I am working with well-formed but non-validated (no DTD, no schema)
computer-generated XML. My task is to make consistent objects from a
large number of these mainframe-generated strings; for instance, one
call might generate the XML
"<RESPONSETYPE1><firstName>Aaron</firstName><firstName>Bob</firstName><l
astName>Amundsen</lastName><lastName>Bickerson</lastName></RESPONSETYPE1
>" and another call might generate
"<RESPONSETYPE2><First>Aaron</First><Last>Amundsen</Last><First>Bob></Fi
rst><Last>Bickerson</Last></RESPONSETYPE2>". My task is to turn these
into .NET objects (say, Customer objects with FirstName and LastName
properties).
Beneath the root element, the data is absolutely flat: repeating data
types are never aggregated into sub-elements. For instance, in
<RESPONSETYPE1> above, I have to match the first <firstName> element
with the first <lastName> element, etc. My question is: am I foolish for
using regular expressions to work with the XML rather than an XML
parser?
My "spike" solution is to use the following method. Can this be done
more efficiently (in either lines of code or significant runtime
performance) with an XML parser? Please note that the mainframe data is
absolutely consistent in generating whitespace, character set, and so
forth. XML's advantages in these areas are not relevant.
public static string[] Elements(string elementName, string xml)
{
ArrayList responses = new ArrayList();
Regex re = new Regex(String.Format("<{0}>(?<value>.*?)</{0}>",
elementName));
MatchCollection matches = re.Matches(xml);
foreach(Match m in matches)
{
string value = m.Groups["value"].Value;
responses.Add(value);
}
return (string[]) responses.ToArray(typeof(string));
}
Thanks in advance,
Larry
