public static DataElement ConvertHTML2DataElement(string html)
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(html);
string sx = doc.DocumentNode.InnerText;
sx = sx.Substring(sx.IndexOf("ELEMENT NO.") + 12, 3);
List <HtmlNode> tables = helpers.SplitTables(doc);
List <ExtendedHTMLNode> tablesMerged = helpers.MergeTablesOnHeaders(tables);
DataElement de = new DataElement();
de.ElementNumber = sx;
foreach (ExtendedHTMLNode d in tablesMerged)
{
if (d.type == ElementType.Vers)
{
de.Version = d.node.InnerText.Replace("VERSION:", "").Trim();
}
if (d.type == ElementType.FYear)
{
de.FirstYear = d.node.InnerText.Replace("FIRST YEAR:", "").Trim();
}
if (d.type == ElementType.LYear)
{
de.LastYear = d.node.InnerText.Replace("LAST YEAR:", "").Trim();
}
if (d.type == ElementType.Fld)
{
de.FieldName = d.node.InnerText.Replace("FIELD NAME:", "").Trim();
}
if (d.type == ElementType.EleNM)
{
de.ElementName = d.node.InnerText.Replace("ELEMENT NAME:", "").Trim();
}
if (d.type == ElementType.Desc)
{
de.Description = d.node.InnerText.Replace("DESCRIPTION:", "").Trim();
}
if (d.type == ElementType.Frmt)
{
int indexOfDataType = 0;
int indexOfUnits = 0;
int indexOfWidth = 0;
indexOfDataType = d.node.InnerText.IndexOf("Data Type:");
indexOfUnits = d.node.InnerText.IndexOf("Units:");
indexOfWidth = d.node.InnerText.IndexOf("Width:");
List <KeyValue> kv = new List <KeyValue>();
KeyValue kvDataType = new KeyValue();
KeyValue kvUnits = new KeyValue();
KeyValue kvWidth = new KeyValue();
kvDataType.Attr = d.node.InnerText.Substring(indexOfDataType, 9).Trim();
kvDataType.Value = d.node.InnerText.Substring(indexOfDataType + 10, indexOfUnits - (indexOfDataType + 11)).Trim();
kvUnits.Attr = d.node.InnerText.Substring(indexOfUnits, 5).Trim();
kvUnits.Value = d.node.InnerText.Substring(indexOfUnits + 6, indexOfWidth - (indexOfUnits + 7)).Trim();
kvWidth.Attr = d.node.InnerText.Substring(indexOfWidth, 5).Trim();
kvWidth.Value = d.node.InnerText.Substring(indexOfWidth + 6).Trim();
kv.Add(kvDataType);
kv.Add(kvUnits);
kv.Add(kvWidth);
de.CodeFormat = kv;
}
if (d.type == ElementType.Clas)
{
bool lastLoopAdded = false;
List <KeyValue> kv = new List <KeyValue>();
KeyValue kvInstance;
List <string> para = d.node.InnerHtml.GetParagraphsListFromHtml();
try
{
for (int i = 1; i < para.Count(); i++)
{
if (!string.IsNullOrWhiteSpace(para[i]))
{
if (!lastLoopAdded)
{
kvInstance = new KeyValue();
kvInstance.Attr = para[i];
kvInstance.Value = para[i + 1];
if (!kv.Contains(kvInstance))
{
kv.Add(kvInstance);
}
lastLoopAdded = true;
}
else
{
lastLoopAdded = false;
}
}
}
}
catch {
kv.Add(new KeyValue {
Attr = "SystemIssue", Value = "Error reading" + d.type.ToString()
});
}
de.Classification = kv;
}
if (d.type == ElementType.CoNt)
{
de.CodingNotes = d.node.InnerText;
}
if (d.type == ElementType.InFi)
{
int indexOfVERSION = 0;
List <string> para = d.node.InnerHtml.GetParagraphsListFromHtml();
List <string> inputFiles = new List <string>();
foreach (string xyz in para)
{
if (xyz.Equals("VERSION"))
{
indexOfVERSION = para.IndexOf(xyz);
break;
}
string[] excludeWords = new string[]
{ "INPUT FILES:",
"HEP - Student",
"HEP - Staff",
"HEP - Applications and Offers" };
if (!excludeWords.Contains(xyz))
{
if (!inputFiles.Contains(xyz))
{
inputFiles.Add(xyz);
}
}
}
de.InputFiles = inputFiles;
List <ChangeRecord> changeHist = new List <ChangeRecord>();
ChangeRecord cr;
try
{
for (int i = indexOfVERSION + 3; i < para.Count();)
{
if (!string.IsNullOrWhiteSpace(para[i]))
{
cr = new ChangeRecord();
cr.Version = para[i];
cr.RevisionDate = para[i + 1];
cr.ReportingYear = para[i + 2];
changeHist.Add(cr);
i = i + 2;
}
}
}
catch {}
de.ChangeHistory = changeHist;
}
if (d.type == ElementType.CHis)
{
System.Console.WriteLine(d.type.ToString());
System.Console.WriteLine(d.node.InnerHtml);
}
}
return(de);
}
public override bool Equals(object obj)
{
KeyValue item = (KeyValue)obj;
return(this.Attr.Equals(item.Attr) && this.Value.Equals(item.Value));
}