def test_process_if_needed_missing_writes(self, xml_parser):
"""If output isn't already present, we should process. If it is
present, we don't need to, unless a dependency has changed."""
with self.cli.isolated_filesystem():
build_tree = xml_parser.reg_text.build_tree
build_tree.return_value = Node()
last_versions = [annual_editions.LastVersionInYear('1111', 2000)]
entry.Version('12', '1000', '1111').write(
Version('1111', date(2000, 1, 1), date(2000, 1, 1)))
entry.Entry('annual', '12', '1000', 2000).write(
b'<ROOT></ROOT>')
annual_editions.process_if_needed('12', '1000', last_versions)
self.assertTrue(build_tree.called)
build_tree.reset_mock()
entry.Entry('tree', '12', '1000', '1111').write(b'tree-here')
annual_editions.process_if_needed('12', '1000', last_versions)
self.assertFalse(build_tree.called)
# Simulate a change to an input file
label_id = str(entry.Annual(12, 1000, 2000))
new_time = timezone.now() + timedelta(hours=1)
DBEntry.objects.filter(label_id=label_id).update(modified=new_time)
annual_editions.process_if_needed('12', '1000', last_versions)
self.assertTrue(build_tree.called)
def test_process_if_needed_missing_writes(monkeypatch):
"""If output isn't already present, we should process. If it is present,
we don't need to, unless a dependency has changed."""
monkeypatch.setattr(annual_editions, 'gpo_cfr', Mock())
build_tree = annual_editions.gpo_cfr.builder.build_tree
build_tree.return_value = Node()
last_versions = [annual_editions.LastVersionInYear('1111', 2000)]
entry.Version('12', '1000', '1111').write(
Version('1111', date(2000, 1, 1), Citation(1, 1)))
entry.Entry('annual', '12', '1000', 2000).write(b'<ROOT></ROOT>')
annual_editions.process_if_needed('12', '1000', last_versions)
assert build_tree.called
build_tree.reset_mock()
entry.Entry('tree', '12', '1000', '1111').write(b'tree-here')
annual_editions.process_if_needed('12', '1000', last_versions)
assert not build_tree.called
# Simulate a change to an input file
label_id = str(entry.Annual(12, 1000, 2000))
new_time = timezone.now() + timedelta(hours=1)
DBEntry.objects.filter(label_id=label_id).update(modified=new_time)
annual_editions.process_if_needed('12', '1000', last_versions)
assert build_tree.called
def test_stale_layers(self):
"""We should have dependencies between all of the layers and their
associated trees. We should also tie the meta layer to the version"""
configured_layers = {'cfr': {'keyterms': None, 'other': None}}
with self.cli.isolated_filesystem(), patch.dict(
layers.LAYER_CLASSES, configured_layers):
version_entry = entry.Version(111, 22, 'aaa')
version_entry.write(Version('aaa', date.today(), date.today()))
tree_entry = entry.Tree(111, 22, 'aaa')
# Use list() to instantiate
self.assertRaises(dependency.Missing, list,
layers.stale_layers(tree_entry, 'cfr'))
entry.Entry('tree', 111, 22, 'bbb').write(b'') # wrong version
self.assertRaises(dependency.Missing, list,
layers.stale_layers(tree_entry, 'cfr'))
entry.Entry('tree', 111, 22, 'aaa').write(b'')
six.assertCountEqual(self, layers.stale_layers(tree_entry, 'cfr'),
['keyterms', 'other'])
self.assertIn(
str(version_entry),
dependency.Graph().dependencies(
str(entry.Layer.cfr(111, 22, 'aaa', 'meta'))))
def test_write_if_needed_no_need_to_recompute(monkeypatch):
"""If all dependencies are up to date and the output is present, there's
no need to write anything"""
monkeypatch.setattr(versions, 'write_to_disk', Mock())
entry.Entry('notice_xml', '111').write(b'content')
entry.Entry('version', 'title', 'part', '111').write(b'out')
versions.write_if_needed('title', 'part', ['111'], {'111': 'xml111'}, {})
assert not versions.write_to_disk.called
def test_deletes_index(tmpdir_setup):
entry.Entry('aaa', 'bbb').write(b'ccc')
entry.Entry('bbb', 'ccc').write(b'ddd')
assert 1 == len(list(entry.Entry("aaa").sub_entries()))
assert 1 == len(list(entry.Entry("bbb").sub_entries()))
CliRunner().invoke(clear)
assert [] == list(entry.Entry().sub_entries())
def test_write_if_needed_no_need_to_recompute(self, write_to_disk):
"""If all dependencies are up to date and the output is present,
there's no need to write anything"""
with self.cli.isolated_filesystem():
entry.Entry('notice_xml', '111').write(b'content')
entry.Entry('version', 'title', 'part', '111').write(b'out')
versions.write_if_needed('title', 'part', ['111'],
{'111': 'xml111'}, {})
self.assertFalse(write_to_disk.called)
def test_process_no_need_to_create(self):
"""If everything is up to date, we don't need to build new versions"""
with CliRunner().isolated_filesystem():
annual = entry.Entry('annual', self.title, self.part, self.year)
tree = entry.Entry('tree', self.title, self.part, self.version_id)
annual.write(b'ANNUAL')
tree.write(b'TREE')
current_version.process_if_needed(self.volume, self.part)
# didn't change
self.assertEqual(annual.read(), b'ANNUAL')
self.assertEqual(tree.read(), b'TREE')
def test_process_no_need_to_create():
"""If everything is up to date, we don't need to build new versions"""
title, part, year = randint(1, 999), randint(1, 999), randint(2000, 2020)
annual = entry.Entry('annual', title, part, year)
tree = entry.Entry('tree', title, part,
'{0}-annual-{1}'.format(year, part))
annual.write(b'ANNUAL')
tree.write(b'TREE')
annual_version.process_if_needed(Volume(year, title, 1), part)
# didn't change
assert annual.read() == b'ANNUAL'
assert tree.read() == b'TREE'
def test_deletes_can_be_focused(tmpdir_setup):
"""If params are provided to delete certain directories, only those
directories should get removed"""
to_delete = ['delroot/aaa/bbb', 'delroot/aaa/ccc',
'root/delsub/aaa', 'root/delsub/bbb']
to_keep = ['root/othersub/aaa', 'root/aaa',
'top-level-file', 'other-root/aaa']
for path in to_delete + to_keep:
entry.Entry(*path.split('/')).write(b'')
CliRunner().invoke(clear, ['delroot', 'root/delsub'])
assert {os.sep.join(c.path)
for c in entry.Entry().sub_entries()} == set(to_keep)
def test_process_creation(self):
"""If no tree is present, we should build one"""
to_patch = 'regparser.commands.current_version.xml_parser'
with CliRunner().isolated_filesystem(), patch(to_patch) as xml_parser:
entry.Entry('annual', self.title, self.part,
self.year).write('<ROOT />')
xml_parser.reg_text.build_tree.return_value = {'my': 'tree'}
current_version.process_if_needed(self.volume, self.part)
tree = entry.Entry('tree', self.title, self.part,
self.version_id).read()
self.assertEqual(json.loads(tree), {'my': 'tree'})
notice = entry.SxS(self.version_id).read()
self.assertEqual(notice['document_number'], self.version_id)
self.assertEqual(notice['cfr_parts'], [str(self.part)])
def test_process_creation(monkeypatch):
"""If no tree is present, we should build one"""
title, part, year = randint(1, 999), randint(1, 999), randint(2000, 2020)
version_id = '{0}-annual-{1}'.format(year, part)
monkeypatch.setattr(annual_version, 'builder', Mock())
entry.Entry('annual', title, part, year).write(b'<ROOT />')
annual_version.builder.build_tree.return_value = {'my': 'tree'}
annual_version.process_if_needed(Volume(year, title, 1), part)
tree = entry.Entry('tree', title, part, version_id).read()
assert json.loads(tree.decode('utf-8')) == {'my': 'tree'}
notice = entry.Notice(version_id).read()
assert notice.version_id == version_id
assert notice.cfr_refs == [TitlePartsRef(title, [part])]
def test_process(monkeypatch):
"""Verify that the correct changes are found"""
compile_regulation = Mock(return_value=Node())
monkeypatch.setattr(fill_with_rules, 'compile_regulation',
compile_regulation)
notice_mock = Mock()
# entry.Notice('new').read().amendments
notice_mock.return_value.read.return_value.amendments = [
{"instruction": "Something something",
"cfr_part": "1000",
"authority": "USC Numbers"},
{"instruction": "More things",
"cfr_part": "1000",
"changes": [["1000-2-b", ["2b changes"]],
["1000-2-c", ["2c changes"]]]},
{"instruction": "Yet more changes",
"cfr_part": "1000",
"changes": [["1000-4-a", ["4a changes"]]]}
]
monkeypatch.setattr(fill_with_rules.entry, 'Notice', notice_mock)
tree_dir = entry.Tree('12', '1000')
(tree_dir / 'old').write(Node())
entry.Entry('notice_xml', 'new').write(b'')
fill_with_rules.process(tree_dir, 'old', 'new')
changes = dict(compile_regulation.call_args[0][1])
assert changes == {"1000-2-b": ["2b changes"],
"1000-2-c": ["2c changes"],
"1000-4-a": ["4a changes"]}
def test_write_if_needed_output_missing(self, write_to_disk):
"""If the output file is missing, we'll always write"""
with self.cli.isolated_filesystem():
entry.Entry('notice_xml', '111').write(b'content')
versions.write_if_needed('title', 'part', ['111'],
{'111': 'xml111'}, {})
self.assertTrue(write_to_disk.called)
def test_process_creation(self):
"""If no tree is present, we should build one"""
to_patch = 'regparser.commands.current_version.xml_parser'
with CliRunner().isolated_filesystem(), patch(to_patch) as xml_parser:
entry.Entry('annual', self.title, self.part,
self.year).write(b'<ROOT />')
xml_parser.reg_text.build_tree.return_value = {'my': 'tree'}
current_version.process_if_needed(self.volume, self.part)
tree = entry.Entry('tree', self.title, self.part,
self.version_id).read()
self.assertEqual(json.loads(tree.decode('utf-8')), {'my': 'tree'})
notice = entry.Notice(self.version_id).read()
self.assertEqual(notice.version_id, self.version_id)
self.assertEqual(notice.cfr_refs,
[TitlePartsRef(self.title, [self.part])])
def test_write_if_needed_delays(self, write_to_disk):
"""Delays introduce dependencies."""
with self.cli.isolated_filesystem():
entry.Entry('notice_xml', '111').write(b'content')
entry.Entry('notice_xml', '222').write(b'content')
entry.Entry('version', 'title', 'part', '111').write(b'out')
versions.write_if_needed(
'title', 'part', ['111'], {'111': 'xml111'},
{'111': versions.Delay('222', 'until-date')})
self.assertFalse(write_to_disk.called)
# Simulate a change to an input file
os.utime(str(entry.Notice('222')), (time() + 1000, time() + 1000))
versions.write_if_needed(
'title', 'part', ['111'], {'111': 'xml111'},
{'111': versions.Delay('222', 'until-date')})
self.assertTrue(write_to_disk.called)
def test_write_if_needed_delays(monkeypatch):
"""Delays introduce dependencies."""
monkeypatch.setattr(versions, 'write_to_disk', Mock())
entry.Entry('notice_xml', '111').write(b'content')
entry.Entry('notice_xml', '222').write(b'content')
entry.Entry('version', 'title', 'part', '111').write(b'out')
versions.write_if_needed('title', 'part', ['111'], {'111': 'xml111'},
{'111': versions.Delay('222', 'until-date')})
assert not versions.write_to_disk.called
# Simulate a change to an input file
label_id = str(entry.Notice('222'))
new_time = timezone.now() + timedelta(hours=1)
DBEntry.objects.filter(label_id=label_id).update(modified=new_time)
versions.write_if_needed('title', 'part', ['111'], {'111': 'xml111'},
{'111': versions.Delay('222', 'until-date')})
assert versions.write_to_disk.called
def test_is_stale():
"""We should raise dependency exceptions when necessary files haven't been
processed. We need SxS entries _and_ the relevant tree"""
with pytest.raises(dependency.Missing):
sxs_layers.is_stale(11, 222, 'aaa')
create_versions()
entry.Entry('sxs', 'aaa').write(b'')
entry.Entry('sxs', 'bbb').write(b'')
with pytest.raises(dependency.Missing):
sxs_layers.is_stale(11, 222, 'aaa')
entry.Entry('tree', 11, 222, 'bbb').write(b'') # Wrong tree
with pytest.raises(dependency.Missing):
sxs_layers.is_stale(11, 222, 'aaa')
entry.Entry('tree', 11, 222, 'aaa').write(b'')
assert sxs_layers.is_stale(11, 222, 'aaa')
def test_fetch_version_ids_no_local(self, fetch_notice_json):
"""If there are no local copies, the document numbers found in the FR
notices should be passed through"""
fetch_notice_json.return_value = [{'document_number': '1'},
{'document_number': '22'}]
with self.cli.isolated_filesystem():
path = entry.Entry("path")
self.assertEqual(['1', '22'],
versions.fetch_version_ids('title', 'part', path))
def test_write_if_needed_delays(self, write_to_disk):
"""Delays introduce dependencies."""
with self.cli.isolated_filesystem():
entry.Entry('notice_xml', '111').write(b'content')
entry.Entry('notice_xml', '222').write(b'content')
entry.Entry('version', 'title', 'part', '111').write(b'out')
versions.write_if_needed(
'title', 'part', ['111'], {'111': 'xml111'},
{'111': versions.Delay('222', 'until-date')})
self.assertFalse(write_to_disk.called)
# Simulate a change to an input file
label_id = str(entry.Notice('222'))
new_time = timezone.now() + timedelta(hours=1)
DBEntry.objects.filter(label_id=label_id).update(modified=new_time)
versions.write_if_needed(
'title', 'part', ['111'], {'111': 'xml111'},
{'111': versions.Delay('222', 'until-date')})
self.assertTrue(write_to_disk.called)
def test_deletes_can_be_focused(self):
"""If params are provided to delete certain directories, only those
directories should get removed"""
with self.cli.isolated_filesystem():
to_delete = ['delroot/aaa/bbb', 'delroot/aaa/ccc',
'root/delsub/aaa', 'root/delsub/bbb']
to_keep = ['root/othersub/aaa', 'root/aaa',
'top-level-file', 'other-root/aaa']
for path in to_delete + to_keep:
entry.Entry(*path.split('/')).write('')
self.cli.invoke(clear, ['delroot', 'root/delsub'])
self.assertItemsEqual(['top-level-file', 'root', 'other-root'],
list(entry.Entry()))
self.assertItemsEqual(['othersub', 'aaa'],
list(entry.Entry('root')))
self.assertItemsEqual(['aaa'],
list(entry.Entry('other-root')))
def test_is_stale(self):
"""We should raise dependency exceptions when necessary files haven't
been processed. We need SxS entries _and_ the relevant tree"""
with CliRunner().isolated_filesystem():
self.assertRaises(dependency.Missing, sxs_layers.is_stale,
11, 222, 'aaa')
self.create_versions()
entry.Entry('sxs', 'aaa').write(b'')
entry.Entry('sxs', 'bbb').write(b'')
self.assertRaises(dependency.Missing, sxs_layers.is_stale,
11, 222, 'aaa')
entry.Entry('tree', 11, 222, 'bbb').write(b'') # Wrong tree
self.assertRaises(dependency.Missing, sxs_layers.is_stale,
11, 222, 'aaa')
entry.Entry('tree', 11, 222, 'aaa').write(b'')
self.assertTrue(sxs_layers.is_stale(11, 222, 'aaa'))
def test_fetch_version_ids_local(self, fetch_notice_json):
"""If a notice is split into multiple entries locally, a single
document number might result in multiple version ids"""
fetch_notice_json.return_value = [{'document_number': '1'},
{'document_number': '22'}]
with self.cli.isolated_filesystem():
path = entry.Entry("path")
(path / '1_20010101').write('v1')
(path / '1_20020202').write('v2')
(path / '22').write('second')
(path / '22-3344').write('unrelated file')
self.assertEqual(['1_20010101', '1_20020202', '22'],
versions.fetch_version_ids('title', 'part', path))
def test_fetch_version_ids_no_local(monkeypatch):
"""If there are no local copies, the document numbers found in the FR
notices should be passed through"""
monkeypatch.setattr(
versions, 'fetch_notice_json',
Mock(return_value=[{
'document_number': '1',
'full_text_xml_url': 'somewhere'
}, {
'document_number': '22',
'full_text_xml_url': 'somewhere'
}]))
path = entry.Entry("path")
assert ['1', '22'] == versions.fetch_version_ids('title', 'part', path)
def test_process_if_needed_missing_writes(self, xml_parser):
"""If output isn't already present, we should process. If it is
present, we don't need to, unless a dependency has changed."""
with self.cli.isolated_filesystem():
build_tree = xml_parser.reg_text.build_tree
build_tree.return_value = Node()
last_versions = [annual_editions.LastVersionInYear('1111', 2000)]
entry.Version('12', '1000', '1111').write(
Version('1111', date(2000, 1, 1), date(2000, 1, 1)))
entry.Entry('annual', '12', '1000', 2000).write(b'<ROOT></ROOT>')
annual_editions.process_if_needed('12', '1000', last_versions)
self.assertTrue(build_tree.called)
build_tree.reset_mock()
entry.Entry('tree', '12', '1000', '1111').write(b'tree-here')
annual_editions.process_if_needed('12', '1000', last_versions)
self.assertFalse(build_tree.called)
# Simulate a change to an input file
os.utime(str(entry.Annual('12', '1000', '2000')),
(time() + 1000, time() + 1000))
annual_editions.process_if_needed('12', '1000', last_versions)
self.assertTrue(build_tree.called)
def test_stale_layers(monkeypatch):
"""We should have dependencies between all of the layers and their
associated trees. We should also tie the meta layer to the version"""
monkeypatch.setattr(layers, 'LAYER_CLASSES',
{'cfr': {
'keyterms': None,
'other': None
}})
version_entry = entry.Version(111, 22, 'aaa')
version_entry.write(Version('aaa', date.today(), Citation(1, 1)))
tree_entry = entry.Tree(111, 22, 'aaa')
with pytest.raises(dependency.Missing):
layers.stale_layers(tree_entry, 'cfr')
entry.Entry('tree', 111, 22, 'bbb').write(b'') # wrong version
with pytest.raises(dependency.Missing):
layers.stale_layers(tree_entry, 'cfr')
entry.Entry('tree', 111, 22, 'aaa').write(b'')
assert set(layers.stale_layers(tree_entry, 'cfr')) == {'keyterms', 'other'}
assert str(version_entry) in dependency.Graph().dependencies(
str(entry.Layer.cfr(111, 22, 'aaa', 'meta')))
def test_deletes_index(self):
with self.cli.isolated_filesystem():
entry.Entry('aaa', 'bbb').write('ccc')
entry.Entry('bbb', 'ccc').write('ddd')
self.assertEqual(1, len(entry.Entry("aaa")))
self.assertEqual(1, len(entry.Entry("bbb")))
self.cli.invoke(clear)
self.assertEqual(0, len(entry.Entry("aaa")))
self.assertEqual(0, len(entry.Entry("bbb")))
def test_fetch_version_ids_skip_no_xml(monkeypatch):
"""We'll skip over all of the versions which don't have XML"""
monkeypatch.setattr(
versions, 'fetch_notice_json',
Mock(return_value=[{
'document_number': '1',
'full_text_xml_url': 'something'
}, {
'document_number': '2',
'full_text_xml_url': None
}, {
'document_number': '3',
'full_text_xml_url': 'somewhere'
}]))
path = entry.Entry("path")
assert ['1', '3'] == versions.fetch_version_ids('title', 'part', path)
def test_rebuild(self):
"""Validate that the `rebuild()` method calculates the correct
"stale" references"""
with CliRunner().isolated_filesystem():
graph = dependency.Graph()
path = entry.Entry('path')
a, b, c, d = [path / char for char in 'abcd']
# (A, B) -> C -> D
graph.add(c, a)
graph.add(c, b)
graph.add(d, c)
# None of the files exist yet; A & B have no dependencies, so they
# are stale due to themselves. C & D are stale due either A or B
self.assert_rebuilt_state(graph,
path,
a='a',
b='b',
c='ab',
d='ab')
b.write(b'bbb')
# B exists now, so dependency errors are only due to A now
self.assert_rebuilt_state(graph, path, a='a', b='', c='a', d='a')
a.write(b'aaa')
# A exists now, too, so C is the bottleneck
self.assert_rebuilt_state(graph, path, a='', b='', c='c', d='c')
c.write(b'ccc')
# Now there's only the final, self-reference
self.assert_rebuilt_state(graph, path, a='', b='', c='', d='d')
d.write(b'ddd')
# Now no one is stale
self.assert_rebuilt_state(graph, path, a='', b='', c='', d='')
self._touch(a, 1000)
# A's been updated. Need to run everything after it
self.assert_rebuilt_state(graph, path, a='', b='', c='a', d='a')
self._touch(d, 2000)
self._touch(c, 3000)
# C and D have been updated, but C's been updated after D
self.assert_rebuilt_state(graph, path, a='', b='', c='', d='c')
def test_writes(self, meta_data, build_notice):
"""If the notice XML is present, we write the parsed version to disk,
even if that version's already present"""
with self.cli.isolated_filesystem():
entry.Notice('1111').write(self.notice_xml)
self.cli.invoke(fetch_sxs, ['1111'])
meta_data.return_value = {'example': 1}
self.assertTrue(build_notice.called)
args, kwargs = build_notice.call_args
self.assertTrue(args[2], {'example': 1})
self.assertTrue(
isinstance(kwargs['xml_to_process'], etree._Element))
build_notice.reset_mock()
entry.Entry('rule_changes', '1111').write('content')
self.cli.invoke(fetch_sxs, ['1111'])
self.assertTrue(build_notice.called)
def test_fetch_version_ids_local(monkeypatch):
"""If a notice is split into multiple entries locally, a single document
number might result in multiple version ids"""
monkeypatch.setattr(
versions, 'fetch_notice_json',
Mock(return_value=[{
'document_number': '1',
'full_text_xml_url': 'somewhere'
}, {
'document_number': '22',
'full_text_xml_url': 'somewhere'
}]))
path = entry.Entry("path")
(path / '1_20010101').write(b'v1')
(path / '1_20020202').write(b'v2')
(path / '22').write(b'second')
(path / '22-3344').write(b'unrelated file')
assert versions.fetch_version_ids(
'title', 'part', path) == ['1_20010101', '1_20020202', '22']