Metadata: Improve data parsing and sanitization #2897

Signed-off-by: Michael Mayer <michael@photoprism.app>

internal/meta/duration.go

@@ -16,6 +16,11 @@ func StringToDuration(s string) (d time.Duration) {
 	}
 
 	s = strings.TrimSpace(s)
+
+	if s == "" {
+		return d
+	}
+
 	sec := DurationSecondsRegexp.FindAllString(s, -1)
 
 	if len(sec) == 1 {

internal/meta/exif.go

@@ -80,9 +80,11 @@ func (data *Data) Exif(fileName string, fileFormat fs.Type, bruteForce bool) (er
 	// Ignore IFD1 data.exif with existing IFD0 values.
 	// see https://github.com/photoprism/photoprism/issues/2231
 	for _, tag := range entries {
-		s := strings.Split(tag.FormattedFirst, "\x00")[0]
-		if tag.TagName != "" && s != "" && (data.exif[tag.TagName] == "" || tag.IfdPath != exif.ThumbnailFqIfdPath) {
-			data.exif[tag.TagName] = s
+		s := strings.Split(tag.FormattedFirst, "\x00")
+		if tag.TagName == "" || len(s) == 0 {
+			// Do nothing.
+		} else if s[0] != "" && (data.exif[tag.TagName] == "" || tag.IfdPath != exif.ThumbnailFqIfdPath) {
+			data.exif[tag.TagName] = s[0]
 		}
 	}
 
@@ -96,13 +98,13 @@ func (data *Data) Exif(fileName string, fileFormat fs.Type, bruteForce bool) (er
 
 	// Find and parse GPS coordinates.
 	if err != nil {
-		log.Debugf("metadata: %s in %s (exif)", err, logName)
+		log.Debugf("metadata: %s in %s (exif collect)", err, logName)
 	} else {
 		var ifd *exif.Ifd
 		if ifd, err = ifdIndex.RootIfd.ChildWithIfdPath(exifcommon.IfdGpsInfoStandardIfdIdentity); err == nil {
 			var gi *exif.GpsInfo
 			if gi, err = ifd.GpsInfo(); err != nil {
-				log.Infof("metadata: %s in %s (exif)", err, logName)
+				log.Debugf("metadata: %s in %s (exif gps-info)", err, logName)
 			} else {
 				if !math.IsNaN(gi.Latitude.Decimal()) && !math.IsNaN(gi.Longitude.Decimal()) {
 					data.Lat = float32(gi.Latitude.Decimal())
@@ -201,8 +203,8 @@ func (data *Data) Exif(fileName string, fileFormat fs.Type, bruteForce bool) (er
 		if i, err := strconv.Atoi(value); err == nil {
 			data.FocalLength = i
 		}
-	} else if value, ok := data.exif["FocalLength"]; ok {
-		values := strings.Split(value, "/")
+	} else if v, ok := data.exif["FocalLength"]; ok {
+		values := strings.Split(v, "/")
 
 		if len(values) == 2 && values[1] != "0" && values[1] != "" {
 			number, _ := strconv.ParseFloat(values[0], 64)

internal/meta/json_exiftool.go

@@ -46,7 +46,7 @@ func (data *Data) Exiftool(jsonData []byte, originalName string) (err error) {
 	jsonValues := j.Map()
 
 	for key, val := range jsonValues {
-		data.json[key] = SanitizeUnicode(val.String())
+		data.json[key] = val.String()
 	}
 
 	if fileName, ok := data.json["FileName"]; ok && fileName != "" && originalName != "" && fileName != originalName {

internal/meta/sanitize.go

@@ -78,19 +78,23 @@ func SanitizeString(s string) string {
 }
 
 // SanitizeUID normalizes unique IDs found in XMP or Exif metadata.
-func SanitizeUID(value string) string {
-	value = SanitizeString(value)
+func SanitizeUID(s string) string {
+	s = SanitizeString(s)
 
-	if start := strings.LastIndex(value, ":"); start != -1 {
-		value = value[start+1:]
+	if len(s) < 15 {
+		return ""
+	}
+
+	if start := strings.LastIndex(s, ":"); start != -1 {
+		s = s[start+1:]
 	}
 
 	// Not a unique ID?
-	if len(value) < 15 || len(value) > 36 {
-		value = ""
+	if len(s) < 15 || len(s) > 36 {
+		s = ""
 	}
 
-	return strings.ToLower(value)
+	return strings.ToLower(s)
 }
 
 // SanitizeTitle normalizes titles and removes unwanted information.
@@ -122,22 +126,25 @@ func SanitizeTitle(title string) string {
 func SanitizeDescription(s string) string {
 	s = SanitizeString(s)
 
-	if s == "" {
+	switch {
+	case s == "":
 		return ""
-	} else if remove := UnwantedDescriptions[s]; remove {
-		s = ""
-	} else if strings.HasPrefix(s, "DCIM\\") && !strings.Contains(s, " ") {
-		s = ""
+	case UnwantedDescriptions[s]:
+		return ""
+	case strings.HasPrefix(s, "DCIM\\") && !strings.Contains(s, " "):
+		return ""
+	default:
+		return s
 	}
-
-	return s
 }
 
 // SanitizeMeta normalizes metadata fields that may contain JSON arrays like keywords and subject.
 func SanitizeMeta(s string) string {
 	if s == "" {
 		return ""
-	} else if strings.HasPrefix(s, "[") && strings.HasSuffix(s, "]") {
+	}
+
+	if strings.HasPrefix(s, "[") && strings.HasSuffix(s, "]") {
 		var words []string
 
 		if err := json.Unmarshal([]byte(s), &words); err != nil {

pkg/clean/unicode_test.go

@@ -10,6 +10,12 @@ 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("Null", func(t *testing.T) {
+		assert.Equal(t, "\x00", Unicode("\x00"))
+	})
+	t.Run("FFFD", func(t *testing.T) {
+		assert.Equal(t, "", Unicode("\uFFFD"))
+	})
 	t.Run("Empty", func(t *testing.T) {
 		assert.Equal(t, "", Unicode(""))
 	})