See the question and my original answer on StackOverflow

There is rarely one unique solution to an html matching problem. Although your solution works fine now and with your sample, the //div expression will search all div elements under root, recursively.

It means if the original Html evolves somehow, you may catch too many things or analyze too many nodes (performance may be an issue with things like // for big documents).

I would suggest something like this, which is more discriminant:

        HtmlDocument doc = new HtmlDocument();

        foreach (HtmlNode node in doc.DocumentNode.SelectNodes("/div/div[starts-with(text(), 'title')]/following-sibling::text()[normalize-space(.) != '']"))

It means

  • Search div elements from the root
  • Then search div elements underneath
  • Filter (using [...]) these elements and only select those with an inner text that start with 'title'
  • Then search all following sibling elements of type text
  • Filter these elements and only select those which are not empty or whitespace only

See this link for some help on XPATH Axes.