def build_non_reg_text(reg_xml):
""" This builds the tree for the non-regulation text such as Appendices
and the Supplement section. """
doc_root = etree.fromstring(reg_xml)
reg_part = int(doc_root.xpath('//REGTEXT')[0].attrib['PART'])
last_section = doc_root.xpath('//REGTEXT/PART/SECTION[last()]')[0]
section_type = None
node_type = None
current_section = None
m_stack = NodeStack()
for child in last_section.getchildren():
if child.tag == 'HD':
p_level = 1
if 'Appendix' in child.text and 'Part' in child.text:
section_type = 'APPENDIX'
node_type = Node.APPENDIX
current_section = headers.parseString(child.text).letter
elif 'Supplement' in child.text and 'Part' in child.text:
section_type = 'SUPPLEMENT'
node_type = Node.INTERP
else:
p_level = 2
if section_type == 'SUPPLEMENT' and 'Appendix' in child.text:
current_section = headers.parseString(child.text).letter
else:
current_section = determine_next_section(m_stack, p_level)
if p_level == 1:
label = [str(reg_part), current_section]
else:
label = [current_section]
n = Node(node_type=node_type, label=label, title=child.text)
last = m_stack.peek()
if len(last) == 0:
m_stack.push_last((p_level, n))
else:
tree_utils.add_to_stack(m_stack, p_level, n)
elif current_section and section_type == 'APPENDIX':
if child.tag == 'EXTRACT':
process_appendix(m_stack, current_section, child)
tree_utils.unwind_stack(m_stack)
elif current_section and section_type == 'SUPPLEMENT':
if child.tag == 'EXTRACT':
process_supplement(reg_part, m_stack, child)
tree_utils.unwind_stack(m_stack)
while m_stack.size() > 1:
tree_utils.unwind_stack(m_stack)
sections = []
for level, section in m_stack.m_stack[0]:
sections.append(section)
return sections
def set_letter(self, appendix):
"""Find (and set) the appendix letter"""
for node in (c for c in appendix.getchildren()
if is_appendix_header(c)):
text = tree_utils.get_node_text(node)
if self.appendix_letter:
logger.warning("Found two appendix headers: %s and %s",
self.appendix_letter, text)
self.appendix_letter = headers.parseString(text).appendix
return self.appendix_letter
def set_letter(self, appendix):
"""Find (and set) the appendix letter"""
for node in (c for c in appendix.getchildren()
if is_appendix_header(c)):
text = tree_utils.get_node_text(node)
if self.appendix_letter:
logging.warning("Found two appendix headers: %s and %s",
self.appendix_letter, text)
self.appendix_letter = headers.parseString(text).appendix
return self.appendix_letter