public void can_process_links_from_this_orgin()
{
_pageElementProcessor = new ATagLinksProcessor(_resourceFactoryMock.Object, _relativaPathProducerMock.Object, _resourceStackMock.Object,
1, InitialHostUri, false);
string testPage = @"<html><head>
</head>
<body>
<a href=""http://www.example.com/blog.html"">
<a href=""http://www.notexample.com/about/contacts"">
</body>
</html>";
HtmlDocument document = new HtmlDocument();
document.LoadHtml(testPage);
_relativaPathProducerMock.Setup(
x =>
x.GetRelativePath(It.Is<Uri>(d => d == InitialHostUri),
It.Is<Uri>(d => d == new Uri("http://www.example.com/blog.html")))).Returns("blog.html");
var resultDocument = _pageElementProcessor.ProcessPage(document, InitialHostUri);
Assert.That(resultDocument.DocumentNode.InnerHtml.Contains("href=\"blog.html\""));
Assert.That(resultDocument.DocumentNode.InnerHtml.Contains("http://www.notexample.com/about/contacts"));
_resourceFactoryMock.Verify(x => x.GetHtmlModel(It.IsAny<Uri>(), 0), Times.Exactly(1));
_relativaPathProducerMock.Verify(x => x.GetRelativePath(It.Is<Uri>(d => d == InitialHostUri), It.IsAny<Uri>()), Times.Exactly(1));
_resourceStackMock.Verify(x => x.AddResourceToQueue(It.IsAny<PageResource>()), Times.Exactly(1));
}
public void should_change_only_already_processed_if_deep_equals_zero()
{
_pageElementProcessor = new ATagLinksProcessor(_resourceFactoryMock.Object, _relativaPathProducerMock.Object, _resourceStackMock.Object,
0, InitialHostUri, true);
string testPage = @"<html><head>
</head>
<body>
<a href=""http://www.example.com/blog.html"">
<a href=""http://www.notexample.com/about/contacts"">
</body>
</html>";
HtmlDocument document = new HtmlDocument();
document.LoadHtml(testPage);
_resourceStackMock.Setup(x => x.WasNotProcessed(new Uri("http://www.notexample.com/about/contacts")))
.Returns(false);
_relativaPathProducerMock.Setup(
x =>
x.GetRelativePath(It.Is<Uri>(d => d == InitialHostUri),
It.Is<Uri>(d => d == new Uri("http://www.notexample.com/about/contacts")))).Returns("..\\www.notexample.com\\about\\contacts.html");
var resultDocument = _pageElementProcessor.ProcessPage(document, InitialHostUri);
Assert.That(resultDocument.DocumentNode.InnerHtml.Contains("href=\"http://www.example.com/blog.html\""));
Assert.That(resultDocument.DocumentNode.InnerHtml.Contains("href=\"..\\www.notexample.com\\about\\contacts.html"));
_resourceFactoryMock.Verify(x => x.GetHtmlModel(It.IsAny<Uri>(), 0), Times.Never());
_relativaPathProducerMock.Verify(x => x.GetRelativePath(It.Is<Uri>(d => d == InitialHostUri), It.IsAny<Uri>()), Times.Exactly(2));
_resourceStackMock.Verify(x => x.AddResourceToQueue(It.IsAny<PageResource>()), Times.Exactly(0));
}
