mirror/opencloud
1
0 Fork 0
mirror of https://github.com/opencloud-eu/opencloud.git synced 2026-10-06 19:01:55 -04:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #3651 from opencloud-eu/fix/search-tika-4-keys

Browse Source 
fix(search): tika key fixes
This commit is contained in:
Dominik Schmidt authored and GitHub committed 2026-10-03 14:27:17 +02:00
2 parents a3255422b8 2282082349
commit 4dec19089d
7 files changed
+39 -30
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

.woodpecker.star
+1 -1
View File
@@ -8,7 +8,7 @@ docker_repo_slug = "opencloudeu/opencloud"
# images
ALPINE_GIT = "alpine/git:latest"
APACHE_TIKA = "apache/tika:4.1.0-SNAPSHOT"
APACHE_TIKA = "apache/tika:4.1.0-full"
CHKO_DOCKER_PUSHRM = "chko/docker-pushrm:1"
CODACY_COVERAGE_REPORTER = "codacy/codacy-coverage-reporter:14.1.3"
COLLABORA_CODE = "collabora/code:24.04.5.1.1"
services/search/pkg/content/tika.go
+4 -4
View File
@@ -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
+5 -6
View File
@@ -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
+16 -16
View File
@@ -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())
services/search/pkg/content/tika_photo.go
+1 -1
View File
@@ -40,7 +40,7 @@ func (t Tika) getPhoto(meta map[string][]string) *libregraph.Photo {
}
}
if v, err := getFirstValue(meta, "Base ISO"); err == nil {
if v, err := getFirstValue(meta, "exif:IsoSpeedRatings"); err == nil {
if i, err := strconv.ParseInt(v, 0, 32); err == nil {
initPhoto()
photo.SetIso(int32(i))
services/search/pkg/content/tika_photo_test.go
+11 -1
View File
@@ -16,7 +16,7 @@ var _ = Describe("getPhoto", func() {
"exif:ExposureTime": {"0.001"},
"exif:FNumber": {"1.8"},
"exif:FocalLength": {"50"},
"Base ISO": {"100"},
"exif:IsoSpeedRatings": {"100"},
"tiff:Orientation": {"1"},
"exif:DateTimeOriginal": {"2018-01-01T12:34:56"},
})
@@ -32,6 +32,16 @@ var _ = Describe("getPhoto", func() {
Expect(photo.TakenDateTime).To(Equal(libregraph.PtrTime(time.Date(2018, 1, 1, 12, 34, 56, 0, time.UTC))))
})
It("takes the iso of the shot, not the canon maker note base iso", func() {
photo := Tika{}.getPhoto(map[string][]string{
"exif:IsoSpeedRatings": {"800"},
"Base ISO": {"256"},
"img:Base ISO": {"256"},
})
Expect(photo).ToNot(BeNil())
Expect(photo.Iso).To(Equal(libregraph.PtrInt32(800)))
})
It("returns nil when no photo metadata is present", func() {
Expect(Tika{}.getPhoto(map[string][]string{})).To(BeNil())
})
tests/acceptance/docker/src/tika.yml
+1 -1
View File
@@ -5,4 +5,4 @@ services:
- tika
command: tika:9998
tika:
image: apache/tika:4.0.0-full
image: apache/tika:4.1.0-full