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();
}
/// <summary>
/// Recursively check all depths.
/// </summary>
/// <param name="lObjectInterface">ObjectInterface.</param>
/// <param name="CallCallback">Do we have to call callback?</param>
/// <param name="depth">Current depth.</param>
/// <returns>FunctionIO with return parameters like ret value and parse count.</returns>
public FunctionIO CheckDeepRelations(ref ObjectInterfacePrototipe lObjectInterface, bool CallCallback, int depth)
{
ObjectAttributeEvaluation Evaluation = new ObjectAttributeEvaluation();
Evaluation.AttributeName = "ParseCount";
if (depth > MaxDepth)
return new FunctionIO(RetVal.False);
bool More = true;
int TotalMove = 0;
int OccurenciesFound = 0;
int StartOffset = lObjectInterface.GetOffset();
bool eof = false;
List<ParserPrimitives> SubPrimitives = cParserObjectRelation.GetSubParserPrimitives();
while (More)
{
int CurrentMove = 0;
int Positives = 0;
ObjectInterfacePrototipe tInterface = lObjectInterface.GetSubObjects();
FunctionIO ret2 = null;
if (tInterface != null)
{
tInterface.GetFirstObject();
if (tInterface.GetCurrentObject() != null)
{
ret2 = CheckDeepRelations(ref tInterface, CallCallback, depth + 1);
if (ret2["RetVal"].Value)
OccurenciesFound += ret2["Occurencies"].Value;
}
}
for (int x = 0; x < SubPrimitives.Count; x++)
{
CurrentMove = 0;
FunctionIO ret = SubPrimitives[x].CheckRelations(ref lObjectInterface, x + 1 == SubPrimitives.Count ? null : SubPrimitives[x + 1], false);
if (ret["RetVal"].Value)
{
CurrentMove = ((ObjectAttributeEvaluation)ret["ParseCount"]).Value;
Positives++;
}
else
break;//Stop execution if we fail.
eof = lObjectInterface.MoveByOffset(CurrentMove);
if (eof)
break;
}
//if (eof)
// break;
lObjectInterface.SetOffset(StartOffset);//We go back to start offset
eof = lObjectInterface.MoveByOffset(TotalMove);//and move for one step and continue checking from there
if (Positives == SubPrimitives.Count)
{
OccurenciesFound++;//We find a new occurency and incrise counter
//We recall our callbacks
if (CallCallback)
{
for (int x = 0; x < SubPrimitives.Count; x++)
{
CurrentMove = 0;
FunctionIO ret = SubPrimitives[x].CheckRelations(ref lObjectInterface, x + 1 == SubPrimitives.Count ? null : SubPrimitives[x + 1], CallCallback);
CurrentMove = ((ObjectAttributeEvaluation)ret["ParseCount"]).Value;
eof = lObjectInterface.MoveByOffset(CurrentMove);
}
}
}
TotalMove++;//We incrise move
lObjectInterface.SetOffset(StartOffset);//We go back to start offset
eof = lObjectInterface.MoveByOffset(TotalMove);//and move for one step and continue checking from there
if (eof)
break;
}
if (OccurenciesFound == 0)
return new FunctionIO(RetVal.False);
Evaluation.Evaluate(TotalMove);
lObjectInterface.SetOffset(StartOffset);
ObjectAttributeInt ParamOccurrenciesFound = new ObjectAttributeInt("Occurencies", OccurenciesFound);
return new FunctionIO(ParamOccurrenciesFound, Evaluation, RetVal.False);
}
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;
}
public override FunctionIO CheckRelations(ref ObjectInterfacePrototipe lObjectInterface, ParserPrimitives NextElement, bool CallCalback)
{
FunctionIO ReturnContainer = new FunctionIO();
ObjectAttributeInt ReturnValue1 = new ObjectAttributeInt("ret", 1);
ObjectAttributeInt ReturnValue0 = new ObjectAttributeInt("ret", 0);
ObjectAttributeEvaluation Evaluation = null;
if (cFunctionParameters == null || cFunctionParameters.GetFunctionParameterByName("ParseCount") == null)
{
Evaluation = new ObjectAttributeEvaluation();
Evaluation.SetAttributeName("ParseCount");
}
else
Evaluation = (ObjectAttributeEvaluation)cFunctionParameters.GetFunctionParameterByName("ParseCount");
bool More = true;
int TotalMove= 0;
int StartOffset = lObjectInterface.GetOffset();
bool eof = false;
while (More)
{
int Positives = 0;
int CurrentMove = 0;
if (NextElement != null)
{
FunctionIO ret = NextElement.CheckRelations(ref lObjectInterface, NextElement.GetNextPrimitive(), false);
if (((ObjectAttributeInt)ret.GetFunctionParameterByName("ret")).GetAttributeValue() == 1)
{
break;
}
}
List<ParserPrimitives> SubPrimitives = cParserObjectRelation.GetSubParserPrimitives();
for (int x = 0; x < SubPrimitives.Count; x++)
{
ParserPrimitives tPrimitives;
if (SubPrimitives.Count == x + 1)
tPrimitives = null;
else
tPrimitives = SubPrimitives[x + 1];
FunctionIO ret = SubPrimitives[x].CheckRelations(ref lObjectInterface, tPrimitives, CallCalback);
if (((ObjectAttributeInt)ret.GetFunctionParameterByName("ret")).GetAttributeValue() == 1)
{
CurrentMove += ((ObjectAttributeEvaluation)ret.GetFunctionParameterByName("ParseCount")).GetEvaluationValue();
Positives++;
}
eof=lObjectInterface.MoveByOffset(CurrentMove);
}
TotalMove += CurrentMove;
if (Positives != SubPrimitives.Count || eof)
{
More = false;
}
}
if (!Evaluation.EvaluateNext(TotalMove))
{
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue0);
return ReturnContainer;
}
lObjectInterface.SetOffset(StartOffset);
ReturnContainer.AddFunctionParameter(Evaluation);
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue1);
return ReturnContainer;
}
public override FunctionIO CheckRelations(ref ObjectInterfacePrototipe lObjectInterface, ParserPrimitives NextElement, bool CallCalback)
{
FunctionIO ReturnContainer = new FunctionIO();
ObjectAttributeInt ReturnValue1 = new ObjectAttributeInt("ret", 1);
ObjectAttributeInt ReturnValue0 = new ObjectAttributeInt("ret", 0);
ObjectAttributeEvaluation Evaluation= null;
if (cFunctionParameters == null || cFunctionParameters.GetFunctionParameterByName("ParseCount") == null)
{
Evaluation = new ObjectAttributeEvaluation();
Evaluation.SetAttributeName("ParseCount");
}
else
Evaluation = (ObjectAttributeEvaluation)cFunctionParameters.GetFunctionParameterByName("ParseCount");
bool result = false;
List<ParserPrimitives> SubPrimitives = cParserObjectRelation.GetSubParserPrimitives();
for (int x = 0; x < SubPrimitives.Count; x++)
{
ParserPrimitives tPrimitives;
if (SubPrimitives.Count == x + 1)
tPrimitives = null;
else
tPrimitives = SubPrimitives[x + 1];
FunctionIO ret = SubPrimitives[x].CheckRelations(ref lObjectInterface, tPrimitives,CallCalback);
if (((ObjectAttributeInt)ret.GetFunctionParameterByName("ret")).GetAttributeValue() == 1)
{
result = true;
Evaluation.SubEvaluate((ObjectAttributeEvaluation)ret.GetFunctionParameterByName("ParseCount"));
}
}
if (result)
{
ReturnContainer.AddFunctionParameter((ObjectAttribute)Evaluation);
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue1);
return ReturnContainer;
}
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue0);
return ReturnContainer;
}
public override FunctionIO CheckRelations(ref ObjectInterfacePrototipe lObjectInterface, ParserPrimitives NextElement, bool CallCalback)
{
FunctionIO ReturnContainer= new FunctionIO();
ObjectAttributeInt ReturnValue1 = new ObjectAttributeInt("ret",1);
ObjectAttributeInt ReturnValue0 = new ObjectAttributeInt("ret",0);
ObjectAttributeEvaluation Evaluation = new ObjectAttributeEvaluation();
Evaluation.SetAttributeName("ParseCount");
if (!CheckCurrentObject(ref lObjectInterface))
{
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue0);
return ReturnContainer;
}
if (cParserPrimitives.Count == 0)
{
if (cCallback != null && CallCalback)
cCallback(lObjectInterface);
ReturnContainer.AddFunctionParameter((ObjectAttribute)Evaluation);
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue1);
return ReturnContainer;
}
//Here we preset next primitives
for (int x = 0; x < cParserPrimitives.Count-1; x++)
{
cParserPrimitives[x].SetNextPrimitive(cParserPrimitives[x + 1]);
}
ObjectInterfacePrototipe SubObjects= lObjectInterface.GetSubObjects();
SubObjects.GetFirstObject();
ObjectAttributeEvaluation ParseCountEvaluation = new ObjectAttributeEvaluation( EvaluationMode.Max, 1, 100000000);
for( int x=0; x< cParserPrimitives.Count; x++ )
{
ParserPrimitives tPrimitives;
if (cParserPrimitives.Count == x + 1)
tPrimitives = null;
else
tPrimitives = cParserPrimitives[x + 1];
FunctionIO ret = cParserPrimitives[x].CheckRelations(ref SubObjects, tPrimitives, CallCalback);
if (((ObjectAttributeInt)ret.GetFunctionParameterByName("ret")).GetAttributeValue() != 1)
{
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue0);
return ReturnContainer;
}
SubObjects.GetOffsetObject(((ObjectAttributeEvaluation)ret.GetFunctionParameterByName("ParseCount")).GetEvaluationValue());
}
if (cCallback != null && CallCalback)
cCallback(lObjectInterface);
ReturnContainer.AddFunctionParameter((ObjectAttribute)Evaluation);
ReturnContainer.AddFunctionParameter((ObjectAttribute)ReturnValue1);
return ReturnContainer;
}
public bool SubEvaluate(ObjectAttributeEvaluation Evaluation)
{
return EvaluateNext(Evaluation.GetEvaluationValue());
}
/// <summary>
/// Gets evaluation based on if ParseCount function parameter is set or creates new ObjectAttrubteEvaluation.
/// </summary>
/// <returns></returns>
protected ObjectAttributeEvaluation GetEvaluation()
{
ObjectAttributeEvaluation Evaluation = null;
ObjectAttribute lParseCount;
if (cFunctionParameters == null || (lParseCount=cFunctionParameters["ParseCount"]) == null)
{
Evaluation = new ObjectAttributeEvaluation();
Evaluation.AttributeName= "ParseCount";
}
else
{
Evaluation = (ObjectAttributeEvaluation)lParseCount;
}
return Evaluation;
}
public override FunctionIO CheckRelations(ref ObjectInterfacePrototipe lObjectInterface, ParserPrimitives NextElement, bool CallCalback)
{
ObjectAttributeEvaluation Evaluation = new ObjectAttributeEvaluation();
Evaluation.AttributeName="ParseCount";
bool negation = GetAttributeByName("negation")==null? false: true;
bool Result = CheckCurrentObject(ref lObjectInterface);
if (negation)
{
if (Result)
return new FunctionIO(RetVal.False);
}
else
{
if (!Result)
return new FunctionIO(RetVal.False);
}
if (cParserPrimitives.Count == 0)
{
if (cCallback != null && CallCalback)
if (!cCallback(lObjectInterface))
return new FunctionIO(Evaluation, RetVal.False);
return new FunctionIO(Evaluation, RetVal.True);
}
//Here we preset next primitives
for (int x = 0; x < cParserPrimitives.Count-1; x++)
cParserPrimitives[x].SetNextPrimitive(cParserPrimitives[x + 1]);
ObjectInterfacePrototipe SubObjects= lObjectInterface.GetSubObjects();
if (SubObjects.GetFirstObject() == null)
return new FunctionIO(Evaluation, RetVal.True);
ObjectAttributeEvaluation ParseCountEvaluation = new ObjectAttributeEvaluation( EvaluationMode.Max, 1, int.MaxValue);
for( int x=0; x< cParserPrimitives.Count; x++ )
{
ParserPrimitives tPrimitives;
if (cParserPrimitives.Count == x + 1)
tPrimitives = null;
else
tPrimitives = cParserPrimitives[x + 1];
FunctionIO ret = cParserPrimitives[x].CheckRelations(ref SubObjects, tPrimitives, CallCalback);
bool tResult= ret["RetVal"].Value;
if (!negation)
if (!tResult)
return new FunctionIO(RetVal.False);
else
if (tResult)
return new FunctionIO(RetVal.False);
SubObjects.GetOffsetObject(ret["ParseCount"].Value);
}
if (cCallback != null && CallCalback)
cCallback(lObjectInterface);
return new FunctionIO(Evaluation, RetVal.True);
}
/// <summary>
/// Evaluate value from other evaluation.
/// </summary>
/// <param name="Evaluation">Another evaluation.</param>
/// <returns>RetTrue if evaluation was a success, false if not.</returns>
public bool Evaluate(ObjectAttributeEvaluation Evaluation)
{
return Evaluate(Evaluation.Value);
}
