Merge branch 'harvest-ocr' of owi92/opencast into r/15.x

Pull request opencast#5757

  Add slide text and segments to Tobira harvest API

modules/tobira/src/main/java/org/opencastproject/tobira/impl/Item.java

@@ -125,6 +125,12 @@ class Item {
 
  final var captions = findCaptions(mp);
 
+ // Get the generated slide text.
+ final var slideText = Arrays.stream(mp.getElements())
+ .filter(mpe -> mpe.getFlavor().eq("mpeg-7/text"))
+ .map(element -> element.getURI())
+ .findFirst();
+
  // Obtain duration from tracks, as that's usually more accurate (stores information from
  // inspect operations). Fall back to `getDcExtent`.
  final var duration = Arrays.stream(mp.getTracks())
@@ -158,6 +164,8 @@ class Item {
  "created", "creator", "title", "extent", "isPartOf", "description", "identifier",
  }))),
  Jsons.p("captions", Jsons.arr(captions)),
+ Jsons.p("slideText", slideText.map(t -> t.toString()).orElse(null)),
+ Jsons.p("segments", Jsons.arr(findSegments(mp))),
  Jsons.p("updated", event.getModified().getTime())
  );
  }
@@ -321,6 +329,17 @@ private static String findThumbnail(MediaPackage mp) {
  .orElse(null);
  }
 
+ private static List<Jsons.Val> findSegments(MediaPackage mp) {
+ return Arrays.stream(mp.getAttachments())
+ .filter(a -> a.getFlavor().getSubtype().equals("segment+preview"))
+ .map(s -> Jsons.obj(
+ Jsons.p("uri", s.toString()),
+ Jsons.p("startTime", s.getReference().getProperty("time"))
+ )
+ )
+ .collect(Collectors.toCollection(ArrayList::new));
+ }
+
  private static Jsons.Val findTimelinePreview(MediaPackage mp) {
  return Arrays.stream(mp.getAttachments())
  .filter(a -> a.getFlavor().getSubtype().equals("timeline+preview"))

modules/tobira/src/main/java/org/opencastproject/tobira/impl/TobiraEndpoint.java

@@ -81,7 +81,7 @@ public class TobiraEndpoint {
 
  // Versioning the Tobira API:
  //
- // Since both Tobira and this API are changing over time, we need some machanism for ensuring they
+ // Since both Tobira and this API are changing over time, we need some mechanism for ensuring they
  // are compatible. We don't want to enforce a 1:1 thing, where a particular Tobira needs one
  // exact API as that makes the update process harder (especially once this module is included in
  // the community version). So instead we have some semver-like versioning here. Increase the
@@ -95,7 +95,7 @@ public class TobiraEndpoint {
  // adding new JSON fields is a non-breaking change. You should also consider whether Tobira needs
  // to resynchronize, i.e. to get new data.
  private static final int VERSION_MAJOR = 1;
- private static final int VERSION_MINOR = 5;
+ private static final int VERSION_MINOR = 6;
  private static final String VERSION = VERSION_MAJOR + "." + VERSION_MINOR;
 
  private SearchService searchService;