download selected files from within a dataset

For code simplicity, still includes bagit support, dataset metadata, etc.

Author: Max Burnette

app/api/Datasets.scala

@@ -2068,7 +2068,9 @@ class  Datasets @Inject()(
     * @return Enumerator to produce array of bytes from a zipped stream containing the bytes of each file
     *         in the dataset
     */
-  def enumeratorFromDataset(dataset: Dataset, chunkSize: Int = 1024 * 8, compression: Int = Deflater.DEFAULT_COMPRESSION, bagit: Boolean, user : Option[User])
+  def enumeratorFromDataset(dataset: Dataset, chunkSize: Int = 1024 * 8,
+                            compression: Int = Deflater.DEFAULT_COMPRESSION, bagit: Boolean,
+                            user : Option[User], fileIDs: Option[List[UUID]])
                            (implicit ec: ExecutionContext): Enumerator[Array[Byte]] = {
     implicit val pec = ec.prepare()
     val dataFolder = if (bagit) "data/" else ""
@@ -2077,7 +2079,17 @@ class  Datasets @Inject()(
 
     // compute list of all files and folder in dataset. This will also make sure
     // that all files and folder names are unique.
-    listFilesInFolder(dataset.files, dataset.folders, dataFolder, filenameMap, inputFiles)
+    fileIDs match {
+      case Some(fids) => {
+        Logger.info("Downloading only some files")
+        Logger.info(fids.toString)
+        listFilesInFolder(fids, List.empty, dataFolder, filenameMap, inputFiles)
+      }
+      case None => {
+        // TODO: folderId match ...
+        listFilesInFolder(dataset.files, dataset.folders, dataFolder, filenameMap, inputFiles)
+      }
+    }
 
     val md5Files = scala.collection.mutable.HashMap.empty[String, MessageDigest] //for the files
     val md5Bag = scala.collection.mutable.HashMap.empty[String, MessageDigest] //for the bag files
@@ -2121,14 +2133,13 @@ class  Datasets @Inject()(
      * the enumerator is finished
      */
 
-    var is: Option[InputStream] = addDatasetInfoToZip(dataFolder,dataset,zip)
+    var is: Option[InputStream] = addDatasetInfoToZip(dataFolder, dataset, zip)
     //digest input stream
     val md5 = MessageDigest.getInstance("MD5")
     md5Files.put(dataFolder+"_info.json",md5)
     is = Some(new DigestInputStream(is.get,md5))
     file_type = 1 //next is metadata
 
-
     Enumerator.generateM({
       is match {
         case Some(inputStream) => {
@@ -2415,7 +2426,7 @@ class  Datasets @Inject()(
 
         // Use custom enumerator to create the zip file on the fly
         // Use a 1MB in memory byte array
-        Ok.chunked(enumeratorFromDataset(dataset,1024*1024, compression,bagit,user)).withHeaders(
+        Ok.chunked(enumeratorFromDataset(dataset,1024*1024, compression, bagit, user, None)).withHeaders(
           CONTENT_TYPE -> "application/zip",
           CONTENT_DISPOSITION -> (FileUtils.encodeAttachment(dataset.name+ ".zip", request.headers.get("user-agent").getOrElse("")))
         )
@@ -2427,6 +2438,30 @@ class  Datasets @Inject()(
     }
   }
 
+  def downloadPartial(id: UUID, fileList: String) = PermissionAction(Permission.DownloadFiles, Some(ResourceRef(ResourceRef.dataset, id))) { implicit request =>
+    implicit val user = request.user
+    datasets.get(id) match {
+      case Some(dataset) => {
+        val fileIDs = fileList.split(',').map(fid => new UUID(fid)).toList
+        val bagit = play.api.Play.configuration.getBoolean("downloadDatasetBagit").getOrElse(true)
+
+        // Increment download count for each file
+        fileIDs.foreach(fid => files.incrementDownloads(fid, user))
+
+        // Use custom enumerator to create the zip file on the fly
+        // Use a 1MB in memory byte array
+        Ok.chunked(enumeratorFromDataset(dataset,1024*1024, -1, bagit, user, Some(fileIDs))).withHeaders(
+          CONTENT_TYPE -> "application/zip",
+          CONTENT_DISPOSITION -> (FileUtils.encodeAttachment(dataset.name+ " (Partial).zip", request.headers.get("user-agent").getOrElse("")))
+        )
+      }
+      // If the dataset wasn't found by ID
+      case None => {
+        NotFound
+      }
+    }
+  }
+
   def updateAccess(id:UUID, access:String) = PermissionAction(Permission.PublicDataset, Some(ResourceRef(ResourceRef.dataset, id))) { implicit request =>
     implicit val user = request.user
     user match {

app/controllers/Application.scala

@@ -333,6 +333,7 @@ class Application @Inject() (files: FileService, collections: CollectionService,
         api.routes.javascript.Datasets.unfollow,
         api.routes.javascript.Datasets.detachFile,
         api.routes.javascript.Datasets.download,
+        api.routes.javascript.Datasets.downloadPartial,
         api.routes.javascript.Datasets.getPreviews,
         api.routes.javascript.Datasets.updateAccess,
         api.routes.javascript.Datasets.addFileEvent,

app/views/datasets/filesAndFolders.scala.html

@@ -89,7 +89,7 @@
                     //'<li><a onclick="showOnlyMarked();"><span class="glyphicon glyphicon-ok"/>&nbsp;Show Only Marked</a></li>' +
                     //'<li><a onclick="showAllFiles();"><span class="glyphicon glyphicon-ok"/>&nbsp;Show All Files</a></li>' +
                     // TODO: Support for downloading by folder and marked list will be implemented in the future
-                    //'<li><a onclick="downloadMarked();"><span class="glyphicon glyphicon-download-alt"/>&nbsp;Download All</a></li>' +
+                    '<li><a onclick="downloadMarked();"><span class="glyphicon glyphicon-download-alt"/>&nbsp;Download All</a></li>' +
                     '<li><a onclick="confirmTagMarked();"><span class="glyphicon glyphicon-tag"/>&nbsp;Tag All</a></li>' +
                     '<li><a onclick="confirmDeleteMarked();"><span class="glyphicon glyphicon-trash"/>&nbsp;Delete All</a></li>' +
                     '<li><a onclick="clearMarked();"><span class="glyphicon glyphicon-erase"/>&nbsp;Clear All</a></li></ul>' +
@@ -160,12 +160,17 @@
     }
 
     function showOnlyMarked() {
-        confirmDeleteMarked('@dataset.id');
+        console.log("Not implemented.")
     }
 
+    // Download selected files as Zip
     function downloadMarked() {
-        // TODO: Downloading selected files as Zip - look at downloading folders tasks
-        console.log("Not implemented.")
+        var selected = $.cookie('[email protected]');
+        if (selected) {
+            window.open(jsRoutes.api.Datasets.downloadPartial('@dataset.id', selected).url, '_blank');
+        } else {
+            notify("No files selected.")
+        }
     }
 
     function confirmTagMarked() {

conf/routes

@@ -574,6 +574,7 @@ GET            /api/datasets/:id/files
 POST           /api/datasets/:id/files                                                  @api.Datasets.uploadToDatasetFile(id: UUID)
 POST           /api/datasets/:id/urls                                                   @api.Datasets.uploadToDatasetJSON(id: UUID)
 GET            /api/datasets/:id/download                                               @api.Datasets.download(id: UUID, compression: Int ?= -1, tracking: Boolean ?= true)
+GET            /api/datasets/:id/downloadPartial                                        @api.Datasets.downloadPartial(id: UUID, fileList: String)
 POST           /api/datasets/:id/comment                                                @api.Datasets.comment(id: UUID)
 POST           /api/datasets/:id/reindex                                                @api.Datasets.reindex(id:UUID, recursive: Boolean ?= true)
 POST           /api/datasets/:id/follow                                                 @api.Datasets.follow(id: UUID)