def test_two_raw_text(self):
h1 = HierarchyLevel.create_raw_text()
h2 = HierarchyLevel.create_raw_text()
h3 = HierarchyLevel(level_1=1, level_2=2, can_be_multiline=False, paragraph_type="raw_text")
self.assertTrue(h1 == h2)
self.assertTrue(h1 >= h2)
self.assertTrue(h1 <= h2)
self.assertTrue(h1 == h3)
self.assertTrue(h1 >= h3)
self.assertTrue(h1 <= h3)
def test_raw_text_greater_than_any_other(self):
list_item = HierarchyLevel(level_1=2, level_2=1, can_be_multiline=False, paragraph_type="list_item")
raw_text = HierarchyLevel.create_raw_text()
self.assertFalse(list_item > raw_text)
self.assertFalse(list_item >= raw_text)
self.assertFalse(list_item == raw_text)
self.assertTrue(list_item < raw_text)
self.assertTrue(list_item <= raw_text)
def _get_lines_with_meta(
self, hierarchy_level_extractor: HierarchyLevelExtractor
) -> List[LineWithMeta]:
"""
:param paragraph_list: list of Paragraph
:return: list of LineWithMeta
"""
lines_with_meta = []
paragraph_id = 0
for i, paragraph in enumerate(self.paragraph_list):
# line with meta:
# {"text": "",
# "type": ""("paragraph" ,"list_item", "raw_text", "style_header"),
# "level": (1,1) or None (hierarchy_level),
# "annotations": [["size", start, end, size], ["bold", start, end, "True"], ...]}
paragraph_properties = ParagraphInfo(paragraph)
line_with_meta = paragraph_properties.get_info()
text = line_with_meta["text"]
paragraph_type = line_with_meta["type"]
level = line_with_meta["level"]
if level:
hierarchy_level = HierarchyLevel(level[0], level[1], False,
paragraph_type)
else:
hierarchy_level = HierarchyLevel.create_raw_text()
dict2annotations = {
"bold": BoldAnnotation,
"italic": ItalicAnnotation,
"underlined": UnderlinedAnnotation,
"size": SizeAnnotation,
"indentation": IndentationAnnotation,
"alignment": AlignmentAnnotation,
"style": StyleAnnotation,
}
annotations = []
for annotation in line_with_meta["annotations"]:
annotations.append(
dict2annotations[annotation[0]](*annotation[1:]))
for object_dict in [self.image_refs, self.diagram_refs]:
if i in object_dict:
for object_uid in object_dict[i]:
annotation = AttachAnnotation(attach_uid=object_uid,
start=0,
end=len(text))
annotations.append(annotation)
if i in self.table_refs:
for table_uid in self.table_refs[i]:
annotation = TableAnnotation(name=table_uid,
start=0,
end=len(text))
annotations.append(annotation)
paragraph_id += 1
metadata = ParagraphMetadata(paragraph_type=paragraph_type,
predicted_classes=None,
page_id=0,
line_id=paragraph_id)
lines_with_meta.append(
LineWithMeta(line=text,
hierarchy_level=hierarchy_level,
metadata=metadata,
annotations=annotations,
uid=paragraph.uid))
lines_with_meta = hierarchy_level_extractor.get_hierarchy_level(
lines_with_meta)
return lines_with_meta
