public void FixRelativeUrl()
{
var absoluteUrl =
UriUtils.CanonicalizeUrl("?aa", "http://www.dianping.com/sh/ss/com");
Assert.Equal("http://www.dianping.com/sh/ss/com?aa", absoluteUrl);
absoluteUrl =
UriUtils.CanonicalizeUrl("../aa", "http://www.dianping.com/sh/ss/com");
Assert.Equal("http://www.dianping.com/sh/aa", absoluteUrl);
absoluteUrl = UriUtils.CanonicalizeUrl("..aa", "http://www.dianping.com/sh/ss/com");
Assert.Equal("http://www.dianping.com/sh/ss/..aa", absoluteUrl);
absoluteUrl =
UriUtils.CanonicalizeUrl("../../aa", "http://www.dianping.com/sh/ss/com/");
Assert.Equal("http://www.dianping.com/sh/aa", absoluteUrl);
absoluteUrl =
UriUtils.CanonicalizeUrl("../../aa", "http://www.dianping.com/sh/ss/com");
Assert.Equal("http://www.dianping.com/aa", absoluteUrl);
// 只有相对路径需要做补充
var selectable1 = new HtmlSelectable("<div><a href=\"/a/b\">aaaaaaab</a></div>", "http://www.b.com");
var value1 = selectable1.Links().First();
Assert.Equal("http://www.b.com/a/b", value1);
// 绝对路径不需要做补充
var selectable2 = new HtmlSelectable("<div><a href=\"http://www.aaaa.com\">aaaaaaab</a></div>",
"http://www.b.com", false);
var value2 = selectable2.SelectList(Selectors.XPath(".//a")).First().Value;
Assert.Equal("aaaaaaab", value2);
}
public void DoNotFixAllRelativeHrefs()
{
var selectable = new HtmlSelectable("<div><a href=\"aaaa.com\">aaaaaaab</a></div>");
var values = selectable.SelectList(Selectors.XPath(".//a")).ToArray();
Assert.Equal("aaaaaaab", values.First().Value);
}
public void RemoveOutboundLinks()
{
// 绝对路径不需要做补充
var selectable2 = new HtmlSelectable("<div><a href=\"http://www.aaaa.com\">aaaaaaab</a></div>",
"http://www.b.com");
var value2 = selectable2.SelectList(Selectors.XPath(".//a"));
Assert.Null(value2);
}
public void XPath2()
{
var selectable = new HtmlSelectable(html2);
var links1 = selectable.Links();
var divs = selectable.SelectList(Selectors.XPath("//div")).ToList();
var link2 = divs[0].Links().ToList();
var link3 = divs[1].Links().ToList();
Assert.Equal(2, links1.Count());
Assert.Single(link2);
Assert.Single(link3);
Assert.Equal("http://whatever.com/aaa", link2[0]);
Assert.Equal("http://whatever.com/bbb", link3[0]);
}