From: Bo Borgerson Date: Wed, 26 Aug 2009 17:03:02 +0000 (-0400) Subject: split: Improve performance in scalar context X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=commitdiff_plain;h=c1a7495a9639962ce446532c71ba34cb952935b4;p=p5sagit%2Fp5-mst-13.2.git split: Improve performance in scalar context Improve the performance of split in scalar context --- diff --git a/pp.c b/pp.c index 8a686cd..e3b7798 100644 --- a/pp.c +++ b/pp.c @@ -4882,11 +4882,13 @@ PP(pp_split) I32 iters = 0; const STRLEN slen = do_utf8 ? utf8_length((U8*)s, (U8*)strend) : (STRLEN)(strend - s); I32 maxiters = slen + 10; + I32 trailing_empty = 0; const char *orig; const I32 origlimit = limit; I32 realarray = 0; I32 base; const I32 gimme = GIMME_V; + const bool gimme_scalar = (GIMME_V == G_SCALAR); const I32 oldsave = PL_savestack_ix; U32 make_mortal = SVs_TEMP; bool multiline = 0; @@ -4985,9 +4987,17 @@ PP(pp_split) if (m >= strend) break; - dstr = newSVpvn_flags(s, m-s, - (do_utf8 ? SVf_UTF8 : 0) | make_mortal); - XPUSHs(dstr); + if (gimme_scalar) { + iters++; + if (m-s == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + dstr = newSVpvn_flags(s, m-s, + (do_utf8 ? SVf_UTF8 : 0) | make_mortal); + XPUSHs(dstr); + } /* skip the whitespace found last */ if (do_utf8) @@ -5015,9 +5025,18 @@ PP(pp_split) m++; if (m >= strend) break; - dstr = newSVpvn_flags(s, m-s, - (do_utf8 ? SVf_UTF8 : 0) | make_mortal); - XPUSHs(dstr); + + if (gimme_scalar) { + iters++; + if (m-s == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + dstr = newSVpvn_flags(s, m-s, + (do_utf8 ? SVf_UTF8 : 0) | make_mortal); + XPUSHs(dstr); + } s = m; } } @@ -5030,34 +5049,49 @@ PP(pp_split) or split //, $str, $i; */ - const U32 items = limit - 1; - if (items < slen) - EXTEND(SP, items); - else - EXTEND(SP, slen); + if (!gimme_scalar) { + const U32 items = limit - 1; + if (items < slen) + EXTEND(SP, items); + else + EXTEND(SP, slen); + } if (do_utf8) { while (--limit) { /* keep track of how many bytes we skip over */ m = s; s += UTF8SKIP(s); - dstr = newSVpvn_flags(m, s-m, SVf_UTF8 | make_mortal); + if (gimme_scalar) { + iters++; + if (s-m == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + dstr = newSVpvn_flags(m, s-m, SVf_UTF8 | make_mortal); - PUSHs(dstr); + PUSHs(dstr); + } if (s >= strend) break; } } else { while (--limit) { - dstr = newSVpvn(s, 1); + if (gimme_scalar) { + iters++; + } else { + dstr = newSVpvn(s, 1); - s++; - if (make_mortal) - sv_2mortal(dstr); + if (make_mortal) + sv_2mortal(dstr); - PUSHs(dstr); + PUSHs(dstr); + } + + s++; if (s >= strend) break; @@ -5079,9 +5113,17 @@ PP(pp_split) ; if (m >= strend) break; - dstr = newSVpvn_flags(s, m-s, - (do_utf8 ? SVf_UTF8 : 0) | make_mortal); - XPUSHs(dstr); + if (gimme_scalar) { + iters++; + if (m-s == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + dstr = newSVpvn_flags(s, m-s, + (do_utf8 ? SVf_UTF8 : 0) | make_mortal); + XPUSHs(dstr); + } /* The rx->minlen is in characters but we want to step * s ahead by bytes. */ if (do_utf8) @@ -5095,9 +5137,17 @@ PP(pp_split) (m = fbm_instr((unsigned char*)s, (unsigned char*)strend, csv, multiline ? FBMrf_MULTILINE : 0)) ) { - dstr = newSVpvn_flags(s, m-s, - (do_utf8 ? SVf_UTF8 : 0) | make_mortal); - XPUSHs(dstr); + if (gimme_scalar) { + iters++; + if (m-s == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + dstr = newSVpvn_flags(s, m-s, + (do_utf8 ? SVf_UTF8 : 0) | make_mortal); + XPUSHs(dstr); + } /* The rx->minlen is in characters but we want to step * s ahead by bytes. */ if (do_utf8) @@ -5127,9 +5177,18 @@ PP(pp_split) strend = s + (strend - m); } m = RX_OFFS(rx)[0].start + orig; - dstr = newSVpvn_flags(s, m-s, - (do_utf8 ? SVf_UTF8 : 0) | make_mortal); - XPUSHs(dstr); + + if (gimme_scalar) { + iters++; + if (m-s == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + dstr = newSVpvn_flags(s, m-s, + (do_utf8 ? SVf_UTF8 : 0) | make_mortal); + XPUSHs(dstr); + } if (RX_NPARENS(rx)) { I32 i; for (i = 1; i <= (I32)RX_NPARENS(rx); i++) { @@ -5139,37 +5198,54 @@ PP(pp_split) /* japhy (07/27/01) -- the (m && s) test doesn't catch parens that didn't match -- they should be set to undef, not the empty string */ - if (m >= orig && s >= orig) { - dstr = newSVpvn_flags(s, m-s, - (do_utf8 ? SVf_UTF8 : 0) - | make_mortal); + if (gimme_scalar) { + iters++; + if (m-s == 0) + trailing_empty++; + else + trailing_empty = 0; + } else { + if (m >= orig && s >= orig) { + dstr = newSVpvn_flags(s, m-s, + (do_utf8 ? SVf_UTF8 : 0) + | make_mortal); + } + else + dstr = &PL_sv_undef; /* undef, not "" */ + XPUSHs(dstr); } - else - dstr = &PL_sv_undef; /* undef, not "" */ - XPUSHs(dstr); + } } s = RX_OFFS(rx)[0].end + orig; } } - iters = (SP - PL_stack_base) - base; + if (!gimme_scalar) { + iters = (SP - PL_stack_base) - base; + } if (iters > maxiters) DIE(aTHX_ "Split loop"); /* keep field after final delim? */ if (s < strend || (iters && origlimit)) { - const STRLEN l = strend - s; - dstr = newSVpvn_flags(s, l, (do_utf8 ? SVf_UTF8 : 0) | make_mortal); - XPUSHs(dstr); + if (!gimme_scalar) { + const STRLEN l = strend - s; + dstr = newSVpvn_flags(s, l, (do_utf8 ? SVf_UTF8 : 0) | make_mortal); + XPUSHs(dstr); + } iters++; } else if (!origlimit) { - while (iters > 0 && (!TOPs || !SvANY(TOPs) || SvCUR(TOPs) == 0)) { - if (TOPs && !make_mortal) - sv_2mortal(TOPs); - iters--; - *SP-- = &PL_sv_undef; + if (gimme_scalar) { + iters -= trailing_empty; + } else { + while (iters > 0 && (!TOPs || !SvANY(TOPs) || SvCUR(TOPs) == 0)) { + if (TOPs && !make_mortal) + sv_2mortal(TOPs); + *SP-- = &PL_sv_undef; + iters--; + } } } diff --git a/t/op/split.t b/t/op/split.t index b3a9741..6b38b43 100644 --- a/t/op/split.t +++ b/t/op/split.t @@ -6,7 +6,7 @@ BEGIN { require './test.pl'; } -plan tests => 136; +plan tests => 250; $FS = ':'; @@ -17,37 +17,61 @@ $_ = 'a:b:c'; is(join(';',$a,$b,$c), 'a;b;c'); @ary = split(/:b:/); +$cnt = split(/:b:/); is(join("$_",@ary), 'aa:b:cc'); +is($cnt, scalar(@ary)); $_ = "abc\n"; my @xyz = (@ary = split(//)); +$cnt = split(//); is(join(".",@ary), "a.b.c.\n"); +is($cnt, scalar(@ary)); $_ = "a:b:c::::"; @ary = split(/:/); +$cnt = split(/:/); is(join(".",@ary), "a.b.c"); +is($cnt, scalar(@ary)); $_ = join(':',split(' '," a b\tc \t d ")); is($_, 'a:b:c:d'); +@ary = split(' '," a b\tc \t d "); +$cnt = split(' '," a b\tc \t d "); +is($cnt, scalar(@ary)); $_ = join(':',split(/ */,"foo bar bie\tdoll")); is($_ , "f:o:o:b:a:r:b:i:e:\t:d:o:l:l"); +@ary = split(/ */,"foo bar bie\tdoll"); +$cnt = split(/ */,"foo bar bie\tdoll"); +is($cnt, scalar(@ary)); $_ = join(':', 'foo', split(/ /,'a b c'), 'bar'); is($_, "foo:a:b::c:bar"); +@ary = split(/ /,'a b c'); +$cnt = split(/ /,'a b c'); +is($cnt, scalar(@ary)); # Can we say how many fields to split to? $_ = join(':', split(' ','1 2 3 4 5 6', 3)); is($_, '1:2:3 4 5 6'); +@ary = split(' ','1 2 3 4 5 6', 3); +$cnt = split(' ','1 2 3 4 5 6', 3); +is($cnt, scalar(@ary)); # Can we do it as a variable? $x = 4; $_ = join(':', split(' ','1 2 3 4 5 6', $x)); is($_, '1:2:3:4 5 6'); +@ary = split(' ','1 2 3 4 5 6', $x); +$cnt = split(' ','1 2 3 4 5 6', $x); +is($cnt, scalar(@ary)); # Does the 999 suppress null field chopping? $_ = join(':', split(/:/,'1:2:3:4:5:6:::', 999)); is($_ , '1:2:3:4:5:6:::'); +@ary = split(/:/,'1:2:3:4:5:6:::', 999); +$cnt = split(/:/,'1:2:3:4:5:6:::', 999); +is($cnt, scalar(@ary)); # Does assignment to a list imply split to one more field than that? $foo = runperl( switches => ['-Dt'], stderr => 1, prog => '($a,$b)=split;' ); @@ -61,10 +85,16 @@ is($_, '1:2 3 4 5 6'); # do subpatterns generate additional fields (without trailing nulls)? $_ = join '|', split(/,|(-)/, "1-10,20,,,"); is($_, "1|-|10||20"); +@ary = split(/,|(-)/, "1-10,20,,,"); +$cnt = split(/,|(-)/, "1-10,20,,,"); +is($cnt, scalar(@ary)); # do subpatterns generate additional fields (with a limit)? $_ = join '|', split(/,|(-)/, "1-10,20,,,", 10); is($_, "1|-|10||20||||||"); +@ary = split(/,|(-)/, "1-10,20,,,", 10); +$cnt = split(/,|(-)/, "1-10,20,,,", 10); +is($cnt, scalar(@ary)); # is the 'two undefs' bug fixed? (undef, $a, undef, $b) = qw(1 2 3 4); @@ -79,40 +109,69 @@ is("$a|$b", "2|4"); # check splitting of null string $_ = join('|', split(/x/, '',-1), 'Z'); is($_, "Z"); +@ary = split(/x/, '',-1); +$cnt = split(/x/, '',-1); +is($cnt, scalar(@ary)); $_ = join('|', split(/x/, '', 1), 'Z'); is($_, "Z"); +@ary = split(/x/, '', 1); +$cnt = split(/x/, '', 1); +is($cnt, scalar(@ary)); $_ = join('|', split(/(p+)/,'',-1), 'Z'); is($_, "Z"); +@ary = split(/(p+)/,'',-1); +$cnt = split(/(p+)/,'',-1); +is($cnt, scalar(@ary)); $_ = join('|', split(/.?/, '',-1), 'Z'); is($_, "Z"); +@ary = split(/.?/, '',-1); +$cnt = split(/.?/, '',-1); +is($cnt, scalar(@ary)); # Are /^/m patterns scanned? $_ = join '|', split(/^a/m, "a b a\na d a", 20); is($_, "| b a\n| d a"); +@ary = split(/^a/m, "a b a\na d a", 20); +$cnt = split(/^a/m, "a b a\na d a", 20); +is($cnt, scalar(@ary)); # Are /$/m patterns scanned? $_ = join '|', split(/a$/m, "a b a\na d a", 20); is($_, "a b |\na d |"); +@ary = split(/a$/m, "a b a\na d a", 20); +$cnt = split(/a$/m, "a b a\na d a", 20); +is($cnt, scalar(@ary)); # Are /^/m patterns scanned? $_ = join '|', split(/^aa/m, "aa b aa\naa d aa", 20); is($_, "| b aa\n| d aa"); +@ary = split(/^aa/m, "aa b aa\naa d aa", 20); +$cnt = split(/^aa/m, "aa b aa\naa d aa", 20); +is($cnt, scalar(@ary)); # Are /$/m patterns scanned? $_ = join '|', split(/aa$/m, "aa b aa\naa d aa", 20); is($_, "aa b |\naa d |"); +@ary = split(/aa$/m, "aa b aa\naa d aa", 20); +$cnt = split(/aa$/m, "aa b aa\naa d aa", 20); +is($cnt, scalar(@ary)); # Greedyness: $_ = "a : b :c: d"; @ary = split(/\s*:\s*/); +$cnt = split(/\s*:\s*/); is(($res = join(".",@ary)), "a.b.c.d", $res); +is($cnt, scalar(@ary)); # use of match result as pattern (!) is('p:q:r:s', join ':', split('abc' =~ /b/, 'p1q1r1s')); +@ary = split('abc' =~ /b/, 'p1q1r1s'); +$cnt = split('abc' =~ /b/, 'p1q1r1s'); +is($cnt, scalar(@ary)); # /^/ treated as /^/m $_ = join ':', split /^/, "ab\ncd\nef\n"; @@ -128,18 +187,26 @@ ok(@list1 == @list2 && # zero-width assertion $_ = join ':', split /(?=\w)/, "rm b"; is($_, "r:m :b"); +@ary = split /(?=\w)/, "rm b"; +$cnt = split /(?=\w)/, "rm b"; +is($cnt, scalar(@ary)); # unicode splittage @ary = map {ord} split //, v1.20.300.4000.50000.4000.300.20.1; +$cnt = split //, v1.20.300.4000.50000.4000.300.20.1; is("@ary", "1 20 300 4000 50000 4000 300 20 1"); +is($cnt, scalar(@ary)); @ary = split(/\x{FE}/, "\x{FF}\x{FE}\x{FD}"); # bug id 20010105.016 +$cnt = split(/\x{FE}/, "\x{FF}\x{FE}\x{FD}"); # bug id 20010105.016 ok(@ary == 2 && $ary[0] eq "\xFF" && $ary[1] eq "\xFD" && $ary[0] eq "\x{FF}" && $ary[1] eq "\x{FD}"); +is($cnt, scalar(@ary)); @ary = split(/(\x{FE}\xFE)/, "\xFF\x{FF}\xFE\x{FE}\xFD\x{FD}"); # variant of 31 +$cnt = split(/(\x{FE}\xFE)/, "\xFF\x{FF}\xFE\x{FE}\xFD\x{FD}"); # variant of 31 ok(@ary == 3 && $ary[0] eq "\xFF\xFF" && $ary[0] eq "\x{FF}\xFF" && @@ -150,16 +217,21 @@ ok(@ary == 3 && $ary[2] eq "\xFD\xFD" && $ary[2] eq "\x{FD}\xFD" && $ary[2] eq "\x{FD}\x{FD}"); +is($cnt, scalar(@ary)); { my @a = map ord, split(//, join("", map chr, (1234, 123, 2345))); + my $c = split(//, join("", map chr, (1234, 123, 2345))); is("@a", "1234 123 2345"); + is($c, scalar(@a)); } { my $x = 'A'; my @a = map ord, split(/$x/, join("", map chr, (1234, ord($x), 2345))); + my $c = split(/$x/, join("", map chr, (1234, ord($x), 2345))); is("@a", "1234 2345"); + is($c, scalar(@a)); } { @@ -171,6 +243,8 @@ ok(@ary == 3 && my $sushi = "\x{b36c}\x{5a8c}\x{ff5b}\x{5079}\x{505b}"; my @charlist = split //, $sushi; + my $charnum = split //, $sushi; + is($charnum, scalar(@charlist)); my $r = ''; foreach my $ch (@charlist) { $r = $r . " " . sprintf "U+%04X", ord($ch); @@ -218,6 +292,8 @@ ok(@ary == 3 && my $a = "ABC\x{263A}"; my @b = split( //, $a ); + my $c = split( //, $a ); + is($c, scalar(@b)); is(scalar @b, 4); @@ -229,8 +305,10 @@ ok(@ary == 3 && { my @a = split(/\xFE/, "\xFF\xFE\xFD"); + my $b = split(/\xFE/, "\xFF\xFE\xFD"); ok(@a == 2 && $a[0] eq "\xFF" && $a[1] eq "\xFD"); + is($b, scalar(@a)); } { @@ -247,6 +325,8 @@ ok(@ary == 3 && { # split /(A)|B/, "1B2" should return (1, undef, 2) my @x = split /(A)|B/, "1B2"; + my $y = split /(A)|B/, "1B2"; + is($y, scalar(@x)); ok($x[0] eq '1' and (not defined $x[1]) and $x[2] eq '2'); } @@ -256,6 +336,8 @@ ok(@ary == 3 && local $SIG{__WARN__} = sub { $warn = join '', @_; chomp $warn }; my $char = "\x{10f1ff}"; my @a = split /\r?\n/, "$char\n"; + my $b = split /\r?\n/, "$char\n"; + is($b, scalar(@a)); ok(@a == 1 && $a[0] eq $char && !defined($warn)); } @@ -267,6 +349,8 @@ ok(@ary == 3 && utf8::upgrade $_ if $u; /(.+)/; my @d = split /[,]/,$1; + my $e = split /[,]/,$1; + is($e, scalar(@d)); is(join (':',@d), 'readin:database:readout', "[perl #18195]"); } } @@ -276,6 +360,8 @@ ok(@ary == 3 && $p="a,b"; utf8::upgrade $p; eval { @a=split(/[, ]+/,$p) }; + eval { $b=split(/[, ]+/,$p) }; + is($b, scalar(@a)); is ("$@-@a-", '-a b-', '#20912 - split() to array with /[]+/ and utf8'); } @@ -335,16 +421,22 @@ ok(@ary == 3 && chop $str; my @res=split(/\s+/,$str); + my $cnt=split(/\s+/,$str); ok(@res == 2 && join('-',@res) eq "A:-:B", "$msg - /\\s+/"); + is($cnt, scalar(@res), "$msg - /\\s+/ (count)"); my $s2 = "$space$space:A:$space$space:B\x{FFFD}"; chop $s2; my @r2 = split(' ',$s2); + my $c2 = split(' ',$s2); ok(@r2 == 2 && join('-', @r2) eq ":A:-:B", "$msg - ' '"); + is($c2, scalar(@r2), "$msg - ' ' (count)"); my @r3 = split(/\s+/, $s2); + my $c3 = split(/\s+/, $s2); ok(@r3 == 3 && join('-', @r3) eq "-:A:-:B", "$msg - /\\s+/ No.2"); + is($c3, scalar(@r3), "$msg - /\\s+/ No.2 (count)"); } } @@ -352,7 +444,11 @@ ok(@ary == 3 && my $src = "ABC \0 FOO \0 XYZ"; my @s = split(" \0 ", $src); my @r = split(/ \0 /, $src); + my $cs = split(" \0 ", $src); + my $cr = split(/ \0 /, $src); is(scalar(@s), 3); + is($cs, 3); + is($cr, 3); is($s[0], "ABC"); is($s[1], "FOO"); is($s[2]," XYZ");