Read double precision numbers from binary file

Hello,

I have to read double precision numbers from a binary file. I'm used to do that in C, which basically looks like that for a LSB first coded file :

char *cAddress = NULL;
double value = 0;

cAddress = (char*)(value);
*cAddress = ucBuffer[uiOffset];
*(cAddress + 1) = ucBuffer[uiOffset + 1];
*(cAddress + 2) = ucBuffer[uiOffset + 2]; 
*(cAddress + 3) = ucBuffer[uiOffset + 3];
*(cAddress + 4) = ucBuffer[uiOffset + 4]; 
*(cAddress + 5) = ucBuffer[uiOffset + 5]; 
*(cAddress + 6) = ucBuffer[uiOffset + 6]; 
*(cAddress + 7) = ucBuffer[uiOffset + 7];

Now, I would like to implement that in Matlab. Do you have any idea on how I should proceed ?

Thanks a lot for supporting

Franck

Answer 1

[James Tursa]

Did you mean to type the following:

cAddress = (char*)(&value); // Added the & for type punning

Otherwise, the method of getting the bytes into the double isn't apparent. In any event, you should be able to simply use fopen with the proper machinefmt option, followed by fread with the '*double' option. Are doubles the only thing(s) in the file?

[Guillaume]

Yes, the OP must have forgotten the & because otherwise the code most likely crashes.

I'd add that the fread implementation of matlab is a lot more robust than the C implementation presented here, as it can deal with difference of endianness between the file and the computer doing the reading (the machinefmt option that James is referring to).

[James Tursa]

... the code most likely crashes."

I don't even think it will compile. I would be surprised if a compiler let you convert a double value directly into a pointer.

[Franck D'Araujo]

You are right, I forgot the &. That's just because the code is a copy/paste from a function where value is already a pointer !

I will take a close look to the fread documentation like proposed.

Thanks for helping

Franck

Answer 2

[Adam]

doc fread

That help page should give you the information you need on precision and other aspects of binary file reading, together with examples. It should be a lot simpler than the raw C equivalent.

[Franck D'Araujo]

Hello, I checked the fread doc, and it's just perfect for my needs. Thank you for pointing it out ! I'm more familiar with C programming than Matlab, so I didn't realize that the Matlab fread function had so many options, particularly the endianness management !

Thanks again for helping

Franck