def test_print_conll_other_tags_ignore_all(capsys):
print_conll_style_tags_for_communication(comm_with_other_tags())
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\n'
'-----\t-----\n'
'1\tThe\n'
'2\tquick\n')
def test_print_conll_ending_tags(capsys):
print_conll_style_tags_for_communication(comm_with_other_tags(),
endings=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\tENDING\n'
'-----\t-----\t------\n'
'1\tThe\t3\n'
'2\tquick\t9\n'
'3\tbrown\t15\n')
def test_print_conll_start_tags(capsys):
print_conll_style_tags_for_communication(comm_with_other_tags(),
starts=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\tSTART\n'
'-----\t-----\t-----\n'
'1\tThe\t0\n'
'2\tquick\t4\n'
'3\tbrown\t10\n')
def test_print_conll_other_tags_ignore_some(capsys):
print_conll_style_tags_for_communication(comm_with_other_tags(),
other_tags=dict(upper=None))
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\tupper\n'
'-----\t-----\t-----\n'
'1\tThe\tTHE\n'
'2\tquick\tQUICK\n')
assert '3\tshe\tSHE\n' in out
def test_print_conll_other_tags_ignore_all(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags())
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\n'
'-----\t-----\n'
'1\tThe\n'
'2\tquick\n'
)
def test_print_conll_other_tags_ignore_some(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags(), other_tags=dict(upper=None))
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\tupper\n'
'-----\t-----\t-----\n'
'1\tThe\tTHE\n'
'2\tquick\tQUICK\n'
)
assert '3\tshe\tSHE\n' in out
def test_print_conll_other_tags_repeated_ner(capsys):
print_conll_style_tags_for_communication(comm_with_other_tags(
'ner', 'ner'),
ner=True,
other_tags=dict(upper=None))
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\tNER\tNER\tupper\n'
'-----\t-----\t---\t---\t-----\n'
'1\tThe\tner_0/0\tner_0/1\tTHE\n'
'2\tquick\tner_1/0\tner_1/1\tQUICK\n')
assert '3\tshe\tner_2/0\tner_2/1\tSHE\n' in out
def test_print_conll_other_tags_repeated_other_tag_filtered(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags('upper', 'ner'),
ner=True,
other_tags=dict(upper=lambda anns: filter(
lambda ann: ann.metadata.tool == 'tool/0', anns)))
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\tNER\tupper\n'
'-----\t-----\t---\t-----\n'
'1\tThe\tner_0/1\tupper_0/0\n'
'2\tquick\tner_1/1\tupper_1/0\n')
assert '3\tshe\tner_2/1\tupper_2/0\n' in out
def test_print_conll_missing_char_offsets(capsys):
comm_without_token_textspans = comm_with_other_tags()
for tokenization in get_tokenizations(comm_without_token_textspans):
for token in get_tokens(tokenization):
token.textSpan = None
print_conll_style_tags_for_communication(comm_without_token_textspans,
char_offsets=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\tCHAR\n'
'-----\t-----\t----\n'
'1\tThe\t\n'
'2\tquick\t\n')
def test_print_conll_start_tags(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags(),
starts=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\tSTART\n'
'-----\t-----\t-----\n'
'1\tThe\t0\n'
'2\tquick\t4\n'
'3\tbrown\t10\n'
)
def test_print_conll_ending_tags(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags(),
endings=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\tENDING\n'
'-----\t-----\t------\n'
'1\tThe\t3\n'
'2\tquick\t9\n'
'3\tbrown\t15\n'
)
def test_print_conll_other_tags_repeated_ner(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags('ner', 'ner'),
ner=True,
other_tags=dict(upper=None))
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\tNER\tNER\tupper\n'
'-----\t-----\t---\t---\t-----\n'
'1\tThe\tner_0/0\tner_0/1\tTHE\n'
'2\tquick\tner_1/0\tner_1/1\tQUICK\n'
)
assert '3\tshe\tner_2/0\tner_2/1\tSHE\n' in out
def test_print_conll_missing_char_offsets(capsys):
comm_without_token_textspans = comm_with_other_tags()
for tokenization in get_tokenizations(comm_without_token_textspans):
for token in get_tokens(tokenization):
token.textSpan = None
print_conll_style_tags_for_communication(
comm_without_token_textspans, char_offsets=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\tCHAR\n'
'-----\t-----\t----\n'
'1\tThe\t\n'
'2\tquick\t\n'
)
def test_print_conll_other_tags_repeated_other_tag_filtered(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags('upper', 'ner'),
ner=True,
other_tags=dict(
upper=lambda anns: filter(lambda ann: ann.metadata.tool == 'tool/0', anns)
))
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\tNER\tupper\n'
'-----\t-----\t---\t-----\n'
'1\tThe\tner_0/1\tupper_0/0\n'
'2\tquick\tner_1/1\tupper_1/0\n'
)
assert '3\tshe\tner_2/1\tupper_2/0\n' in out
def test_print_conll_missing_tags(capsys):
# We don't use comm_with_other_tags() here because we want to test
# the case where:
# tokenization.TokenTaggingList = None
comm = create_comm(
'quick', '''\
The quick brown fox jumped
over the lazy dog .
Or did she ?
''')
print_conll_style_tags_for_communication(comm, ner=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith('INDEX\tTOKEN\n'
'-----\t-----\n'
'1\tThe\n'
'2\tquick\n')
def test_print_conll_missing_tags(capsys):
# We don't use comm_with_other_tags() here because we want to test
# the case where:
# tokenization.TokenTaggingList = None
comm = create_comm('quick', '''\
The quick brown fox jumped
over the lazy dog .
Or did she ?
''')
print_conll_style_tags_for_communication(comm, ner=True)
(out, err) = capsys.readouterr()
assert err == ''
assert out.startswith(
'INDEX\tTOKEN\n'
'-----\t-----\n'
'1\tThe\n'
'2\tquick\n'
)
def test_print_conll_char_offsets(capsys):
print_conll_style_tags_for_communication(
comm_with_other_tags(), char_offsets=True)
def test_print_conll_char_offsets(capsys):
print_conll_style_tags_for_communication(comm_with_other_tags(),
char_offsets=True)
