public void ParseString_ReturnElementWithText_WhenHaveTextAfterChildren(string html)
{
// Arrange
// Act
HtmlNodeElement element = (HtmlNodeElement)htmlParser.ParseString(html).FirstOrDefault();
// Assert
Assert.Single(element.Texts());
Assert.Equal("Text", element.Texts()[0].HtmlString.Value);
Assert.Equal(element.Children.LastOrDefault().UId, element.Texts()[0].AfterElementUId);
}
public void ParseString_ReturnElementWithText_WhenTagHaveTextWhitoutChildren(string html)
{
// Arrange
// Act
HtmlNodeElement element = (HtmlNodeElement)htmlParser.ParseString(html).FirstOrDefault();
// Assert
Assert.NotEmpty(element.Texts());
Assert.Equal("Text", element.Texts()[0].HtmlString.Value);
Assert.Null(element.Texts()[0].AfterElementUId);
}
public void ParseString_ReturnElementWithText_WhenHaveTextBetweenChildren(string html)
{
// Arrange
// Act
HtmlNodeElement element = (HtmlNodeElement)htmlParser.ParseString(html).FirstOrDefault();
// Assert
Assert.Equal(3, element.Texts().Length);
Assert.Equal("Text", element.Texts()[0].HtmlString.Value);
Assert.Equal("Text1", element.Texts()[1].HtmlString.Value);
Assert.Equal("Text2", element.Texts()[2].HtmlString.Value);
}
/// <summary>
/// Convert html document string to HtmlDocument.
/// </summary>
/// <param name="html">Valid html document string.</param>
/// <returns cref="HtmlDocument">HtmlDocument instance represent given html string.</returns>
public HtmlDocument ParsePage(string html)
{
if (html.IsNullOrEmpty())
{
throw new ArgumentNullException("html");
}
IHtmlElementsCollection elements = ParseString(html);
HtmlNodeElement element = (HtmlNodeElement)elements.FirstOrDefault(x => x.TagName == "html");
HtmlDocument document = new HtmlDocument();
document.Attributes = element.Attributes;
document.Children = element.Children;
document.Text(element.Texts());
document.Doctype = (HtmlDocTypeElement)elements.FirstOrDefault(x => x.TagName.IsEqualIgnoreCase("!DOCTYPE"));
return(document);
}
/// <summary>
/// Create html content.
/// </summary>
/// <param name="htmlElement">Object of type HtmlNodeElement, for convert to standart HTML document content.</param>
/// <returns>Html content repsresent specified node element.</returns>
public IHtmlContent CreateHtmlContent(HtmlNodeElement htmlElement)
{
if (htmlElement.Children.IsNullOrEmpty())
{
return(new HtmlString(
$"{CreateStartTag(htmlElement.TagName)}{htmlElement.Text()}{CreateEndTag(htmlElement.TagName)}"
.Insert((htmlElement.TagName.Length + 1), htmlElement.Attributes.IsNullOrEmpty() ? "" : $" {String.Join(" ", htmlElement.Attributes.Select(x => CreateAtribute(x)))}")));
}
List <IHtmlContent> childContents = new List <IHtmlContent>();
List <HtmlNodeElementText> knownPossitionedTexts = new List <HtmlNodeElementText>();
foreach (var text in htmlElement.Texts())
{
if (text.AfterElementUId == null || !htmlElement.Children.Any(x => x.UId == text.AfterElementUId))
{
childContents.Add(text.HtmlString);
}
else
{
knownPossitionedTexts.Add(text);
}
}
foreach (var child in htmlElement.Children)
{
childContents.Add(CreateHtmlContent(child));
if (knownPossitionedTexts.Any(x => x.AfterElementUId == child.UId))
{
childContents.Add(knownPossitionedTexts.FirstOrDefault(x => x.AfterElementUId == child.UId).HtmlString);
}
}
return(new HtmlString(
$"{CreateStartTag(htmlElement.TagName)}{String.Join("", childContents.Select(x => x.ToString()).ToArray())}{CreateEndTag(htmlElement.TagName)}"
.Insert((htmlElement.TagName.Length + 1), htmlElement.Attributes.IsNullOrEmpty() ? "" : $" {String.Join(" ", htmlElement.Attributes.Select(x => CreateAtribute(x)))}")));
}