#!/usr/bin/env ruby
require 'rubygems'
require 'hpricot'

doc = Hpricot(open(ARGV[0]))

# Nuke the credits
doc.search("//meta[@name='GENERATOR']").remove
doc.search("//meta[@name='generator']").remove
els = doc.search("/html/body/blockquote/em/text()")
els.each { |e|
  if e.to_s =~ /This document was/
    doc.search(e.parent.parent.previous.previous.xpath).remove
    doc.search(e.parent.parent.xpath).remove
  end
}

# and HTML comments
doc.search("//comment()").remove

# clean up the subsections
doc.search("//h2[@class='section']").remove_attr(:class)
doc.search("//h3[@class='subsection']").remove_attr(:class)
doc.search("//h4[@class='subsubsection']").remove_attr(:class)


# Clean up the footnotes and references DL lists

doc.search("//dl[@class='thebibliography']").set(:class, "refs")
doc.search("//dl/dt[@class='dt-thefootnotes']").remove_attr(:class)
doc.search("//dl/dd[@class='dd-thefootnotes']").remove_attr(:class)
doc.search("//dl/dt[@class='dt-thebibliography']").remove_attr(:class)
doc.search("//dl/dd[@class='dd-thebibliography']").remove_attr(:class)
(doc/"//dl[@class='thefootnotes']").set(:class, 'footnotes')

# Squish IDs into parents and remove unnecessary a's
["//h2/a", "//h3/a", "//h4/a"].each { |h|
    doc.search(h).each { |a|
        a.parent.set_attribute(:id, a.get_attribute(:id))
        a.parent.replace_child(a, Hpricot::Text.new(a.inner_html))
    }
}

["//dl[@class='footnotes']/dt/a",
 "//dl[@class='refs']/dt/a"].each { |thing|
    doc.search(thing).each { |a|
        a.parent.set_attribute(:id, a.get_attribute(:id))
        a.parent.inner_html = a.inner_html
    }
}

(doc/"//a").remove_attr(:name)
(doc/"//td").remove_attr(:nowrap)

# Clean up some of the figures
(doc/"//blockquote[@class='table']").each { |e|
  e.swap("<div class=\"figure\">\n"+e.inner_html+"\n</div>\n")
}
(doc/"//blockquote[@class='figure']").each { |e|
  e.swap("<div class=\"figure\">\n"+e.inner_html+"\n</div>\n")
}
(doc/"//div[@class='caption']").each { |e|
  e.swap("<p class=\"caption\">\n"+e.inner_html+"\n</p>\n")
}

(doc/"//div[@class='figure']/div[@class='center']/img").each { |e|
  e.parent.swap("<p class=\"fig\">"+e.parent.inner_html+"</p>\n")
}

(doc/"//div[@class='figure']/div[@class='center']").each { |e|
    e.swap(e.inner_html)
}

(doc/"//div[@class='figure']/p[@class='caption']/table/tr/td").each { |e|
    e.parent.parent.swap(e.inner_html)
}

(doc/"//div[@class='figure']/p[@class='caption']/font").each { |e|
    e.swap(e.inner_html)
}

# Move a trailing <a id> label into the div
(doc/"//div[@class='figure']/p[@class='caption']").each { |cap|
    ns = cap.next_sibling
    if (ns.name == "a" && ns.get_attribute(:id))
        theid = ns.get_attribute(:id)
        ns.parent.children.delete(ns)
        cap.parent.set_attribute(:id, theid)
    end
}

doc.search("//dl[@class='refs']/dt/font").each { |f|
    f.parent.inner_html = f.inner_html
}

puts doc.to_html