public void TestDeserializeDocument()
{
var serializer = new HSerializer();
var hdoc = serializer.DeserializeDocument(new StringReader("<html><body><h1>Document</h1><p>Content & more &entity;.</p></body></html>"));
Assert.Same(Encoding.UTF8, hdoc.Encoding);
// Document with one root
Assert.Equal(1, hdoc.Nodes().Count());
Assert.Equal("html", hdoc.Root.Name);
// Root with one body
Assert.Equal(1, hdoc.Root.Nodes().Count());
HElement body = hdoc.Root.FirstNode as HElement;
Assert.NotNull(body);
Assert.Equal("body", body.Name);
// Body contains two elements
Assert.Equal(2, body.Nodes().Count());
var elms = body.Elements().ToArray();
Assert.Equal(2, elms.Length);
// First h1
Assert.Equal("h1", elms[0].Name);
Assert.Equal(1, elms[0].Nodes().Count());
Assert.IsType <HText>(elms[0].FirstNode);
Assert.Equal("Document", ((HText)elms[0].FirstNode).Value);
// Second p
Assert.Equal("p", elms[1].Name);
Assert.Equal(1, elms[1].Nodes().Count());
Assert.IsType <HText>(elms[1].FirstNode);
Assert.Equal("Content & more &entity;.", ((HText)elms[1].FirstNode).Value);
// Test from stream
using (var ms = new MemoryStream(Encoding.ASCII.GetBytes("<html><body><h1>Document</h1><p>Content & more &entity;.</p></body></html>")))
{
serializer.RemoveUnknownOrInvalidEntities = true;
hdoc = serializer.DeserializeDocument(new StreamReader(ms, Encoding.ASCII));
Assert.Same(Encoding.ASCII, hdoc.Encoding);
Assert.Equal(6, hdoc.DescendantNodes().Count());
Assert.Equal(0, hdoc.ParseErrors.Length);
HElement elm = hdoc.Root.Elements("body").First().Elements().Last();
Assert.Equal("p", elm.Name);
Assert.Equal(1, elm.Nodes().Count());
Assert.IsType <HText>(elm.FirstNode);
Assert.Equal("Content & more .", ((HText)elm.FirstNode).Value);
}
Assert.Throws <ArgumentNullException>(() => serializer.DeserializeDocument(null));
}
public void TestDeserializeDocument_WithErrors()
{
var serializer = new HSerializer();
var hdoc = serializer.DeserializeDocument(new StringReader("<html><body><h1>Document</h1><p class=>Content & more.</p></body></html>"));
Assert.Same(Encoding.UTF8, hdoc.Encoding);
// Document with one root
Assert.Equal(1, hdoc.Nodes().Count());
Assert.Equal("html", hdoc.Root.Name);
// Root with one body
Assert.Equal(1, hdoc.Root.Nodes().Count());
HElement body = hdoc.Root.FirstNode as HElement;
Assert.NotNull(body);
Assert.Equal("body", body.Name);
// Body contains two elements
Assert.Equal(2, body.Nodes().Count());
var elms = body.Elements().ToArray();
Assert.Equal(2, elms.Length);
// First h1
Assert.Equal("h1", elms[0].Name);
Assert.Equal(1, elms[0].Nodes().Count());
Assert.IsType <HText>(elms[0].FirstNode);
Assert.Equal("Document", ((HText)elms[0].FirstNode).Value);
// Second p
Assert.Equal("p", elms[1].Name);
Assert.Equal(1, elms[1].Nodes().Count());
Assert.IsType <HText>(elms[1].FirstNode);
Assert.Equal("Content & more.", ((HText)elms[1].FirstNode).Value);
// Check errors
Assert.Equal(1, hdoc.ParseErrors.Length);
Assert.Equal("Attribute value expected.", hdoc.ParseErrors[0].Message);
}
public void TestDeserializeTestPage1()
{
String pageContent;
using (var pageStream = this.GetType().Assembly.GetManifestResourceStream("HDoc.Tests.Resources.TestPage1.html"))
using (var reader = new StreamReader(pageStream))
pageContent = reader.ReadToEnd();
using (var pageStream = this.GetType().Assembly.GetManifestResourceStream("HDoc.Tests.Resources.TestPage1.html"))
using (var reader = new StreamReader(pageStream))
{
var serializer = new HSerializer();
var doc = serializer.DeserializeDocument(reader);
Assert.Same(Encoding.UTF8, doc.Encoding);
Assert.Null(doc.XmlDeclaration);
Assert.Equal(StandardDoctype.Html5, doc.DocumentType.StandardType);
var nodes = doc.Nodes().ToArray();
Assert.Equal(3, nodes.Length);
Assert.IsType <HDocumentType>(nodes[0]);
Assert.IsType <HText>(nodes[1]);
Assert.IsType <HElement>(nodes[2]);
nodes = doc.DescendantNodes().ToArray();
Assert.Equal(129, nodes.Length);
var elms = doc.Descendants().ToArray();
Assert.Equal(46, elms.Length);
// Correct false HTML source
String tmp = pageContent
.Replace("src=\"http://placekitten.com/g/64/64\">", "src=\"http://placekitten.com/g/64/64\" />")
.Replace("a = b < c;", "a = b < c;")
.Replace("s = \"<html>\";", "s = "<html>";")
.Replace("français", "français")
;
Assert.Equal(tmp, serializer.Serialize(doc));
}
}
