Merge pull request #736 from boarwell/encoding-charset

Add encoding charset sniffing

Author: thornjad

lib/core/index.js

@@ -5,13 +5,16 @@
 const path = require('path');
 const fs = require('fs');
 const url = require('url');
+const { Readable } = require('stream');
+const buffer = require('buffer');
 const mime = require('mime');
 const urlJoin = require('url-join');
 const showDir = require('./show-dir');
 const version = require('../../package.json').version;
 const status = require('./status-handlers');
 const generateEtag = require('./etag');
 const optsParser = require('./opts');
+const htmlEncodingSniffer = require('html-encoding-sniffer');
 
 let httpServerCore = null;
 
@@ -234,8 +237,17 @@ module.exports = function createMiddleware(_dir, _options) {
       let cacheControl = cache;
       let stream = null;
       if (contentType && isTextFile(contentType)) {
-        // Assume text types are utf8
-        contentType += '; charset=UTF-8';
+        if (stat.size < buffer.constants.MAX_LENGTH) {
+          const bytes = fs.readFileSync(file);
+          const sniffedEncoding = htmlEncodingSniffer(bytes, {
+            defaultEncoding: 'UTF-8'
+          });
+          contentType += `; charset=${sniffedEncoding}`;
+          stream = Readable.from(bytes)
+        } else {
+          // Assume text types are utf8
+          contentType += '; charset=UTF-8';
+        }
       }
 
       if (file === gzippedFile) { // is .gz picked up
@@ -317,7 +329,10 @@ module.exports = function createMiddleware(_dir, _options) {
         return;
       }
 
-      stream = fs.createReadStream(file);
+      // stream may already have been assigned during encoding sniffing.
+      if (stream === null) {
+        stream = fs.createReadStream(file);
+      }
 
       stream.pipe(res);
       stream.on('error', (err) => {