Try fix lists (74aefcd5) · Commits · Centre for Testing and Interoperability / Markdown specifications development / Specification tools

generateBaseline/postprocessing.py

+112 −34

Original line number	Diff line number	Diff line
		@@ -1031,6 +1031,8 @@ def update_lists(docx_input, docx_output):
		counter_b1 = 0
		counter_b2 = 0
		counter_b3 = 0
		counter_b4 = 0
		counter_b5 = 0
		counter_compact = 0
		counter_table = 0
		counter_numbered = 0
		@@ -1162,10 +1164,9 @@ def update_lists(docx_input, docx_output):
		else:
		# Take all elements except pPr (usually only runs and hyperlinks) in this paragraph and append them to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None:
		if previous_para is not None and previous_para.tag == f"{{{ns['w']}}}p":
		if (previous_para.xpath('./w:pPr/w:pStyle[@w:val="SourceCode"]', namespaces=ns)):
		compact_style.set(f"{{{ns['w']}}}val", "B1")
		#continue
		else:
		break_run = OxmlElement('w:r')
		break_run.append(OxmlElement('w:br'))
		@@ -1173,11 +1174,12 @@ def update_lists(docx_input, docx_output):
		for element in para.xpath('./*', namespaces=ns):
		if element.tag != f"{{{ns['w']}}}pPr":
		previous_para.append(element)
		# 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")
		else:
		compact_style.set(f"{{{ns['w']}}}val", "B1")
		#Remove numPr from pPr
		#if not para_removed:
		# numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		@@ -1201,23 +1203,22 @@ def update_lists(docx_input, docx_output):
		else:
		# Take all elements except pPr (usually only runs and hyperlinks) in this paragraph and append them to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None:
		if previous_para is not None and previous_para.tag == f"{{{ns['w']}}}p":
		if (previous_para.xpath('./w:pPr/w:pStyle[@w:val="SourceCode"]', namespaces=ns)):
		compact_style.set(f"{{{ns['w']}}}val", "B2")
		#ontinue
		else:
		if previous_para is not None:
		break_run = OxmlElement('w:r')
		break_run.append(OxmlElement('w:br'))
		previous_para.append(break_run)
		for element in para.xpath('./*', namespaces=ns):
		if element.tag != f"{{{ns['w']}}}pPr":
		previous_para.append(element)
		# 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")
		else:
		compact_style.set(f"{{{ns['w']}}}val", "B2")
		#Remove numPr from pPr
		#if not para_removed:
		# numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		@@ -1241,28 +1242,105 @@ def update_lists(docx_input, docx_output):
		else:
		# Take all elements except pPr (usually only runs and hyperlinks) in this paragraph and append them to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None:
		if previous_para is not None and previous_para.tag == f"{{{ns['w']}}}p":
		if (previous_para.xpath('./w:pPr/w:pStyle[@w:val="SourceCode"]', namespaces=ns)):
		compact_style.set(f"{{{ns['w']}}}val", "B3")
		#continue
		else:
		if previous_para is not None:
		break_run = OxmlElement('w:r')
		break_run.append(OxmlElement('w:br'))
		previous_para.append(break_run)
		for element in para.xpath('./*', namespaces=ns):
		if element.tag != f"{{{ns['w']}}}pPr":
		previous_para.append(element)
		# 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")
		else:
		compact_style.set(f"{{{ns['w']}}}val", "B3")
		#Remove numPr from pPr
		#if not para_removed:
		# numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		# pPr.remove(numPr)
		counter_b3 += 1
		elif para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="3"]', namespaces=ns):
		if para.xpath('ancestor::w:tbl', namespaces=ns):
		abstract_id_tb4 = get_style_abstract_id(styles_root, numbering_root, "TB4")
		if abstract_id_tb4 is not None and abstract_num_id is not None:
		update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_tb4)

		compact_style.set(f"{{{ns['w']}}}val", "TB4")
		counter_table += 1
		else:
		if not lvl_text == " ":
		abstract_id_b4 = get_style_abstract_id(styles_root, numbering_root, "B4")
		if abstract_id_b4 is not None and abstract_num_id is not None:
		update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_b4)

		compact_style.set(f"{{{ns['w']}}}val", "B4")
		else:
		# Take all elements except pPr (usually only runs and hyperlinks) in this paragraph and append them to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None and previous_para.tag == f"{{{ns['w']}}}p":
		if (previous_para.xpath('./w:pPr/w:pStyle[@w:val="SourceCode"]', namespaces=ns)):
		compact_style.set(f"{{{ns['w']}}}val", "B4")
		else:
		break_run = OxmlElement('w:r')
		break_run.append(OxmlElement('w:br'))
		previous_para.append(break_run)
		for element in para.xpath('./*', namespaces=ns):
		if element.tag != f"{{{ns['w']}}}pPr":
		previous_para.append(element)
		para.getparent().remove(para)
		para_removed = True
		#print(f'Removed the original paragraph')
		#compact_style.set(f"{{{ns['w']}}}val", "FP")
		else:
		compact_style.set(f"{{{ns['w']}}}val", "B4")
		#Remove numPr from pPr
		#if not para_removed:
		# numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		# pPr.remove(numPr)
		counter_b4 += 1
		elif para.xpath('./w:pPr/w:numPr/w:ilvl[@w:val="4"]', namespaces=ns):
		if para.xpath('ancestor::w:tbl', namespaces=ns):
		abstract_id_tb5 = get_style_abstract_id(styles_root, numbering_root, "TB5")
		if abstract_id_tb5 is not None and abstract_num_id is not None:
		update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_tb5)

		compact_style.set(f"{{{ns['w']}}}val", "TB5")
		counter_table += 1
		else:
		if not lvl_text == " ":
		abstract_id_b5 = get_style_abstract_id(styles_root, numbering_root, "B5")
		if abstract_id_b5 is not None and abstract_num_id is not None:
		update_list_instance_to_style(numbering_root, abstract_num_id, abstract_id_b5)

		compact_style.set(f"{{{ns['w']}}}val", "B5")
		else:
		# Take all elements except pPr (usually only runs and hyperlinks) in this paragraph and append them to the previous paragraph. New run starting by <w:br/>
		previous_para = para.getprevious()
		if previous_para is not None and previous_para.tag == f"{{{ns['w']}}}p":
		if (previous_para.xpath('./w:pPr/w:pStyle[@w:val="SourceCode"]', namespaces=ns)):
		compact_style.set(f"{{{ns['w']}}}val", "B5")
		else:
		break_run = OxmlElement('w:r')
		break_run.append(OxmlElement('w:br'))
		previous_para.append(break_run)
		for element in para.xpath('./*', namespaces=ns):
		if element.tag != f"{{{ns['w']}}}pPr":
		previous_para.append(element)
		para.getparent().remove(para)
		para_removed = True
		#print(f'Removed the original paragraph')
		#compact_style.set(f"{{{ns['w']}}}val", "FP")
		else:
		compact_style.set(f"{{{ns['w']}}}val", "B5")
		#Remove numPr from pPr
		#if not para_removed:
		# numPr = pPr.xpath('./w:numPr', namespaces=ns)[0]
		# pPr.remove(numPr)
		counter_b5 += 1

		pPr.insert(0, compact_style)
		counter_compact += 1
		@@ -1436,7 +1514,7 @@ def update_lists(docx_input, docx_output):
		processed_paras.add(id(para))
		parent.remove(para)"""

		print(f'Updated {counter_b1} B1 style paragraphs, {counter_b2} B2 style paragraphs, {counter_b3} B3 style paragraphs')
		print(f'Updated {counter_b1} B1 style paragraphs, {counter_b2} B2 style paragraphs, {counter_b3} B3 style paragraphs, {counter_b4} B4 style paragraphs, {counter_b5} B5 style paragraphs')
		print(f'Updated {counter_compact} Compact style paragraphs, {counter_table} unnumbered list items in tables, {counter_regular} outside tables')
		print(f'Updated {counter_numbered} numbered list items')