Element#siblings_at, Element#nodes_at

  >> doc.at(:h3)
  => {elem h3 {text "bottles and cans"} h3}
  >> doc.at(:h3).siblings_at(0..2)
  => #<Hpricot::Elements[
       {elem <h3> {text "bottles and cans"} </h3>}, 
       {elem <p> {text "just clap your hands"} </p>}, 
       {elem <ul> {elem <li> {text "2 turntables"} </li>}
         {elem <li> {text "a microphone"} </li>} </ul>}]>

Basically: grab this element, as well as the two siblings below it. The nodes_at method will include text elements and comments.

  >> doc.at(:h3).nodes_at(0..2)
  => #<Hpricot::Elements[
       {elem <h3> {text "bottles and cans"} </h3>},
       {text "\n"}, 
       {elem <p> {text "just clap your hands"} </p>}]>

You can also do stuff like nodes_at(-2, 2, 5) to grab specific elements. Nodes positioned at two places above, two places below and five places below the selected element. (doc)

text() and comment()

  >> doc.search("p/text()")
  => #<Hpricot::Elements[{text "just clap your hands"}]>
  >> doc.at("//comment()")
  => {comment "<!-- insert mp3 of applause here -->"}

Element#to_original_html

  >> doc = Hpricot("<p>a bunch of <b>messy <i>messy</b> html that" +
       "doesn't</u> match up<!_ egg _!>")
  => #<Hpricot::Doc {elem <p> {text "a bunch of "} {elem <b> {text "messy "} 
       {elem <i> {text "messy"}} </b>} {text " html that doesn't"} 
       {bogusetag</uu</u>>} {text " match up<!_ egg _!>"}}>
  >> puts doc.to_original_html
  <p>a bunch of <b>messy <i>messy</b> html that doesn't match up<!_ egg _!>

XPath indices

  >> doc.at("li[1]")
  => {elem li {text "a microphone"} li}

Those indices work like E:nth-of-type.

Version 0.5 of Hpricot is nearing. Please test the latest gem to help me figure out any subtleties. Also, rdoc is now included. So, yeah: HELP!

 gem install hpricot --source code.whytheluckystiff.net