cpd.cpp

这个程序主要是实现检查是否抄袭作业

#include "cpd.h"

using namespace std;

// Debug functions
/*
void OutputHashTable(HawkEye he, long Key)
{
    cout << "==Outputting Hash Table==" << endl;
    FileBlock *ndPtr= he.HashTable[Key];
    while( ndPtr != NULL ){
	cout << ndPtr->HashKey << endl;
	ndPtr= ndPtr->Next;
    }
    cout << "==Output Completed==" << endl;
}
*/

void HawkEye::OutputFileBlock(FileBlock *ndPtr)
{
#ifdef DEBUGFUNCTION
    if(ndPtr == NULL)
	cout << "Current FileBlock is NULL!!!" << endl;
    else{
    cout << "==Outputting File Block==" << endl;
    cout << "StartPoint: " << ndPtr->StartPoint << endl;
    cout << "Length: " << ndPtr->Length << endl;
    cout << "FileName: " << ndPtr->FileName << endl;
    cout << "HashKey: " << ndPtr->HashKey << endl;
    cout << "==Output Completed==" << endl;
    }
#endif
/*
    cout << "==========Hash table statistic===========" << endl;
    for(i= 0; i< MAX; i++){
	FileBlock *fbp= HashTable[i];
	int counter= 0;
	while( fbp != NULL ){
	    counter++;
	    fbp=fbp->Next;
	}
	cout << i << ':' << counter << endl;
    }
    */
}

/*
void Scan(long key)
{
    FileBlock *ndPtr= HashTable[key];
    int Counter= 0;
    while( ndPtr != NULL ){
	Counter++;
	ndPtr= ndPtr->Next;
    }
    cout << "Scanning completed, 1003 has " << Counter << " elements." << endl;
}
*/

// Independent function
// Repo
string HawkEye::Suck(FileBlock *fbp)
{
#ifdef DEBUG_Suck
    cout << "To Be Sucked: " << endl;
    OutputFileBlock(fbp);
#endif
    string ToBeReturned= "";
    fstream fin;
    fin.open( (fbp->FileName).c_str(), fstream::in );
    if( !fin.good() )
	cout << "Open file error in Suck()" << endl;
    fin.seekg(fbp->StartPoint);
    char ch;
    for(i= 0; i< fbp->Length; i++){
	fin.read( &ch, sizeof(char) );
	ToBeReturned+= ch;
    }
    fin.close();
#ifdef DEBUG_Suck
    cout << "Sucked Out: " << ToBeReturned << endl;
#endif
    return ToBeReturned;
}

string HawkEye::GetFileName(string AbsPath)
{
    int dashpos, dotpos;
    dashpos= AbsPath.rfind('/');
    if(dashpos == (int)string::npos)
	//! If it is already the relative path
	dashpos= -1;
    dotpos= AbsPath.rfind('.');
    return AbsPath.substr(dashpos+1, dotpos-dashpos-1);
}

string HawkEye::GetSuffix(string AbsPath)
{
    unsigned int dashpos, dotpos;
    dotpos= AbsPath.rfind('.');
    if(dotpos == string::npos)
	return "";
    dashpos= AbsPath.rfind('/');
    if(dashpos == string::npos)
	//! If it is already the relative path
	dashpos= 0;
    return '.' + AbsPath.substr(dotpos+1, AbsPath.size()-dotpos-1);
}

// Independent function
int HawkEye::LCS(string astr, string bstr)
{
    /*
    unsigned int asize= astr.size(),
		 bsize= bstr.size();
    int counter[asize+1][bsize+1];
    for(i= 1; i<= asize; i++)
	counter[i][0]= 0;
    for(i= 0; i<= bsize; i++)
	counter[0][i]= 0;
    for(i= 1; i<= asize; i++)
	for(j= 1; j<= bsize; j++)
	    if( astr[i-1] == bstr[j-1] )
		counter[i][j]= counter[i-1][j-1] + 1;
	    else{
		if( counter[i-1][j] > counter[i][j-1] )
		    counter[i][j]= counter[i-1][j];
		else
		    counter[i][j]= counter[i][j-1];
	    }
    return counter[asize][bsize];
    */
    return 0;
}

// Independent function
// inFileName may use abs path or rel path
// outFileName may use abs path or rel path too, 
// auto saved into repository
bool HawkEye::ReturnKiller(string inFileName, string outFileName)
{
    fstream fin, fout;

    cout << "$$fin=" << inFileName << endl;
    fin.open( inFileName.c_str(), fstream::in);
    assert( fin.good() );

    outFileName= GetFileName(outFileName) + GetSuffix(outFileName);
    cout << "$$fout=" << REPOSITORYPATH + outFileName << endl;
    fout.open( (REPOSITORYPATH + outFileName).c_str(), fstream::out);
    assert( fout.good() );

    char ch;
    while( fin.read( &ch, sizeof(char) ) )
	if( ch != '\n' )
	    fout << ch;
    fin.close();
    fout.close();
    return true;
}

// repo 
void HawkEye::SaveHashTable()
{
    fstream fout;
    fout.open( (REPOSITORYPATH + m_HashTableName).c_str(), fstream::out );
    fout << " " << endl;
    if(fout.good())
	for(i= 0; i< MAX; i++)
	    if( NULL != HashTable[i] ){
		FileBlock *ndPtr= HashTable[i];
		while( NULL != ndPtr ){
		    FileBlock *successor= ndPtr->Next;
		    fout << ndPtr->HashKey << ' ' << ndPtr->StartPoint << ' ' << ndPtr->Length << ' ' << ndPtr->FileName << ' ';
		    ndPtr= successor;
		}
	    }
    fout.close();
}

// repo 
void HawkEye::LoadHashTable()
{
    long HashKey;
    fstream fin;
    fin.open( (REPOSITORYPATH + m_HashTableName).c_str(), fstream::in );
    if( !fin.good() ){
	cout << "No document in repository yet!" << endl;
	cout << "Please add document into repository first" << endl;
    }else{
	while( fin >> HashKey ){
	    FileBlock *ndPtr= new FileBlock;
	    ndPtr->HashKey= HashKey;
	    fin >> ndPtr->StartPoint >> ndPtr->Length >> ndPtr->FileName;
	    ndPtr->Next= NULL;
	    Insert(ndPtr);
	}
    }
    fin.close();
}

void HawkEye::CleanUpHashTalbe()
{
    for(i= 0; i< MAX; i++)
	if( NULL != HashTable[i] ){
	    FileBlock *ndPtr= HashTable[i];
	    FileBlock *ndPtrToNext;
	    while( NULL != ndPtr->Next)
	    {
		ndPtrToNext= ndPtr->Next;
		delete ndPtr;
		ndPtr= ndPtrToNext;
	    }
	    delete ndPtr;
	}
}

//! Rebuild hash table
//!
//! Generate new hash table from current repository
//! repo
void HawkEye::RebuildHashTable()
{
    fstream fin;
    fin.open( (REPOSITORYPATH + "filelist.lst").c_str(), fstream::in);
    string filename;
    while( fin >> filename )
    {
    cout<<filename<<"  before ins_Chunk"<<endl;
	Ins_Chunk(REPOSITORYPATH + filename);
   }
    fin.close();
    SaveHashTable();
}

//! Choose a table
//!
//! Set another hash table for current use
//! \param filename Then file name of new hash table
void HawkEye::SetHashTable(string filename)
{
    m_HashTableName= filename;
}

//! Change the threshold of least string length
//!
//! Set the least string length in bytes, one can 
//! adjust it dynamically by invoking this function
//! \param threshold The least string length
void HawkEye::SetStringSizeTreshold(int threshold)
{
    String_Size_Treshold= threshold;
}

//! Constructor
//!
//! initialize HashTable
HawkEye::HawkEye()
{
    SetHashTable("hstb.txt");
    for(i= 0; i< MAX; i++)
	HashTable[i]= NULL;
    SetStringSizeTreshold(2);
}

//! Destructor
//!
//! Maybe Clean up HashTable
//! But nothing to do now
HawkEye::~HawkEye()
{
    unsigned int size= OutputMessage.size();
    if( (size != 0) && (size > String_Size_Treshold) ){
	OutputMessage[size-2]='i';
	OutputMessage[size-1]='s';
	OutputMessage+= " Violated.";
	cout << OutputMessage << endl;
    }
    CleanUpHashTalbe();
}

//! Check Document into Repository
//! \param PathAndName a file name string
void HawkEye::CheckIn(string PathAndName)
{
    fstream FileListFin, FileListFout;
    FileListFin.open( (REPOSITORYPATH + "filelist.lst").c_str(), fstream::in);

    string Name= GetFileName(PathAndName);
    string Suffix= GetSuffix(PathAndName);
    if( FileListFin.good() ){
	//! Check for if it already exsits
	char EntryInList[300];
	while( !FileListFin.eof() ){
	    FileListFin.getline(EntryInList, 300);
	    if( Name == EntryInList ){
		cout << "File already exsit! Adding failed" << endl;
		exit(1);
	    }
	}                  
    }

    //! Open or create(if not exist) file list
    FileListFout.open( (REPOSITORYPATH + "filelist.lst").c_str(), fstream::out|fstream::app);

    //! Add it into file list