def test_should_add_lb_element_before_token_without_tag(self):
tei_parent = _lines_to_tei(
E.front(),
[TeiLine([]), TeiLine([_tei_text(TOKEN_1, None)])])
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == [TeiTagNames.LB]
assert child_elements[0].tail == TOKEN_1
def test_should_preserve_space_after_lb_within_tokens_with_same_tag(self):
tei_parent = _lines_to_tei(E.front(), [
TeiLine([_tei_text(TOKEN_1, TAG_1)]),
TeiLine([_tei_text(' ' + TOKEN_2, TAG_1)])
])
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == [TAG_1]
assert _to_xml(child_elements[0]) == (
'<{tag1}>{token1}<{lb}/> {token2}</{tag1}>'.format(
tag1=TAG_1, token1=TOKEN_1, token2=TOKEN_2, lb=TeiTagNames.LB))
def test_should_not_include_standalone_space_after_lb_in_tag_before_other_different_tag(
self):
tei_parent = _lines_to_tei(E.front(), [
TeiLine([_tei_text(TOKEN_1, TAG_1)]),
TeiLine([_tei_text(' ', None),
_tei_text(TOKEN_2, TAG_2)])
])
assert _to_xml(tei_parent) == (
'<front><{tag1}>{token1}<{lb}/></{tag1}> <{tag2}>{token2}</{tag2}></front>'
.format(tag1=TAG_1,
tag2=TAG_2,
token1=TOKEN_1,
token2=TOKEN_2,
lb=TeiTagNames.LB))
def test_should_map_tag_to_tei_path(self):
tei_parent = _lines_to_tei(E.front(),
[TeiLine([_tei_text(TOKEN_1, TAG_1)])],
tag_to_tei_path_mapping={TAG_1: TAG_2})
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == [TAG_2]
assert child_elements[0].text == TOKEN_1
def test_should_map_tag_to_nested_tei_path(self):
tei_parent = _lines_to_tei(
E.front(), [TeiLine([_tei_text(TOKEN_1, TAG_1)])],
tag_to_tei_path_mapping={TAG_1: 'parent/child'})
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == ['parent']
nested_child_elements = list(child_elements[0])
assert [c.tag for c in nested_child_elements] == ['child']
assert nested_child_elements[0].text == TOKEN_1
def test_should_combine_tokens(self):
tei_parent = _lines_to_tei(E.front(), [
TeiLine(
[_tei_text(TOKEN_1, TAG_1),
_tei_text(' ' + TOKEN_2, TAG_1)])
])
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == [TAG_1]
assert child_elements[0].text == ' '.join([TOKEN_1, TOKEN_2])
def test_should_use_common_path_between_similar_nested_tag_paths(self):
tei_parent = _lines_to_tei(
E.front(),
[TeiLine([_tei_text(TOKEN_1, TAG_1),
_tei_text(TOKEN_2, TAG_2)])],
tag_to_tei_path_mapping={
TAG_1: 'parent/child1',
TAG_2: 'parent/child2'
})
assert _to_xml(tei_parent) == (
'<front><parent><child1>{token1}</child1>'
'<child2>{token2}</child2></parent></front>'.format(
token1=TOKEN_1, token2=TOKEN_2))
def test_should_apply_default_tag(self):
tei_parent = _lines_to_tei(
E.front(), [TeiLine([_tei_text(TOKEN_1, '')])],
tag_to_tei_path_mapping={DEFAULT_TAG_KEY: 'other'})
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == ['other']
def test_should_convert_single_token(self):
tei_parent = _lines_to_tei(E.front(),
[TeiLine([_tei_text(TOKEN_1, TAG_1)])])
child_elements = list(tei_parent)
assert [c.tag for c in child_elements] == [TAG_1]
assert [c.text for c in child_elements] == [TOKEN_1]