private void TrimNodeOnStart(HtmlNode node)
{
var textNode = node as HtmlText;
if (textNode != null)
{
TrimNodeOnStart(textNode);
}
if (node is HtmlComment && settings.RemoveComments)
{
var comment = (HtmlComment)node;
bool keepComment = false;
foreach (var regex in settings.KeepCommentsRegex)
{
var result = regex.Match(comment.Slice.Text, comment.Slice.Start, comment.Slice.Length);
if (result.Success)
{
keepComment = true;
break;
}
}
// Don't remove special ignoring comments
if (!keepComment)
{
node.Remove();
}
}
// Remove HTML script
if (settings.RemoveJavaScript && node is HtmlElement && ((HtmlElement)node).Descriptor?.Name == "script")
{
node.Remove();
}
// If current node requires preserving formatting inside it we need to trim all pending text node that we collected before
var nodeName = node is HtmlElement ? ((HtmlElement)node).Name : null;
if (settings.CollapseWhitespaces && !string.IsNullOrEmpty(nodeName) && settings.TagsWithNonCollapsableWhitespaces.ContainsKey(nodeName))
{
TrimPendingTextNodes();
}
}
private void TrimNodeOnStart(HtmlNode node)
{
var textNode = node as HtmlText;
if (textNode != null)
{
TrimNodeOnStart(textNode);
}
if (node is HtmlComment && settings.RemoveComments)
{
var comment = (HtmlComment)node;
bool keepComment = false;
foreach (var regex in settings.KeepCommentsRegex)
{
var result = regex.Match(comment.Slice.Text, comment.Slice.Start, comment.Slice.Length);
if (result.Success)
{
keepComment = true;
break;
}
}
// Don't remove special ignoring comments
if (!keepComment)
{
node.Remove();
}
}
// Remove HTML script
if (settings.RemoveJavaScript && node is HtmlElement && ((HtmlElement)node).Descriptor.Name == "script")
{
node.Remove();
}
}
private void TrimNodeOnStart(HtmlNode node)
{
var textNode = node as HtmlText;
if (textNode != null)
{
TrimNodeOnStart(textNode);
}
if (node is HtmlComment && settings.RemoveComments)
{
var comment = (HtmlComment)node;
// Don't remove special ignoring comments
if (!comment.Slice.StartsWith("!"))
{
node.Remove();
}
}
}
