getting an array of objects out of XML using linq

Question

I was wondering is there a way that I can use Linq to XML to get an array of objects out of an xml string... I am used to using XPath to do all of my dirty work, and Xpath seems so intuitive. But, I keep hearing about how great linq is and once you get it, then it makes you life much easier. The jury is still out on that one, but then again, I am not that great with Linq. Still learning. But suppose I have an object...

class PatientClass
{
   //public int Item_ID { get; set; }
   public int PatientId { get; set; }
   public int EMPIID { get; set; }
}

Suppose I have another object...

public class TemplateModel
{
    List<PatientACOModel> Template { set; get; }
}

which is simply a list of the first object...

suppose I have an xml document that looks like so...

<dataTemplateSpecification id="id1" name="name1" >
<templates xmlns="">
<template>
  <elements>
    <element id="element0" name="PatientId" display="Patient ID" dataType="String" visable="true" readOnly="false" value="4563">
      <mapping path="//Template/TemplateData/ACOData/PATIENT_ID" />
    </element>
    <element id="element1" name="PopulationPatientID" display="Population Patient ID" dataType="String" visable="true" readOnly="true" enc="2098" value="6407">
      <mapping path="//Template/TemplateData/ACOData/POPULATION_PATIENT_ID" />
    </element>
    <element id="element2" name="EMPIID" display="EMPIID" dataType="String" visable="true" readOnly="true" value="">
      <mapping path="//Template/TemplateData/ACOData/EMPI" />
    </element>        
  </elements>
</template>
<template>
  <elements>
    <element id="element0" name="PatientId" display="Patient ID" dataType="String" visable="true" readOnly="false" value="4563">
      <mapping path="//Template/TemplateData/ACOData/PATIENT_ID" />
    </element>
    <element id="element1" name="PopulationPatientID" display="Population Patient ID" dataType="String" visable="true" readOnly="true" enc="2098" value="6407">
      <mapping path="//Template/TemplateData/ACOData/POPULATION_PATIENT_ID" />
    </element>
    <element id="element2" name="EMPIID" display="EMPIID" dataType="String" visable="true" readOnly="true" value="">
      <mapping path="//Template/TemplateData/ACOData/EMPI" />
    </element>        
  </elements>
</template>
<template>
  <elements>
    <element id="element0" name="PatientId" display="Patient ID" dataType="String" visable="true" readOnly="false" value="4563">
      <mapping path="//Template/TemplateData/ACOData/PATIENT_ID" />
    </element>
    <element id="element1" name="PopulationPatientID" display="Population Patient ID" dataType="String" visable="true" readOnly="true" enc="2098" value="6407">
      <mapping path="//Template/TemplateData/ACOData/POPULATION_PATIENT_ID" />
    </element>
    <element id="element2" name="EMPIID" display="EMPIID" dataType="String" visable="true" readOnly="true" value="">
      <mapping path="//Template/TemplateData/ACOData/EMPI" />
    </element>        
  </elements>
</template>
</templates>
</dataTemplateSpecification>

You see, dataTemplateSpecification/templates/template would be an instance of the PatientClass that I have above. whereas dataTemplateSpecification/templates/ would be an instance of the TemplateModel object (a list of PatientClasses... I have them named PatientACOModel in the List, but they are essentially the same thing... one simply doesn't have as many variables as the other).

Now I use this to parse out a patient object...

IEnumerable<PatientClass> template = (IEnumerable<PatientClass>)(from templates in xDocument.Descendants("dataTemplateSpecification")//elem.XPathSelectElements(string.Format("//templates/template[./elements/element[@name=\"PopulationPatientID\"and @value='{0}' and @enc='{1}']]", "1", 0))
                                               select new PatientClass
                                               {
                                                  PatientId = int.Parse(templates.Descendants("element").Single(el => el.Attribute("name").Value=="PatientId").ToString()),//XPathSelectElement("elements/element[@name='PatientId']").Attribute("value").Value),
                                                  EMPIID = int.Parse(templates.Descendants("element").Single(el => el.Attribute("name").Value=="EMPIID").ToString()),//XPathSelectElement("elements/element[@name='EMPIID']").Attribute("value").Value),
                                               }

It is currently returning null values, but I am working on that... But How could I get a list of these patients. I would probably need a super query to handle the list of stuff and sub query to get the Patient info right?

So something like this...

IEnumerable<TemplateModel> template = (IEnumerable<TemplateModel>)(from templates in elem.XPathSelectElements("//templates/template")
                                                 select new TemplateModel
                                                 {
                                                     TemplateModel = 
                                                     (from pat in templates
                                                      select new PatientClass
                                                      {
                                                          PatientId = int.Parse(templates.XPathSelectElement("elements/element[@name='PatientId']").Attribute("value").Value),
                                                          EMPIID = int.Parse(templates.XPathSelectElement("elements/element[@name='EMPIID']").Attribute("value").Value),
                                                     )
                                                  }

That would seem logical to me. But perhaps I don't understand the basis of Linq

Is there some reason you're not just serializing your objects in the standard way? — Ann L.
– Ann L., Commented Mar 13, 2012 at 14:39
That would be one seriously ugly object. For some reason, my architect is a big fan of this xml pattern. It probably has something to do with the industry that we are in. — SoftwareSavant
– SoftwareSavant, Commented Mar 13, 2012 at 15:11
One possibility would be to use XSL to transform the data from a metadata specification to somethign more concrete (i.e. an element named PatientID with a value of whatever, rather than the specification of an element to be called PatientID). You could then transform that into an object using XML serialization. But I'm guessing it's the metadata-to-concrete-anything step that's the killer, right? — Ann L.
– Ann L., Commented Mar 13, 2012 at 16:00

Phil · Accepted Answer · 2012-03-13 17:34:36Z

2

Give this a try (because you like XPath):

var s="<dataTemplateSpecification .../>";

var element = XElement.Parse(s);

var patients = element.XPathSelectElements("//elements").Select (
    e => new Patient
        {
            PatientId = (int)e.XPathSelectElement("//element[@id='element0']").Attribute("value"),
            PopulationPatientId = (int)e.XPathSelectElement("//element[@id='element1']").Attribute("value"),
            EmpId = (string)e.XPathSelectElement("//element[@id='element2']").Attribute("value"),
        }
    );

Where the patient class is:

class Patient{
    public int PatientId{get;set;}
    public int PopulationPatientId{get;set;}
    public string EmpId{get;set;}
}

edited Mar 13, 2012 at 17:34

answered Mar 13, 2012 at 17:25

Phil

43.2k9 gold badges102 silver badges102 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

devgeezer · Accepted Answer · 2012-03-13 17:22:12Z

1

We use hand-rolled linq-to-xml at work, so I'll take a swing.

One quick note is that all of the EMPIID elements have a value = "", which will cause a FormatException when calling int.Parse. There are a series of explicit cast operators on XAttribute that elegantly get you to common .net types.

Here's code that I might write for the parsing you need to do (staying pure to the xnode style of parsing without using XPath):

IEnumerable<TemplateModel> templates =
    from dataTemplate in xDocument.Descendants("dataTemplateSpecification")
    select new TemplateModel
    {
        TemplateModel = 
            (from template in dataTemplate.Element("templates").Elements("template")
            let elements = template.Element("elements").Elements("element")
            select new PatientClass
            {
                PatientId = (int)elements.Single(e => (string)e.Attribute["name"] == "PatientId").Attribute("value"), 
                EMPIID = (int)elements.Single(e = (string)e.Attribute["name"] == "EMPIID").Attribute("value"),
            }).ToList()
    };

The type of the query is already IEnumerable<TemplateModel> so there is no immediate need to cast the result. In most cases, my preference is to simply leave the type declared as var but your tastes may differ - leaving it as var means that I don't have to correct types if I apply grouping or other LINQ transforms for instance. I introduced a range variable elements using the let keyword; this is useful for calculating a value just once and reusing it. Since the type of the TemplateModel property on the TemplateModel type is a list, I call the ToList extension to set the property. I call the cast operator for the attributes to convert their values to int and string rather than calling int.Parse and ToString.

The capabilities of LINQ permit further processing (filtering, folding and projection) in a very elegant manner. Being able to seamlessly transition from linq-to-xml to working in linq-to-objects makes LINQ very compelling, IMHO. Anyhow, come on in to the LINQ pool, the water's fine. :)

answered Mar 13, 2012 at 17:22

devgeezer

4,2191 gold badge22 silver badges26 bronze badges

2 Comments

SoftwareSavant Over a year ago

good answer. But when there is an empty string, I always get an Input string was not in a correct format. Exception. Same thing for DateTime attributes I guess, what would I do with empty strings?

devgeezer Over a year ago

You might try changing to a nullable type for the value types - for example: int? aka Nullable<int> and DateTime? respectively. XAttribute has conversion operators for these types as well.

Collectives™ on Stack Overflow

getting an array of objects out of XML using linq

2 Answers 2

Comments

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related