feat: support charset (#442)

Author: SunsetTechuila

package-lock.json

@@ -6,7 +6,7 @@
   "version": "0.17.4",
   "icon": "EditorConfig_icon.png",
   "engines": {
-    "vscode": "^1.98.0"
+    "vscode": "^1.100.0"
   },
   "author": "EditorConfig Team",
   "license": "MIT",
@@ -114,7 +114,7 @@
   "devDependencies": {
     "@types/mocha": "^10.0.10",
     "@types/node": "^20.19.0",
-    "@types/vscode": "~1.98.0",
+    "@types/vscode": "~1.100.0",
     "@typescript-eslint/eslint-plugin": "^8.34.0",
     "@typescript-eslint/parser": "^8.34.0",
     "@vscode/test-electron": "^2.5.2",

package.json

@@ -8,20 +8,31 @@ import {
 	window,
 	workspace,
 } from 'vscode'
+import { KnownProps } from 'editorconfig'
+
 import {
 	InsertFinalNewline,
 	PreSaveTransformation,
 	SetEndOfLine,
 	TrimTrailingWhitespace,
 } from './transformations'
-
 import {
 	applyTextEditorOptions,
 	resolveCoreConfig,
 	resolveFile,
 	resolveTextEditorOptions,
 } from './api'
 
+type Charset = Exclude<KnownProps['charset'], undefined | 'unset'>
+type EncodingMap = Record<Charset, TextDocument['encoding']>
+const encodingMap = {
+	'utf-8': 'utf8',
+	'utf-8-bom': 'utf8bom',
+	'utf-16le': 'utf16le',
+	'utf-16be': 'utf16be',
+	latin1: 'iso88591',
+} as const satisfies EncodingMap
+
 export default class DocumentWatcher {
 	private disposable: Disposable
 	private preSaveTransformations: PreSaveTransformation[] = [
@@ -65,6 +76,8 @@ export default class DocumentWatcher {
 				if (path.basename(doc.fileName) === '.editorconfig') {
 					this.log('.editorconfig file saved.')
 				}
+				// in case document was dirty on open/text editor change
+				this.handleDocumentEncoding(doc)
 			}),
 		)
 
@@ -78,6 +91,10 @@ export default class DocumentWatcher {
 			}),
 		)
 
+		subscriptions.push(
+			workspace.onDidOpenTextDocument(this.handleDocumentEncoding),
+		)
+
 		this.disposable = Disposable.from.apply(this, subscriptions)
 		this.log('Document watcher initialized')
 	}
@@ -156,6 +173,39 @@ export default class DocumentWatcher {
 				onNoActiveTextEditor: this.onNoActiveTextEditor,
 				onSuccess: this.onSuccess,
 			})
+			this.handleDocumentEncoding(editor.document)
 		}
 	}
+
+	private async handleDocumentEncoding(document: TextDocument) {
+		const relativePath = workspace.asRelativePath(document.fileName)
+		const editorconfigSettings = await resolveCoreConfig(document, {
+			onBeforeResolve: this.onBeforeResolve,
+		})
+
+		const { charset } = editorconfigSettings
+		this.log(`${relativePath}: Target charset is`, charset ?? 'not set')
+		if (!charset) {
+			return
+		}
+		if (!(charset in encodingMap)) {
+			this.log(`${relativePath}: Unsupported charset`)
+			return
+		}
+
+		const targetEncoding = encodingMap[charset as keyof typeof encodingMap]
+		if (targetEncoding === document.encoding) {
+			return
+		}
+
+		if (document.isDirty) {
+			this.log(`${relativePath}: Cannot change encoding, document is dirty`)
+			return
+		}
+
+		this.log(`${relativePath}: Re-opening document with ${targetEncoding} encoding...`)
+		await workspace.openTextDocument(document.uri, {
+			encoding: targetEncoding,
+		})
+	}
 }

src/DocumentWatcher.ts

@@ -0,0 +1,4 @@
+root = true
+
+[*]
+charset = latin1

src/test/suite/fixtures/charset/latin1/.editorconfig

@@ -0,0 +1,4 @@
+root = true
+
+[*]
+charset = unset

src/test/suite/fixtures/charset/unset/.editorconfig

@@ -0,0 +1,4 @@
+root = true
+
+[*]
+charset = utf-16be

src/test/suite/fixtures/charset/utf-16be/.editorconfig

@@ -0,0 +1,4 @@
+root = true
+
+[*]
+charset = utf-16le

src/test/suite/fixtures/charset/utf-16le/.editorconfig

@@ -0,0 +1,4 @@
+root = true
+
+[*]
+charset = utf-8-bom

src/test/suite/fixtures/charset/utf-8-bom/.editorconfig

@@ -0,0 +1,4 @@
+root = true
+
+[*]
+charset = utf-8

src/test/suite/fixtures/charset/utf-8/.editorconfig

@@ -315,6 +315,51 @@ suite('EditorConfig extension', function () {
 			'editor selection end line changed',
 		)
 	})
+
+	test('charset (utf-8)', async () => {
+		const document = await withSetting('charset', 'utf-8').createDoc()
+		assert.strictEqual(
+			document.encoding,
+			'utf8',
+			`document encoding is ${document.encoding} instead of utf8`,
+		)
+	})
+
+	test('charset (utf-8-bom)', async () => {
+		const document = await withSetting('charset', 'utf-8-bom').createDoc()
+		assert.strictEqual(
+			document.encoding,
+			'utf8bom',
+			`document encoding is ${document.encoding} instead of utf8bom`,
+		)
+	})
+
+	test('charset (utf-16le)', async () => {
+		const document = await withSetting('charset', 'utf-16le').createDoc()
+		assert.strictEqual(
+			document.encoding,
+			'utf16le',
+			`document encoding is ${document.encoding} instead of utf16le`,
+		)
+	})
+
+	test('charset (utf-16be)', async () => {
+		const document = await withSetting('charset', 'utf-16be').createDoc()
+		assert.strictEqual(
+			document.encoding,
+			'utf16be',
+			`document encoding is ${document.encoding} instead of utf16be`,
+		)
+	})
+
+	test('charset (latin1)', async () => {
+		const document = await withSetting('charset', 'latin1').createDoc()
+		assert.strictEqual(
+			document.encoding,
+			'iso88591',
+			`document encoding is ${document.encoding} instead of iso88591`,
+		)
+	})
 })
 
 function withSetting(
@@ -327,11 +372,13 @@ function withSetting(
 ) {
 	return {
 		async getText() {
-			return (await createDoc(options.contents, options.fileName)).getText()
+			return (
+				await this.createDoc(options.contents, options.fileName)
+			).getText()
 		},
 		saveText(text: string) {
 			return new Promise<string>(async resolve => {
-				const doc = await createDoc(options.contents, options.fileName)
+				const doc = await this.createDoc(options.contents, options.fileName)
 				workspace.onDidChangeTextDocument(doc.save)
 				workspace.onDidSaveTextDocument(savedDoc => {
 					assert.strictEqual(savedDoc.isDirty, false, 'dirty saved doc')
@@ -346,15 +393,15 @@ function withSetting(
 				)
 			})
 		},
-	}
-	async function createDoc(contents = '', name = 'test') {
-		const uri = await utils.createFile(
-			contents,
-			getFixturePath([rule, value, name]),
-		)
-		const doc = await workspace.openTextDocument(uri)
-		await window.showTextDocument(doc)
-		await wait(50) // wait for EditorConfig to apply new settings
-		return doc
+		async createDoc(contents = '', name = 'test') {
+			const uri = await utils.createFile(
+				contents,
+				getFixturePath([rule, value, name]),
+			)
+			const doc = await workspace.openTextDocument(uri)
+			await window.showTextDocument(doc)
+			await wait(50) // wait for EditorConfig to apply new settings
+			return doc
+		},
 	}
 }