IPE :: App Dev :: File IO

What is a File?

Typically we think of files as a container for data. Largely due to the naming format Windows uses to describe the things a user interacts with, a user typically associates a “file” with data written to the hard drive.

While by extension this may be true, a more abstract definition is more accurate. A “file” is an object (a container) that Windows can manipulate. This object (container) is called a File Stream. For simplicity, we can use the fstream library to manipulate text based files.

#include <fstream>

To verify a file exists, we can try to open and check for success:

bool fileExists(char *fileName)
{
	bool retVal;
	std::ifstream inFile(fileName);			//instatiate file stream object, and open it;
	retVal = inFile.good();
	inFile.close();
	return retVal;
}

Similarly we can create a file just by opening it:

void createFile(char *fileName)
{
	std::fstream targetFileCreate;			//instatiate file stream object;
	targetFileCreate.open(fileName, std::fstream::out);
	targetFileCreate << std::flush;
	targetFileCreate.close();
}

With a solid understanding of pointers and character arrays (see Fundamentals), reading a character file is easy:

NOTES: this function takes a NULL pointer as an argument and returns a dynamically allocated array that must be deleted to avoid memory leaks. Alternatively use readCSVDimensions() (presented below) to predetermine csv file dimensions and rewrite function to pre-allocate memory.

int readFile(char** charBuffer, char* fileName, int* bufferLength) {
	int retVal = 0;
	if (*charBuffer == NULL) {
		std::ifstream inFile(fileName, std::fstream::in);	//instatiate file stream object, and open it;
		if (inFile.is_open()) {
			inFile.seekg(0, inFile.end);				//find end of file
			*bufferLength = 1 + (int)inFile.tellg();		//get cursor location
			inFile.seekg(0, inFile.beg);				//place curser at beginning
			*charBuffer = new char[*bufferLength];		//allocate memory
			memset(*charBuffer, '\0', *bufferLength);	//set whole char array to null chars
			inFile.read(*charBuffer, *bufferLength - 1);	//read file less null char
			inFile.close();								//close file
			retVal = 0;	//success
		}
		else {
			*bufferLength = 0;
			retVal = -2; //file could not open
		}
	}
	else {
		std::cout << "Error in alx::fileIO::readFile()\n" <<std::endl;
		*bufferLength = 0;
		retVal = -1; //charBuffer was not NULL
	}
	return retVal;
}

int main()
{
	char* charBuffer = NULL;
	char fileName[70] = "file.txt";
	int bufferLength = 0;

	//readFile allocates memory internally
	int retVal = readFile(&charBuffer, fileName, &bufferLength);
	std::cout << "file contents: " << charBuffer << std::endl;

	//delete memory
	delete[] charBuffer;
}

To write simple character files (for lengthier files see writeCSV2() as an example):

int writeToFile(char* fileName, char* content, int writeMode) {
	//writeModes: 0: open or create, 1. append or create
	std::ofstream outFile;			//instatiate file stream object;
	switch (writeMode)
	{
	case 1: { //create or append
		outFile.open(fileName, std::ofstream::out | std::ofstream::app);
		break;
	}
	default: //open or create
		outFile.open(fileName, std::ofstream::out | std::ofstream::trunc);
		break;
	}
	outFile << content;
	outFile.close();
	return 0;
}

int main()
{
    char charBuffer[30] = "contents to write to file...\n";
    char fileName[70] = "file.txt";
	writeToFile(fileName, charBuffer, 0);
}

Reading CSV Files

To read comma separated value files (.CSV) we need to convert to and from specific data types and format/parse the lines of text relative to rows and columns of the data set.

int readCSVDimensions(char* fileChars, int* rows, int* columns) {
	//Note: assumes a rectangular table of values with specific EOL and seperation characters
	*rows = 0;
	*columns = 1;
	//determine dinensions
	char* ptrLineEnd = strchr(fileChars, '\n');
	char* ptrValSep = strchr(fileChars, ',');
	while (ptrLineEnd > ptrValSep) {
		(*columns)++;
		ptrValSep = strchr(ptrValSep + 1, ',');
	}
	while (ptrLineEnd != NULL) {
		(*rows)++;
		ptrLineEnd = strchr(ptrLineEnd + 1, '\n');
	}
	std::cout << "   dimensions: " << *rows << " x " << *columns << std::endl;
	return 0;
}

int readCSV2D(char* fileName, double*** data, int* rows, int* columns){
	//returns a dynamically allocated array** you must delete it externally
	//WARNING: uses strtod which can return a zero value on failure!!!
	int retVal = 0;
	if (*data == NULL) {
		*rows = 0;
		*columns = 0;
		char* charBuffer = NULL;
		int charBufferLength = 0;
		int charRetVal = readFile(&charBuffer, fileName, &charBufferLength);
		if (charRetVal == 0) {
			//std::cout <<"file contents" << charBuffer << std::endl;
			if (charBufferLength > 0) {
				readCSVDimensions(charBuffer, rows, columns);
				//allocate memory
				(*data) = new double* [*rows];
				for (int r = 0; r < *rows; r++)
					(*data)[r] = new double[*columns];
				//step through seperations characters to parse data
				char* ptrValStart = charBuffer;					//pointer to a character in Buffer, init to first character
				char* ptrValEnd;
				//char* ptrLineEnd;
				bool endOfFile = false;
				for (int r = 0; r < *rows; r++) {			//for each row
					for (int c = 0; c < *columns; c++) {		//for each column
						//read next value and get new end ptr; if error val is zero padded!!!!
						(*data)[r][c] = strtod(ptrValStart, &ptrValEnd);
						ptrValStart = ptrValEnd + 1;		//reset start ptr
					}
				}
				delete[] charBuffer;
			}
			else {
				retVal = -3; //file is empty
				std::cout << "ERROR file is empty" << std::endl;
			}
		}
		else{
			retVal = -2; //could not read file
			std::cout << "ERROR reading file" << std::endl;
		}
	}
	else {
		std::cout << "ERROR passing NULL value" << std::endl;
		retVal = -1;
		*rows = 0;
		*columns = 0;
	}
	return retVal;
}

int main()
{
    int rows = 0;
    int columns = 0;
    char fileName[12] = "C:/data.csv";
    double** data = NULL;
    readCSV2D(fileName, &data, &rows, &columns);

    for (int i = 0; i < rows; i++)
        for (int j = 0; j < columns;j++)
            std::cout << data[i][j] << std::endl;

    for (int i = 0; i < rows; i++)
        delete[] data[i];
    delete[] data;
    data = NULL;
	
}

Writing CSV Files

Writing simple CSV files is straightforward when allowing fstream to automatically determine the periodicity of systems calls (when to actually write the data to disk):

int writeCSV(char* fileName, double** data2D, int rows, int columns, int writeMode) {
	//writeModes: 0: overwrite or create, 1. append or create
	std::ofstream outFile;		//instatiate file stream object;
	outFile.precision(9);		//sets precision for floating point values in stream
	switch (writeMode)
	{
	case 1: { //create or append
		outFile.open(fileName, std::ofstream::out | std::ofstream::app);
		outFile.flush();		//push content to file (if any)
		break;
	}
	default: //overwrite or create
		outFile.open(fileName, std::ofstream::out | std::ofstream::trunc);
		break;
	}
	//count characters needed to conver double to char

	//then write n characters to file
	for (int h = 0; h < rows; h++) { //for each row
		for (int w = 0; w < columns - 1; w++) {
			outFile << data2D[h][w] << ", ";
		}
		outFile << data2D[h][columns - 1] << "\n";
	}
	outFile.close();	//flush and close
	return 0;
}

When writing large files (millions of data elements), it is advantageous to pre-parse the data and force control of the fstream object to minimize system calls. This significantly reduces write time (For example: when using VS2019 targeting Windows 10, write times of 15 minutes can be reduced to less than 1 minute).

int writeCSV2(char* fileName, double** data2D, int rows, int columns, int writeMode) {
	//writeModes: 0: overwrite or create, 1. append or create
	std::ofstream outFile;		//instatiate file stream object;
	outFile.precision(9);		//sets precision for floating point values in stream
	switch (writeMode)
	{
	case 1: { //create or append
		outFile.open(fileName, std::ofstream::out | std::ofstream::app);
		outFile.flush();		//push content to file (if any)
		break;
	}
	default: //overwrite or create
		outFile.open(fileName, std::ofstream::out | std::ofstream::trunc);
		break;
	}
	//count characters needed to conver double to char

	//then write n characters to file
	const int bufferSize = 1650000;
	char* charBuffer = new char[bufferSize];	//prealocate character buffer
	char charValue[255];
	int charValueLength = 0;
	memset(charBuffer, '\0', bufferSize);
	int bytesInBuffer = 0;
	int h = 0, w = 0;

	while (true) {
		if (w < (columns - 1)) {
			double val = data2D[h][w];
			sprintf_s(charValue, 255, "%.4f, ", data2D[h][w]);
			charValueLength =  strlen(charValue);
			memcpy(charBuffer + bytesInBuffer, charValue, charValueLength);  //bufferSize - bytesInBuffer
			bytesInBuffer += charValueLength;
			w++;
		}
		else {
			sprintf_s(charBuffer + bytesInBuffer, bufferSize - bytesInBuffer, "%.4f\n", data2D[h][w]);
			bytesInBuffer = strlen(charBuffer);
			w = 0;
			h++;
		}
		if (bytesInBuffer > 1649900) {
			std::cout << "pushing to file..." << std::endl;
			outFile << charBuffer;		//write to file
			memset(charBuffer, '/0', bufferSize);
			bytesInBuffer = 0;
		}
		if (h > (rows-1)) {
			std::cout << "pushing to file..." << std::endl;
			outFile << charBuffer;		//write to file
			outFile.close();			//flush and close
			break;
		}
	}
	delete[] charBuffer;
	charBuffer = NULL;
	outFile.close();	//flush and close
	return 0;
}