reading combined ascii & binary file in c++

[jjhall]

I need to open and read a file that has a combined ascii and binary format. Don't ask me why it is this way; I think it is kind of stupid, but I didn't write the program that generated the file. Anyway, there is a short header that is written in ascii, then the data is in binary. I have tried opening the file in ascii format first with

ifstream data("filename", ios::in);

Then I read to a buffer until I reach the end of the header. Next, I store the location with

int location;
location = data.tellg();

I then close the file and reopen it with

ifstream data_binar("filename", ios::in | ios::binary);

However, when I read the next field and use cout to print it to the screen, it is a bunch of symbols as if it is still binary. What do I need to do to recognize this as an int (or float, etc.)?

Thanks,

Jeremiah

[jjhall]

I've played around with the binary mode some more, and I get confusing results. I have tried opening the file using

file.open(filename, ios::in | ios::binary);

. Then I use the seekg function to move to the beginning of the binary data. (I know where the position of the binary data is explicitly.) However, when I use the stream extraction operator, or the file.read function, I get the same jibberish as when I open the file without using ios::binary. Also, it reads the binary data at the head of the file correctly, as if it is not a binary stream at all.

Has anybody dealt with files that have both ascii and binary data in them?

[weaknessforcats]

You need to know the format of the file. An int of 123 followedby a float of 45.67 would look like 12345.67. You can't know there is an int and a float here.

I recommend a flat file. That is, a file with all records the same length and each field with a constant width. This way you know ahead of time that there is an int followed by a float. Just read the correct number of bytes base don yur record format.

Another trick is to use a field separator: 123|45.67| etc... This way you fetch bytes until you get the separator. Then convert your buffer into the correct type by using a conversion function.

[jjhall]

The problem is that I already have the files in a certain format, and I want to use them as they are. I know that all of the data in the binary section of the file will be doubles. The data is written with the C command fwrite, so I am not sure what type of delimiter is used, but I am assuming that it is either a space or a null character (?).

I think there is also a possibility that there is an endian issue. I don't know very much about this, but the documentation for the files indicated that there was a character written to the file to determine whether it is big or little endian. Could that affect my ability to read binary data from the file?

[weaknessforcats]

Absolutely.

Little endian (Windows) writes an int LSB first so when the file is read the LSB is read first and put into the low memory location.

Big endian (Unix/Linux) writes the MSB first.

You will need to interrogate the endian character in the fiule inorder to kniow how to read the file.

If the variables have separators like a space or null character, then you should be able to read them without ambiguity.

[jjhall]

Absolutely.

Little endian (Windows) writes an int LSB first so when the file is read the LSB is read first and put into the low memory location.

Big endian (Unix/Linux) writes the MSB first.

You will need to interrogate the endian character in the fiule inorder to kniow how to read the file.

If the variables have separators like a space or null character, then you should be able to read them without ambiguity.

The file was written on an SGI machine, and I'm trying to read it on a Linux box. How do I interrogate the endian character? Is it the first character of the file?

Also, what are the LSB and MSB?

Thanks for all your help.

[weaknessforcats]

The endian character should be one byte. Probably a char. Endian issues do not apply if there is one byte so you can read this indicator regardless. Based on what it says you read the reat of the file either little endial or big endian.

LSB is Least Significant Byte.
MSB is Most Significant Byte.