<CodeActionKind>> {
 485        Some(vec![
 486            CodeActionKind::EMPTY,
 487            CodeActionKind::QUICKFIX,
 488            CodeActionKind::REFACTOR,
 489            CodeActionKind::REFACTOR_EXTRACT,
 490            CodeActionKind::SOURCE,
 491        ])
 492    }
 493
 494    fn disk_based_diagnostic_sources(&self) -> Vec<String> {
 495        Default::default()
 496    }
 497
 498    fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
 499        None
 500    }
 501
 502    fn language_ids(&self) -> HashMap<String, String> {
 503        Default::default()
 504    }
 505
 506    #[cfg(any(test, feature = "test-support"))]
 507    fn as_fake(&self) -> Option<&FakeLspAdapter> {
 508        None
 509    }
 510}
 511
 512async fn try_fetch_server_binary<L: LspAdapter + 'static + Send + Sync + ?Sized>(
 513    adapter: &L,
 514    delegate: &Arc<dyn LspAdapterDelegate>,
 515    container_dir: PathBuf,
 516    cx: &mut AsyncAppContext,
 517) -> Result<LanguageServerBinary> {
 518    if let Some(task) = adapter.will_fetch_server(delegate, cx) {
 519        task.await?;
 520    }
 521
 522    let name = adapter.name();
 523    log::info!("fetching latest version of language server {:?}", name.0);
 524    delegate.update_status(name.clone(), LanguageServerBinaryStatus::CheckingForUpdate);
 525    let latest_version = adapter
 526        .fetch_latest_server_version(delegate.as_ref())
 527        .await?;
 528
 529    log::info!("downloading language server {:?}", name.0);
 530    delegate.update_status(adapter.name(), LanguageServerBinaryStatus::Downloading);
 531    let binary = adapter
 532        .fetch_server_binary(latest_version, container_dir, delegate.as_ref())
 533        .await;
 534
 535    delegate.update_status(name.clone(), LanguageServerBinaryStatus::None);
 536    binary
 537}
 538
 539#[derive(Clone, Debug, Default, PartialEq, Eq)]
 540pub struct CodeLabel {
 541    /// The text to display.
 542    pub text: String,
 543    /// Syntax highlighting runs.
 544    pub runs: Vec<(Range<usize>, HighlightId)>,
 545    /// The portion of the text that should be used in fuzzy filtering.
 546    pub filter_range: Range<usize>,
 547}
 548
 549#[derive(Clone, Deserialize, JsonSchema)]
 550pub struct LanguageConfig {
 551    /// Human-readable name of the language.
 552    pub name: Arc<str>,
 553    /// The name of this language for a Markdown code fence block
 554    pub code_fence_block_name: Option<Arc<str>>,
 555    // The name of the grammar in a WASM bundle (experimental).
 556    pub grammar: Option<Arc<str>>,
 557    /// The criteria for matching this language to a given file.
 558    #[serde(flatten)]
 559    pub matcher: LanguageMatcher,
 560    /// List of bracket types in a language.
 561    #[serde(default)]
 562    #[schemars(schema_with = "bracket_pair_config_json_schema")]
 563    pub brackets: BracketPairConfig,
 564    /// If set to true, auto indentation uses last non empty line to determine
 565    /// the indentation level for a new line.
 566    #[serde(default = "auto_indent_using_last_non_empty_line_default")]
 567    pub auto_indent_using_last_non_empty_line: bool,
 568    /// A regex that is used to determine whether the indentation level should be
 569    /// increased in the following line.
 570    #[serde(default, deserialize_with = "deserialize_regex")]
 571    #[schemars(schema_with = "regex_json_schema")]
 572    pub increase_indent_pattern: Option<Regex>,
 573    /// A regex that is used to determine whether the indentation level should be
 574    /// decreased in the following line.
 575    #[serde(default, deserialize_with = "deserialize_regex")]
 576    #[schemars(schema_with = "regex_json_schema")]
 577    pub decrease_indent_pattern: Option<Regex>,
 578    /// A list of characters that trigger the automatic insertion of a closing
 579    /// bracket when they immediately precede the point where an opening
 580    /// bracket is inserted.
 581    #[serde(default)]
 582    pub autoclose_before: String,
 583    /// A placeholder used internally by Semantic Index.
 584    #[serde(default)]
 585    pub collapsed_placeholder: String,
 586    /// A line comment string that is inserted in e.g. `toggle comments` action.
 587    /// A language can have multiple flavours of line comments. All of the provided line comments are
 588    /// used for comment continuations on the next line, but only the first one is used for Editor::ToggleComments.
 589    #[serde(default)]
 590    pub line_comments: Vec<Arc<str>>,
 591    /// Starting and closing characters of a block comment.
 592    #[serde(default)]
 593    pub block_comment: Option<(Arc<str>, Arc<str>)>,
 594    /// A list of language servers that are allowed to run on subranges of a given language.
 595    #[serde(default)]
 596    pub scope_opt_in_language_servers: Vec<String>,
 597    #[serde(default)]
 598    pub overrides: HashMap<String, LanguageConfigOverride>,
 599    /// A list of characters that Zed should treat as word characters for the
 600    /// purpose of features that operate on word boundaries, like 'move to next word end'
 601    /// or a whole-word search in buffer search.
 602    #[serde(default)]
 603    pub word_characters: HashSet<char>,
 604    /// Whether to indent lines using tab characters, as opposed to multiple
 605    /// spaces.
 606    #[serde(default)]
 607    pub hard_tabs: Option<bool>,
 608    /// How many columns a tab should occupy.
 609    #[serde(default)]
 610    pub tab_size: Option<NonZeroU32>,
 611    /// How to soft-wrap long lines of text.
 612    #[serde(default)]
 613    pub soft_wrap: Option<SoftWrap>,
 614    /// The name of a Prettier parser that will be used for this language when no file path is available.
 615    /// If there's a parser name in the language settings, that will be used instead.
 616    #[serde(default)]
 617    pub prettier_parser_name: Option<String>,
 618}
 619
 620#[derive(Clone, Debug, Serialize, Deserialize, Default, JsonSchema)]
 621pub struct LanguageMatcher {
 622    /// 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`.
 623    #[serde(default)]
 624    pub path_suffixes: Vec<String>,
 625    /// A regex pattern that determines whether the language should be assigned to a file or not.
 626    #[serde(
 627        default,
 628        serialize_with = "serialize_regex",
 629        deserialize_with = "deserialize_regex"
 630    )]
 631    #[schemars(schema_with = "regex_json_schema")]
 632    pub first_line_pattern: Option<Regex>,
 633}
 634
 635/// Represents a language for the given range. Some languages (e.g. HTML)
 636/// interleave several languages together, thus a single buffer might actually contain
 637/// several nested scopes.
 638#[derive(Clone, Debug)]
 639pub struct LanguageScope {
 640    language: Arc<Language>,
 641    override_id: Option<u32>,
 642}
 643
 644#[derive(Clone, Deserialize, Default, Debug, JsonSchema)]
 645pub struct LanguageConfigOverride {
 646    #[serde(default)]
 647    pub line_comments: Override<Vec<Arc<str>>>,
 648    #[serde(default)]
 649    pub block_comment: Override<(Arc<str>, Arc<str>)>,
 650    #[serde(skip_deserializing)]
 651    #[schemars(skip)]
 652    pub disabled_bracket_ixs: Vec<u16>,
 653    #[serde(default)]
 654    pub word_characters: Override<HashSet<char>>,
 655    #[serde(default)]
 656    pub opt_into_language_servers: Vec<String>,
 657}
 658
 659#[derive(Clone, Deserialize, Debug, Serialize, JsonSchema)]
 660#[serde(untagged)]
 661pub enum Override<T> {
 662    Remove { remove: bool },
 663    Set(T),
 664}
 665
 666impl<T> Default for Override<T> {
 667    fn default() -> Self {
 668        Override::Remove { remove: false }
 669    }
 670}
 671
 672impl<T> Override<T> {
 673    fn as_option<'a>(this: Option<&'a Self>, original: Option<&'a T>) -> Option<&'a T> {
 674        match this {
 675            Some(Self::Set(value)) => Some(value),
 676            Some(Self::Remove { remove: true }) => None,
 677            Some(Self::Remove { remove: false }) | None => original,
 678        }
 679    }
 680}
 681
 682impl Default for LanguageConfig {
 683    fn default() -> Self {
 684        Self {
 685            name: "".into(),
 686            code_fence_block_name: None,
 687            grammar: None,
 688            matcher: LanguageMatcher::default(),
 689            brackets: Default::default(),
 690            auto_indent_using_last_non_empty_line: auto_indent_using_last_non_empty_line_default(),
 691            increase_indent_pattern: Default::default(),
 692            decrease_indent_pattern: Default::default(),
 693            autoclose_before: Default::default(),
 694            line_comments: Default::default(),
 695            block_comment: Default::default(),
 696            scope_opt_in_language_servers: Default::default(),
 697            overrides: Default::default(),
 698            word_characters: Default::default(),
 699            collapsed_placeholder: Default::default(),
 700            hard_tabs: None,
 701            tab_size: None,
 702            soft_wrap: None,
 703            prettier_parser_name: None,
 704        }
 705    }
 706}
 707
 708fn auto_indent_using_last_non_empty_line_default() -> bool {
 709    true
 710}
 711
 712fn deserialize_regex<'de, D: Deserializer<'de>>(d: D) -> Result<Option<Regex>, D::Error> {
 713    let source = Option::<String>::deserialize(d)?;
 714    if let Some(source) = source {
 715        Ok(Some(regex::Regex::new(&source).map_err(de::Error::custom)?))
 716    } else {
 717        Ok(None)
 718    }
 719}
 720
 721fn regex_json_schema(_: &mut SchemaGenerator) -> Schema {
 722    Schema::Object(SchemaObject {
 723        instance_type: Some(InstanceType::String.into()),
 724        ..Default::default()
 725    })
 726}
 727
 728fn serialize_regex<S>(regex: &Option<Regex>, serializer: S) -> Result<S::Ok, S::Error>
 729where
 730    S: Serializer,
 731{
 732    match regex {
 733        Some(regex) => serializer.serialize_str(regex.as_str()),
 734        None => serializer.serialize_none(),
 735    }
 736}
 737
 738#[doc(hidden)]
 739#[cfg(any(test, feature = "test-support"))]
 740pub struct FakeLspAdapter {
 741    pub name: &'static str,
 742    pub initialization_options: Option<Value>,
 743    pub capabilities: lsp::ServerCapabilities,
 744    pub initializer: Option<Box<dyn 'static + Send + Sync + Fn(&mut lsp::FakeLanguageServer)>>,
 745    pub disk_based_diagnostics_progress_token: Option<String>,
 746    pub disk_based_diagnostics_sources: Vec<String>,
 747    pub prettier_plugins: Vec<&'static str>,
 748    pub language_server_binary: LanguageServerBinary,
 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}
 830
 831#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Clone, Copy)]
 832pub struct GrammarId(pub usize);
 833
 834impl GrammarId {
 835    pub(crate) fn new() -> Self {
 836        Self(NEXT_GRAMMAR_ID.fetch_add(1, SeqCst))
 837    }
 838}
 839
 840pub struct Grammar {
 841    id: GrammarId,
 842    pub ts_language: tree_sitter::Language,
 843    pub(crate) error_query: Query,
 844    pub(crate) highlights_query: Option<Query>,
 845    pub(crate) brackets_config: Option<BracketConfig>,
 846    pub(crate) redactions_config: Option<RedactionConfig>,
 847    pub(crate) runnable_config: Option<RunnableConfig>,
 848    pub(crate) indents_config: Option<IndentConfig>,
 849    pub outline_config: Option<OutlineConfig>,
 850    pub embedding_config: Option<EmbeddingConfig>,
 851    pub(crate) injection_config: Option<InjectionConfig>,
 852    pub(crate) override_config: Option<OverrideConfig>,
 853    pub(crate) highlight_map: Mutex<HighlightMap>,
 854}
 855
 856struct IndentConfig {
 857    query: Query,
 858    indent_capture_ix: u32,
 859    start_capture_ix: Option<u32>,
 860    end_capture_ix: Option<u32>,
 861    outdent_capture_ix: Option<u32>,
 862}
 863
 864pub struct OutlineConfig {
 865    pub query: Query,
 866    pub item_capture_ix: u32,
 867    pub name_capture_ix: u32,
 868    pub context_capture_ix: Option<u32>,
 869    pub extra_context_capture_ix: Option<u32>,
 870    pub open_capture_ix: Option<u32>,
 871    pub close_capture_ix: Option<u32>,
 872}
 873
 874#[derive(Debug)]
 875pub struct EmbeddingConfig {
 876    pub query: Query,
 877    pub item_capture_ix: u32,
 878    pub name_capture_ix: Option<u32>,
 879    pub context_capture_ix: Option<u32>,
 880    pub collapse_capture_ix: Option<u32>,
 881    pub keep_capture_ix: Option<u32>,
 882}
 883
 884struct InjectionConfig {
 885    query: Query,
 886    content_capture_ix: u32,
 887    language_capture_ix: Option<u32>,
 888    patterns: Vec<InjectionPatternConfig>,
 889}
 890
 891struct RedactionConfig {
 892    pub query: Query,
 893    pub redaction_capture_ix: u32,
 894}
 895
 896#[derive(Clone, Debug, PartialEq)]
 897enum RunnableCapture {
 898    Named(SharedString),
 899    Run,
 900}
 901
 902struct RunnableConfig {
 903    pub query: Query,
 904    /// A mapping from capture indice to capture kind
 905    pub extra_captures: Vec<RunnableCapture>,
 906}
 907
 908struct OverrideConfig {
 909    query: Query,
 910    values: HashMap<u32, (String, LanguageConfigOverride)>,
 911}
 912
 913#[derive(Default, Clone)]
 914struct InjectionPatternConfig {
 915    language: Option<Box<str>>,
 916    combined: bool,
 917}
 918
 919struct BracketConfig {
 920    query: Query,
 921    open_capture_ix: u32,
 922    close_capture_ix: u32,
 923}
 924
 925impl Language {
 926    pub fn new(config: LanguageConfig, ts_language: Option<tree_sitter::Language>) -> Self {
 927        Self::new_with_id(LanguageId::new(), config, ts_language)
 928    }
 929
 930    fn new_with_id(
 931        id: LanguageId,
 932        config: LanguageConfig,
 933        ts_language: Option<tree_sitter::Language>,
 934    ) -> Self {
 935        Self {
 936            id,
 937            config,
 938            grammar: ts_language.map(|ts_language| {
 939                Arc::new(Grammar {
 940                    id: GrammarId::new(),
 941                    highlights_query: None,
 942                    brackets_config: None,
 943                    outline_config: None,
 944                    embedding_config: None,
 945                    indents_config: None,
 946                    injection_config: None,
 947                    override_config: None,
 948                    redactions_config: None,
 949                    runnable_config: None,
 950                    error_query: Query::new(&ts_language, "(ERROR) @error").unwrap(),
 951                    ts_language,
 952                    highlight_map: Default::default(),
 953                })
 954            }),
 955            context_provider: None,
 956        }
 957    }
 958
 959    pub fn with_context_provider(mut self, provider: Option<Arc<dyn ContextProvider>>) -> Self {
 960        self.context_provider = provider;
 961        self
 962    }
 963
 964    pub fn with_queries(mut self, queries: LanguageQueries) -> Result<Self> {
 965        if let Some(query) = queries.highlights {
 966            self = self
 967                .with_highlights_query(query.as_ref())
 968                .context("Error loading highlights query")?;
 969        }
 970        if let Some(query) = queries.brackets {
 971            self = self
 972                .with_brackets_query(query.as_ref())
 973                .context("Error loading brackets query")?;
 974        }
 975        if let Some(query) = queries.indents {
 976            self = self
 977                .with_indents_query(query.as_ref())
 978                .context("Error loading indents query")?;
 979        }
 980        if let Some(query) = queries.outline {
 981            self = self
 982                .with_outline_query(query.as_ref())
 983                .context("Error loading outline query")?;
 984        }
 985        if let Some(query) = queries.embedding {
 986            self = self
 987                .with_embedding_query(query.as_ref())
 988                .context("Error loading embedding query")?;
 989        }
 990        if let Some(query) = queries.injections {
 991            self = self
 992                .with_injection_query(query.as_ref())
 993                .context("Error loading injection query")?;
 994        }
 995        if let Some(query) = queries.overrides {
 996            self = self
 997                .with_override_query(query.as_ref())
 998                .context("Error loading override query")?;
 999        }
1000        if let Some(query) = queries.redactions {
1001            self = self
1002                .with_redaction_query(query.as_ref())
1003                .context("Error loading redaction query")?;
1004        }
1005        if let Some(query) = queries.runnables {
1006            self = self
1007                .with_runnable_query(query.as_ref())
1008                .context("Error loading tests query")?;
1009        }
1010        Ok(self)
1011    }
1012
1013    pub fn with_highlights_query(mut self, source: &str) -> Result<Self> {
1014        let grammar = self
1015            .grammar_mut()
1016            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1017        grammar.highlights_query = Some(Query::new(&grammar.ts_language, source)?);
1018        Ok(self)
1019    }
1020
1021    pub fn with_runnable_query(mut self, source: &str) -> Result<Self> {
1022        let grammar = self
1023            .grammar_mut()
1024            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1025
1026        let query = Query::new(&grammar.ts_language, source)?;
1027        let mut extra_captures = Vec::with_capacity(query.capture_names().len());
1028
1029        for name in query.capture_names().iter() {
1030            let kind = if *name == "run" {
1031                RunnableCapture::Run
1032            } else {
1033                RunnableCapture::Named(name.to_string().into())
1034            };
1035            extra_captures.push(kind);
1036        }
1037
1038        grammar.runnable_config = Some(RunnableConfig {
1039            extra_captures,
1040            query,
1041        });
1042
1043        Ok(self)
1044    }
1045
1046    pub fn with_outline_query(mut self, source: &str) -> Result<Self> {
1047        let grammar = self
1048            .grammar_mut()
1049            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1050        let query = Query::new(&grammar.ts_language, source)?;
1051        let mut item_capture_ix = None;
1052        let mut name_capture_ix = None;
1053        let mut context_capture_ix = None;
1054        let mut extra_context_capture_ix = None;
1055        let mut open_capture_ix = None;
1056        let mut close_capture_ix = None;
1057        get_capture_indices(
1058            &query,
1059            &mut [
1060                ("item", &mut item_capture_ix),
1061                ("name", &mut name_capture_ix),
1062                ("context", &mut context_capture_ix),
1063                ("context.extra", &mut extra_context_capture_ix),
1064                ("open", &mut open_capture_ix),
1065                ("close", &mut close_capture_ix),
1066            ],
1067        );
1068        if let Some((item_capture_ix, name_capture_ix)) = item_capture_ix.zip(name_capture_ix) {
1069            grammar.outline_config = Some(OutlineConfig {
1070                query,
1071                item_capture_ix,
1072                name_capture_ix,
1073                context_capture_ix,
1074                extra_context_capture_ix,
1075                open_capture_ix,
1076                close_capture_ix,
1077            });
1078        }
1079        Ok(self)
1080    }
1081
1082    pub fn with_embedding_query(mut self, source: &str) -> Result<Self> {
1083        let grammar = self
1084            .grammar_mut()
1085            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1086        let query = Query::new(&grammar.ts_language, source)?;
1087        let mut item_capture_ix = None;
1088        let mut name_capture_ix = None;
1089        let mut context_capture_ix = None;
1090        let mut collapse_capture_ix = None;
1091        let mut keep_capture_ix = None;
1092        get_capture_indices(
1093            &query,
1094            &mut [
1095                ("item", &mut item_capture_ix),
1096                ("name", &mut name_capture_ix),
1097                ("context", &mut context_capture_ix),
1098                ("keep", &mut keep_capture_ix),
1099                ("collapse", &mut collapse_capture_ix),
1100            ],
1101        );
1102        if let Some(item_capture_ix) = item_capture_ix {
1103            grammar.embedding_config = Some(EmbeddingConfig {
1104                query,
1105                item_capture_ix,
1106                name_capture_ix,
1107                context_capture_ix,
1108                collapse_capture_ix,
1109                keep_capture_ix,
1110            });
1111        }
1112        Ok(self)
1113    }
1114
1115    pub fn with_brackets_query(mut self, source: &str) -> Result<Self> {
1116        let grammar = self
1117            .grammar_mut()
1118            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1119        let query = Query::new(&grammar.ts_language, source)?;
1120        let mut open_capture_ix = None;
1121        let mut close_capture_ix = None;
1122        get_capture_indices(
1123            &query,
1124            &mut [
1125                ("open", &mut open_capture_ix),
1126                ("close", &mut close_capture_ix),
1127            ],
1128        );
1129        if let Some((open_capture_ix, close_capture_ix)) = open_capture_ix.zip(close_capture_ix) {
1130            grammar.brackets_config = Some(BracketConfig {
1131                query,
1132                open_capture_ix,
1133                close_capture_ix,
1134            });
1135        }
1136        Ok(self)
1137    }
1138
1139    pub fn with_indents_query(mut self, source: &str) -> Result<Self> {
1140        let grammar = self
1141            .grammar_mut()
1142            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1143        let query = Query::new(&grammar.ts_language, source)?;
1144        let mut indent_capture_ix = None;
1145        let mut start_capture_ix = None;
1146        let mut end_capture_ix = None;
1147        let mut outdent_capture_ix = None;
1148        get_capture_indices(
1149            &query,
1150            &mut [
1151                ("indent", &mut indent_capture_ix),
1152                ("start", &mut start_capture_ix),
1153                ("end", &mut end_capture_ix),
1154                ("outdent", &mut outdent_capture_ix),
1155            ],
1156        );
1157        if let Some(indent_capture_ix) = indent_capture_ix {
1158            grammar.indents_config = Some(IndentConfig {
1159                query,
1160                indent_capture_ix,
1161                start_capture_ix,
1162                end_capture_ix,
1163                outdent_capture_ix,
1164            });
1165        }
1166        Ok(self)
1167    }
1168
1169    pub fn with_injection_query(mut self, source: &str) -> Result<Self> {
1170        let grammar = self
1171            .grammar_mut()
1172            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1173        let query = Query::new(&grammar.ts_language, source)?;
1174        let mut language_capture_ix = None;
1175        let mut content_capture_ix = None;
1176        get_capture_indices(
1177            &query,
1178            &mut [
1179                ("language", &mut language_capture_ix),
1180                ("content", &mut content_capture_ix),
1181            ],
1182        );
1183        let patterns = (0..query.pattern_count())
1184            .map(|ix| {
1185                let mut config = InjectionPatternConfig::default();
1186                for setting in query.property_settings(ix) {
1187                    match setting.key.as_ref() {
1188                        "language" => {
1189                            config.language.clone_from(&setting.value);
1190                        }
1191                        "combined" => {
1192                            config.combined = true;
1193                        }
1194                        _ => {}
1195                    }
1196                }
1197                config
1198            })
1199            .collect();
1200        if let Some(content_capture_ix) = content_capture_ix {
1201            grammar.injection_config = Some(InjectionConfig {
1202                query,
1203                language_capture_ix,
1204                content_capture_ix,
1205                patterns,
1206            });
1207        }
1208        Ok(self)
1209    }
1210
1211    pub fn with_override_query(mut self, source: &str) -> anyhow::Result<Self> {
1212        let query = {
1213            let grammar = self
1214                .grammar
1215                .as_ref()
1216                .ok_or_else(|| anyhow!("no grammar for language"))?;
1217            Query::new(&grammar.ts_language, source)?
1218        };
1219
1220        let mut override_configs_by_id = HashMap::default();
1221        for (ix, name) in query.capture_names().iter().enumerate() {
1222            if !name.starts_with('_') {
1223                let value = self.config.overrides.remove(*name).unwrap_or_default();
1224                for server_name in &value.opt_into_language_servers {
1225                    if !self
1226                        .config
1227                        .scope_opt_in_language_servers
1228                        .contains(server_name)
1229                    {
1230                        util::debug_panic!("Server {server_name:?} has been opted-in by scope {name:?} but has not been marked as an opt-in server");
1231                    }
1232                }
1233
1234                override_configs_by_id.insert(ix as u32, (name.to_string(), value));
1235            }
1236        }
1237
1238        if !self.config.overrides.is_empty() {
1239            let keys = self.config.overrides.keys().collect::<Vec<_>>();
1240            Err(anyhow!(
1241                "language {:?} has overrides in config not in query: {keys:?}",
1242                self.config.name
1243            ))?;
1244        }
1245
1246        for disabled_scope_name in self
1247            .config
1248            .brackets
1249            .disabled_scopes_by_bracket_ix
1250            .iter()
1251            .flatten()
1252        {
1253            if !override_configs_by_id
1254                .values()
1255                .any(|(scope_name, _)| scope_name == disabled_scope_name)
1256            {
1257                Err(anyhow!(
1258                    "language {:?} has overrides in config not in query: {disabled_scope_name:?}",
1259                    self.config.name
1260                ))?;
1261            }
1262        }
1263
1264        for (name, override_config) in override_configs_by_id.values_mut() {
1265            override_config.disabled_bracket_ixs = self
1266                .config
1267                .brackets
1268                .disabled_scopes_by_bracket_ix
1269                .iter()
1270                .enumerate()
1271                .filter_map(|(ix, disabled_scope_names)| {
1272                    if disabled_scope_names.contains(name) {
1273                        Some(ix as u16)
1274                    } else {
1275                        None
1276                    }
1277                })
1278                .collect();
1279        }
1280
1281        self.config.brackets.disabled_scopes_by_bracket_ix.clear();
1282
1283        let grammar = self
1284            .grammar_mut()
1285            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1286        grammar.override_config = Some(OverrideConfig {
1287            query,
1288            values: override_configs_by_id,
1289        });
1290        Ok(self)
1291    }
1292
1293    pub fn with_redaction_query(mut self, source: &str) -> anyhow::Result<Self> {
1294        let grammar = self
1295            .grammar_mut()
1296            .ok_or_else(|| anyhow!("cannot mutate grammar"))?;
1297
1298        let query = Query::new(&grammar.ts_language, source)?;
1299        let mut redaction_capture_ix = None;
1300        get_capture_indices(&query, &mut [("redact", &mut redaction_capture_ix)]);
1301
1302        if let Some(redaction_capture_ix) = redaction_capture_ix {
1303            grammar.redactions_config = Some(RedactionConfig {
1304                query,
1305                redaction_capture_ix,
1306            });
1307        }
1308
1309        Ok(self)
1310    }
1311
1312    fn grammar_mut(&mut self) -> Option<&mut Grammar> {
1313        Arc::get_mut(self.grammar.as_mut()?)
1314    }
1315
1316    pub fn name(&self) -> Arc<str> {
1317        self.config.name.clone()
1318    }
1319
1320    pub fn code_fence_block_name(&self) -> Arc<str> {
1321        self.config
1322            .code_fence_block_name
1323            .clone()
1324            .unwrap_or_else(|| self.config.name.to_lowercase().into())
1325    }
1326
1327    pub fn context_provider(&self) -> Option<Arc<dyn ContextProvider>> {
1328        self.context_provider.clone()
1329    }
1330
1331    pub fn highlight_text<'a>(
1332        self: &'a Arc<Self>,
1333        text: &'a Rope,
1334        range: Range<usize>,
1335    ) -> Vec<(Range<usize>, HighlightId)> {
1336        let mut result = Vec::new();
1337        if let Some(grammar) = &self.grammar {
1338            let tree = grammar.parse_text(text, None);
1339            let captures =
1340                SyntaxSnapshot::single_tree_captures(range.clone(), text, &tree, self, |grammar| {
1341                    grammar.highlights_query.as_ref()
1342                });
1343            let highlight_maps = vec![grammar.highlight_map()];
1344            let mut offset = 0;
1345            for chunk in BufferChunks::new(text, range, Some((captures, highlight_maps)), vec![]) {
1346                let end_offset = offset + chunk.text.len();
1347                if let Some(highlight_id) = chunk.syntax_highlight_id {
1348                    if !highlight_id.is_default() {
1349                        result.push((offset..end_offset, highlight_id));
1350                    }
1351                }
1352                offset = end_offset;
1353            }
1354        }
1355        result
1356    }
1357
1358    pub fn path_suffixes(&self) -> &[String] {
1359        &self.config.matcher.path_suffixes
1360    }
1361
1362    pub fn should_autoclose_before(&self, c: char) -> bool {
1363        c.is_whitespace() || self.config.autoclose_before.contains(c)
1364    }
1365
1366    pub fn set_theme(&self, theme: &SyntaxTheme) {
1367        if let Some(grammar) = self.grammar.as_ref() {
1368            if let Some(highlights_query) = &grammar.highlights_query {
1369                *grammar.highlight_map.lock() =
1370                    HighlightMap::new(highlights_query.capture_names(), theme);
1371            }
1372        }
1373    }
1374
1375    pub fn grammar(&self) -> Option<&Arc<Grammar>> {
1376        self.grammar.as_ref()
1377    }
1378
1379    pub fn default_scope(self: &Arc<Self>) -> LanguageScope {
1380        LanguageScope {
1381            language: self.clone(),
1382            override_id: None,
1383        }
1384    }
1385
1386    pub fn lsp_id(&self) -> String {
1387        match self.config.name.as_ref() {
1388            "Plain Text" => "plaintext".to_string(),
1389            language_name => language_name.to_lowercase(),
1390        }
1391    }
1392
1393    pub fn prettier_parser_name(&self) -> Option<&str> {
1394        self.config.prettier_parser_name.as_deref()
1395    }
1396}
1397
1398impl LanguageScope {
1399    pub fn collapsed_placeholder(&self) -> &str {
1400        self.language.config.collapsed_placeholder.as_ref()
1401    }
1402
1403    /// Returns line prefix that is inserted in e.g. line continuations or
1404    /// in `toggle comments` action.
1405    pub fn line_comment_prefixes(&self) -> &[Arc<str>] {
1406        Override::as_option(
1407            self.config_override().map(|o| &o.line_comments),
1408            Some(&self.language.config.line_comments),
1409        )
1410        .map_or(&[] as &[_], |e| e.as_slice())
1411    }
1412
1413    pub fn block_comment_delimiters(&self) -> Option<(&Arc<str>, &Arc<str>)> {
1414        Override::as_option(
1415            self.config_override().map(|o| &o.block_comment),
1416            self.language.config.block_comment.as_ref(),
1417        )
1418        .map(|e| (&e.0, &e.1))
1419    }
1420
1421    /// Returns a list of language-specific word characters.
1422    ///
1423    /// By default, Zed treats alphanumeric characters (and '_') as word characters for
1424    /// the purpose of actions like 'move to next word end` or whole-word search.
1425    /// It additionally accounts for language's additional word characters.
1426    pub fn word_characters(&self) -> Option<&HashSet<char>> {
1427        Override::as_option(
1428            self.config_override().map(|o| &o.word_characters),
1429            Some(&self.language.config.word_characters),
1430        )
1431    }
1432
1433    /// Returns a list of bracket pairs for a given language with an additional
1434    /// piece of information about whether the particular bracket pair is currently active for a given language.
1435    pub fn brackets(&self) -> impl Iterator<Item = (&BracketPair, bool)> {
1436        let mut disabled_ids = self
1437            .config_override()
1438            .map_or(&[] as _, |o| o.disabled_bracket_ixs.as_slice());
1439        self.language
1440            .config
1441            .brackets
1442            .pairs
1443            .iter()
1444            .enumerate()
1445            .map(move |(ix, bracket)| {
1446                let mut is_enabled = true;
1447                if let Some(next_disabled_ix) = disabled_ids.first() {
1448                    if ix == *next_disabled_ix as usize {
1449                        disabled_ids = &disabled_ids[1..];
1450                        is_enabled = false;
1451                    }
1452                }
1453                (bracket, is_enabled)
1454            })
1455    }
1456
1457    pub fn should_autoclose_before(&self, c: char) -> bool {
1458        c.is_whitespace() || self.language.config.autoclose_before.contains(c)
1459    }
1460
1461    pub fn language_allowed(&self, name: &LanguageServerName) -> bool {
1462        let config = &self.language.config;
1463        let opt_in_servers = &config.scope_opt_in_language_servers;
1464        if opt_in_servers.iter().any(|o| *o == *name.0) {
1465            if let Some(over) = self.config_override() {
1466                over.opt_into_language_servers.iter().any(|o| *o == *name.0)
1467            } else {
1468                false
1469            }
1470        } else {
1471            true
1472        }
1473    }
1474
1475    fn config_override(&self) -> Option<&LanguageConfigOverride> {
1476        let id = self.override_id?;
1477        let grammar = self.language.grammar.as_ref()?;
1478        let override_config = grammar.override_config.as_ref()?;
1479        override_config.values.get(&id).map(|e| &e.1)
1480    }
1481}
1482
1483impl Hash for Language {
1484    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
1485        self.id.hash(state)
1486    }
1487}
1488
1489impl PartialEq for Language {
1490    fn eq(&self, other: &Self) -> bool {
1491        self.id.eq(&other.id)
1492    }
1493}
1494
1495impl Eq for Language {}
1496
1497impl Debug for Language {
1498    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
1499        f.debug_struct("Language")
1500            .field("name", &self.config.name)
1501            .finish()
1502    }
1503}
1504
1505impl Grammar {
1506    pub fn id(&self) -> GrammarId {
1507        self.id
1508    }
1509
1510    fn parse_text(&self, text: &Rope, old_tree: Option<Tree>) -> Tree {
1511        with_parser(|parser| {
1512            parser
1513                .set_language(&self.ts_language)
1514                .expect("incompatible grammar");
1515            let mut chunks = text.chunks_in_range(0..text.len());
1516            parser
1517                .parse_with(
1518                    &mut move |offset, _| {
1519                        chunks.seek(offset);
1520                        chunks.next().unwrap_or("").as_bytes()
1521                    },
1522                    old_tree.as_ref(),
1523                )
1524                .unwrap()
1525        })
1526    }
1527
1528    pub fn highlight_map(&self) -> HighlightMap {
1529        self.highlight_map.lock().clone()
1530    }
1531
1532    pub fn highlight_id_for_name(&self, name: &str) -> Option<HighlightId> {
1533        let capture_id = self
1534            .highlights_query
1535            .as_ref()?
1536            .capture_index_for_name(name)?;
1537        Some(self.highlight_map.lock().get(capture_id))
1538    }
1539}
1540
1541impl CodeLabel {
1542    pub fn plain(text: String, filter_text: Option<&str>) -> Self {
1543        let mut result = Self {
1544            runs: Vec::new(),
1545            filter_range: 0..text.len(),
1546            text,
1547        };
1548        if let Some(filter_text) = filter_text {
1549            if let Some(ix) = result.text.find(filter_text) {
1550                result.filter_range = ix..ix + filter_text.len();
1551            }
1552        }
1553        result
1554    }
1555
1556    pub fn push_str(&mut self, text: &str, highlight: Option<HighlightId>) {
1557        let start_ix = self.text.len();
1558        self.text.push_str(text);
1559        let end_ix = self.text.len();
1560        if let Some(highlight) = highlight {
1561            self.runs.push((start_ix..end_ix, highlight));
1562        }
1563    }
1564}
1565
1566impl Ord for LanguageMatcher {
1567    fn cmp(&self, other: &Self) -> std::cmp::Ordering {
1568        self.path_suffixes.cmp(&other.path_suffixes).then_with(|| {
1569            self.first_line_pattern
1570                .as_ref()
1571                .map(Regex::as_str)
1572                .cmp(&other.first_line_pattern.as_ref().map(Regex::as_str))
1573        })
1574    }
1575}
1576
1577impl PartialOrd for LanguageMatcher {
1578    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1579        Some(self.cmp(other))
1580    }
1581}
1582
1583impl Eq for LanguageMatcher {}
1584
1585impl PartialEq for LanguageMatcher {
1586    fn eq(&self, other: &Self) -> bool {
1587        self.path_suffixes == other.path_suffixes
1588            && self.first_line_pattern.as_ref().map(Regex::as_str)
1589                == other.first_line_pattern.as_ref().map(Regex::as_str)
1590    }
1591}
1592
1593#[cfg(any(test, feature = "test-support"))]
1594impl Default for FakeLspAdapter {
1595    fn default() -> Self {
1596        Self {
1597            name: "the-fake-language-server",
1598            capabilities: lsp::LanguageServer::full_capabilities(),
1599            initializer: None,
1600            disk_based_diagnostics_progress_token: None,
1601            initialization_options: None,
1602            disk_based_diagnostics_sources: Vec::new(),
1603            prettier_plugins: Vec::new(),
1604            language_server_binary: LanguageServerBinary {
1605                path: "/the/fake/lsp/path".into(),
1606                arguments: vec![],
1607                env: Default::default(),
1608            },
1609        }
1610    }
1611}
1612
1613#[cfg(any(test, feature = "test-support"))]
1614#[async_trait(?Send)]
1615impl LspAdapter for FakeLspAdapter {
1616    fn name(&self) -> LanguageServerName {
1617        LanguageServerName(self.name.into())
1618    }
1619
1620    fn get_language_server_command<'a>(
1621        self: Arc<Self>,
1622        _: Arc<Language>,
1623        _: Arc<Path>,
1624        _: Arc<dyn LspAdapterDelegate>,
1625        _: futures::lock::MutexGuard<'a, Option<LanguageServerBinary>>,
1626        _: &'a mut AsyncAppContext,
1627    ) -> Pin<Box<dyn 'a + Future<Output = Result<LanguageServerBinary>>>> {
1628        async move { Ok(self.language_server_binary.clone()) }.boxed_local()
1629    }
1630
1631    async fn fetch_latest_server_version(
1632        &self,
1633        _: &dyn LspAdapterDelegate,
1634    ) -> Result<Box<dyn 'static + Send + Any>> {
1635        unreachable!();
1636    }
1637
1638    async fn fetch_server_binary(
1639        &self,
1640        _: Box<dyn 'static + Send + Any>,
1641        _: PathBuf,
1642        _: &dyn LspAdapterDelegate,
1643    ) -> Result<LanguageServerBinary> {
1644        unreachable!();
1645    }
1646
1647    async fn cached_server_binary(
1648        &self,
1649        _: PathBuf,
1650        _: &dyn LspAdapterDelegate,
1651    ) -> Option<LanguageServerBinary> {
1652        unreachable!();
1653    }
1654
1655    async fn installation_test_binary(&self, _: PathBuf) -> Option<LanguageServerBinary> {
1656        unreachable!();
1657    }
1658
1659    fn process_diagnostics(&self, _: &mut lsp::PublishDiagnosticsParams) {}
1660
1661    fn disk_based_diagnostic_sources(&self) -> Vec<String> {
1662        self.disk_based_diagnostics_sources.clone()
1663    }
1664
1665    fn disk_based_diagnostics_progress_token(&self) -> Option<String> {
1666        self.disk_based_diagnostics_progress_token.clone()
1667    }
1668
1669    async fn initialization_options(
1670        self: Arc<Self>,
1671        _: &Arc<dyn LspAdapterDelegate>,
1672    ) -> Result<Option<Value>> {
1673        Ok(self.initialization_options.clone())
1674    }
1675
1676    fn as_fake(&self) -> Option<&FakeLspAdapter> {
1677        Some(self)
1678    }
1679}
1680
1681fn get_capture_indices(query: &Query, captures: &mut [(&str, &mut Option<u32>)]) {
1682    for (ix, name) in query.capture_names().iter().enumerate() {
1683        for (capture_name, index) in captures.iter_mut() {
1684            if capture_name == name {
1685                **index = Some(ix as u32);
1686                break;
1687            }
1688        }
1689    }
1690}
1691
1692pub fn point_to_lsp(point: PointUtf16) -> lsp::Position {
1693    lsp::Position::new(point.row, point.column)
1694}
1695
1696pub fn point_from_lsp(point: lsp::Position) -> Unclipped<PointUtf16> {
1697    Unclipped(PointUtf16::new(point.line, point.character))
1698}
1699
1700pub fn range_to_lsp(range: Range<PointUtf16>) -> lsp::Range {
1701    lsp::Range {
1702        start: point_to_lsp(range.start),
1703        end: point_to_lsp(range.end),
1704    }
1705}
1706
1707pub fn range_from_lsp(range: lsp::Range) -> Range<Unclipped<PointUtf16>> {
1708    let mut start = point_from_lsp(range.start);
1709    let mut end = point_from_lsp(range.end);
1710    if start > end {
1711        mem::swap(&mut start, &mut end);
1712    }
1713    start..end
1714}
1715
1716#[cfg(test)]
1717mod tests {
1718    use super::*;
1719    use gpui::TestAppContext;
1720
1721    #[gpui::test(iterations = 10)]
1722    async fn test_language_loading(cx: &mut TestAppContext) {
1723        let languages = LanguageRegistry::test(cx.executor());
1724        let languages = Arc::new(languages);
1725        languages.register_native_grammars([
1726            ("json", tree_sitter_json::language()),
1727            ("rust", tree_sitter_rust::language()),
1728        ]);
1729        languages.register_test_language(LanguageConfig {
1730            name: "JSON".into(),
1731            grammar: Some("json".into()),
1732            matcher: LanguageMatcher {
1733                path_suffixes: vec!["json".into()],
1734                ..Default::default()
1735            },
1736            ..Default::default()
1737        });
1738        languages.register_test_language(LanguageConfig {
1739            name: "Rust".into(),
1740            grammar: Some("rust".into()),
1741            matcher: LanguageMatcher {
1742                path_suffixes: vec!["rs".into()],
1743                ..Default::default()
1744            },
1745            ..Default::default()
1746        });
1747        assert_eq!(
1748            languages.language_names(),
1749            &[
1750                "JSON".to_string(),
1751                "Plain Text".to_string(),
1752                "Rust".to_string(),
1753            ]
1754        );
1755
1756        let rust1 = languages.language_for_name("Rust");
1757        let rust2 = languages.language_for_name("Rust");
1758
1759        // Ensure language is still listed even if it's being loaded.
1760        assert_eq!(
1761            languages.language_names(),
1762            &[
1763                "JSON".to_string(),
1764                "Plain Text".to_string(),
1765                "Rust".to_string(),
1766            ]
1767        );
1768
1769        let (rust1, rust2) = futures::join!(rust1, rust2);
1770        assert!(Arc::ptr_eq(&rust1.unwrap(), &rust2.unwrap()));
1771
1772        // Ensure language is still listed even after loading it.
1773        assert_eq!(
1774            languages.language_names(),
1775            &[
1776                "JSON".to_string(),
1777                "Plain Text".to_string(),
1778                "Rust".to_string(),
1779            ]
1780        );
1781
1782        // Loading an unknown language returns an error.
1783        assert!(languages.language_for_name("Unknown").await.is_err());
1784    }
1785}