From a70c57ffd1049356d9ad23e48b4249e0f4073381 Mon Sep 17 00:00:00 2001
From: Dave Henderson <dhenderson@gmail.com>
Date: Thu, 29 Nov 2018 11:31:48 -0500
Subject: [PATCH] textfile collector: Ensure that only UTF8 textfiles are
 parsed

Signed-off-by: Dave Henderson <dhenderson@gmail.com>
---
 collector/textfile.go      | 25 ++++++++++++++++++++++++-
 collector/textfile_test.go | 35 +++++++++++++++++++++++++++++++----
 2 files changed, 55 insertions(+), 5 deletions(-)

diff --git a/collector/textfile.go b/collector/textfile.go
index 02cb7a40..b1bd33c0 100644
--- a/collector/textfile.go
+++ b/collector/textfile.go
@@ -237,7 +237,12 @@ fileLoop:
 			continue
 		}
 		var parser expfmt.TextParser
-		r := utfbom.SkipOnly(carriageReturnFilteringReader{r: file})
+		r, encoding := utfbom.Skip(carriageReturnFilteringReader{r: file})
+		if err := checkBOM(encoding); err != nil {
+			log.Errorf("Invalid file encoding detected in %s: %s - file must be UTF8", path, err.Error())
+			error = 1.0
+			continue
+		}
 		parsedFamilies, err := parser.TextToMetricFamilies(r)
 		file.Close()
 		if err != nil {
@@ -281,3 +286,21 @@ fileLoop:
 	)
 	return nil
 }
+
+func checkBOM(encoding utfbom.Encoding) error {
+	bom := ""
+	switch encoding {
+	case utfbom.Unknown, utfbom.UTF8:
+		return nil
+	case utfbom.UTF16BigEndian:
+		bom = "UTF16BigEndian"
+	case utfbom.UTF16LittleEndian:
+		bom = "UTF16LittleEndian"
+	case utfbom.UTF32BigEndian:
+		bom = "UTF32BigEndian"
+	case utfbom.UTF32LittleEndian:
+		bom = "UTF32LittleEndian"
+	}
+
+	return fmt.Errorf(bom)
+}
diff --git a/collector/textfile_test.go b/collector/textfile_test.go
index 231aefff..d3717146 100644
--- a/collector/textfile_test.go
+++ b/collector/textfile_test.go
@@ -1,14 +1,15 @@
 package collector
 
 import (
-	"testing"
-	"strings"
+	"github.com/dimchansky/utfbom"
 	"io/ioutil"
+	"strings"
+	"testing"
 )
 
 func TestCRFilter(t *testing.T) {
 	sr := strings.NewReader("line 1\r\nline 2")
-	cr := carriageReturnFilteringReader{ r: sr }
+	cr := carriageReturnFilteringReader{r: sr}
 	b, err := ioutil.ReadAll(cr)
 	if err != nil {
 		t.Error(err)
@@ -17,4 +18,30 @@ func TestCRFilter(t *testing.T) {
 	if string(b) != "line 1\nline 2" {
 		t.Errorf("Unexpected output %q", b)
 	}
-}
\ No newline at end of file
+}
+
+func TestCheckBOM(t *testing.T) {
+	testdata := []struct {
+		encoding utfbom.Encoding
+		err      string
+	}{
+		{utfbom.Unknown, ""},
+		{utfbom.UTF8, ""},
+		{utfbom.UTF16BigEndian, "UTF16BigEndian"},
+		{utfbom.UTF16LittleEndian, "UTF16LittleEndian"},
+		{utfbom.UTF32BigEndian, "UTF32BigEndian"},
+		{utfbom.UTF32LittleEndian, "UTF32LittleEndian"},
+	}
+	for _, d := range testdata {
+		err := checkBOM(d.encoding)
+		if d.err == "" && err != nil {
+			t.Error(err)
+		}
+		if d.err != "" && err == nil {
+			t.Errorf("Missing expected error %s", d.err)
+		}
+		if err != nil && !strings.Contains(err.Error(), d.err) {
+			t.Error(err)
+		}
+	}
+}