handle split logs when data stream is large

livelog/extractor/writer.go

@@ -7,41 +7,49 @@ import (
 	"regexp"
 )
 
-const Esc = "\u001B"
-
 var (
-	prefix = Esc + "]1338;"
+	prefix = []byte("\u001B]1338;")
+	suffix = []byte("\u001B]0m")
 	re     = regexp.MustCompilePOSIX("\u001B]1338;((.*?)\u001B]0m)")
 )
 
 type Writer struct {
-	base io.Writer
-	file []byte
+	base    io.Writer
+	file    []byte
+	chunked bool
 }
 
 func New(w io.Writer) *Writer {
-	return &Writer{w, nil}
+	return &Writer{w, nil, false}
 }
 
 func (e *Writer) Write(p []byte) (n int, err error) {
-	if bytes.HasPrefix(p, []byte(prefix)) == false {
+	if bytes.HasPrefix(p, prefix) == false && e.chunked == false {
 		return e.base.Write(p)
 	}
-	card := re.FindStringSubmatch(string(p))
-	if len(card) != 0 {
-		data, err := base64.StdEncoding.DecodeString(card[len(card)-1:][0])
-		if err == nil {
-			e.file = data
-		}
-		return e.base.Write([]byte(""))
-	}
-	return e.base.Write(p)
+
+	// if the data does not include the ansi suffix,
+	// it exceeds the size of the buffer and is chunked.
+	e.chunked = !bytes.Contains(p, suffix)
+
+	// trim the ansi prefix and suffix from the data,
+	// and also trim any spacing or newlines that could
+	// cause confusion.
+	p = bytes.TrimSpace(p)
+	p = bytes.TrimPrefix(p, prefix)
+	p = bytes.TrimSuffix(p, suffix)
+
+	e.file = append(e.file, p...)
+	return n, nil
 }
 
 func (e *Writer) File() ([]byte, bool) {
-	if len(e.file) > 0 {
-		return e.file, true
-	} else {
+	if len(e.file) == 0 {
+		return nil, false
+	}
+	data, err := base64.StdEncoding.DecodeString(string(e.file))
+	if err != nil {
 		return nil, false
 	}
+	return data, true
 }