Re: Gawk match() and numbers in scientific notation

On Tuesday 6 May 2008 13:16, Hermann Peifer wrote:

> Hi,
> 
> 
> I am somehwat puzzled with match() results for numbers in scientific
> notation. See below.
> 
> $ cat testdata
> 100
> 100e-3
> 100E3
> 
> I am wondering what kind of uppercase character is matched in record
> 2:
> 
> $ gawk '{print $1,match($1,/[A-Z]/)}' testdata
> 100 0
> 100e-3 4
> 100E3 4

Try this with LC_ALL=C (always a good idea when working with
locale-sensitive data):

$ LC_ALL=C gawk '{print $1,match($1,/[A-Z]/)}' testdata
100 0
100e-3 0
100E3 4

> I am also wondering about the non-matches in these cases:
> 
> $ gawk 'BEGIN{print match(100e-3,/[A-Z]/)}'
> 0
> $ gawk 'BEGIN{print match(100E3,/[A-Z]/)}'
> 0

Unlike before, here you are asking awk to do number-to-string conversion.
The problem is (I think) that the numbers lose the Es when they are
converted to strings:

$ gawk 'BEGIN{printf "%s\n", 100E3}'
100000
$ gawk 'BEGIN{printf "%s\n", 100e-3}'
0.1

Compare with this (which also shows the influence of the locale):

$ gawk 'BEGIN{print match("100E3",/[A-Z]/)}'
4
$ gawk 'BEGIN{print match("100e-3",/[A-Z]/)}'
4
$ LC_ALL=C gawk 'BEGIN{print match("100E3",/[A-Z]/)}'
4
$ LC_ALL=C gawk 'BEGIN{print match("100e-3",/[A-Z]/)}'
0


Of course, I stand to be corrected for anything wrong I could have
written.

-- 
All the commands are tested with bash and GNU tools, so they may use
nonstandard features. I try to mention when something is nonstandard (if
I'm aware of that), but I may miss something. Corrections are welcome.