Merge M4A development into main (#1825)

* BXC-4630 derivative MP3 from WAV (#1789)

* generate mp3 derivatives from wav files

* generate mp3 derivatives from wav files

* fix destroy derivative and enhancement tests

* fix javadoc, unique startuporder value and route id, add log statements/test/assertions, use FcrepoJmsConstants, update context file

* remove constructor, rename bean to addAudioAccessCopyProcessor, add test, update context files

* rename bean addAudioAccessCopyProcessor in context file

* add audio/wave and audio/x-wave to mimetypes

* switch mp3 to m4a

* add audioDerivativeProcessor to service-context

* change audio temp path from -access to -audio

* add audio mimetypes to pattern (#1798)

* use m4a derivates for access

* fix test

* add audio access datastream to isAudio, setSoundContent, and IIIFv3ViewableFilter

* soundContent conditional, add contentObject id to getAccessPath, fix audioDs resolution

* use cover viewer if audio access datastream present

* remove changes to FullRecordController, add audio datastream to hasViewableFiles and performQuery

* fix test

* add audio derivative to DerivativeServiceTest

* Code climate

---------

Co-authored-by: krwong <[email protected]>
Co-authored-by: krwong <[email protected]>

Author: bbpennel

auth-api/src/main/java/edu/unc/lib/boxc/auth/api/services/DatastreamPermissionUtil.java

@@ -23,6 +23,7 @@ public class DatastreamPermissionUtil {
         DS_PERMISSION_MAP = new EnumMap<>(DatastreamType.class);
         DS_PERMISSION_MAP.put(DatastreamType.FULLTEXT_EXTRACTION, Permission.viewHidden);
         DS_PERMISSION_MAP.put(DatastreamType.JP2_ACCESS_COPY, Permission.viewAccessCopies);
+        DS_PERMISSION_MAP.put(DatastreamType.AUDIO_ACCESS_COPY, Permission.viewAccessCopies);
         DS_PERMISSION_MAP.put(DatastreamType.ACCESS_SURROGATE, Permission.viewAccessCopies);
         DS_PERMISSION_MAP.put(DatastreamType.MD_DESCRIPTIVE, Permission.viewMetadata);
         DS_PERMISSION_MAP.put(DatastreamType.MD_DESCRIPTIVE_HISTORY, Permission.viewHidden);

indexing-solr/src/test/java/edu/unc/lib/boxc/indexing/solr/filter/SetDatastreamFilterTest.java

@@ -45,6 +45,7 @@
 import java.util.stream.Collectors;
 
 import static edu.unc.lib.boxc.indexing.solr.test.MockRepositoryObjectHelpers.makeFileObject;
+import static edu.unc.lib.boxc.model.api.DatastreamType.AUDIO_ACCESS_COPY;
 import static edu.unc.lib.boxc.model.api.DatastreamType.JP2_ACCESS_COPY;
 import static edu.unc.lib.boxc.model.api.DatastreamType.ORIGINAL_FILE;
 import static edu.unc.lib.boxc.model.api.DatastreamType.TECHNICAL_METADATA;
@@ -118,6 +119,7 @@ public class SetDatastreamFilterTest {
     private static final long PREMIS_SIZE = 893l;
 
     private static final long JP2_SIZE = 11;
+    private static final long AUDIO_SIZE = 11;
 
     private AutoCloseable closeable;
 
@@ -255,6 +257,8 @@ public void fileObjectAudioOnlyBinaryTest() throws Exception {
                 fileResource(TECHNICAL_METADATA.getId(), FILE2_SIZE, FILE2_MIMETYPE, FILE2_NAME, FILE2_DIGEST));
         when(binObj2.getBinaryStream()).thenReturn(getClass().getResourceAsStream("/datastream/techmd_mp3.xml"));
         when(fileObj.getBinaryObjects()).thenReturn(Arrays.asList(binObj, binObj2));
+        List<Derivative> derivs = makeAudioDerivative();
+        when(derivativeService.getDerivatives(pid)).thenReturn(derivs);
         dip.setContentObject(fileObj);
 
         filter.filter(dip);
@@ -263,6 +267,8 @@ public void fileObjectAudioOnlyBinaryTest() throws Exception {
                 FILE_MP3_SIZE, FILE_MP3_MIMETYPE, FILE_MP3_NAME, FILE_MP3_DIGEST, null, FILE_MP3_EXTENT);
         assertContainsDatastream(idb.getDatastream(), TECHNICAL_METADATA.getId(),
                 FILE2_SIZE, FILE2_MIMETYPE, FILE2_NAME, FILE2_DIGEST, null, null);
+        assertContainsDatastream(idb.getDatastream(), AUDIO_ACCESS_COPY.getId(),
+                AUDIO_SIZE, AUDIO_ACCESS_COPY.getMimetype(), "access.m4a", null, null, null);
     }
 
     @Test
@@ -297,6 +303,8 @@ public void fileObjectAudioOnlyBinaryWithDotMillisecondsSeperatorTest() throws E
                 fileResource(TECHNICAL_METADATA.getId(), FILE2_SIZE, FILE2_MIMETYPE, FILE2_NAME, FILE2_DIGEST));
         when(binObj2.getBinaryStream()).thenReturn(getClass().getResourceAsStream("/datastream/techmd_dot_separated_milliseconds.xml"));
         when(fileObj.getBinaryObjects()).thenReturn(Arrays.asList(binObj, binObj2));
+        List<Derivative> derivs = makeAudioDerivative();
+        when(derivativeService.getDerivatives(pid)).thenReturn(derivs);
         dip.setContentObject(fileObj);
 
         filter.filter(dip);
@@ -305,6 +313,8 @@ public void fileObjectAudioOnlyBinaryWithDotMillisecondsSeperatorTest() throws E
                 FILE_MP3_SIZE, FILE_MP3_MIMETYPE, FILE_MP3_NAME, FILE_MP3_DIGEST, null, FILE_MP3_EXTENT);
         assertContainsDatastream(idb.getDatastream(), TECHNICAL_METADATA.getId(),
                 FILE2_SIZE, FILE2_MIMETYPE, FILE2_NAME, FILE2_DIGEST, null, null);
+        assertContainsDatastream(idb.getDatastream(), AUDIO_ACCESS_COPY.getId(),
+                AUDIO_SIZE, AUDIO_ACCESS_COPY.getMimetype(), "access.m4a", null, null, null);
     }
 
     @Test
@@ -768,4 +778,11 @@ private List<Derivative> makeJP2Derivative() throws IOException {
 
         return List.of(new Derivative(JP2_ACCESS_COPY, jp2File));
     }
+
+    private List<Derivative> makeAudioDerivative() throws IOException {
+        File m4aFile = derivDir.resolve("access.m4a").toFile();
+        FileUtils.write(m4aFile, "m4a content", "UTF-8");
+
+        return List.of(new Derivative(AUDIO_ACCESS_COPY, m4aFile));
+    }
 }

model-api/src/main/java/edu/unc/lib/boxc/model/api/DatastreamType.java

@@ -13,6 +13,7 @@
  */
 public enum DatastreamType {
     ACCESS_SURROGATE("access_surrogate", "application/octet-stream", null, null, EXTERNAL),
+    AUDIO_ACCESS_COPY("audio", "audio/aac", "m4a", null, EXTERNAL),
     FULLTEXT_EXTRACTION("fulltext", "text/plain", "txt", null, EXTERNAL),
     JP2_ACCESS_COPY("jp2", "image/jp2", "jp2", null, EXTERNAL),
     MD_DESCRIPTIVE("md_descriptive", "text/xml", "xml", METADATA_CONTAINER, INTERNAL),

model-fcrepo/src/test/java/edu/unc/lib/boxc/model/fcrepo/services/DerivativeServiceTest.java

@@ -14,6 +14,7 @@
 import java.nio.file.Paths;
 import java.util.List;
 
+import static edu.unc.lib.boxc.model.api.DatastreamType.AUDIO_ACCESS_COPY;
 import static edu.unc.lib.boxc.model.api.DatastreamType.FULLTEXT_EXTRACTION;
 import static edu.unc.lib.boxc.model.api.DatastreamType.JP2_ACCESS_COPY;
 import static edu.unc.lib.boxc.model.api.DatastreamType.ORIGINAL_FILE;
@@ -73,18 +74,22 @@ public void testGetDerivativeNotExist() throws Exception {
     public void testGetDerivatives() throws Exception {
         File originalDerivFile1 = createDerivative(pid, FULLTEXT_EXTRACTION);
         File originalDerivFil21 = createDerivative(pid, JP2_ACCESS_COPY);
+        File originalDerivFile3 = createDerivative(pid, AUDIO_ACCESS_COPY);
 
         List<Derivative> derivs = derivativeService.getDerivatives(pid);
-        assertEquals(2, derivs.size());
+        assertEquals(3, derivs.size());
 
         Derivative textDeriv = findDerivative(derivs, FULLTEXT_EXTRACTION);
         Derivative jp2Deriv = findDerivative(derivs, JP2_ACCESS_COPY);
+        Derivative audioDeriv = findDerivative(derivs, AUDIO_ACCESS_COPY);
 
         assertNotNull(textDeriv);
         assertNotNull(jp2Deriv);
+        assertNotNull(audioDeriv);
 
         assertEquals(originalDerivFile1, textDeriv.getFile());
         assertEquals(originalDerivFil21, jp2Deriv.getFile());
+        assertEquals(originalDerivFile3, audioDeriv.getFile());
     }
 
     @Test

search-solr/src/main/java/edu/unc/lib/boxc/search/solr/filters/IIIFv3ViewableFilter.java

@@ -23,7 +23,8 @@ public String toFilterString() {
         var fileTypeFilter = getFileTypes().stream().map( type -> fileTypeField + ":" + type)
                 .collect(Collectors.joining(" OR ", "(", ")"));
         var datastreamFilter = SearchFieldKey.DATASTREAM.getSolrField() + ":" + DatastreamType.JP2_ACCESS_COPY.getId() + "|*";
-        return "(" + fileTypeFilter + ") OR (" + datastreamFilter + ")";
+        var datastreamFilterAudio = SearchFieldKey.DATASTREAM.getSolrField() + ":" + DatastreamType.AUDIO_ACCESS_COPY.getId() + "|*";
+        return "(" + fileTypeFilter + ") OR (" + datastreamFilter + ") OR (" + datastreamFilterAudio + ")";
     }
 
     @Override

services-camel-app/src/main/java/edu/unc/lib/boxc/services/camel/audio/AudioDerivativeProcessor.java

@@ -0,0 +1,51 @@
+package edu.unc.lib.boxc.services.camel.audio;
+
+import edu.unc.lib.boxc.services.camel.util.CdrFcrepoHeaders;
+import org.apache.camel.Exchange;
+import org.apache.camel.Message;
+import org.apache.camel.Processor;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+import java.util.regex.Pattern;
+
+/**
+ * Processor which validates and prepares audio objects for producing derivatives
+ * @author krwong
+ */
+public class AudioDerivativeProcessor implements Processor {
+    private static final Logger log = LoggerFactory.getLogger(AudioDerivativeProcessor.class);
+
+    private static final Pattern MIMETYPE_PATTERN =
+            Pattern.compile("^(audio.(basic|mpeg|mp4|x-aiff|x-ms-wma|x-wave|x-wav|wav|wave|3gpp))$");
+
+    /**
+     * Returns true if the subject of the exchange is a binary which
+     * is eligible for having audio derivatives generated from it.
+     * @param exchange
+     * @return
+     */
+    public static boolean allowedAudioType(Exchange exchange) {
+        Message in = exchange.getIn();
+        String mimetype = (String) in.getHeader(CdrFcrepoHeaders.CdrBinaryMimeType);
+        String binPath = (String) in.getHeader(CdrFcrepoHeaders.CdrBinaryPath);
+
+        if (!MIMETYPE_PATTERN.matcher(mimetype).matches()) {
+            log.debug("File type {} on object {} is not applicable for audio derivatives", mimetype, binPath);
+            return false;
+        }
+
+        log.debug("Object {} with type {} is permitted for audio derivatives", binPath, mimetype);
+        return true;
+    }
+
+    @Override
+    public void process(Exchange exchange) throws Exception {
+        Message in = exchange.getIn();
+        String mimetype = (String) in.getHeader(CdrFcrepoHeaders.CdrBinaryMimeType);
+
+        String binPath = (String) in.getHeader(CdrFcrepoHeaders.CdrBinaryPath);
+        log.debug("Keeping existing audio path as {} for type {}", binPath, mimetype);
+        in.setHeader(CdrFcrepoHeaders.CdrAudioPath, binPath);
+    }
+}

services-camel-app/src/main/java/edu/unc/lib/boxc/services/camel/audio/AudioEnhancementsRouter.java

@@ -0,0 +1,62 @@
+package edu.unc.lib.boxc.services.camel.audio;
+
+import edu.unc.lib.boxc.model.api.exceptions.RepositoryException;
+import edu.unc.lib.boxc.services.camel.images.AddDerivativeProcessor;
+import edu.unc.lib.boxc.services.camel.util.CdrFcrepoHeaders;
+import org.apache.camel.BeanInject;
+import org.apache.camel.LoggingLevel;
+import org.apache.camel.builder.RouteBuilder;
+import org.apache.camel.spi.UuidGenerator;
+import org.apache.camel.support.DefaultUuidGenerator;
+import org.slf4j.Logger;
+
+import static org.slf4j.LoggerFactory.getLogger;
+
+/**
+ * Router which triggers the creation of audio derivatives
+ * @author krwong
+ */
+public class AudioEnhancementsRouter extends RouteBuilder {
+    private static final Logger log = getLogger(AudioEnhancementsRouter.class);
+
+    @BeanInject(value = "addAudioAccessCopyProcessor")
+    private AddDerivativeProcessor addAudioAccessCopyProcessor;
+
+    private UuidGenerator uuidGenerator;
+
+    /**
+     * Configure the audio enhancement route workflow.
+     */
+    @Override
+    public void configure() throws Exception {
+        AudioDerivativeProcessor audioDerivProcessor = new AudioDerivativeProcessor();
+
+        uuidGenerator = new DefaultUuidGenerator();
+
+        onException(RepositoryException.class)
+                .redeliveryDelay("{{error.retryDelay}}")
+                .maximumRedeliveries("{{error.maxRedeliveries}}")
+                .backOffMultiplier("{{error.backOffMultiplier}}")
+                .retryAttemptedLogLevel(LoggingLevel.WARN);
+
+        from("direct:process.enhancement.audioAccessCopy")
+                .routeId("AudioAccessCopy")
+                .startupOrder(25)
+                .log(LoggingLevel.DEBUG, log, "Access copy triggered")
+                .filter().method(addAudioAccessCopyProcessor, "needsRun")
+                .filter().method(audioDerivProcessor, "allowedAudioType")
+                    .bean(audioDerivProcessor)
+                    .log(LoggingLevel.INFO, log, "Creating/Updating AAC access copy for ${headers[CdrAudioPath]}")
+                    // Generate an random identifier to avoid derivative collisions
+                    .setBody(exchange -> uuidGenerator.generateUuid())
+                    .setHeader(CdrFcrepoHeaders.CdrTempPath, simple("${properties:services.tempDirectory}/${body}-audio"))
+                    .doTry()
+                        .recipientList(simple("exec:/bin/sh?args=${properties:cdr.enhancement.bin}/convertWav.sh "
+                                + "${headers[CdrAudioPath]} ${headers[CdrTempPath]}"))
+                        .bean(addAudioAccessCopyProcessor)
+                    .endDoTry()
+                    .doFinally()
+                        .bean(addAudioAccessCopyProcessor, "cleanupTempFile")
+                .end();
+    }
+}

services-camel-app/src/main/java/edu/unc/lib/boxc/services/camel/destroyDerivatives/DestroyDerivativesRouter.java

@@ -2,6 +2,8 @@
 
 import edu.unc.lib.boxc.services.camel.fulltext.FulltextProcessor;
 import edu.unc.lib.boxc.services.camel.images.ImageDerivativeProcessor;
+import edu.unc.lib.boxc.services.camel.audio.AudioDerivativeProcessor;
+
 import org.apache.camel.BeanInject;
 import org.apache.camel.LoggingLevel;
 import org.apache.camel.PropertyInject;
@@ -28,6 +30,9 @@ public class DestroyDerivativesRouter extends RouteBuilder {
     @BeanInject(value = "destroyFulltextProcessor")
     private DestroyDerivativesProcessor destroyFulltextProcessor;
 
+    @BeanInject(value = "destroyAudioProcessor")
+    private DestroyDerivativesProcessor destroyAudioProcessor;
+
     private String destroyDerivativesStreamCamel;
     private long errorRetryDelay;
     private int errorMaxRedeliveries;
@@ -51,6 +56,8 @@ public void configure() throws Exception {
                         .to("direct:image.derivatives.destroy")
                     .when(method(FulltextProcessor.class, "allowedTextType"))
                         .to("direct:fulltext.derivatives.destroy")
+                    .when(method(AudioDerivativeProcessor.class, "allowedAudioType"))
+                        .to("direct:audio.derivatives.destroy")
                 .end();
 
         from("direct:fulltext.derivatives.destroy")
@@ -64,6 +71,12 @@ public void configure() throws Exception {
                 .startupOrder(202)
                 .log(LoggingLevel.DEBUG, log, "Destroying access copy derivatives")
                 .bean(destroyAccessCopyProcessor);
+
+        from("direct:audio.derivatives.destroy")
+                .routeId("CdrDestroyAudio")
+                .startupOrder(199)
+                .log(LoggingLevel.DEBUG, log, "Destroying derivative audio files")
+                .bean(destroyAudioProcessor);
     }
 
     public void setDestroyedMsgProcessor(DestroyedMsgProcessor destroyedMsgProcessor) {
@@ -78,6 +91,10 @@ public void setDestroyFulltextProcessor(DestroyDerivativesProcessor destroyFullt
         this.destroyFulltextProcessor = destroyFulltextProcessor;
     }
 
+    public void setDestroyAudioProcessor(DestroyDerivativesProcessor destroyAudioProcessor) {
+        this.destroyAudioProcessor = destroyAudioProcessor;
+    }
+
     @PropertyInject("cdr.destroy.derivatives.stream.camel")
     public void setDestroyDerivativesStreamCamel(String destroyDerivativesStreamCamel) {
         this.destroyDerivativesStreamCamel = destroyDerivativesStreamCamel;

services-camel-app/src/main/java/edu/unc/lib/boxc/services/camel/enhancements/EnhancementRouter.java

@@ -37,7 +37,7 @@ public class EnhancementRouter extends RouteBuilder {
     @PropertyInject(value = "cdr.enhancement.processingThreads")
     private Integer enhancementThreads;
 
-    private static final String DEFAULT_ENHANCEMENTS = "imageAccessCopy,extractFulltext";
+    private static final String DEFAULT_ENHANCEMENTS = "imageAccessCopy,extractFulltext,audioAccessCopy";
     @Override
     public void configure() throws Exception {
 

services-camel-app/src/main/java/edu/unc/lib/boxc/services/camel/util/CdrFcrepoHeaders.java

@@ -22,6 +22,9 @@ public abstract class CdrFcrepoHeaders {
     // URI identifying the location
     public static final String CdrImagePath = "CdrImagePath";
 
+    // URI identifying the location
+    public static final String CdrAudioPath = "CdrAudioPath";
+
     // File path for a temp file
     public static final String CdrTempPath = "CdrTempPath";
 

services-camel-app/src/main/webapp/WEB-INF/service-context.xml

@@ -223,6 +223,14 @@
         <constructor-arg value="${cdr.enhancement.path.fulltext}" />
     </bean>
 
+    <bean id="addAudioAccessCopyProcessor" class="edu.unc.lib.boxc.services.camel.images.AddDerivativeProcessor">
+        <constructor-arg value="#{T(edu.unc.lib.boxc.model.api.DatastreamType).AUDIO_ACCESS_COPY.getExtension()}" />
+        <constructor-arg value="${cdr.enhancement.path.audio}" />
+    </bean>
+
+    <bean id="audioAccessCopyProcessor" class="edu.unc.lib.boxc.services.camel.audio.AudioDerivativeProcessor">
+    </bean>
+
     <bean id="destroyedMsgProcessor" class="edu.unc.lib.boxc.services.camel.destroyDerivatives.DestroyedMsgProcessor">
         <property name="jp2BasePath" value="${cdr.enhancement.path.jp2}" />
     </bean>
@@ -241,6 +249,11 @@
         <constructor-arg value="#{T(edu.unc.lib.boxc.model.api.DatastreamType).FULLTEXT_EXTRACTION.getExtension()}" />
         <constructor-arg value="${cdr.enhancement.path.fulltext}" />
     </bean>
+
+    <bean id="destroyAudioProcessor" class="edu.unc.lib.boxc.services.camel.destroyDerivatives.DestroyDerivativesProcessor">
+        <constructor-arg value="#{T(edu.unc.lib.boxc.model.api.DatastreamType).AUDIO_ACCESS_COPY.getExtension()}" />
+        <constructor-arg value="${cdr.enhancement.path.audio}" />
+    </bean>
 
     <bean id="indexingMessageProcessor" class="edu.unc.lib.boxc.services.camel.triplesReindexing.IndexingMessageProcessor">
     </bean>
@@ -547,6 +560,7 @@
         <camel:package>edu.unc.lib.boxc.services.camel.enhancements</camel:package>
         <camel:package>edu.unc.lib.boxc.services.camel.images</camel:package>
         <camel:package>edu.unc.lib.boxc.services.camel.fulltext</camel:package>
+        <camel:package>edu.unc.lib.boxc.services.camel.audio</camel:package>
         <camel:package>edu.unc.lib.boxc.services.camel.solr</camel:package>
         <camel:package>edu.unc.lib.boxc.services.camel.binaryCleanup</camel:package>
         <!-- Initialize metaServicesRouter after the routes it depends on -->