Parsing xml

[Robert Klemme]

require 'mechanize'
mech = WWW::Mechanize.new
mech.get 'http://www.shoe-g.com/index.rdf'
doc = Nokogiri(mech.page.body)
titles = (doc / 'title').map(&:text)

Here's a similar solution using REXML:

require 'open-uri'
require 'rexml/document'

body = open('http://www.shoe-g.com/index.rdf').read
doc = REXML:ocument.new body
titles = doc.elements.to_enumeach, '//title').map(&:text)

Five lines as well...

Cheers

robert

 

[Bill Kelly]

Regex is not stateful, thus you can't use it to parse XML. Oh there
are ways to hack yourself around some limitations and get some
results, but you are going to spend a TON of time making very
unreadable Regex that will die at the presense of slightest malformed
XML.

But, regexps work just fine to lex XML. The parser, then,
becomes a bit of ruby that accepts tokens from the regexp
lexer.

Handling the most common syntactic elements of an XML doc
this way (tags, text, cdata) is relatively trivial.

On the other hand, as we can see from the BNF, handling the
full XML spec is complicated: http://pastie.org/pastes/427101

. . .

In any case, I'm fully on board with the "why reinvent the
wheel?" replies in this thread.

I just had to visit this territory recently because REXML
doesn't work properly in a $SAFE = 4 sandbox.


Regards,

Bill

 

[Simon Krahnke]

I don't see what you could add to the regexp to handle nested tags. You can't
really handle nested structures with regular expressions.

I don't see how the regex above he wrote doesn't already do that.

If you modify it to /<some-tag.*?>(.*?)<\/some-tag>/ it will ignore
attributes, too. Put so simple it can of course problems when there are
other elements whose names start with "some-tag".

ttfn, simon .... l

 

[Sebastian Hungerecker]

I don't see how the regex above he wrote doesn't already do that.

document = "<some-tag> lala <some-tag> lulu </some-tag> lili </some-tag>"
document.match(/<some-tag>(.+?)<\/some-tag>/)[1]
=> " lala <some-tag> lulu "