write binary with struct.pack_into

[Chris Angelico]

On Sat, Oct 6, 2012 at 1:27 PM, palmeira <palmeira at gmail.com> wrote:
> import struct
> bloco='>%df' %(252)  #Binary format
>
> # READ
> fa=open('testIN.bin')
> my_array=struct.unpack_from(bloco,fa.read()[0*4:251*4])    # my_aray = 252
> elements array
> ## This read is OK!
>
> #WRITE
> fb=open('testOUT.bin')
> test=struct.pack_into(bloco,fb.write()[0*4:251*4])  # ERROR in this WRITE

You have a beautiful parallel here, but I think that's where your
problem is. In the READ section, you have fa.read() which will read
the whole file, and then you slice the resulting string. That's pretty
inefficient for large files, but it'll work.

But when you write, that completely does not work. (Even assuming
you've opened read/write, per Dennis's comment.)
>>> fb.write()
Traceback (most recent call last):
  File "<pyshell#37>", line 1, in <module>
    fb.write()
TypeError: write() takes exactly 1 argument (0 given)

It needs an argument of what to write out, it doesn't return a
writable buffer suitable for pack_into.

I recommend you completely rewrite your file handling to use actual
seeks and file writes. Also, you'll want (I think) to use binary mode
on your files; character encodings don't mean much when you're working
with arrays of numbers.

Finally, when you post issues, "ERROR in this WRITE" isn't very
helpful. Please please post the full traceback (like in my trivial
example above), as problems will be much more visible.

Hope that's of some value!

ChrisA