marco@0: #!usr/bin/env/ python
marco@1: 
marco@3: """ SWORD2 DSpace bulk uploader - v0.2
marco@1: 
marco@1: A python script to submit large numbers of files to a SWORD2-compatible repository, specifically DSpace 1.8x.
marco@1: Built on the SWORD2 python client library: https://bitbucket.org/beno/python-sword2/overview
marco@1: 
marco@1: Dependencies:
marco@1: 
marco@1: - python 2.X
marco@1: 
marco@1: - sword2 library: https://bitbucket.org/beno/python-sword2/src
marco@1: 
marco@1: -----------------------------------
marco@1: Copyright 2012 Marco Fabiani
marco@1: Copyright 2012 Queen Mary, University of London
marco@1: -----------------------------------
marco@1: """
marco@0: 
marco@0: import argparse, getpass, zipfile,os,sys
marco@0: from sword2 import *
marco@0: 
marco@0: # Parse arguments
marco@0: parser = argparse.ArgumentParser(description="Bulk upload to DSpace using SWORDv2.",epilog="If the submission is created successfully, it will remain open to be completed with the necessary metadata and licenses, using the DSpace web interface. The submission can be found in the \"My Account -> Submissions\" section of the user's area.")
marco@0: parser.add_argument("data", type=str, nargs=1,
marco@1:                    help="Accepts: METSDSpaceSIP packages, zip files, directories, single files. NOTE: METSDSpaceSIP packages are only accepted by Collections with a workflow!")
marco@0: parser.add_argument("--username", dest="user_name", type=str,nargs=1, help="DSpace username.")
marco@0: parser.add_argument("--title", dest="title", type=str,nargs=1, help="Title (ignored for METS packages).")
marco@0: parser.add_argument("--author", dest="author", type=str,nargs="+", help="Author(s) (ignored for METS packages). Accepts multiple entries in the format \"Surname, Name\"")
marco@0: parser.add_argument("--date", dest="date", type=str,nargs=1, help="Date of creation (string) (ignored for METS packages).")
marco@1: parser.add_argument("--servicedoc", dest="dspaceurl", type=str,nargs=1, help="Url of the SWORDv2 service document (default: http://c4dm.eecs.qmul.ac.uk/smdmrd-test/swordv2/servicedocument")
marco@0: 
marco@0: args = parser.parse_args()
marco@0: data = args.data[0]
marco@0: if args.dspaceurl == None:
marco@0: 	dspaceurl = "http://c4dm.eecs.qmul.ac.uk/smdmrd-test/swordv2/servicedocument"
marco@0: else:
marco@0: 	dspaceurl = args.dspaceurl[0]
marco@0: 
marco@0: 
marco@0: try:
marco@0: 	# Connect to SWORD server
marco@0: 	attempts = 3 #  Number of attempts left to connect to server
marco@0: 	connected = False
marco@0: 	while attempts>0 and not connected:
marco@0: 		print "Connecting to SWORD server. Remaining attempts: ", attempts
marco@0: 		# Try to login, get service document
marco@0: 		# Get username and password
marco@0: 		if args.user_name == None:
marco@0: 			user_name = raw_input("DSpace Username: ")
marco@0: 		else:
marco@0: 			user_name = args.user_name[0]
marco@0: 			print "DSpace Username: ",user_name
marco@0: 		user_pass = getpass.getpass("DSpace password:")
marco@0: 		# Connect to the server
marco@0: 		c = Connection(dspaceurl, user_name=user_name, user_pass=user_pass,keep_history=False)
marco@3: 		
marco@0: 		# Get service document
marco@0: 		try:
marco@0: 			c.get_service_document()
marco@3: 		except: # Server error
marco@3: 			print "Server unreachable!"
marco@3: 			break
marco@3: 		if c.sd != None:
marco@3: 			connected = True
marco@3: 		else:
marco@0: 			attempts-=1
marco@0: 			print "Incorrect username and/or password"
marco@3: 			
marco@0: 			
marco@0: 	if connected:
marco@0: 		# List available collections
marco@0: 		print "------------------------"
marco@0: 		print "Welcome to the ",c.workspaces[0][0], "repository"
marco@0: 		print "Available Collections: "
marco@0: 		numColl = len(c.workspaces[0][1])
marco@0: 		for ctr in range(numColl):
marco@0: 			coll = c.workspaces[0][1][ctr]
marco@0: 			print ctr+1,":",coll.title
marco@0: 		# Select a collection to deposit into
marco@0: 		sel = -1
marco@0: 		while (sel<=0 or sel>numColl):
marco@0: 			sel = input("Select a Collection to submit your files into: ")
marco@0: 		collection = c.workspaces[0][1][sel-1]
marco@0: 		print "Selected Collection: ",collection.title
marco@0: 		
marco@0: 		# Create a submission: build the zip files
marco@0: 		temp = True # delete the zip file at the end of the upload	
marco@0: 		if zipfile.is_zipfile(data):
marco@0: 			zipf = data
marco@0: 			temp = False
marco@0: 		elif os.path.isfile(data): # This is a single file
marco@0: 			dataname = os.path.basename(data)
marco@0: 			zipf = os.path.splitext(dataname)[0]+".zip"
marco@0: 			myzip = zipfile.ZipFile(zipf, "w")
marco@3: 			myzip.write(data,os.path.basename(data))
marco@0: 			myzip.close()
marco@3: 		elif os.path.isdir(data): # This is a directory, zip all the files and maintain the structure, but start from the base only...
marco@0: 			dataname = os.path.basename(os.path.normpath(data))
marco@0: 			zipf = dataname+".zip"
marco@0: 			myzip = zipfile.ZipFile(zipf, "w")
marco@0: 			# get the directory structure
marco@3: 			print "Creating a zip archive for submission..."
marco@0: 			for root, dirs, files in os.walk(data):
marco@0: 				for name in files:
marco@3: 					myzip.write(os.path.join(root,name),
marco@3: 						os.path.relpath(os.path.join(root,name),data))
marco@0: 			myzip.close()
marco@0: 		else:
marco@0: 			print "Couldn't find the data."
marco@0: 			sys.exit()
marco@0: 		
marco@0: 		#Check if this is a METSDSpaceSIP: see if there is a mets.xml file in the zip
marco@0: 		myzip = zipfile.ZipFile(zipf)
marco@0: 		if "mets.xml" in myzip.namelist():
marco@0: 			packaging = "http://purl.org/net/sword/package/METSDSpaceSIP"
marco@0: 			type = "METS"
marco@0: 		else:
marco@0: 			packaging = "http://purl.org/net/sword/package/SimpleZip"
marco@0: 			type = "SimpleZip"
marco@0: 		
marco@0: 		print "------------------------"
marco@0: 		print "This is a ",type," submission"
marco@3: 		print "Uploading files..."
marco@0: 		myzip.close()
marco@0: 
marco@0: 		payload = open(zipf, "rb")
marco@0: 		try:
marco@0: 			receipt_dep = c.create(col_iri = collection.href,
marco@0: 			payload = payload,
marco@0: 			filename = zipf,
marco@0: 			mimetype = "application/zip",
marco@0: 			packaging = packaging)
marco@0: 			print type, " submission successful."
marco@0: 			if type == "SimpleZip":
marco@0: 				# If some of the additional arguments for author, title, date etc. have been specified, update the metadata
marco@0: 				if (args.title != None) or (args.author != None) or (args.date != None):
marco@0: 					entry = Entry()	
marco@0: 					print "------------------------"
marco@0: 					print "Updating with additional metadata"
marco@0: 					if  args.title != None:
marco@0: 						entry.add_fields(dcterms_title = args.title[0])
marco@0: 					if args.author != None:
marco@0: 						for creator in args.author:
marco@0: 							entry.add_fields(dcterms_creator=creator)
marco@0: 					if args.date != None:
marco@0: 						entry.add_fields(dcterms_created = args.date[0])
marco@0: 					try:
marco@0: 						receipt_update = c.update(dr = receipt_dep , metadata_entry = entry, in_progress = True) # in_progress is True: we don't want to close the submission
marco@0: 						print "Additional metadata updated successfully."
marco@0: 					except:
marco@0: 						print "Server error"
marco@0: 			print "------------------------"
marco@0: 			print "You will find this submission in the \"Submissions\" list in your DSpace account. To complete/edit it with metadata and licenses, click on the title and then on \"Resume\"."			
marco@0: 		except:
marco@0: 			print "Error! Couldn't submit the file!"
marco@0: 			if type == "METS": # Just guessing: not sure this is the problem...
marco@0: 				print "To submit a METS package, the collection MUST have a workflow!"
marco@0: 		payload.close()
marco@3: 		if temp:
marco@3: 			os.remove(zipf)
marco@0: 		
marco@0: 	else: # Failed to connect to SWORDv2 Server
marco@0: 		print "Couldn't connect to the server."
marco@0: 		if attempts == 0:
marco@0: 			print "Invalid credentials entered 3 times."
marco@0: 	
marco@0: except KeyboardInterrupt:
marco@0: 	print "------------------------"
marco@3: 	print "\nSubmission aborted by user."