fix: Handle first chunk shorter than 7 bytes correctly when next chunks are larger (#7)

Author: sttk

index.js

@@ -4,28 +4,30 @@ var through = require('through2');
 var removeBom = require('remove-bom-buffer');
 
 function removeBomStream() {
-  var completed = false;
+  var state = 0; // 0:Not removed, -1:In removing, 1:Already removed
   var buffer = Buffer.alloc(0);
 
   return through(onChunk, onFlush);
 
   function removeAndCleanup(data) {
-    completed = true;
+    state = 1; // Already removed
 
     buffer = null;
 
     return removeBom(data);
   }
 
   function onChunk(data, enc, cb) {
-    if (completed) {
+    if (state === 1) {
       return cb(null, data);
     }
 
-    if (data.length >= 7) {
+    if (state === 0 /* Not removed */ && data.length >= 7) {
       return cb(null, removeAndCleanup(data));
     }
 
+    state = -1; // In removing
+
     var bufferLength = buffer.length;
     var chunkLength = data.length;
     var totalLength = bufferLength + chunkLength;
@@ -39,7 +41,7 @@ function removeBomStream() {
   }
 
   function onFlush(cb) {
-    if (completed || !buffer) {
+    if (state === 2 /* Already removed */ || !buffer) {
       return cb();
     }
 

test/index.js

@@ -29,7 +29,23 @@ describe('removeBomStream', function () {
     );
   });
 
-  it('removes the BOM from a UTF8 buffer', function (done) {
+  it('ignores UTF8 buffer without a BOM even if first chunk is shorter than 7 chars but second and subsequent are larger', function(done) {
+    var filepath = path.join(__dirname, './fixtures/test.txt');
+    var fileContent = fs.readFileSync(filepath, 'utf-8');
+
+    var rmBom = removeBomStream();
+    var output = '';
+    rmBom.on('data', function(d) {
+      output += d.toString();
+    });
+    rmBom.write(Buffer.from(fileContent.slice(0, 5)));
+    rmBom.write(Buffer.from(fileContent.slice(5)));
+
+    expect(output).toEqual(fileContent);
+    done();
+  });
+
+  it('removes the BOM from a UTF8 buffer', function(done) {
     var filepath = path.join(__dirname, './fixtures/bom-utf8.txt');
 
     var expected = fs.readFileSync(filepath).slice(3);
@@ -81,7 +97,23 @@ describe('removeBomStream', function () {
     );
   });
 
-  it('does not remove the BOM from a UTF16BE buffer', function (done) {
+  it('remove the BOM from a UTF8 buffer even if first chunk is shorter than 7 chars but second and subsequent are larger', function(done) {
+    var filepath = path.join(__dirname, './fixtures/bom-utf8.txt');
+    var fileContent = fs.readFileSync(filepath, 'utf-8');
+
+    var rmBom = removeBomStream();
+    var output = '';
+    rmBom.on('data', function(d) {
+      output += d.toString();
+    });
+    rmBom.write(Buffer.from(fileContent.slice(0, 5)));
+    rmBom.write(Buffer.from(fileContent.slice(5)));
+
+    expect(output).toEqual(fileContent.slice(1));
+    done();
+  });
+
+  it('does not remove the BOM from a UTF16BE buffer', function(done) {
     var filepath = path.join(__dirname, './fixtures/bom-utf16be.txt');
 
     var expected = fs.readFileSync(filepath);