def build_tree(reg_xml):
doc = etree.fromstring(reg_xml)
preprocess_xml(doc)
reg_part = get_reg_part(doc)
title = get_title(doc)
tree = Node("", [], [reg_part], title)
part = doc.xpath('//PART')[0]
subpart_xmls = [c for c in part.getchildren() if c.tag == 'SUBPART']
if len(subpart_xmls) > 0:
subparts = [build_subpart(reg_part, s) for s in subpart_xmls]
tree.children = subparts
else:
section_xmls = [c for c in part.getchildren() if c.tag == 'SECTION']
sections = []
for section_xml in section_xmls:
sections.extend(build_from_section(reg_part, section_xml))
empty_part = reg_text.build_empty_part(reg_part)
empty_part.children = sections
tree.children = [empty_part]
non_reg_sections = build_non_reg_text(doc, reg_part)
tree.children += non_reg_sections
return tree
def build_tree(reg_xml):
doc = etree.fromstring(reg_xml)
preprocess_xml(doc)
reg_part = get_reg_part(doc)
title = get_title(doc)
tree = Node("", [], [reg_part], title)
part = doc.xpath('//PART')[0]
subpart_xmls = [c for c in part.getchildren() if c.tag == 'SUBPART']
if len(subpart_xmls) > 0:
subparts = [build_subpart(reg_part, s) for s in subpart_xmls]
tree.children = subparts
else:
section_xmls = [c for c in part.getchildren() if c.tag == 'SECTION']
sections = []
for section_xml in section_xmls:
sections.extend(build_from_section(reg_part, section_xml))
empty_part = reg_text.build_empty_part(reg_part)
empty_part.children = sections
tree.children = [empty_part]
non_reg_sections = build_non_reg_text(doc, reg_part)
tree.children += non_reg_sections
return tree
def build_tree(reg_xml):
logger.info("Build tree %s", reg_xml)
preprocess_xml(reg_xml)
reg_part = get_reg_part(reg_xml)
title = get_title(reg_xml)
tree = Node("", [], [reg_part], title)
part = reg_xml.xpath('//PART')[0]
# Build a list of SUBPARTs, then pull SUBJGRPs into that list:
subpart_and_subjgrp_xmls = []
for subpart in part.xpath('./SUBPART|./SUBJGRP'):
subpart_and_subjgrp_xmls.append(subpart)
# SUBJGRPS can be nested, particularly inside SUBPARTs
for subjgrp in subpart.xpath('./SUBJGRP'):
subpart_and_subjgrp_xmls.append(subjgrp)
if len(subpart_and_subjgrp_xmls) > 0:
subthings = []
letter_list = []
for subthing in subpart_and_subjgrp_xmls:
if subthing.tag == "SUBPART":
subthings.append(build_subpart(reg_part, subthing))
elif subthing.tag == "SUBJGRP":
built_subjgrp = build_subjgrp(reg_part, subthing, letter_list)
letter_list.append(built_subjgrp.label[-1])
subthings.append(built_subjgrp)
tree.children = subthings
else:
section_xmls = [c for c in part.getchildren() if c.tag == 'SECTION']
sections = []
for section_xml in section_xmls:
sections.extend(build_from_section(reg_part, section_xml))
empty_part = reg_text.build_empty_part(reg_part)
empty_part.children = sections
tree.children = [empty_part]
non_reg_sections = build_non_reg_text(reg_xml, reg_part)
tree.children += non_reg_sections
return tree
def build_tree(reg_xml):
logger.info("Build tree %s", reg_xml)
preprocess_xml(reg_xml)
reg_part = get_reg_part(reg_xml)
title = get_title(reg_xml)
tree = Node("", [], [reg_part], title)
part = reg_xml.xpath('//PART')[0]
# Build a list of SUBPARTs, then pull SUBJGRPs into that list:
subpart_and_subjgrp_xmls = []
for subpart in part.xpath('./SUBPART|./SUBJGRP'):
subpart_and_subjgrp_xmls.append(subpart)
# SUBJGRPS can be nested, particularly inside SUBPARTs
for subjgrp in subpart.xpath('./SUBJGRP'):
subpart_and_subjgrp_xmls.append(subjgrp)
if len(subpart_and_subjgrp_xmls) > 0:
subthings = []
letter_list = []
for subthing in subpart_and_subjgrp_xmls:
if subthing.tag == "SUBPART":
subthings.append(build_subpart(reg_part, subthing))
elif subthing.tag == "SUBJGRP":
built_subjgrp = build_subjgrp(reg_part, subthing, letter_list)
letter_list.append(built_subjgrp.label[-1])
subthings.append(built_subjgrp)
tree.children = subthings
else:
section_xmls = [c for c in part.getchildren() if c.tag == 'SECTION']
sections = []
for section_xml in section_xmls:
sections.extend(build_from_section(reg_part, section_xml))
empty_part = reg_text.build_empty_part(reg_part)
empty_part.children = sections
tree.children = [empty_part]
non_reg_sections = build_non_reg_text(reg_xml, reg_part)
tree.children += non_reg_sections
return tree
def build_tree(reg_xml):
doc = etree.fromstring(reg_xml)
reg_part = doc.xpath('//REGTEXT')[0].attrib['PART']
parent = doc.xpath('//REGTEXT/PART/HD')[0]
title = parent.text
tree = Node("", [], [reg_part], title)
part = doc.xpath('//REGTEXT/PART')[0]
html_parser = HTMLParser.HTMLParser()
sections = []
for child in part.getchildren():
if child.tag == 'SECTION':
sections.append(build_section(reg_part, child))
tree.children = sections
non_reg_sections = build_non_reg_text(reg_xml)
tree.children += non_reg_sections
return tree
