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

@@ -307,6 +307,9 @@
         },
         "livePhoto": {
           "$ref": "#/definitions/v0LivePhoto"
+        },
+        "preview": {
+          "$ref": "#/definitions/v0Preview"
         }
       }
     },
@@ -478,6 +481,20 @@
         }
       }
     },
+    "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;
@@ -110,6 +117,7 @@ message Entity {
 	MotionPhoto motionPhoto = 21;
 	Video video = 22;
 	LivePhoto livePhoto = 23;
+	Preview preview = 24;
 }
 
 message Match {

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

@@ -8,6 +8,7 @@ import (
 	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"
 )
 
@@ -34,24 +35,47 @@ func setDriveItemThumbnails(item *libregraph.DriveItem, res *provider.ResourceIn
 	}
 }
 
-// previewThumbnailSet returns nil when the thumbnailer cannot render the resource.
+// previewThumbnailSet returns nil when no preview can be produced for the resource.
 func previewThumbnailSet(res *provider.ResourceInfo, baseURL string) *libregraph.ThumbnailSet {
-	if !thumbnail.IsMimeTypeSupported(res.GetMimeType()) {
+	if !thumbnail.HasPreview(res) {
 		return nil
 	}
-	return thumbnailSetFor(baseURL, storagespace.FormatResourceID(res.GetId()))
+
+	itemID := storagespace.FormatResourceID(res.GetId())
+	set := thumbnailSetFor(baseURL, itemID)
+	// only the source carries exact dimensions, the boxes above are requests
+	if w, h := previewSourceDimensions(res); w > 0 && h > 0 {
+		url := fmt.Sprintf("%s&x=%d&y=%d", previewBaseURL(baseURL, itemID), w, h)
+		set.Source = &libregraph.Thumbnail{Url: &url, Width: &w, Height: &h}
+	}
+	return set
+}
+
+// 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
 }
 
 // thumbnailSetFor builds the urls of the WebDAV preview endpoint.
 func thumbnailSetFor(baseURL, itemID string) *libregraph.ThumbnailSet {
-	base := fmt.Sprintf("%s/dav/spaces/%s?scalingup=0&preview=1&processor=thumbnail", baseURL, itemID)
+	base := previewBaseURL(baseURL, itemID)
 	return &libregraph.ThumbnailSet{
 		Small:  previewThumbnail(base, thumbnailBoxSmall),
 		Medium: previewThumbnail(base, thumbnailBoxMedium),
 		Large:  previewThumbnail(base, thumbnailBoxLarge),
 	}
 }
 
+func previewBaseURL(baseURL, itemID string) string {
+	return fmt.Sprintf("%s/dav/spaces/%s?scalingup=0&preview=1&processor=thumbnail", baseURL, itemID)
+}
+
 func previewThumbnail(base string, box int32) *libregraph.Thumbnail {
 	url := fmt.Sprintf("%s&x=%d&y=%d", base, box, box)
 	return &libregraph.Thumbnail{Url: &url}

services/search/pkg/bleve/backend.go

@@ -125,16 +125,18 @@ func (b *Backend) Search(_ context.Context, sir *searchService.SearchIndexReques
 					ResourceId: resourceIDtoSearchID(rootID),
 					Path:       getFieldValue[string](hit.Fields, "Path"),
 				},
-				Id:          resourceIDtoSearchID(rID),
-				Name:        getFieldValue[string](hit.Fields, "Name"),
-				ParentId:    resourceIDtoSearchID(pID),
-				Size:        uint64(getFieldValue[float64](hit.Fields, "Size")),
-				Type:        uint64(getFieldValue[float64](hit.Fields, "Type")),
-				MimeType:    getFieldValue[string](hit.Fields, "MimeType"),
-				Deleted:     getFieldValue[bool](hit.Fields, "Deleted"),
-				Tags:        getFieldSliceValue[string](hit.Fields, "Tags"),
-				Favorites:   getFieldSliceValue[string](hit.Fields, "Favorites"),
-				Highlights:  getFragmentValue(hit.Fragments, "Content", 0),
+				Id:         resourceIDtoSearchID(rID),
+				Name:       getFieldValue[string](hit.Fields, "Name"),
+				ParentId:   resourceIDtoSearchID(pID),
+				Size:       uint64(getFieldValue[float64](hit.Fields, "Size")),
+				Type:       uint64(getFieldValue[float64](hit.Fields, "Type")),
+				MimeType:   getFieldValue[string](hit.Fields, "MimeType"),
+				Deleted:    getFieldValue[bool](hit.Fields, "Deleted"),
+				Tags:       getFieldSliceValue[string](hit.Fields, "Tags"),
+				Favorites:  getFieldSliceValue[string](hit.Fields, "Favorites"),
+				Highlights: getFragmentValue(hit.Fragments, "Content", 0),
+				Preview:    hitToFacet[searchMessage.Preview](hit.Fields, "preview"),
+
 				Audio:       hitToFacet[searchMessage.Audio](hit.Fields, "audio"),
 				Image:       hitToFacet[searchMessage.Image](hit.Fields, "image"),
 				Location:    hitToFacet[searchMessage.GeoCoordinates](hit.Fields, "location"),

services/search/pkg/bleve/testdata/mapping.golden.json

@@ -1053,6 +1053,38 @@
           }
         }
       },
+      "preview": {
+        "enabled": true,
+        "dynamic": true,
+        "properties": {
+          "height": {
+            "enabled": true,
+            "dynamic": true,
+            "fields": [
+              {
+                "type": "number",
+                "store": true,
+                "index": true,
+                "include_in_all": true,
+                "docvalues": true
+              }
+            ]
+          },
+          "width": {
+            "enabled": true,
+            "dynamic": true,
+            "fields": [
+              {
+                "type": "number",
+                "store": true,
+                "index": true,
+                "include_in_all": true,
+                "docvalues": true
+              }
+            ]
+          }
+        }
+      },
       "video": {
         "enabled": true,
         "dynamic": true,

services/search/pkg/content/content.go

@@ -15,14 +15,16 @@ func init() {
 // Document wraps all resource meta fields,
 // it is used as a content extraction result.
 type Document struct {
-	Title       string                     `json:"Title"`
-	Name        string                     `json:"Name"`
-	Content     string                     `json:"Content"`
-	Size        uint64                     `json:"Size"`
-	Mtime       *time.Time                 `json:"Mtime,omitempty"`
-	MimeType    string                     `json:"MimeType"`
-	Tags        []string                   `json:"Tags"`
-	Favorites   []string                   `json:"Favorites"`
+	Title     string     `json:"Title"`
+	Name      string     `json:"Name"`
+	Content   string     `json:"Content"`
+	Size      uint64     `json:"Size"`
+	Mtime     *time.Time `json:"Mtime,omitempty"`
+	MimeType  string     `json:"MimeType"`
+	Tags      []string   `json:"Tags"`
+	Favorites []string   `json:"Favorites"`
+	Preview   *Preview   `json:"preview,omitempty"`
+
 	Audio       *libregraph.Audio          `json:"audio,omitempty"`
 	Image       *libregraph.Image          `json:"image,omitempty"`
 	Location    *libregraph.GeoCoordinates `json:"location,omitempty"`
@@ -32,6 +34,21 @@ type Document struct {
 	LivePhoto   *libregraph.LivePhoto      `json:"livePhoto,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 {
 	return strings.TrimSpace(stopwords.CleanString(content, langCode, true))
 }

services/search/pkg/content/tika.go

@@ -82,7 +82,11 @@ 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 {
 		title, err := getFirstValue(meta, "dc:title")
 		if err != nil {
@@ -98,42 +102,27 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,
 		} else if content, err := getFirstValue(meta, "X-TIKA:content"); err == nil {
 			doc.Content = strings.TrimSpace(fmt.Sprintf("%s %s", doc.Content, content))
 		}
-
-		// keep facets from earlier entries, an embedded resource's meta
-		// (e.g. cover art) must not reset them
-		if v := t.getLocation(meta); v != nil {
-			doc.Location = v
-		}
-		if v := t.getImage(meta); v != nil {
-			doc.Image = v
-		}
-		if v := t.getPhoto(meta); v != nil {
-			doc.Photo = v
-		}
-		if v := t.getAudio(meta); v != nil {
-			doc.Audio = v
-		}
-		if v := t.getLivePhoto(meta); v != nil {
-			doc.LivePhoto = v
-		}
-	}
-
-	if len(metas) > 0 {
-		// the video facet says the file is a video, so it comes from the file
-		// itself: the clip tika extracts from a motion photo must not make its
-		// image look like one
-		doc.Video = t.getVideo(metas[0])
 	}
 
 	// a motion photo is the xmp on the file itself plus the video tika extracted
 	// from it. The xmp alone proves nothing: a share can keep it and strip the
 	// appended video.
-	if len(metas) > 0 {
-		if i := slices.IndexFunc(metas[1:], isVideo); i >= 0 {
-			doc.MotionPhoto = t.getMotionPhoto(metas[0], metas[i+1])
-		}
+	if i := slices.IndexFunc(metas[1:], isVideo); i >= 0 {
+		doc.MotionPhoto = t.getMotionPhoto(metas[0], metas[i+1])
 	}
 
+	doc.Preview = getPreview(ri.GetMimeType(), metas)
+
+	// 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.Video = t.getVideo(container)
+	doc.LivePhoto = t.getLivePhoto(container)
+
 	if langCode := t.detectLanguage(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
+}