Naive spellbook integration

Cargo.lock

@@ -1441,6 +1441,7 @@ dependencies = [
  "serde",
  "serde_json",
  "slotmap",
+ "spellbook",
  "tempfile",
  "thiserror",
  "tokio",
@@ -2167,6 +2168,16 @@ dependencies = [
  "windows-sys 0.52.0",
 ]
 
+[[package]]
+name = "spellbook"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "82fee930378e69527fbeb7e5f737e3ddaf32e675a04fc9375de676aa8de84693"
+dependencies = [
+ "ahash",
+ "hashbrown",
+]
+
 [[package]]
 name = "static_assertions"
 version = "1.1.0"

Cargo.toml

@@ -42,6 +42,7 @@ tree-sitter = { version = "0.22" }
 nucleo = "0.5.0"
 slotmap = "1.0.7"
 thiserror = "1.0"
+spellbook = "0.1"
 
 [workspace.package]
 version = "24.7.0"

helix-loader/src/grammar.rs

@@ -592,6 +592,6 @@ fn mtime(path: &Path) -> Result<SystemTime> {
 /// Gives the contents of a file from a language's `runtime/queries/<lang>`
 /// directory
 pub fn load_runtime_file(language: &str, filename: &str) -> Result<String, std::io::Error> {
-    let path = crate::runtime_file(&PathBuf::new().join("queries").join(language).join(filename));
+    let path = crate::runtime_file(PathBuf::new().join("queries").join(language).join(filename));
     std::fs::read_to_string(path)
 }

helix-loader/src/lib.rs

@@ -107,7 +107,8 @@ fn find_runtime_file(rel_path: &Path) -> Option<PathBuf> {
 /// The valid runtime directories are searched in priority order and the first
 /// file found to exist is returned, otherwise the path to the final attempt
 /// that failed.
-pub fn runtime_file(rel_path: &Path) -> PathBuf {
+pub fn runtime_file<P: AsRef<Path>>(rel_path: P) -> PathBuf {
+    let rel_path = rel_path.as_ref();
     find_runtime_file(rel_path).unwrap_or_else(|| {
         RUNTIME_DIRS
             .last()
@@ -132,6 +133,14 @@ pub fn cache_dir() -> PathBuf {
     path
 }
 
+pub fn state_dir() -> PathBuf {
+    let strategy = choose_base_strategy().expect("Unable to find the cache directory!");
+    // BaseStrategy always return Some in `state_dir`.
+    let mut path = strategy.state_dir().unwrap();
+    path.push("helix");
+    path
+}
+
 pub fn config_file() -> PathBuf {
     CONFIG_FILE.get().map(|path| path.to_path_buf()).unwrap()
 }
@@ -152,6 +161,11 @@ pub fn default_log_file() -> PathBuf {
     cache_dir().join("helix.log")
 }
 
+// HACK: there should a personal dictionary per-locale.
+pub fn personal_dictionary_file() -> PathBuf {
+    state_dir().join("personal-dictionary.txt")
+}
+
 /// Merge two TOML documents, merging values from `right` onto `left`
 ///
 /// When an array exists in both `left` and `right`, `right`'s array is

helix-term/src/commands.rs

@@ -567,6 +567,7 @@ impl MappableCommand {
         command_palette, "Open command palette",
         goto_word, "Jump to a two-character label",
         extend_to_word, "Extend to a two-character label",
+        add_word_to_personal_dictionary, "Add the word under the primary cursor to the personal dictionary for the current locale",
     );
 }
 
@@ -6298,3 +6299,61 @@ fn jump_to_word(cx: &mut Context, behaviour: Movement) {
     }
     jump_to_label(cx, words, behaviour)
 }
+
+// HACK: this should be folded into code actions.
+fn add_word_to_personal_dictionary(cx: &mut Context) {
+    let (view, doc) = current_ref!(cx.editor);
+    let text = doc.text().slice(..);
+    let selection = doc.selection(view.id).primary();
+    let range = if selection.len() == 1 {
+        textobject::textobject_word(text, selection, textobject::TextObject::Inside, 1, false)
+    } else {
+        selection
+    };
+    let word = range.fragment(text);
+
+    let prompt = ui::Prompt::new(
+        "add-word:".into(),
+        None,
+        ui::completers::none,
+        move |cx, input: &str, event: PromptEvent| {
+            fn append_word(word: &str) -> std::io::Result<()> {
+                use std::io::Write;
+                let path = helix_loader::state_dir().join("personal-dictionary.txt");
+                let mut file = std::fs::OpenOptions::new()
+                    .create(true)
+                    .append(true)
+                    .open(path)?;
+                file.write_all(word.as_bytes())?;
+                file.write_all(helix_core::NATIVE_LINE_ENDING.as_str().as_bytes())?;
+                file.sync_data()?;
+
+                Ok(())
+            }
+
+            if event != PromptEvent::Validate {
+                return;
+            }
+
+            if let Err(err) = cx.editor.dictionary.add(input) {
+                cx.editor.set_error(format!(
+                    "Failed to add \"{input}\" to the dictionary: {err}"
+                ));
+                return;
+            }
+
+            if let Err(err) = append_word(input) {
+                cx.editor.set_error(format!(
+                    "Failed to persist \"{input}\" to the on-disk dictionary: {err}"
+                ));
+                return;
+            }
+
+            cx.editor
+                .set_status(format!("Added \"{input}\" to the dictionary"));
+        },
+    )
+    .with_line(word.into(), cx.editor);
+
+    cx.push_layer(Box::new(prompt));
+}

helix-term/src/keymap/default.rs

@@ -229,6 +229,7 @@ pub fn default() -> HashMap<Mode, KeyTrie> {
             "D" => workspace_diagnostics_picker,
             "g" => changed_file_picker,
             "a" => code_action,
+            "A" => add_word_to_personal_dictionary,
             "'" => last_picker,
             "G" => { "Debug (experimental)" sticky=true
                 "l" => dap_launch,

helix-term/src/ui/editor.rs

@@ -21,6 +21,7 @@ use helix_core::{
     unicode::width::UnicodeWidthStr,
     visual_offset_from_block, Change, Position, Range, Selection, Transaction,
 };
+use helix_stdx::rope::RopeSliceExt;
 use helix_view::{
     annotations::diagnostics::DiagnosticFilter,
     document::{Mode, SavePoint, SCRATCH_BUFFER_NAME},
@@ -28,7 +29,7 @@ use helix_view::{
     graphics::{Color, CursorKind, Modifier, Rect, Style},
     input::{KeyEvent, MouseButton, MouseEvent, MouseEventKind},
     keyboard::{KeyCode, KeyModifiers},
-    Document, Editor, Theme, View,
+    Dictionary, Document, Editor, Theme, View,
 };
 use std::{mem::take, num::NonZeroUsize, path::PathBuf, rc::Rc, sync::Arc};
 
@@ -144,6 +145,10 @@ impl EditorView {
             }
             overlay_highlights = Box::new(syntax::merge(overlay_highlights, diagnostic));
         }
+        let spell = Self::doc_spell_highlights(&editor.dictionary, doc, view, theme);
+        if !spell.is_empty() {
+            overlay_highlights = Box::new(syntax::merge(overlay_highlights, spell));
+        }
 
         if is_focused {
             let highlights = syntax::merge(
@@ -460,6 +465,55 @@ impl EditorView {
         ]
     }
 
+    pub fn doc_spell_highlights(
+        dict: &Dictionary,
+        doc: &Document,
+        view: &View,
+        theme: &Theme,
+    ) -> Vec<(usize, std::ops::Range<usize>)> {
+        // This is **very** ***very*** naive and not at all reflective of what the actual
+        // integration will look like. Doing this per-render is very needlessly expensive.
+        // Instead it should be done in the background and possibly incrementally (only
+        // re-checking ranges that are affected by document changes). However regex-cursor
+        // is very fast and so is spellbook (degenerate cases max out at 1μs in a release
+        // build on my machine, i.e. a worst case throughput of 2 million words / second) so
+        // this is suitable for my testing. I mostly want to find cases where spellbook's
+        // results are surprising.
+        // Also we want to use tree-sitter to mark nodes as ones that should be spellchecked
+        // and maybe specify strategies for doing tokenization (try to tokenize prose vs.
+        // programming languages).
+        // Plus these should really be proper diagnostics so that we can pull them up in the
+        // diagnostics picker and jump to them.
+        use helix_stdx::rope::Regex;
+        use once_cell::sync::Lazy;
+        use std::borrow::Cow;
+        static WORDS: Lazy<Regex> = Lazy::new(|| Regex::new(r#"[0-9A-Z]*(['-]?[a-z]+)*"#).unwrap());
+
+        let mut spans = Vec::new();
+        let error = theme.find_scope_index("diagnostic.error").unwrap();
+
+        let text = doc.text().slice(..);
+        let start = text.line_to_char(text.char_to_line(doc.view_offset(view.id).anchor));
+        let end = text.line_to_char(view.estimate_last_doc_line(doc) + 1);
+
+        for match_ in WORDS.find_iter(text.regex_input_at(start..end)) {
+            let range = text.byte_to_char(match_.start())..text.byte_to_char(match_.end());
+            // TODO: consider how to allow passing the RopeSlice to spellbook:
+            // * Use an Input trait like regex-cursor?
+            // * Accept `impl Iterator<Item = char>`?
+            // * Maybe spellbook should have an internal `String` buffer and it should try to copy
+            //   the word into that? Only in the best case do you not have to allocate at all.
+            //   Maybe we should use a single string buffer and perform all changes to the string
+            //   in-place instead of using `replace` from the stdlib and Cows.
+            let word = Cow::from(text.slice(range.clone()));
+            if !dict.check(&word) {
+                spans.push((error, range))
+            }
+        }
+
+        spans
+    }
+
     /// Get highlight spans for selections in a document view.
     pub fn doc_selection_highlights(
         mode: Mode,

helix-view/Cargo.toml

@@ -52,6 +52,8 @@ log = "~0.4"
 parking_lot = "0.12.3"
 thiserror.workspace = true
 
+spellbook.workspace = true
+
 [target.'cfg(windows)'.dependencies]
 clipboard-win = { version = "5.4", features = ["std"] }
 

helix-view/src/editor.rs

@@ -10,7 +10,7 @@ use crate::{
     register::Registers,
     theme::{self, Theme},
     tree::{self, Tree},
-    Document, DocumentId, View, ViewId,
+    Dictionary, Document, DocumentId, View, ViewId,
 };
 use dap::StackFrame;
 use helix_vcs::DiffProviderRegistry;
@@ -1078,6 +1078,9 @@ pub struct Editor {
 
     pub mouse_down_range: Option<Range>,
     pub cursor_cache: CursorCache,
+
+    /// HACK:
+    pub dictionary: Dictionary,
 }
 
 pub type Motion = Box<dyn Fn(&mut Editor)>;
@@ -1157,6 +1160,30 @@ impl Editor {
         // HAXX: offset the render area height by 1 to account for prompt/commandline
         area.height -= 1;
 
+        // HACK: what's the right interface for Spellbook to expose so we don't have to
+        // read these entire files into strings? (See associated TODO in Spellbook.)
+        let aff =
+            std::fs::read_to_string(helix_loader::runtime_file("dictionaries/en_US/en_US.aff"))
+                .unwrap();
+        let dic =
+            std::fs::read_to_string(helix_loader::runtime_file("dictionaries/en_US/en_US.dic"))
+                .unwrap();
+        // HACK: All this stuff should happen off the main thread.
+        let mut dictionary = Dictionary::new(&aff, &dic).unwrap();
+        if let Ok(file) = std::fs::File::open(helix_loader::personal_dictionary_file()) {
+            use std::io::{BufRead as _, BufReader};
+            let reader = BufReader::with_capacity(8 * 1024, file);
+            for line in reader.lines() {
+                let line = line.unwrap();
+                let line = line.trim();
+                if line.is_empty() {
+                    continue;
+                }
+
+                dictionary.add(line).unwrap();
+            }
+        }
+
         Self {
             mode: Mode::Normal,
             tree: Tree::new(area),
@@ -1195,6 +1222,7 @@ impl Editor {
             handlers,
             mouse_down_range: None,
             cursor_cache: CursorCache::default(),
+            dictionary,
         }
     }
 

helix-view/src/lib.rs

@@ -75,5 +75,6 @@ pub fn align_view(doc: &mut Document, view: &View, align: Align) {
 pub use document::Document;
 pub use editor::Editor;
 use helix_core::char_idx_at_visual_offset;
+pub use spellbook::Dictionary;
 pub use theme::Theme;
 pub use view::View;