def test01_dump_zip_change_list(self):
cl=ChangeDumpManifest()
cl.add( Resource('http://ex.org/a', length=7, path='resync/test/testdata/a', change="updated") )
cl.add( Resource('http://ex.org/b', length=21, path='resync/test/testdata/b', change="updated") )
d=Dump()
zipf=os.path.join(self.tmpdir,"test01_dump.zip")
d.write_zip(cl,zipf) # positional args
self.assertTrue( os.path.exists(zipf) )
self.assertTrue( zipfile.is_zipfile(zipf) )
zo=zipfile.ZipFile(zipf,'r')
self.assertEqual( len(zo.namelist()), 3 )
zo.close()
os.unlink(zipf)
def test01_dump_zip_change_list(self):
cl = ChangeDumpManifest()
cl.add(Resource('http://ex.org/a', length=7,
path='tests/testdata/a', change="updated"))
cl.add(Resource('http://ex.org/b', length=21,
path='tests/testdata/b', change="updated"))
d = Dump()
zipf = os.path.join(self.tmpdir, "test01_dump.zip")
d.write_zip(cl, zipf) # positional args
self.assertTrue(os.path.exists(zipf))
self.assertTrue(zipfile.is_zipfile(zipf))
zo = zipfile.ZipFile(zipf, 'r')
self.assertEqual(len(zo.namelist()), 3)
zo.close()
os.unlink(zipf)
def test_build_ex_23(self):
cdm = ChangeDumpManifest()
cdm.up = "http://example.com/dataset1/capabilitylist.xml"
cdm.md_from = "2013-01-02T00:00:00Z"
cdm.md_until = "2013-01-03T00:00:00Z"
cdm.add( Resource(uri="http://example.com/res7.html",
lastmod="2013-01-02T12:00:00Z",
change="created",
md5="1c1b0e264fa9b7e1e9aa6f9db8d6362b",
length=4339,
mime_type="text/html",
path="/changes/res7.html") )
cdm.add( Resource(uri="http://example.com/res9.pdf",
lastmod="2013-01-02T13:00:00Z",
change="updated",
md5="f906610c3d4aa745cb2b986f25b37c5a",
length=38297,
mime_type="application/pdf",
path="/changes/res9.pdf") )
cdm.add( Resource(uri="http://example.com/res5.tiff",
lastmod="2013-01-02T19:00:00Z",
change="deleted") )
cdm.add( Resource(uri="http://example.com/res7.html",
lastmod="2013-01-02T20:00:00Z",
change="updated",
md5="0988647082c8bc51778894a48ec3b576",
length="5426", #should also take string
mime_type="text/html",
path="/changes/res7-v2.html") )
self._assert_xml_equal_ex( cdm.as_xml(), 'resourcesync_ex_23' )
def get_change_dump_manifest_xml(self, record_id):
"""Get change dump manifest xml.
:param record_id: Identifier of record
:return xml
"""
if not self._is_record_in_index(record_id) or not self._validation():
return None
cdm = ChangeDumpManifest()
cdm.up = '{}resync/{}/changedump.xml'.format(request.url_root,
self.repository_id)
if self.change_dump_manifest:
prev_id, prev_ver_id = record_id.split(".")
current_record = WekoRecord.get_record_by_pid(record_id)
from .utils import get_pid
prev_record_pid = get_pid('{}.{}'.format(prev_id,
str(int(prev_ver_id) -
1)))
if prev_record_pid:
prev_record = WekoRecord.get_record(
id_=prev_record_pid.object_uuid)
else:
prev_record = None
if current_record:
list_file = [file for file in current_record.files]
current_checksum = [
file.info().get('checksum')
for file in current_record.files
]
prev_checksum = []
if prev_record:
list_file.extend([file for file in prev_record.files])
prev_checksum = [
file.info().get('checksum')
for file in prev_record.files
]
for file in list_file:
file_info = file.info()
change = None
if file_info.get('checksum') in prev_checksum:
if file_info.get('checksum') in current_checksum:
change = None
if file_info.get('checksum') not in current_checksum:
change = 'deleted'
else:
if file_info.get('checksum') in current_checksum:
change = 'created'
path = 'recid_{}/{}'.format(current_record.get('recid'),
file_info.get('key'))
lastmod = str(datetime.datetime.utcnow().replace(
tzinfo=datetime.timezone.utc).isoformat())
if change:
re = Resource(
'{}record/{}/files/{}'.format(
request.url_root, current_record.get('recid'),
file_info.get('key')),
lastmod=lastmod,
sha256=file_info.get('checksum').split(':')[1],
length=str(file_info.get('size')),
path=path if change != 'delete' else '',
change=change)
cdm.add(re)
return cdm.as_xml()
