internal IEnumerable<XElement> FindCandidatesForArticleContent(XDocument document)
{
var paraElements = document.GetElementsByTagName("p");
var candidateElements = new HashSet<XElement>();
_elementsScores.Clear();
foreach (var paraElement in paraElements)
{
string innerText = GetInnerText(paraElement);
if (innerText.Length < _MinParagraphLength)
{
continue;
}
var parentElement = paraElement.Parent;
var grandParentElement = parentElement != null ? parentElement.Parent : null;
int score = 1; // 1 point for having a paragraph
// Add points for any comma-segments within this paragraph.
score += GetSegmentsCount(innerText, ',');
// For every PARAGRAPH_SEGMENT_LENGTH characters in this paragraph, add another point. Up to MAX_POINTS_FOR_SEGMENTS_COUNT points.
score += Math.Min(innerText.Length / _ParagraphSegmentLength, _MaxPointsForSegmentsCount);
// Add the score to the parent.
if (parentElement != null && (parentElement.Name == null || !"html".Equals(parentElement.Name.LocalName, StringComparison.OrdinalIgnoreCase)))
{
candidateElements.Add(parentElement);
AddPointsToElementScore(parentElement, score);
}
// Add half the score to the grandparent.
if (grandParentElement != null && (grandParentElement.Name == null || !"html".Equals(grandParentElement.Name.LocalName, StringComparison.OrdinalIgnoreCase)))
{
candidateElements.Add(grandParentElement);
AddPointsToElementScore(grandParentElement, score / 2);
}
}
return candidateElements;
}
private static void ResolveElementsUrls(XDocument document, string tagName, string attributeName, string url, Func<AttributeTransformationInput, AttributeTransformationResult> attributeValueTransformer)
{
if (document == null)
{
throw new ArgumentNullException("document");
}
if (string.IsNullOrEmpty(url))
{
throw new ArgumentNullException("url");
}
var elements = document.GetElementsByTagName(tagName);
foreach (var element in elements)
{
var attributeValue = element.GetAttributeValue(attributeName, null);
if (attributeValue == null)
{
continue;
}
attributeValue = ResolveElementUrl(attributeValue, url);
if (!string.IsNullOrEmpty(attributeValue))
{
AttributeTransformationResult attributeTransformationResult;
if (attributeValueTransformer != null)
{
attributeTransformationResult = attributeValueTransformer.Invoke(new AttributeTransformationInput { AttributeValue = attributeValue, Element = element });
}
else
{
attributeTransformationResult = new AttributeTransformationResult { TransformedValue = attributeValue };
}
element.SetAttributeValue(attributeName, attributeTransformationResult.TransformedValue);
if (!string.IsNullOrEmpty(attributeTransformationResult.OriginalValueAttributeName))
{
element.SetAttributeValue(attributeTransformationResult.OriginalValueAttributeName, attributeValue);
}
}
}
}
internal void GlueDocument(XDocument document, XElement articleTitleElement, XElement articleContentElement)
{
var documentBody = GetOrCreateBody(document);
/* Include readability.css stylesheet. */
var headElement = document.GetElementsByTagName("head").FirstOrDefault();
if (headElement == null)
{
headElement = new XElement("head");
documentBody.AddBeforeSelf(headElement);
}
var styleElement = new XElement("style");
styleElement.SetAttributeValue("type", "text/css");
var readabilityStylesheetStream = Assembly.GetExecutingAssembly().GetManifestResourceStream(_ReadabilityStylesheetResourceName);
if (readabilityStylesheetStream == null)
{
throw new InternalErrorException("Couldn't load the NReadability stylesheet embedded resource.");
}
using (var sr = new StreamReader(readabilityStylesheetStream))
{
styleElement.SetInnerHtml(sr.ReadToEnd());
}
headElement.Add(styleElement);
/* Apply reading style to body. */
string readingStyleClass = GetReadingStyleClass(_readingStyle);
documentBody.SetClass(readingStyleClass);
documentBody.SetStyle("display: block;");
/* Create inner div. */
var innerDiv = new XElement("div");
innerDiv.SetId(InnerDivId);
innerDiv.SetClass(GetReadingMarginClass(_readingMargin) + " " + GetReadingSizeClass(_readingSize));
if (articleTitleElement != null)
{
innerDiv.Add(articleTitleElement);
}
if (articleContentElement != null)
{
innerDiv.Add(articleContentElement);
}
/* Create overlay div. */
var overlayDiv = new XElement("div");
overlayDiv.SetId(OverlayDivId);
overlayDiv.SetClass(readingStyleClass);
overlayDiv.Add(innerDiv);
/* Clear the old HTML, insert the new content. */
documentBody.RemoveAll();
documentBody.Add(overlayDiv);
}
private static XElement TryFindArticleContentElement(XDocument document, string articleContentElementHint)
{
if (document == null)
{
throw new ArgumentNullException("document");
}
if (string.IsNullOrEmpty(articleContentElementHint))
{
throw new ArgumentException("Argument can't be null nor empty.", "articleContentElementHint");
}
return document
.GetElementsByTagName(articleContentElementHint)
.FirstOrDefault();
}
internal void GlueDocument(XDocument document, XElement articleTitleElement, XElement articleContentElement)
{
XElement documentBody = GetOrCreateBody(document);
/* Include readability.css stylesheet. */
XElement headElement = document.GetElementsByTagName("head").FirstOrDefault();
if (headElement == null)
{
headElement = new XElement("head");
documentBody.AddBeforeSelf(headElement);
}
XElement styleElement = new XElement("style");
styleElement.SetAttributeValue("type", "text/css");
Stream readabilityStylesheetStream = typeof(NReadabilityTranscoder).GetTypeInfo().Assembly.GetManifestResourceStream(_ReadabilityStylesheetResourceName);
if (readabilityStylesheetStream == null)
{
throw new InternalErrorException("Couldn't load the NReadability stylesheet embedded resource.");
}
using (var sr = new StreamReader(readabilityStylesheetStream))
{
styleElement.SetInnerHtml(sr.ReadToEnd());
}
headElement.Add(styleElement);
/* Apply reading style to body. */
string readingStyleClass = GetReadingStyleClass(_readingStyle);
documentBody.SetClass(readingStyleClass);
documentBody.SetStyle("display: block;");
documentBody.RemoveAll();
if (articleTitleElement != null)
{
documentBody.Add(articleTitleElement);
}
if (articleContentElement != null)
{
documentBody.Add(articleContentElement);
}
}
using System;
