Reading/writing binary problem

Ruby
1

Hi,

I have some binary files which I need to delete some data at the
beginning of the file (to extract the raw data out of the file). I got a
matlab file from a colleque but I would prefer if I could do it with
ruby.

This matlab code does the trick:
function m=write_rad(filename)
file=fopen(filename);
f= fread(file,‘uint16’);

[r,k]=size(f);
if k==1;
		t=r-1048576;
		m=f(t+1:r,:);
end
end % function

How can I do this in ruby? I’m completely lost.

Thanks
Wolfgang

2

Wolfgang wrote:

[r,k]=size(f);
if k==1;
       t=r-1048576;
       m=f(t+1:r,:);
end
end % function

How can I do this in ruby? I’m completely lost.

An easy solution is to copy the file:

Untested:

def cut_prefix(file, length)
raise ArgumentError, “negative length” if length < 0

bak = file + “.bak”
File.rename file, bak

File.open(bak,“rb”) do |io_r|
io_r.seek length

 File.open(file, "wb") do |io_w|
   while ( buffer = io_r.read(4096) )
     io_w.write(buffer)
   end
 end

end
end

Another solution would be to do it internally with continuously seeking,
reading, seeking, writing etc.

Kind regards

robert
3

On Wed, 29 Mar 2006, Wolfgang wrote:

Wolfgang
this sould be easy if i understood that method and the nature of your
data :wink:

here’s what i understand this method to do :

open the file

 file=fopen(filename);

read all available unit16 quantities into an [m, n] array

 f= fread(file,'uint16');

check the size of this array, r is the number of elements read

 [r,k]=size(f);

k will be one iff data was read into r dimension

 if k==1;

here i assume 1048576 is the ‘row_size’ of 1mb. we set t to be the

offset

of the last row

 t=r-1048576;

because matlab is evil and uses one based indexing, we inc t by one

and

extract the last row/block of data. but here it seems like the ith

index

should be t+1:r-1048576 no? otherwise it seems like this would be

out of

bounds?

 m=f(t+1:r,:);

in any case this looks alot like you are trying to extract the last 1mb
of
uint16 quantities from a data file - is that right? let me know and
i’ll show
you how to do this.

regards.

-a

4

On Thu, 30 Mar 2006, Wolfgang wrote:

PS: I will also test the example of Robert

ok. so, just to clarify, you want to read a 2mb matrix out of a file
after
skiping 13312 bytes of crap. this will do that

def extract_data path
open(path){|f| f.seek(13312); f.read(102410242)}
end

here data returned is a buffer. if you want a matrix object i
highly
suggest installing the narray module and using something like

require ‘narray’
def extract_matrix path
na = NArray.sint 1024, 1024
open(path){|f| f.seek(13312); na[] = f.read(102410242)}
na
end

this will return a numerical matrix with nice operators. you can write
it to
file with

open(‘m.dat’,‘w’){|f| f.write na}

and this will write the matix in binary.

regards.

-a

5

On Mar 29, 2006, at 6:31 PM, [email protected] wrote:

open(path){|f| f.seek(13312); f.read(102410242)}

If you are using Windows don’t forget to open it in binary mode:

open(…, ‘rb’) …

– Daniel

6

Hello,

yes I want to extract a matrix of 1024x1024 of binary data (2bytes for
each value). But ahead of this data is 1048576 bit (13312 bytes) of
crap. I don’t know matlab at all and I don’t like to start this routine
when I want to import that data into my application, which can not skip
this header. If it is easier to do by a linux command (its running in
cygwin) any help is also welcome.

Wolfgang

PS: I will also test the example of Robert

[email protected] schrieb:

7

On Thu, 30 Mar 2006, Daniel H. wrote:

On Mar 29, 2006, at 6:31 PM, [email protected] wrote:

open(path){|f| f.seek(13312); f.read(102410242)}

If you are using Windows don’t forget to open it in binary mode:

open(…, ‘rb’) …

– Daniel

ah. indeed.

-a

8

Thank you all,

this worked perfectly!

Wolfgang

[email protected] schrieb:

9

Daniel H. wrote:

On Mar 29, 2006, at 6:31 PM, [email protected] wrote:

open(path){|f| f.seek(13312); f.read(102410242)}

If you are using Windows don’t forget to open it in binary mode:

open(…, ‘rb’) …

In fact I recommend to use that switch regardless of OS used: it’s more
portable and also serves as documentation hint that the file is actually
binary.

Kind regards

robert