Hopefully simple question, input streams

Hopefully this is a simple question. I searched out for it, but found nothing. Basically I just need to know some simple info.

Basically I am reading in a file in binary using ifstream. All is fine and good, then I hit some spaces, and then it gets all wacky from there.

I setup a char array of 4 bytes and read in 4 bytes at a time. This works for a while until I hit a part of the file and then it starts to read spaces and messes up what I am reading.

So for example:

...

Code:

---

char *tempPtr;
char tempVal[4];

in.read( tempVal, 4 );  //say for instance a line in hex 34F2
//format the string into hex
long value = strtol(  tempVal, &tempPtr, 16 );

---

then it hits a line with a space, ( I can see the raw file as well, and in a hex editor it ignores these, so I can assume there is a way to decide what is a is not a space / \n), and then gets a little funky...

same code as before, on a particular 8 bytes...

Code:

---

char *tempPtr;
char tempVal[4];

in.read( tempVal, 4 );  //line is 0000 0106 0DEE 05AB
//format the string into hex
CString temp;
temp.Format( "%02X%02X%02X%02X", tempVal[0], tempVal[1], tempVal[2], tempVal[3] ); 

in.read( tempVal, 4 );  // read in the second part of the above line
//format the string into hex
CString temp1;
temp1.Format( "%02X%02X%02X%02X", tempVal[0], tempVal[1], tempVal[2], tempVal[3] ); 

CString joined = temp + temp1;

---

This should come out as normal, but when I hit the spaces it comes out more like this...

0000 0106 0DFFFFEEFFFF05AB

which to me seems impossible, but hey, I am no expert either :).

Any simple mistakes anyone sees with that? I am hesitant to eradicate all FF's ( space in ascii ) because there are of course numbers which have FF in them. Any ideas on this? I am kinda not able to spot it, but perhaps someone has a good way to id them. Thanks! :)

looks to me like you wrote the file yourself... bad times if you're really reading it in as binary...

...and for some reason those reads look wrong...

No I didn't write it myself. I can hex edit it, its generated and readable in a separate program, so I know this one is "correct". And yes it is written as a binary file, 'tis how they decided to format the damn thing :(. The reads are correct according to the standard, unless I somehow misread it...

well if it were a true binary file, there should be no problem with spaces... you have to read it in the same way you wrote it out, unless you're paying close attention to what you're doing to it.

In other words, I need to see how it was written to see what's wrong with how it's being read =P

Haha well I can post the file when I get to the computer with it. This was a file written via a java applet, that is, using DataInputStream.

well that could be your problem...

I think you meant the file was written with DataOutputStream, which, by definition, outputs something meant to be read in by DataInputStream. How exactly it does that depends on Sun's implementation.

you need to know exactly how Sun wrote the data out to the file, which means you need to remember what method you used to write it out, and then in C++ figure out how sun's data types differ from your platform's data types and make the conversion.

for example, writeBytes(String s) writes each character as one Byte discarding the high-order bits, whereas writeChars(String s) writeas each character as two Bytes. On my system, each character is saved in one Byte. Therefore, if I used writeChars(String s), I'd have to discard the upper 8 bits somehow (possibly with a double-read). I'd also have to make sure all the data was packed tight, with no empty spaces (because that's how fstream reads binary files)

the point is: DataOutputStream outputs data in a way that can be read by DataInputStream, but that doesn't necessarily mean it's a true binary file. Any spaces in a real binary file are data that the text editor is trying to display in an incompatible character set (ASCII).

Here's some code that writes and then reads a binary data file in C++:

Code:

---

#include <iostream>
#include <fstream>

int main()
{
        char abc[7]="ABCDEF";
        double num=3.14159;
        std::fstream file("test.dat",std::ios::out|std::ios::trunc|std::ios::binary);
       
        file.write(abc,sizeof(char)*7);
        file.write(reinterpret_cast<char*>(&num),sizeof(num));
       
        file.close();
        file.clear();
        abc[0]='\0';
        num=0.0;

        file.open("test.dat",std::ios::in|std::ios::binary);
        file.read(abc,sizeof(char)*7);
        file.read(reinterpret_cast<char*>(&num),sizeof(num));
        file.close();

        std::cout<<abc<<std::endl<<num<<std::endl;
        return 0;
}

---

and here's what the data file looks like in GVIM:

Code:

---

ABCDEF^@n~F^[ðù!        @

---

(the blue characters are control characters, and really shouldn't be visible)

Quote:

---

Originally Posted by dpro

Reading 0000 0106 0DEE 05AB

This should come out as normal, but when I hit the spaces it comes out more like this...

0000 0106 0DFFFFEEFFFF05AB

which to me seems impossible, but hey, I am no expert either :).

---

EE is a negative value and is printing as FFFFEE. Make the variable tempVal unsigned and it should print properly.

I would agree with you major_small, except it is completely readable. hex workshop reads the values correctly every time it loads the file. This would indicate that it should work.

WaltP, that seemed to work, it is reading in the correct value. Thanks, I will see how this works for the rest and go from there, thanks again!