5 replies to this topic

[nuclear123]

any suggestions to improve this is greatly appreciated!(yes it compiles and works properly)

// Sequential_Sort.cpp : Defines the entry point for the console application.
//

#include "stdafx.h"
#include <iostream>
#include <fstream>
#include <Windows.h>
#include <WinBase.h>

/**
* Class: Array
*
* Custom Array class used to read input from a file
* and store the contents to specific members of Node.
* 1 Node is equivalent to 1 person.
*
*/

class Array {
private:
class Node {
private:
	friend class Array;
	int m_SSid;
	int m_age;
	int m_commulativeContribution;
	int m_yearStarted;
public:
};
	Node        *m_headPtr;
	int          m_ArraySize;
	std::fstream m_dataFile;
public:
	Array( int size, std::string fileName ); 
	~Array();
	void populateArray();
	void displayArray();
	void displayElement( int index );
	int  sequentialSearch( int SSid );
};

/**
* Constructor: Array()
*
* The constructor must accept the size of the list to be used
* as well as the name of the file which contains the data of
* each Node/Person.  If the file fails to Open for input then
* throw an exception which is caught and the error is stored in
* a log file.  If successfull then allocate space for our custom
* array depending on the size specified by the user/programmer.
* If unable to allocate memory on the heap for the custom array close
* the file and then throw an exception which is caught by the caller 
* and stored into a log file.
*
* @access public
*/

Array::Array( int size, std::string fileName )
	: m_ArraySize( size ) 
{
	m_dataFile.open( fileName.c_str(), std::ios::in );
	if( m_dataFile.fail() )
		throw "Unable To Open The Specified File";
	m_headPtr = new Node[size]; // allocate memory for Array
	if( m_headPtr == NULL ) {
		m_dataFile.close();
		throw "Unable To Allocate Sufficient Memory On The Heap";
	}
}

Array::~Array()
{
	m_dataFile.close();
	delete [] m_headPtr;
}

/**
* Function: populateArray()
*
* Simply progresses through the custom array and populates it
* with data from the specified input file.  Must clear eof/status
* flags for the ofstream and then reset file cursor to the beginning
* of the file in order to make this function call to be successfull
* multiple times
*
* @return void
* @access private
*/
void Array::populateArray()
{
	m_dataFile.clear(); 
	m_dataFile.seekp( 0L, std::ios::beg ); // (offset, base(where to start))
	for( int i = 0; i < m_ArraySize; i++ ) {
   		m_dataFile >> m_headPtr[i].m_SSid;
		m_dataFile >> m_headPtr[i].m_age;
		m_dataFile >> m_headPtr[i].m_commulativeContribution;
		m_dataFile >> m_headPtr[i].m_yearStarted;
	}
}


/**
* Function: displayArray()
*
* Simply Progresses through the whole list
* and displays the datamembers of each Node/Person
* element in the list.
*
* @return void
* @access private
*/
void Array::displayArray()
{
	for( int i = 0; i < m_ArraySize; i++ ) {
		std::cout << m_headPtr[i].m_SSid << std::endl;
		std::cout << m_headPtr[i].m_age << std::endl;
		std::cout << m_headPtr[i].m_commulativeContribution << std::endl;
		std::cout << m_headPtr[i].m_yearStarted << std::endl << std::endl;
	}
}


/**
* Function: displayElement()
*
* Recieves an Index to where the located Person/Node
* object is located in the list, and then dereferences
* the object's data members there and displays them.
*
* @return void
* @access private
*/
void Array::displayElement( int i )
{
	std::cout << m_headPtr[i].m_SSid << std::endl;
	std::cout << m_headPtr[i].m_age << std::endl;
	std::cout << m_headPtr[i].m_commulativeContribution << std::endl;
	std::cout << m_headPtr[i].m_yearStarted << std::endl;
}


/**
* Function: sequentialSearch()
*
* Simply Progresses through the whole list
* and when a match for the specified passed in SSid
* is found, then return the index to which that
* SSid number was found. Returns -1 if no match was found
*
* @return bool
* @access private
*/
int Array::sequentialSearch( int SSid )
{
	for( int i = 0; i < m_ArraySize; i++ )
		if( m_headPtr[i].m_SSid == SSid )
			return i;
	return -1;
}

/**
* Function: add_log()
*
* Whenever an exception is thrown and caught
* this function is called to write to a text
* file what exception was thrown, and when it 
* happened.
*
* @return void
*  Free Function
*/
void add_log(char buffer[100])
{
	std::ofstream outFile;
	outFile.open("Log.txt", std::ios::app);
	SYSTEMTIME st;

	GetLocalTime(&st);
	outFile << "===================================" << std::endl;
	outFile << "Date: " << st.wMonth << "/" << st.wDay << "/" << st.wYear << std::endl;
	outFile << "Time: " << st.wHour << ":" << st.wMinute << std::endl;
	outFile << "===================================" << std::endl;
	outFile << "[ERROR] - " << buffer << std::endl;
	outFile.close();
}

/**
* Function: holdScreen()
*
* This function is pretty self explanatory.
* All it does is hold the console screen open.
* It does this by ignoring any input left in the
* input stream, and then asking the user to type
* a character which is pretty much used to exit
* the program in this case
*
* @return void
*/
inline void holdScreen()
{
	std::cin.ignore();
	std::getchar();
}


int _tmain(int argc, _TCHAR* argv[])
{
	try {
	Array Array1( 2, "datafile.txt" );
	Array1.populateArray();
	Array1.displayArray();
	int index = Array1.sequentialSearch( 645 );
	if( index == -1 )
		std::cout << "The SSid that was searched for was not found in the Array" << std::endl;
	else
		std::cout << "The SSid was found at index " << index << std::endl;
	}
	catch( char *string ) {
		add_log( string );
	}



	holdScreen();
	return 0;
}

[Hodgman]

Implementation problems:
1) Functions like displayElement, which do not modify members, should be const
2) Complex types, such as std::string should be passed by const-reference, instead of passed by value.
3) Array should definitely be a template class, with Node as the template parameter.
4) The check of m_headPtr against NULL is useless -- the basic new operator never returns NULL (it throws a std::bad_alloc on failure).
5) You're not following the rule of three, which means your class is dangerous and will cause crashes if an instance is ever copied/assigned.

Design problems:
This design constantly violates the law of demeter. You've tied together:
A ) a particular application-specific data structure (Node)
B ) functionality for deserialising 'Nodes' from a file stream (which also requires the user to guess how many items are in the file...)
C ) functionality for displaying 'Nodes' to a hard-coded output stream (this logic is even duplicated in displayArray/displayElement)
D ) a linear search function (std::find)
E ) array resource management (std::vector)

That's 5 different areas of functionality, all tightly coupled together. There is no need for any of those areas to be directly connected to each other like that.

[Hodgman]

Could you plz explain how/what i can do to make my member functions loosely coupled? Also how can i stop from duplicating logic like you say?

I'll reply to your PM here.


First, to decouple the array class, we can strip the it back so it only deals with array memory management - nothing else.

template<class T> class Array

{
private:
	typedef T Node;
	Node*  m_headPtr;
	size_t m_ArraySize;
public:
	Array()  : m_ArraySize(), m_headPtr() {}
	Array( size_t size ) : m_ArraySize( size ), m_headPtr( new Node[size] ) {}
	Array( const Array& obj ) : m_ArraySize(), m_headPtr() { *this = obj; }
	~Array() { delete [] m_headPtr; }
	Array& operator=( const Array& obj );
	void resize( size_t );
	size_t size() const { return m_ArraySize; }
	Node& operator[]( size_t i ) { return m_headPtr[i]; }
	const Node& operator[]( size_t i ) const { return m_headPtr[i]; }
};

template<class T> Array<T>& Array<T>::operator=( const Array<T>& obj )
{
	if( this != &obj )//skip over this code if performing self assignment (e.g. "a = a;")
	{
		delete [] m_headPtr;
		m_ArraySize = obj.m_ArraySize;
		m_headPtr   = new Node[obj.m_ArraySize];
		for( int i = 0; i < m_ArraySize; i++ )
			m_headPtr[i] = obj.m_headPtr[i];
	}
	return *this;
}

template<class T> void Array<T>::resize( size_t size )
{
	m_ArraySize = size;
	Node* pNew  = new Node[size];
	for( int i = 0; i < m_ArraySize; i++ )
		pNew[i] = m_headPtr[i];
	delete [] m_headPtr;
	m_headPtr = pNew;
}

Some of the logic that was in populate/display can be moved into overloaded operators for your node structure:

struct Node {
		int m_SSid;
		int m_age;
		int m_commulativeContribution;
		int m_yearStarted;
};
std::istream& operator>>( std::fstream& in, Node& out )
{
	in >> out.m_SSid;
	in >> out.m_age;
	in >> out.m_commulativeContribution;
	in >> out.m_yearStarted;
	return in;
}
std::ostream& operator<<( std::ostream& out, const Node& in )
{
	out << in.m_SSid << std::endl;
	out << in.m_age << std::endl;
	out << in.m_commulativeContribution << std::endl;
	out << in.m_yearStarted << std::endl;
	return out;
}

...Which makes the populate/display functions a bit more reusable (they work for any kind of array/node now). They also don't have to be member functions, because they don't need to access any private variables.

//print out an array's values
template<class Array>void displayArray( const Array& in )
{
	for( size_t i = 0; i != in.size(); ++i )
		std::cout << in[i];
}

//read array values from a file
template<class Array> void populateArray( Array& out, std::fstream& in, size_t size )
{
	in.clear(); 
	in.seekp( 0L, std::ios::beg ); // (offset, base(where to start))
	out.resize( size );
	for( int i = 0; i < size; i++ )
		in >> out[i];
}

//helper overload takes a file-name instead of a file
template<class Array> void populateArray( Array& out, const std::string& fileName, size_t size )
{
	std::fstream dataFile;
	dataFile.open( fileName.c_str(), std::ios::in );
	if( dataFile.fail() )
		throw "Unable To Open The Specified File";
	populateArray( out, dataFile, size );
	dataFile.close();
}

The search function used to be hard-coded to only search for the "m_SSid" field. If you turn it into a template then you can reuse this logic for many different kinds of searches

template<class Array, class T> int sequentialSearch( const Array& data, const T& find )
{
	for( int i = 0; i < data.size(); i++ )
		if( data[i] == find )
			return i;
	return -1;
}

//pass a 'FindSsid' object to sequentialSearch in order to search nodes based on m_SSid
struct FindSsid {
	int ssid;
	FindSsid(int i) : ssid(i) {}
};
bool operator==( const Node& lhs, const FindSsid& rhs ) {
	return lhs.m_SSid == rhs.ssid;
}

//putting it all together
int main( int argc, char** argv )
{
	try {
		Array<Record> array1;
		populateArray( array1, "datafile.txt", 2 );
		displayArray( array1 );
		int index = sequentialSearch( array1, FindSsid(645) );
		if( index == -1 )
			std::cout << "The SSid that was searched for was not found in the Array" << std::endl;
		else
			std::cout << "The SSid was found at index " << index << std::endl;
	}
	catch( const char *string ) {
		add_log( string );
	}
	catch( const std::bad_alloc& exception ) {
		add_log( "Unable to Allocate the Specified memory" );
	}
	catch( ... ) {
		add_log( "Unknown error" );
	}
	holdScreen();
	return 0;}

[nuclear123]

thanks a bunch mate! am currently at work, but when i get home i shall study this and understand how it's differs from my code! and make according changes!

typedef T Node; <--- why instantiate a Node object on creation of an array? is this a mistake?

[nobodynews]

thanks a bunch mate! am currently at work, but when i get home i shall study this and understand how it's differs from my code! and make according changes!

typedef T Node; <--- why instantiate a Node object on creation of an array? is this a mistake?

That doesn't instantiate anything. A typedef creates a type alias so that Node can be used in place of T.

[Hodgman]

<--- why instantiate a Node object on creation of an array? is this a mistake?

Sorry, that was unnecessarily confusing.

I did that because the Array class was already written to use 'Node' everywhere, but I wanted it to use 'T' instead. That typedef says that wherever you see the word Node used (inside the array class and it's functions ONLY), then it's actually a 'T' (NOT the Node from your first post).

e.g. given a Array<int>, then Array<int>::Node would be a private typedef for 'int'.

template<class T> class Array
{
private:
        typedef T Node; //Inside the scope of "Array", pretend that the word "Node" is actually "T".
....
....
template<class T> void Array<T>::resize( size_t size )
{//We're inside the scope of "Array::" here
...
    //this doesn't actually make one of your 'Node' structs, this makes a 'T' (whatever that is)
    Node* pNew = new Node[size];   //i.e.   T* pNew = new T[size];
...

I could've done a find&replace with 'Node' and 'T' instead :/