Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

format() not behaving as expected

56 views
Skip to first unread message

Josh English

unread,
Jun 29, 2012, 12:31:53 PM6/29/12
to
I have a list of tuples, and usually print them using:

print c, " ".join(map(str, list_of_tuples))

This is beginning to feel clunky (but gives me essentially what I want), and I thought there was a better, more concise, way to achieve this, so I explored the new string format and format() function:

>>> c = (1,3)
>>> s = "{0[0]}"
>>> print s.format(c)
'1'
>>> print format(c,s)
Traceback (most recent call last):
File "<interactive input>", line 1, in <module>
ValueError: Invalid conversion specification

I'm running *** Python 2.7.2 (default, Jun 12 2011, 15:08:59) [MSC v.1500 32 bit (Intel)] on win32. ***
(This is actually a PortablePython run on a Windows 7 machine)

Any idea why one form works and the other doesn't?

MRAB

unread,
Jun 29, 2012, 1:02:45 PM6/29/12
to pytho...@python.org
The ".format" method accepts multiple arguments, so the placeholders in
the format string need to specify which argument to format as well as
how to format it (the format specification after the ":").

The "format" function, on the other hand, accepts only a single
argument to format, so it needs only the format specification, and
therefore can't accept subscripting or attributes.

>>> c = "foo"
>>> print "{0:s}".format(c)
foo
>>> format(c, "s")
'foo'

Steven D'Aprano

unread,
Jun 29, 2012, 1:08:20 PM6/29/12
to
On Fri, 29 Jun 2012 09:31:53 -0700, Josh English wrote:

> I have a list of tuples, and usually print them using:
>
> print c, " ".join(map(str, list_of_tuples))
>
> This is beginning to feel clunky (but gives me essentially what I want),
> and I thought there was a better, more concise, way to achieve this, so
> I explored the new string format and format() function:
>
>>>> c = (1,3)
>>>> s = "{0[0]}"
>>>> print s.format(c)
> '1'

That's not actually the output copied and pasted. You have quotes around
the string, which you don't get if you pass it to the print command.


>>>> print format(c,s)
> Traceback (most recent call last):
> File "<interactive input>", line 1, in <module>
> ValueError: Invalid conversion specification
[...]
> Any idea why one form works and the other doesn't?

Because the format built-in function is not the same as the format string
method.

The string method takes brace substitutions, like "{0[0]}" which returns
the first item of the first argument.

The format function takes a format spec, not a brace substitution. For
example:

>>> format(c, '<10s')
'(1, 3) '
>>> format(c, '>10s')
' (1, 3)'
>>> format(c, '*>10s')
'****(1, 3)'

The details of the format spec are in the Fine Manual:

http://docs.python.org/library/string.html#formatspec

although sadly all the examples are about using brace substitutions, not
format specs.

(Personally, I find the documentation about format to be less than
helpful.)

You can also read the PEP that introduced the new formatting, but keep in
mind that there have been some changes since the PEP was written, so it
may not quite match the current status quo.

http://www.python.org/dev/peps/pep-3101/


--
Steven

Josh English

unread,
Jun 29, 2012, 1:19:30 PM6/29/12
to pytho...@python.org
On Friday, June 29, 2012 10:02:45 AM UTC-7, MRAB wrote:
>
> The ".format" method accepts multiple arguments, so the placeholders in
> the format string need to specify which argument to format as well as
> how to format it (the format specification after the ":").
>
> The "format" function, on the other hand, accepts only a single
> argument to format, so it needs only the format specification, and
> therefore can't accept subscripting or attributes.
>
> >>> c = "foo"
> >>> print "{0:s}".format(c)
> foo
> >>> format(c, "s")
> 'foo'

Thank you. That's beginning to make sense to me. If I understand this, everything between the braces is the format specification, and the format specification doesn't include the braces, right?

Josh

Josh English

unread,
Jun 29, 2012, 1:19:30 PM6/29/12
to comp.lan...@googlegroups.com, pytho...@python.org
On Friday, June 29, 2012 10:02:45 AM UTC-7, MRAB wrote:
>
> The ".format" method accepts multiple arguments, so the placeholders in
> the format string need to specify which argument to format as well as
> how to format it (the format specification after the ":").
>
> The "format" function, on the other hand, accepts only a single
> argument to format, so it needs only the format specification, and
> therefore can't accept subscripting or attributes.
>
> >>> c = "foo"
> >>> print "{0:s}".format(c)
> foo
> >>> format(c, "s")
> 'foo'

Josh English

unread,
Jun 29, 2012, 1:24:58 PM6/29/12
to
On Friday, June 29, 2012 10:08:20 AM UTC-7, Steven D&#39;Aprano wrote:
> >>>> c = (1,3)
> >>>> s = "{0[0]}"
> >>>> print s.format(c)
> > '1'
>
> That's not actually the output copied and pasted. You have quotes around
> the string, which you don't get if you pass it to the print command.
>

Mea culpa. I typed it in manually because the direct copy and paste was rather ugly full of errors because of many haplographies.



> >>>> print format(c,s)
> > Traceback (most recent call last):
> > File "<interactive input>", line 1, in <module>
> > ValueError: Invalid conversion specification
> [...]
> > Any idea why one form works and the other doesn't?
>
> Because the format built-in function is not the same as the format string
> method.
...
>
> (Personally, I find the documentation about format to be less than
> helpful.)
>

Thanks. I think it's coming together.

Either way, this format seems uglier than what I had before, and it's longer to type out. I suppose that's just programmers preference.

Josh

MRAB

unread,
Jun 29, 2012, 1:41:29 PM6/29/12
to pytho...@python.org
No, the format specification is the part after the ":" (if present), as
in the example.

Here are more examples:

>>> c = "foo"
>>> print "{0}".format(c)
foo
>>> format(c, "")
'foo'
>>> print "To 2 decimal places: {0:0.2f}".format(1.2345)
To 2 decimal places: 1.23
>>> print format(1.2345, "0.02f")
1.23
0 new messages