def test_revision_generator_notices(self, init, merge_changes):
init.return_value = None
b = Builder() # Don't need parameters as init's been mocked out
aaaa = {'document_number': 'aaaa', 'effective_on': '2012-12-12',
'publication_date': '2011-11-11', 'changes': []}
bbbb = {'document_number': 'bbbb', 'effective_on': '2012-12-12',
'publication_date': '2011-11-12', 'changes': []}
cccc = {'document_number': 'cccc', 'effective_on': '2013-01-01',
'publication_date': '2012-01-01', 'changes': []}
b.notices = [aaaa, bbbb, cccc]
b.eff_notices = {'2012-12-12': [aaaa, bbbb], '2013-01-01': [cccc]}
b.doc_number = 'aaaa'
tree = Node(label=['1111'])
version_list = []
notice_lists = []
for notice, _, _, notices in b.revision_generator(tree):
version_list.append(notice['document_number'])
notice_lists.append(notices)
self.assertEqual(['bbbb', 'cccc'], version_list)
self.assertEqual(2, len(notice_lists))
self.assertEqual(2, len(notice_lists[0]))
self.assertTrue(aaaa in notice_lists[0])
self.assertTrue(bbbb in notice_lists[0])
self.assertEqual(3, len(notice_lists[1]))
self.assertTrue(aaaa in notice_lists[1])
self.assertTrue(bbbb in notice_lists[1])
self.assertTrue(cccc in notice_lists[1])
def parse_regulation(args):
""" Run the parser on the specified command-line arguments. Broken out into
separate function to assist in profiling.
"""
with codecs.open(args.filename, 'r', 'utf-8') as f:
reg = f.read()
doc_number = args.notice
act_title_and_section = [args.act_title, args.act_section]
# First, the regulation tree
reg_tree = Builder.reg_tree(reg)
builder = Builder(cfr_title=args.title,
cfr_part=reg_tree.label_id(),
doc_number=doc_number)
builder.write_notices()
# Always do at least the first reg
logger.info("Version %s", doc_number)
builder.write_regulation(reg_tree)
layer_cache = LayerCacheAggregator()
builder.gen_and_write_layers(reg_tree, act_title_and_section, layer_cache)
layer_cache.replace_using(reg_tree)
if args.generate_diffs:
generate_diffs(doc_number, reg_tree, act_title_and_section, builder,
layer_cache)
def test_layer_cache(self, init):
"""Integration test for layer caching"""
init.return_value = None
cache = LayerCacheAggregator()
b = Builder() # Don't need parameters as init's been mocked out
b.cfr_title, b.cfr_part, b.doc_number = 15, '111', '111-222'
b.writer = Mock()
b.checkpointer = NullCheckpointer()
write = b.writer.layer.return_value.write
tree = Node(label=["1234"],
children=[
Node(label=["1234", "1"],
children=[
Node("See paragraph (b)",
label=["1234", "1", "a"]),
Node("This is b", label=["1234", "1", "b"])
])
])
b.gen_and_write_layers(tree, [], cache, [])
arg = write.call_args_list[3][0][0]
self.assertEqual(['1234-1-a'], arg.keys())
cache.replace_using(tree)
write.reset_mock()
tree.children[0].children[1].text = "References paragraph (a)"
b.gen_and_write_layers(tree, [], cache, [])
arg = write.call_args_list[3][0][0]
self.assertEqual(['1234-1-a'], arg.keys())
write.reset_mock()
tree.children[0].children[0].text = "Contains no references"
b.gen_and_write_layers(tree, [], cache, [])
arg = write.call_args_list[3][0][0]
self.assertEqual(['1234-1-a'], arg.keys())
write.reset_mock()
notice = {'document_number': '111-222'}
cache.invalidate_by_notice(notice)
b.gen_and_write_layers(tree, [], cache, [])
arg = write.call_args_list[3][0][0]
self.assertEqual(['1234-1-a'], arg.keys())
write.reset_mock()
notice['changes'] = {'1234-1-b': 'some change'}
cache.invalidate_by_notice(notice)
b.gen_and_write_layers(tree, [], cache, [])
arg = write.call_args_list[3][0][0]
self.assertEqual(['1234-1-a', '1234-1-b'], list(sorted(arg.keys())))
write.reset_mock()
notice['changes'] = {'1234-Subpart-A': 'some change'}
cache.invalidate_by_notice(notice)
b.gen_and_write_layers(tree, [], cache, [])
arg = write.call_args_list[3][0][0]
self.assertEqual(['1234-1-b'], list(sorted(arg.keys())))
def build_by_notice(filename,
title,
act_title,
act_section,
notice_doc_numbers,
doc_number=None,
checkpoint=None):
with codecs.open(filename, 'r', 'utf-8') as f:
reg = f.read()
file_digest = hashlib.sha256(reg.encode('utf-8')).hexdigest()
if checkpoint:
checkpointer = Checkpointer(checkpoint)
else:
checkpointer = NullCheckpointer()
# build the initial tree
reg_tree = checkpointer.checkpoint("init-tree-" + file_digest,
lambda: Builder.reg_tree(reg))
title_part = reg_tree.label_id()
if doc_number is None:
doc_number = Builder.determine_doc_number(reg, title, title_part)
checkpointer.suffix = ":".join(
["", title_part, str(args.title), doc_number])
# create the builder
builder = Builder(cfr_title=title,
cfr_part=title_part,
doc_number=doc_number,
checkpointer=checkpointer)
builder.fetch_notices_json()
for notice in notice_doc_numbers:
builder.build_notice_from_doc_number(notice)
builder.write_regulation(reg_tree)
layer_cache = LayerCacheAggregator()
act_title_and_section = [act_title, act_section]
builder.gen_and_write_layers(reg_tree, act_title_and_section, layer_cache)
layer_cache.replace_using(reg_tree)
if args.generate_diffs:
generate_diffs(reg_tree, act_title_and_section, builder, layer_cache)
"notice_doc_# act_title act_section (Generate diffs? " +
"True/False)")
print(" e.g. python build_from.py rege.txt 12 2011-31725 15 1693 " +
"False")
exit()
with codecs.open(sys.argv[1], 'r', 'utf-8') as f:
reg = f.read()
doc_number = sys.argv[3]
# First, the regulation tree
reg_tree = Builder.reg_tree(reg)
builder = Builder(cfr_title=int(sys.argv[2]),
cfr_part=reg_tree.label_id(),
doc_number=doc_number)
# Didn't include the provided version
if not any(n['document_number'] == doc_number for n in builder.notices):
print "Could not find notice_doc_#, %s" % doc_number
exit()
builder.write_notices()
# Always do at least the first reg
logger.info("Version %s", doc_number)
builder.write_regulation(reg_tree)
layer_cache = LayerCacheAggregator()
builder.gen_and_write_layers(reg_tree, sys.argv[4:6], layer_cache)
layer_cache.replace_using(reg_tree)
required=False,
default=True)
args = parser.parse_args()
with codecs.open(args.filename, 'r', 'utf-8') as f:
reg = f.read()
doc_number = args.notice
act_title_and_section = [args.act_title, args.act_section]
# First, the regulation tree
reg_tree = Builder.reg_tree(reg)
builder = Builder(cfr_title=args.title,
cfr_part=reg_tree.label_id(),
doc_number=doc_number)
builder.write_notices()
# Always do at least the first reg
logger.info("Version %s", doc_number)
builder.write_regulation(reg_tree)
layer_cache = LayerCacheAggregator()
builder.gen_and_write_layers(reg_tree, act_title_and_section, layer_cache)
layer_cache.replace_using(reg_tree)
# this used to assume implicitly that if gen-diffs was not specified it was
# True; changed it to explicit check
if args.generate_diffs:
all_versions = {doc_number: reg_tree}
