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