First, an introduction to my scenario in how Hpricot¹ became of use:

The Scenario

Imagine that you have an article, and when you create that article you specify headings with ‘h3′ tags. My goal was to parse a body of text for all h3 tags, grab them and put them in a table of contents. That part was easy with some natural regexp. Next, I wanted to link each header in the table of contents to it’s respective heading with an anchor link.

The Problem

This is where the trouble started to occur. I got to the point where the name anchor was being updated on every h3 tag, but only if no other attribute existed in the ‘h3′ tag. This is due to the fragility of the regexp I used. To no avail with this cryptic demon, I scoured the internets for a solution; along came a mythical angel on IRC that mentioned using Hpricot to solve the problem I was having.

The thought of not having to use regexp, and replacing it with a pretty parser? Oh this is my bread and butter of programming enjoyment! Luckily, the day after all, was in fact saved due to Hpricot.

Hpricotified Solution

Here’s the victorious code that brought me to my solution:

  def chapter_titles
    headers = self.body.scan(/<h3.*?>(.+?)<\/h3>/)
  end
 
  def set_headers
    body = Hpricot(self.body)
    (body/"h3").each do |each|
      value = each.inner_html
      each.set_attribute("name", "#{value}")
    end
    self.body = body.to_s
  end

Now, I could probably create a simple version of chapter_titles with Hpricot, as a challenge maybe you should do it!

Regexpified Lame Version

Here’s the old regexp version of set_headers

  def set_chapter_names
   self.body.scan(/<h3>(.+?)<\/h3>/).each do |h|
     header = h.to_s
     header.gsub!(/^\s+|\s+$/, '')
     self.body.gsub!("<h3>#{h}</h3>", "<h3 name=\"#{header}\">#{h}</h3>")
   end
  end

As you can see, that’s hideous compared to the Hpricot version.

If anyone is concerned about speed with Hpricot, I don’t think it’s an issue. First of all, it’s coded by Why The Lucky Stiff, that almost requires a “nuff’ said”, but it’s scanner is also “…coded in C”². Doubt the speed now?

If you’d like to take Hpricot for a spin and give it a try, just check out it’s home page.

What are your experiences with Hpricot if you’ve used it before?

1. http://code.whytheluckystiff.net/hpricot/ [↩]
2. From the first paragraph of http://code.whytheluckystiff.net/hpricot/ [↩]

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site. Back to Top.