Offline, privacy-first grammar checker. Fast, open-source, Rust-powered
fix(obsidian): problems when converting from Rust char indexes (#2104)
Elijah Potteraeb303b4fa20
diff --git a/harper-wasm/src/lib.rs b/harper-wasm/src/lib.rsindex 66e47d900..f257c542b 100644--- a/harper-wasm/src/lib.rs+++ b/harper-wasm/src/lib.rs@@ -280,7 +280,9 @@ impl Linter {.into_iter().map(|l| {let problem_text = l.span.get_content_string(&source);- Lint::new(l, problem_text, language)+ let span = Into::<Span>::into(l.span).to_js_indices(source.as_slice());++ Lint::new(l, span, problem_text, language)}).collect(),})@@ -310,7 +312,8 @@ impl Linter {.into_iter().map(|l| {let problem_text = l.span.get_content_string(&source);- Lint::new(l, problem_text, language)+ let span = Into::<Span>::into(l.span).to_js_indices(source.as_slice());+ Lint::new(l, span, problem_text, language)}).collect()}@@ -471,6 +474,8 @@ impl Suggestion {#[wasm_bindgen]pub struct Lint {inner: harper_core::linting::Lint,+ /// Indexed in a proverbial JS string+ span: Span,/// The problematic text that produced this lint.problem_text: String,language: Language,@@ -480,11 +485,13 @@ pub struct Lint {impl Lint {pub(crate) fn new(inner: harper_core::linting::Lint,+ span: Span,problem_text: String,language: Language,) -> Self {Self {inner,+ span,problem_text,language,}@@ -521,7 +528,7 @@ impl Lint {/// Get the location of the problematic text.pub fn span(&self) -> Span {- self.inner.span.into()+ self.span}/// Get a description of the error.@@ -535,6 +542,13 @@ impl Lint {}}+/// Convert Harper's character index into a UTF-16 code unit index understood by JS.+fn char_idx_to_js_str_idx(char_idx: usize, char_str: &[char]) -> usize {+ char_str.iter().take(char_idx).fold(0usize, |acc, ch| {+ acc + if (*ch as u32) <= 0xFFFF { 1 } else { 2 }+ })+}+#[wasm_bindgen]pub fn get_default_lint_config_as_json() -> String {let config =@@ -577,6 +591,15 @@ impl Span {}}+impl Span {+ pub fn to_js_indices(&self, source: &[char]) -> Self {+ Self::new(+ char_idx_to_js_str_idx(self.start, source),+ char_idx_to_js_str_idx(self.end, source),+ )+ }+}+impl From<Span> for harper_core::Span<char> {fn from(value: Span) -> Self {harper_core::Span::new(value.start, value.end)diff --git a/packages/harper.js/src/Linter.test.ts b/packages/harper.js/src/Linter.test.tsindex b217c9fa9..977ab6acf 100644--- a/packages/harper.js/src/Linter.test.ts+++ b/packages/harper.js/src/Linter.test.ts@@ -333,6 +333,20 @@ for (const [linterName, Linter] of Object.entries(linters)) {const newLinter = new Linter({ binary });await newLinter.importStatsFile(stats);});++ test(`${linterName} emits the correct span indices`, async () => {+ const text = '✉️👋👍✉️🚀✉️🌴 This is to show the offset issue sdssda is it there?';++ const linter = new LocalLinter({ binary });+ const lints = await linter.lint(text);++ const span = lints[0].span();++ expect(span.start).toBe(48);+ expect(span.end).toBe(54);++ expect(text.slice(span.start, span.end)).toBe('sdssda');+ });}test('Linters have the same config format', async () => {diff --git a/packages/obsidian-plugin/src/State.ts b/packages/obsidian-plugin/src/State.tsindex a5e31bc71..9849b2192 100644--- a/packages/obsidian-plugin/src/State.ts+++ b/packages/obsidian-plugin/src/State.ts@@ -1,10 +1,10 @@import type { Extension, StateField } from '@codemirror/state';import type { LintConfig, Linter, Suggestion } from 'harper.js';import { binaryInlined, type Dialect, LocalLinter, SuggestionKind, WorkerLinter } from 'harper.js';-import { toArray } from 'lodash-es';import { minimatch } from 'minimatch';import type { MarkdownFileInfo, MarkdownView, Workspace } from 'obsidian';import { linter } from './lint';+import { charIndexToCodeUnit } from './textUtils';export type Settings = {ignoredLints?: string;@@ -128,16 +128,13 @@ export default class State {}const text = view.state.doc.sliceString(-1);- const chars = toArray(text);+ const chars = Array.from(text);const lints = await this.harper.lint(text);return lints.map((lint) => {const span = lint.span();- span.start = charIndexToCodePointIndex(span.start, chars);- span.end = charIndexToCodePointIndex(span.end, chars);-const actions = lint.suggestions().map((sug) => {return {name:@@ -334,26 +331,11 @@ export default class State {}}-/** Harper returns positions based on char indexes,- * but Obsidian identifies locations in documents based on Unicode code points.- * This converts between from the former to the latter.*/-function charIndexToCodePointIndex(index: number, sourceChars: string[]): number {- let traversed = 0;-- for (let i = 0; i < index; i++) {- const delta = sourceChars[i].length;-- traversed += delta;- }-- return traversed;-}-function suggestionToLabel(sug: Suggestion) {if (sug.kind() === SuggestionKind.Remove) {return 'Remove';} else if (sug.kind() === SuggestionKind.Replace) {- return `“Replace with ${sug.get_replacement_text()}”`;+ return `Replace with “${sug.get_replacement_text()}”`;} else if (sug.kind() === SuggestionKind.InsertAfter) {return `Insert “${sug.get_replacement_text()}” after this.`;}diff --git a/packages/obsidian-plugin/src/textUtils.test.ts b/packages/obsidian-plugin/src/textUtils.test.tsindex fdd527db2..15c03b9af 100644--- a/packages/obsidian-plugin/src/textUtils.test.ts+++ b/packages/obsidian-plugin/src/textUtils.test.ts@@ -1,4 +1,5 @@-import { expect, test } from 'vitest';+import { binary, LocalLinter } from 'harper.js';+import { describe, expect, test } from 'vitest';import { linesToString, stringToLines } from './textUtils';test('Dictionary values are reversible', () => {