gistfile1.rb

# This is for breaking up XML docs
# into multiple chunks by a given node type.
# For example, in TEI the "pb" tag exists to signify page breaks...
# Using this code, you could break up the doc into multiple documents
# based on the position/depth of the individual "pb" tags.

require 'rubygems'
require 'nokogiri'

def next_node(start, e=[], &b)
  e << start
  if start.children.size > 0
    yield start.children.first
    next_node start.children.first, e, &b
  end
  if n = start.next
    yield n
    next_node n, e, &b
  end
  if ! e.include?(start) and n = start.parent.next
    yield n
    next_node n, e, &b
  end
end

module XMLMethods
  
  def previous_nodes_r
    ps = previous_sibling
    ((parent.previous_nodes_r rescue []) + (ps.previous_nodes_r rescue []) + [previous_sibling]).compact.uniq
  end
  
end

def fragment_doc(source_doc, pattern)
  pages = []
  source_doc.search(pattern).each do |test_pb|
    doc = Nokogiri::XML(source_doc.to_xml)
    pb = doc.at(test_pb.path)
    deletions = []
    mode = nil
    after_target_node = false
    next_node(doc) do |e|
      e.extend XMLMethods
      if e == pb
        deletions += e.previous_nodes_r
        after_target_node = true
      end
      if after_target_node and e != pb and e.name == pb.name
        mode = :delete
      end
      deletions << e if mode == :delete and ! pb.ancestors.include?(e)
    end
    deletions.each do |e|
      e.remove
    end
    pages << doc
  end
  pages
end

data = %Q(
<div>
  <pre>
  	<pb id="pb1"/>
  	<div id="one">
  		<p>1 some text</p>
  		<p>2 some text</p>
  		<p>3 some text</p>
  		<p id="prePb2">4 some text</p>
  		<pb id="pb2"/>
  		<p>5 some text</p>
  		<p>6 some text</p>
  		<div id="one-one">
  			<p>1-1 some text</p>
  			<p>1-2 some text</p>
  			<p>1-3 some text</p>
  			<pb id="pb3"/>
  			<p>1-4 some text</p>
  			<p>1-5 some text</p>
  		</div>
  	</div>
  	<pb id="pb4"/>
  	<div id="two">
  		<p>1 some text</p>
  		<p>2 some text</p>
  		<p>3 some text</p>
  		<p>4 some text</p>
  		<pb id="pb5"/>
  		<p>5 some text</p>
  		<p>6 some text</p>
  		<div id="two-one">
  			<pb id="pb6"/>
  			<p>2-1 some text</p>
  			<p>2-2 some text</p>
  			<p>2-3 some text</p>
  			<p>2-4 some text</p>
  			<p>3-5 some text</p>
  		</div>
  		<p id="ending1">ENDING TEXT 1</p>
  	</div>
  </pre>
	<p id="ending2">ENDING TEXT 2</p>
</div>
)

xml = Nokogiri::XML(data)
pages = fragment_doc(xml, 'pb')
puts pages.inspect