private void ProcessDataTypeColumn(HL7SchemaRepository repo, string hl7Version, DataType dt)
{
HtmlNode scrapedData = null;
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Scraping table information for data type {0} version {1}.", dt.Name, hl7Version));
try
{
string url = string.Format("http://hl7-definition.caristix.com:9010/Default.aspx?version=HL7%20v{0}&dataType={1}", hl7Version, dt.Name);
var Webget = new HtmlWeb();
HtmlDocument doc = Webget.Load(url);
scrapedData = doc.DocumentNode.SelectSingleNode("//table[@id='grdResult']");
}
catch (Exception ex)
{
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Failed to scrape DataTypeColumn for version {0} column {1}. ERROR: {2}", hl7Version, dt.Name, ex.Message));
}
if (scrapedData != null)
{
foreach (HtmlNode node in scrapedData.ChildNodes)
{
if (node.Name == "tr")
{
bool isValid = true;
const string tdPattern = @"<td\b[^>]*?>(?<V>[\s\S]*?)</\s*td>";
StringBuilder strNode = new StringBuilder();
foreach (Match match in Regex.Matches(node.InnerHtml, tdPattern, RegexOptions.IgnoreCase))
{
string value = match.Groups["V"].Value;
strNode.Append(value + "|");
}
if (strNode.Length > 0)
{
strNode.Remove(strNode.Length - 1, 1);
DataTypeColumn newDataTypeColumn = new DataTypeColumn();
try
{
string[] splitData = strNode.ToString().Split('|');
newDataTypeColumn.Sequence = splitData[0].Split('.')[1].Parse <long>();
newDataTypeColumn.Length = splitData[1].Parse <long>();
newDataTypeColumn.IsOptional = bool.Parse(splitData[3] == "O" ? "true" : "false");
newDataTypeColumn.Name = splitData[5];
newDataTypeColumn.DataType1 = dt;
const string aPattern = @"<a\b[^>]*?>(?<V>[\s\S]*?)</\s*a>";
foreach (Match match in Regex.Matches(splitData[2], aPattern, RegexOptions.IgnoreCase))
{
string value = match.Groups["V"].Value;
newDataTypeColumn.DataType = DataTypeMap(value);
}
}
catch (Exception ex)
{
_fileLogger.LogMessage(LogMessageType.ERROR, ex.Message);
isValid = false;
}
if (isValid)
{
repo.AddNew <DataTypeColumn>(newDataTypeColumn);
}
}
}
}
}
}
private void CreateSegment(string version, string segment)
{
HL7SchemaRepository repo = new HL7SchemaRepository();
HtmlNodeCollection scrapedSegmentData = null;
try
{
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Starting segment {0}, version {1}", segment, version));
string url = string.Format("http://hl7-definition.caristix.com:9010/Default.aspx?version=HL7%20v{0}&segment={1}", version, segment);
var Webget = new HtmlWeb();
var doc = Webget.Load(url);
scrapedSegmentData = doc.DocumentNode.SelectNodes("/html[1]/body[1]/div[5]/div[1]/div[3]/table[1]/tr");
}
catch (Exception ex)
{
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Failure scraping segment {0}, version {1}. ERROR: {2}", segment, version, ex.Message));
}
if (scrapedSegmentData != null)
{
foreach (HtmlNode node in scrapedSegmentData)
{
if (node.Name == "tr")
{
Segment s = new Segment();
s.SegmentId = segment;
s.Version = version;
bool isValid = true;
const string tdPattern = @"<td\b[^>]*?>(?<V>[\s\S]*?)</\s*td>";
StringBuilder strNode = new StringBuilder();
foreach (Match match in Regex.Matches(node.InnerHtml, tdPattern, RegexOptions.IgnoreCase))
{
string value = match.Groups["V"].Value;
strNode.Append(value + "|");
}
if (strNode.Length > 0)
{
strNode.Remove(strNode.Length - 1, 1);
try
{
string[] splitData = strNode.ToString().Split('|');
//0 Sequence string, split on .
string[] sequence = splitData[0].Split('.');
s.Sequence = long.Parse(sequence[1]);
//index 1 = length
s.Length = long.Parse(splitData[1]);
//index 2 - the type. needs to run agains regex to get
const string aPattern = @"<a\b[^>]*?>(?<V>[\s\S]*?)</\s*a>";
foreach (Match match in Regex.Matches(splitData[2], aPattern, RegexOptions.IgnoreCase))
{
string value = match.Groups["V"].Value;
s.DataType = DataTypeMap(value);
}
//index 3 optional
s.IsRequired = bool.Parse(splitData[3] == "R" ? "true" : "false");
//index 4 is repeating
s.IsRepeating = bool.Parse(splitData[4] == "*" ? "true" : "false");
//index 6 name
s.Name = splitData[6];
}
catch (Exception ex)
{
_fileLogger.LogMessage(LogMessageType.ERROR, ex.Message);
isValid = false;
}
if (isValid)
{
repo.AddNew <Segment>(s);
}
}
}
}
}
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Ending segment {0}, version {1}", segment, version));
}
private void ScrapeDataTypeBy(string hl7Version)
{
HtmlNode scrapedData = null;
try
{
string url = string.Format("http://hl7-definition.caristix.com:9010/Default.aspx?version=HL7%20v{0}", hl7Version);
var Webget = new HtmlWeb();
HtmlDocument doc = Webget.Load(url);
scrapedData = doc.DocumentNode.SelectSingleNode("//select[@id='cbxDataType']");
}
catch (Exception ex)
{
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Failed to scrape DataType for version {0}. ERROR: {1}", hl7Version, ex.Message));
}
if (scrapedData != null)
{
Parallel.ForEach(scrapedData.ChildNodes, option => {
if (option.InnerText != "- Choose a data type" & option.InnerText != "\r\n\t" & !string.IsNullOrEmpty(option.InnerText))
{
bool isValid = true;
HL7SchemaRepository repo = new HL7SchemaRepository();
DataType dt = new DataType();
try
{
dt.Name = option.InnerText;
dt.Version = hl7Version;
}
catch (Exception ex)
{
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Error creating DataType {0} version {1}. ERROR: {2}", option.InnerText, hl7Version, ex.Message));
isValid = false;
}
if (isValid)
{
repo.AddNew <DataType>(dt);
_fileLogger.LogMessage(LogMessageType.INFO, string.Format("Created DataType {0} version {1}", option.InnerText, hl7Version));
}
repo.Dispose();
}
});
#region standard for each
//foreach (HtmlNode option in scrapedData.ChildNodes)
//{
// if (option.InnerText != "- Choose a data type" & option.InnerText != "\r\n\t" & !string.IsNullOrEmpty(option.InnerText))
// {
// bool isValid = true;
// HL7SchemaRepository repo = new HL7SchemaRepository();
// DataType dt = new DataType();
// try
// {
// dt.Name = option.InnerText;
// dt.Version = hl7Version;
// }
// catch (Exception ex)
// {
// _fileLogger.LogMessage(LogMessageType.INFO, string.Format("Error creating DataType {0} version {1}. ERROR: {2}", option.InnerText, hl7Version, ex.Message));
// isValid = false;
// }
// if (isValid)
// {
// repo.AddNew<DataType>(dt);
// _fileLogger.LogMessage(LogMessageType.INFO, string.Format("Created DataType {0} version {1}", option.InnerText, hl7Version));
// }
// repo.Dispose();
// }
//}
#endregion
}
}