def test_should_ignore_sub_fields_if_excluded(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper,
segment_affiliation: bool, sub_fields: str, expected_country: str):
target_jats_xml = etree.tostring(
get_target_xml_node(affiliation_nodes=[
E.aff(E.label(LABEL_1), ' ', E.country(COUNTRY_1)),
]))
test_helper.tei_raw_file_path.write_bytes(
etree.tostring(
get_affiliation_tei_node(
[TEI_E.affiliation(TEI_E.other(LABEL_1), ' ',
COUNTRY_1)])))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict, 'matcher': 'simple',
'segment-affiliation': segment_affiliation,
'fields': 'author_aff',
'sub-fields': sub_fields,
'preserve-sub-tags': True,
'no-preserve-sub-fields': 'author_aff-label'
}),
save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
first_aff = get_first_affiliation(tei_auto_root)
assert get_tei_xpath_text(first_aff, './tei:marker') == LABEL_1
assert get_tei_xpath_text(
first_aff, './tei:address/tei:country') == expected_country
def test_should_auto_annotate_single_figure_with_label_and_caption(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper):
target_figure_content_nodes = [
E.label(LABEL_1),
' ',
E.caption(E.p(TEXT_1))
]
target_jats_xml = etree.tostring(
get_target_xml_node(body_nodes=[
E.fig(*target_figure_content_nodes),
])
)
test_helper.tei_raw_file_path.write_bytes(etree.tostring(
get_training_tei_node([
E.figure(get_nodes_text(target_figure_content_nodes))
])
))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict,
'matcher': 'simple',
'fields': 'figure'
}), save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
first_figure = get_first_figure(tei_auto_root)
assert get_tei_xpath_text(first_figure, './/label') == (
LABEL_1
)
assert get_tei_xpath_text(first_figure, './figDesc') == (
TEXT_1
)
def test_should_not_preserve_existing_sub_tags_if_provided_by_excluded(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper,
segment_affiliation: bool):
target_jats_xml = etree.tostring(
get_target_xml_node(affiliation_nodes=[
E.aff(E.label(LABEL_1), ' ' + LABEL_2 + ' ' + TEXT_1),
]))
test_helper.tei_raw_file_path.write_bytes(
etree.tostring(
get_affiliation_tei_node([
TEI_E.affiliation(TEI_E.other(LABEL_1), ' ',
TEI_E.marker(LABEL_2), ' ',
TEI_E.preserved(TEXT_1))
])))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict, 'matcher': 'simple',
'segment-affiliation': segment_affiliation,
'fields': 'author_aff',
'preserve-sub-tags': True,
'no-preserve-sub-fields': 'author_aff-label'
}),
save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
first_aff = get_first_affiliation(tei_auto_root)
assert get_tei_xpath_text(first_aff, './tei:marker') == LABEL_1
assert get_tei_xpath_text(first_aff, './tei:preserved') == TEXT_1
def test_should_group_preserved_and_auto_annotated_address_fields(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper):
target_jats_xml = etree.tostring(
get_target_xml_node(affiliation_nodes=[
E.aff(E.label(LABEL_1), TEXT_1 + ' ' + CITY_1 +
', ', E.country(COUNTRY_1)),
]))
test_helper.tei_raw_file_path.write_bytes(
etree.tostring(
get_affiliation_tei_node([
TEI_E.affiliation(
TEI_E.marker(LABEL_1), ' ' + TEXT_1 + ' ',
TEI_E.address(TEI_E.settlement(CITY_1), ', ',
TEI_E.country(COUNTRY_1)))
])))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict, 'matcher': 'simple',
'segment-affiliation': True,
'preserve-sub-tags': True,
'no-preserve-sub-fields': 'author_aff-label,author_aff-country',
'fields': 'author_aff'
}),
save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
tei_auto_aff = get_first_affiliation(tei_auto_root)
assert get_tei_xpath_text(tei_auto_aff, './tei:marker') == LABEL_1
assert get_tei_xpath_text(tei_auto_aff,
'./tei:address/tei:settlement') == CITY_1
assert get_tei_xpath_text(tei_auto_aff,
'./tei:address/tei:country') == COUNTRY_1
assert get_tei_xpath_text(tei_auto_aff, './tei:address',
delimiter='|') == (CITY_1 + ', ' + COUNTRY_1)
def test_should_preserve_existing_sub_tags(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper,
segment_affiliation: bool):
untagged_text = ' some affiliation '
other1_sub_tag_text = 'other-sub-tag1'
other2_sub_tag_text = 'other-sub-tag2'
target_jats_xml = etree.tostring(
get_target_xml_node(affiliation_nodes=[
E.aff(other1_sub_tag_text + LABEL_1 + untagged_text +
other2_sub_tag_text),
]))
test_helper.tei_raw_file_path.write_bytes(
etree.tostring(
get_affiliation_tei_node([
TEI_E.affiliation(TEI_E.other1(other1_sub_tag_text),
TEI_E.marker(LABEL_1),
get_nodes_text([untagged_text]),
TEI_E.other2(other2_sub_tag_text))
])))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict, 'matcher': 'simple',
'segment-affiliation': segment_affiliation,
'fields': 'author_aff',
'preserve-sub-tags': True
}),
save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
first_aff = get_first_affiliation(tei_auto_root)
assert get_tei_xpath_text(first_aff, './tei:marker') == LABEL_1
assert get_tei_xpath_text(first_aff,
'./tei:other1') == other1_sub_tag_text
assert get_tei_xpath_text(first_aff,
'./tei:other2') == other2_sub_tag_text
def test_should_auto_annotate_single_affiliation_with_all_supported_fields(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper):
target_reference_content_nodes = [
E('label', LABEL_1), ' ',
E('addr-line',
E('named-content', {'content-type': 'department'},
DEPARTMENT_1)), ' ',
E.institution(INSTITUTION_1), ' ',
E('addr-line', E('named-content', {'content-type': 'city'},
CITY_1)), ' ',
E('addr-line',
E('named-content', {'content-type': 'postcode'}, POST_CODE_1)),
' ',
E('addr-line',
E('named-content', {'content-type': 'state'}, STATE_1)), ' ',
E.country(COUNTRY_1)
]
target_jats_xml = etree.tostring(
get_target_xml_node(affiliation_nodes=[
E.aff(*target_reference_content_nodes),
]))
test_helper.tei_raw_file_path.write_bytes(
etree.tostring(
get_affiliation_tei_node([
TEI_E.affiliation(
get_nodes_text(target_reference_content_nodes))
])))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict, 'matcher': 'simple',
'fields': 'author_aff'
}),
save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
first_aff = get_first_affiliation(tei_auto_root)
assert get_tei_xpath_text(first_aff, './tei:marker') == LABEL_1
assert get_tei_xpath_text(
first_aff, './tei:orgName[@type="department"]') == (DEPARTMENT_1)
assert get_tei_xpath_text(
first_aff, './tei:orgName[@type="institution"]') == (INSTITUTION_1)
assert get_tei_xpath_text(first_aff,
'./tei:address/tei:settlement') == CITY_1
assert get_tei_xpath_text(first_aff,
'./tei:address/tei:postCode') == POST_CODE_1
assert get_tei_xpath_text(first_aff,
'./tei:address/tei:region') == STATE_1
assert get_tei_xpath_text(first_aff,
'./tei:address/tei:country') == COUNTRY_1
def test_should_auto_annotate_single_affiliation_with_single_field(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper):
target_reference_content_nodes = [E('label', LABEL_1), ' Some text']
target_jats_xml = etree.tostring(
get_target_xml_node(affiliation_nodes=[
E.aff(*target_reference_content_nodes),
]))
test_helper.tei_raw_file_path.write_bytes(
etree.tostring(
get_affiliation_tei_node([
TEI_E.affiliation(
get_nodes_text(target_reference_content_nodes))
])))
LOGGER.debug('target_jats_xml: %s', target_jats_xml)
test_helper.xml_file_path.write_bytes(target_jats_xml)
main(dict_to_args({
**test_helper.main_args_dict, 'matcher': 'simple',
'fields': 'author_aff'
}),
save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
first_aff = get_first_affiliation(tei_auto_root)
assert get_tei_xpath_text(first_aff, './tei:marker') == (LABEL_1)