Nokogiri: create xml from string with `?` in field name - ruby-on-rails

Question

Controller response includes "spec?" field:
r = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<hash type=\"array\">\n <item><spec? type=\"boolean\">false</spec?>\n </item>\n <hash>\n"
When trying to create xml from it with Nokogiri.xml(r) receive literally:
<?xml version="1.0" encoding="UTF-8"?>
<hash type="array">
<item><spec type=" type="boolean">false/spec">
</spec>item>
<hash>
</hash></item></hash>
which is something strange;
My question is:
is it possible to create xml from string using Nokogiri, parsing or removing ? and other non-xml-standart chars, at stage of Nokogiri.XML()?
Desirible result:
Nokogiri.xml(r) do |config|
config.maybe_some_configs?
end #=>
<?xml version="1.0" encoding="UTF-8"?>
<hash type="array">
<item><spec type="boolean">false</spec></item>
</hash>

Answer 1

The proper way to parse a string into an XML DOM is Nokogiri::XML or Nokogiri.XML or Nokogiri::XML.parse, but not using xml.
Also, XML tags can't contain ?. See the spec for more information. You'll have to dig through the "Names and Tokens" section and decode hexadecimal character descriptions to figure out the ranges of characters allowed, but a hint is that ? is character code 0x3f.
Which leads to the fact that the XML in r is invalid:
<?xml version="1.0" encoding="UTF-8"?>
<hash type="array">
<item><spec? type="boolean">false</spec?>
</item>
<hash>
Which, when parsed results in:
irb(main):012:0> doc = Nokogiri::XML(r)
#<Nokogiri::XML::Document:0x80c8014c name="document" children=[#<Nokogiri::XML::Element:0x80c7399c name="hash" attributes=[#<Nokogiri::XML::Attr:0x80c733e8 name="type" value="array">] children=[#<Nokogiri::XML::Text:0x80c6e26c "\n ">, #<Nokogiri::XML::Element:0x80c6df60 name="item" children=[#<Nokogiri::XML::Element:0x80c6d970 name="spec">, #<Nokogiri::XML::Text:0x80c6d09c "? type=\"boolean\">false">]>, #<Nokogiri::XML::Text:0x80c6ca34 "?>\n ">]>]>
irb(main):013:0> doc.errors
[
[0] #<Nokogiri::XML::SyntaxError: error parsing attribute name>,
[1] #<Nokogiri::XML::SyntaxError: attributes construct error>,
[2] #<Nokogiri::XML::SyntaxError: Couldn't find end of Start Tag spec line 3>,
[3] #<Nokogiri::XML::SyntaxError: expected '>'>,
[4] #<Nokogiri::XML::SyntaxError: Opening and ending tag mismatch: item line 3 and spec>,
[5] #<Nokogiri::XML::SyntaxError: Opening and ending tag mismatch: hash line 2 and item>,
[6] #<Nokogiri::XML::SyntaxError: Extra content at the end of the document>
]
As a result, Nokogiri is having to do some fixup in the DOM to try to make sense of it. The resulting XML looks like:
irb(main):014:0> puts doc.to_xml
<?xml version="1.0" encoding="UTF-8"?>
<hash type="array">
<item><spec/>? type="boolean">false</item>?>
</hash>
The way to fix it is to give Nokogiri valid XML. Either fix the source of the XML, if you control it, or fix the problems in the string before passing it to Nokogiri.
By its definition, XML is a strict format, and Nokogiri honors that and, trying to be friendly, makes it possible for you to check errors to see if its empty?. If it's not, odds are good you shouldn't continue using the source until you've determined the problems and fixed whatever causes the parsing problems. Sometimes the problem is fairly benign, and you can ignore it, but in either case you should at least be aware of it.
Pre-massaging the data before Nokogiri sees it isn't hard:
doc = Nokogiri::XML(r.gsub('spec?', 'spec'))
irb(main):024:0> puts doc.to_xml
<?xml version="1.0" encoding="UTF-8"?>
<hash type="array">
<item><spec type="boolean">false</spec>
</item>
<hash>
</hash></hash>
nil
irb(main):025:0> doc.errors
[
[0] #<Nokogiri::XML::SyntaxError: Premature end of data in tag hash line 5>,
[1] #<Nokogiri::XML::SyntaxError: Premature end of data in tag hash line 2>
]
That's a start, but not an attempt to fix it for you completely. I'm teaching you to fish, not handing out fish.