def prepared_eadxml(request, archive, filename):
"""On GET, serves out a prepared version of the EAD file in the specified
archive subversion directory. Response header is set so the user should
be prompted to download the xml, with a filename matching that of
the original document.
On POST, commits the prepared version of the EAD file to the subversion
directory of the specified archive, with a log message indicating the user
who requested the commit.
Steps taken to prepare a document are documented in
:meth:`~findingaids.fa_admin.utils.prep_ead`.
:param filename: name of the file to prep; should be base filename only,
document will be pulled from the configured source directory.
"""
# find relative to svn path if associated with an archive
prepped_xml = cache.get(filename)
arch = get_object_or_404(Archive, slug=archive)
fullpath = os.path.join(arch.svn_local_path, filename)
if prepped_xml is None:
try:
ead = load_xmlobject_from_file(fullpath, FindingAid) # validate or not?
except XMLSyntaxError, e:
# xml is not well-formed : return 500 with error message
return HttpResponseServerError("Could not load document: %s" % e)
# flash meesage that appear on the screen for user, message itself is generated in utils.py
with message_logging(request, 'findingaids.fa_admin.utils', logging.INFO):
try:
ead = utils.prep_ead(ead, filename)
prepped_xml = ead.serializeDocument()
cache.set(filename, prepped_xml)
except Exception as e:
# any exception on prep is most likely ark generation
return HttpResponseServerError('Failed to prep the document: ' + str(e))
def handle(self, *args, **options):
verbosity = int(options['verbosity'])
self._setup_logging(verbosity)
# check for required settings
if not hasattr(settings, 'EXISTDB_ROOT_COLLECTION') or not settings.EXISTDB_ROOT_COLLECTION:
raise CommandError("EXISTDB_ROOT_COLLECTION setting is missing")
return
if verbosity == self.v_all:
print "Preparing documents from all defined Archives"
updated = 0
unchanged = 0
errored = 0
if len(args):
files = args
else:
files = set()
svn = svn_client()
for archive in Archive.objects.all():
# update to make sure we have latest version of everything
svn.update(str(archive.svn_local_path)) # apparently can't handle unicode
files.update(set(glob.iglob(os.path.join(archive.svn_local_path, '*.xml'))))
for file in files:
try:
ead = load_xmlobject_from_file(file, FindingAid)
orig_xml = ead.serializeDocument(pretty=True)
ead = utils.prep_ead(ead, file)
# sanity check before saving
dbpath = settings.EXISTDB_ROOT_COLLECTION + "/" + os.path.basename(file)
errors = utils.check_ead(file, dbpath, xml=ead.serializeDocument())
if errors:
errored += 1
print "Prepared EAD for %s does not pass sanity checks, not saving." % file
if verbosity >= self.v_normal:
print "Errors found:"
for err in errors:
# some errors include a list of error instances - display nicely
if isinstance(err, list):
for suberr in err:
print " %s" % suberr
else:
print " %s" % err
elif orig_xml == ead.serializeDocument(pretty=True):
if verbosity >= self.v_normal:
print "No changes made to %s" % file
unchanged += 1
else:
with open(file, 'w') as f:
ead.serializeDocument(f, pretty=True)
if verbosity >= self.v_normal:
print "Updated %s" % file
updated += 1
except XMLSyntaxError, e:
# xml is not well-formed
print "Error: failed to load %s (document not well-formed XML?)" \
% file
errored += 1
except Exception, e:
# catch any other exceptions
print "Error: failed to prep %s : %s" % (file, e)
errored += 1
def test_prep_ead(self):
# valid fixtures is an ead with series/subseries, and index
# - clear out fixture ark url to trigger generating a new one (simulated)
del(self.valid_ead.eadid.url)
del(self.valid_ead.eadid.identifier)
ead = utils.prep_ead(self.valid_ead, self.valid_eadfile)
self.assert_(isinstance(ead, FindingAid), "prep_ead should return an instance of FindingAid")
self.assertEqual(u'hartsfield558', ead.eadid.value)
self.assertEqual(u'hartsfield558_series1', ead.dsc.c[0].id)
self.assertEqual(u'hartsfield558_subseries6.1', ead.dsc.c[5].c[0].id)
self.assertEqual(u'hartsfield558_index1', ead.archdesc.index[0].id)
# ark should be generated and stored in eadid url
self.assertEqual(MockDjangoPidmanClient.test_ark, ead.eadid.url)
# short-form ark should be stored in identifier attribute
self.assert_(MockDjangoPidmanClient.test_ark.endswith(ead.eadid.identifier))
# ead with no series
eadfile = os.path.join(settings.BASE_DIR, 'fa', 'tests',
'fixtures', 'pittsfreeman1036.xml')
ead = load_xmlobject_from_file(eadfile, FindingAid)
ead = utils.prep_ead(ead, eadfile)
self.assert_(isinstance(ead, FindingAid), "prep_ead should return an instance of FindingAid")
self.assertEqual(u'pittsfreeman1036', ead.eadid.value)
# series with no unitid
eadfile = os.path.join(settings.BASE_DIR, 'fa', 'tests',
'fixtures', 'raoul548.xml')
ead = load_xmlobject_from_file(eadfile, FindingAid)
ead = utils.prep_ead(ead, eadfile)
self.assertEqual(u'raoul548_series3', ead.dsc.c[2].id)
# whitespace cleanup
ead = utils.prep_ead(self.invalid_ead, self.invalid_eadfile)
# - no leading whitespace in list title
# ead.archdesc.origination is getting normalized, so can't be used for testing
origination = ead.node.xpath('//e:origination/e:persname', namespaces={'e': EAD_NAMESPACE})
self.assertEqual(u'Hartsfield, William Berry.', origination[0].text)
# test the node text directly (does not include unitdate)
self.assertEqual(u'William Berry Hartsfield papers, ', ead.unittitle.node.text)
self.assertEqual(u'Gone with the wind (Motion picture)',
ead.archdesc.controlaccess.controlaccess[0].title[0].value)
self.assertEqual(u'Allen, Ivan.',
ead.archdesc.controlaccess.controlaccess[1].person_name[0].value)
self.assertEqual(u'Mines and mineral resources--Georgia.',
ead.archdesc.controlaccess.controlaccess[3].subject[1].value)
# unicode characters
self.assertEqual(u'Motion pictures--Georgia. \u2026',
ead.archdesc.controlaccess.controlaccess[3].subject[2].value)
self.assertEqual(u'Motion pictures.',
ead.archdesc.controlaccess.controlaccess[-1].genre_form[0].value)
# remaining errors after clean-up:
# 1 - duplicate origination
# 2 - > 2 containers in a did (summary error and list of problem dids)
# 2 - 1 container in a did (summary error and list of problem dids)
# = 5
self.assertEqual(5, len(utils.check_eadxml(ead)),
"only 3 errors (duplicate origination, 3 containers in a did, 1 container in a did) should be left in invalid test fixture after cleaning")
# special case - unittitle begins with a <title>
eadfile = os.path.join(settings.BASE_DIR, 'fa', 'tests',
'fixtures', 'pittsfreeman1036.xml')
ead = load_xmlobject_from_file(eadfile, FindingAid)
ead = utils.prep_ead(ead, eadfile)
self.assertFalse(unicode(ead.list_title).startswith('None'),
'cleaned unittitle with leading <title> should not start with "None"')
:param filename: name of the file to prep; should be base filename only,
document will be pulled from the configured source directory.
"""
# find relative to svn path if associated with an archive
arch = get_object_or_404(Archive, slug=archive)
fullpath = os.path.join(arch.svn_local_path, filename)
try:
ead = load_xmlobject_from_file(fullpath, FindingAid) # validate or not?
except XMLSyntaxError, e:
# xml is not well-formed : return 500 with error message
return HttpResponseServerError("Could not load document: %s" % e)
with message_logging(request, 'findingaids.fa_admin.utils', logging.INFO):
try:
ead = utils.prep_ead(ead, filename)
except Exception as e:
# any exception on prep is most likely ark generation
return HttpResponseServerError('Failed to prep the document: ' + str(e))
# on GET, display the xml and make available for download
if request.method == 'GET':
prepped_xml = ead.serializeDocument()
response = HttpResponse(prepped_xml, mimetype='application/xml')
response['Content-Disposition'] = "attachment; filename=%s" % filename
return response
# on POST, save to file and commit to subversion
if request.method == 'POST':
file_path = os.path.join(arch.svn_local_path, filename)
with open(file_path, 'w') as xmlfile:
