how to keep blank result for Nokogiri's NodeSet.search method

i want to run the search method of Nokogiri::XML::NodeSet based on one NodeSet called nodeset for some xpath rule like below:

nodeset.search(rule)

the above code returns a NodeSet, but that doesn't contain the ones which can not match the rule. My intention is t开发者_JAVA技巧hat: if element in nodeset is matched the rule, ok please return the matched result; if not matched please return a blank string in the result, so that i can know which element in caller nodeset is matched, which element in caller nodeset is not matched.

Could someone tell me how to do it? i will appreciate your help very much.

---

Nokogiri NodeSet support set operations similar to Ruby arrays. Instead of keeping blanks in your matched set, find out the missed items after the fact:

require 'nokogiri'

doc = Nokogiri::XML <<-ENDXML
<root>
  <a id="a1" class="foo">
    <a id="a1a" class="foo" />
    <a id="a1b" class="foo" andalso="this" />
  </a>
  <a id="a2" class="foo" andalso="this">
    <a id="a2a" class="bar" />
    <a id="a2b" class="bar" andalso="this" />
  </a>
  <a id="a3" class="foo" andalso="this" />
</root>
ENDXML

foos = doc.xpath('//a[@class="foo"]')
p foos.map{ |e| e['id'] }
#=> ["a1", "a1a", "a1b", "a2"]

subselect = foos.xpath('self::*[@andalso="this"]')
p subselect.map{ |e| e['id'] }
#=> ["a1b", "a2", "a3"]

missed = foos - subselect
p missed.map{ |e| e['id'] }
#=> ["a1", "a1a"]

If you really want non-nodes in the result, you'll have to use #map instead of #search or other Nokogiri methods and get an Array instead of a NodeSet:

subselect = foos.map do |el|
  if el['andalso']=='this'
    el
  else
    ""
  end
end
p subselect.map{ |e| e=="" ? "" : e['id'] }
#=> ["", "", "a1b", "a2", "a3"]

---

I don't know nokogiri well enough to know how well this will work but I suspect the following example may suggest a way forward. The following assumes that NodeSet behaves like a ruby array which it does according to its API docs [1]

a = (0..9).to_a
 => [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
evens = a.select { |i| i % 2 == 0 }
 => [0, 2, 4, 6, 8]
odds = a - evens
 => [1, 3, 5, 7, 9]

I believe you should be able to do something similar with your nodeset so that when your search has been performed, you can find the non-matched nodes by subtracting the new nodeset from the original one.

[1] http://nokogiri.rubyforge.org/nokogiri/Nokogiri/XML/NodeSet.html#M000448

---

Here's how I'd go about it:

require 'nokogiri'

xml = <<EOT
<xml>
  <find_node>foo</find_node>
  <ignore_node>bar</ignore_node>
  <find_node>foo</find_node>
  <ignore_node>bar</ignore_node>
</xml>
EOT

# parse the document...
doc = Nokogiri::XML(xml)

# find the nodes we want...
desired_nodes = doc.search('//find_node')

# see if it's working...
desired_nodes.map{ |n| n.to_xml } # => ["<find_node>foo</find_node>", "<find_node>foo</find_node>"]

# walk the tree, grabbing the text or '' depending on whether the node is a hit or a miss...
node_result = doc.search('/xml/*').map{ |n| desired_nodes.include?(n) ? n.text : '' }

# ** here's the result **
node_result # => ["foo", "", "foo", ""]

---

# if we wanted to we could grab the desired_nodes' text...
desired_nodes.map{ |n| n.text } # => ["foo", "foo"]

# or find the ignored nodes...
ignored_nodes = doc.search('/xml/*') - desired_nodes
ignored_nodes.map{ |n| n.to_xml } # => ["<ignore_node>bar</ignore_node>", "<ignore_node>bar</ignore_node>"]

# ...and grab the ignored_nodes' text...
ignored_nodes.map{ |n| n.text } # => ["bar", "bar"]