Adding a soft carriage return for text belonging to an element list (ca9cf684) · Commits · Centre for Testing and Interoperability / Markdown specifications development / Specification tools

generateBaseline/postprocessing.py

+103 −36

Original line number	Diff line number	Diff line
		@@ -913,20 +913,23 @@ def update_unnumbered_lists(docx_input, docx_output):
		namespaces=ns
		)
		if not abstract_num:
		return (None, None, None)
		return (None, None, None, None)

		# Check the format for this level
		lvl = abstract_num[0].xpath(
		numFmt_elem = abstract_num[0].xpath(
		f'./w:lvl[@w:ilvl="{ilvl}"]/w:numFmt',
		namespaces=ns
		)
		if not lvl:
		return (None, None, None)
		if not numFmt_elem:
		return (None, None, None, None)

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

		lvl_text_elem = abstract_num[0].xpath(f'./w:lvl[@w:ilvl="{ilvl}"]/w:lvlText', namespaces=ns)
		if not lvl_text_elem:
		lvl_text = None

		lvl_text = lvl_text_elem[0].get(f"{{{ns['w']}}}val")
		# Numbered formats
		numbered_formats = ['decimal', 'lowerLetter', 'upperLetter',
		'lowerRoman', 'upperRoman', 'arabic', 'ordinal',
		@@ -951,17 +954,17 @@ def update_unnumbered_lists(docx_input, docx_output):
		unnumbered_formats = ['bullet', 'circle', 'square', 'dash', 'diamond',
		'check', 'arrow', 'arrowhead', 'rtArrow', 'hyphen']

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

		except Exception as e:
		# If any error occurs, return None to fall back to heuristic
		return (None, None, None)
		return (None, None, None, None)
		counter_regular = 0
		counter_b1 = 0
		counter_b2 = 0
		@@ -1041,6 +1044,7 @@ def update_unnumbered_lists(docx_input, docx_output):
		abstract.set(qn('w:val'), abstract_id_bn)

		for para in paragraphs:
		para_removed = False
		# Skip if already processed
		if id(para) in processed_paras:
		continue
		@@ -1056,7 +1060,7 @@ def update_unnumbered_lists(docx_input, docx_output):
		else:
		compact_style = OxmlElement('w:pStyle')
		# Check if it is a numbered list and get the format
		is_numbered, num_format, abstract_num_id = is_numbered_list(para, numbering_root)
		is_numbered, num_format, lvl_text, 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
		#pStyle = OxmlElement('w:pStyle')
		# If format is decimal (numbers), use BN; otherwise use BL
		@@ -1076,6 +1080,8 @@ def update_unnumbered_lists(docx_input, docx_output):
		#numPr.remove(numPr.xpath('./w:numId', namespaces=ns)[0])
		counter_numbered += 1
		else:
		#print(f'The lvl_text is {lvl_text} and the abstract_num_id is {abstract_num_id}')

		if para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="0"]', namespaces=ns):
		if para.xpath('ancestor::w:tbl', namespaces=ns):
		abstract_id_tb1 = get_style_abstract_id(styles_root, numbering_root, "TB1")
		@@ -1085,12 +1091,32 @@ def update_unnumbered_lists(docx_input, docx_output):
		compact_style.set(f"{{{ns['w']}}}val", "TB1")
		counter_table += 1
		else:
		if not lvl_text == " ":
		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")
		else:
		#print(f'Found text belonging to a numbered list item')
		# Take the text from the paragraph
		text = get_para_text(para)
		# Insert it as new run to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None:
		new_run = OxmlElement('w:r')
		new_run.append(OxmlElement('w:br'))
		text_t = OxmlElement('w:t')
		text_t.text = text
		new_run.append(text_t)
		previous_para.append(new_run)
		# Remove the original paragraph
		para.getparent().remove(para)
		para_removed = True
		#print(f'Removed the original paragraph')
		#compact_style.set(f"{{{ns['w']}}}val", "FP")
		#Remove numPr from pPr
		if not para_removed:
		numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		pPr.remove(numPr)
		counter_b1 += 1
		@@ -1103,12 +1129,32 @@ def update_unnumbered_lists(docx_input, docx_output):
		compact_style.set(f"{{{ns['w']}}}val", "TB2")
		counter_table += 1
		else:
		if not lvl_text == " ":
		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")
		else:
		#print(f'Found text belonging to a numbered list item')
		# Take the text from the paragraph
		text = get_para_text(para)
		# Insert it as new run to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None:
		new_run = OxmlElement('w:r')
		new_run.append(OxmlElement('w:br'))
		text_t = OxmlElement('w:t')
		text_t.text = text
		new_run.append(text_t)
		previous_para.append(new_run)
		# Remove the original paragraph
		para.getparent().remove(para)
		para_removed = True
		#print(f'Removed the original paragraph')
		#compact_style.set(f"{{{ns['w']}}}val", "FP")
		#Remove numPr from pPr
		if not para_removed:
		numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		pPr.remove(numPr)
		counter_b2 += 1
		@@ -1121,15 +1167,36 @@ def update_unnumbered_lists(docx_input, docx_output):
		compact_style.set(f"{{{ns['w']}}}val", "TB3")
		counter_table += 1
		else:
		if not lvl_text == " ":
		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")
		else:
		#print(f'Found text belonging to a numbered list item')
		# Take the text from the paragraph
		text = get_para_text(para)
		# Insert it as new run to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None:
		new_run = OxmlElement('w:r')
		new_run.append(OxmlElement('w:br'))
		text_t = OxmlElement('w:t')
		text_t.text = text
		new_run.append(text_t)
		previous_para.append(new_run)
		# Remove the original paragraph
		para.getparent().remove(para)
		para_removed = True
		#print(f'Removed the original paragraph')
		#compact_style.set(f"{{{ns['w']}}}val", "FP")
		#Remove numPr from pPr
		if not para_removed:
		numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		pPr.remove(numPr)
		counter_b3 += 1

		pPr.insert(0, compact_style)
		counter_compact += 1
		continue