Question

Here’s an example of the data:

<animals><name>George</name><description>A big brown fox.</description></animals>

It really doesn’t get more complicated than that. I want to modify all text in the elements. (In this case, encrypt it).

What I’ve come up with so far is:

xml_data.gsub(/(<.*>)(.+)(<\/.*>)(?=<)/, "#{$1}#{$2.encrypt_string}#{$3}")

But, it only replaces the last element’s text. So I’m obviously missing something.

I invite any suggestions (including using REXML). I must use libraries standard with Ruby 1.8.7.

There is no chance of the XML being malformed because I wrote the process that produces it.

Thank you in advance!

Answer 1

Don’t use regular expressions for this, use a real parser such as Nokogiri:

s = '<animals><name>George</name><description>A big brown fox.</description></animals>'
d = Nokogiri::XML(s)
d.search('//text()').each { |n| n.content = n.content.encrypt_string }
s2 = d.to_xml(:save_with => Nokogiri::XML::Node::SaveOptions::NO_DECLARATION).strip

Assuming of course that you have monkey patched encrypt_string into String somewhere.

As far as your regex goes, (.+) is greedy and will happily consume </close_tag>, you have similar problems with .*.

---

If you must use a regex (and it seems that you have choice), then you should tighten up your regex and switch to the block form of gsub to get sensible $1 and $2:

xml_data.gsub(/<([^>]+)>([^<]+)<\/\1>/) { "<#{$1}>#{$2.encrypt_string}</#{$1}>" }

Using [^>]+ and [^<]+ keeps you within the tags you want and the \1 back-reference is an easy to way match the opening and closing tags. For example, using upcase in place of encrypt_string does this:

>> s = '<animals><name>George</name><description>A big brown fox.</description></animals>'
>> s.gsub(/<([^>]+)>([^<]+)<\/\1>/) { "<#{$1}>#{$2.upcase}</#{$1}>" }
=> "<animals><name>GEORGE</name><description>A BIG BROWN FOX.</description></animals>"