fix: Fixed encoding detection for gutter (close #526) (#590)

Author: edgardmessias

package-lock.json

@@ -46,7 +46,6 @@
     "style-check": "npx prettylint src/**/*.ts"
   },
   "dependencies": {
-    "is-utf8": "^0.2.1",
     "minimatch": "^3.0.4",
     "original-fs": "^1.0.0",
     "semver": "^6.0.0",

package.json

@@ -190,7 +190,7 @@ export enum PropStatus {
 
 export interface ICpOptions extends SpawnOptions {
   cwd?: string;
-  encoding?: string;
+  encoding?: string | null;
   log?: boolean;
   username?: string;
   password?: string;

src/common/types.ts

@@ -0,0 +1,72 @@
+import { jschardet } from "./vscodeModules";
+
+jschardet.Constants.MINIMUM_THRESHOLD = 0.2;
+jschardet.MacCyrillicModel.mTypicalPositiveRatio += 0.001;
+
+function detectEncodingByBOM(buffer: Buffer): string | null {
+  if (!buffer || buffer.length < 2) {
+    return null;
+  }
+
+  const b0 = buffer.readUInt8(0);
+  const b1 = buffer.readUInt8(1);
+
+  // UTF-16 BE
+  if (b0 === 0xfe && b1 === 0xff) {
+    return "utf16be";
+  }
+
+  // UTF-16 LE
+  if (b0 === 0xff && b1 === 0xfe) {
+    return "utf16le";
+  }
+
+  if (buffer.length < 3) {
+    return null;
+  }
+
+  const b2 = buffer.readUInt8(2);
+
+  // UTF-8
+  if (b0 === 0xef && b1 === 0xbb && b2 === 0xbf) {
+    return "utf8";
+  }
+
+  return null;
+}
+
+const IGNORE_ENCODINGS = ["ascii", "utf-8", "utf-16", "utf-32"];
+
+const JSCHARDET_TO_ICONV_ENCODINGS: { [name: string]: string } = {
+  ibm866: "cp866",
+  big5: "cp950"
+};
+
+export function detectEncoding(buffer: Buffer): string | null {
+  const result = detectEncodingByBOM(buffer);
+
+  if (result) {
+    return result;
+  }
+
+  const detected = jschardet.detect(buffer);
+
+  if (!detected || !detected.encoding || detected.confidence < 0.8) {
+    return null;
+  }
+
+  const encoding = detected.encoding;
+
+  // Ignore encodings that cannot guess correctly
+  // (http://chardet.readthedocs.io/en/latest/supported-encodings.html)
+  if (0 <= IGNORE_ENCODINGS.indexOf(encoding.toLowerCase())) {
+    return null;
+  }
+
+  const normalizedEncodingName = encoding
+    .replace(/[^a-zA-Z0-9]/g, "")
+    .toLowerCase();
+  const mapped = JSCHARDET_TO_ICONV_ENCODINGS[normalizedEncodingName];
+
+  return mapped || normalizedEncodingName;
+}

src/encoding.ts

@@ -1,6 +1,5 @@
 import * as cp from "child_process";
 import { EventEmitter } from "events";
-import isUtf8 = require("is-utf8");
 import * as proc from "process";
 import { Readable } from "stream";
 import {
@@ -9,12 +8,13 @@ import {
   IExecutionResult,
   ISvnOptions
 } from "./common/types";
+import * as encodeUtil from "./encoding";
 import { configuration } from "./helpers/configuration";
 import { parseInfoXml } from "./infoParser";
 import SvnError from "./svnError";
 import { Repository } from "./svnRepository";
 import { dispose, IDisposable, toDisposable } from "./util";
-import { iconv, jschardet } from "./vscodeModules";
+import { iconv } from "./vscodeModules";
 
 export const svnErrorCodes: { [key: string]: string } = {
   AuthorizationFailed: "E170001",
@@ -102,9 +102,14 @@ export class Svn {
     // Force non interactive environment
     args.push("--non-interactive");
 
-    let encoding = options.encoding || "";
+    let encoding: string | undefined | null = options.encoding;
     delete options.encoding;
 
+    // SVN with '--xml' always return 'UTF-8', and jschardet detects this encoding: 'TIS-620'
+    if (args.includes("--xml")) {
+      encoding = "utf8";
+    }
+
     const defaults: cp.SpawnOptions = {
       env: proc.env
     };
@@ -156,35 +161,20 @@ export class Svn {
 
     dispose(disposables);
 
-    // SVN with '--xml' always return 'UTF-8', and jschardet detects this encoding: 'TIS-620'
-    if (args.includes("--xml")) {
-      encoding = "utf8";
-    } else if (encoding === "") {
-      encoding = "utf8"; // Initial encoding
-
-      const defaultEncoding = configuration.get<string>("default.encoding");
-      if (defaultEncoding) {
-        if (!iconv.encodingExists(defaultEncoding)) {
-          this.logOutput(
-            "svn.default.encoding: Invalid Parameter: '" +
-            defaultEncoding +
-            "'.\n"
-          );
-        } else if (!isUtf8(stdout)) {
-          encoding = defaultEncoding;
-        }
-      } else {
-        jschardet.MacCyrillicModel.mTypicalPositiveRatio += 0.001;
-
-        const encodingGuess = jschardet.detect(stdout);
-
-        if (
-          encodingGuess.confidence > 0.8 &&
-          iconv.encodingExists(encodingGuess.encoding)
-        ) {
-          encoding = encodingGuess.encoding;
-        }
+    if (!encoding) {
+      encoding = encodeUtil.detectEncoding(stdout);
+    }
+
+    // if not detected
+    if (!encoding) {
+      encoding = configuration.get<string>("default.encoding");
+    }
+
+    if (!iconv.encodingExists(encoding)) {
+      if (encoding) {
+        console.warn(`SVN: The encoding "${encoding}" is invalid`);
       }
+      encoding = "utf8";
     }
 
     const decodedStdout = iconv.decode(stdout, encoding);