disappearing digit with regex and tr

zentara has asked for the wisdom of the Perl Monks concerning the following question:

Hi, Someone on the gtk2 perl maillist was asking a question about number text returned from a label, being in European and commafied format, and not acting as a number,

Well I thought, this ought to be easy with a tr or regex, but it isn't. So before I blow my mind on Friday, I figure I would ask about this.

The code below, takes a float, converts it to European format, then I attempt, to tr ( or regex ) the comma's out. I keep losing the digit 7. ???? I can't make any sense why, although I did find a clunky way (shown below).

Can the regex experts explain what is happening?

#!/usr/bin/perl

my $num_in = 1234567.89;
print "$num_in\n";
my $num_out =  
scalar reverse join "", map { s/(\d{2})\./$1,\b/; s/^(\d{1,3})$/\.$1/;
+ $_ } (reverse sprintf("%.2f", $num_in)) =~ m/.{1,3}/g;
print "$num_out\n";

my $get_text = $num_out;
print "$get_text\n";

#this loses the 7  
# $get_text  =~ s/\.//g;  
# tr is similar
 $get_text  =~ tr/.//d;  

#this clunky method works to remove commas
# $get_text  =~ s/\./ /g;
#  # $get_text  =~ s/\s+//g; #won't work to remove spaces 
# $get_text  =~ s/\s+//;
# $get_text  =~ s/\s+//;

#this then works to change decimal point to . 
$get_text =~ tr/,/./d;

# now if I try to force numeric context it fails
#$get_text = $get_text + 0.00;

print "$get_text\n";
[download]

I'm not really a human, but I play one on earth. Cogito ergo sum a bum

Comment on disappearing digit with regex and tr Download Code

Replies are listed 'Best First'.
Re: disappearing digit with regex and tr by ikegami (Patriarch) on Aug 18, 2006 at 15:04 UTC
I don't know why (don't have time to find out), but you're printing a backspace. `Before the s///: length = 14 31 2e 32 33 34 2e 35 36 37 2e 08 2c 38 39 1 . 2 3 4 . 5 6 7 . ^H , 8 9 After the s///: length = 11 31 32 33 34 35 36 37 08 2c 38 39 1 2 3 4 5 6 7 ^H , 8 9` [download] Update: It's the `\b`. `\b` means backspace in a string literal. Update: Here's a quickly written solution: `my $num_in = 1234567.89; my ($i, $d) = split /\./, $num_in; $i = reverse($i); $i =~ s/\G(\d{3})(?!$)/$1./g; $i = reverse($i); my $num_out = "$i,$d"; print($num_out, "\n");` [download]	[reply] [d/l] [select]
Re: disappearing digit with regex and tr by Velaki (Chaplain) on Aug 18, 2006 at 15:13 UTC
After running your snippet of code and piping it through `cat -v`, I get: `1234567.89 1.234.567.^H,89 1.234.567.^H,89 1234567^H.89` [download] It looks like you have a remnant of a word boundry in your regex in that long statement, but because it's in the replacement side of the substitution, it's being treated as a backspace. Perhaps you meant it to go in the search portion of the substitution section? Hope this helped, -v. "Perl. There is no substitute."	[reply] [d/l] [select]
Re: disappearing digit with regex and tr by duff (Parson) on Aug 18, 2006 at 15:06 UTC
I'd say it has something to do with that backspace you're inserting into the string in the first substitution of your map. Don't do that :-) you can see that the missing 7 is really still there by piping the output of your program through some utility like `xxd` to get a raw dump. duff	[reply] [d/l]
Re: disappearing digit with regex and tr by zentara (Cardinal) on Aug 18, 2006 at 15:45 UTC
Thanks. I'll have to wait till tommorrow to upvote you guys, I'm all out today. I guess I deserve what I got for not checking that map carefully. Sneaky backspace. I'm not really a human, but I play one on earth. Cogito ergo sum a bum	[reply]
Golf, anyone? :) by suaveant (Parson) on Aug 18, 2006 at 17:40 UTC
`perl -e '$_=1234567;s/\./,/;s/(?=(\d\d\d)+(,\|\Z))/./g;print'` - Ant - Some of my best work - (1 2 3)	[reply] [d/l]
Re: Golf, anyone? :) by Sidhekin (Priest) on Aug 18, 2006 at 17:56 UTC
`perl -e '$_=1234567;s/\./,/;s/(?=(\d\d\d)+(,\|\Z))/./g;print'` That doesn't handle 234567 nor 234567.89 too well. :) My best, based on yours, obviously: `y/./,/;s/\d(?=(\d\d\d)+(,\|\Z))/$&./g; # 37, counting the semicolon` [download] Update: Ow, too obviously. \Z is just long-hand for $ ... so slice one off ... and update2 \d\d\d can be shortened to \d{3} for one more: `y/./,/;s/\d(?=(\d{3})+(,\|$))/$&./g; # 35, counting the semicolon` [download] ... or, to demonstrate ... `sidhekin@blackbox:~$ perl -le 'for (@ARGV) { y/./,/;s/\d(?=(\d{3})+(,\| +$))/$&./g;print }' 1234567.89 234567.89 1234567 234567 1.234.567,89 234.567,89 1.234.567 234.567 sidhekin@blackbox:~$` [download] `print "Just another Perl ${\(trickster and hacker)},"` The Sidhekin proves Sidhe did it!	[reply] [d/l] [select]
Re^2: Golf, anyone? :) by suaveant (Parson) on Aug 18, 2006 at 18:02 UTC
Ooops. Just needs a little positive lookbehind. `perl -le 'for(@ARGV){s/\./,/;s/(?<=\d)(?=(\d\d\d)+(,\|\Z))/./g;print}' +1234567.89 234567.89 1234567 234567 1.234.567,89 234.567,89 1.234.567 234.567` [download] update.. Ok.. yours is shorter... mine is probably more efficient for not invoking $& at least :) - Ant - Some of my best work - (1 2 3)	[reply] [d/l]