Fix unicode chunking issue, ignore BOM mark

lib/eventsource.js

@@ -10,6 +10,18 @@ var httpsOptions = [
   'rejectUnauthorized', 'secureProtocol', 'servername'
 ]
 
+var bom = [239, 187, 191]
+var colon = 58
+var space = 32
+var lineFeed = 10
+var carriageReturn = 13
+
+function hasBom (buf) {
+  return bom.every(function (charCode, index) {
+    return buf[index] === charCode
+  })
+}
+
 /**
  * Creates a new EventSource object
  *
@@ -160,16 +172,21 @@ function EventSource (url, eventSourceInitDict) {
 
       // text/event-stream parser adapted from webkit's
       // Source/WebCore/page/EventSource.cpp
-      var buf = ''
+      var isFirst = true
+      var buf
       res.on('data', function (chunk) {
-        buf += chunk
+        buf = buf ? Buffer.concat([buf, chunk]) : chunk
+        if (isFirst && hasBom(buf)) {
+          buf = buf.slice(bom.length)
+        }
 
+        isFirst = false
         var pos = 0
         var length = buf.length
 
         while (pos < length) {
           if (discardTrailingNewline) {
-            if (buf[pos] === '\n') {
+            if (buf[pos] === lineFeed) {
               ++pos
             }
             discardTrailingNewline = false
@@ -181,14 +198,14 @@ function EventSource (url, eventSourceInitDict) {
 
           for (var i = pos; lineLength < 0 && i < length; ++i) {
             c = buf[i]
-            if (c === ':') {
+            if (c === colon) {
               if (fieldLength < 0) {
                 fieldLength = i - pos
               }
-            } else if (c === '\r') {
+            } else if (c === carriageReturn) {
               discardTrailingNewline = true
               lineLength = i - pos
-            } else if (c === '\n') {
+            } else if (c === lineFeed) {
               lineLength = i - pos
             }
           }
@@ -203,7 +220,7 @@ function EventSource (url, eventSourceInitDict) {
         }
 
         if (pos === length) {
-          buf = ''
+          buf = void 0
         } else if (pos > 0) {
           buf = buf.slice(pos)
         }
@@ -245,19 +262,19 @@ function EventSource (url, eventSourceInitDict) {
     } else if (fieldLength > 0) {
       var noValue = fieldLength < 0
       var step = 0
-      var field = buf.slice(pos, pos + (noValue ? lineLength : fieldLength))
+      var field = buf.slice(pos, pos + (noValue ? lineLength : fieldLength)).toString()
 
       if (noValue) {
         step = lineLength
-      } else if (buf[pos + fieldLength + 1] !== ' ') {
+      } else if (buf[pos + fieldLength + 1] !== space) {
         step = fieldLength + 1
       } else {
         step = fieldLength + 2
       }
       pos += step
 
       var valueLength = lineLength - step
-      var value = buf.slice(pos, pos + valueLength)
+      var value = buf.slice(pos, pos + valueLength).toString()
 
       if (field === 'data') {
         data += value + '\n'

package.json

@@ -30,6 +30,7 @@
     }
   ],
   "devDependencies": {
+    "buffer-from": "^1.1.1",
     "express": "^4.15.3",
     "mocha": "^3.5.3",
     "nyc": "^11.2.1",

test/eventsource_test.js

@@ -1,5 +1,6 @@
 /* eslint-disable no-new */
 var EventSource = require('../lib/eventsource')
+var bufferFrom = require('buffer-from')
 var path = require('path')
 var http = require('http')
 var https = require('https')
@@ -167,6 +168,24 @@ describe('Parser', function () {
     })
   })
 
+  it('ignores byte-order mark', function (done) {
+    createServer(function (err, server) {
+      if (err) return done(err)
+
+      server.on('request', function (req, res) {
+        res.writeHead(200, {'Content-Type': 'text/event-stream'})
+        res.write('\uFEFF')
+        res.write('data: foo\n\n')
+        res.end()
+      })
+      var es = new EventSource(server.url)
+      es.onmessage = function (m) {
+        assert.equal('foo', m.data)
+        server.close(done)
+      }
+    })
+  })
+
   it('parses one one-line message in two chunks', function (done) {
     createServer(function (err, server) {
       if (err) return done(err)
@@ -215,7 +234,7 @@ describe('Parser', function () {
     })
   })
 
-  it('parses really chopped up unicode data', function (done) {
+  it('parses chopped up unicode data', function (done) {
     createServer(function (err, server) {
       if (err) return done(err)
 
@@ -237,6 +256,30 @@ describe('Parser', function () {
     })
   })
 
+  it('parses really chopped up unicode data', function (done) {
+    createServer(function (err, server) {
+      if (err) return done(err)
+
+      server.on('request', function (req, res) {
+        const msg = bufferFrom('data: Aslak Hellesøy is the original author\n\n')
+        res.writeHead(200, {'Content-Type': 'text/event-stream'})
+
+        // Slice in the middle of a unicode sequence (ø), making sure that one data
+        // chunk will contain the first byte and the second chunk will get the other
+        res.write(msg.slice(0, 19), 'binary', function () {
+          res.write(msg.slice(19))
+        })
+      })
+
+      var es = new EventSource(server.url)
+
+      es.onmessage = function (m) {
+        assert.equal('Aslak Hellesøy is the original author', m.data)
+        server.close(done)
+      }
+    })
+  })
+
   it('accepts CRLF as separator', function (done) {
     createServer(function (err, server) {
       if (err) return done(err)