Metadata: Improve titles generated from file and folder names #3447

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

internal/entity/photo_title_test.go

@@ -142,7 +142,7 @@ func TestPhoto_UpdateTitle(t *testing.T) {
 		if err != nil {
 			t.Fatal(err)
 		}
-		assert.Equal(t, "Photo / Germany / 2016", m.PhotoTitle)
+		assert.Equal(t, "longlonglonglonglongcity / 2016", m.PhotoTitle)
 	})
 	t.Run("photo with location without loc name and short city", func(t *testing.T) {
 		m := PhotoFixtures.Get("Photo12")
@@ -152,7 +152,7 @@ func TestPhoto_UpdateTitle(t *testing.T) {
 		if err != nil {
 			t.Fatal(err)
 		}
-		assert.Equal(t, "Photo / shortcity / 2016", m.PhotoTitle)
+		assert.Equal(t, "shortcity / Germany / 2016", m.PhotoTitle)
 	})
 	t.Run("no location original name", func(t *testing.T) {
 		m := PhotoFixtures.Get("19800101_000002_D640C559")
@@ -188,7 +188,7 @@ func TestPhoto_UpdateTitle(t *testing.T) {
 		if len(m.SubjectNames()) > 0 {
 			assert.Equal(t, "Actress A / 1990", m.PhotoTitle)
 		} else {
-			assert.Equal(t, "Bridge / 1990", m.PhotoTitle)
+			assert.Equal(t, "Bridge1 / 1990", m.PhotoTitle)
 		}
 	})
 	t.Run("no location no labels no takenAt", func(t *testing.T) {
@@ -199,7 +199,7 @@ func TestPhoto_UpdateTitle(t *testing.T) {
 		if err != nil {
 			t.Fatal(err)
 		}
-		assert.Equal(t, "Photo", m.PhotoTitle)
+		assert.Equal(t, "Photo20", m.PhotoTitle)
 	})
 	t.Run("OnePerson", func(t *testing.T) {
 		m := PhotoFixtures.Get("Photo10")

pkg/txt/file_title.go

@@ -7,7 +7,7 @@ import (
 	"github.com/photoprism/photoprism/pkg/fs"
 )
 
-var FileTitleRegexp = regexp.MustCompile("[\\p{L}\\-,':&+!?！]{1,}|( [&+] )?")
+var FileTitleRegexp = regexp.MustCompile("[\\p{L}\\d,':&+!?！]{1,}[\\-]{0,}|( [&+] )?")
 
 // FileTitle returns the string with the first characters of each word converted to uppercase.
 func FileTitle(s string) string {
@@ -23,7 +23,12 @@ func FileTitle(s string) string {
 	found := 0
 
 	for _, w := range words {
-		w = strings.ToLower(w)
+		// Ignore purely numeric values, such as those in timestamps.
+		if found == 0 && IsDateNumber(w) {
+			continue
+		} else if IsNumeric(w) {
+			continue
+		}
 
 		// Ignore ASCII strings < 3 characters at the beginning.
 		if IsASCII(w) && (len(w) < 3 && found == 0 || len(w) == 1) {
@@ -39,7 +44,7 @@ func FileTitle(s string) string {
 
 		found++
 
-		if found > 10 {
+		if found > 16 {
 			break
 		}
 	}
@@ -59,7 +64,7 @@ func FileTitle(s string) string {
 		title = strings.TrimSuffix(title, " "+w)
 	}
 
-	if len(title) <= 4 && IsASCII(title) {
+	if len(title) <= 4 && IsASCII(title) || IsDateNumber(title) {
 		return ""
 	}
 

pkg/txt/file_title_test.go

@@ -32,10 +32,10 @@ func TestFileTitle(t *testing.T) {
 		assert.Equal(t, "Photo Lover", FileTitle("photo-lover"))
 	})
 	t.Run("nyc", func(t *testing.T) {
-		assert.Equal(t, "Bridge in, or by, NYC", FileTitle("BRIDGE in, or by, nyc"))
+		assert.Equal(t, "BRIDGE in, or by, NYC", FileTitle("BRIDGE in, or by, nyc"))
 	})
 	t.Run("apple", func(t *testing.T) {
-		assert.Equal(t, "Phil Unveils iPhone, iPad, iPod, 'airpods', Airpod, AirPlay, iMac or MacBook", FileTitle("phil unveils iphone, ipad, ipod, 'airpods', airpod, airplay, imac or macbook 11 pro and max"))
+		assert.Equal(t, "Phil Unveils iPhone, iPad, iPod, 'airpods', Airpod, AirPlay, iMac or MacBook Pro and Max", FileTitle("phil unveils iphone, ipad, ipod, 'airpods', airpod, airplay, imac or macbook 11 pro and max"))
 	})
 	t.Run("IMG_4568", func(t *testing.T) {
 		assert.Equal(t, "", FileTitle("IMG_4568"))
@@ -53,7 +53,7 @@ func TestFileTitle(t *testing.T) {
 		assert.Equal(t, "Tim Robbins / TIFF", FileTitle("tim-robbins--tiff-2012_7999233420_o.jpg"))
 	})
 	t.Run("20200102-204030-Berlin-Germany-2020-3h4.jpg", func(t *testing.T) {
-		assert.Equal(t, "Berlin Germany", FileTitle("20200102-204030-Berlin-Germany-2020-3h4.jpg"))
+		assert.Equal(t, "Berlin Germany 2020", FileTitle("20200102-204030-Berlin-Germany-2020-3h4.jpg"))
 	})
 	t.Run("changing-of-the-guard--buckingham-palace_7925318070_o.jpg", func(t *testing.T) {
 		assert.Equal(t, "Changing of the Guard / Buckingham Palace", FileTitle("changing-of-the-guard--buckingham-palace_7925318070_o.jpg"))
@@ -79,10 +79,10 @@ func TestFileTitle(t *testing.T) {
 		assert.Equal(t, "Cyka Swappable Mag", FileTitle("Cyka - swappable mag (82405706) .jpg"))
 	})
 	t.Run("issue_361_d", func(t *testing.T) {
-		assert.Equal(t, "Dishwasher Friedrich the Smol", FileTitle("dishwasher1910 - Friedrich the smol (82201574) 1ページ.jpg"))
+		assert.Equal(t, "Dishwasher1910 Friedrich the Smol", FileTitle("dishwasher1910 - Friedrich the smol (82201574) 1ページ.jpg"))
 	})
 	t.Run("issue_361_e", func(t *testing.T) {
-		assert.Equal(t, "Eaycddvu Aafuur", FileTitle("EaycddvU0AAfuUR.jpg"))
+		assert.Equal(t, "EaycddvU0AAfuUR", FileTitle("EaycddvU0AAfuUR.jpg"))
 	})
 	t.Run("Eigene Bilder 1013/2007/oldies/neumühle", func(t *testing.T) {
 		// TODO: Normalize strings, see https://godoc.org/golang.org/x/text/unicode/norm
@@ -92,9 +92,11 @@ func TestFileTitle(t *testing.T) {
 		assert.Equal(t, "Neumühle", FileTitle("Neumühle"))
 	})
 	t.Run("IQVG4929", func(t *testing.T) {
-		assert.Equal(t, "", FileTitle("IQVG4929.jpg"))
+		assert.Equal(t, "IQVG4929", FileTitle("IQVG4929.jpg"))
+	})
+	t.Run("IMG_1234", func(t *testing.T) {
+		assert.Equal(t, "", FileTitle("IMG_1234.jpg"))
 	})
-
 	t.Run("du,-ich,-er, Sie und es", func(t *testing.T) {
 		assert.Equal(t, "Du, Ich, Er, Sie und Es", FileTitle("du,-ich,-er, Sie und es"))
 	})
@@ -114,7 +116,7 @@ func TestFileTitle(t *testing.T) {
 		assert.Equal(t, "Boston New Year's", FileTitle("boston new year's"))
 	})
 	t.Run("Screenshot", func(t *testing.T) {
-		assert.Equal(t, "Screenshot", FileTitle("Screenshot 2020-05-04 at 14:25:01.jpeg"))
+		assert.Equal(t, "Screenshot 2020 05", FileTitle("Screenshot 2020-05-04 at 14:25:01.jpeg"))
 	})
 	t.Run("HD", func(t *testing.T) {
 		assert.Equal(t, "Desktop Nebula HD Wallpapers", FileTitle("Desktop-Nebula-hd-Wallpapers.jpeg"))
@@ -125,4 +127,25 @@ func TestFileTitle(t *testing.T) {
 	t.Run("ImgNonCommercialPics", func(t *testing.T) {
 		assert.Equal(t, "Non Commercial Pics", FileTitle("Img Non Commercial Pics"))
 	})
+	t.Run("Birthday", func(t *testing.T) {
+		assert.Equal(t, "40th Birthday in Berlin", FileTitle("2024-10-23 40th Birthday in Berlin.jpg"))
+	})
+	t.Run("February2nd", func(t *testing.T) {
+		assert.Equal(t, "February 2nd", FileTitle("2024-10-23 February 2nd.jpg"))
+	})
+	t.Run("Boeing737", func(t *testing.T) {
+		assert.Equal(t, "Boeing", FileTitle("Boeing 737.jpg"))
+	})
+	t.Run("Boeing747-8F", func(t *testing.T) {
+		assert.Equal(t, "Boeing 747 8F", FileTitle("Boeing 747-8F.jpg"))
+	})
+	t.Run("Boeing747-100SR", func(t *testing.T) {
+		assert.Equal(t, "Boeing 747 100SR", FileTitle("Boeing 747-100SR.jpg"))
+	})
+	t.Run("Apostrophe", func(t *testing.T) {
+		assert.Equal(t, "Ma'am", FileTitle("Ma'am"))
+	})
+	t.Run("Download", func(t *testing.T) {
+		assert.Equal(t, "Tourist Attraction Berlin", FileTitle("20170812-185131-Tourist-Attraction-Berlin-2017.jpg"))
+	})
 }

pkg/txt/is.go

@@ -27,6 +27,55 @@ func IsASCII(s string) bool {
 	return true
 }
 
+// IsNumeric tests if the string only starts and ends with an ascii number character.
+func IsNumeric(s string) bool {
+	if s == "" {
+		return false
+	}
+
+	if s[0] < 48 || s[0] > 57 {
+		return false
+	}
+
+	if l := len(s); l < 2 {
+		return false
+	} else if s[l-1] < 48 || s[l-1] > 57 {
+		return false
+	}
+
+	return true
+}
+
+// IsNumber tests if the string only contains ascii number characters.
+func IsNumber(s string) bool {
+	if s == "" {
+		return false
+	}
+
+	for i := 0; i < len(s); i++ {
+		if s[i] < 48 || s[i] > 57 {
+			return false
+		}
+	}
+
+	return true
+}
+
+// IsDateNumber tests if the string only contains numeric characters, common delimiters like "-" and "_".
+func IsDateNumber(s string) bool {
+	if s == "" {
+		return false
+	}
+
+	for i := 0; i < len(s); i++ {
+		if (s[i] < 48 || s[i] > 57) && s[i] != '_' && s[i] != '-' {
+			return false
+		}
+	}
+
+	return true
+}
+
 // IsLatin reports whether the string only contains latin letters.
 func IsLatin(s string) bool {
 	if s == "" {

pkg/txt/is_test.go

@@ -68,6 +68,66 @@ func TestIsASCII(t *testing.T) {
 	})
 }
 
+func TestIsNumeric(t *testing.T) {
+	t.Run("Empty", func(t *testing.T) {
+		assert.False(t, IsNumeric(""))
+	})
+	t.Run("123", func(t *testing.T) {
+		assert.True(t, IsNumeric("123"))
+	})
+	t.Run("123.", func(t *testing.T) {
+		assert.False(t, IsNumeric("123."))
+	})
+	t.Run("2024-10-23", func(t *testing.T) {
+		assert.True(t, IsNumeric("2024-10-23"))
+	})
+	t.Run("20200102-204030", func(t *testing.T) {
+		assert.True(t, IsNumeric("20200102-204030"))
+	})
+	t.Run("ABC", func(t *testing.T) {
+		assert.False(t, IsNumeric("ABC"))
+	})
+}
+
+func TestIsNumber(t *testing.T) {
+	t.Run("Empty", func(t *testing.T) {
+		assert.False(t, IsNumber(""))
+	})
+	t.Run("123", func(t *testing.T) {
+		assert.True(t, IsNumber("123"))
+	})
+	t.Run("123.", func(t *testing.T) {
+		assert.False(t, IsNumber("123."))
+	})
+	t.Run("2024-10-23", func(t *testing.T) {
+		assert.False(t, IsNumber("2024-10-23"))
+	})
+	t.Run("ABC", func(t *testing.T) {
+		assert.False(t, IsNumber("ABC"))
+	})
+}
+
+func TestIsDateNumber(t *testing.T) {
+	t.Run("Empty", func(t *testing.T) {
+		assert.False(t, IsDateNumber(""))
+	})
+	t.Run("123", func(t *testing.T) {
+		assert.True(t, IsDateNumber("123"))
+	})
+	t.Run("123.", func(t *testing.T) {
+		assert.False(t, IsDateNumber("123."))
+	})
+	t.Run("2024-10-23", func(t *testing.T) {
+		assert.True(t, IsDateNumber("2024-10-23"))
+	})
+	t.Run("20200102-204030", func(t *testing.T) {
+		assert.True(t, IsDateNumber("20200102-204030"))
+	})
+	t.Run("ABC", func(t *testing.T) {
+		assert.False(t, IsDateNumber("ABC"))
+	})
+}
+
 func TestIsLatin(t *testing.T) {
 	t.Run("The quick brown fox.", func(t *testing.T) {
 		assert.False(t, IsLatin("The quick brown fox."))