investigate

package.json

@@ -68,6 +68,7 @@
     "@vscode/sqlite3": "5.0.8",
     "@vscode/sudo-prompt": "9.3.1",
     "@vscode/vscode-languagedetection": "1.0.21",
+    "chardet": "^1.4.0",
     "graceful-fs": "4.2.8",
     "http-proxy-agent": "^2.1.0",
     "https-proxy-agent": "^2.2.3",

src/vs/workbench/services/textfile/common/encoding.ts

@@ -316,27 +316,22 @@ const IGNORE_ENCODINGS = ['ascii', 'utf-16', 'utf-32'];
  * Guesses the encoding from buffer.
  */
 async function guessEncodingByBuffer(buffer: VSBuffer): Promise<string | null> {
-	const jschardet = await import('jschardet');
+	const nodechardet = await import('chardet');
 
 	// ensure to limit buffer for guessing due to https://github.com/aadsm/jschardet/issues/53
 	const limitedBuffer = buffer.slice(0, AUTO_ENCODING_GUESS_MAX_BYTES);
 
-	// before guessing jschardet calls toString('binary') on input if it is a Buffer,
-	// since we are using it inside browser environment as well we do conversion ourselves
-	// https://github.com/aadsm/jschardet/blob/v2.1.1/src/index.js#L36-L40
-	const binaryString = encodeLatin1(limitedBuffer.buffer);
-
-	const guessed = jschardet.detect(binaryString);
-	if (!guessed || !guessed.encoding) {
+	const guessed = nodechardet.detect(limitedBuffer.buffer);
+	if (!guessed) {
 		return null;
 	}
 
-	const enc = guessed.encoding.toLowerCase();
+	const enc = guessed.toLowerCase();
 	if (0 <= IGNORE_ENCODINGS.indexOf(enc)) {
 		return null; // see comment above why we ignore some encodings
 	}
 
-	return toIconvLiteEncoding(guessed.encoding);
+	return toIconvLiteEncoding(guessed);
 }
 
 const JSCHARDET_TO_ICONV_ENCODINGS: { [name: string]: string } = {
@@ -351,15 +346,6 @@ function toIconvLiteEncoding(encodingName: string): string {
 	return mapped || normalizedEncodingName;
 }
 
-function encodeLatin1(buffer: Uint8Array): string {
-	let result = '';
-	for (let i = 0; i < buffer.length; i++) {
-		result += String.fromCharCode(buffer[i]);
-	}
-
-	return result;
-}
-
 /**
  * The encodings that are allowed in a settings file don't match the canonical encoding labels specified by WHATWG.
  * See https://encoding.spec.whatwg.org/#names-and-labels

yarn.lock

@@ -2551,6 +2551,11 @@ chardet@^0.7.0:
   resolved "https://registry.yarnpkg.com/chardet/-/chardet-0.7.0.tgz#90094849f0937f2eedc2425d0d28a9e5f0cbad9e"
   integrity sha512-mT8iDcrh03qDGRRmoA2hmBJnxpllMR+0/0qlzjqZES6NdiWDcZkCNAk4rPFZ9Q85r27unkiNNg8ZOiwZXBHwcA==
 
+chardet@^1.4.0:
+  version "1.4.0"
+  resolved "https://registry.yarnpkg.com/chardet/-/chardet-1.4.0.tgz#278748f260219990fb2167dbfb1b253ca26b41ea"
+  integrity sha512-NpwMDdSIprbYx1CLnfbxEIarI0Z+s9MssEgggMNheGM+WD68yOhV7IEA/3r6tr0yTRgQD0HuZJDw32s99i6L+A==
+
 charenc@~0.0.1:
   version "0.0.2"
   resolved "https://registry.yarnpkg.com/charenc/-/charenc-0.0.2.tgz#c0a1d2f3a7092e03774bfa83f14c0fc5790a8667"