def _apply_styles(old_properties: BaseProperties,
styles: List[BeautifulSoup]):
"""
applies all styles to old_properties
:param old_properties: properties for changing
:param styles: styles in order to apply
"""
# hierarchy of styles: defaults -> paragraph -> numbering -> character
for current_style in styles: # apply styles in reverse order
if current_style.pPr:
change_paragraph_properties(old_properties, current_style.pPr)
if current_style.rPr:
change_run_properties(old_properties, current_style.rPr)
def _get_numbering_formatting(self) -> Optional[Run]:
"""
if the paragraph is a list item applies it's properties to the paragraph
adds numbering run to the list of paragraph runs
:returns: numbering run if there is the text in numbering else None
"""
if self.xml.numPr and self.numbering_extractor:
numbering_run = Run(self, self.styles_extractor)
self.numbering_extractor.parse(self.xml.numPr, self, numbering_run)
if numbering_run.text:
if self.xml.pPr.rPr:
change_run_properties(numbering_run, self.xml.pPr.rPr)
return numbering_run
return None
def parse(self,
xml: BeautifulSoup,
paragraph_properties: BaseProperties,
run_properties: BaseProperties):
"""
parses numPr content and extracts properties for paragraph for given numId and list level
changes old_paragraph properties according to list properties
changes run_properties adding text of numeration and it's properties
:param xml: BeautifulSoup tree with numPr from document.xml or styles.xml (style content)
:param paragraph_properties: Paragraph for changing
:param run_properties: Run for changing
"""
if not xml:
return
ilvl, num_id = xml.ilvl, xml.numId
if not num_id or num_id['w:val'] not in self.num_list:
return
else:
num_id = num_id['w:val']
if not ilvl:
try:
style_id = xml['w:styleId']
num = self.num_list[num_id]
# find link on this styleId in the levels list
for level_num, level in num.levels.items():
if 'styleId' in level and level['styleId'] == style_id:
ilvl = level_num
except KeyError:
return
else:
ilvl = ilvl['w:val']
lvl_info = self.num_list[num_id].get_level_info(ilvl)
text = self._get_list_text(ilvl, num_id)
if lvl_info['styleId']:
self.styles_extractor.parse(lvl_info['styleId'], paragraph_properties, "numbering")
if lvl_info['pPr']:
change_paragraph_properties(paragraph_properties, lvl_info['pPr'])
if lvl_info['rPr']:
change_run_properties(run_properties, lvl_info['rPr'])
change_run_properties(paragraph_properties, lvl_info['rPr'])
run_properties.text = text
paragraph_properties.list_level = self.levels_count
def _make_run_list(self):
"""
makes runs of the paragraph and adds them to the paragraph list
"""
run_list = self.xml.find_all('w:r')
for run_tree in run_list:
new_run = Run(self, self.styles_extractor)
if run_tree.rStyle:
self.styles_extractor.parse(run_tree.rStyle['w:val'], new_run,
"character")
if self.xml.pPr and self.xml.pPr.rPr:
change_run_properties(new_run, self.xml.pPr.rPr)
if run_tree.rPr:
change_run_properties(new_run, run_tree.rPr)
new_run.get_text(run_tree)
if not new_run.text:
continue
if self.runs and self.runs[-1] == new_run:
self.runs[-1].text += new_run.text
else:
self.runs.append(new_run)