Reading from a binary file.

[Technyque]

Hello all

I'm trying to read information from a binary file (employees.bin), into a struct. However, I'm unsure on why my data is not being read correctly.

#include <fstream>
#include <iostream>
#include <iomanip>
#include <string>
#include <ctime>
#include <cstring>
#include <cstdlib>
#include <conio.h> // for "getch()"
using namespace std;
const int MAX = 100;



struct empdata
{
    //data members
        char firstname[13];
	char lastname[13];
	char idnumber[12];
	char datehired[9];
        double payrate ;
	double hoursworked ;
	double grosspay;
};




int main ()
{
        empdata EmployeeRecord;
	int index = 0 , limit = 0;



	fstream labfile;
	labfile.open("testEmployees.bin", ios::in | ios::binary);



	if(labfile.fail())
	{
		cout << "Unable to open the data file!!!" <<endl;
		cout << "Please make sure your bin file is in the same location as the program!"<<endl<<endl<<endl;
		system("PAUSE");
		return 999;

	}

        labfile.seekg(0,std::ios::beg);
	labfile.read(reinterpret_cast<char*>(&EmployeeRecord), sizeof(EmployeeRecord));


	while(!labfile.eof())
	{
**Here im just testing to see if the correct data went in.
        cout<<EmployeeRecord.firstname<< endl;
	    cout<<EmployeeRecord.lastname<< endl;
	    cout<<EmployeeRecord.idnumber<< endl;
	    cout<<EmployeeRecord.datehired<< endl;
	    cout<<EmployeeRecord.payrate<< endl;
	    cout<<EmployeeRecord.hoursworked<< endl;

	    labfile.read(reinterpret_cast<char*>(&EmployeeRecord), sizeof(EmployeeRecord));
	    getch();
	}

    labfile.close();
	getch();
	return(0);
}

And i get:

Andrew
Conyers
219-52-2044
02231988
3.43597e+010<<<<wrong, should be payrate
1.72632e-306<<<<wrong, should be hours worked

The payrate and hoursworked are wrong when reading the first record.
Once I hit enter so that it reads the file for the next record, all of the values are off.

Been looking for an answer for a while and haven't been able to find it.

Any help is appreciated. Thanks in advance.

[ahura24]

this is the format of scientific !!! i dont know what you write in your file and then what you read from of but test it in begining of your code :

cout.setf(ios::fixed, ios::floatfield);

[Technyque]

tested it, now I get

34359746702.000000<<payrate
0.000000<<hours worked

BTW, I'm not writing anything to a file, just reading. He asks me to make a struct with the data members: first, last, ID, date hired ,pay and hours.

Then declare a single object variable named EmployeeRecord and write a loop control structure to read data from a .bin file my professor made, one record at a time, into the employeerecord variable.

Does it have anything to do with atof or null terminator?

Also is there anything I can read to help, went over the chapter twice but it only explains how to do it the basic way.

[ceruleus]

I'm not sure if I fully understand what you're trying to do here but I would recommend that rather than trying to read the whole struct in one go you should try to take each piece. So if your struct was something like:

struct MyStruct
{
  int i;
  char c;
}

//read each piece individually like this
int readNumber = 0;
char readChar = '';
labfile.read( &readNumber, sizeof(int) );
labfile.read( &readChar, sizeof(char) );

maybe I'm missing the point here... hope it helps!

[andywestken]

Were you given the struct definition? Sizes and all?

[andywestken]

For your struct def

int main()
{
	size_t a = sizeof(empdata);
	size_t b = sizeof(char) * (13 + 13 + 12 + 9) + sizeof(double) * 3;

	std::cout << a << std::endl;
	std::cout << b << std::endl;

	return 0;
}

produces the following output

72
71

Assuming you haven't overridden the compilers default packing?

So you are reading one additional byte cf. the members declared in struct.

[Technyque]

@ceruleus
These worked one by one for the chars in the struct, but I'm still not getting the right info for the doubles...
'

	
    labfile.read(reinterpret_cast<char*>(&EmployeeRecord.firstname), sizeof(EmployeeRecord.firstname));
    labfile.read(reinterpret_cast<char*>(&EmployeeRecord.lastname), sizeof(EmployeeRecord.lastname));
    labfile.read(reinterpret_cast<char*>(&EmployeeRecord.idnumber), sizeof(EmployeeRecord.idnumber));
    labfile.read(reinterpret_cast<char*>(&EmployeeRecord.datehired), sizeof(EmployeeRecord.datehired));


This didnt work
   labfile.read(reinterpret_cast<char*>(&EmployeeRecord.payrate), sizeof(EmployeeRecord.payrate));
    labfile.read(reinterpret_cast<char*>(&EmployeeRecord.hoursworked), sizeof(EmployeeRecord.hoursworked));

Still getting weird info for payrate and hoursworked(looks like 3.43597e+010), when it should return in values of payrate(999.99) and hoursworked(99.9)

..............

@andy this is the info he gave us, along with a pdf file that says what he wants us to do in the assignment and a bin file(testemployees.bin).

He said-
"For you information, in case you need it, I used the following character counts for the char array string values in the structure used to create the binary data file. It should not matter, but I want you to have the details just in case."
first name and last name: 12+1 (13) <-- that means a maximum of 12 characters + 1 for the NULL
id: 11+1 (12)
date hired: 8+1 (9)
pay rate: float
hours worked: float


And the pdf. file with the assignment says:

"Use the structure definition from lab 4(that's the struct I'm using) that represents the employee data type. Declare a Single object variable and name it employeerecord. Write a looping control structure to read data, one record at a time, into this employeerecord variable. NOTE: The data file I created is a binary data file(testemployees.bin)."

The assignment asks me to do other stuff after that but I can't even get the data to read properly into the struct. Been trying to find a solution for days now but I can't figure it out. If not then i'll just have to wait until after a test tomorrow to ask him about it, but i'm a bit worried about something like this showing up on the test.

[ahura24]

@Technyque (8):
labfile.read(reinterpret_cast<char*>(&EmployeeRecord.firstname), sizeof(EmployeeRecord.firstname));

this code not right because firstname is the address of first block which array be started !! then you can wrote
reinterpret_cast<char*>(EmployeeRecord.firstname)
or
reinterpret_cast<char*>(&EmployeeRecord.firstname[0])
i dont know but i read in complete reference STL , reinterpret_cast is not good idea then you just write
(char *)&obj is better ! but i dont know why !!
your first code dont have any problem just are you sure about your file ?!!! if you dont know about inside of them , then how do you know its true ?!!!

you can test yourself just debug this code and see yourself :

struct empdata
{
	//data members
	char firstname[13];
	char lastname[13];
	char idnumber[12];
	char datehired[9];
	double payrate ;
	double hoursworked ;
	double grosspay;

	empdata()
	{
		strcpy(firstname, "asdf");
		strcpy(lastname, "fdfdf");
		strcpy(idnumber, "");
		strcpy(datehired, "aaaaa");

		payrate = 123;
		hoursworked = 111;
		grosspay = 0;
	}
};

void main()
{
	empdata e;

	empdata rr; // just for check!!

	// just for check !!!! 
	rr.grosspay = 111111;
	rr.hoursworked = 0;
	strcpy(rr.firstname, "");

	ofstream fout("a.dat", ios::out | ios::binary);
	fout.write((const char *)&e, sizeof( empdata ));
	fout.close();

	ifstream fin("a.dat", ios::in | ios::binary);
	fin.read((char *)&rr, sizeof( empdata ));
	fin.close();
}

[ceruleus]

@OP you said that this doesn't work:

labfile.read(reinterpret_cast<char*>(&EmployeeRecord.payrate), sizeof(EmployeeRecord.payrate));
    labfile.read(reinterpret_cast<char*>(&EmployeeRecord.hoursworked), sizeof(EmployeeRecord.hoursworked));

That makes sense since you are casting a double to a char, so of course it won't work. When using basic types (like double, int, char, etc...) you should just use a c-style cast:

labfile.read( (double)&EmployeeRecord.payrate, sizeof( double ) );
    labfile.read( (double)&EmployeeRecord.hoursworked, sizeof( double ) );

edit: also, I edited the above... I'm not sure if your second parameter to read is valid, use sizeof(type) instead

[Technyque]

@ahura24


Bingo! I ran your code to create "a.dat", and using my original code it's reading all the data from "a.dat" like it's supposed to. I guess this means the .bin file that my prof gave us may need to be fixed?

Thanks for all the help, will mark as solved later today after I talk to him.

[andywestken]

I would be interested to here what your prof's solution is.

For binary data it is standard practice to write or read a single record (= structure) at a time, as you were attempting to do earlier on.

[Technyque]

would this be by single record? I'm pretty sure he wanted it read like this but i guess ill see.

labfile.read(reinterpret_cast<char*>(&EmployeeRecord), sizeof(EmployeeRecord));

@ceruleus,
I get: invalid cast from type 'double*' to type 'double', when it try it that way.

I'm thinking it may be something wrong with the .bin file he provided us with, I'll let ya'll know what he says.

[ceruleus]

@Technyque then change the (double)&EmployeeRecord.payrate to (double*)&EmployeeRecord.payrate

The point I was trying to get across is that I think you don't need to use reinterpret_cast or any of those fancy object casts. You should probably just be using a c-style cast like the above code. I'm really not sure though can someone step in and correct me/agree with me here?

[andywestken]

I am a bit concerned about the comment about c-style versus reinterpret cast

I have just checked Sutter and Alexandrescu (C++ Coding Standard) and it states "Don't use C-style casts".

Ever!

[Technyque]

Just talked to someone in my class. Never felt like such a noob in my life lol...

The .bin file good, I was supposed to use float instead of double in my struct. Don't know how I missed that, I even posted the email he sent our class with the directions, which states that those 2 values were float.

Thanks everyone that tried to help and sorry for wasting your time.

[andywestken]

:-)

The advantage of mistakes is that they help to "burn" facts into your mind!