public static void Test1()
{
string HtmlCode = "<html><head></head><body><div><a id='abcd'>Test</a><p id='defgh'><a>test2<a/>test3</p><a>1</a><a>2</a><a>3</a><a>4</a></div></body></html>";
HTMLInterface TestInterface = new HTMLInterface(HtmlCode);
TestInterface.GetFirstObject();
List<ObjectAttribute> RepresentAAttributes1 = new List<ObjectAttribute>();
RepresentAAttributes1.Add(new ObjectAttributeString("name", "A"));
RepresentAAttributes1.Add(new ObjectAttributeRegex("id", "[a-z]+"));
RepresentAAttributes1.Add(new ObjectAttributeRegex("innerText", "[A-Z]+"));
List<ObjectAttribute> RepresentAAttributes2 = new List<ObjectAttribute>();
RepresentAAttributes2.Add(new ObjectAttributeString("name", "A"));
RepresentAAttributes2.Add(new ObjectAttributeRegex("id", "[a-z]+"));
RepresentAAttributes2.Add(new ObjectAttributeRegex("innerText", "Test"));
List<ObjectAttribute> RepresentAAttributes3 = new List<ObjectAttribute>();
RepresentAAttributes3.Add(new ObjectAttributeString("name", "A"));
RepresentAAttributes3.Add(new ObjectAttributeRegex("innerText", "test2"));
List<ObjectAttribute> RepresentAAttributes4 = new List<ObjectAttribute>();
RepresentAAttributes4.Add(new ObjectAttributeString("name", "A"));
RepresentAAttributes4.Add(new ObjectAttributeRegex("innerText", "\\d+"));
List<ObjectAttribute> RepresentPAttributes = new List<ObjectAttribute>();
RepresentPAttributes.Add(new ObjectAttributeString("name", "P"));
RepresentPAttributes.Add(new ObjectAttributeRegex("id", "def[a-z]+"));
ParserObject RepresentA1 = new ParserObject(RepresentAAttributes1);
ParserObject RepresentA2 = new ParserObject(RepresentAAttributes2);
ParserObject RepresentA3 = new ParserObject(RepresentAAttributes3);
ParserObject RepresentA4 = new ParserObject(RepresentAAttributes4);
ParserObject RepresentP = new ParserObject(RepresentPAttributes);
RepresentP.AddSubParserPrimitive(RepresentA3);
List<ObjectAttribute> RepresentDIVAttributes = new List<ObjectAttribute>();
RepresentDIVAttributes.Add(new ObjectAttributeString("name", "DIV"));
ParserObject RepresentDIV = new ParserObject(RepresentDIVAttributes);
ParserObjectRelation RelationAND = new ParserObjectRelation();
RelationAND.SetRelationFunction(new RelationFunctionAND(ref RelationAND, null));
RelationAND.AddSubParserPrimitive(RepresentA1);
RelationAND.AddSubParserPrimitive(RepresentA2);
ParserObjectRelation RelationOR = new ParserObjectRelation();
RelationOR.SetRelationFunction(new RelationFunctionOR(ref RelationOR, null));
RelationOR.AddSubParserPrimitive(RelationAND);
RelationOR.AddSubParserPrimitive(RepresentP);
ParserObjectRelation RelationMORE = new ParserObjectRelation();
RelationMORE.SetRelationFunction(new RelationFunctionMORE(ref RelationMORE, null));
RelationMORE.AddSubParserPrimitive(RepresentA4);
RepresentDIV.AddSubParserPrimitive(RelationOR);
RepresentDIV.AddSubParserPrimitive(RepresentP);
RepresentDIV.AddSubParserPrimitive(RelationMORE);
ObjectInterfacePrototipe tInterface = (ObjectInterfacePrototipe)TestInterface;
FunctionIO Result = RepresentDIV.CheckRelations(ref tInterface, null, true);
Result.GetType();
}
public static void Test2()
{
string HtmlCode = "<html><head></head><body><div><div>a</div><div>b</div><div>c</div><p>a</p><p>b</p><p>c</p><a>test</a></div></body></html>";
HTMLInterface TestInterface = new HTMLInterface(HtmlCode);
TestInterface.GetFirstObject();
List<ObjectAttribute> RepresentAAttributes1 = new List<ObjectAttribute>();
RepresentAAttributes1.Add(new ObjectAttributeRegex("name", ".+"));
//RepresentAAttributes1.Add(new ObjectAttributeRegex("innerText", "[a-z]"));
ParserObject RepresentA1 = new ParserObject(RepresentAAttributes1);
List<ObjectAttribute> RepresentPAttributes = new List<ObjectAttribute>();
RepresentPAttributes.Add(new ObjectAttributeRegex("name", "P"));
RepresentPAttributes.Add(new ObjectAttributeRegex("innerText", "[a-z]"));
ParserObject RepresentP = new ParserObject(RepresentPAttributes);
List<ObjectAttribute> RepresentAAttributes2 = new List<ObjectAttribute>();
RepresentAAttributes2.Add(new ObjectAttributeString("name", "A"));
RepresentAAttributes2.Add(new ObjectAttributeRegex("innerText", "[a-z]+"));
ParserObject RepresentA2 = new ParserObject(RepresentAAttributes2);
ObjectAttributeEvaluation evaluation = new ObjectAttributeEvaluation(EvaluationMode.Max, 1, 4);
evaluation.AttributeName="ParseCount";
FunctionIO RelationMOREInput = new FunctionIO();
RelationMOREInput.AddFunctionParameter(evaluation);
TestCallback CallbackClass = new TestCallback();
ParserObjectRelation RelationMORE = new ParserObjectRelation();
RelationMORE.SetRelationFunction(new RelationFunctionMORE(ref RelationMORE, RelationMOREInput));
RelationMORE.AddSubParserPrimitive(RepresentA1);
RelationMORE.SetCallback(CallbackClass.Call);
ParserObjectRelation RelationMORE2 = new ParserObjectRelation();
RelationMORE2.SetRelationFunction(new RelationFunctionMORE(ref RelationMORE2, RelationMOREInput));
RelationMORE2.AddSubParserPrimitive(RepresentP);
RelationMORE2.SetCallback(CallbackClass.Call);
List<ObjectAttribute> RepresentDIVAttributes = new List<ObjectAttribute>();
RepresentDIVAttributes.Add(new ObjectAttributeString("name", "DIV"));
ParserObject RepresentDIV = new ParserObject(RepresentDIVAttributes);
RepresentDIV.AddSubParserPrimitive(RelationMORE);
RepresentDIV.AddSubParserPrimitive(RelationMORE2);
RepresentDIV.AddSubParserPrimitive(RepresentA2);
ObjectInterfacePrototipe tInterface = (ObjectInterfacePrototipe)TestInterface;
FunctionIO Result = RepresentDIV.CheckRelations(ref tInterface, null, true);
Result.GetType();
}
private ParserPrimitives ParseNode( XmlNode ElementToParse )
{
XmlNodeType NodeType= ElementToParse.NodeType;
if (NodeType != XmlNodeType.Element)
return null;
ParserPrimitives ParserPrimitive = null;
if (ElementToParse.Name == "object")
{
ParserPrimitive = new ParserObject();
List<ObjectAttribute> Attributes= new List<ObjectAttribute>();
foreach( XmlAttribute Attribute in ElementToParse.Attributes )
{
if (Attribute.Name == "callback")
{
ParserPrimitive.SetCallback(cCallbackStore.GetCallbackEntryByName(Attribute.Value).GetCallbackFunction());
continue;
}
Attributes.Add(new ObjectAttributeRegex(Attribute.Name, Attribute.Value));
}
((ParserObject)ParserPrimitive).SetObjectAttributes(Attributes);
}
else if (ElementToParse.Name == "relation")
{
ParserPrimitive = new ParserObjectRelation();
string RelationFunctionName="";
int min_evaluation = 0, max_evaluation = 0;
EvaluationMode evaluation_mode= EvaluationMode.Max;
FunctionIO FunctionInput = new FunctionIO();
foreach (XmlAttribute Attribute in ElementToParse.Attributes)
{
if (Attribute.Name == "callback")
{
ParserPrimitive.SetCallback(cCallbackStore.GetCallbackEntryByName(Attribute.Value).GetCallbackFunction());
continue;
}
else if (Attribute.Name == "relation_function")
{
RelationFunctionName = Attribute.Value;
continue;
}
else if (Attribute.Name == "evaluation_min")
{
min_evaluation = System.Int32.Parse(Attribute.Value);
continue;
}
else if (Attribute.Name == "evaluation_max")
{
min_evaluation = System.Int32.Parse(Attribute.Value);
continue;
}
else if (Attribute.Name == "evaluation_mode")
{
switch (Attribute.Value)
{
case "min":
evaluation_mode = EvaluationMode.Min;
break;
case "max":
evaluation_mode = EvaluationMode.Max;
break;
}
continue;
}
FunctionInput.AddFunctionParameter( new ObjectAttributeString(Attribute.Name, Attribute.Value) );
}
ObjectAttributeEvaluation evaluation= new ObjectAttributeEvaluation( evaluation_mode, min_evaluation, max_evaluation );
FunctionInput.AddFunctionParameter( evaluation );
RelationFunction tRelationFunction= null;
ParserObjectRelation Relation = (ParserObjectRelation)ParserPrimitive;
switch( RelationFunctionName )
{
case "more":
tRelationFunction = new RelationFunctionMORE(ref Relation, FunctionInput);
break;
case "and":
tRelationFunction = new RelationFunctionAND(ref Relation, FunctionInput);
break;
case "or":
tRelationFunction = new RelationFunctionOR(ref Relation, FunctionInput);
break;
}
Relation.SetRelationFunction(tRelationFunction);
}
return ParserPrimitive;
}
/// <summary>
/// Relation function constructor.
/// </summary>
/// <param name="lParserObjectRelation">Parent ObjectRelation.</param>
/// <param name="lFunctionParameters">Input parameters for relation function.</param>
public RelationFunctionAND(ref ParserObjectRelation lParserObjectRelation, FunctionIO lFunctionParameters)
: base(ref lParserObjectRelation, lFunctionParameters)
{
}
public RelationFunction CreateRelationFunctionByName( string RelationFunctionName, ParserObjectRelation lParserObjectRelation, FunctionIO lFunctionInput )
{
if (cFunctionEntries[RelationFunctionName] != null)
{
//Creates a new instance of specifficed class.
return (RelationFunction)Activator.CreateInstance(cFunctionEntries[RelationFunctionName].AssociatedType, new object[] { lParserObjectRelation, lFunctionInput });
}
return null;
}
public RelationFunction(ref ParserObjectRelation lParserObjectRelation, FunctionIO lFunctionParameters)
{
cParserObjectRelation = lParserObjectRelation;
cFunctionParameters = lFunctionParameters;
}
