Merged in DSC-1619 (pull request DSpace#2224)

DSC-1619 Approved-by: Giuseppe Digilio
4Science · May 29, 2024 · 747a74d · 747a74d
2 parents 1cc78a8 + 06e6199
commit 747a74d
Show file tree

Hide file tree

Showing 12 changed files with 267 additions and 22 deletions.
diff --git a/dspace-api/src/main/java/org/dspace/iiif/consumer/IIIFCacheEventConsumer.java b/dspace-api/src/main/java/org/dspace/iiif/consumer/IIIFCacheEventConsumer.java
@@ -19,6 +19,8 @@
 import org.dspace.core.Context;
 import org.dspace.event.Consumer;
 import org.dspace.event.Event;
+import org.dspace.services.ConfigurationService;
+import org.dspace.utils.DSpace;
 
 
 /**
@@ -37,12 +39,18 @@ public class IIIFCacheEventConsumer implements Consumer {
     // Collects modified bitstreams for individual removal from canvas dimension cache.
     private final Set<DSpaceObject> toEvictFromCanvasCache = new HashSet<>();
 
+    private ConfigurationService configurationService;
+
     @Override
     public void initialize() throws Exception {
+        configurationService = new DSpace().getConfigurationService();
     }
 
     @Override
     public void consume(Context ctx, Event event) throws Exception {
+        if (!configurationService.getBooleanProperty("iiif.enabled", false)) {
+            return;
+        }
         int st = event.getSubjectType();
         if (!(st == Constants.BUNDLE || st == Constants.ITEM || st == Constants.BITSTREAM)) {
             return;
@@ -136,6 +144,9 @@ private void addToCacheEviction(DSpaceObject subject, DSpaceObject subject2, int
 
     @Override
     public void end(Context ctx) throws Exception {
+        if (!configurationService.getBooleanProperty("iiif.enabled", false)) {
+            return;
+        }
         // Get the eviction service beans.
         ManifestsCacheEvictService manifestsCacheEvictService = CacheEvictBeanLocator.getManifestsCacheEvictService();
         CanvasCacheEvictService canvasCacheEvictService = CacheEvictBeanLocator.getCanvasCacheEvictService();

diff --git a/dspace-server-webapp/src/test/java/org/dspace/app/rest/iiif/IIIFControllerIT.java b/dspace-server-webapp/src/test/java/org/dspace/app/rest/iiif/IIIFControllerIT.java
@@ -131,7 +131,7 @@ public void findOneIIIFSearchableItemWithDefaultDimensionsIT() throws Exception
                 .andExpect(status().isOk())
                 .andExpect(jsonPath("$.@context", is("http://iiif.io/api/presentation/2/context.json")))
                 .andExpect(jsonPath("$.service.profile", is("http://iiif.io/api/search/0/search")))
-                .andExpect(jsonPath("$.thumbnail.@id", containsString("/iiif/2/"
+                .andExpect(jsonPath("$.thumbnail.@id", containsString("/iiif-server/"
                         + bitstream1.getID())))
                 .andExpect(jsonPath("$.metadata[0].label", is("Title")))
                 .andExpect(jsonPath("$.metadata[0].value", is("Public item 1")))

diff --git a/dspace/config/dspace.cfg b/dspace/config/dspace.cfg
@@ -508,6 +508,8 @@ filter.org.dspace.app.mediafilter.TikaTextExtractionFilter.inputFormats = OpenDo
 filter.org.dspace.app.mediafilter.TikaTextExtractionFilter.inputFormats = OpenDocument Text
 filter.org.dspace.app.mediafilter.TikaTextExtractionFilter.inputFormats = RTF
 filter.org.dspace.app.mediafilter.TikaTextExtractionFilter.inputFormats = Text
+filter.org.dspace.app.mediafilter.TikaTextExtractionFilter.inputFormats = TTML
+filter.org.dspace.app.mediafilter.TikaTextExtractionFilter.inputFormats = WebVTT
 filter.org.dspace.app.mediafilter.JPEGFilter.inputFormats = BMP, GIF, JPEG, PNG
 filter.org.dspace.app.mediafilter.BrandedPreviewJPEGFilter.inputFormats = BMP, GIF, JPEG, PNG
 filter.org.dspace.app.mediafilter.ImageMagickImageThumbnailFilter.inputFormats = BMP, GIF, PNG, JPG, TIFF, JPEG, JPEG 2000
@@ -802,7 +804,7 @@ event.dispatcher.default.class = org.dspace.event.BasicDispatcher
 # Add rdf here, if you are using dspace-rdf to export your repository content as RDF.
 # Add iiif here, if you are using dspace-iiif.
 # Add orcidqueue here, if the integration with ORCID is configured and wish to enable the synchronization queue functionality
-event.dispatcher.default.consumers = versioning, discovery, eperson, dedup, crisconsumer, orcidqueue, audit, nbeventsdelete, referenceresolver, orcidwebhook, itemenhancer, customurl, reciprocal, filetypemetadataenhancer, authoritylink
+event.dispatcher.default.consumers = versioning, discovery, eperson, dedup, crisconsumer, orcidqueue, audit, nbeventsdelete, referenceresolver, orcidwebhook, itemenhancer, customurl, iiif, reciprocal, filetypemetadataenhancer, authoritylink
 
 # enable the item enhancer poller
 related-item-enhancer-poller.enabled = true
@@ -1018,6 +1020,7 @@ registry.metadata.load = workflow-types.xml
 registry.metadata.load = openaire4-types.xml
 registry.metadata.load = dspace-types.xml
 registry.metadata.load = iiif-types.xml
+registry.metadata.load = av-types.xml
 registry.metadata.load = bitstream-types.xml
 registry.metadata.load = dataquality-types.xml
 
@@ -1962,6 +1965,7 @@ include = ${module_dir}/authentication-shibboleth.cfg
 include = ${module_dir}/authentication-x509.cfg
 include = ${module_dir}/authentication-orcid.cfg
 include = ${module_dir}/authority.cfg
+include = ${module_dir}/av-stream.cfg
 include = ${module_dir}/bulkedit.cfg
 include = ${module_dir}/citation-page.cfg
 include = ${module_dir}/clamav.cfg

diff --git a/dspace/config/local-orchestrator.cfg.EXAMPLE b/dspace/config/local-orchestrator.cfg.EXAMPLE
@@ -0,0 +1,4 @@
+# EXAMPLE "local-orchestrator-<env>.cfg" for the 4Science DSpace Orchestrator Addon.
+
+solr-database-resync.cron = -
+related-item-enhancer-poller.enabled = false
diff --git a/dspace/config/local.cfg.EXAMPLE b/dspace/config/local.cfg.EXAMPLE
@@ -326,14 +326,8 @@ usage-statistics.dbfile=/var/lib/GeoIP/GeoLite2-City.mmdb
 # IIIF and OCR Integration #
 ############################
 
-## IIIF configuration
-iiif.enabled = false
-# iiif.image.server =
-
-## OCR configuration
-# iiif.search.url = ${solr.server}/${solr.multicorePrefix}ocr
-# iiif.search.plugin = org.dspace.app.iiif.service.WordHighlightSolrSearch
-# ocr.tesseract.path =
+## IIIF configuration (see modules/iiif.cfg)
+#iiif.enabled = true
 
 #########################################################
 ### LOGIN MIUR configuration ###

diff --git a/dspace/config/modules/av-stream.cfg b/dspace/config/modules/av-stream.cfg
@@ -0,0 +1,66 @@
+#---------------------------------------------------------------#
+#-------------AV STREAMING CONFIGURATIONS-----------------------#
+#---------------------------------------------------------------#
+# Configuration properties used by the AV Streaming viewer      #
+# This configuration file is related to the 4Science AV addon   #
+# for more information see https://4science.com/dspace-add-ons/ #
+#---------------------------------------------------------------#
+
+video.valid-mimetype = video/webm, video/x-matroska, video/mpeg, video/quicktime,\
+            video/mp4, video/x-msvideo, video/x-flv, video/divx, video/3gpp, video/avi
+
+audio.valid-mimetype = audio/x-mpeg, audio/x-pn-realaudio, audio/mpeg, audio/x-aiff,\
+            audio/basic, audio/x-wav, audio/mp4, audio/ogg, audio/flac,\
+            audio/aac, audio/ac3, audio/x-ms-wma, audio/vnd.wave, audio/mp3,\
+            video/ogg
+
+
+# Only subtitles in VTT format are supported by the Video JS Player
+subtitles.supported.mimetypes = text/vtt
+
+audiowaveform.path = /usr/bin/audiowaveform
+ffmpeg.path = /usr/bin/ffmpeg
+ffprobe.path = /usr/bin/ffprobe
+m4box.path = /usr/bin/MP4Box
+
+stream-hq.height = 720
+stream-hq.width = 1280
+stream-mq.height = 480
+stream-mq.width = 640
+stream-lq.height = 180
+stream-lq.width = 320
+stream-original.enabled = true
+
+# use the follow configuration when the curation task have to work only on a certain bundle name
+#input-audio-bundle-name = ORIGINAL
+#input-video-bundle-name = ORIGINAL
+
+# use the following to edit the max width for a thumbnail (Default is 500px)
+# Height is set to keep the aspect ratio so it cannot be edited
+
+filter-media.video.thumbnail.max-width = 500
+filter-media.audio.thumbnail.max-width = 200
+filter-media.audio.thumbnail.max-height = 100
+
+# The following configuration allows to skip the retrieval
+# of several scenes in a specified time period (seconds)
+#scene.discading.interval = 30
+
+scene.thumbnail.max-width = 500
+video.thumbnail.max-number = 20
+
+### Configure the transcribe service ###
+
+# Choose the correct service plugin
+av-stream.transcribe-service.plugin = whisper-service
+#av-stream.transcribe-service.plugin = aws-transcribe-service
+
+
+### WHISPER Transcribe Configuration ###
+
+whisper.fullpath = ~/.local/bin/whisper
+
+### Whisper models are: tiny - base - small - medium - large
+whisper.default.transcription-model = base
+
+whisper.transcription-format = vtt
diff --git a/dspace/config/modules/deduplication.cfg b/dspace/config/modules/deduplication.cfg
@@ -35,8 +35,9 @@ deduplication.facet-limit = 150
 #------------DEDUPLICATION / DATAQUALITY CONFIGURATIONS------------#
 #------------------------------------------------------------------#
 #                                                                  #
-# Configurations for the Deduplication / DataQuality features      #
-#                                                                  #
+# The configurations below are related to the extra                #
+# functionalities provided by the 4Science Data Quality addon      #
+# for more information see https://4science.com/dspace-add-ons/    #
 #------------------------------------------------------------------#
 # metadata here listed will be excluded by merge tool logic
 merge.excluded-metadata = dc.description.provenance, datacite.rights, dc.date.accessioned, dc.date.modified, dc.date.available, dspace.entity.type
diff --git a/dspace/config/modules/iiif.cfg b/dspace/config/modules/iiif.cfg
@@ -5,7 +5,7 @@ iiif.enabled = false
 
 # Public base url of a iiif server able to serve DSpace images. The bitstream
 # uuid is appended to this URL
-iiif.image.server = http://localhost:8182/iiif/2/
+iiif.image.server = ${dspace.ui.url}/iiif-server/
 
 # Base URL of the search service used to support the (experimental) IIIF Search
 # capability. The actual path will depend on how search is being supported.
@@ -110,6 +110,15 @@ iiif.search.rows = 120
 # If you choose to encode utf8 for solr indexing, set this property to true.
 iiif.search.index.xml.encode = false
 
+#------------------------------------------------------------------#
+#------------        IIIF Plus CONFIGURATIONS          ------------#
+#------------------------------------------------------------------#
+#                                                                  #
+# The configurations below are related to the extra                #
+# functionalities provided by the 4Science IIIF Plus addon         #
+# for more information see https://4science.com/dspace-add-ons/    #
+#------------------------------------------------------------------#
+
 # Valid formats
 iiif.valid-formats = JPEG
 iiif.valid-formats = GIF

diff --git a/dspace/config/modules/networklab.cfg b/dspace/config/modules/networklab.cfg
@@ -1,7 +1,9 @@
 #---------------------------------------------------------------------#
 #-------------------Networklab Addon CONFIGURATIONS-------------------#
 #---------------------------------------------------------------------#
-#         Configuration properties used by the addon module           #
+# This configuration file is related to the 4Science                  #
+# Network Labs addon                                                  #
+# for more information see https://4science.com/dspace-add-ons/       #
 #---------------------------------------------------------------------#
 
 # Url of the neo4j instance

diff --git a/dspace/config/modules/ocr.cfg b/dspace/config/modules/ocr.cfg
@@ -1,4 +1,6 @@
 #### OCR CONFIGURATION ####
+# This configuration file is related to the 4Science OCR addon
+# for more information see https://4science.com/dspace-add-ons/
 
 # IIIF Search settings are defined in iiif module file, so see the iiif.cfg file
 

diff --git a/dspace/config/registries/av-types.xml b/dspace/config/registries/av-types.xml
@@ -0,0 +1,151 @@
+<dspace-dc-types>
+
+    <dspace-header>
+        <title>DSpace Audio/Video Addon Schema</title>
+        <contributor.author>Stefano Maffei</contributor.author>
+    </dspace-header>
+
+    <dc-schema>
+        <name>bitstream</name>
+        <namespace>https://4science.it/audiovideo</namespace>
+    </dc-schema>
+
+    <dc-schema>
+        <name>dash</name>
+        <namespace>https://4science.it/dash</namespace>
+    </dc-schema>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>dashbundle</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>dash</schema>
+        <element>manifest</element>
+        <qualifier></qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>height</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>width</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>duration</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>format</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>codec</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>fps</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>audioformat</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>video</element>
+        <qualifier>audiocodec</qualifier>
+        <scope_note></scope_note>
+    </dc-type>
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>audio</element>
+        <qualifier>peaks</qualifier>
+        <scope_note></scope_note>
+    </dc-type>  
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>scene</element>
+        <qualifier>timestamp</qualifier>
+        <scope_note></scope_note>
+    </dc-type>      
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>category</element>
+        <qualifier></qualifier>
+        <scope_note></scope_note>
+    </dc-type>          
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>transcribe</element>
+        <qualifier>request</qualifier>
+        <scope_note></scope_note>
+    </dc-type>              
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>transcribe</element>
+        <qualifier>processed</qualifier>
+        <scope_note></scope_note>
+    </dc-type>                  
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>whisper</element>
+        <qualifier>model</qualifier>
+        <scope_note></scope_note>
+    </dc-type>  
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>transcribe</element>
+        <qualifier>request</qualifier>
+        <scope_note></scope_note>
+    </dc-type>              
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>transcribe</element>
+        <qualifier>processed</qualifier>
+        <scope_note></scope_note>
+    </dc-type>                  
+
+    <dc-type>
+        <schema>bitstream</schema>
+        <element>whisper</element>
+        <qualifier>model</qualifier>
+        <scope_note></scope_note>
+    </dc-type>  
+
+</dspace-dc-types>