How to load and merge a dataset of XML docs

I would like to consume a dataset of XML documents, and merge them into a single document containing only distinct elements.

To illustrate, I have a dataset as:

r, x
-- -------------------------------
1, <root><a>111</a></root>
2, <root><a>222</a><b>222</b></root>
3, <root><c>333</c></root>

would result in:

<a>111</a><b>222</b><c>333</c>

The <a> element from r=2 is not merged since we already have an element = <a> from r=1. I need only merge new elements, starting with r=1 going forward.

I am able to iterate over the list, but having difficulty comparing and merging. The code below fails to identify <a>222</a> as a duplicate. Is it possibly comparing the element values as well?

 using (SqlDataReader dsReader = cmd.ExecuteReader())
            {
                XDocument baseDoc = new XDocument();
                XDocument childDoc = new XDocument();

                while (dsReader.Read())
                {
                    // this is the base doc, merge forward from here
                    if (dsReader["r"].ToString() == "1")
                    {
                        baseDoc = XDocument.Parse(dsReader["x"].ToString());
                        SqlContext.Pipe.Send("start:" + baseDoc.ToString());

                    }
                    // this is a child doc, do merge operation
                    else
                    {
                        childDoc = XDoc开发者_如何学JAVAument.Parse(dsReader["x"].ToString());

                        // find elements only present in child
                        var childOnly = (childDoc.Descendants("root").Elements()).Except(baseDoc.Descendants("root").Elements());
                        foreach (var e in childOnly)
                        {
                            baseDoc.Root.Add(e);
                        }
                    }
                }
            }

I am bit confused about baseDoc and childDoc usage in your code. I hope I correctly understood your question. Here is my proposal:

        using (SqlDataReader dsReader = cmd.ExecuteReader())
        {
            XElement result = new XElement("root");
            while (dsReader.Read())
            {
                // Read source
                XDocument srcDoc = XDocument.Parse(dsReader["x"].ToString());

                // Construct result element
                foreach (XElement baseElement in srcDoc.Descendants("root").Elements())
                    if (result.Element(baseElement.Name) == null)   // skip already added nodes
                        result.Add(new XElement(baseElement.Name, baseElement.Value));

            }
            // Construct result string from sub-elements (to avoid "<root>..</root>" in output)
            string str = "";
            foreach (XElement element in result.Elements())
                str += element.ToString();

            // send the result
            SqlContext.Pipe.Send("start:" + str);
        }

Note that my code ignores r-numbering. I use order as it comes from sql data reader. If rows are not sorted by "r", then additional sort is required before my code.