def test3_changeset(self):
src = ChangeSet()
src.add( ResourceChange('a',timestamp=1) )
src.add( ResourceChange('b',timestamp=2) )
src.add( ResourceChange('c',timestamp=3) )
src.add( ResourceChange('d',timestamp=4))
src.add( ResourceChange('e',timestamp=5) )
self.assertEqual(len(src), 5, "5 things in src")
def generate(self):
"""Generates an inventory of changes"""
changeset = ChangeSet()
for change in self.changes:
changeset.add(change)
changeset.capabilities[self.current_changeset_uri()] = {
"rel": "current http://www.openarchives.org/rs/changeset"}
if self.previous_changeset_uri() is not None:
changeset.capabilities[self.previous_changeset_uri()] = {
"rel": "previous http://www.openarchives.org/rs/changeset"}
return changeset
def generate(self, from_changeid=None):
"""Generates an inventory of changes"""
if from_changeid==None:
from_changeid=self.first_change_id
from_changeid = int(from_changeid)
changeset = ChangeSet()
for change in self.changes_from(from_changeid):
changeset.add(change)
changeset.capabilities[self.next_changeset_uri()] = {
"rel": "next http://www.openarchives.org/rs/changeset"}
changeset.capabilities[self.current_changeset_uri(from_changeid)] = {
"rel": "current http://www.openarchives.org/rs/changeset"}
return changeset
def test1_set_with_repeats(self):
src = ChangeSet()
src.add( ResourceChange('a',timestamp=1) )
src.add( ResourceChange('b',timestamp=1) )
src.add( ResourceChange('c',timestamp=1) )
src.add( ResourceChange('a',timestamp=2) )
src.add( ResourceChange('b',timestamp=2) )
self.assertEqual(len(src), 5, "5 changes in changeset")
def changeset_sitemap(
self, outfile=None, ref_sitemap=None, newref_sitemap=None, empty=None, capabilities=None, dump=None
):
changeset = ChangeSet()
changeset.capabilities = capabilities
if not empty:
# 1. Get and parse reference sitemap
old_inv = self.read_reference_sitemap(ref_sitemap)
# 2. Depending on whether a newref_sitemap was specified, either read that
# or build inventory from files on disk
if newref_sitemap is None:
# Get inventory from disk
new_inv = self.inventory
else:
new_inv = self.read_reference_sitemap(newref_sitemap, name="new reference")
# 3. Calculate changeset
(same, updated, deleted, created) = old_inv.compare(new_inv)
changeset.add_changed_resources(updated, changetype="UPDATED")
changeset.add_changed_resources(deleted, changetype="DELETED")
changeset.add_changed_resources(created, changetype="CREATED")
# 4. Write out changeset
s = Sitemap(pretty_xml=True, allow_multifile=self.allow_multifile, mapper=self.mapper)
if self.max_sitemap_entries is not None:
s.max_sitemap_entries = self.max_sitemap_entries
if outfile is None:
print s.resources_as_xml(changeset, changeset=True)
else:
s.write(changeset, basename=outfile, changeset=True)
self.write_dump_if_requested(changeset, dump)
def test5_add_changed_resources(self):
added = Inventory()
added.add( Resource('a',timestamp=1) )
added.add( Resource('d',timestamp=4))
self.assertEqual(len(added), 2, "2 things in added inventory")
changes = ChangeSet()
changes.add_changed_resources( added, changetype='created' )
self.assertEqual(len(changes), 2, "2 things added")
i = iter(changes)
first = i.next()
self.assertEqual(first.uri, 'a', "changes[0].uri=a")
self.assertEqual(first.timestamp, 1, "changes[0].timestamp=1")
self.assertEqual(first.changetype, 'created', "changes[0].changetype=created")
second = i.next()
self.assertEqual(second.timestamp, 4, "changes[1].timestamp=4")
self.assertEqual(second.changetype, 'created', "changes[1].changetype=created")
# Now add some with updated (one same, one diff)
updated = Inventory()
updated.add( Resource('a',timestamp=5) )
updated.add( Resource('b',timestamp=6))
self.assertEqual(len(updated), 2, "2 things in updated inventory")
changes.add_changed_resources( updated, changetype='updated' )
self.assertEqual(len(changes), 4, "4 = 2 old + 2 things updated")
# Make new inventory from the changes which should not have dupes
dst = Inventory()
dst.add( changes, replace=True )
self.assertEqual(len(dst), 3, "3 unique resources")
self.assertEqual(dst.resources['a'].timestamp, 5 ) # 5 was later in last the 1
self.assertEqual(dst.resources['a'].changetype, 'updated')
self.assertEqual(dst.resources['b'].timestamp, 6)
self.assertEqual(dst.resources['b'].changetype, 'updated')
self.assertEqual(dst.resources['d'].timestamp, 4)
self.assertEqual(dst.resources['d'].changetype, 'created')
def test2_with_repeats_again(self):
r1 = ResourceChange(uri='a',size=1)
r2 = ResourceChange(uri='b',size=2)
i = ChangeSet()
i.add(r1)
i.add(r2)
self.assertEqual( len(i), 2 )
# Can add another ResourceChange with same URI
r1d = ResourceChange(uri='a',size=10)
i.add(r1d)
self.assertEqual( len(i), 3 )
def test4_iter(self):
i = ChangeSet()
i.add( ResourceChange('a',timestamp=1) )
i.add( ResourceChange('b',timestamp=2) )
i.add( ResourceChange('c',timestamp=3) )
i.add( ResourceChange('d',timestamp=4) )
resources=[]
for r in i:
resources.append(r)
self.assertEqual(len(resources), 4)
self.assertEqual( resources[0].uri, 'a')
self.assertEqual( resources[3].uri, 'd')
def changeset_sitemap(self,
outfile=None,
ref_sitemap=None,
newref_sitemap=None,
empty=None,
capabilities=None,
dump=None):
changeset = ChangeSet()
changeset.capabilities = capabilities
if (not empty):
# 1. Get and parse reference sitemap
old_inv = self.read_reference_sitemap(ref_sitemap)
# 2. Depending on whether a newref_sitemap was specified, either read that
# or build inventory from files on disk
if (newref_sitemap is None):
# Get inventory from disk
new_inv = self.inventory
else:
new_inv = self.read_reference_sitemap(newref_sitemap,
name='new reference')
# 3. Calculate changeset
(same, updated, deleted, created) = old_inv.compare(new_inv)
changeset.add_changed_resources(updated, changetype='UPDATED')
changeset.add_changed_resources(deleted, changetype='DELETED')
changeset.add_changed_resources(created, changetype='CREATED')
# 4. Write out changeset
s = Sitemap(pretty_xml=True,
allow_multifile=self.allow_multifile,
mapper=self.mapper)
if (self.max_sitemap_entries is not None):
s.max_sitemap_entries = self.max_sitemap_entries
if (outfile is None):
print s.resources_as_xml(changeset, changeset=True)
else:
s.write(changeset, basename=outfile, changeset=True)
self.write_dump_if_requested(changeset, dump)