Xpath to attributes

ruby-talk
1

Is there anyway to do generic xpath's, for both elements and attributes
(ie //customer/phone/@type --> "home")? REXML can do xpath to fetch
elements (rexml_element.text('//customer/phone')) but can't seem to use
xpath to find attributes.

2

d = REXML::Document.new <<EOX
    <customer>
      <phone type='home'>0115</phone>
      <phone type='mobi'>07807</phone>
    </customer>
  EOX

  m = REXML::XPath.match(d, '//customer/phone[@type = "home"]')
   p m.to_s
  # => <phone type='home'>0115</phone>

ยทยทยท

On Wed, 2006-02-22 at 10:33 +0900, listrecv@gmail.com wrote:

Is there anyway to do generic xpath's, for both elements and attributes
(ie //customer/phone/@type --> "home")? REXML can do xpath to fetch
elements (rexml_element.text('//customer/phone')) but can't seem to use
xpath to find attributes.

--
Ross Bamford - rosco@roscopeco.REMOVE.co.uk

3

listrecv@gmail.com wrote:

Is there anyway to do generic xpath's, for both elements and attributes
(ie //customer/phone/@type --> "home")? REXML can do xpath to fetch
elements (rexml_element.text('//customer/phone')) but can't seem to use
xpath to find attributes.

class Array; alias atr first; alias txt last end
class String
  def xtag(s)
    scan( %r!
              < #{s} (?: \s+ ( [^>]* ) )? / >
              >
              < #{s} (?: \s+ ( [^>]* ) )? >
              ( .*? ) </ #{s} >
          !mx ).
      map{ |unpaired, attr, data| h = { }
        attr = ( unpaired || attr )
        if attr
          attr.scan( %r! ( \S+ ) = ( ["'] ) ( .*? ) \2 !x ){ |k,q,v|
            h[k] = v }
        end
        [ h, data ]
      }
  end
  def xpath(s)
    s.scan(%r! [^/"]+ (?: "[^"]*" )? !x).inject([[nil,self]]){|ary,str|
      if "@" == str[0,1]
        str =~ /@(.*?)="(.*)"/
        ary.select{|a,t| a[$1] == $2 }
      else
        return if == ary
        ary[0].txt.xtag(str)
      end
    }
  end
end

p DATA.read.xpath('customer/@loc="south"/phone/@type="mobi"')

__END__
<customer loc="north">
  <phone type='home'>0115</phone>
  <phone type='mobi'>07807</phone>
</customer>
<customer loc="south">
  <phone type='home'>0319</phone>
  <phone type='mobi'>09802</phone>
</customer>

4

Thanks.

But I'd like to just extract the attribute value via Xpath.

If my Xpath is correct, I could just do '//customer/phone/@type' to get
'home'. Is there any way I can do this in Ruby?

5

listrecv@gmail.com wrote:

Thanks.

But I'd like to just extract the attribute value via Xpath.

If my Xpath is correct, I could just do '//customer/phone/@type' to get
'home'. Is there any way I can do this in Ruby?

p DATA.read.xchain('customer/@loc="south"/phone').first.atr['type']

__END__
<customer loc="north">
  <phone type='home'>0115</phone>
  <phone type='mobi'>07807</phone>
</customer>
<customer loc="south">
  <phone type='home'>0319</phone>
</customer>

The rest:

class Array; alias atr first; alias txt last end
class String
  def xtag(s)
    result =
    scan( %r!
              < #{s} (?: \s+ ( [^>]* ) )? / >
              >
              < #{s} (?: \s+ ( [^>]* ) )? >
              ( .*? ) </ #{s} >
          !mx ) \
      { |unpaired, attr, data| h = { }
        attr = ( unpaired || attr )
        attr.scan( %r{ ( \S+ ) = ( ["'] ) ( .*? ) \2 }x ){ |k,q,v|
          h[k] = v } if attr
        block_given? ? ( yield [ h, data ] ) : result << [ h, data ]
      }
    result
  end
  def xchain(s)
    s.scan(%r! [^/"]+ (?: "[^"]*" )? !x).inject([[nil,self]]){|ary,str|
      if "@" == str[0,1]
        str =~ /@(.*?)="(.*)"/
        ary.select{|a,t| a[$1] == $2 }
      else
        return if == ary
        ary[0].txt.xtag(str)
      end
    }
  end
end