// Create workflow start provenance message
private static void recordStart(Context c, Item myitem)
throws SQLException, IOException, AuthorizeException {
// Get non-internal format bitstreams
Bitstream[] bitstreams = myitem.getNonInternalBitstreams();
// get date
DCDate now = DCDate.getCurrent();
// Create provenance description
String provmessage = "";
if (myitem.getSubmitter() != null) {
provmessage =
"Submitted by "
+ myitem.getSubmitter().getFullName()
+ " ("
+ myitem.getSubmitter().getEmail()
+ ") on "
+ now.toString()
+ "\n";
} else
// null submitter
{
provmessage = "Submitted by unknown (probably automated) on" + now.toString() + "\n";
}
// add sizes and checksums of bitstreams
provmessage += InstallItem.getBitstreamProvenanceMessage(myitem);
// Add message to the DC
myitem.addDC("description", "provenance", "en", provmessage);
myitem.update();
}
/**
* rejects an item - rejection means undoing a submit - WorkspaceItem is created, and the
* WorkflowItem is removed, user is emailed rejection_message.
*
* @param c Context
* @param wi WorkflowItem to operate on
* @param e EPerson doing the operation
* @param rejection_message message to email to user
*/
public static WorkspaceItem reject(
Context c, WorkflowItem wi, EPerson e, String rejection_message)
throws SQLException, AuthorizeException, IOException {
// authorize a DSpaceActions.REJECT
// stop workflow
deleteTasks(c, wi);
// rejection provenance
Item myitem = wi.getItem();
// Get current date
String now = DCDate.getCurrent().toString();
// Get user's name + email address
String usersName = getEPersonName(e);
// Here's what happened
String provDescription =
"Rejected by " + usersName + ", reason: " + rejection_message + " on " + now + " (GMT) ";
// Add to item as a DC field
myitem.addDC("description", "provenance", "en", provDescription);
myitem.update();
// convert into personal workspace
WorkspaceItem wsi = returnToWorkspace(c, wi);
// notify that it's been rejected
notifyOfReject(c, wi, e, rejection_message);
log.info(
LogManager.getHeader(
c,
"reject_workflow",
"workflow_item_id="
+ wi.getID()
+ "item_id="
+ wi.getItem().getID()
+ "collection_id="
+ wi.getCollection().getID()
+ "eperson_id="
+ e.getID()));
return wsi;
}
// Record approval provenance statement
private static void recordApproval(Context c, WorkflowItem wi, EPerson e)
throws SQLException, IOException, AuthorizeException {
Item item = wi.getItem();
// Get user's name + email address
String usersName = getEPersonName(e);
// Get current date
String now = DCDate.getCurrent().toString();
// Here's what happened
String provDescription =
"Approved for entry into archive by " + usersName + " on " + now + " (GMT) ";
// add bitstream descriptions (name, size, checksums)
provDescription += InstallItem.getBitstreamProvenanceMessage(item);
// Add to item as a DC field
item.addDC("description", "provenance", "en", provDescription);
item.update();
}
private void crosswalkPDF(Context context, Item item, InputStream metadata)
throws CrosswalkException, IOException, SQLException, AuthorizeException {
COSDocument cos = null;
try {
PDFParser parser = new PDFParser(metadata);
parser.parse();
cos = parser.getDocument();
// sanity check: PDFBox breaks on encrypted documents, so give up.
if (cos.getEncryptionDictionary() != null)
throw new MetadataValidationException(
"This packager cannot accept an encrypted PDF document.");
/* PDF to DC "crosswalk":
*
* NOTE: This is not in a crosswalk plugin because (a) it isn't
* useful anywhere else, and more importantly, (b) the source
* data is not XML so it doesn't fit the plugin's interface.
*
* pattern of crosswalk -- PDF dict entries to DC:
* Title -> title.null
* Author -> contributor.author
* CreationDate -> date.created
* ModDate -> date.created
* Creator -> description.provenance (application that created orig)
* Producer -> description.provenance (convertor to pdf)
* Subject -> description.abstract
* Keywords -> subject.other
* date is java.util.Calendar
*/
PDDocument pd = new PDDocument(cos);
PDDocumentInformation docinfo = pd.getDocumentInformation();
String title = docinfo.getTitle();
// sanity check: item must have a title.
if (title == null)
throw new MetadataValidationException(
"This PDF file is unacceptable, it does not have a value for \"Title\" in its Info dictionary.");
log.debug("PDF Info dict title=\"" + title + "\"");
item.addDC("title", null, "en", title);
String value;
Calendar date;
if ((value = docinfo.getAuthor()) != null) {
item.addDC("contributor", "author", null, value);
log.debug("PDF Info dict author=\"" + value + "\"");
}
if ((value = docinfo.getCreator()) != null)
item.addDC(
"description",
"provenance",
"en",
"Application that created the original document: " + value);
if ((value = docinfo.getProducer()) != null)
item.addDC(
"description", "provenance", "en", "Original document converted to PDF by: " + value);
if ((value = docinfo.getSubject()) != null)
item.addDC("description", "abstract", null, value);
if ((value = docinfo.getKeywords()) != null) item.addDC("subject", "other", null, value);
// Take either CreationDate or ModDate as "date.created",
// Too bad there's no place to put "last modified" in the DC.
Calendar calValue;
if ((calValue = docinfo.getCreationDate()) == null) calValue = docinfo.getModificationDate();
if (calValue != null)
item.addDC("date", "created", null, (new DCDate(calValue.getTime())).toString());
item.update();
} finally {
if (cos != null) cos.close();
}
}
