Nokogiri extract object that is not wrapped in tags - ruby

Question

I'm loading regulations to a database and putting them in a tree hierarchy. When the XML Im scraping is set up as below, it is trivial to scrape it:
<CHAPTER>
<PART>
<SUBPART>
<SECTION>
<HD>Section Title</HD>
</SECTION>
<SECTION> ... </SECTION>
<APPENDIX>
<HD>Appendix Title</HD>
<P>Appendix content...</P>
<FOO>More content in unexpected tags</FOO>
</APPENDIX>
</SUBPART>
<SUBPART> ... </SUBPART>
</PART>
</CHAPTER>
Since I have to know what the parent ID is, I do something along this line:
parent_id = 1
doc.xpath("//chapter/part/subpart").each do |subpart|
title = subpart.xpath("hd").first.text
# add is a method that creates object and saves it to database, returning its id
id = add(title,'SUBPART',parent_id)
subpart.xpath('section').each do |section|
title = section.xpath('hd').first.text
add(title,'SECTION',id)
end
subpart.xpath('appendix').each do |app|
title = section.xpath('hd').first.text
content = app.to_s
add(title,'APPENDIX',id,content) #content is an optional input
end
end
However, the XML is not always set up in such a logical way. Sometimes, the appendices are not wrapped in tags :(
When this is the case, the XML looks like this:
<EXTRACT>
<HD SOURCE="HD1">Appendix A to § 1926.60—Substance Data Sheet, for 4-4′ Methylenedianiline</HD>
<NOTE>
<HD SOURCE="HED">Note:</HD>
<P>The requirements applicable to construction work under this appendix A are identical to those set forth in appendix A to § 1910.1050 of this chapter.</P>
</NOTE>
<HD SOURCE="HD1">Appendix B to § 1926.60—Substance Technical Guidelines, MDA</HD>
<NOTE>
<HD SOURCE="HED">Note:</HD>
<P>The requirements applicable to construction work under this appendix B are identical to those set forth in appendix B to § 1910.1050 of this chapter.</P>
</NOTE>
<HD SOURCE="HD1">Appendix C to § 1926.60—Medical Surveillance Guidelines for MDA</HD>
<NOTE>
<HD SOURCE="HED">Note:</HD>
<P>The requirements applicable to construction work under this appendix C are identical to those set forth in appendix C to § 1910.1050 of this chapter.</P>
</NOTE>
<HD SOURCE="HD1">Appendix D to § 1926.60—Sampling and Analytical Methods for MDA Monitoring and Measurement Procedures</HD>
<NOTE>
<HD SOURCE="HED">Note:</HD>
<P>The requirements applicable to construction work under this appendix D are identical to those set forth in appendix D to § 1910.1050 of this chapter.</P>
</NOTE>
</EXTRACT>
<CITA>
[57 FR 35681, Aug. 10, 1992, as amended at 57 FR 49649, Nov. 3, 1992; 61 FR 5510, Feb. 13, 1996; 61 FR 31431, June 20, 1996; 63 FR 1296, Jan. 8, 1998; 69 FR 70373, Dec. 6, 2004; 70 FR 1143, Jan. 5, 2005; 71 FR 16674, Apr. 3, 2006; 71 FR 50191, Aug. 24, 2006; 73 FR 75588, Dec. 12, 2008; 76 FR 33611, June 8, 2011; 77 FR 17889, Mar. 26, 2012]
</CITA>
The only way I can think of extracting these appendices is to iterate through the <EXTRACT> node and check the tags to see if its name is "HD" and "Appendix" is in the text. Then save everything after until I hit the next <HD> with "Appendix" in the text.
Feels like a pretty clunky solution. Is there a better way to do this?