def assertUnlabeledFormFieldDictTransformCorrect(self,
form_fields,
generated_fields,
read_results=None):
if generated_fields is None:
return
for idx, expected in enumerate(generated_fields):
self.assertEqual(adjust_confidence(expected.confidence),
form_fields["field-" + str(idx)].confidence)
self.assertEqual(expected.key.text,
form_fields["field-" + str(idx)].label_data.text)
self.assertBoundingBoxTransformCorrect(
form_fields["field-" + str(idx)].label_data.bounding_box,
expected.key.bounding_box)
if read_results:
self.assertFieldElementsTransFormCorrect(
form_fields["field-" + str(idx)].label_data.field_elements,
expected.key.elements, read_results)
self.assertEqual(expected.value.text,
form_fields["field-" + str(idx)].value_data.text)
self.assertBoundingBoxTransformCorrect(
form_fields["field-" + str(idx)].value_data.bounding_box,
expected.value.bounding_box)
if read_results:
self.assertFieldElementsTransFormCorrect(
form_fields["field-" + str(idx)].value_data.field_elements,
expected.value.elements, read_results)
def assertTablesTransformCorrect(self,
layout,
expected_layout,
read_results=None,
**kwargs):
for table, expected_table in zip(layout, expected_layout):
self.assertEqual(table.row_count, expected_table.rows)
self.assertEqual(table.column_count, expected_table.columns)
self.assertBoundingBoxTransformCorrect(table.bounding_box,
expected_table.bounding_box)
for cell, expected_cell in zip(table.cells, expected_table.cells):
self.assertEqual(table.page_number, cell.page_number)
self.assertEqual(cell.text, expected_cell.text)
self.assertEqual(cell.row_index, expected_cell.row_index)
self.assertEqual(cell.column_index, expected_cell.column_index)
self.assertEqual(
cell.row_span, expected_cell.row_span
if expected_cell.row_span is not None else 1)
self.assertEqual(
cell.column_span, expected_cell.column_span
if expected_cell.column_span is not None else 1)
self.assertEqual(cell.confidence,
adjust_confidence(expected_cell.confidence))
self.assertEqual(
cell.is_header, expected_cell.is_header
if expected_cell.is_header is not None else False)
self.assertEqual(
cell.is_footer, expected_cell.is_footer
if expected_cell.is_footer is not None else False)
self.assertBoundingBoxTransformCorrect(
cell.bounding_box, expected_cell.bounding_box)
self.assertFieldElementsTransFormCorrect(
cell.field_elements, expected_cell.elements, read_results)
def assertFormWordTransformCorrect(self, word, expected):
self.assertEqual(word.text, expected.text)
self.assertEqual(word.confidence,
adjust_confidence(expected.confidence))
self.assertEqual(word.kind, "word")
self.assertBoundingBoxTransformCorrect(word.bounding_box,
expected.bounding_box)
def assertTablesTransformCorrect(self,
layout,
expected_layout,
read_results=None,
**kwargs):
for table, expected_table in zip(layout, expected_layout):
assert table.row_count == expected_table.rows
assert table.column_count == expected_table.columns
self.assertBoundingBoxTransformCorrect(table.bounding_box,
expected_table.bounding_box)
for cell, expected_cell in zip(table.cells, expected_table.cells):
assert table.page_number == cell.page_number
assert cell.text == expected_cell.text
assert cell.row_index == expected_cell.row_index
assert cell.column_index == expected_cell.column_index
assert cell.row_span == expected_cell.row_span if expected_cell.row_span is not None else 1
assert cell.column_span == expected_cell.column_span if expected_cell.column_span is not None else 1
assert cell.confidence == adjust_confidence(
expected_cell.confidence)
assert cell.is_header == (expected_cell.is_header
if expected_cell.is_header
is not None else False)
assert cell.is_footer == (expected_cell.is_footer
if expected_cell.is_footer
is not None else False)
self.assertBoundingBoxTransformCorrect(
cell.bounding_box, expected_cell.bounding_box)
self.assertFieldElementsTransFormCorrect(
cell.field_elements, expected_cell.elements, read_results)
def assertFormSelectionMarkTransformCorrect(self, selection_mark,
expected):
self.assertEqual(selection_mark.kind, "selectionMark")
self.assertEqual(selection_mark.confidence,
adjust_confidence(expected.confidence))
self.assertEqual(selection_mark.state, expected.state)
self.assertBoundingBoxTransformCorrect(selection_mark.bounding_box,
expected.bounding_box)
def assertDocumentSelectionMarkTransformCorrect(self, selection_mark,
expected):
assert selection_mark.kind == "selectionMark"
assert selection_mark.confidence == adjust_confidence(
expected.confidence)
assert selection_mark.state == expected.state
self.assertBoundingBoxTransformCorrect(selection_mark.bounding_box,
expected.bounding_box)
def assertFormFieldsTransformCorrect(self, form_fields, generated_fields, read_results=None):
if generated_fields is None:
return
for label, expected in generated_fields.items():
field_type = expected.type
self.assertEqual(adjust_value_type(field_type), form_fields[label].value_type)
self.assertEqual(label, form_fields[label].name)
self.assertEqual(adjust_confidence(expected.confidence), form_fields[label].confidence)
self.assertFormFieldValueTransformCorrect(form_fields[label], expected, read_results)
def assertFormFieldsTransformCorrect(self, form_fields, generated_fields, read_results=None):
if generated_fields is None:
return
for label, expected in generated_fields.items():
if expected is None: # None value occurs with labeled tables and empty cells
continue
field_type = expected.type
self.assertEqual(adjust_value_type(field_type), form_fields[label].value_type)
self.assertEqual(label, form_fields[label].name)
self.assertEqual(adjust_confidence(expected.confidence), form_fields[label].confidence)
self.assertFormFieldValueTransformCorrect(form_fields[label], expected, read_results)
def assertFormWordTransformCorrect(self, word, expected):
assert word.text == expected.text
assert word.confidence == adjust_confidence(expected.confidence)
assert word.kind == "word"
self.assertBoundingBoxTransformCorrect(word.bounding_box,
expected.bounding_box)
