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_tolerate_invalid_closing_content_element(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper):
target_figure_content_nodes_1 = [
E.label(LABEL_1),
' ',
E.caption(E.p(TEXT_1))
]
target_figure_content_nodes_2 = [
E.label(LABEL_2),
' ',
E.caption(E.p(TEXT_2))
]
target_jats_xml = etree.tostring(
get_target_xml_node(body_nodes=[
E.fig(*target_figure_content_nodes_1),
E.fig(*target_figure_content_nodes_2)
])
)
test_helper.tei_raw_file_path.write_text(''.join([
'<tei><text>',
'<figure>',
get_nodes_text(target_figure_content_nodes_1),
'</content>',
'</figure>',
'<figure>',
get_nodes_text(target_figure_content_nodes_2),
'</content>',
'</figure>',
'</text></tei>'
]))
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()
assert get_tei_xpath_text_list(tei_auto_root, './/figure//label') == [
LABEL_1, LABEL_2
]
assert get_tei_xpath_text_list(tei_auto_root, './/figure/figDesc') == [
TEXT_1, TEXT_2
]
def test_should_segment_figures_if_enabled(
self, test_helper: SingleFileAutoAnnotateEndToEndTestHelper):
target_figure_content_nodes_1 = [
E.label(LABEL_1),
' ',
E.caption(E.p(TEXT_1))
]
target_figure_content_nodes_2 = [
E.label(LABEL_2),
' ',
E.caption(E.p(TEXT_2))
]
target_jats_xml = etree.tostring(
get_target_xml_node(body_nodes=[
E.fig(*target_figure_content_nodes_1),
E.fig(*target_figure_content_nodes_2)
])
)
test_helper.tei_raw_file_path.write_bytes(etree.tostring(
get_training_tei_node([
E.figure(get_nodes_text(
target_figure_content_nodes_1
+ [' ']
+ target_figure_content_nodes_2
))
])
))
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-figures': True,
'fields': 'figure'
}), save_main_session=False)
tei_auto_root = test_helper.get_tei_auto_root()
assert get_text_content_list(get_all_figures(tei_auto_root)) == [
get_nodes_text(target_figure_content_nodes_1),
get_nodes_text(target_figure_content_nodes_2)
]