public override Tag CreateTag(TagData tagData, CompositeTagData compositeTagData)
{
string link = ExtractLink(compositeTagData.StartTag, tagData.UrlBeingParsed);
int mailto = link.IndexOf("mailto");
bool mailLink = false;
if (mailto == 0)
{
// yes it is
mailto = link.IndexOf(":");
link = link.Substring(mailto + 1);
mailLink = true;
}
int javascript = link.IndexOf("javascript:");
bool javascriptLink = false;
if (javascript == 0)
{
link = link.Substring(11); // this magic number is "javascript:".length()
javascriptLink = true;
}
string accessKey = GetAccessKey(compositeTagData.StartTag);
string myLinkText = compositeTagData.Children.ToString();
LinkTag linkTag = new LinkTag(tagData, compositeTagData,
new LinkData(link, myLinkText, accessKey, mailLink, javascriptLink));
linkTag.ThisScanner = this;
return(linkTag);
}
public void Scan()
{
CreateParser("<A HREF=\"mytest.html\"><IMG SRC=\"abcd.jpg\">Hello World</A>", "http://www.yahoo.com");
// Register the image scanner
LinkScanner linkScanner = new LinkScanner("-l");
parser.AddScanner(linkScanner);
parser.AddScanner(linkScanner.CreateImageScanner("-i"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node should be a link node");
LinkTag linkTag = (LinkTag)node[0];
// Get the link data and cross-check
Node[] dataNode = new AbstractNode[10];
int i = 0;
foreach (Node nestedNode in linkTag)
{
dataNode[i++] = nestedNode;
}
Assert.AreEqual(2, i, "Number of data nodes");
Assert.IsTrue(dataNode[0] is ImageTag, "First data node should be an Image Node");
Assert.IsTrue(dataNode[1] is StringNode, "Second data node shouls be a String Node");
// Check the contents of each data node
ImageTag imageTag = (ImageTag)dataNode[0];
Assert.AreEqual("http://www.yahoo.com/abcd.jpg", imageTag.ImageURL, "Image URL");
StringNode stringNode = (StringNode)dataNode[1];
Assert.AreEqual("Hello World", stringNode.Text, "String Contents");
}
public void BadImageInLinkBug()
{
CreateParser(
"<a href=\"registration.asp?EventID=1272\"><img border=\"0\" src=\"\\images\\register.gif\"</a>",
"http://www.fedpage.com/Event.asp?EventID=1272");
parser.RegisterScanners();
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node should be a HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
// Get the image tag from the link
Node[] insideNodes = new AbstractNode[10];
int j = 0;
foreach (Node nestedNode in linkTag)
{
insideNodes[j++] = nestedNode;
}
Assert.AreEqual(1, j, "Number of contained internal nodes");
Assert.IsTrue(insideNodes[0] is ImageTag);
ImageTag imageTag = (ImageTag)insideNodes[0];
Assert.AreEqual("http://www.fedpage.com/images/register.gif", imageTag.ImageURL, "Image Tag Location");
}
public void RelativeLinkNotHTMLBug()
{
CreateParser("<A HREF=\"newpage.html\">New Page</A>", "http://www.mysite.com/books/some.asp");
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node should be a HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("http://www.mysite.com/books/newpage.html", linkTag.Link, "Link");
}
public void NotFTPLink()
{
CreateParser("<A HREF=\"ftp.html\">my ftp</A>", "http://www.cj.com/");
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node should be a HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("my ftp", linkTag.ToPlainTextString(), "Link Plain Text");
Assert.IsFalse(linkTag.FTPLink, "Link is not a FTP site");
}
public void NotJavascriptLink()
{
CreateParser("<A HREF=\"javascript_not.html\">say hello</A>", "http://www.cj.com/");
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node should be a HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("say hello", linkTag.ToPlainTextString(), "Link Plain Text");
Assert.IsFalse(linkTag.JavascriptLink, "Link is not a Javascript command");
}
public void MailtoLink()
{
CreateParser("<A HREF=\"mailto:[email protected]\">[email protected]</A>", "http://www.cj.com/");
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node should be a HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("*****@*****.**", linkTag.ToPlainTextString(), "Link Plain Text");
Assert.IsTrue(linkTag.MailLink, "Link is a mail link");
}
public void RelativeLinkScan3()
{
CreateParser("<A HREF=\"../abc/def/mytest.html\"> Hello World</A>", "http://www.yahoo.com/ghi");
// Register the image scanner
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node identified should be HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
AssertStringEquals("Expected Link", "http://www.yahoo.com/abc/def/mytest.html", linkTag.Link);
}
public void LinkContainsEqualTo()
{
CreateParser("<a class=rlbA href=/news/866201.asp?0sl=-" + "32>Shoe bomber handed life sentence</a>");
parser.RegisterScanners();
ParseAndAssertNodeCount(1);
AssertType("node type", typeof(LinkTag), node[0]);
LinkTag linkTag = (LinkTag)node[0];
AssertStringEquals("link text", "Shoe bomber handed life sentence", linkTag.LinkText);
AssertStringEquals("link url", "/news/866201.asp?0sl=-32", linkTag.Link);
}
public void ExtractLinkInvertedCommasBug2()
{
CreateParser(
"<a href=\"http://cbc.ca/artsCanada/stories/greatnorth271202\" class=\"lgblacku\">Vancouver schools plan 'Great Northern Way'</a>");
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "The node should be a link tag");
LinkTag linkTag = (LinkTag)node[0];
AssertStringEquals("Extracted Text", "Vancouver schools plan 'Great Northern Way'", linkTag.LinkText);
}
public void TagSymbolsInLinkText()
{
CreateParser("<a href=\"/cataclysm/Langy-AnEmpireReborn-Ch2.shtml#story\"" +
"><< An Empire Reborn: Chapter 2 <<</a>");
parser.RegisterScanners();
ParseAndAssertNodeCount(1);
AssertType("node", typeof(LinkTag), node[0]);
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("<< An Empire Reborn: Chapter 2 <<", linkTag.LinkText, "link text");
}
public void LinkSpacesBug()
{
CreateParser("<a href=\"http://www.kizna.com/servlets/SomeServlet?name=Sam Joseph\">Click Here</A>");
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "The node should be a link tag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("http://www.kizna.com/servlets/SomeServlet?name=Sam Joseph", linkTag.Link,
"Link URL of link tag");
Assert.AreEqual("Click Here", linkTag.LinkText, "Link Text of link tag");
}
public void LinkDataContents()
{
CreateParser(
"<a href=\"http://transfer.go.com/cgi/atransfer.pl?goto=http://www.signs.movies.com&name=114332&srvc=nws&context=283&guid=4AD5723D-C802-4310-A388-0B24E1A79689\" target=\"_new\"><img src=\"http://ad.abcnews.com/ad/sponsors/buena_vista_pictures/bvpi-ban0003.gif\" width=468 height=60 border=\"0\" alt=\"See Signs in Theaters 8-2 - Starring Mel Gibson\" align=><font face=\"verdana,arial,helvetica\" SIZE=\"1\"><b></b></font></a>",
"http://transfer.go.com");
// Register the image scanner
LinkScanner linkScanner = new LinkScanner("-l");
parser.AddScanner(linkScanner);
parser.AddScanner(linkScanner.CreateImageScanner("-i"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node 0 should be a link tag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual(
"http://transfer.go.com/cgi/atransfer.pl?goto=http://www.signs.movies.com&name=114332&srvc=nws&context=283&guid=4AD5723D-C802-4310-A388-0B24E1A79689",
linkTag.Link, "Link URL");
Assert.AreEqual("", linkTag.LinkText, "Link Text");
Node[] containedNodes = new AbstractNode[10];
int i = 0;
foreach (Node nestedNode in linkTag)
{
containedNodes[i++] = nestedNode;
}
Assert.AreEqual(5, i, "There should be 5 contained nodes in the link tag");
Assert.IsTrue(containedNodes[0] is ImageTag, "First contained node should be an image tag");
ImageTag imageTag = (ImageTag)containedNodes[0];
Assert.AreEqual("http://ad.abcnews.com/ad/sponsors/buena_vista_pictures/bvpi-ban0003.gif", imageTag.ImageURL,
"Image Location");
Assert.AreEqual("60", imageTag["HEIGHT"], "Image Height");
Assert.AreEqual("468", imageTag["WIDTH"], "Image Width");
Assert.AreEqual("0", imageTag["BORDER"], "Image Border");
Assert.AreEqual("See Signs in Theaters 8-2 - Starring Mel Gibson", imageTag["ALT"], "Image Alt");
Assert.IsTrue(containedNodes[1] is Tag, "Second contained node should be Tag");
Tag tag1 = (Tag)containedNodes[1];
Assert.AreEqual("font face=\"verdana,arial,helvetica\" SIZE=\"1\"", tag1.Text, "Tag Contents");
Assert.IsTrue(containedNodes[2] is Tag, "Third contained node should be Tag");
Tag tag2 = (Tag)containedNodes[2];
Assert.AreEqual("b", tag2.Text, "Tag Contents");
Assert.IsTrue(containedNodes[3] is EndTag, "Fourth contained node should be HTMLEndTag");
EndTag endTag1 = (EndTag)containedNodes[3];
Assert.AreEqual("b", endTag1.Text, "Fourth Tag contents");
Assert.IsTrue(containedNodes[4] is EndTag, "Fifth contained node should be HTMLEndTag");
EndTag endTag2 = (EndTag)containedNodes[4];
Assert.AreEqual("font", endTag2.Text, "Fifth Tag contents");
}
public void BrokenLink()
{
CreateParser(
"<a href=\"faq.html\">" + "<br>\n" + "<img src=\"images/46revues.gif\" " + "width=\"100\" " +
"height=\"46\" " + "border=\"0\" " + "alt=\"Rejoignez revues.org!\" " + "align=\"middle\">",
"http://www.yahoo.com");
// Register the image scanner
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node 0 should be a link tag");
LinkTag linkTag = (LinkTag)node[0];
Assert.IsNotNull(linkTag.ToString());
}
public void QueryLink()
{
CreateParser(
"<A \n" +
"HREF=\"/cgi-bin/view_search?query_text=postdate>20020701&txt_clr=White&bg_clr=Red&url=http://localhost/Testing/Report1.html\">20020702 Report 1</A>",
"http://transfer.go.com");
// Register the image scanner
parser.RegisterScanners();
ParseAndAssertNodeCount(1);
Assert.IsTrue(node[0] is LinkTag, "Node 1 should be a link tag");
LinkTag linkTag = (LinkTag)node[0];
AssertStringEquals("Resolved Link",
"http://transfer.go.com/cgi-bin/view_search?query_text=postdate>20020701&txt_clr=White&bg_clr=Red&url=http://localhost/Testing/Report1.html",
linkTag.Link);
Assert.AreEqual("20020702 Report 1", linkTag.LinkText, "Resolved Link Text");
}
public void ErroneousLinkBugFromYahoo2()
{
CreateParser(
"<td>" + "<a href=s/8741>" +
"<img src=\"http://us.i1.yimg.com/us.yimg.com/i/i16/mov_popc.gif\" height=16 width=16 border=0>" +
"</td>" + "<td nowrap> \n" + "<a href=s/7509><b>Yahoo! Movies</b></a>" + "</td>",
"http://www.yahoo.com");
parser.RegisterScanners();
Node[] linkNodes = parser.ExtractAllNodesThatAre(typeof(LinkTag));
Assert.AreEqual(2, linkNodes.Length, "number of links");
LinkTag linkTag = (LinkTag)linkNodes[0];
AssertStringEquals("Link", "http://www.yahoo.com/s/8741", linkTag.Link);
// Verify the link data
AssertStringEquals("Link Text", "", linkTag.LinkText);
}
public void ErroneousLinkBugFromYahoo()
{
CreateParser(
"<a href=s/8741>" + "<img src=\"http://us.i1.yimg.com/us.yimg.com/i/i16/mov_popc.gif\" " + "height=16 " +
"width=16 " + "border=0>" + "This is a test\n" + "<a href=s/7509>" + "<b>Yahoo! Movies</b>" + "</a>",
"http://www.yahoo.com");
parser.RegisterScanners();
ParseAndAssertNodeCount(2);
// The first node should be a Tag
Assert.IsTrue(node[0] is LinkTag, "First node should be a HTMLLinkTag");
// The second node should be a HTMLStringNode
Assert.IsTrue(node[1] is LinkTag, "Second node should be a HTMLLinkTag");
LinkTag linkTag = (LinkTag)node[0];
Assert.AreEqual("http://www.yahoo.com/s/8741", linkTag.Link, "Link");
// Verify the link data
Assert.AreEqual("This is a test\r\n", linkTag.LinkText, "Link Text");
}
public void MultipleLineBug()
{
CreateParser("<LI><font color=\"FF0000\" size=-1><b>Tech Samachar:</b></font><a \n" +
"href=\"http://ads.samachar.com/bin/redirect/tech.txt?http://www.samachar.com/tech\n" +
"nical.html\"> Journalism 3.0</a> by Rajesh Jain");
Parser.LineSeparator = "\r\n";
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(8);
Assert.IsTrue(node[6] is LinkTag, "Seventh node should be a link tag");
LinkTag linkTag = (LinkTag)node[6];
string exp = "http://ads.samachar.com/bin/redirect/tech.txt?http://www.samachar.com/technical.html";
AssertStringEquals("Link URL of link tag", exp, linkTag.Link);
Assert.AreEqual(" Journalism 3.0", linkTag.LinkText, "Link Text of link tag");
Assert.IsTrue(node[7] is StringNode, "Eight node should be a string node");
StringNode stringNode = (StringNode)node[7];
Assert.AreEqual(" by Rajesh Jain", stringNode.Text, "String node contents");
}
public virtual void VisitLinkTag(LinkTag linkTag)
{
}
