Merge 0efdb40d57 into 8960e66bd3

reader.go

@@ -5,6 +5,8 @@ import (
 	"syscall"
 )
 
+const bufferSize = 8 * 1024
+
 type Reader struct {
 	source            io.Reader
 	converter         *Converter
@@ -33,12 +35,12 @@ func NewReaderFromConverter(source io.Reader, converter *Converter) (reader *Rea
 	reader.converter = converter
 
 	// create 8K buffers
-	reader.buffer = make([]byte, 8*1024)
+	reader.buffer = make([]byte, bufferSize)
 
 	return reader
 }
 
-func (this *Reader) fillBuffer() {
+func (this *Reader) fillBuffer() int {
 	// slide existing data to beginning
 	if this.readPos > 0 {
 		// copy current bytes - is this guaranteed safe?
@@ -59,6 +61,7 @@ func (this *Reader) fillBuffer() {
 	if err != nil {
 		this.err = err
 	}
+	return bytesRead
 }
 
 // implement the io.Reader interface
@@ -85,6 +88,18 @@ func (this *Reader) Read(p []byte) (n int, err error) {
 
 	// if we experienced an iconv error, check it
 	if err != nil {
+		// EINVAL:
+		// An incomplete multibyte sequence is encountered in the input,
+		// and the input byte sequence terminates after it.
+		if err == syscall.EINVAL {
+			// If we can read new data, then this should NOT be
+			// considered as an error.
+			newData := this.fillBuffer()
+			if newData > 0 {
+				return n, nil
+			}
+		}
+
 		// E2BIG errors can be ignored (we'll get them often) as long
 		// as at least 1 byte was written. If we experienced an E2BIG
 		// and no bytes were written then the buffer is too small for
@@ -92,6 +107,13 @@ func (this *Reader) Read(p []byte) (n int, err error) {
 		if err != syscall.E2BIG || bytesWritten == 0 {
 			// track anything else
 			this.err = err
+		} else {
+			// Should not return this.err
+			// If we got EOF from source in last fillBuffer() call, and
+			// there is still more data to process in buffer, in this
+			// case, if we return this.err(=EOF), then data in buffer
+			// will be lost.
+			return n, nil
 		}
 	}
 

reader_test.go

@@ -0,0 +1,38 @@
+package iconv
+
+import (
+	"bytes"
+	"io/ioutil"
+	"testing"
+)
+
+func GbkToUtf8(src []byte) ([]byte, error) {
+	reader, err := NewReader(bytes.NewReader(src), "gbk", "utf-8")
+	if err != nil {
+		return nil, err
+	}
+	return ioutil.ReadAll(reader)
+}
+
+func Utf8ToGbk(src []byte) ([]byte, error) {
+	reader, err := NewReader(bytes.NewReader(src), "utf-8", "gbk")
+	reader.buffer = make([]byte, 16)
+	if err != nil {
+		return nil, err
+	}
+	return ioutil.ReadAll(reader)
+}
+
+func TestReaderWithDataLargerThanBuffer(t *testing.T) {
+	chars := []byte("梅")
+	for len(chars) < bufferSize*2 {
+		t.Logf("input size: %d", len(chars))
+		chars = append(chars, chars...)
+		_, err := Utf8ToGbk(chars)
+		if err != nil {
+			t.Fail()
+			t.Logf("failed with %d bytes data", len(chars))
+			return
+		}
+	}
+}