public List <ViseonHeaderModel> GetViseonElements(HtmlDocument doc)
{
try
{
var headers = new List <ViseonHeaderModel>();
foreach (var type in ViseonStaticData.HeaderElements)
{
var list = doc.DocumentNode.GetElementsByTagName(type);
foreach (var item in list)
{
headers.Add(new ViseonHeaderModel()
{
HtmlTagName = item.Name,
Text = item.InnerText,
WordCount = WordCounting.CountWords(item.InnerText),
});
}
}
return(headers);
}
catch (Exception ex)
{
return(new List <ViseonHeaderModel>());
}
}
public ViseonTitleModel GetViseonElement(HtmlDocument doc)
{
try
{
var list = doc.DocumentNode.GetElementsByTagName(ViseonStaticData.TitleElement);
var title = list.FirstOrDefault();
// no title found
if (title == null)
{
return new ViseonTitleModel()
{
HtmlTagName = ViseonStaticData.TitleElement,
Text = "", WordCount = 0
}
}
;
return(new ViseonTitleModel()
{
Text = title.InnerText,
WordCount = WordCounting.CountWords(title.InnerText),
HtmlTagName = ViseonStaticData.TitleElement,
CharacterCount = title.InnerText.Length
});
}
catch (Exception e)
{
Console.WriteLine(e);
throw;
}
}
public ViseonDescriptionModel GetViseonElement(HtmlDocument doc)
{
try
{
var list = doc.DocumentNode.GetElementsByTagName(ViseonStaticData.MetaElements);
HtmlNode desc = null;
foreach (var htmlNode in list)
{
if (htmlNode.Attributes[ViseonStaticData.Meta.NameProp] == null)
{
continue;
}
if (!String.Equals(htmlNode.Attributes[ViseonStaticData.Meta.NameProp].Value, ViseonStaticData.MetaDescription, StringComparison.CurrentCultureIgnoreCase))
{
continue;
}
desc = htmlNode;
break;
}
// no desc
if (desc == null)
{
return new ViseonDescriptionModel()
{
HtmlTagName = ViseonStaticData.MetaDescription,
Text = "",
WordCount = 0
}
}
;
var text = desc.Attributes[ViseonStaticData.Meta.ContentProp]?.Value;
return(new ViseonDescriptionModel()
{
Text = text ?? "",
WordCount = text == null ? 0 : WordCounting.CountWords(text),
HtmlTagName = ViseonStaticData.MetaDescription,
CharacterCount = text?.Length ?? 0
});
}
catch (Exception e)
{
Console.WriteLine(e);
throw;
}
}
public List <ViseonParagraphModel> GetViseonElements(HtmlDocument doc)
{
try
{
var paras = doc.DocumentNode.GetElementsByName(ViseonStaticData.ParaElement);
var result = new List <ViseonParagraphModel>();
foreach (var htmlNode in paras)
{
result.Add(new ViseonParagraphModel()
{
HtmlTagName = ViseonStaticData.ParaElement,
Text = htmlNode.InnerText,
WordCount = WordCounting.CountWords(htmlNode.InnerText)
});
}
return(result);
}
catch (Exception ex)
{
return(new List <ViseonParagraphModel>());
}
}