homepage

I found some unexpected behavior while working with the struct module.
>>> import struct
This works as expected:
>>> struct.pack('1s1s','3','4')
'34'
In this case, with bad input, the error message says I need 2 arguments, when I provide 2 arguments.
>>> struct.pack('1s1s','33')
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
struct.error: pack requires exactly 2 arguments

Also, I tested this on Windows in Python 3.2.3 and Windows in Python 2.7.2

Hmm. Yes, that's not very clear. The same message is used both for struct.pack and for Struct.pack (which struct.pack is really just a convenient alias for); it makes a bit more sense for the latter:
>>> struct.Struct('1s1s').pack('33')
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
struct.error: pack requires exactly 2 arguments

Funny. struct.pack(fmt, args...) is just an alias to struct.Struct(fmt).pack(args...). The error message should be changed to explicitly state that we are talking about the data for packing, and not about the arguments of function. Or should remove mention of the number of arguments at all (leave only "too much" or "too little").

It might help if the error message also stated how many arguments were actually received, like the TypeError message already does for bad function / method calls. E.g., 
"struct.error: pack expected 2 items for packing (got 1)"

> It might help if the error message also stated how many arguments were actually received, like the TypeError message already does for bad function / method calls. E.g., 
> 
> "struct.error: pack expected 2 items for packing (got 1)"
Yes, this would be useful. But seldom implemented.
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
TypeError: not enough arguments for format string
>>> '%s %s'%(123,456,789)
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
TypeError: not all arguments converted during string formatting
struct.pack also inconsistent in other error messages.
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
struct.error: argument for 's' must be a bytes object
>>> struct.pack('i', '123')
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
struct.error: required argument is not an integer
For "s" is mentioned format, and for "i" no. It would be helpful to
mention also the number of the item.

Changed the error message as suggested by http://bugs.python.org/issue14897#msg161493
Also made similar change to struct.pack_into

A second try, now taking into account buffer and offset params in pack_into.

LGTM.

New changeset 626ed0e06fd6 by Petri Lehtinen in branch '2.7':
#14897: Enhance error messages of struct.pack and struct.pack_into
http://hg.python.org/cpython/rev/626ed0e06fd6
New changeset a555bd4026b0 by Petri Lehtinen in branch '3.2':
#14897: Enhance error messages of struct.pack and struct.pack_into
http://hg.python.org/cpython/rev/a555bd4026b0
New changeset 70d5906e0461 by Petri Lehtinen in branch '3.3':
#14897: Enhance error messages of struct.pack and struct.pack_into
http://hg.python.org/cpython/rev/70d5906e0461
New changeset ebc588a3db51 by Petri Lehtinen in branch 'default':
#14897: Enhance error messages of struct.pack and struct.pack_into
http://hg.python.org/cpython/rev/ebc588a3db51 

Applied, thanks!

Matti Mäki, can you please submit a contributor form?
http://python.org/psf/contrib/contrib-form/
http://python.org/psf/contrib/ 

> "Enhance error messages of struct.pack and struct.pack_into"
You probably should have used the word 'fix' rather than 'enhance' here: else it smells like you're putting new features into a maintenance release. :-)

> You probably should have used the word 'fix' rather than 'enhance' here: else it smells like you're putting new features into a maintenance release. :-)
Ah, true. But unfixable now :(

I filled up the contributor form and gave it to Petri Lehtinen 2012年10月22日

Matti Mäki wrote:
> I filled up the contributor form and gave it to Petri Lehtinen 2012年10月22日
Yeah, and I posted it (along with 14 other contributor forms received
at PyCon Finland) to the PSF last week. They just seem to be a bit
slow to process them.