Need to extract only decimal numbers for a glob of text [SOLVED]

Question

If you have a look at /dev/zero's thread here, you'll see that users have been posting the output of his script which are numbers that range from 2 to 5 decimal places.  If I dump this entire thread to txt file, how can I:
1) Delete everything except for numbers of the following formats (where 'x' is a digit and '.' is a decimal point)?
2) Format the output to be one target per line?
x.xx
x.xxx
x.xxxx
x.xxxxx
xx.xx
xx.xxx
xx.xxxx
xx.xxxxx
I have experimented with some sed strings but am not making any traction.  Perhaps your perl or awk ninjas have a good solution?
Here is the source file which was generated from a copy/paste of that thread into an empty text file: http://pastebin.com/ZkRFhFAr
Last edited by graysky (2012-12-21 11:19:05)

Answer 1

Ooh, a challenge...
* 1 hour later *
#!/usr/bin/perl -an
@ary = grep /^\d{1,2}\.\d{2,5}$/, @F;
print join("\n", @ary) . "\n" if @ary;
Run as ./script.pl unixness_thread.txt
Not perfect, it grabs 10.04 which is an Ubuntu version in context, and 4.10 which is an Xfce version... but not too far off. Stick a "sort -n" on the end and knock off the obvious outliers and you'll be half there.
Edit -- just noticed it skips at least one number in parentheses: (8.73086). Was thinking the default splitting behavior would not be a problem but I was wrong. Not sure how to fix this but I think it can be done with -F.
Edit again -- changing the shebang as follows seems to work:
#!/usr/bin/perl -an -F/[^\d\.]/
Last edited by Trent (2012-12-21 01:55:08)