Backend: Sanitize metadata titles and descriptions

Signed-off-by: Michael Mayer <michael@liquidbytes.net>

Makefile

@@ -109,6 +109,9 @@ acceptance:
 acceptance-firefox:
 	$(info Running JS acceptance tests in Firefox...)
 	(cd frontend &&	npm run acceptance-firefox && cd ..)
+reset-photoprism-db:
+	$(info Purging photoprism database...)
+	mysql < scripts/reset-photoprism-db.sql
 reset-test-db:
 	$(info Purging test databases...)
 	mysql < scripts/reset-test-db.sql

internal/entity/password.go

@@ -3,6 +3,7 @@ package entity
 import (
 	"time"
 
+	"github.com/photoprism/photoprism/pkg/txt"
 	"golang.org/x/crypto/bcrypt"
 )
 
@@ -17,14 +18,14 @@ type Password struct {
 // NewPassword creates a new password instance.
 func NewPassword(uid, password string) Password {
 	if uid == "" {
-		panic("password: uid must not be empty")
+		panic("auth: can't set password without uid")
 	}
 
 	m := Password{UID: uid}
 
 	if password != "" {
 		if err := m.SetPassword(password); err != nil {
-			log.Errorf("password: %s (set password)", err)
+			log.Errorf("auth: failed setting password for %s", uid)
 		}
 	}
 
@@ -68,7 +69,7 @@ func FindPassword(uid string) *Password {
 	if err := Db().Where("uid = ?", uid).First(&result).Error; err == nil {
 		return &result
 	} else {
-		log.Errorf("password: %s (not found)", err)
+		log.Errorf("auth: no password for %s", txt.Quote(uid))
 	}
 
 	return nil

internal/entity/person.go

@@ -7,6 +7,7 @@ import (
 	"github.com/jinzhu/gorm"
 	"github.com/photoprism/photoprism/internal/acl"
 	"github.com/photoprism/photoprism/pkg/rnd"
+	"github.com/photoprism/photoprism/pkg/txt"
 )
 
 type People []Person
@@ -133,7 +134,7 @@ func FindPersonByUserName(userName string) *Person {
 	if err := Db().Where("user_name = ?", userName).First(&result).Error; err == nil {
 		return &result
 	} else {
-		log.Errorf("user: %s", err)
+		log.Errorf("auth: user %s not found", txt.Quote(userName))
 		return nil
 	}
 }
@@ -149,7 +150,7 @@ func FindPersonByUID(uid string) *Person {
 	if err := Db().Where("person_uid = ?", uid).First(&result).Error; err == nil {
 		return &result
 	} else {
-		log.Errorf("user: %s", err)
+		log.Errorf("auth: user %s not found", txt.Quote(uid))
 		return nil
 	}
 }

internal/meta/data.go

@@ -15,7 +15,7 @@ type Data struct {
 	Duration     time.Duration `meta:"Duration,MediaDuration,TrackDuration"`
 	Codec        string        `meta:"CompressorID,Compression,FileType"`
 	Title        string        `meta:"Title"`
-	Subject      string        `meta:"Subject,PersonInImage"`
+	Subject      string        `meta:"Subject,PersonInImage,ObjectName"`
 	Keywords     string        `meta:"Keywords"`
 	Comment      string        `meta:"-"`
 	Artist       string        `meta:"Artist,Creator"`

internal/meta/exif.go

@@ -342,7 +342,7 @@ func (data *Data) Exif(fileName string) (err error) {
 	}
 
 	if value, ok := tags["ImageDescription"]; ok {
-		data.Description = SanitizeString(value)
+		data.Description = SanitizeDescription(value)
 	}
 
 	data.All = tags

internal/meta/json.go

@@ -187,5 +187,8 @@ func (data *Data) JSON(jsonName, originalName string) (err error) {
 		data.InstanceID = SanitizeUID(data.InstanceID)
 	}
 
+	data.Title = SanitizeTitle(data.Title)
+	data.Description = SanitizeDescription(data.Description)
+
 	return nil
 }

internal/meta/sanitize.go

@@ -1,9 +1,16 @@
 package meta
 
 import (
+	"regexp"
 	"strings"
 )
 
+var DscTitleRegexp = regexp.MustCompile("\\D{3}[\\d_]\\d{4}(.JPG)?")
+
+var UnwantedDescriptions = map[string]bool{
+	"OLYMPUS DIGITAL CAMERA": true,
+}
+
 // SanitizeString removes unwanted character from an exif value string.
 func SanitizeString(value string) string {
 	value = strings.TrimSpace(value)
@@ -25,3 +32,25 @@ func SanitizeUID(value string) string {
 
 	return strings.ToLower(value)
 }
+
+// SanitizeTitle normalizes titles and removes unwanted information.
+func SanitizeTitle(value string) string {
+	value = SanitizeString(value)
+
+	if dsc := DscTitleRegexp.FindString(value); dsc == value {
+		value = ""
+	}
+
+	return value
+}
+
+// SanitizeDescription normalizes descriptions and removes unwanted information.
+func SanitizeDescription(value string) string {
+	value = SanitizeString(value)
+
+	if remove := UnwantedDescriptions[value]; remove {
+		value = ""
+	}
+
+	return value
+}

internal/meta/sanitize_test.go

@@ -0,0 +1,55 @@
+package meta
+
+import "testing"
+
+func TestSanitizeTitle(t *testing.T) {
+	t.Run("IMG_0599", func(t *testing.T) {
+		result := SanitizeTitle("IMG_0599")
+
+		if result != "" {
+			t.Fatal("result should be empty")
+		}
+	})
+
+	t.Run("IMG_0599.JPG", func(t *testing.T) {
+		result := SanitizeTitle("IMG_0599.JPG")
+
+		if result != "" {
+			t.Fatal("result should be empty")
+		}
+	})
+
+	t.Run("IMG_0599 ABC", func(t *testing.T) {
+		result := SanitizeTitle("IMG_0599 ABC")
+
+		if result != "IMG_0599 ABC" {
+			t.Fatal("result should be IMG_0599 ABC")
+		}
+	})
+
+	t.Run("DSC10599", func(t *testing.T) {
+		result := SanitizeTitle("DSC10599")
+
+		if result != "" {
+			t.Fatal("result should be empty")
+		}
+	})
+}
+
+func TestSanitizeDescription(t *testing.T) {
+	t.Run("IMG_0599", func(t *testing.T) {
+		result := SanitizeDescription("IMG_0599")
+
+		if result == "" {
+			t.Fatal("result should not be empty")
+		}
+	})
+
+	t.Run("OLYMPUS DIGITAL CAMERA", func(t *testing.T) {
+		result := SanitizeDescription("OLYMPUS DIGITAL CAMERA")
+
+		if result != "" {
+			t.Fatal("result should be empty")
+		}
+	})
+}

internal/meta/xmp_document.go

@@ -203,29 +203,29 @@ func (doc *XmpDocument) Load(filename string) error {
 }
 
 func (doc *XmpDocument) Title() string {
-	return doc.RDF.Description.Title.Alt.Li.Text
+	return SanitizeTitle(doc.RDF.Description.Title.Alt.Li.Text)
 }
 
 func (doc *XmpDocument) Artist() string {
-	return doc.RDF.Description.Creator.Seq.Li
+	return SanitizeString(doc.RDF.Description.Creator.Seq.Li)
 }
 
 func (doc *XmpDocument) Description() string {
-	return doc.RDF.Description.Description.Alt.Li.Text
+	return SanitizeDescription(doc.RDF.Description.Description.Alt.Li.Text)
 }
 
 func (doc *XmpDocument) Copyright() string {
-	return doc.RDF.Description.Rights.Alt.Li.Text
+	return SanitizeString(doc.RDF.Description.Rights.Alt.Li.Text)
 }
 
 func (doc *XmpDocument) CameraMake() string {
-	return doc.RDF.Description.Make
+	return SanitizeString(doc.RDF.Description.Make)
 }
 
 func (doc *XmpDocument) CameraModel() string {
-	return doc.RDF.Description.Model
+	return SanitizeString(doc.RDF.Description.Model)
 }
 
 func (doc *XmpDocument) LensModel() string {
-	return doc.RDF.Description.LensModel
+	return SanitizeString(doc.RDF.Description.LensModel)
 }

scripts/reset-photoprism-db.sql

@@ -0,0 +1,2 @@
+DROP DATABASE IF EXISTS photoprism;
+CREATE DATABASE IF NOT EXISTS photoprism;