give div's in article content the same weight as p's

This commit is contained in:
2014-02-16 13:52:12 +01:00
parent 9cc55170b7
commit b6c610ec85
2 changed files with 49 additions and 42 deletions
@@ -1300,12 +1300,18 @@ namespace ReadSharp.Ports.NReadability
/* Do these last as the previous stuff may have removed junk that will affect these. */
CleanConditionally(articleContentElement, "table");
CleanConditionally(articleContentElement, "ul");
CleanConditionally(articleContentElement, "div", true);
//CleanConditionally(articleContentElement, "div", true);
/* Remove extra paragraphs. */
IEnumerable<XElement> paraElements = articleContentElement.GetElementsByTagName("p");
List<XElement> paraElements = articleContentElement.GetElementsByTagName("p").ToList();
IEnumerable<XElement> divElements = articleContentElement.GetElementsByTagName("div");
var elementsToRemove = new List<XElement>();
if (divElements != null)
{
paraElements.AddRange(divElements);
}
foreach (XElement paraElement in paraElements)
{
string innerText = GetInnerText(paraElement, false);