def test_has_parent_definitions_indicator_p_marker(self):
     t = Terms(None)
     stack = ParentStack()
     stack.add(
         0,
         Node("(a) Definitions. For purposes of this " +
              "section except blah"))
     self.assertTrue(t.has_parent_definitions_indicator(stack))
Esempio n. 2
0
 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))
Esempio n. 3
0
    def test_node_defintions_act(self):
        t = Terms(None)
        stack = ParentStack()
        stack.add(0, Node('Definitions', label=['9999']))

        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_defintions_act(self):
        t = Terms(None)
        stack = ParentStack()
        stack.add(0, Node('Definitions', label=['9999']))

        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_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_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_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)
Esempio n. 9
0
 def test_node_definitions_multiple_xml(self):
     """Find xml definitions which are separated by `and`"""
     stack = ParentStack().add(0, Node(label=['9999']))
     winter = Node("(4) Cold and dreary mean winter.", label=['9999', '4'])
     winter.tagged_text = ('(4) <E T="03">Cold</E> and '
                           '<E T="03">dreary</E> mean winter.')
     inc, _ = Terms(None).node_definitions(winter, stack)
     self.assertEqual(len(inc), 2)
     cold, dreary = inc
     self.assertEqual(cold, Ref('cold', '9999-4', 4))
     self.assertEqual(dreary, Ref('dreary', '9999-4', 13))
Esempio n. 10
0
 def test_node_definitions_xml_or(self):
     """Find xml definitions which are separated by `or`"""
     stack = ParentStack().add(0, Node(label=['9999']))
     tamale = Node("(i) Hot tamale or tamale means nom nom",
                   label=['9999', '4'])
     tamale.tagged_text = ('(i) <E T="03">Hot tamale</E> or <E T="03"> '
                           'tamale</E> means nom nom ')
     inc, _ = Terms(None).node_definitions(tamale, stack)
     self.assertEqual(len(inc), 2)
     hot, tamale = inc
     self.assertEqual(hot, Ref('hot tamale', '9999-4', 4))
     self.assertEqual(tamale, Ref('tamale', '9999-4', 18))
Esempio n. 11
0
 def test_node_definitions_xml_commas(self):
     """Find xml definitions which have commas separating them"""
     stack = ParentStack().add(0, Node(label=['9999']))
     summer = Node("(i) Hot, humid, or dry means summer.",
                   label=['9999', '4'])
     summer.tagged_text = ('(i) <E T="03">Hot</E>, <E T="03">humid</E>, '
                           'or <E T="03">dry</E> means summer.')
     inc, _ = Terms(None).node_definitions(summer, stack)
     self.assertEqual(len(inc), 3)
     hot, humid, dry = inc
     self.assertEqual(hot, Ref('hot', '9999-4', 4))
     self.assertEqual(humid, Ref('humid', '9999-4', 9))
     self.assertEqual(dry, Ref('dry', '9999-4', 19))
Esempio n. 12
0
    def test_node_definitions_too_long(self):
        """Don't find definitions which are too long"""
        stack = ParentStack().add(0, Node('Definitions', label=['9999']))

        text = u"""“I declare under the penalties of perjury that this—(insert
        type of document, such as, statement, application, request,
        certificate), including the documents submitted in support thereof,
        has been examined by me and, to the best of my knowledge and belief,
        is true, correct, and complete.”"""
        node = Node(u'```extract\n{}\n```'.format(text))
        included, excluded = Terms(None).node_definitions(node, stack)
        self.assertEqual([], included)
        self.assertEqual([], 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_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)))
Esempio n. 15
0
    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)
Esempio n. 16
0
class ScopeFinderTest(TestCase):
    def setUp(self):
        self.finder = ScopeFinder()
        self.stack = ParentStack()

    def add_nodes(self, length):
        """There's a common prefix of nodes we'll add"""
        label = ['1000', '3', 'd', '6', 'iii']
        for i in range(length):
            self.stack.add(i, Node(label=label[:i+1]))

    def assert_scope(self, *scopes):
        self.assertEqual(list(scopes), self.finder.determine_scope(self.stack))

    def test_determine_scope_default(self):
        """Defaults to the entire reg"""
        self.add_nodes(2)
        self.assert_scope(('1000',))

    def test_determine_scope_this_part(self):
        """Definitions scoped to a part also cover the interpretations for
        that part"""
        self.add_nodes(1)
        self.stack.add(1, Node('For the purposes of this part, blah blah',
                               label=['1001', '3']))
        self.assert_scope(('1001',), ('1001', Node.INTERP_MARK))

    def test_determine_scope_this_subpart(self):
        """Subpart scope gets expanded to include other sections in the same
        subpart"""
        self.finder.subpart_map = {
            'SubPart 1': ['A', '3'],
            'Other': []
        }
        self.add_nodes(2)
        self.stack.add(2, Node('For the purposes of this subpart, yada yada',
                               label=['1000', '3', 'c']))
        self.assert_scope(('1000', 'A'), ('1000', '3'),
                          ('1000', 'A', Node.INTERP_MARK),
                          ('1000', '3', Node.INTERP_MARK))

    def test_determine_scope_this_section(self):
        """Section scope can be triggered in a child paragraph"""
        self.add_nodes(2)
        self.stack.add(2, Node('For the purposes of this section, blah blah',
                               label=['1000', '3', 'd']))
        self.assert_scope(('1000', '3'), ('1000', '3', Node.INTERP_MARK))

    def test_determine_scope_this_paragraph(self):
        """Paragraph scope is tied to the paragraph that determined it.
        Previous paragraph scopes won't apply to adjacent children"""
        self.add_nodes(2)
        self.stack.add(2, Node('For the purposes of this section, blah blah',
                               label=['1000', '3', 'd']))
        self.stack.add(3, Node('For the purposes of this paragraph, blah blah',
                               label=['1000', '3', 'd', '5']))
        self.assert_scope(('1000', '3', 'd', '5'),
                          ('1000', '3', 'd', '5', Node.INTERP_MARK))

        self.stack.add(3, Node(label=['1002', '3', 'd', '6']))
        self.assert_scope(('1000', '3'), ('1000', '3', Node.INTERP_MARK))

        self.stack.add(3, Node('Blah as used in this paragraph, blah blah',
                               label=['1000', '3', 'd', '7']))
        self.assert_scope(('1000', '3', 'd', '7'),
                          ('1000', '3', 'd', '7', Node.INTERP_MARK))

    def test_determine_scope_purposes_of_specific_paragraph(self):
        self.add_nodes(4)
        self.stack.add(
            4, Node(u'For the purposes of this § 1000.3(d)(6)(i), blah',
                    label=['1000', '3', 'd', '6', 'i']))
        self.assert_scope(('1000', '3', 'd', '6', 'i'),
                          ('1000', '3', 'd', '6', 'i', Node.INTERP_MARK))

    def test_determine_scope_purposes_of_specific_section(self):
        self.add_nodes(4)
        self.stack.add(4, Node(u'For the purposes of § 1000.3, blah',
                               label=['1000', '3', 'd', '6', 'ii']))
        self.assert_scope(('1000', '3'), ('1000', '3', Node.INTERP_MARK))

    def test_determine_scope_as_used_in_thi_section(self):
        self.add_nodes(4)
        self.stack.add(4, Node('As used in this section, blah blah',
                               label=['1000', '3', 'd', '6', 'iii']))
        self.assert_scope(('1000', '3'), ('1000', '3', Node.INTERP_MARK))

    def test_subpart_scope(self):
        self.finder.subpart_map = {
            None: ['1', '2', '3'],
            'A': ['7', '5', '0'],
            'Q': ['99', 'abc', 'q']
        }
        self.assertEqual([['111', '1'], ['111', '2'], ['111', '3']],
                         self.finder.subpart_scope(['111', '3']))
        self.assertEqual([['115', '7'], ['115', '5'], ['115', '0']],
                         self.finder.subpart_scope(['115', '5']))
        self.assertEqual([['62', '99'], ['62', 'abc'], ['62', 'q']],
                         self.finder.subpart_scope(['62', 'abc']))
        self.assertEqual([], self.finder.subpart_scope(['71', 'Z']))
    def test_determine_scope(self):
        stack = ParentStack()
        t = Terms(None)

        stack.add(0, Node(label=['1000']))
        stack.add(1, Node(label=['1000', '1']))

        # Defaults to the entire reg
        self.assertEqual([('1000',)], t.determine_scope(stack))

        stack.add(1, Node('For the purposes of this part, blah blah',
                          label=['1001', '2']))
        self.assertEqual([('1001',), ('1001', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        t.subpart_map = {
            'SubPart 1': ['A', '3'],
            'Other': []
        }
        stack.add(1, Node(label=['1000', '3']))
        stack.add(2, Node('For the purposes of this subpart, yada yada',
                          label=['1000', '3', 'c']))
        self.assertEqual([('1000', 'A'), ('1000', '3'),
                          ('1000', 'A', Node.INTERP_MARK),
                          ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(2, Node('For the purposes of this section, blah blah',
                          label=['1000', '3', 'd']))
        self.assertEqual([('1000', '3'), ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(3, Node('For the purposes of this paragraph, blah blah',
                          label=['1000', '3', 'd', '5']))
        self.assertEqual([('1000', '3', 'd', '5'),
                          ('1000', '3', 'd', '5', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(3, Node(label=['1002', '3', 'd', '6']))
        self.assertEqual([('1000', '3'), ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(3, Node('Blah as used in this paragraph, blah blah',
                          label=['1000', '3', 'd', '7']))
        self.assertEqual([('1000', '3', 'd', '7'),
                          ('1000', '3', 'd', '7', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(4, Node(u'For the purposes of this § 1000.3(d)(6)(i), blah',
                          label=['1000', '3', 'd', '6', 'i']))
        self.assertEqual([('1000', '3', 'd', '6', 'i'),
                          ('1000', '3', 'd', '6', 'i', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(4, Node(u'For the purposes of § 1000.3, blah',
                          label=['1000', '3', 'd', '6', 'ii']))
        self.assertEqual([('1000', '3'),
                          ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(4, Node('As used in this section, blah blah',
                          label=['1000', '3', 'd', '6', 'iii']))
        self.assertEqual(
            [('1000', '3'), ('1000', '3', Node.INTERP_MARK)],
            t.determine_scope(stack))
    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)
    def test_has_parent_definitions_indicator_the_term_means(self):
        t = Terms(None)
        stack = ParentStack()
        stack.add(0, Node('Contains no terms or definitions'))
        self.assertFalse(t.has_parent_definitions_indicator(stack))
        stack.add(1, Node("(a) The term Bob means awesome"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
        stack.add(2, Node("No defs either"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))

        stack.pop()
        stack.pop()
        stack.add(1, Node(u"(a) “Term” means some stuff"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))

        stack.pop()
        stack.add(1, Node("(a) The term Bob refers to"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
    def test_has_parent_definitions_indicator(self):
        t = Terms(None)
        stack = ParentStack()
        stack.add(0, Node("This has no defs"))
        self.assertFalse(t.has_parent_definitions_indicator(stack))
        stack.add(1, Node("No Def", title="No def"))
        self.assertFalse(t.has_parent_definitions_indicator(stack))
        stack.add(2, Node("Tomatoes do not meet the definition 'vegetable'"))
        self.assertFalse(t.has_parent_definitions_indicator(stack))

        stack.add(3, Node("Definition. This has a definition."))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
        stack.pop()
        self.assertFalse(t.has_parent_definitions_indicator(stack))

        stack.add(3, Node("Definitions. This has multiple!"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
        stack.pop()
        self.assertFalse(t.has_parent_definitions_indicator(stack))

        stack.add(3, Node("No body", title="But Definition is in the title"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
class SmartQuotesTest(TestCase):
    def setUp(self):
        self.stack = ParentStack()
        self.finder = def_finders.SmartQuotes(self.stack)
        self.depth = 0

    def check_indicator(self, expected, text, title=None):
        """Common pattern for adding a node to the stack and then verifying
        the `has_def_indicator` method"""
        self.stack.add(self.depth, Node(text, title=title))
        self.assertEqual(self.finder.has_def_indicator(), expected)
        self.depth += 1

    def pop_and_check(self, expected=False):
        """Common pattern for popping the stack and then verifying the
        `has_def_indicator` method"""
        self.stack.pop()
        self.assertEqual(self.finder.has_def_indicator(), expected)
        self.depth -= 1

    def test_has_def_indicator(self):
        self.check_indicator(False, "This has no defs")
        self.check_indicator(False, "No Def", title="No def")
        self.check_indicator(
            False, "Tomatoes do not meet the definition 'vegetable'")
        self.check_indicator(True, "Definition. This has a definition.")
        self.pop_and_check()
        self.check_indicator(True, "Definitions. This has multiple!")
        self.pop_and_check()
        self.check_indicator(True, "No body",
                             title="But Definition is in the title")

    def test_has_def_indicator_p_marker(self):
        self.check_indicator(
            True,
            "(a) Definitions. For purposes of this section except blah")

    def test_has_def_indicator_the_term_means(self):
        self.check_indicator(False, 'Contains no terms or definitions')
        self.check_indicator(True, "(a) The term Bob means awesome")
        self.check_indicator(True, "No defs either")

        self.pop_and_check(expected=True)
        self.pop_and_check()
        self.check_indicator(True, u"(a) “Term” means some stuff")

        self.pop_and_check()
        self.check_indicator(True, "(a) The term Bob refers to")

    def assert_finds_definition(self, text, *expected):
        """Check that the definition is _not_ found when it has no
        "Definition" parent and _is_ found when such a parent exists"""
        self.stack.add(0, Node(label=['999']))
        node = Node(text)
        self.assertEqual([], self.finder.find(node))

        self.stack.add(1, Node("Definitions", label=['999', '1']))
        actual = self.finder.find(node)
        self.assertEqual(len(expected), len(actual))
        for expected_ref, actual_ref in zip(expected, actual):
            self.assertEqual(expected_ref.term, actual_ref.term)
            self.assertEqual(expected_ref.start, actual_ref.start)

        self.stack.pop()

    def test_find(self):
        """Tests several examples involving smart quotes"""
        self.assert_finds_definition(
            u'This has a “worD” and then more',
            def_finders.Ref('word', None, 12))
        self.assert_finds_definition(
            u'I have “anotheR word” term and “moree”',
            def_finders.Ref('another word', None, 8),
            def_finders.Ref('moree', None, 32))
        self.assert_finds_definition(
            u'But the child “DoeS sEe”?',
            def_finders.Ref('does see', None, 15))
        self.assert_finds_definition(
            u'Start with “this,”', def_finders.Ref('this', None, 12))
        self.assert_finds_definition(
            u'Start with “this;”', def_finders.Ref('this', None, 12))
        self.assert_finds_definition(
            u'Start with “this.”', def_finders.Ref('this', None, 12))
        self.assert_finds_definition(
            u'As do “subchildren”', def_finders.Ref('subchildren', None, 7))
    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'(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)
    def test_has_parent_definitions_indicator_the_term_means(self):
        t = Terms(None)
        stack = ParentStack()
        stack.add(0, Node('Contains no terms or definitions'))
        self.assertFalse(t.has_parent_definitions_indicator(stack))
        stack.add(1, Node("(a) The term Bob means awesome"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
        stack.add(2, Node("No defs either"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))

        stack.pop()
        stack.pop()
        stack.add(1, Node(u"(a) “Term” means some stuff"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))

        stack.pop()
        stack.add(1, Node("(a) The term Bob refers to"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
    def test_has_parent_definitions_indicator(self):
        t = Terms(None)
        stack = ParentStack()
        stack.add(0, Node("This has no defs"))
        self.assertFalse(t.has_parent_definitions_indicator(stack))
        stack.add(1, Node("No Def", title="No def"))
        self.assertFalse(t.has_parent_definitions_indicator(stack))
        stack.add(2, Node("Tomatoes do not meet the definition 'vegetable'"))
        self.assertFalse(t.has_parent_definitions_indicator(stack))

        stack.add(3, Node("Definition. This has a definition."))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
        stack.pop()
        self.assertFalse(t.has_parent_definitions_indicator(stack))

        stack.add(3, Node("Definitions. This has multiple!"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
        stack.pop()
        self.assertFalse(t.has_parent_definitions_indicator(stack))

        stack.add(3, Node("No body", title="But Definition is in the title"))
        self.assertTrue(t.has_parent_definitions_indicator(stack))
 def setUp(self):
     self.stack = ParentStack()
     self.finder = def_finders.SmartQuotes(self.stack)
     self.depth = 0
    def test_determine_scope(self):
        stack = ParentStack()
        t = Terms(None)

        stack.add(0, Node(label=['1000']))
        stack.add(1, Node(label=['1000', '1']))

        # Defaults to the entire reg
        self.assertEqual([('1000',)], t.determine_scope(stack))

        stack.add(1, Node('For the purposes of this part, blah blah',
                          label=['1001', '2']))
        self.assertEqual([('1001',), ('1001', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        t.subpart_map = {
            'SubPart 1': ['A', '3'],
            'Other': []
        }
        stack.add(1, Node(label=['1000', '3']))
        stack.add(2, Node('For the purposes of this subpart, yada yada',
                          label=['1000', '3', 'c']))
        self.assertEqual([('1000', 'A'), ('1000', '3'),
                          ('1000', 'A', Node.INTERP_MARK),
                          ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(2, Node('For the purposes of this section, blah blah',
                          label=['1000', '3', 'd']))
        self.assertEqual([('1000', '3'), ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(3, Node('For the purposes of this paragraph, blah blah',
                          label=['1000', '3', 'd', '5']))
        self.assertEqual([('1000', '3', 'd', '5'),
                          ('1000', '3', 'd', '5', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(3, Node(label=['1002', '3', 'd', '6']))
        self.assertEqual([('1000', '3'), ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))
        stack.add(4, Node(u'For the purposes of this § 1000.3(d)(6)(i), blah',
                          label=['1000', '3', 'd', '6', 'i']))
        self.assertEqual([('1000', '3', 'd', '6', 'i'),
                          ('1000', '3', 'd', '6', 'i', Node.INTERP_MARK)],
                         t.determine_scope(stack))

        stack.add(4, Node(u'For the purposes of § 1000.3, blah',
                          label=['1000', '3', 'd', '6', 'ii']))
        self.assertEqual([('1000', '3'),
                          ('1000', '3', Node.INTERP_MARK)],
                         t.determine_scope(stack))
Esempio n. 27
0
 def setUp(self):
     self.finder = ScopeFinder()
     self.stack = ParentStack()
 def test_has_parent_definitions_indicator_p_marker(self):
     t = Terms(None)
     stack = ParentStack()
     stack.add(0, Node("(a) Definitions. For purposes of this " +
                       "section except blah"))
     self.assertTrue(t.has_parent_definitions_indicator(stack))