private void AddItem(ParseDataItem item)
{
if (_items == null)
{
_items = new List <ParseDataItem>();
}
_items.Add(item);
}
private void AddChild(ParseDataItem child)
{
if (child != null)
{
if (_childItems == null)
{
_childItems = new List <ParseDataItem>();
}
_childItems.Add(child);
}
}
public void Load(string filePath)
{
FileName = Path.GetFileName(filePath);
FilePath = Path.GetFullPath(filePath);
using (FileStream fs = File.Open(filePath, FileMode.Open, FileAccess.Read, FileShare.ReadWrite))
{
using (BufferedStream bs = new BufferedStream(fs))
{
using (StreamReader sr = new StreamReader(bs))
{
bool insideBody = false;
while (sr.Peek() >= 0)
{
string tag = string.Empty;
if (!insideBody)
{
do
{
tag = StreamReaderExtensions.ReadUntil(sr, '>');
if (tag.ToLower().Contains("<body"))
{
insideBody = true;
}
} while (!insideBody);
}
ParseDataItem dataItem = new ParseDataItem(sr, this, null);
if (dataItem.HasData)
{
AddItem(dataItem);
}
}
sr.Dispose();
}
bs.Dispose();
}
fs.Dispose();
}
}
public ParseDataItem(StreamReader sr, ParseDocument parentDocument, ParseDataItem parentDataItem)
{
ParentDocument = parentDocument;
ParentDataItem = parentDataItem;
string nextTag = string.Empty;
int associatedChildCount = 0;
Stack <string> tagStack = new Stack <string>();
do
{
while (sr.Peek() > 0)
{
char firstChar;
if (string.IsNullOrEmpty(nextTag))
{
firstChar = (char)sr.Read();
while (sr.Peek() > 0 && char.IsControl(firstChar))
{
firstChar = (char)sr.Read();
}
}
else
{
firstChar = nextTag[0];
}
string toTest = string.Empty;
string toTestNode = firstChar.Equals('/') ? nextTag : string.Empty;
if (!string.IsNullOrEmpty(toTestNode))
{
if (toTestNode.Contains(" ") || toTestNode.StartsWith("br"))
{
toTest = toTestNode.Substring(toTestNode.StartsWith("/") ? 1 : 0, toTestNode.StartsWith("/") ? toTestNode.IndexOf(" ") - 1 : toTestNode.IndexOf(" ")).Trim();
}
else
{
toTest = toTestNode.Substring(toTestNode.StartsWith("/") ? 1 : 0, toTestNode.StartsWith("/") ? toTestNode.IndexOf(">") - 1 : toTestNode.IndexOf(">")).Trim();
}
}
if (firstChar.Equals('<') ||
((firstChar.Equals('/') && tagStack.Peek() != DATA_TAG) ||
(firstChar.Equals('/') && tagStack.Peek() == DATA_TAG && HasData && (!string.IsNullOrEmpty(toTest) && !tagsToProcess.Contains(toTest))) ||
(firstChar.Equals('/') && tagStack.Peek() == DATA_TAG && HasData && reservedHeaders.Contains(HTMLDecodedHeader.ToUpper().Trim())) ||
(firstChar.Equals('/') && tagStack.Peek() == DATA_TAG && !reservedHeaders.Contains(HTMLDecodedHeader.ToUpper().Trim()))))
{
string tagNode = string.IsNullOrEmpty(nextTag) ? StreamReaderExtensions.ReadUntil(sr, '>') : nextTag.StartsWith("<") ? nextTag.Substring(1) : nextTag;
nextTag = string.Empty;
bool isOpenNode = !(firstChar.Equals('/') || tagNode.StartsWith("/"));
string tag = string.Empty;
if (tagNode.Contains(" ") || tagNode.StartsWith("br"))
{
tag = tagNode.Substring(tagNode.StartsWith("/") ? 1 : 0, tagNode.StartsWith("/") ? tagNode.IndexOf(" ") - 1 : tagNode.IndexOf(" ")).Trim();
}
else
{
tag = tagNode.Substring(tagNode.StartsWith("/") ? 1 : 0, tagNode.StartsWith("/") ? tagNode.IndexOf(">") - 1 : tagNode.IndexOf(">")).Trim();
}
if (tagStack.Any() && tagStack.Peek().Trim().ToLower() == DATA_TAG && tag.Trim().ToLower() == TABLE_TAG)
{
if (isOpenNode)
{
ParseDataItem newChild = new ParseDataItem(sr, ParentDocument, this);
if (newChild.HasData)
{
AddChild(newChild);
associatedChildCount++;
}
}
}
else
{
if (tagsToProcess.Contains(tag))
{
if (isOpenNode)
{
tagStack.Push(tag);
}
else
{
if (tagStack.Peek().Equals(tag))
{
tagStack.Pop();
}
if (tag.Equals(DATA_TAG))
{
if (HTMLDecodedValues != null && HTMLDecodedValues.Any())
{
ValueCounts.Last().AssociatedChildCount = associatedChildCount;
}
associatedChildCount = 0;
}
}
}
if (!tagStack.Any())
{
break;
}
}
}
else
{
if (tagStack != null && tagStack.Count > 0)
{
bool done = false;
int dataLoopCount = 0;
do
{
string control = Int32.TryParse(firstChar.ToString(), out int temp) || firstChar.ToString().ToUpper().Equals("X") ||
firstChar.ToString().ToUpper().Equals("Y") ? string.Empty : firstChar + "<";
string text = dataLoopCount == 0 ? firstChar + StreamReaderExtensions.ReadUntil(sr, '<') : StreamReaderExtensions.ReadUntil(sr, '<');
if (!text.Equals(control) && !text.StartsWith("br /") && !text.Equals("<"))
{
text = new string(text.Substring(0, text.Length - 1).Where(c => !char.IsControl(c)).ToArray());
if (!string.IsNullOrEmpty(text))
{
if (tagStack.Peek().Trim().ToLower() == HEADER_TAG)
{
HTMLDecodedHeader = text;
}
else if (tagStack.Peek().Trim().ToLower() == DATA_TAG)
{
if (HTMLDecodedValues != null && HTMLDecodedValues.Any())
{
ValueCounts.Last().AssociatedChildCount = associatedChildCount;
}
associatedChildCount = 0;
AddValue(text);
}
}
nextTag = StreamReaderExtensions.ReadUntil(sr, '>');
if (!nextTag.StartsWith("br /") && !nextTag.StartsWith("br/"))
{
done = true;
}
dataLoopCount++;
}
else
{
nextTag = StreamReaderExtensions.ReadUntil(sr, '>');
string tempTag = nextTag;
tempTag = tempTag.Replace(">", "");
if (tagsToProcess.Contains(tempTag))
{
nextTag = "<" + nextTag;
}
done = true;
}
} while (!done);
}
}
}
} while (tagStack.Any() && sr.Peek() > 0);
}
public ParseDataItem(ref string fileContents, ParseDocument parentDocument, ParseDataItem parentDataItem)
{
ParentDocument = parentDocument;
ParentDataItem = parentDataItem;
Stack <string> tags = new Stack <string>();
do
{
RemoveUnnecessaryTags(ref fileContents);
int toRemoveLength = 0;
if (fileContents.StartsWith("<"))
{
string tagNode = fileContents.Substring(0, fileContents.IndexOf(">") + 1);
bool isOpenNode = !tagNode.StartsWith("</");
toRemoveLength = tagNode.Length;
string tag = string.Empty;
if (tagNode.Contains(" "))
{
tag = tagNode.Substring(1, tagNode.IndexOf(" ")).Trim();
}
else
{
tag = tagNode.Substring(1, tagNode.IndexOf(">") - 1).Trim();
}
if (tags.Any() && tags.Peek().Trim().ToUpper() == DATA_TAG && tag.Trim().ToUpper() == TABLE_TAG)
{
AddChild(new ParseDataItem(ref fileContents, ParentDocument, this));
}
else
{
if (isOpenNode)
{
tags.Push(tag);
}
else
{
tags.Pop();
}
fileContents = fileContents.Substring(toRemoveLength);
}
}
else
{
do
{
RemoveUnnecessaryTags(ref fileContents);
string text = fileContents.Substring(0, fileContents.IndexOf("<"));
toRemoveLength = text.Length;
if (tags.Peek().Trim().ToUpper() == HEADER_TAG)
{
HTMLDecodedHeader = text;
}
else if (tags.Peek().Trim().ToUpper() == DATA_TAG)
{
AddValue(text);
}
fileContents = fileContents.Substring(toRemoveLength);
RemoveUnnecessaryTags(ref fileContents);
} while (!fileContents.StartsWith("<"));
}
RemoveUnnecessaryTags(ref fileContents);
} while (tags.Any());
}
