public static void SetInnerHtml(this XElement element, string html)
{
if (element == null)
{
throw new ArgumentNullException(nameof(element));
}
if (html == null)
{
throw new ArgumentNullException(nameof(html));
}
element.RemoveAll();
var tmpElement = new SgmlDomBuilder().BuildDocument(html);
if (tmpElement.Root == null)
{
return;
}
foreach (var node in tmpElement.Root.Nodes())
{
element.Add(node);
}
}
/// <summary>
/// Initializes a new instance of NReadabilityTranscoder. Allows setting all options.
/// </summary>
/// <param name="dontStripUnlikelys">Determines whether elements that are unlikely to be a part of main content will be removed.</param>
/// <param name="dontNormalizeSpacesInTextContent">Determines whether spaces in InnerText properties of elements will be normalized automatically (eg. whether double spaces will be replaced with single spaces).</param>
/// <param name="dontWeightClasses">Determines whether 'weight-class' algorithm will be used when cleaning content.</param>
/// <param name="readingStyle">Styling for the extracted article.</param>
/// <param name="readingMargin">Margin for the extracted article.</param>
/// <param name="readingSize">Font size for the extracted article.</param>
private NReadabilityTranscoder(
bool dontStripUnlikelys,
bool dontNormalizeSpacesInTextContent,
bool dontWeightClasses,
ReadingStyle readingStyle,
ReadingMargin readingMargin,
ReadingSize readingSize)
{
_dontStripUnlikelys = dontStripUnlikelys;
_dontNormalizeSpacesInTextContent = dontNormalizeSpacesInTextContent;
_dontWeightClasses = dontWeightClasses;
_readingStyle = readingStyle;
_readingMargin = readingMargin;
_readingSize = readingSize;
_sgmlDomBuilder = new SgmlDomBuilder();
_sgmlDomSerializer = new SgmlDomSerializer();
_elementsScores = new Dictionary<XElement, float>();
}
private static void AssertHtmlContentIsEmpty(string content)
{
if (content != null)
{
content = content.Trim();
}
var document = new SgmlDomBuilder().BuildDocument(content);
int count = 0;
foreach (var node in document.DescendantNodes())
{
var element = node as XElement;
if (element != null)
{
var name = element.Name.LocalName;
if (!"html".Equals(name, StringComparison.OrdinalIgnoreCase)
&& !"head".Equals(name, StringComparison.OrdinalIgnoreCase)
&& !"meta".Equals(name, StringComparison.OrdinalIgnoreCase)
&& !"body".Equals(name, StringComparison.OrdinalIgnoreCase))
{
count++;
}
}
}
Assert.AreEqual(0, count);
}
static NReadabilityTranscoderTests_Old()
{
_sgmlDomBuilder = new SgmlDomBuilder();
_sgmlDomSerializer = new SgmlDomSerializer();
}
public static void SetInnerHtml(this XElement element, string html)
{
if (element == null)
{
throw new ArgumentNullException("element");
}
if (html == null)
{
throw new ArgumentNullException("html");
}
element.RemoveAll();
var tmpElement = new SgmlDomBuilder().BuildDocument(html);
if (tmpElement.Root == null)
{
return;
}
foreach (var node in tmpElement.Root.Nodes())
{
element.Add(node);
}
}
/// <summary>
/// Initializes a new instance of NReadabilityTranscoder. Allows setting all options.
/// </summary>
/// <param name="dontStripUnlikelys">Determines whether elements that are unlikely to be a part of main content will be removed.</param>
/// <param name="dontNormalizeSpacesInTextContent">Determines whether spaces in InnerText properties of elements will be normalized automatically (eg. whether double spaces will be replaced with single spaces).</param>
/// <param name="dontWeightClasses">Determines whether 'weight-class' algorithm will be used when cleaning content.</param>
/// <param name="divIdHints">针对特定网站抽取特定div id=的节点作为内容块 </param>
private NReadabilityTranscoder(
bool dontStripUnlikelys,
bool dontNormalizeSpacesInTextContent,
bool dontWeightClasses,
Dictionary<Regex, string> divIdHints = null)
{
_dontStripUnlikelys = dontStripUnlikelys;
_dontNormalizeSpacesInTextContent = dontNormalizeSpacesInTextContent;
_dontWeightClasses = dontWeightClasses;
_sgmlDomBuilder = new SgmlDomBuilder();
_sgmlDomSerializer = new SgmlDomSerializer();
_elementsScores = new Dictionary<XElement, float>();
_articleContentDivIdHints2 = divIdHints;
}