fix(search): read the tika 4 live photo keys

Tika 4 prefixes its metadata keys (mp4:, img:), so the live photo facet was never built there. Live photo support needs tika 4: tika 3 got the video keys only in 3.3.2 and never the still image time. The content key is read through one lookup, marked tika 4 and tika 3 legacy.

services/search/pkg/content/tika.go

@@ -95,10 +95,10 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,
 			doc.Title = strings.TrimSpace(fmt.Sprintf("%s %s", doc.Title, title))
 		}
 
-		// tika 4 renamed the meta prefix from X-TIKA: to tk:
-		if content, err := getFirstValue(meta, "tk:content"); err == nil {
-			doc.Content = strings.TrimSpace(fmt.Sprintf("%s %s", doc.Content, content))
-		} else if content, err := getFirstValue(meta, "X-TIKA:content"); err == nil {
+		if content, err := getFirstValue(meta,
+			"tk:content",     // tika 4
+			"X-TIKA:content", // tika 3 legacy
+		); err == nil {
 			doc.Content = strings.TrimSpace(fmt.Sprintf("%s %s", doc.Content, content))
 		}
 	}

services/search/pkg/content/tika_live_photo.go

@@ -10,9 +10,8 @@ import (
 func (t Tika) getLivePhoto(meta map[string][]string) *libregraph.LivePhoto {
 	// ContentId pairs the two halves, without it there is no live photo. The
 	// video carries it in the QuickTime item list, the still image in the Apple
-	// maker note, which tika surfaces as "Content Identifier". A file is only
-	// ever one half, so both keys are read.
-	contentID, err := getFirstValue(meta, "com.apple.quicktime.content.identifier", "Content Identifier")
+	// maker note. A file is only ever one half, so both keys are read.
+	contentID, err := getFirstValue(meta, "mp4:com.apple.quicktime.content.identifier", "img:Content Identifier")
 	if err != nil || contentID == "" {
 		return nil
 	}
@@ -25,19 +24,19 @@ func (t Tika) getLivePhoto(meta map[string][]string) *libregraph.LivePhoto {
 		}
 	}
 
-	if v, err := getFirstValue(meta, "com.apple.quicktime.live-photo.auto"); err == nil {
+	if v, err := getFirstValue(meta, "mp4:com.apple.quicktime.live-photo.auto"); err == nil {
 		if b, err := strconv.ParseBool(v); err == nil {
 			livePhoto.SetAuto(b)
 		}
 	}
 
-	if v, err := getFirstValue(meta, "com.apple.quicktime.live-photo.vitality-score"); err == nil {
+	if v, err := getFirstValue(meta, "mp4:com.apple.quicktime.live-photo.vitality-score"); err == nil {
 		if f, err := strconv.ParseFloat(v, 64); err == nil {
 			livePhoto.SetVitalityScore(f)
 		}
 	}
 
-	if v, err := getFirstValue(meta, "com.apple.quicktime.live-photo.vitality-scoring-version"); err == nil {
+	if v, err := getFirstValue(meta, "mp4:com.apple.quicktime.live-photo.vitality-scoring-version"); err == nil {
 		if i, err := strconv.ParseInt(v, 10, 64); err == nil {
 			livePhoto.SetVitalityScoringVersion(i)
 		}

services/search/pkg/content/tika_live_photo_test.go

@@ -9,12 +9,12 @@ import (
 var _ = Describe("getLivePhoto", func() {
 	It("maps the video half of a live photo", func() {
 		livePhoto := Tika{}.getLivePhoto(map[string][]string{
-			"Content-Type":                                            {"video/quicktime"},
-			"com.apple.quicktime.content.identifier":                  {"6F1A2B3C-1234-4E5F-9A8B-0011223344CC"},
-			"quicktime:still-image-time":                              {"1500000"},
-			"com.apple.quicktime.live-photo.auto":                     {"1"},
-			"com.apple.quicktime.live-photo.vitality-score":           {"0.75"},
-			"com.apple.quicktime.live-photo.vitality-scoring-version": {"4"},
+			"Content-Type": {"video/quicktime"},
+			"mp4:com.apple.quicktime.content.identifier":                  {"6F1A2B3C-1234-4E5F-9A8B-0011223344CC"},
+			"quicktime:still-image-time":                                  {"1500000"},
+			"mp4:com.apple.quicktime.live-photo.auto":                     {"1"},
+			"mp4:com.apple.quicktime.live-photo.vitality-score":           {"0.75"},
+			"mp4:com.apple.quicktime.live-photo.vitality-scoring-version": {"4"},
 		})
 		Expect(livePhoto).ToNot(BeNil())
 
@@ -27,8 +27,8 @@ var _ = Describe("getLivePhoto", func() {
 
 	It("maps the still half via the Apple maker-note content identifier", func() {
 		livePhoto := Tika{}.getLivePhoto(map[string][]string{
-			"Content-Type":       {"image/heic"},
-			"Content Identifier": {"6F1A2B3C-1234-4E5F-9A8B-0011223344CC"},
+			"Content-Type":           {"image/heic"},
+			"img:Content Identifier": {"6F1A2B3C-1234-4E5F-9A8B-0011223344CC"},
 		})
 		Expect(livePhoto).ToNot(BeNil())
 		Expect(livePhoto.ContentId).To(Equal("6F1A2B3C-1234-4E5F-9A8B-0011223344CC"))
@@ -39,15 +39,15 @@ var _ = Describe("getLivePhoto", func() {
 			"Content-Type": {"image/jpeg"},
 		})).To(BeNil())
 		Expect(Tika{}.getLivePhoto(map[string][]string{
-			"Content-Type":       {"image/heic"},
-			"Content Identifier": {""},
+			"Content-Type":           {"image/heic"},
+			"img:Content Identifier": {""},
 		})).To(BeNil(), "an empty pairing id is no live photo")
 	})
 
 	It("rounds a fractional still-image-time", func() {
 		livePhoto := Tika{}.getLivePhoto(map[string][]string{
-			"com.apple.quicktime.content.identifier": {"6F1A2B3C"},
-			"quicktime:still-image-time":             {"1500000.7"},
+			"mp4:com.apple.quicktime.content.identifier": {"6F1A2B3C"},
+			"quicktime:still-image-time":                 {"1500000.7"},
 		})
 		Expect(livePhoto).ToNot(BeNil())
 		Expect(livePhoto.StillImageTimeUs).To(Equal(libregraph.PtrInt64(1500001)))
@@ -55,9 +55,9 @@ var _ = Describe("getLivePhoto", func() {
 
 	It("keeps the facet when optional values are malformed", func() {
 		livePhoto := Tika{}.getLivePhoto(map[string][]string{
-			"com.apple.quicktime.content.identifier":        {"6F1A2B3C"},
-			"com.apple.quicktime.live-photo.auto":           {"maybe"},
-			"com.apple.quicktime.live-photo.vitality-score": {"very"},
+			"mp4:com.apple.quicktime.content.identifier":        {"6F1A2B3C"},
+			"mp4:com.apple.quicktime.live-photo.auto":           {"maybe"},
+			"mp4:com.apple.quicktime.live-photo.vitality-score": {"very"},
 		})
 		Expect(livePhoto).ToNot(BeNil())
 		Expect(livePhoto.Auto).To(BeNil())
@@ -66,7 +66,7 @@ var _ = Describe("getLivePhoto", func() {
 
 	It("leaves the video-only fields empty on the still half", func() {
 		livePhoto := Tika{}.getLivePhoto(map[string][]string{
-			"Content Identifier": {"6F1A2B3C"},
+			"img:Content Identifier": {"6F1A2B3C"},
 		})
 		Expect(livePhoto).ToNot(BeNil())
 		Expect(livePhoto.StillImageTimeUs).To(BeNil())