Consider also that regex are so optimezed so that the second option, using substr, 'may' become a 'little' faster then regexes only when you operate against huge data in a foreach loop:

#10 fields in the array iterating over:
Benchmark: running Regex, Substr for at least 4 CPU seconds...
     Regex: 4.58121 wallclock secs ( 4.18 usr +  0.02 sys =  4.20 CPU)
+ @ 3968640.46/s (n=16656384)
    Substr: 4.38291 wallclock secs ( 4.32 usr +  0.00 sys =  4.32 CPU)
+ @ 3927240.92/s (n=16969608)

#100 fields
Benchmark: running Regex, Substr for at least 4 CPU seconds...
     Regex: 4.56133 wallclock secs ( 4.12 usr + -0.01 sys =  4.11 CPU)
+ @ 3995786.36/s (n=16402703)
    Substr: 4.29298 wallclock secs ( 4.03 usr +  0.00 sys =  4.03 CPU)
+ @ 4190639.34/s (n=16871514)

#1000 fields ###Regex is faster!! (tested 5 times)
Benchmark: running Regex, Substr for at least 4 CPU seconds...
     Regex: 4.1849 wallclock secs ( 4.03 usr +  0.01 sys =  4.04 CPU) 
+@ 4139558.42/s (n=16723816)
    Substr: 5.5707 wallclock secs ( 4.26 usr +  0.00 sys =  4.26 CPU) 
+@ 3843370.27/s (n=16368914)

#10000 fields
Benchmark: running Regex, Substr for at least 4 CPU seconds...
     Regex: 4.87697 wallclock secs ( 4.68 usr +  0.00 sys =  4.68 CPU)
+ @ 3600221.37/s (n=16849036)
    Substr: 4.21117 wallclock secs ( 3.99 usr +  0.03 sys =  4.02 CPU)
+ @ 4352730.93/s (n=17519742)

#BUT ALSO 10000 fields
Benchmark: running Regex, Substr for at least 4 CPU seconds...
     Regex: 4.32358 wallclock secs ( 4.09 usr + -0.02 sys =  4.07 CPU)
+ @ 3946014.74/s (n=16064226)
    Substr: 4.44415 wallclock secs ( 4.10 usr +  0.03 sys =  4.13 CPU)
+ @ 3892862.84/s (n=16093095)
[download]

L*