def test_find_all(self):
root = make_xml(b"<one><two><three/><three/></two></one>")
# Can't find the root element
result = find_all(root, "one")
self.assertEqual(result, [])
result = find_all(root, "three")
expected = ["three", "three"]
self.assertEqual(list(elements_to_tags(result)), expected)
result = find_all(root, "two")
self.assertEqual(list(elements_to_tags(result)), ["two"])
def test_find_all(self):
root = make_xml(b'<one><two><three/><three/></two></one>')
# Can't find the root element
result = find_all(root, 'one')
self.assertEqual(result, [])
result = find_all(root, 'three')
expected = ['three', 'three']
self.assertEqual(list(elements_to_tags(result)), expected)
result = find_all(root, 'two')
self.assertEqual(list(elements_to_tags(result)), ['two'])
def _get_rowspan(self, el, v_merge):
current_row = self.pre_processor.row_index(el)
current_col = self.pre_processor.column_index(el)
rowspan = 1
result = ''
tbl = find_ancestor_with_tag(self.pre_processor, el, 'tbl')
# We only want table cells that have a higher row_index that is greater
# than the current_row and that are on the current_col
if tbl is None:
return ''
tcs = [
tc for tc in find_all(tbl, 'tc')
if self.pre_processor.row_index(tc) >= current_row and
self.pre_processor.column_index(tc) == current_col
]
restart_in_v_merge = False
if v_merge is not None and 'val' in v_merge.attrib:
restart_in_v_merge = 'restart' in v_merge.attrib['val']
def increment_rowspan(tc):
if not restart_in_v_merge:
return False
if not self.pre_processor.vmerge_continue(tc):
return False
return True
for tc in tcs:
if increment_rowspan(tc):
rowspan += 1
else:
rowspan = 1
if rowspan > 1:
result = rowspan
return str(result)
def test_get_image_sizes(self):
parser = XMLDocx2Html(
document_xml=self.get_xml(),
relationships=self.relationships,
)
tree = parse_xml_from_string(self.get_xml())
els = []
els.extend(find_all(tree, 'drawing'))
els.extend(find_all(tree, 'pict'))
image_ids = []
for el in els:
image_ids.append(parser._get_image_size(el))
expected = [
('40px', '20px'),
('41pt', '21pt'),
]
self.assertEqual(
set(image_ids),
set(expected),
)
def test_get_image_sizes(self):
parser = XMLDocx2Html(
document_xml=self.get_xml(),
rels_dict=self.relationship_dict,
)
tree = parse_xml_from_string(self.get_xml())
els = []
els.extend(find_all(tree, 'drawing'))
els.extend(find_all(tree, 'pict'))
image_ids = []
for el in els:
image_ids.append(parser._get_image_size(el))
expected = [
('40px', '20px'),
('41pt', '21pt'),
]
self.assertEqual(
set(image_ids),
set(expected),
)
def test_get_image_id(self):
parser = XMLDocx2Html(
document_xml=self.get_xml(),
rels_dict=self.relationship_dict,
)
tree = parse_xml_from_string(self.get_xml())
els = []
els.extend(find_all(tree, 'drawing'))
els.extend(find_all(tree, 'pict'))
image_ids = []
for el in els:
image_ids.append(parser._get_image_id(el))
expected = [
'rId0',
'rId1',
]
self.assertEqual(
set(image_ids),
set(expected),
)
def test_get_image_id(self):
parser = XMLDocx2Html(
document_xml=self.get_xml(),
relationships=self.relationships,
)
tree = parse_xml_from_string(self.get_xml())
els = []
els.extend(find_all(tree, 'drawing'))
els.extend(find_all(tree, 'pict'))
image_ids = []
for el in els:
image_ids.append(parser._get_image_id(el))
expected = [
'rId0',
'rId1',
]
self.assertEqual(
set(image_ids),
set(expected),
)
def _parse_styles(self):
if self.styles_text is None:
return {}
tree = parse_xml_from_string(self.styles_text)
styles_dict = {}
for style in find_all(tree, "style"):
style_val = find_first(style, "name").attrib["val"]
run_properties = find_first(style, "rPr")
styles_dict[style.attrib["styleId"]] = {
"style_name": style_val,
"default_run_properties": self._parse_run_properties(run_properties),
}
return styles_dict
def _parse_styles(self):
if self.styles_text is None:
return {}
tree = parse_xml_from_string(self.styles_text)
styles_dict = {}
for style in find_all(tree, 'style'):
style_val = find_first(style, 'name').attrib['val']
run_properties = find_first(style, 'rPr')
styles_dict[style.attrib['styleId']] = {
'style_name': style_val,
'default_run_properties': self._parse_run_properties(
run_properties,
),
}
return styles_dict
def _parse_styles(self):
if self.styles_text is None:
return {}
tree = parse_xml_from_string(self.styles_text)
styles_dict = {}
for style in find_all(tree, 'style'):
style_val = find_first(style, 'name').attrib['val']
run_properties = find_first(style, 'rPr')
styles_dict[style.attrib['styleId']] = {
'style_name':
style_val,
'default_run_properties':
self._parse_run_properties(run_properties, ),
}
return styles_dict
def _parse_styles(self):
styles_part = self.document.main_document_part.style_definitions_part
if not styles_part:
return {}
styles_root = styles_part.root_element
styles_dict = {}
for style in find_all(styles_root, 'style'):
style_val = find_first(style, 'name').attrib['val']
run_properties = find_first(style, 'rPr')
styles_dict[style.attrib['styleId']] = {
'style_name':
style_val,
'default_run_properties':
self._parse_run_properties(run_properties, ),
}
return styles_dict
def _parse_styles(self):
styles_part = self.document.main_document_part.style_definitions_part
if not styles_part:
return {}
styles_root = styles_part.root_element
styles_dict = {}
for style in find_all(styles_root, 'style'):
style_val = find_first(style, 'name').attrib['val']
run_properties = find_first(style, 'rPr')
styles_dict[style.attrib['styleId']] = {
'style_name': style_val,
'default_run_properties': self._parse_run_properties(
run_properties,
),
}
return styles_dict
