Parsing large files/xml (#149)

* bug: parsing large files/xml

* tweak: more test cases

Author: vdjagilev

formatter/file.go

@@ -1,8 +1,6 @@
 package formatter
 
 import (
-	"bufio"
-	"errors"
 	"io"
 	"os"
 )
@@ -20,21 +18,6 @@ type InputFileConfig struct {
 	Source  io.ReadCloser
 }
 
-// ReadContents reads content from stdin or provided file-path
-func (i *InputFileConfig) ReadContents() ([]byte, error) {
-	var err error
-	var content []byte
-	if i.Source == nil {
-		return nil, errors.New("no reading source is defined")
-	}
-	scanner := bufio.NewScanner(i.Source)
-	for scanner.Scan() {
-		content = append(content, scanner.Bytes()...)
-	}
-	err = scanner.Err()
-	return content, err
-}
-
 // ExistsOpen tries to open a file for reading, returning an error if it fails
 func (i *InputFileConfig) ExistsOpen() error {
 	f, err := os.Open(i.Path)

formatter/workflow.go

@@ -86,12 +86,15 @@ func (w *MainWorkflow) Execute() (err error) {
 
 // parse reads & unmarshalles the input file into NMAPRun struct
 func (w *MainWorkflow) parse() (run NMAPRun, err error) {
-	input, err := w.Config.InputFileConfig.ReadContents()
-	if err != nil {
-		return
+	if w.Config.InputFileConfig.Source == nil {
+		return run, fmt.Errorf("no input file is defined")
 	}
-	if err = xml.Unmarshal(input, &run); err != nil {
+	d := xml.NewDecoder(w.Config.InputFileConfig.Source)
+	_, err = d.Token()
+	if err != nil {
 		return
 	}
-	return run, nil
+
+	err = d.Decode(&run)
+	return
 }

formatter/workflow_test.go

@@ -50,6 +50,16 @@ func TestMainWorkflow_parse(t *testing.T) {
 			<nmaprun></nmaprun>`,
 			fileName: "main_workflow_parse_3_test",
 		},
+		{
+			name: "Bad XML file",
+			w: &MainWorkflow{
+				Config: &Config{},
+			},
+			wantNMAPRun: NMAPRun{},
+			wantErr:     true,
+			fileContent: "<?x< version=",
+			fileName:    "main_workflow_parse_4_test_wrong_xml",
+		},
 		{
 			name: "XML file with some matching output",
 			w: &MainWorkflow{