Re^2: Can you write a faster code to perform this task?

Replies are listed 'Best First'.
Re^3: Can you write a faster code to perform this task? by trizen (Hermit) on Sep 29, 2014 at 11:50 UTC
What about something like this? `use 5.014; say 'iiiiiiiiMMMMMMMMMMMooooooooooooMMMMMMMMMMiiiiiMMMMMMMMoooo' =~ tr/M/~/sr=~tr/~//; # => prints: 3` [download] Update: just to be sure, it would be safer to replace 'M' with something much exotic, like '\0', which should be OK, assuming that you're reading text files. (`tr/M/\0/sr =~ tr/\0//`)	[reply] [d/l] [select]
Re^4: Can you write a faster code to perform this task? by Lotus1 (Vicar) on Sep 29, 2014 at 18:30 UTC
I tried it with replacing 'M' with 'M' and it worked. No need to worry about exotic characters. `use 5.014; say 'iiiiiiiiMMMMMMMMMMMooooooooooooMMMMMMMMMMiiiiiMMMMMMMMoooo' =~ tr/M/M/sr=~tr/M//; # => prints: 3` [download]	[reply] [d/l]
Re^5: Can you write a faster code to perform this task? by BrowserUk (Patriarch) on Sep 29, 2014 at 19:38 UTC
No need for a new perl and the /r option either. This works fine: `#! perl -slw use strict; use Time::HiRes qw[ time ]; my $start = time; my $count=0; while( <DATA> ) { # $count += ()= m{M+}g; tr/M/M/s; $count += tr/M/M/; } print $count."\n"; printf "Took %9f secs\n", time() - $start;` [download] With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday' Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error. "Science is about questioning the status quo. Questioning authority". In the absence of evidence, opinion is indistinguishable from prejudice.	[reply] [d/l]
Re^4: Can you write a faster code to perform this task? by LanX (Saint) on Sep 29, 2014 at 12:07 UTC
Well TIMTOWTDI with two chained tr is not necessarily fast. :) But who knows, try to benchmark... :) Cheers Rolf _{(addicted to the Perl Programming Language and ☆☆☆☆ :)}	[reply]
Re^5: Can you write a faster code to perform this task? by Your Mother (Archbishop) on Sep 29, 2014 at 16:02 UTC
`tr` is MUCH faster than `s` though. I tried to work this out but didn’t hit on trizen’s cleverness. So… `# (v5.18.2) built for darwin-2level 'tr' => sub { $str =~ tr/M/~/sr=~tr/~//; }, -- count 3 regex 3 regex2 3 split 3 tr 3 Rate regex2 split regex count tr regex2 169354/s -- -62% -65% -75% -94% split 446500/s 164% -- -7% -34% -85% regex 481882/s 185% 8% -- -29% -84% count 674634/s 298% 51% 40% -- -78% tr 3044813/s 1698% 582% 532% 351% --` [download] Update: as trizen noted in his code, the `/r` flag to `tr` came in at 5.14. It won’t work on older Perls (update: <- stupid apostrophe begone).	[reply] [d/l] [select]
Re^6: Can you write a faster code to perform this task? by LanX (Saint) on Sep 29, 2014 at 17:56 UTC
Re^3: Can you write a faster code to perform this task? by Laurent_R (Canon) on Sep 29, 2014 at 17:25 UTC
True, Rolf, I mis-read the OP's requirement. I thought he wanted the number of "M". But, OTOH, trizen has found a way to use `tr///` which seems to be very fast.	[reply] [d/l]