[reportlab-users] What arguments does fp_str expect?

[Peter]

On Tue, Mar 24, 2009 at 4:55 PM, Robin Becker <robin at reportlab.com> wrote:

>

> Peter wrote:

>>

>> Hi all,

>>

>> Earlier today, I had a problem in the fp_str function in

>> reportlab/lib/utils.py where it was being passed three floats held as

>> strings (perhaps an RGB color tuple, I have not checked yet).  The

>> root problem may have been elsewhere (possibly in my own code), but it

>> would help to know what arguments this function expects.  There is no

>> docstring, but the comment at the top suggests string arguments might

>> be expected.

>

> fp_str is supposed to convert floats/ints into strings. It's not supposed to

> be able to convert arbitrary objects (including strings). Probably all this

> ought to disappear now that we have fixed point. The older versions of

> acrobat reader used 16.16 fixed point arithmetic so the precision required

> in the files was strictly limited.

>

> The *a argument in the python version indicates that we convert (or expect

> to convert) more than one value. As a special feature if there is only one

> argument and that argument is of type list/sequence then we make the

> arguments the contents of a[0].


OK - thanks!

That means the error I was seeing was triggered elsewhere.  I'll have
to investigate further...

Could you update the fp_str function's docstring with this information?

It took me a while to work out what the code was actually doing.  My
instinct would have been something more like this:

def fp_str(*a):
    if len(a)==1 and isSeqType(a[0]): a = a[0]
    s = []
    A = s.append
    for value in a:
        if abs(value) <= 1e-6:
            A('0')
        else:
            #Can use at most 6 dp, don't want leading or trailing zeros
            A(('%.6f' % value).strip("0"))
    return " ".join(s)
assert fp_str([0.000000123, 0.00000123, 0.0000123, 0.000123 , 0.123,
1.23, 12.3, -12.3]) \
       == '0 .000001 .000012 .000123 .123 1.23 12.3 -12.3'

Presumably your somewhat more cryptic version is much faster, and
avoids international floating point issues (comma versus period)?

As an aside, utils.pyt contains several uses of string.join(a_list,
a_sep) rather than a_sep.join(a_list), which might be worth fixing.

Peter