Nokogiri: irregular divs - ruby

Question

Trying to deal with irregular content in div elements. Namely what comes after the h3 titles. There is no set content under the h3 headings. However, I need to associate whatever text is there with the heading. There could be a ul or just a span or both. The main thing is not combining all the text under the h3 headings.
I have been able to navigate to my div using the .css operator. Each div contains one or more of 4 h3 headings followed by a comment or a list if there is more than one comment.
How can I separate whatever follows a h3 tag ending before the next tag (if there is one)?
You can see a sample of the div I'm working with here (I can grab whatever is between the h2 because its the same for every div):
<div class="inspection_container">
<h2 class="inspection_date_title">
<div class="calendar_list">
<span>Mar</span><strong>4</strong>
</div>Routine Inspection<small>Inspected Mar. 4, 2014</small>
</h2>
<h3>Actions taken by inspector</h3>
<ul>
<li class="Comment">
<strong>Consultation / Technical Assistance</strong><p>Instructions are given to the owner/operator to assist them with taking the proper actions to meet regulations.</p>
</li>
</ul>
</div>
<div class="inspection_container">
<h2 class="inspection_date_title">
<div class="calendar_list">
<span>Sep</span><strong>4</strong>
</div>Re-inspection<small>Inspected Sep. 4, 2013</small>
</h2>
<h3>Not in compliance</h3>
<ul>
<li class="X">
<strong>Premise is clean/sanitary</strong><p>Food premise is to be maintained in a clean and sanitary condition.</p>
</li>
</ul>
<h3>Actions taken by inspector</h3>
<ul>
<li class="Comment">
<strong>Consultation / Technical Assistance</strong><p>Instructions are given to the owner/operator to assist them with taking the proper actions to meet regulations.</p>
</li>
</ul>
</div>
<div class="inspection_container">
<h2 class="inspection_date_title">
<div class="calendar_list">
<span>Aug</span><strong>30</strong>
</div>Routine Inspection<small>Inspected Aug. 30, 2013</small>
</h2>
<h3>Not in compliance</h3>
<ul>
<li class="X">
<strong>Washrooms are cleaned regularly</strong><p>Washrooms are to be kept clean, sanitary, in good repair and must be supplied with liquid soap in a dispenser, single service/paper towels, cloth roller towel or hot air dryer and hot and cold running water.</p>
</li>
<li class="X">
<strong>Building interior is well-maintained</strong><p>Walls, floors and ceilings are to be maintained and in good repair.</p>
</li>
<li class="X">
<strong>Premise is clean/sanitary</strong><p>Food premise is to be maintained in a clean and sanitary condition.</p>
</li>
</ul>
<h3>Actions taken by inspector</h3>
<ul>
<li class="Comment">
<strong>Consultation / Technical Assistance</strong><p>Instructions are given to the owner/operator to assist them with taking the proper actions to meet regulations.</p>
</li>
</ul>
</div>

Answer 1

Provided that:
You only have intertwined h3 and ul elements till the end of the wrapping div
no other element can appear in this structure instead of ul
no other element can appear in this structure instead of h3
and that your example is representative, this should do the trick.
//ul[count(following-sibling::h3) = count(following-sibling::ul)]
If you have other elements in the same place as the ul but there is always only one between the h3s, you can use this expression
//ul[count(following-sibling::h3) = count(following-sibling::*[not(local-name() = 'h3')])]
As for grouping the h3 elements and the ul elements following them immediately, I don't think this is feasible in XPath alone. You'll need to do this in Ruby. I suggest searching for the div elements and parsing them imperatively, while counting the nodes and grouping the odd and even h3s and uls together