feat(search): index and expose the video facet

Author: dschmidt

protogen/gen/opencloud/messages/search/v0/search.pb.go

@@ -298,6 +298,9 @@
           "items": {
             "type": "string"
           }
+        },
+        "video": {
+          "$ref": "#/definitions/v0Video"
         }
       }
     },
@@ -531,6 +534,49 @@
           "format": "int32"
         }
       }
+    },
+    "v0Video": {
+      "type": "object",
+      "properties": {
+        "audioBitsPerSample": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "audioChannels": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "audioFormat": {
+          "type": "string"
+        },
+        "audioSamplesPerSecond": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "bitrate": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "duration": {
+          "type": "string",
+          "format": "int64"
+        },
+        "fourCC": {
+          "type": "string"
+        },
+        "frameRate": {
+          "type": "number",
+          "format": "double"
+        },
+        "height": {
+          "type": "integer",
+          "format": "int32"
+        },
+        "width": {
+          "type": "integer",
+          "format": "int32"
+        }
+      }
     }
   },
   "externalDocs": {

protogen/gen/opencloud/messages/search/v0/search.pb.web.go

@@ -59,6 +59,19 @@ message Photo {
 	optional google.protobuf.Timestamp takenDateTime = 9;
 }
 
+message Video {
+	optional int32 audioBitsPerSample = 1;
+	optional int32 audioChannels = 2;
+	optional string audioFormat = 3;
+	optional int32 audioSamplesPerSecond = 4;
+	optional int32 bitrate = 5;
+	optional int64 duration = 6;
+	optional string fourCC = 7;
+	optional double frameRate = 8;
+	optional int32 height = 9;
+	optional int32 width = 10;
+}
+
 message Entity {
 	Reference ref = 1;
 	ResourceID id = 2;
@@ -80,6 +93,7 @@ message Entity {
 	Image image = 18;
 	Photo photo = 19;
 	repeated string favorites = 20;
+	Video video = 22;
 }
 
 message Match {

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

@@ -456,6 +456,7 @@ func cs3ResourceToDriveItem(logger *log.Logger, publicBaseURL *url.URL, res *sto
 		driveItem.Image = metadataToFacet[libregraph.Image](metadata, "image")
 		driveItem.Location = metadataToFacet[libregraph.GeoCoordinates](metadata, "location")
 		driveItem.Photo = metadataToFacet[libregraph.Photo](metadata, "photo")
+		driveItem.Video = metadataToFacet[libregraph.Video](metadata, "video")
 	}
 
 	return driveItem, nil

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

@@ -139,6 +139,7 @@ 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"),
+				Video:      hitToFacet[searchMessage.Video](hit.Fields, "video"),
 			},
 		}
 

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

@@ -27,6 +27,7 @@ type Document struct {
 	Image     *libregraph.Image          `json:"image,omitempty"`
 	Location  *libregraph.GeoCoordinates `json:"location,omitempty"`
 	Photo     *libregraph.Photo          `json:"photo,omitempty"`
+	Video     *libregraph.Video          `json:"video,omitempty"`
 }
 
 func CleanString(content, langCode string) string {

services/search/pkg/bleve/backend.go

@@ -108,6 +108,9 @@ 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 langCode, _ := t.tika.LanguageString(ctx, doc.Content); langCode != "" && t.CleanStopWords {

services/search/pkg/content/content.go

@@ -2,11 +2,18 @@ package content
 
 import (
 	"strconv"
+	"strings"
 
 	libregraph "github.com/opencloud-eu/libre-graph-api-go"
 )
 
 func (t Tika) getImage(meta map[string][]string) *libregraph.Image {
+	// tiff:ImageWidth/Length are also set for videos; the content type is what
+	// tells an image apart.
+	if ct, err := getFirstValue(meta, "Content-Type"); err != nil || !strings.HasPrefix(ct, "image/") {
+		return nil
+	}
+
 	var image *libregraph.Image
 	initImage := func() {
 		if image == nil {

services/search/pkg/content/tika.go

@@ -9,6 +9,7 @@ import (
 var _ = Describe("getImage", func() {
 	It("maps the image dimensions to the image facet", func() {
 		image := Tika{}.getImage(map[string][]string{
+			"Content-Type":     {"image/jpeg"},
 			"tiff:ImageWidth":  {"100"},
 			"tiff:ImageLength": {"200"},
 		})
@@ -17,6 +18,14 @@ var _ = Describe("getImage", func() {
 		Expect(image.Height).To(Equal(libregraph.PtrInt32(200)))
 	})
 
+	It("returns nil for a video even though it carries the same tiff dimensions", func() {
+		Expect(Tika{}.getImage(map[string][]string{
+			"Content-Type":     {"video/mp4"},
+			"tiff:ImageWidth":  {"1920"},
+			"tiff:ImageLength": {"1080"},
+		})).To(BeNil())
+	})
+
 	It("returns nil when no image metadata is present", func() {
 		Expect(Tika{}.getImage(map[string][]string{})).To(BeNil())
 	})