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