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 CompositeTagWithDeadlock()
{
CreateParser("<custom>" + "<another>something" + "</custom>" + "<custom>" + "<another>else</another>" +
"</custom>");
parser.AddScanner(new AnotherScanner(true));
CustomTag customTag = ParseCustomTag(2);
int x = customTag.ChildCount;
Assert.AreEqual(1, customTag.ChildCount, "child count");
Assert.IsFalse(customTag.EmptyXmlTag, "custom tag should not be xml end tag");
Assert.AreEqual(0, customTag.StartTag.ElementBegin, "starting loc");
Assert.AreEqual(7, customTag.StartTag.ElementEnd, "ending loc");
Assert.AreEqual(1, customTag.tagData.StartLine, "starting line position");
Assert.AreEqual(1, customTag.tagData.EndLine, "ending line position");
AnotherTag anotherTag = (AnotherTag)customTag[0];
Assert.AreEqual(1, anotherTag.ChildCount, "anotherTag child count");
StringNode stringNode = (StringNode)anotherTag[0];
AssertStringEquals("anotherTag child text", "something", stringNode.ToPlainTextString());
AssertStringEquals("first custom tag html", "<CUSTOM><ANOTHER>something</ANOTHER></CUSTOM>",
customTag.ToHtml());
customTag = (CustomTag)node[1];
AssertStringEquals("second custom tag html", "<CUSTOM><ANOTHER>else</ANOTHER></CUSTOM>", customTag.ToHtml());
}
public override void VisitStringNode(StringNode stringNode)
{
string text = stringNode.Text;
if (!preTagBeingProcessed)
{
text = Translate.Decode(text);
text = ReplaceNonBreakingSpaceWithOrdinarySpace(text);
}
textAccumulator.Append(text);
}
public void EmptyTag3()
{
string testHTML = "<html><body>text<>text</body></html>";
CreateParser(testHTML);
parser.RegisterScanners();
ParseAndAssertNodeCount(5);
Assert.IsTrue(node[2] is StringNode, "Third node should be a string node");
StringNode stringNode = (StringNode)node[2];
Assert.AreEqual("text<>text", stringNode.Text, "Third node has incorrect text");
}
public void EmptyTag6()
{
string testHTML = "<html><body>text<>\ntext</body></html>";
CreateParser(testHTML);
parser.RegisterScanners();
Parser.LineSeparator = "\r\n"; // actually a static method
ParseAndAssertNodeCount(5);
Assert.IsTrue(node[2] is StringNode, "Third node should be a string node");
StringNode stringNode = (StringNode)node[2];
string actual = stringNode.Text;
Assert.AreEqual("text<>\r\ntext", actual, "Third node has incorrect text");
}
public void ErroneousLinkBug()
{
CreateParser("<p>Site Comments?<br>" + "<a href=\"mailto:[email protected]?subject=Site Comments\">" +
"Mail Us" + "<a>" + "</p>");
parser.RegisterScanners();
ParseAndAssertNodeCount(6);
// The first node should be a Tag
Assert.IsTrue(node[0] is Tag, "First node should be a Tag");
// The second node should be a HTMLStringNode
Assert.IsTrue(node[1] is StringNode, "Second node should be a HTMLStringNode");
StringNode stringNode = (StringNode)node[1];
Assert.AreEqual("Site Comments?", stringNode.Text, "Text of the StringNode");
Assert.IsTrue(node[2] is Tag, "Third node should be a tag");
}
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 void FreshMeatBug()
{
CreateParser("<a>Revision</a>", "http://www.yahoo.com");
// Register the image scanner
parser.AddScanner(new LinkScanner("-l"));
ParseAndAssertNodeCount(3);
Assert.IsTrue(node[0] is Tag, "Node 0 should be a tag");
Tag tag = (Tag)node[0];
Assert.AreEqual("a", tag.Text, "Tag Contents");
Assert.IsTrue(node[1] is StringNode, "Node 1 should be a string node");
StringNode stringNode = (StringNode)node[1];
Assert.AreEqual("Revision", stringNode.Text, "StringNode Contents");
Assert.IsTrue(node[2] is EndTag, "Node 2 should be a string node");
EndTag endTag = (EndTag)node[2];
Assert.AreEqual("a", endTag.Text, "End Tag Contents");
}
public void CompositeTagWithOneTextChild()
{
CreateParser("<Custom>" + "Hello" + "</Custom>");
CustomTag customTag = ParseCustomTag(1);
int x = customTag.ChildCount;
Assert.AreEqual(1, customTag.ChildCount, "child count");
Assert.IsFalse(customTag.EmptyXmlTag, "custom tag should not be xml end tag");
Assert.AreEqual(0, customTag.StartTag.ElementBegin, "starting loc");
Assert.AreEqual(7, customTag.StartTag.ElementEnd, "ending loc");
Assert.AreEqual(1, customTag.tagData.StartLine, "starting line position");
Assert.AreEqual(1, customTag.tagData.EndLine, "ending line position");
Node child = customTag[0];
AssertType("child", typeof(StringNode), child);
StringNode text = (StringNode)child;
AssertStringEquals("child text", "Hello", child.ToPlainTextString());
}
public void CompositeTagWithNestedTag()
{
CreateParser("<Custom>" + "<Another>" + "Hello" + "</Another>" + "<Custom/>" + "</Custom>" + "<Custom/>");
parser.AddScanner(new CustomScanner(this));
parser.AddScanner(new AnotherScanner());
ParseAndAssertNodeCount(2);
AssertType("first node", typeof(CustomTag), this.node[0]);
AssertType("second node", typeof(CustomTag), this.node[1]);
CustomTag customTag = (CustomTag)this.node[0];
Node node = customTag[0];
AssertType("first child", typeof(AnotherTag), node);
AnotherTag anotherTag = (AnotherTag)node;
Assert.AreEqual(1, anotherTag.ChildCount, "another tag children count");
node = anotherTag[0];
AssertType("nested child", typeof(StringNode), node);
StringNode text = (StringNode)node;
Assert.AreEqual("Hello", text.ToPlainTextString(), "text");
}
public override void VisitStringNode(StringNode stringNode)
{
string stringToBeSearched = stringNode.Text.ToUpper();
if (!multipleSearchesWithinStrings && stringToBeSearched.IndexOf(stringToFind) != -1)
{
stringFound = true;
foundCount++;
}
else if (multipleSearchesWithinStrings)
{
int index = -1;
do
{
index = stringToBeSearched.IndexOf(stringToFind, index + 1);
if (index != -1)
{
foundCount++;
}
} while (index != -1);
}
}
public static string ExtractXMLData(Node node, string tagName, NodeReader reader)
{
try
{
string xmlData = "";
bool xmlTagFound = IsXMLTagFound(node, tagName);
if (xmlTagFound)
{
try
{
do
{
node = reader.ReadElement();
if (node != null)
{
if (node is StringNode)
{
StringNode stringNode = (StringNode)node;
if (xmlData.Length > 0)
{
xmlData += " ";
}
xmlData += stringNode.Text;
}
else if (!(node is org.htmlparser.tags.EndTag))
{
xmlTagFound = false;
}
}
} while (node is StringNode);
}
catch (System.Exception e)
{
throw new ParserException(
"HTMLTagScanner.extractXMLData() : error while trying to find xml tag", e);
}
}
if (xmlTagFound)
{
if (node != null)
{
if (node is org.htmlparser.tags.EndTag)
{
org.htmlparser.tags.EndTag endTag = (org.htmlparser.tags.EndTag)node;
if (!endTag.Text.Equals(tagName))
{
xmlTagFound = false;
}
}
}
}
if (xmlTagFound)
{
return(xmlData);
}
else
{
return(null);
}
}
catch (System.Exception e)
{
throw new ParserException(
"HTMLTagScanner.extractXMLData() : Error occurred while trying to extract xml tag", e);
}
}
/// <summary> Locate the StringNode within the input string, by parsing from the given position
/// </summary>
/// <param name="reader">HTML reader to be provided so as to allow reading of next line
/// </param>
/// <param name="input">Input String
/// </param>
/// <param name="position">Position to start parsing from
/// </param>
/// <param name="balance_quotes">If <code>true</code> enter ignoring state on
/// encountering quotes.
///
/// </param>
public virtual Node Find(NodeReader reader, string input, int position, bool balance_quotes)
{
StringBuilder textBuffer = new StringBuilder();
int state = BEFORE_PARSE_BEGINS_STATE;
int textBegin = position;
int textEnd = position;
int inputLen = input.Length;
char ch;
char ignore_ender = '\"';
for (int i = position; (i < inputLen && state != PARSE_COMPLETED_STATE); i++)
{
ch = input[i];
if (ch == '<' && state != PARSE_IGNORE_STATE)
{
if (BeginTag(input, i))
{
state = PARSE_COMPLETED_STATE;
textEnd = i - 1;
}
}
if (balance_quotes && (ch == '\'' || ch == '"'))
{
if (state == PARSE_IGNORE_STATE)
{
if (ch == ignore_ender)
{
state = PARSE_HAS_BEGUN_STATE;
}
}
else
{
ignore_ender = ch;
state = PARSE_IGNORE_STATE;
}
}
if (state == BEFORE_PARSE_BEGINS_STATE)
{
state = PARSE_HAS_BEGUN_STATE;
}
if (state == PARSE_HAS_BEGUN_STATE || state == PARSE_IGNORE_STATE)
{
textBuffer.Append(input[i]);
}
// Patch by Cedric Rosa
if (state == BEFORE_PARSE_BEGINS_STATE && i == inputLen - 1)
{
state = PARSE_HAS_BEGUN_STATE;
}
if (state == PARSE_HAS_BEGUN_STATE && i == inputLen - 1)
{
do
{
input = reader.GetNextLine();
if (input != null && input.Length == 0)
{
textBuffer.Append(Parser.LineSeparator);
}
} while (input != null && input.Length == 0);
if (input == null)
{
textEnd = i;
state = PARSE_COMPLETED_STATE;
}
else
{
textBuffer.Append(Parser.LineSeparator);
inputLen = input.Length;
i = -1;
}
}
}
return(StringNode.CreateStringNode(textBuffer, textBegin, textEnd, reader.Parser.ShouldDecodeNodes));
}
public virtual void VisitStringNode(StringNode stringNode)
{
}
