public void Rows_WhenCalled_AllTRElementsReturned()
{
var doc = HtmlDocumentLoader.LoadHtml(GetHtml());
var table = HtmlTable.GetByPath(doc, HtmlPath.Parse("/BODY[0]/TABLE[0]"));
Assert.That(table.Rows.Count, Is.EqualTo(4));
}
public void GetPath_SimplePath_PathFound()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div id='xx'/></body></html>");
var path = doc.GetElementById("xx").GetPath();
Assert.That(path.ToString(), Is.EqualTo("/BODY[0]/DIV[0]"));
}
public void GetPath_MultipleElementsWithSameTagName_CorrectPathWithCorrectChildrenPosFound()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div/><div/><div><P/><P/><p id='xx'/></div></body></html>");
var path = doc.GetElementById("xx").GetPath();
Assert.That(path.ToString(), Is.EqualTo("/BODY[0]/DIV[2]/P[2]"));
}
public void Smoke()
{
//arrange
var loader = new HtmlDocumentLoader();
//act/assert
Assert.That(loader, Is.Not.Null);
}
public void GetByName_FormExists_FormFound()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><form name='xx' id='xx'/></body></html>");
var form = HtmlForm.GetByName(doc, "xx");
Assert.That(form.FormElement, Is.EqualTo(doc.GetElementById("xx")));
}
public void GetByName_MultipleForms_CorrectFormFound()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div><p><form name='NO' id='NO'/></p><form name='xx' id='xx'/></div></body></html>");
var form = HtmlForm.GetByName(doc, "xx");
Assert.That(form.FormElement, Is.EqualTo(doc.GetElementById("xx")));
}
public void GetByName_NoFormWithThisName_ReturnsNull()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div id='xx'/></body></html>");
var form = HtmlForm.GetByName(doc, "not-existing");
Assert.That(form, Is.Null);
}
public void GetRoot_SomeChild_RootReturned()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div id='xx'/></body></html>");
var root = doc.GetElementById("xx").GetRoot();
Assert.That(root, Is.EqualTo(doc.Body.Parent));
}
public void Ctor_NotATableElement_Throws()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div id='xx'/></body></html>");
var element = doc.GetElementById("xx");
var ex = Assert.Throws <ArgumentException>(() => new HtmlTable(element));
Assert.That(ex.Message, Does.Contain("not a html table element"));
}
public void GetCell_WithRowAndColumn_ReturnsCorrectCell()
{
var doc = HtmlDocumentLoader.LoadHtml(GetHtml());
var table = HtmlTable.GetByPath(doc, HtmlPath.Parse("/BODY[0]/TABLE[0]"));
var element = table.GetCell(1, 2);
Assert.That(element.Id, Is.EqualTo("c12"));
}
public void GetColumn_WithCell_ReturnsCorrectTDs()
{
var doc = HtmlDocumentLoader.LoadHtml(GetHtml());
var table = HtmlTable.GetByPath(doc, HtmlPath.Parse("/BODY[0]/TABLE[0]"));
var column = table.GetColumn(doc.GetElementById("c12"));
Assert.That(column.Select(td => td.Id), Is.EquivalentTo(new[] { "c02", "c12", "c22", "c32" }));
}
public void GetRow_WithCell_ReturnsCorrectTDs()
{
var doc = HtmlDocumentLoader.LoadHtml(GetHtml());
var table = HtmlTable.GetByPath(doc, HtmlPath.Parse("/BODY[0]/TABLE[0]"));
var row = table.GetRow(doc.GetElementById("c12"));
Assert.That(row[0].Parent.Id, Is.EqualTo("row1"));
}
public void ColumnIndexOf_WithCell_ReturnsCorrectIndex()
{
var doc = HtmlDocumentLoader.LoadHtml(GetHtml());
var table = HtmlTable.GetByPath(doc, HtmlPath.Parse("/BODY[0]/TABLE[0]"));
var idx = table.ColumnIndexOf(doc.GetElementById("c12"));
Assert.That(idx, Is.EqualTo(2));
}
public void GetElementByPath_SimplePath_ElementFound()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div id='xx'/></body></html>");
var path = HtmlPath.Parse("/BODY[0]/div[0]");
var element = doc.GetElementByPath(path);
Assert.That(element, Is.EqualTo(doc.GetElementById("xx")));
}
public void GetElementByPath_MultipleElementsWithSameTagName_ElementFound()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div/><div/><div><P/><P/><p id='xx'/></div></body></html>");
var path = HtmlPath.Parse("/BODY[0]/div[2]/P[2]");
var element = doc.GetElementByPath(path);
Assert.That(element, Is.EqualTo(doc.GetElementById("xx")));
}
public void GetElementByPath_NoElementWithSuchPath_ReturnsNull()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body><div id='xx'/></body></html>");
var path = HtmlPath.Parse("/html[0]/BODY[0]/P[0]");
var element = doc.GetElementByPath(path);
Assert.That(element, Is.Null);
}
public void GetElementByPath_PathPointsToBody_BodyReturned()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body></body></html>");
var path = HtmlPath.Parse("/body[0]");
var element = doc.GetElementByPath(path);
Assert.That(element, Is.EqualTo(doc.Body));
}
public void LoadDocument()
{
//arrange
var loader = new HtmlDocumentLoader();
//act
var document = loader.LoadDocument(@"http://cnn.com");
//post
Assert.That(document, Is.Not.Null);
}
static void Main(string[] args)
{
var logger = new ConsoleLogger();
try
{
var loader = new HtmlDocumentLoader();
var repository = new CrawlerRepository();
//var crawlers = new ICrawler[] { new RabotaUaCrawler(logger), new CareersStackoverfowComCrawler(logger) };
var crawlers = new ICrawler[] { new CareersStackoverfowComCrawler(logger) };
foreach (var crawler in crawlers)
{
crawler.Crawl(loader, repository);
}
}
catch (Exception e)
{
logger.Log("FAILED exception caught in Main() method. Exception message: " + e.Message);
logger.Log(e.StackTrace);
}
}
public static IHtmlDocument LoadHtml(string html)
{
return(HtmlDocumentLoader.LoadHtml(html));
}
public void GetRoot_WithRoot_RootReturned()
{
var doc = HtmlDocumentLoader.LoadHtml("<html><body></body></html>");
Assert.That(doc.Body.GetRoot(), Is.EqualTo(doc.Body.Parent));
}
