public HtmlExtractor Seek(IElementPredicate predicate)
{
lastMatch = null;
SeekWithin(predicate, null);
return this;
}
public HtmlExtractor Seek(IElementPredicate predicate)
{
lastMatch = null;
SeekWithin(predicate, null);
return(this);
}
/// <summary>
/// Does not do tag balancing.
/// </summary>
public string CollectHtmlUntilPredicate(IElementPredicate elementp)
{
StringBuilder result = new StringBuilder();
Element el;
while (null != (el = Next()))
{
if (elementp.IsMatch(el))
{
break;
}
result.Append(html, el.Offset, el.Length);
}
return(result.ToString());
}
public HtmlExtractor MatchNext(IElementPredicate predicate, bool ignoreWhitespace)
{
lastMatch = null;
Element e;
do
{
e = parser.Next();
}while (e != null && ignoreWhitespace && IsWhitespaceOrZeroLengthText(e));
if (e != null && predicate.IsMatch(e))
{
lastMatch = e;
}
return(this);
}
public HtmlExtractor SeekWithin(IElementPredicate predicate, IElementPredicate withinPredicate)
{
lastMatch = null;
Element e;
while (null != (e = parser.Next()))
{
if (predicate.IsMatch(e))
{
lastMatch = e;
break;
}
if (withinPredicate != null && withinPredicate.IsMatch(e))
{
break;
}
}
return(this);
}
public SmartPredicate(string criterion)
{
actualPredicate = HtmlExtractor.Parse(criterion);
}
public PredicatePair(IElementPredicate match, IElementPredicate stop)
{
_match = match;
_stop = stop;
}
/// <summary>
/// Does not do tag balancing.
/// </summary>
public string CollectTextUntilPredicate(IElementPredicate elementp)
{
return(HtmlUtils.HTMLToPlainText(CollectHtmlUntilPredicate(elementp)));
}
public HtmlExtractor SeekWithin(IElementPredicate predicate, IElementPredicate withinPredicate)
{
lastMatch = null;
Element e;
while (null != (e = parser.Next()))
{
if (predicate.IsMatch(e))
{
lastMatch = e;
break;
}
if (withinPredicate != null && withinPredicate.IsMatch(e))
break;
}
return this;
}
public SmartPredicate(string criterion)
{
actualPredicate = HtmlExtractor.Parse(criterion);
}
public PredicatePair(IElementPredicate match, IElementPredicate stop)
{
_match = match;
_stop = stop;
}
/// <summary>
/// Does not do tag balancing.
/// </summary>
public string CollectHtmlUntilPredicate(IElementPredicate elementp)
{
StringBuilder result = new StringBuilder();
Element el;
while (null != (el = Next()))
{
if (elementp.IsMatch(el))
break;
result.Append(html, el.Offset, el.Length);
}
return result.ToString();
}
/// <summary>
/// Does not do tag balancing.
/// </summary>
public string CollectTextUntilPredicate(IElementPredicate elementp)
{
return HtmlUtils.HTMLToPlainText(CollectHtmlUntilPredicate(elementp));
}
public HtmlExtractor MatchNext(IElementPredicate predicate, bool ignoreWhitespace)
{
lastMatch = null;
Element e;
do
{
e = parser.Next();
}
while (e != null && ignoreWhitespace && IsWhitespaceOrZeroLengthText(e));
if (e != null && predicate.IsMatch(e))
lastMatch = e;
return this;
}
