feat(search): motion photo facet

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

@@ -299,6 +299,9 @@
             "type": "string"
           }
         },
+        "motionPhoto": {
+          "$ref": "#/definitions/v0MotionPhoto"
+        },
         "video": {
           "$ref": "#/definitions/v0Video"
         }
@@ -393,6 +396,23 @@
         }
       }
     },
+    "v0MotionPhoto": {
+      "type": "object",
+      "properties": {
+        "version": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "presentationTimestampUs": {
+          "type": "string",
+          "format": "int64"
+        },
+        "videoSize": {
+          "type": "string",
+          "format": "int64"
+        }
+      }
+    },
     "v0Photo": {
       "type": "object",
       "properties": {

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

@@ -72,6 +72,12 @@ message Video {
 	optional int32 width = 10;
 }
 
+message MotionPhoto {
+	optional int32 version = 1;
+	optional int64 presentationTimestampUs = 2;
+	optional int64 videoSize = 3;
+}
+
 message Entity {
 	Reference ref = 1;
 	ResourceID id = 2;
@@ -93,6 +99,7 @@ message Entity {
 	Image image = 18;
 	Photo photo = 19;
 	repeated string favorites = 20;
+	MotionPhoto motionPhoto = 21;
 	Video video = 22;
 }
 

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

@@ -485,6 +485,7 @@ func cs3ResourceToDriveItem(logger *log.Logger, publicBaseURL *url.URL, res *sto
 		driveItem.Location = metadataToFacet[libregraph.GeoCoordinates](metadata, "location")
 		driveItem.Photo = metadataToFacet[libregraph.Photo](metadata, "photo")
 		driveItem.Video = metadataToFacet[libregraph.Video](metadata, "video")
+		driveItem.LibreGraphMotionPhoto = metadataToFacet[libregraph.MotionPhoto](metadata, "motionPhoto")
 		driveItem.LibreGraphMeFollowing = libregraph.PtrBool(metadata[_favoriteMetadataKey] == "1")
 		if t := metadata["tags"]; t != "" {
 			driveItem.LibreGraphTags = tags.New(t).AsSlice()

services/search/pkg/bleve/backend.go

@@ -125,21 +125,22 @@ 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),
-				Audio:      hitToFacet[searchMessage.Audio](hit.Fields, "audio"),
-				Image:      hitToFacet[searchMessage.Image](hit.Fields, "image"),
-				Location:   hitToFacet[searchMessage.GeoCoordinates](hit.Fields, "location"),
-				Photo:      hitToFacet[searchMessage.Photo](hit.Fields, "photo"),
-				Video:      hitToFacet[searchMessage.Video](hit.Fields, "video"),
+				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),
+				Audio:       hitToFacet[searchMessage.Audio](hit.Fields, "audio"),
+				Image:       hitToFacet[searchMessage.Image](hit.Fields, "image"),
+				Location:    hitToFacet[searchMessage.GeoCoordinates](hit.Fields, "location"),
+				Photo:       hitToFacet[searchMessage.Photo](hit.Fields, "photo"),
+				Video:       hitToFacet[searchMessage.Video](hit.Fields, "video"),
+				MotionPhoto: hitToFacet[searchMessage.MotionPhoto](hit.Fields, "motionPhoto"),
 			},
 		}
 

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

@@ -760,6 +760,51 @@
           }
         ]
       },
+      "motionPhoto": {
+        "enabled": true,
+        "dynamic": true,
+        "properties": {
+          "presentationTimestampUs": {
+            "enabled": true,
+            "dynamic": true,
+            "fields": [
+              {
+                "type": "number",
+                "store": true,
+                "index": true,
+                "include_in_all": true,
+                "docvalues": true
+              }
+            ]
+          },
+          "version": {
+            "enabled": true,
+            "dynamic": true,
+            "fields": [
+              {
+                "type": "number",
+                "store": true,
+                "index": true,
+                "include_in_all": true,
+                "docvalues": true
+              }
+            ]
+          },
+          "videoSize": {
+            "enabled": true,
+            "dynamic": true,
+            "fields": [
+              {
+                "type": "number",
+                "store": true,
+                "index": true,
+                "include_in_all": true,
+                "docvalues": true
+              }
+            ]
+          }
+        }
+      },
       "photo": {
         "enabled": true,
         "dynamic": true,

services/search/pkg/content/content.go

@@ -15,19 +15,20 @@ 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"`
-	Audio     *libregraph.Audio          `json:"audio,omitempty"`
-	Image     *libregraph.Image          `json:"image,omitempty"`
-	Location  *libregraph.GeoCoordinates `json:"location,omitempty"`
-	Photo     *libregraph.Photo          `json:"photo,omitempty"`
-	Video     *libregraph.Video          `json:"video,omitempty"`
+	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"`
+	Audio       *libregraph.Audio          `json:"audio,omitempty"`
+	Image       *libregraph.Image          `json:"image,omitempty"`
+	Location    *libregraph.GeoCoordinates `json:"location,omitempty"`
+	Photo       *libregraph.Photo          `json:"photo,omitempty"`
+	Video       *libregraph.Video          `json:"video,omitempty"`
+	MotionPhoto *libregraph.MotionPhoto    `json:"motionPhoto,omitempty"`
 }
 
 func CleanString(content, langCode string) string {

services/search/pkg/content/extractor.go

@@ -2,7 +2,6 @@ package content
 
 import (
 	"context"
-	"errors"
 	"fmt"
 
 	provider "github.com/cs3org/go-cs3apis/cs3/storage/provider/v1beta1"
@@ -13,19 +12,13 @@ type Extractor interface {
 	Extract(ctx context.Context, ri *provider.ResourceInfo) (Document, error)
 }
 
-func getFirstValue(m map[string][]string, key string) (string, error) {
-	if m == nil {
-		return "", errors.New("undefined map")
+// getFirstValue returns the first metadata value present among keys, trying them
+// in order. It errors when the map is nil or none of the keys holds a value.
+func getFirstValue(m map[string][]string, keys ...string) (string, error) {
+	for _, key := range keys {
+		if v, ok := m[key]; ok && len(v) > 0 {
+			return v[0], nil
+		}
 	}
-
-	v, ok := m[key]
-	if !ok {
-		return "", fmt.Errorf("unknown key: %v", key)
-	}
-
-	if len(v) == 0 {
-		return "", fmt.Errorf("no values for: %v", key)
-	}
-
-	return v[0], nil
+	return "", fmt.Errorf("no value for keys: %v", keys)
 }

services/search/pkg/content/tika.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"slices"
 	"strings"
 
 	gateway "github.com/cs3org/go-cs3apis/cs3/gateway/v1beta1"
@@ -112,8 +113,21 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,
 		if v := t.getAudio(meta); v != nil {
 			doc.Audio = v
 		}
-		if v := t.getVideo(meta); v != nil {
-			doc.Video = 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])
 		}
 	}
 

services/search/pkg/content/tika_motion_photo.go

@@ -0,0 +1,63 @@
+package content
+
+import (
+	libregraph "github.com/opencloud-eu/libre-graph-api-go"
+	"strconv"
+	"strings"
+)
+
+// getMotionPhoto reads Google Motion Photo XMP, which Tika exposes under the
+// canonical Camera/Container prefixes. It covers both the current MotionPhoto
+// scheme and the legacy MicroVideo scheme. videoSize (needed to range-fetch the
+// video) comes from the video tika extracted, and is required: without it the
+// facet is dropped.
+func (t Tika) getMotionPhoto(meta, video map[string][]string) *libregraph.MotionPhoto {
+	// the marker is what makes this a motion photo rather than a picture that
+	// happens to carry a video: per the spec only a value of 1 counts, every
+	// other value means "treat as a still image".
+	if v, err := getFirstValue(meta, "Camera:MotionPhoto", "Camera:MicroVideo"); err != nil || v != "1" {
+		return nil
+	}
+
+	var motionPhoto *libregraph.MotionPhoto
+	initMotionPhoto := func() {
+		if motionPhoto == nil {
+			motionPhoto = libregraph.NewMotionPhoto()
+		}
+	}
+
+	if v, err := getFirstValue(meta, "Camera:MotionPhotoVersion", "Camera:MicroVideoVersion"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 32); err == nil {
+			initMotionPhoto()
+			motionPhoto.SetVersion(int32(i))
+		}
+	}
+
+	if v, err := getFirstValue(meta, "Camera:MotionPhotoPresentationTimestampUs", "Camera:MicroVideoPresentationTimestampUs"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 64); err == nil {
+			initMotionPhoto()
+			motionPhoto.SetPresentationTimestampUs(i)
+		}
+	}
+
+	if v, err := getFirstValue(video, "Content-Length"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 64); err == nil {
+			initMotionPhoto()
+			motionPhoto.SetVideoSize(i)
+		}
+	}
+
+	if motionPhoto == nil || !motionPhoto.HasVideoSize() {
+		return nil
+	}
+	return motionPhoto
+}
+
+// isVideo reports whether meta describes a video. Tika emits the video appended
+// to a motion photo as an embedded document, and it only does so when the bytes
+// the xmp advertises are really there: a shared motion photo can keep the xmp
+// and lose the video.
+func isVideo(meta map[string][]string) bool {
+	v, err := getFirstValue(meta, "Content-Type")
+	return err == nil && strings.HasPrefix(v, "video/")
+}