Fix for restarting numbering in lists (6d9bdea1) · Commits · Centre for Testing and Interoperability / Markdown specifications development / Specification tools

generateBaseline/postprocessing.py

+78 −14

Original line number	Original line	Diff line number	Diff line
	@@ -852,6 +852,13 @@ def update_unnumbered_lists(docx_input, docx_output):
	except KeyError:		except KeyError:
	numbering_root = None		numbering_root = None

			# Try to read styles.xml, if it doesn't exist, styles_root will be None
			try:
			styles_data = zin.read("word/styles.xml")
			styles_root = etree.fromstring(styles_data)
			except KeyError:
			styles_root = None

	root = etree.fromstring(xml_data)		root = etree.fromstring(xml_data)

	def is_numbered_list(para, numbering_root):		def is_numbered_list(para, numbering_root):
	@@ -888,16 +895,16 @@ def update_unnumbered_lists(docx_input, docx_output):
	namespaces=ns		namespaces=ns
	)		)
	if not num_elem:		if not num_elem:
	return (None, None)		return (None, None, None)

	# Get the abstractNumId		# Get the abstractNumId
	abstract_num_id_elem = num_elem[0].xpath('./w:abstractNumId', namespaces=ns)		abstract_num_id_elem = num_elem[0].xpath('./w:abstractNumId', namespaces=ns)
	if not abstract_num_id_elem:		if not abstract_num_id_elem:
	return (None, None)		return (None, None, None)

	abstract_num_id = abstract_num_id_elem[0].get(f"{{{ns['w']}}}val")		abstract_num_id = abstract_num_id_elem[0].get(f"{{{ns['w']}}}val")
	if abstract_num_id is None:		if abstract_num_id is None:
	return (None, None)		return (None, None, None)

	# Find the abstractNum		# Find the abstractNum
	abstract_num = numbering_root.xpath(		abstract_num = numbering_root.xpath(
	@@ -905,7 +912,7 @@ def update_unnumbered_lists(docx_input, docx_output):
	namespaces=ns		namespaces=ns
	)		)
	if not abstract_num:		if not abstract_num:
	return (None, None)		return (None, None, None)

	# Check the format for this level		# Check the format for this level
	lvl = abstract_num[0].xpath(		lvl = abstract_num[0].xpath(
	@@ -913,11 +920,11 @@ def update_unnumbered_lists(docx_input, docx_output):
	namespaces=ns		namespaces=ns
	)		)
	if not lvl:		if not lvl:
	return (None, None)		return (None, None, None)

	num_fmt = lvl[0].get(f"{{{ns['w']}}}val")		num_fmt = lvl[0].get(f"{{{ns['w']}}}val")
	if num_fmt is None:		if num_fmt is None:
	return (None, None)		return (None, None, None)

	# Numbered formats		# Numbered formats
	numbered_formats = ['decimal', 'lowerLetter', 'upperLetter',		numbered_formats = ['decimal', 'lowerLetter', 'upperLetter',
	@@ -944,16 +951,16 @@ def update_unnumbered_lists(docx_input, docx_output):
	'check', 'arrow', 'arrowhead', 'rtArrow', 'hyphen']		'check', 'arrow', 'arrowhead', 'rtArrow', 'hyphen']

	if num_fmt in numbered_formats:		if num_fmt in numbered_formats:
	return (True, num_fmt) # Return (True, format string)		return (True, num_fmt, abstract_num_id) # Return (True, format string)
	elif num_fmt in unnumbered_formats:		elif num_fmt in unnumbered_formats:
	return (False, num_fmt) # Return (False, format string)		return (False, num_fmt, abstract_num_id) # Return (False, format string)
	else:		else:
	# Unknown format, default to unnumbered		# Unknown format, default to unnumbered
	return (False, num_fmt)		return (False, num_fmt, abstract_num_id)

	except Exception as e:		except Exception as e:
	# If any error occurs, return None to fall back to heuristic		# If any error occurs, return None to fall back to heuristic
	return (None, None)		return (None, None, None)
	counter_regular = 0		counter_regular = 0
	counter_b1 = 0		counter_b1 = 0
	counter_b2 = 0		counter_b2 = 0
	@@ -1001,6 +1008,33 @@ def update_unnumbered_lists(docx_input, docx_output):
	text += text_elem.text		text += text_elem.text
	return text		return text

			def get_style_abstract_id(styles_root, numbering_root, style_id):
			"""
			Find the abstractNumId used by the given style.
			"""
			styles = styles_root
			style = styles.xpath(f'//w:style[@w:styleId="{style_id}"]', namespaces=ns)[0]

			num_id_el = style.xpath('.//w:numId', namespaces=ns)
			if not num_id_el:
			return None
			#raise ValueError(f"Style {style_id} has no numId")

			# Remove numPr from style if it exists
			num_pr = style.xpath('.//w:numPr', namespaces=ns)
			if num_pr:
			for el in num_pr:
			el.getparent().remove(el)

			return num_id_el[0].get(qn('w:val'))

			def update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_bn):

			for num in numbering_root.xpath(f'//w:num[@w:abstractNumId="{abstract_num_id}"]', namespaces=ns):
			abstract = num.xpath('./w:abstractNumId', namespaces=ns)[0]
			old_id = abstract.get(qn('w:val'))
			abstract.set(qn('w:val'), abstract_id_bn)

	for para in paragraphs:		for para in paragraphs:
	# Skip if already processed		# Skip if already processed
	if id(para) in processed_paras:		if id(para) in processed_paras:
	@@ -1017,25 +1051,39 @@ def update_unnumbered_lists(docx_input, docx_output):
	else:		else:
	compact_style = OxmlElement('w:pStyle')		compact_style = OxmlElement('w:pStyle')
	# Check if it is a numbered list and get the format		# Check if it is a numbered list and get the format
	is_numbered, num_format = is_numbered_list(para, numbering_root)		is_numbered, num_format, abstract_num_id = is_numbered_list(para, numbering_root)
	if is_numbered: # is_numbered is True if numbered, False if unnumbered, None if cannot determine		if is_numbered: # is_numbered is True if numbered, False if unnumbered, None if cannot determine
	#pStyle = OxmlElement('w:pStyle')		#pStyle = OxmlElement('w:pStyle')
	# If format is decimal (numbers), use BN; otherwise use BL		# If format is decimal (numbers), use BN; otherwise use BL
	if num_format == 'decimal':		if num_format == 'decimal':
			abstract_id_bn = get_style_abstract_id(styles_root, numbering_root, "BN")
			if abstract_id_bn is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_bn)
	compact_style.set(f"{{{ns['w']}}}val", "BN")		compact_style.set(f"{{{ns['w']}}}val", "BN")
	else:		else:
			abstract_id_bl = get_style_abstract_id(styles_root, numbering_root, "BL")
			if abstract_id_bl is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_bl)
	compact_style.set(f"{{{ns['w']}}}val", "BL")		compact_style.set(f"{{{ns['w']}}}val", "BL")
	pPr.insert(0, compact_style)		pPr.insert(0, compact_style)
	# Remove numId from numPr		# Remove numId from numPr
	numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]		#numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
	numPr.remove(numPr.xpath('./w:numId', namespaces=ns)[0])		#numPr.remove(numPr.xpath('./w:numId', namespaces=ns)[0])
	counter_numbered += 1		counter_numbered += 1
	else:		else:
	if para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="0"]', namespaces=ns):		if para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="0"]', namespaces=ns):
	if para.xpath('ancestor::w:tbl', namespaces=ns):		if para.xpath('ancestor::w:tbl', namespaces=ns):
			abstract_id_tb1 = get_style_abstract_id(styles_root, numbering_root, "TB1")
			if abstract_id_tb1 is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_tb1)

	compact_style.set(f"{{{ns['w']}}}val", "TB1")		compact_style.set(f"{{{ns['w']}}}val", "TB1")
	counter_table += 1		counter_table += 1
	else:		else:
			abstract_id_b1 = get_style_abstract_id(styles_root, numbering_root, "B1")
			if abstract_id_b1 is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_b1)

	compact_style.set(f"{{{ns['w']}}}val", "B1")		compact_style.set(f"{{{ns['w']}}}val", "B1")
	#Remove numPr from pPr		#Remove numPr from pPr
	numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]		numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
	@@ -1043,9 +1091,17 @@ def update_unnumbered_lists(docx_input, docx_output):
	counter_b1 += 1		counter_b1 += 1
	elif para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="1"]', namespaces=ns):		elif para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="1"]', namespaces=ns):
	if para.xpath('ancestor::w:tbl', namespaces=ns):		if para.xpath('ancestor::w:tbl', namespaces=ns):
			abstract_id_tb2 = get_style_abstract_id(styles_root, numbering_root, "TB2")
			if abstract_id_tb2 is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_tb2)

	compact_style.set(f"{{{ns['w']}}}val", "TB2")		compact_style.set(f"{{{ns['w']}}}val", "TB2")
	counter_table += 1		counter_table += 1
	else:		else:
			abstract_id_b2 = get_style_abstract_id(styles_root, numbering_root, "B2")
			if abstract_id_b2 is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_b2)

	compact_style.set(f"{{{ns['w']}}}val", "B2")		compact_style.set(f"{{{ns['w']}}}val", "B2")
	#Remove numPr from pPr		#Remove numPr from pPr
	numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]		numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
	@@ -1053,9 +1109,17 @@ def update_unnumbered_lists(docx_input, docx_output):
	counter_b2 += 1		counter_b2 += 1
	elif para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="2"]', namespaces=ns):		elif para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="2"]', namespaces=ns):
	if para.xpath('ancestor::w:tbl', namespaces=ns):		if para.xpath('ancestor::w:tbl', namespaces=ns):
			abstract_id_tb3 = get_style_abstract_id(styles_root, numbering_root, "TB3")
			if abstract_id_tb3 is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_tb3)

	compact_style.set(f"{{{ns['w']}}}val", "TB3")		compact_style.set(f"{{{ns['w']}}}val", "TB3")
	counter_table += 1		counter_table += 1
	else:		else:
			abstract_id_b3 = get_style_abstract_id(styles_root, numbering_root, "B3")
			if abstract_id_b3 is not None and abstract_num_id is not None:
			update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_b3)

	compact_style.set(f"{{{ns['w']}}}val", "B3")		compact_style.set(f"{{{ns['w']}}}val", "B3")
	#Remove numPr from pPr		#Remove numPr from pPr
	numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]		numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]

Admin message