def iter_semantic_content_for_entity_blocks( # pylint: disable=arguments-differ
self, entity_tokens: Iterable[Tuple[str, LayoutBlock]],
**kwargs) -> Iterable[SemanticContentWrapper]:
entity_tokens = list(entity_tokens)
LOGGER.debug('entity_tokens: %s', entity_tokens)
figure: Optional[SemanticFigure] = None
for name, layout_block in entity_tokens:
if not figure:
figure = SemanticFigure()
semantic_content = self.get_semantic_content_for_entity_name(
name, layout_block=layout_block)
figure.add_content(semantic_content)
if figure:
yield figure
def test_should_add_asset_citation_for_resolved_figure(self):
semantic_document = SemanticDocument()
semantic_document.body_section.add_content(
SemanticSection([
SemanticParagraph([
SemanticTextContentWrapper(
layout_block=LayoutBlock.for_text('See')),
SemanticFigureCitation(
layout_block=LayoutBlock.for_text('Fig 1'),
target_content_id='fig_0')
]),
SemanticFigure([
SemanticLabel(
layout_block=LayoutBlock.for_text('Figure 1'))
],
content_id='fig_0')
]))
tei_document = get_tei_for_semantic_document(semantic_document)
LOGGER.debug('tei xml: %r', etree.tostring(tei_document.root))
assert tei_document.get_xpath_text_content_list(
'//tei:body/tei:div/tei:p') == ['See Fig 1']
assert tei_document.get_xpath_text_content_list(
'//tei:body/tei:div/tei:p/tei:ref[@type="figure"]') == ['Fig 1']
assert tei_document.get_xpath_text_content_list(
'//tei:body/tei:div/tei:p/tei:ref[@type="figure"]/@target') == [
'#fig_0'
]
def test_should_add_section_figures_to_back(self):
semantic_document = SemanticDocument()
semantic_document.back_section.add_content(
SemanticSection([
SemanticFigure([
SemanticLabel(
layout_block=LayoutBlock.for_text('Label 1')),
SemanticCaption(
layout_block=LayoutBlock.for_text('Caption 1'))
],
content_id='fig_0')
]))
tei_document = get_tei_for_semantic_document(semantic_document)
LOGGER.debug('tei xml: %r', etree.tostring(tei_document.root))
figure_xpath = (
'//tei:back/tei:div[@type="annex"]/tei:figure[not(contains(@type, "table"))]'
)
assert tei_document.get_xpath_text_content_list(
f'{figure_xpath}/tei:head') == ['Label 1']
assert tei_document.get_xpath_text_content_list(
f'{figure_xpath}/tei:label') == ['Label 1']
assert tei_document.get_xpath_text_content_list(
f'{figure_xpath}/tei:figDesc') == ['Caption 1']
assert tei_document.get_xpath_text_content_list(
f'{figure_xpath}/@xml:id') == ['fig_0']
assert not tei_document.xpath(
'//tei:back/tei:div[@type="annex"]/tei:div')
def test_should_ignore_layout_graphic_without_related_block(self):
semantic_graphic_1 = SemanticGraphic(layout_graphic=LayoutGraphic(
coordinates=FAR_AWAY_COORDINATES_1, related_block=None))
candidate_semantic_content_1 = SemanticFigure(
[SemanticLabel(layout_block=LayoutBlock.for_text('Figure 1'))])
result = GraphicRelatedBlockTextGraphicMatcher().get_graphic_matches(
semantic_graphic_list=[semantic_graphic_1],
candidate_semantic_content_list=[candidate_semantic_content_1])
LOGGER.debug('result: %r', result)
assert not result.graphic_matches
assert result.unmatched_graphics == [semantic_graphic_1]
def test_should_render_graphic_element(self):
semantic_figure = SemanticFigure([
SemanticLabel(layout_block=LayoutBlock.for_text('Label 1')),
SemanticCaption(layout_block=LayoutBlock.for_text('Caption 1')),
SemanticGraphic(layout_graphic=LayoutGraphic(
local_file_path='image1.png'))
],
content_id='fig_0')
result = _get_wrapped_figure_tei_element(semantic_figure)
assert result.get_xpath_text_content_list(
f'{FIGURE_XPATH}/tei:graphic')
def test_should_ignore_layout_graphic_without_local_path(
self, ocr_model_mock: MagicMock):
ocr_model_mock.predict_single.return_value.get_text.side_effect = RuntimeError
semantic_graphic_1 = SemanticGraphic(layout_graphic=LayoutGraphic(
coordinates=FAR_AWAY_COORDINATES_1, local_file_path=None))
candidate_semantic_content_1 = SemanticFigure(
[SemanticLabel(layout_block=LayoutBlock.for_text('Figure 1'))])
result = OpticalCharacterRecognitionGraphicMatcher(
ocr_model=ocr_model_mock).get_graphic_matches(
semantic_graphic_list=[semantic_graphic_1],
candidate_semantic_content_list=[candidate_semantic_content_1])
LOGGER.debug('result: %r', result)
assert not result.graphic_matches
assert result.unmatched_graphics == [semantic_graphic_1]
def test_should_render_label_description_and_id(self):
semantic_figure = SemanticFigure([
SemanticLabel(layout_block=LayoutBlock.for_text('Label 1')),
SemanticCaption(layout_block=LayoutBlock.for_text('Caption 1'))
],
content_id='fig_0')
result = _get_wrapped_figure_tei_element(semantic_figure)
assert result.get_xpath_text_content_list(
f'{FIGURE_XPATH}/tei:head') == ['Label 1']
assert result.get_xpath_text_content_list(
f'{FIGURE_XPATH}/tei:label') == ['Label 1']
assert result.get_xpath_text_content_list(
f'{FIGURE_XPATH}/tei:figDesc') == ['Caption 1']
assert result.get_xpath_text_content_list(
f'{FIGURE_XPATH}/@xml:id') == ['fig_0']
def test_should_match_based_on_figure_label(self, related_text: str,
figure_label: str,
should_match: bool):
semantic_graphic_1 = SemanticGraphic(layout_graphic=LayoutGraphic(
coordinates=FAR_AWAY_COORDINATES_1,
related_block=LayoutBlock.for_text(related_text)))
candidate_semantic_content_1 = SemanticFigure(
[SemanticLabel(layout_block=LayoutBlock.for_text(figure_label))])
result = GraphicRelatedBlockTextGraphicMatcher().get_graphic_matches(
semantic_graphic_list=[semantic_graphic_1],
candidate_semantic_content_list=[candidate_semantic_content_1])
LOGGER.debug('result: %r', result)
if should_match:
assert len(result) == 1
first_match = result.graphic_matches[0]
assert first_match.semantic_graphic == semantic_graphic_1
else:
assert not result.graphic_matches
assert result.unmatched_graphics == [semantic_graphic_1]
def test_should_match_based_on_figure_label(
self, ocr_model_mock: MagicMock, ocr_text: str, figure_label: str,
should_match: bool, tmp_path: Path):
local_graphic_path = tmp_path / 'image.png'
PIL.Image.new('RGB', (10, 10), (0, 1, 2)).save(local_graphic_path)
ocr_model_mock.predict_single.return_value.get_text.return_value = ocr_text
semantic_graphic_1 = SemanticGraphic(layout_graphic=LayoutGraphic(
coordinates=FAR_AWAY_COORDINATES_1,
local_file_path=str(local_graphic_path)))
candidate_semantic_content_1 = SemanticFigure(
[SemanticLabel(layout_block=LayoutBlock.for_text(figure_label))])
result = OpticalCharacterRecognitionGraphicMatcher(
ocr_model=ocr_model_mock).get_graphic_matches(
semantic_graphic_list=[semantic_graphic_1],
candidate_semantic_content_list=[candidate_semantic_content_1])
LOGGER.debug('result: %r', result)
if should_match:
assert len(result) == 1
first_match = result.graphic_matches[0]
assert first_match.semantic_graphic == semantic_graphic_1
else:
assert not result.graphic_matches
assert result.unmatched_graphics == [semantic_graphic_1]
def _get_semantic_content_for_page_coordinates(
coordinates: LayoutPageCoordinates) -> SemanticContentWrapper:
return SemanticFigure(layout_block=LayoutBlock.for_tokens(
[LayoutToken(text='dummy', coordinates=coordinates)]))