Reading from a binary file in a portable way

[John Purser]

"Steve Holden" <sholden at holdenweb.com> wrote in message
news:p9yv8.11123$eS3.9630 at atlpnn01.usenetserver.com...
> "jb" <egrossde at yahoo.de> wrote ...
> >
> > I should like to read a (binary) file x.y in a buffer and then do things
> > like getting the 16 bit integer that is at the offset n, n+1 in the file
> > (maybe lowbyte, then higyhbyte.
> >
> > Previously I did it only under MS-Windows where for example I could open
> the
> > file with the flag O_BINARY but this is no longer possible under Linux.
> >
> > What would a solution that works under both Unix and MS-Windows look
like?
> >
> Such a solution would probably use the mmap module. Although there are
minor
> differences on the two platforms it should be possible to write something
> that works well on both.
>
> regards
>  Steve


I just did this.  An outline of my code would look like:

import struct

fInput = file('im1.d', 'r+b')

fInput.seek(512)

record = fInput.read(180)

item_number = struct.unpack('> i', record[1:5])

print item_number

fInput.close()

It worked nicely on Windows 2000 (Python 2.2) and Debian Linux (Python 2.1).
I was reading and writing to the input file and needed data from the whole
record so there's some extra lines.  I'd suggest writing a fuction that
positioned the file pointer in the correct position based on record number,
record length, data slice, header, etc.

Good Luck,

John Purser