internal static async Task <IEnumerable <Common.IDocument> > ProcessElementsAsync(
Common.IDocument input,
IExecutionContext context,
string querySelector,
bool first,
Action <Common.IDocument, IExecutionContext, IElement, Dictionary <string, object> > processElement)
{
// Parse the HTML content
IHtmlDocument htmlDocument = await input.ParseHtmlAsync(context, HtmlParser);
if (htmlDocument == null)
{
return(input.Yield());
}
// Evaluate the query selector
try
{
if (!string.IsNullOrWhiteSpace(querySelector))
{
IElement[] elements = first
? new[] { htmlDocument.QuerySelector(querySelector) }
: htmlDocument.QuerySelectorAll(querySelector).ToArray();
if (elements.Length > 0 && elements[0] != null)
{
INode clone = htmlDocument.Clone(true); // Clone the document so we know if it changed
Dictionary <string, object> metadata = new Dictionary <string, object>();
foreach (IElement element in elements)
{
processElement(input, context, element, metadata);
}
if (htmlDocument.Equals(clone))
{
// Elements were not edited so return the original document or clone it with new metadata
return(metadata.Count == 0 ? input.Yield() : input.Clone(metadata).Yield());
}
// Elements were edited so get the new content
using (Stream contentStream = await context.GetContentStreamAsync())
{
using (StreamWriter writer = contentStream.GetWriter())
{
htmlDocument.ToHtml(writer, ProcessingInstructionFormatter.Instance);
writer.Flush();
IContentProvider contentProvider = context.GetContentProvider(contentStream, MediaTypes.Html);
return(metadata.Count == 0
? input.Clone(contentProvider).Yield()
: input.Clone(metadata, contentProvider).Yield());
}
}
}
}
return(input.Yield());
}
catch (Exception ex)
{
context.LogWarning("Exception while processing HTML for {0}: {1}", input.ToSafeDisplayString(), ex.Message);
return(input.Yield());
}
}
