def test_node_defintions_act(self): t = Terms(None) stack = ParentStack() stack.add(0, Node('Definitions', label=['9999'])) node = Node(u'“Act” means some reference to 99 U.S.C. 1234') included, excluded = t.node_definitions(node, stack) self.assertEqual([], included) self.assertEqual(1, len(excluded)) node = Node(u'“Act” means something else entirely') included, excluded = t.node_definitions(node, stack) self.assertEqual(1, len(included)) self.assertEqual([], excluded)
def test_node_definitions_needs_term(self): t = Terms(None) stack = ParentStack() stack.add(0, Node('Definitions', label=['9999'])) node = Node(u"However, for purposes of rescission under §§ 1111.15 " + u"and 1111.13, and for purposes of §§ 1111.12(a)(1), " + u"and 1111.46(d)(4), the term means all calendar " + u"days...") self.assertEqual(([], []), t.node_definitions(node, stack))
def test_node_definitions_exclusion(self): n1 = Node(u'“Bologna” is a type of deli meat', label=['111', '1']) n2 = Node(u'Let us not forget that the term “bologna” does not ' + 'include turtle meat', label=['111', '1', 'a']) t = Terms(Node(label=['111'], children=[n1, n2])) t.pre_process() stack = ParentStack() stack.add(1, Node('Definitions')) included, excluded = t.node_definitions(n1, stack) self.assertEqual([Ref('bologna', '111-1', (1, 8))], included) self.assertEqual([], excluded) t.scoped_terms[('111', '1')] = included included, excluded = t.node_definitions(n2, stack) self.assertEqual([], included) self.assertEqual([Ref('bologna', '111-1-a', (33, 40))], excluded)
def test_node_definitions_multiple_xml(self): t = Terms(None) stack = ParentStack() stack.add(0, Node(label=['9999'])) winter = Node("(4) Cold and dreary mean winter.", label=['9999', '4']) tagged = '(4) <E T="03">Cold</E> and <E T="03">dreary</E> mean ' tagged += 'winter.' winter.tagged_text = tagged inc, _ = t.node_definitions(winter, stack) self.assertEqual(len(inc), 2) cold, dreary = inc self.assertEqual(cold, Ref('cold', '9999-4', (4, 8))) self.assertEqual(dreary, Ref('dreary', '9999-4', (13, 19))) summer = Node("(i) Hot, humid, or dry means summer.", label=['9999', '4']) tagged = '(i) <E T="03">Hot</E>, <E T="03">humid</E>, or ' tagged += '<E T="03">dry</E> means summer.' summer.tagged_text = tagged inc, _ = t.node_definitions(summer, stack) self.assertEqual(len(inc), 3) hot, humid, dry = inc self.assertEqual(hot, Ref('hot', '9999-4', (4, 7))) self.assertEqual(humid, Ref('humid', '9999-4', (9, 14))) self.assertEqual(dry, Ref('dry', '9999-4', (19, 22))) tamale = Node("(i) Hot tamale or tamale means nom nom", label=['9999', '4']) tagged = '(i) <E T="03">Hot tamale</E> or <E T="03"> tamale</E> ' tagged += 'means nom nom ' tamale.tagged_text = tagged inc, _ = t.node_definitions(tamale, stack) self.assertEqual(len(inc), 2) hot, tamale = inc self.assertEqual(hot, Ref('hot tamale', '9999-4', (4, 14))) self.assertEqual(tamale, Ref('tamale', '9999-4', (18, 24)))
def test_node_definitions_no_def(self): """Verify that none of the matchers match certain strings""" t = Terms(None) stack = ParentStack() stack.add(0, Node(label=['999'])) stack.add(1, Node('Definitions', label=['999', '1'])) no_defs = [ 'This has no defs', 'Also has no terms', 'Still no terms, but', 'the next one does' ] for txt in no_defs: defs, exc = t.node_definitions(Node(txt), stack) self.assertEqual([], defs) self.assertEqual([], exc)
def test_node_definitions(self): t = Terms(None) smart_quotes = [ (u'This has a “worD” and then more', [Ref('word', 'aaa', (12, 16))]), (u'I have “anotheR word” term and “moree”', [Ref('another word', 'bbb', (8, 20)), Ref('moree', 'bbb', (32, 37))]), (u'But the child “DoeS sEe”?', [Ref('does see', 'ccc', (15, 23))]), (u'Start with “this,”', [Ref('this', 'hhh', (12, 16))]), (u'Start with “this;”', [Ref('this', 'iii', (12, 16))]), (u'Start with “this.”', [Ref('this', 'jjj', (12, 16))]), (u'As do “subchildren”', [Ref('subchildren', 'ddd', (7, 18))])] no_defs = [ u'This has no defs', u'Also has no terms', u'Still no terms, but', u'the next one does'] xml_defs = [ (u'(4) Thing means a thing that is defined', u'(4) <E T="03">Thing</E> means a thing that is defined', Ref('thing', 'eee', (4, 9))), (u'(e) Well-meaning lawyers means people who do weird things', u'(e) <E T="03">Well-meaning lawyers</E> means people who do ' + 'weird things', Ref('well-meaning lawyers', 'fff', (4, 24))), (u'(e) Words have the same meaning as in a dictionary', u'(e) <E T="03">Words</E> have the same meaning as in a ' + 'dictionary', Ref('words', 'ffg', (4, 9))), (u'(e) Banana has the same meaning as bonono', u'(e) <E T="03">Banana</E> has the same meaning as bonono', Ref('banana', 'fgf', (4, 10))), (u'(f) Huge billowy clouds means I want to take a nap', u'(f) <E T="03">Huge billowy clouds</E> means I want to take a ' + 'nap', Ref('huge billowy clouds', 'ggg', (4, 23)))] xml_no_defs = [ (u'(d) Term1 or term2 means stuff', u'(d) <E T="03">Term1</E> or <E T="03">term2></E> means stuff'), (u'This term means should not match', u'<E T="03">This term</E> means should not match')] scope_term_defs = [ ('For purposes of this section, the term blue means the color', Ref('blue', '11-11', (39, 43))), ('For purposes of paragraph (a)(1) of this section, the term ' + 'cool bro means hip cat', Ref('cool bro', '11-22', (59, 67))), ('For purposes of this paragraph, po jo means "poor Joe"', Ref('po jo', '11-33', (32, 37)))] stack = ParentStack() stack.add(0, Node(label=['999'])) for txt in no_defs: defs, exc = t.node_definitions(Node(txt), stack) self.assertEqual([], defs) self.assertEqual([], exc) for txt, refs in smart_quotes: defs, exc = t.node_definitions(Node(txt), stack) self.assertEqual([], defs) self.assertEqual([], exc) for txt, xml in xml_no_defs: node = Node(txt) node.tagged_text = xml defs, exc = t.node_definitions(node, stack) self.assertEqual([], defs) self.assertEqual([], exc) for txt, xml, ref in xml_defs: node = Node(txt, label=[ref.label]) node.tagged_text = xml defs, exc = t.node_definitions(node, stack) self.assertEqual([ref], defs) self.assertEqual([], exc) for txt, ref in scope_term_defs: defs, exc = t.node_definitions( Node(txt, label=ref.label.split('-')), stack) self.assertEqual([ref], defs) self.assertEqual([], exc) # smart quotes are affected by the parent stack.add(1, Node('Definitions', label=['999', '1'])) for txt in no_defs: defs, exc = t.node_definitions(Node(txt), stack) self.assertEqual([], defs) self.assertEqual([], exc) for txt, refs in smart_quotes: defs, exc = t.node_definitions(Node(txt, label=[refs[0].label]), stack) self.assertEqual(refs, defs) self.assertEqual([], exc) for txt, xml in xml_no_defs: node = Node(txt) node.tagged_text = xml defs, exc = t.node_definitions(node, stack) self.assertEqual([], defs) self.assertEqual([], exc) for txt, xml, ref in xml_defs: node = Node(txt, label=[ref.label]) node.tagged_text = xml defs, exc = t.node_definitions(node, stack) self.assertEqual([ref], defs) self.assertEqual([], exc)