def test01_as_xml(self):
rd = ChangeDump()
rd.add( Resource('a.zip',timestamp=1) )
rd.add( Resource('b.zip',timestamp=2) )
xml = rd.as_xml()
#print(xml)
self.assertTrue( re.search(r'<rs:md .*capability="changedump"', xml), 'XML has capability' )
self.assertTrue( re.search(r'<url><loc>a.zip</loc><lastmod>1970-01-01T00:00:01Z</lastmod></url>', xml), 'XML has resource a' )
def test01_as_xml(self):
rd = ChangeDump()
rd.add( Resource('a.zip',timestamp=1) )
rd.add( Resource('b.zip',timestamp=2) )
xml = rd.as_xml()
print xml
self.assertTrue( re.search(r'<rs:md .*capability="changedump"', xml), 'XML has capability' )
self.assertTrue( re.search(r'<url><loc>a.zip</loc><lastmod>1970-01-01T00:00:01Z</lastmod></url>', xml), 'XML has resource a' )
def test01_as_xml(self):
rd = ChangeDump()
rd.add(Resource("a.zip", timestamp=1))
rd.add(Resource("b.zip", timestamp=2))
xml = rd.as_xml()
# print(xml)
self.assertTrue(re.search(r'<rs:md .*capability="changedump"', xml), "XML has capability")
self.assertTrue(
re.search(r"<url><loc>a.zip</loc><lastmod>1970-01-01T00:00:01Z</lastmod></url>", xml), "XML has resource a"
)
def test10_parse(self):
xml='<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n\
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:rs="http://www.openarchives.org/rs/terms/">\
<rs:md capability="changedump" from="2013-01-01"/>\
<url><loc>http://example.com/a.zip</loc><lastmod>2012-03-14T18:37:36Z</lastmod><rs:md length="12345" /></url>\
<url><loc>http://example.com/b.zip</loc><lastmod>2012-03-14T18:37:36Z</lastmod><rs:md length="56789" /></url>\
</urlset>'
rd=ChangeDump()
rd.parse(fh=io.StringIO(xml))
self.assertEqual( len(rd.resources), 2, 'got 2 resource dumps')
self.assertEqual( rd.md['capability'], 'changedump', 'capability set' )
self.assertEqual( rd.md_from, '2013-01-01' )
self.assertTrue( 'http://example.com/a.zip' in rd.resources )
self.assertTrue( rd.resources['http://example.com/a.zip'].length, 12345 )
self.assertTrue( 'http://example.com/b.zip' in rd.resources )
self.assertTrue( rd.resources['http://example.com/b.zip'].length, 56789 )
def test10_parse(self):
xml = '<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n\
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:rs="http://www.openarchives.org/rs/terms/">\
<rs:md capability="changedump" from="2013-01-01"/>\
<url><loc>http://example.com/a.zip</loc><lastmod>2012-03-14T18:37:36Z</lastmod><rs:md length="12345" /></url>\
<url><loc>http://example.com/b.zip</loc><lastmod>2012-03-14T18:37:36Z</lastmod><rs:md length="56789" /></url>\
</urlset>'
rd = ChangeDump()
rd.parse(fh=io.StringIO(xml))
self.assertEqual(len(rd.resources), 2, 'got 2 resource dumps')
self.assertEqual(rd.md['capability'], 'changedump', 'capability set')
self.assertEqual(rd.md_from, '2013-01-01')
self.assertTrue('http://example.com/a.zip' in rd.resources)
self.assertTrue(rd.resources['http://example.com/a.zip'].length, 12345)
self.assertTrue('http://example.com/b.zip' in rd.resources)
self.assertTrue(rd.resources['http://example.com/b.zip'].length, 56789)
def test_build_ex_22(self):
"""Change Dump with three dump files"""
cd = ChangeDump()
cd.up = 'http://example.com/dataset1/capabilitylist.xml'
cd.md_from="2013-01-01T00:00:00Z"
z1 = Resource( uri='http://example.com/20130101-changedump.zip',
lastmod='2013-01-01T23:59:59Z',
length=3109,
md_from="2013-01-01T00:00:00Z",
md_until="2013-01-02T00:00:00Z",
mime_type="application/zip" )
z1.contents='http://example.com/20130101-changedump-manifest.xml'
z2 = Resource( uri='http://example.com/20130102-changedump.zip',
lastmod='2013-01-02T23:59:59Z',
length=6629,
md_from="2013-01-02T00:00:00Z",
md_until="2013-01-03T00:00:00Z",
mime_type="application/zip" )
z2.contents='http://example.com/20130102-changedump-manifest.xml'
z3 = Resource( uri='http://example.com/20130103-changedump.zip',
lastmod='2013-01-03T23:59:59Z',
length=8124,
md_from="2013-01-03T00:00:00Z",
md_until="2013-01-04T00:00:00Z",
mime_type="application/zip" )
z3.contents='http://example.com/20130103-changedump-manifest.xml'
cd.add( [z1, z2, z3] )
ex_xml = self._open_ex('resourcesync_ex_22').read()
self._assert_xml_equal( cd.as_xml(), ex_xml )
def test12_parse_bad_capability(self):
# the <rs:md capability="bad_capability".. should give error
xml='<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n\
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:rs="http://www.openarchives.org/rs/terms/">\
<rs:md capability="bad_capability" from="2013-01-01"/>\
<url><loc>http://example.com/bad_res_1</loc><lastmod>2012-03-14T18:37:36Z</lastmod></url>\
</urlset>'
rd=ChangeDump()
self.assertRaises( SitemapParseError, rd.parse, fh=io.StringIO(xml) )
def test11_parse_no_capability(self):
# For a resource dump this should be an error
xml='<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n\
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:rs="http://www.openarchives.org/rs/terms/">\
<rs:md at="2013-01-01"/>\
<url><loc>http://example.com/a.zip</loc><lastmod>2012-03-14T18:37:36Z</lastmod><rs:md length="12" /></url>\
</urlset>'
rd=ChangeDump()
self.assertRaises( SitemapParseError, rd.parse, fh=io.StringIO(xml) )
def test_build_ex_13(self):
"""Capability List document with 4 entries"""
cl = CapabilityList()
cl.describedby = 'http://example.com/info_about_set1_of_resources.xml'
cl.up = 'http://example.com/resourcesync_description.xml'
cl.add_capability( capability=ResourceList( uri='http://example.com/dataset1/resourcelist.xml' ) )
cl.add_capability( capability=ResourceDump( uri='http://example.com/dataset1/resourcedump.xml' ) )
cl.add_capability( capability=ChangeList( uri='http://example.com/dataset1/changelist.xml' ) )
cl.add_capability( capability=ChangeDump( uri='http://example.com/dataset1/changedump.xml' ) )
ex_xml = self._open_ex('resourcesync_ex_13').read()
self._assert_xml_equal( cl.as_xml(), ex_xml )
def get_change_dump_xml(self, from_date):
"""
Get change dump xml.
:return: Updated Change List info
"""
if not self._validation():
return None
change_dump = ChangeDump()
change_dump.up = '{}resync/capability.xml'.format(request.url_root)
change_dump.index = '{}resync/{}/changedump.xml'.format(
request.url_root, self.repository_id)
record_changes = self._get_record_changes_with_interval(from_date)
for data in record_changes:
try:
next_ch = self._next_change(data, record_changes)
if data.get('status') == 'deleted':
continue
loc = '{}resync/{}/{}/change_dump_content.zip'.format(
request.url_root, self.repository_id,
'{}.{}'.format(data.get('record_id'),
data.get('record_version')))
rc = Resource(loc,
lastmod=data.get("updated"),
mime_type='application/zip',
md_from=data.get('updated'),
md_until=datetime.datetime.utcnow().replace(
tzinfo=datetime.timezone.utc).isoformat(),
ln=[])
if next_ch and next_ch.get('updated'):
rc.md_until = next_ch.get('updated')
if self.change_dump_manifest:
ln = {
'rel':
'contents',
'href':
'{}resync/{}/{}/changedump_manifest.xml'.format(
request.url_root, self.repository_id,
'{}.{}'.format(data.get('record_id'),
data.get('record_version'))),
'type':
'application/xml'
}
rc.ln.append(ln)
change_dump.add(rc)
except Exception:
current_app.logger.error('-' * 60)
traceback.print_exc(file=sys.stdout)
current_app.logger.error('-' * 60)
continue
return change_dump.as_xml()
