cpd.cpp

这个程序主要是实现检查是否抄袭作业

    FileListFout << Name << endl;
   //............
	cout<<Name<<"\t"<<Suffix<<endl;
	//..................
    //! Copy it into repository
    ReturnKiller(PathAndName, Name);
    //!Ins_Chunk(PathAndName);

    RebuildHashTable();
    FileListFin.close();
    FileListFout.close();
}

//! Examine if a file is probably a plagiarism
//! \param PathAndName a file name string
void HawkEye::Exam(string PathAndName)
{
    ReturnKiller(PathAndName, "t.tmp");
    Evl_Chunk(REPOSITORYPATH + "t.tmp");
}

bool HawkEye::Compare(string StrA, string StrB)
{
#ifdef DEBUG_Compare
    cout << "StrA: " << StrA << endl << "StrB: " << StrB << endl;
#endif
    unsigned long Length= StrA.size();
    if( Length != StrB.size() )
	return false;
    for(i= 0; i< Length; i++)
	if( StrA[i] != StrB[i] )
	    return false;
    return true;
}

long HawkEye::Hash(string str)
{
    long h= 0, a= 33, chnum;
    for(i= 0; i< str.size(); i++){
	// The Chinese Char is smaller than zero(-127 - 128 not 0 - 255)
	chnum= 0xFF + str[i];
	h= (a*h + chnum) % MAX;
    }
    return h;
}

void HawkEye::Insert(FileBlock *ToBeLinked)
{
#ifdef DEBUG_Insert
    cout << "Into Insert()" << endl;
    OutputFileBlock(ToBeLinked);
#endif
    if( NULL == HashTable[ToBeLinked->HashKey] ){
	// If it is empty
	HashTable[ToBeLinked->HashKey]= ToBeLinked;
    }else{
	// Add it to the head of the chain
	ToBeLinked->Next= HashTable[ToBeLinked->HashKey]->Next;
	HashTable[ToBeLinked->HashKey]->Next= ToBeLinked;
    }
#ifdef DEBUG_Insert
    cout << "Out Insert()" << endl;
    OutputFileBlock(ToBeLinked);
#endif
}

//! No Repo
//! Receive abs path!!!
void HawkEye::Ins_Chunk(string PathAndName)
{
    //!PathAndName= GetFileName(PathAndName) + GetSuffix(PathAndName);
    cout<<"hello in Ins_Chunk"<<endl;
    char Ch1, Ch2;
    string Str;
    fstream fin;
    fin.open( PathAndName.c_str(), fstream::in );
    if( !fin.good() ){
	cout << "Error: Can't find input file in Ins_Chunk()!: " << PathAndName << endl;
	cin >> Str;
	exit(1);
    }
    
    long PosCounter= 0;

Ins_Entry:
    while( fin.read(&Ch1, sizeof(char)) ){
	PosCounter++;
#ifdef DEBUG_Ins_Chunk
	//cout << "Ch1: " << Ch1 << endl;
#endif
	// If it can read two chars,check if it is double cut char
	// If Ch1 is the last one,that is,there is no char left,skip this step
	if( fin.read(&Ch2, sizeof(char)) ){
	    PosCounter++;
	    // Is Ch1+Ch2 compose a DoubleCutChar?
	    for(i= 0; i< DoubleCutChar.size(); i+= 2)
		///////////////////////////////Simplify here///////////////////////////////////////
		if( DoubleCutChar[i] == Ch1 && DoubleCutChar[i+1] == Ch2 ){
		    if( Str.size() == 0 ){
			goto Ins_Entry;
		    }else{
			FileBlock *ndPtr= new FileBlock;
			// Go back two chars
			ndPtr->StartPoint= PosCounter - Str.size() - 2;
			ndPtr->Length= Str.size();
			ndPtr->FileName= PathAndName;
			ndPtr->Next= NULL;
			ndPtr->HashKey= Hash(Str);
#ifdef DEBUG_Ins_Chunk
			cout << "Before Double Insertion" << endl;
			cout << Str << endl;
			OutputFileBlock(ndPtr);
#endif
			Insert(ndPtr);
			Str= "";
			goto Ins_Entry;
		    }
		}
#ifdef DEBUG_Ins_Chunk
	    //cout << "Ch2: " << Ch2 << endl;
#endif
	    fin.putback(Ch2);
	    PosCounter--;

#ifdef DEBUG_Ins_Chunk
	    //cout << "Into Single" << endl;
#endif
	    // Single
	    for(i= 0; i< SingleCutChar.size(); i++)
		if( SingleCutChar[i] == Ch1 )
		    if( Str.size() == 0 )
			goto Ins_Entry;
		    else{
			FileBlock *ndPtr= new FileBlock;
			// Go back one char
			ndPtr->StartPoint= PosCounter - Str.size() - 1;
			ndPtr->Length= Str.size();
			ndPtr->FileName= PathAndName;
			ndPtr->Next= NULL;
			ndPtr->HashKey= Hash(Str);
#ifdef DEBUG_Ins_Chunk
			cout << "Before Single Insertion" << endl;
			cout << Str << endl;
			OutputFileBlock(ndPtr);
#endif
			Insert(ndPtr);
			Str= "";
			goto Ins_Entry;
		    }
	    Str+= Ch1;
	    goto Ins_Entry;
	}
    }
    fin.close();

    // The final check for string that ends with no puncture
    if( Str.size() != 0 ){
	FileBlock *ndPtr= new FileBlock;
	Str+=Ch1;
	// Go back one char
	ndPtr->StartPoint= PosCounter - Str.size();
	ndPtr->Length= Str.size();
	ndPtr->FileName= PathAndName;
	ndPtr->Next= NULL;
	ndPtr->HashKey= Hash(Str);
#ifdef DEBUG_Ins_Chunk
	cout << "Before Last Insertion" << endl;
	cout << "String: " << Str << endl;
	cout << "PosCounter: " << PosCounter << endl;
	cout << Str << endl;
	OutputFileBlock(ndPtr);
#endif
	Insert(ndPtr);
    }
}

void HawkEye::Process(string str, ProcessMode pm)
{
    if( pm == CONCATENATE ){
	OutputMessage+= str + "，";
    }
    else if( pm == PRINT ){
	if( OutputMessage.size() >= String_Size_Treshold ){
	    OutputMessage+= str + " is Violated ";
	    cout << OutputMessage << endl;
	}
	OutputMessage= "";
    }
}

void HawkEye::Evaluate(string Str)
{
    static bool LastOneIsViolated= false;
    FileBlock *ndPtr= HashTable[Hash(Str)];
#ifdef DEBUG_Evaluate
    if( ndPtr == NULL ){
	cout << "No hash table yet?" << endl;
    }else{
	cout << "String into Evaluate: " << Str << endl;
	cout << "Hash Key in Evaluate(): " << ndPtr->HashKey << endl;
    }
    // OutputHashTable(1003);
    OutputFileBlock(ndPtr);
#endif
    LastOneIsViolated= false;
    while( ndPtr != NULL ){
#ifdef DEBUG_Evaluate
	cout << "And The Original String is: " << Suck(ndPtr) << endl;
	OutputFileBlock(ndPtr);
#endif
	////////////////////////////////////////////////////
	if( Compare( Str, Suck(ndPtr) ) )
	    LastOneIsViolated= true;

	ndPtr= ndPtr->Next;
    }
    // Process violation
    if( LastOneIsViolated )
	Process(Str, CONCATENATE);
    else
	if( OutputMessage.size() != 0 )
	    Process(Str, PRINT);
}

//! No Repo
void HawkEye::Evl_Chunk(string PathAndName)
{
    char Ch1, Ch2;
    string Str;
    fstream fin;
    fin.open( PathAndName.c_str(), fstream::in );
    if( !fin.good() ){
	cout << "Error: Can't find input file in Evl_Chunk()!" << endl;
	cin >> Str;
	exit(1);
    }
    long PosCounter= 0;

Evl_Entry:
    while( fin.read(&Ch1, sizeof(char)) ){
	PosCounter++;
#ifdef DEBUG_Evl_Chunk
	//cout << "Ch1: " << Ch1 << endl;
#endif
	// If it can read two chars,check if it is double cut char
	// If Ch1 is the last one,that is,there is no char left,skip this step
	if( fin.read(&Ch2, sizeof(char)) ){
	    PosCounter++;
	    // Is Ch1+Ch2 compose a DoubleCutChar?
	    for(i= 0; i< DoubleCutChar.size(); i+= 2)
		///////////////////////////////Simplify here///////////////////////////////////////
		if( DoubleCutChar[i] == Ch1 && DoubleCutChar[i+1] == Ch2 ){
		    if( Str.size() == 0 ){
			goto Evl_Entry;
		    }else{
#ifdef DEBUG_Evl_Chunk
			cout << "Before Double Evaluation" << endl;
			cout << Str << endl;
#endif
			Evaluate(Str);
			Str= "";
			goto Evl_Entry;
		    }
		}
#ifdef DEBUG_Evl_Chunk
	    //	    cout << "Ch2: " << Ch2 << endl;
#endif
	    fin.putback(Ch2);
	    PosCounter--;

#ifdef DEBUG_Evl_Chunk
	    //cout << "Into Single" << endl;
#endif
	    // Single
	    for(i= 0; i< SingleCutChar.size(); i++)
		if( SingleCutChar[i] == Ch1 )
		    if( Str.size() == 0 )
			goto Evl_Entry;
		    else{
#ifdef DEBUG_Evl_Chunk
			cout << "Before Single Evaluation" << endl;
			cout << Str << endl;
#endif
			Evaluate(Str);
			Str= "";
			goto Evl_Entry;
		    }
	    Str+= Ch1;
	    goto Evl_Entry;
	}
    }
    fin.close();

    // The final check for string that ends with no puncture
    if( Str.size() != 0 ){
	Str+=Ch1;
#ifdef DEBUG_Evl_Chunk
	cout << "Before Last Evaluation" << endl;
	cout << "String: " << Str << endl;
	cout << "PosCounter: " << PosCounter << endl;
	cout << Str << endl;
#endif
	Evaluate(Str);
    }
}