feat(search): honest has-preview prop and thumbnails relationship

pkg/conversions/strings.go

@@ -1,6 +1,7 @@
 package conversions
 
 import (
+	"strconv"
 	"strings"
 )
 
@@ -14,3 +15,12 @@ func StringToSliceString(src string, sep string) []string {
 
 	return parts
 }
+
+// StringToInt32 parses s as a base-10 int32, returning fallback on any error.
+func StringToInt32(s string, fallback int32) int32 {
+	v, err := strconv.ParseInt(s, 10, 32)
+	if err != nil {
+		return fallback
+	}
+	return int32(v)
+}

protogen/gen/opencloud/services/search/v0/search.swagger.json

@@ -288,6 +288,9 @@
           "items": {
             "type": "string"
           }
+        },
+        "preview": {
+          "$ref": "#/definitions/v0Preview"
         }
       }
     },
@@ -391,6 +394,20 @@
         }
       }
     },
+    "v0Preview": {
+      "type": "object",
+      "properties": {
+        "width": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "height": {
+          "type": "integer",
+          "format": "int32"
+        }
+      },
+      "description": "Preview carries the dimensions of an embedded preview (e.g. audio cover art).\nIts presence signals that a preview is available for the resource."
+    },
     "v0Reference": {
       "type": "object",
       "properties": {

protogen/proto/opencloud/messages/search/v0/search.proto

@@ -41,6 +41,13 @@ message Image {
 	optional int32 height = 2;
 }
 
+// Preview carries the dimensions of an embedded preview (e.g. audio cover art).
+// Its presence signals that a preview is available for the resource.
+message Preview {
+	optional int32 width = 1;
+	optional int32 height = 2;
+}
+
 message GeoCoordinates {
   optional double altitude  = 1;
   optional double latitude = 2;
@@ -80,6 +87,7 @@ message Entity {
 	Image image = 18;
 	Photo photo = 19;
 	repeated string favorites = 20;
+	Preview preview = 21;
 }
 
 message Match {

services/graph/pkg/service/v0/driveitems.go

@@ -209,6 +209,7 @@ func (g Graph) GetRootDriveChildren(w http.ResponseWriter, r *http.Request) {
 		errorcode.GeneralException.Render(w, r, http.StatusInternalServerError, err.Error())
 		return
 	}
+	g.setDriveItemsThumbnails(r, files, lRes.GetInfos())
 
 	render.Status(r, http.StatusOK)
 	render.JSON(w, r, &ListResponse{Value: files})
@@ -341,6 +342,7 @@ func (g Graph) GetDriveItemChildren(w http.ResponseWriter, r *http.Request) {
 		errorcode.GeneralException.Render(w, r, http.StatusInternalServerError, err.Error())
 		return
 	}
+	g.setDriveItemsThumbnails(r, files, res.GetInfos())
 
 	render.Status(r, http.StatusOK)
 	render.JSON(w, r, &ListResponse{Value: files})

services/graph/pkg/service/v0/thumbnails.go

@@ -0,0 +1,81 @@
+package svc
+
+import (
+	"fmt"
+	"net/http"
+	"strings"
+
+	provider "github.com/cs3org/go-cs3apis/cs3/storage/provider/v1beta1"
+	libregraph "github.com/opencloud-eu/libre-graph-api-go"
+	"github.com/opencloud-eu/reva/v2/pkg/storagespace"
+
+	"github.com/opencloud-eu/opencloud/pkg/conversions"
+	"github.com/opencloud-eu/opencloud/services/thumbnails/pkg/thumbnail"
+)
+
+func shouldExpand(r *http.Request, relationship string) bool {
+	return strings.Contains(r.URL.Query().Get("$expand"), relationship)
+}
+
+func (g Graph) setDriveItemsThumbnails(r *http.Request, items []*libregraph.DriveItem, infos []*provider.ResourceInfo) {
+	if !shouldExpand(r, "thumbnails") {
+		return
+	}
+	for i := range items {
+		if i < len(infos) {
+			setDriveItemThumbnails(items[i], infos[i], g.config.Commons.OpenCloudURL)
+		}
+	}
+}
+
+// Requested box sizes for the thumbnail URLs. Small/medium/large carry no
+// dimensions: the endpoint rounds the box to a configured resolution and fits
+// within it, so the served size is not known here. Only the source dimensions
+// (below) are exact.
+const (
+	thumbnailBoxSmall  = 36
+	thumbnailBoxMedium = 48
+	thumbnailBoxLarge  = 96
+)
+
+func setDriveItemThumbnails(item *libregraph.DriveItem, res *provider.ResourceInfo, baseURL string) {
+	if set := previewThumbnailSet(res, baseURL); set != nil {
+		item.SetThumbnails([]libregraph.ThumbnailSet{*set})
+	}
+}
+
+func previewThumbnailSet(res *provider.ResourceInfo, baseURL string) *libregraph.ThumbnailSet {
+	if !thumbnail.HasPreview(res) {
+		return nil
+	}
+
+	base := fmt.Sprintf("%s/dav/spaces/%s?scalingup=0&preview=1&processor=thumbnail",
+		baseURL, storagespace.FormatResourceID(res.GetId()))
+
+	set := &libregraph.ThumbnailSet{
+		Small:  previewThumbnail(base, thumbnailBoxSmall),
+		Medium: previewThumbnail(base, thumbnailBoxMedium),
+		Large:  previewThumbnail(base, thumbnailBoxLarge),
+	}
+	if w, h := previewSourceDimensions(res); w > 0 && h > 0 {
+		url := fmt.Sprintf("%s&x=%d&y=%d", base, w, h)
+		set.Source = &libregraph.Thumbnail{Url: &url, Width: &w, Height: &h}
+	}
+	return set
+}
+
+func previewThumbnail(base string, box int32) *libregraph.Thumbnail {
+	url := fmt.Sprintf("%s&x=%d&y=%d", base, box, box)
+	return &libregraph.Thumbnail{Url: &url}
+}
+
+// previewSourceDimensions: audio cover from oc.preview, images from the image facet.
+func previewSourceDimensions(res *provider.ResourceInfo) (int32, int32) {
+	if w, h := thumbnail.PreviewDimensions(res); w > 0 && h > 0 {
+		return w, h
+	}
+	meta := res.GetArbitraryMetadata().GetMetadata()
+	w := conversions.StringToInt32(meta["libre.graph.image.width"], 0)
+	h := conversions.StringToInt32(meta["libre.graph.image.height"], 0)
+	return w, h
+}

services/search/pkg/bleve/backend.go

@@ -141,6 +141,7 @@ func (b *Backend) Search(_ context.Context, sir *searchService.SearchIndexReques
 				Image:      hitToFacet[searchMessage.Image](hit.Fields, "image"),
 				Location:   hitToFacet[searchMessage.GeoCoordinates](hit.Fields, "location"),
 				Photo:      hitToFacet[searchMessage.Photo](hit.Fields, "photo"),
+				Preview:    hitToFacet[searchMessage.Preview](hit.Fields, "preview"),
 			},
 		}
 

services/search/pkg/content/content.go

@@ -27,6 +27,22 @@ type Document struct {
 	Image     *libregraph.Image          `json:"image,omitempty"`
 	Location  *libregraph.GeoCoordinates `json:"location,omitempty"`
 	Photo     *libregraph.Photo          `json:"photo,omitempty"`
+	Preview   *Preview                   `json:"preview,omitempty"`
+}
+
+// Preview holds the dimensions of an embedded preview (e.g. audio cover art).
+// Internal signal, not a Graph facet; its presence marks that a preview exists.
+type Preview struct {
+	Width  int32 `json:"width"`
+	Height int32 `json:"height"`
+}
+
+// ToMap flows Preview through the shared facet flattening, under oc.preview.
+func (p Preview) ToMap() (map[string]interface{}, error) {
+	return map[string]interface{}{
+		"width":  p.Width,
+		"height": p.Height,
+	}, nil
 }
 
 func CleanString(content, langCode string) string {

services/search/pkg/content/tika.go

@@ -77,7 +77,11 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,
 	if err != nil {
 		return doc, err
 	}
+	if len(metas) == 0 {
+		return doc, nil
+	}
 
+	// Title and content aggregate across the container and embedded resources.
 	for _, meta := range metas {
 		if title, err := getFirstValue(meta, "title"); err == nil {
 			doc.Title = strings.TrimSpace(fmt.Sprintf("%s %s", doc.Title, title))
@@ -86,13 +90,18 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,
 		if content, err := getFirstValue(meta, "X-TIKA:content"); err == nil {
 			doc.Content = strings.TrimSpace(fmt.Sprintf("%s %s", doc.Content, content))
 		}
-
-		doc.Location = t.getLocation(meta)
-		doc.Image = t.getImage(meta)
-		doc.Photo = t.getPhoto(meta)
-		doc.Audio = t.getAudio(meta)
 	}
 
+	// Facets come from the container (first entry) only; embedded resources like
+	// audio cover art must not leak in (the cover becomes the preview instead).
+	container := metas[0]
+	doc.Location = t.getLocation(container)
+	doc.Image = t.getImage(container)
+	doc.Photo = t.getPhoto(container)
+	doc.Audio = t.getAudio(container)
+
+	doc.Preview = getPreview(ri.GetMimeType(), metas)
+
 	if langCode, _ := t.tika.LanguageString(ctx, doc.Content); langCode != "" && t.CleanStopWords {
 		doc.Content = CleanString(doc.Content, langCode)
 	}

services/search/pkg/content/tika_preview.go

@@ -0,0 +1,47 @@
+package content
+
+import (
+	"strconv"
+	"strings"
+
+	"github.com/opencloud-eu/opencloud/services/thumbnails/pkg/thumbnail"
+)
+
+// frontCoverDescription is the picture type Tika reports (as dc:description) for
+// the front cover, matching the thumbnailer's cover selection.
+const frontCoverDescription = "Cover (front)"
+
+// getPreview returns the dimensions of an audio file's embedded cover art from
+// Tika's recursive metadata, preferring the front cover (dc:description) and
+// falling back to the first image, matching the thumbnailer's selection. It only
+// runs for EmbeddedPreviewMimeTypes.
+func getPreview(mimeType string, metas []map[string][]string) *Preview {
+	if _, ok := thumbnail.EmbeddedPreviewMimeTypes[mimeType]; !ok {
+		return nil
+	}
+	var first *Preview
+	for _, meta := range metas {
+		ct, err := getFirstValue(meta, "Content-Type")
+		if err != nil || !strings.HasPrefix(ct, "image/") {
+			continue
+		}
+		w, wErr := getFirstValue(meta, "tiff:ImageWidth")
+		h, hErr := getFirstValue(meta, "tiff:ImageLength")
+		if wErr != nil || hErr != nil {
+			continue
+		}
+		width, wErr := strconv.ParseInt(w, 10, 32)
+		height, hErr := strconv.ParseInt(h, 10, 32)
+		if wErr != nil || hErr != nil || width <= 0 || height <= 0 {
+			continue
+		}
+		preview := &Preview{Width: int32(width), Height: int32(height)}
+		if desc, _ := getFirstValue(meta, "dc:description"); desc == frontCoverDescription {
+			return preview
+		}
+		if first == nil {
+			first = preview
+		}
+	}
+	return first
+}

services/search/pkg/content/tika_preview_test.go

@@ -0,0 +1,49 @@
+package content
+
+import (
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+)
+
+var _ = Describe("getPreview", func() {
+	audio := map[string][]string{"Content-Type": {"audio/mpeg"}}
+	cover := map[string][]string{
+		"Content-Type":     {"image/jpeg"},
+		"tiff:ImageWidth":  {"500"},
+		"tiff:ImageLength": {"400"},
+	}
+	coverNoDims := map[string][]string{"Content-Type": {"image/jpeg"}}
+
+	It("returns the embedded cover dimensions for audio", func() {
+		p := getPreview("audio/mpeg", []map[string][]string{audio, cover})
+		Expect(p).ToNot(BeNil())
+		Expect(*p).To(Equal(Preview{Width: 500, Height: 400}))
+	})
+
+	It("returns nil when the audio has no cover", func() {
+		Expect(getPreview("audio/mpeg", []map[string][]string{audio})).To(BeNil())
+	})
+
+	It("returns nil when the cover lacks dimensions", func() {
+		Expect(getPreview("audio/mpeg", []map[string][]string{audio, coverNoDims})).To(BeNil())
+	})
+
+	It("prefers the front cover over an earlier back cover", func() {
+		back := map[string][]string{
+			"Content-Type": {"image/jpeg"}, "dc:description": {"Cover (back)"},
+			"tiff:ImageWidth": {"30"}, "tiff:ImageLength": {"30"},
+		}
+		front := map[string][]string{
+			"Content-Type": {"image/jpeg"}, "dc:description": {"Cover (front)"},
+			"tiff:ImageWidth": {"64"}, "tiff:ImageLength": {"40"},
+		}
+		p := getPreview("audio/mpeg", []map[string][]string{audio, back, front})
+		Expect(p).ToNot(BeNil())
+		Expect(*p).To(Equal(Preview{Width: 64, Height: 40}))
+	})
+
+	It("is gated to embedded-preview types", func() {
+		// an image is unconditional; its preview is not driven by oc.preview.
+		Expect(getPreview("image/png", []map[string][]string{cover})).To(BeNil())
+	})
+})

services/search/pkg/opensearch/internal/convert/opensearch.go

@@ -82,6 +82,7 @@ func OpenSearchHitToMatch(hit opensearchgoAPI.SearchHit) (*searchMessage.Match,
 			Image:    copyFacet[searchMessage.Image](resource.Image),
 			Location: copyFacet[searchMessage.GeoCoordinates](resource.Location),
 			Photo:    copyFacet[searchMessage.Photo](resource.Photo),
+			Preview:  copyFacet[searchMessage.Preview](resource.Preview),
 		},
 	}
 

services/search/pkg/search/service.go

@@ -633,6 +633,7 @@ func (s *Service) doUpsertItem(ref *provider.Reference, batch BatchOperator) {
 	facetToMetadata(metadata, doc.Image, "libre.graph.image.")
 	facetToMetadata(metadata, doc.Location, "libre.graph.location.")
 	facetToMetadata(metadata, doc.Photo, "libre.graph.photo.")
+	facetToMetadata(metadata, doc.Preview, "oc.preview.")
 	if len(metadata) == 0 {
 		return
 	}