private static void AddAttribute(XmlReadInstructions instruction, string name, string xpath, HashSet<string> addedColumns)
{
if (addedColumns.Contains(name))
{
int index = 2;
while (addedColumns.Contains(name + index)) index++;
name = name + index;
}
addedColumns.Add(name);
instruction.Columns.Add(new XmlColumn
{
Name = name,
XPath = xpath,
});
}
private static void CollectColumns(XmlReadInstructions instruction, Node current, HashSet<string> addedColumns, Node relativeTo)
{
if (current.HasText)
{
AddAttribute(instruction, current.Name, current.GetXPath(relativeTo), addedColumns);
}
foreach (var attr in current.AttributeList)
{
if (attr.Contains(":")) continue;
string xpath = current.GetXPath(relativeTo);
if (!String.IsNullOrEmpty(xpath) && !xpath.EndsWith("/")) xpath += "/";
xpath += "@" + attr;
AddAttribute(instruction, attr, xpath, addedColumns);
}
foreach (var child in current.ChildList)
{
if (child.IsRepetetive) continue;
CollectColumns(instruction, child, addedColumns, relativeTo);
}
}
public static List<XmlReadInstructions> AnalyseXmlReader(System.Xml.XmlReader reader, bool globalUniqueColumnNames)
{
var root = new Node();
var current = root;
var resultSets = new Dictionary<string, Node>();
while (reader.Read())
{
switch (reader.NodeType)
{
case XmlNodeType.Element:
current = current.SubElement(reader.Name);
if (reader.HasAttributes)
{
reader.MoveToFirstAttribute();
do
{
current.Attribute(reader.Name);
} while (reader.MoveToNextAttribute());
reader.MoveToElement();
}
if (current.CurrentCount > 1)
{
string xpath = current.AbsXPath;
if (!resultSets.ContainsKey(xpath))
{
resultSets[xpath] = current;
current.IsRepetetive = true;
}
}
if (reader.IsEmptyElement) current = current.Parent;
break;
case XmlNodeType.Text:
if (!String.IsNullOrWhiteSpace(reader.Value))
{
current.HasText = true;
}
break;
case XmlNodeType.XmlDeclaration:
case XmlNodeType.ProcessingInstruction:
case XmlNodeType.Comment:
continue;
case XmlNodeType.EndElement:
current = current.Parent;
break;
}
}
// remove repetetive parents. Remains only innermost repetetives
foreach (var resultSet in resultSets.Values.ToList())
{
var node = resultSet;
node = node.Parent;
while (node != null && !node.IsRepetetive) node = node.Parent;
if (node != null)
{
resultSets.Remove(node.AbsXPath);
node.IsRepetetive = false;
}
}
if (!resultSets.Any())
{
resultSets["/"] = root;
}
var res = new List<XmlReadInstructions>();
var addedColumns = new HashSet<string>();
var collectionNames = new HashSet<string>();
foreach (var resultSet in resultSets.Values)
{
var instruction = new XmlReadInstructions();
instruction.XPath = resultSet.AbsXPath ?? "/";
string collectionName = resultSet.Name;
if (collectionNames.Contains(collectionName))
{
int index = 2;
while (collectionNames.Contains(collectionName + index)) index++;
collectionName = collectionName + index;
}
instruction.CollectionName = collectionName;
if (!globalUniqueColumnNames) addedColumns.Clear();
CollectColumns(instruction, root, addedColumns, resultSet);
if (resultSet != root)
{
CollectColumns(instruction, resultSet, addedColumns, resultSet);
}
res.Add(instruction);
}
return res;
}