From ae130dc5007ba782dd7e2437c229322ae0b9d770 Mon Sep 17 00:00:00 2001
From: Michael Mayer <michael@photoprism.app>
Date: Tue, 15 Nov 2022 14:45:21 +0100
Subject: [PATCH] Metadata: Sanitize bad Unicode strings #2897

Signed-off-by: Michael Mayer <michael@photoprism.app>
---
 internal/meta/json_exiftool.go | 23 +++++++++++------------
 internal/meta/sanitize.go      |  3 ++-
 pkg/clean/unicode.go           | 21 +++++++++++++++++++++
 pkg/clean/unicode_test.go      | 16 ++++++++++++++++
 4 files changed, 50 insertions(+), 13 deletions(-)
 create mode 100644 pkg/clean/unicode.go
 create mode 100644 pkg/clean/unicode_test.go

diff --git a/internal/meta/json_exiftool.go b/internal/meta/json_exiftool.go
index a7ce1da55..23c56e31d 100644
--- a/internal/meta/json_exiftool.go
+++ b/internal/meta/json_exiftool.go
@@ -9,15 +9,14 @@ import (
 	"strings"
 	"time"
 
-	"github.com/photoprism/photoprism/pkg/video"
-
-	"github.com/photoprism/photoprism/pkg/projection"
-
-	"github.com/photoprism/photoprism/pkg/clean"
-	"github.com/photoprism/photoprism/pkg/rnd"
-	"github.com/photoprism/photoprism/pkg/txt"
 	"github.com/tidwall/gjson"
 	"gopkg.in/photoprism/go-tz.v2/tz"
+
+	"github.com/photoprism/photoprism/pkg/clean"
+	"github.com/photoprism/photoprism/pkg/projection"
+	"github.com/photoprism/photoprism/pkg/rnd"
+	"github.com/photoprism/photoprism/pkg/txt"
+	"github.com/photoprism/photoprism/pkg/video"
 )
 
 const MimeVideoMP4 = "video/mp4"
@@ -47,7 +46,7 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 	jsonValues := j.Map()
 
 	for key, val := range jsonValues {
-		data.json[key] = val.String()
+		data.json[key] = SanitizeString(val.String())
 	}
 
 	if fileName, ok := data.json["FileName"]; ok && fileName != "" && originalName != "" && fileName != originalName {
@@ -134,22 +133,22 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 				}
 			case []string:
 				existing := fieldValue.Interface().([]string)
-				fieldValue.Set(reflect.ValueOf(txt.AddToWords(existing, strings.TrimSpace(jsonValue.String()))))
+				fieldValue.Set(reflect.ValueOf(txt.AddToWords(existing, SanitizeString(jsonValue.String()))))
 			case Keywords:
 				existing := fieldValue.Interface().(Keywords)
-				fieldValue.Set(reflect.ValueOf(txt.AddToWords(existing, strings.TrimSpace(jsonValue.String()))))
+				fieldValue.Set(reflect.ValueOf(txt.AddToWords(existing, SanitizeString(jsonValue.String()))))
 			case projection.Type:
 				if !fieldValue.IsZero() {
 					continue
 				}
 
-				fieldValue.Set(reflect.ValueOf(projection.Type(strings.TrimSpace(jsonValue.String()))))
+				fieldValue.Set(reflect.ValueOf(projection.Type(SanitizeString(jsonValue.String()))))
 			case string:
 				if !fieldValue.IsZero() {
 					continue
 				}
 
-				fieldValue.SetString(strings.TrimSpace(jsonValue.String()))
+				fieldValue.SetString(SanitizeString(jsonValue.String()))
 			case bool:
 				if !fieldValue.IsZero() {
 					continue
diff --git a/internal/meta/sanitize.go b/internal/meta/sanitize.go
index 6dd72c01b..4f4247062 100644
--- a/internal/meta/sanitize.go
+++ b/internal/meta/sanitize.go
@@ -5,6 +5,7 @@ import (
 	"regexp"
 	"strings"
 
+	"github.com/photoprism/photoprism/pkg/clean"
 	"github.com/photoprism/photoprism/pkg/fs"
 	"github.com/photoprism/photoprism/pkg/txt"
 )
@@ -64,7 +65,7 @@ func SanitizeString(s string) string {
 
 	s = strings.TrimSpace(s)
 
-	return strings.Replace(s, "\"", "", -1)
+	return clean.Unicode(strings.Replace(s, "\"", "", -1))
 }
 
 // SanitizeUID normalizes unique IDs found in XMP or Exif metadata.
diff --git a/pkg/clean/unicode.go b/pkg/clean/unicode.go
new file mode 100644
index 000000000..490a4ed05
--- /dev/null
+++ b/pkg/clean/unicode.go
@@ -0,0 +1,21 @@
+package clean
+
+import "strings"
+
+// Unicode returns a string a valid unicode.
+func Unicode(s string) string {
+	if s == "" {
+		return ""
+	}
+
+	var b strings.Builder
+
+	for _, c := range s {
+		if c == '\uFFFD' {
+			continue
+		}
+		b.WriteRune(c)
+	}
+
+	return b.String()
+}
diff --git a/pkg/clean/unicode_test.go b/pkg/clean/unicode_test.go
new file mode 100644
index 000000000..62cad0f4b
--- /dev/null
+++ b/pkg/clean/unicode_test.go
@@ -0,0 +1,16 @@
+package clean
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestUnicode(t *testing.T) {
+	t.Run("Valid", func(t *testing.T) {
+		assert.Equal(t, "Naïve bonds and futures surge as inflation eases 🚀🚀🚀", Unicode("Naïve bonds and futures surge as inflation eases 🚀🚀🚀"))
+	})
+	t.Run("Empty", func(t *testing.T) {
+		assert.Equal(t, "", Unicode(""))
+	})
+}