diff --git a/main.go b/main.go
index 6c3ce26..25b1169 100644
--- a/main.go
+++ b/main.go
@@ -12,6 +12,7 @@
 	"regexp"
 	"runtime"
 	"strconv"
+	"strings"
 	"sync"
 	"time"
 
@@ -76,23 +77,28 @@
 					mimeEncoding: mimeEncoding,
 					size:         file.Size(),
 				}
-				if extension == ".pdf" {
-					wg.Add(1)
-					go processPDF(&fileData)
-				} else if extension == ".xml" {
+				if *configurationFolder == "" {
+					if extension == ".pdf" {
+						wg.Add(1)
+						go processPDF(&fileData)
+					} else if extension == ".xml" && strings.Contains(fileData.mimetype, "text") {
+						wg.Add(1)
+						go processXML(&fileData)
+					} else {
+						logrus.WithFields(logrus.Fields{
+							"corpusName":   fileData.corpusName,
+							"name":         fileData.name,
+							"startAt":      fileData.startAt,
+							"extension":    fileData.extension,
+							"path":         fileData.path,
+							"mimetype":     fileData.mimetype,
+							"mimeEncoding": fileData.mimeEncoding,
+							"size":         fileData.size,
+						}).Info("")
+					}
+				} else {
 					wg.Add(1)
 					go processXML(&fileData)
-				} else {
-					logrus.WithFields(logrus.Fields{
-						"corpusName":   fileData.corpusName,
-						"name":         fileData.name,
-						"startAt":      fileData.startAt,
-						"extension":    fileData.extension,
-						"path":         fileData.path,
-						"mimetype":     fileData.mimetype,
-						"mimeEncoding": fileData.mimeEncoding,
-						"size":         fileData.size,
-					}).Info("")
 				}
 			}
 		}
@@ -112,7 +118,7 @@
 	}
 
 	if *withWordCount == true {
-		fmt.Println("Count number word is enabled")
+		fmt.Println(color.InBlue("Count number word is enabled"))
 	}
 	// init logger
 	sec := time.Now().Unix()
@@ -126,7 +132,10 @@
 		fmt.Println(err)
 		os.Exit(1)
 	}
-	logrus.SetFormatter(&logrus.JSONFormatter{})
+	logrus.SetFormatter(&logrus.JSONFormatter{
+		DisableHTMLEscape: true,
+		DisableTimestamp:  true,
+	})
 	logrus.SetOutput(outputFile)
 }
 
diff --git a/xml.go b/xml.go
index 5f622ab..d5cdc8a 100644
--- a/xml.go
+++ b/xml.go
@@ -10,7 +10,7 @@
 )
 
 var regexDoctype = regexp.MustCompile("<!DOCTYPE.*([[sS]*?])?>")
-var regexDtd = regexp.MustCompile("[a-zA-Z0-9-]*.dtd")
+var regexDtd = regexp.MustCompile("[a-zA-Z0-9-_./:]*.dtd")
 
 func processXML(message *LogMessage) {
 	// queue for read xml (limit number of parallel read files)