public void Parse(Stream stream, Action <BracketPipeTextFragment> callback)
{
bool skip = false;
if (Distinct)
{
cache = new FastLookupCache <string>(CacheSize);
}
else
{
cache = null;
}
var filter = GetFilter(Filter);
if (filter == null)
{
throw new FetchoException("{0} is an invalid filter type", Filter);
}
var tag = new Stack <string>();
using (var reader = new HtmlReader(stream))
{
foreach (var c in reader)
{
switch (c.Type)
{
case HtmlTokenType.StartTag:
if (c.Value == ScriptHtmlTag)
{
skip = true;
}
if (c.Value == StyleHtmlTag)
{
skip = true;
}
tag.Push(c.Value);
break;
case HtmlTokenType.EndTag:
if (c.Value == ScriptHtmlTag)
{
skip = false;
}
if (c.Value == StyleHtmlTag)
{
skip = false;
}
if (tag.Count > 0)
{
tag.Pop();
}
break;
default:
break;
}
if (!skip && c.Type == HtmlTokenType.Text)
{
if (c.Value.Length.IsBetween(MinimumLength, MaximumLength))
{
if (cache == null || !cache.Contains(c.Value))
{
cache?.Enqueue(c.Value);
string tagvalue = string.Empty;
if (tag.Count > 0)
{
tagvalue = tag.Peek();
}
var fragment = new BracketPipeTextFragment(tagvalue, c.Value);
if (filter(fragment))
{
callback(fragment);
}
}
}
}
}
}
}
private bool GenericTagFilter(string[] validTags, BracketPipeTextFragment fragment) => validTags.Any(x => x == fragment.Tag);
