如何使用Perl正则表达式从等式中提取具有非零系数的变量？

Question

I wish to grab the non-zero coefficients of a particular number arrangement. I have the set of numbers below and I wish to strictly extract with regex the non-zero coefficient ( x2 , x3 , x5 ) from

x1 *0.000000+ x2 *-0.100000+ x3 *0.850000+ x4 *0.000000+ x5 *0.000056+ x6 *0.000000 

I tried doing this

(\*[-+]?[0]*\.?[0]+)

but am not really sure as to declare that it shouldn't consider the "*0.000000"

Answer 1

Match x-digit not followed by space-star-zero-dot-zeroes-endOfWord:

x\d+(?! \*0\.0+\b)

See live demo .

Answer 2

I would recommend using the fact that you have full-blown programming language at your disposal to simplify the regex to capture all coefficients, and then use grep to eliminate the zeros:

#!/usr/bin/env perl

use strict;
use warnings;

my $eqn = "x1 *0.000000+ x2 *-0.100000+ x3 *0.850000+ x4 *0.000000+ x5 *0.000056+ x6 *0.000000 ";

my %eqn = ($eqn =~ /(x[0-9]+) \s [*] (-?[0-9.]+) [+]/gx);

print "$_\n" for grep $eqn{$_} != 0, keys %eqn;

Output (order will not be consistent unless you sort variable names):

x3 
x5 
x2

This might look like too much extra work, but having the equation parsed into a hash which maps variables to coefficients may actually save you work down the road depending on the exact nature of what you are doing. I am assuming you are doing some kind of model selection based on coefficient magnitudes.

Answer 3

Do:

x\d+(?!\s*\*0\.0*(?:[+\s]|$))

• x\\d+ gets x followed by one or more digits
• The zero width negative lookahead pattern (?!\\s*\\*0\\.0*(?:[+\\s]|$)) ensures that the match is not followed by zero co-efficients

Demo

---

More robust, avoiding false positives in the match:

x\d+(?!\s*\*0\.0*(?:[+\s]|$))(?=\s*\*-?\d+(?:\.\d*)?(?:[+\s]|$))

Demo