2 # Copyright (C) 2000-2003 Stephen McCamant. All rights reserved.
3 # This program is free software; you can redistribute and/or modify it
4 # under the same terms as Perl itself.
6 # Note: we need to keep track of how many use declarations/BEGIN
7 # blocks this module uses, so we can avoid printing them when user
8 # asks for the BEGIN blocks in her program. Update the comments and
9 # the count in concise_specials if you add or delete one. The
10 # -MO=Concise counts as use #1.
13 use warnings; # uses #3 and #4, since warnings uses Carp
15 use Exporter (); # use #5
17 our $VERSION = "0.98";
18 our @ISA = qw(Exporter);
19 our @EXPORT_OK = qw( set_style set_style_standard add_callback
20 concise_subref concise_cv concise_main
21 add_style walk_output compile reset_sequence );
23 ( io => [qw( walk_output compile reset_sequence )],
24 style => [qw( add_style set_style_standard )],
25 cb => [qw( add_callback )],
26 mech => [qw( concise_subref concise_cv concise_main )], );
29 use B qw(class ppname main_start main_root main_cv cstring svref_2object
30 SVf_IOK SVf_NOK SVf_POK SVf_IVisUV SVf_FAKE OPf_KIDS OPf_SPECIAL
31 CVf_ANON PAD_FAKELEX_ANON PAD_FAKELEX_MULTI SVf_ROK);
35 ["(?(#label =>\n)?)(*( )*)#class (#addr) #name (?([#targ])?) "
36 . "#svclass~(?((#svaddr))?)~#svval~(?(label \"#coplabel\")?)\n",
37 "(*( )*)goto #class (#addr)\n",
40 ["#hyphseq2 (*( (x( ;)x))*)<#classsym> #exname#arg(?([#targarglife])?)"
41 . "~#flags(?(/#private)?)(?(:#hints)?)(x(;~->#next)x)\n"
42 , " (*( )*) goto #seq\n",
43 "(?(<#seq>)?)#exname#arg(?([#targarglife])?)"],
45 ["(x(;(*( )*))x)#noise#arg(?([#targarg])?)(x( ;\n)x)",
47 "(?(#seq)?)#noise#arg(?([#targarg])?)"],
49 ["#class (#addr)\n\top_next\t\t#nextaddr\n\top_sibling\t#sibaddr\n\t"
50 . "op_ppaddr\tPL_ppaddr[OP_#NAME]\n\top_type\t\t#typenum\n"
51 . "\top_flags\t#flagval\n\top_private\t#privval\t#hintsval\n"
52 . "(?(\top_first\t#firstaddr\n)?)(?(\top_last\t\t#lastaddr\n)?)"
53 . "(?(\top_sv\t\t#svaddr\n)?)",
56 "env" => [$ENV{B_CONCISE_FORMAT}, $ENV{B_CONCISE_GOTO_FORMAT},
57 $ENV{B_CONCISE_TREE_FORMAT}],
60 # Renderings, ie how Concise prints, is controlled by these vars
62 our $stylename; # selects current style from %style
63 my $order = "basic"; # how optree is walked & printed: basic, exec, tree
65 # rendering mechanics:
66 # these 'formats' are the line-rendering templates
67 # they're updated from %style when $stylename changes
68 my ($format, $gotofmt, $treefmt);
71 my $base = 36; # how <sequence#> is displayed
72 my $big_endian = 1; # more <sequence#> display
73 my $tree_style = 0; # tree-order details
74 my $banner = 1; # print banner before optree is traversed
75 my $do_main = 0; # force printing of main routine
76 my $show_src; # show source code
78 # another factor: can affect all styles!
79 our @callbacks; # allow external management
81 set_style_standard("concise");
87 ($format, $gotofmt, $treefmt) = @_;
88 #warn "set_style: deprecated, use set_style_standard instead\n"; # someday
89 die "expecting 3 style-format args\n" unless @_ == 3;
93 my ($newstyle,@args) = @_;
94 die "style '$newstyle' already exists, choose a new name\n"
95 if exists $style{$newstyle};
96 die "expecting 3 style-format args\n" unless @args == 3;
97 $style{$newstyle} = [@args];
98 $stylename = $newstyle; # update rendering state
101 sub set_style_standard {
102 ($stylename) = @_; # update rendering state
103 die "err: style '$stylename' unknown\n" unless exists $style{$stylename};
104 set_style(@{$style{$stylename}});
111 # output handle, used with all Concise-output printing
112 our $walkHandle; # public for your convenience
113 BEGIN { $walkHandle = \*STDOUT }
115 sub walk_output { # updates $walkHandle
117 return $walkHandle unless $handle; # allow use as accessor
119 if (ref $handle eq 'SCALAR') {
121 die "no perlio in this build, can't call walk_output (\\\$scalar)\n"
122 unless $Config::Config{useperlio};
123 # in 5.8+, open(FILEHANDLE,MODE,REFERENCE) writes to string
124 open my $tmp, '>', $handle; # but cant re-set existing STDOUT
125 $walkHandle = $tmp; # so use my $tmp as intermediate var
128 my $iotype = ref $handle;
129 die "expecting argument/object that can print\n"
130 unless $iotype eq 'GLOB' or $iotype and $handle->can('print');
131 $walkHandle = $handle;
135 my($order, $coderef, $name) = @_;
136 my $codeobj = svref_2object($coderef);
138 return concise_stashref(@_)
139 unless ref($codeobj) =~ '^B::(?:CV|FM)\z';
140 concise_cv_obj($order, $codeobj, $name);
143 sub concise_stashref {
146 foreach my $k (sort keys %$h) {
147 next unless defined $h->{$k};
149 my $coderef = *s{CODE} or next;
151 print "FUNC: ", *s, "\n";
152 my $codeobj = svref_2object($coderef);
153 next unless ref $codeobj eq 'B::CV';
154 eval { concise_cv_obj($order, $codeobj, $k) };
155 warn "err $@ on $codeobj" if $@;
159 # This should have been called concise_subref, but it was exported
160 # under this name in versions before 0.56
161 *concise_cv = \&concise_subref;
164 my ($order, $cv, $name) = @_;
165 # name is either a string, or a CODE ref (copy of $cv arg??)
169 if (ref($cv->XSUBANY) =~ /B::(\w+)/) {
170 print $walkHandle "$name is a constant sub, optimized to a $1\n";
174 print $walkHandle "$name is XS code\n";
177 if (class($cv->START) eq "NULL") {
179 if (ref $name eq 'CODE') {
180 print $walkHandle "coderef $name has no START\n";
182 elsif (exists &$name) {
183 print $walkHandle "$name exists in stash, but has no START\n";
186 print $walkHandle "$name not in symbol table\n";
190 sequence($cv->START);
191 if ($order eq "exec") {
192 walk_exec($cv->START);
194 elsif ($order eq "basic") {
195 # walk_topdown($cv->ROOT, sub { $_[0]->concise($_[1]) }, 0);
196 my $root = $cv->ROOT;
197 unless (ref $root eq 'B::NULL') {
198 walk_topdown($root, sub { $_[0]->concise($_[1]) }, 0);
200 print $walkHandle "B::NULL encountered doing ROOT on $cv. avoiding disaster\n";
203 print $walkHandle tree($cv->ROOT, 0);
209 sequence(main_start);
211 if ($order eq "exec") {
212 return if class(main_start) eq "NULL";
213 walk_exec(main_start);
214 } elsif ($order eq "tree") {
215 return if class(main_root) eq "NULL";
216 print $walkHandle tree(main_root, 0);
217 } elsif ($order eq "basic") {
218 return if class(main_root) eq "NULL";
219 walk_topdown(main_root,
220 sub { $_[0]->concise($_[1]) }, 0);
224 sub concise_specials {
225 my($name, $order, @cv_s) = @_;
227 if ($name eq "BEGIN") {
228 splice(@cv_s, 0, 8); # skip 7 BEGIN blocks in this file. NOW 8 ??
229 } elsif ($name eq "CHECK") {
230 pop @cv_s; # skip the CHECK block that calls us
233 print $walkHandle "$name $i:\n";
235 concise_cv_obj($order, $cv, $name);
239 my $start_sym = "\e(0"; # "\cN" sometimes also works
240 my $end_sym = "\e(B"; # "\cO" respectively
242 my @tree_decorations =
243 ([" ", "--", "+-", "|-", "| ", "`-", "-", 1],
244 [" ", "-", "+", "+", "|", "`", "", 0],
245 [" ", map("$start_sym$_$end_sym", "qq", "wq", "tq", "x ", "mq", "q"), 1],
246 [" ", map("$start_sym$_$end_sym", "q", "w", "t", "x", "m"), "", 0],
249 my @render_packs; # collect -stash=<packages>
252 # set rendering state from options and args
255 @options = grep(/^-/, @_);
256 @args = grep(!/^-/, @_);
258 for my $o (@options) {
260 if ($o eq "-basic") {
262 } elsif ($o eq "-exec") {
264 } elsif ($o eq "-tree") {
268 elsif ($o eq "-compact") {
270 } elsif ($o eq "-loose") {
272 } elsif ($o eq "-vt") {
274 } elsif ($o eq "-ascii") {
278 elsif ($o =~ /^-base(\d+)$/) {
280 } elsif ($o eq "-bigendian") {
282 } elsif ($o eq "-littleendian") {
285 # miscellaneous, presentation
286 elsif ($o eq "-nobanner") {
288 } elsif ($o eq "-banner") {
291 elsif ($o eq "-main") {
293 } elsif ($o eq "-nomain") {
295 } elsif ($o eq "-src") {
298 elsif ($o =~ /^-stash=(.*)/) {
301 if (! %{$pkg.'::'}) {
305 if (!$Config::Config{usedl}
306 && keys %{$pkg.'::'} == 1
307 && $pkg->can('bootstrap')) {
308 # It is something that we're statically linked to, but hasn't
313 push @render_packs, $pkg;
316 elsif (exists $style{substr($o, 1)}) {
317 $stylename = substr($o, 1);
318 set_style_standard($stylename);
320 warn "Option $o unrecognized";
327 my (@args) = compileOpts(@_);
329 my @newargs = compileOpts(@_); # accept new rendering options
330 warn "disregarding non-options: @newargs\n" if @newargs;
332 for my $objname (@args) {
333 next unless $objname; # skip null args to avoid noisy responses
335 if ($objname eq "BEGIN") {
336 concise_specials("BEGIN", $order,
337 B::begin_av->isa("B::AV") ?
338 B::begin_av->ARRAY : ());
339 } elsif ($objname eq "INIT") {
340 concise_specials("INIT", $order,
341 B::init_av->isa("B::AV") ?
342 B::init_av->ARRAY : ());
343 } elsif ($objname eq "CHECK") {
344 concise_specials("CHECK", $order,
345 B::check_av->isa("B::AV") ?
346 B::check_av->ARRAY : ());
347 } elsif ($objname eq "UNITCHECK") {
348 concise_specials("UNITCHECK", $order,
349 B::unitcheck_av->isa("B::AV") ?
350 B::unitcheck_av->ARRAY : ());
351 } elsif ($objname eq "END") {
352 concise_specials("END", $order,
353 B::end_av->isa("B::AV") ?
354 B::end_av->ARRAY : ());
357 # convert function names to subrefs
359 print $walkHandle "B::Concise::compile($objname)\n"
361 concise_subref($order, ($objname)x2);
364 $objname = "main::" . $objname unless $objname =~ /::/;
366 my $glob = \*$objname;
367 unless (*$glob{CODE} || *$glob{FORMAT}) {
368 print $walkHandle "$objname:\n" if $banner;
369 print $walkHandle "err: unknown function ($objname)\n";
372 if (my $objref = *$glob{CODE}) {
373 print $walkHandle "$objname:\n" if $banner;
374 concise_subref($order, $objref, $objname);
376 if (my $objref = *$glob{FORMAT}) {
377 print $walkHandle "$objname (FORMAT):\n"
379 concise_subref($order, $objref, $objname);
384 for my $pkg (@render_packs) {
386 concise_stashref($order, \%{$pkg.'::'});
389 if (!@args or $do_main or @render_packs) {
390 print $walkHandle "main program:\n" if $do_main;
391 concise_main($order);
393 return @args; # something
398 my $lastnext; # remembers op-chain, used to insert gotos
400 my %opclass = ('OP' => "0", 'UNOP' => "1", 'BINOP' => "2", 'LOGOP' => "|",
401 'LISTOP' => "@", 'PMOP' => "/", 'SVOP' => "\$", 'GVOP' => "*",
402 'PVOP' => '"', 'LOOP' => "{", 'COP' => ";", 'PADOP' => "#");
404 no warnings 'qw'; # "Possible attempt to put comments..."; use #7
406 qw'# () sc ( @? 1 $* gv *{ m$ m@ m% m? p/ *$ $ $# & a& pt \\ s\\ rf bl
407 ` *? <> ?? ?/ r/ c/ // qr s/ /c y/ = @= C sC Cp sp df un BM po +1 +I
408 -1 -I 1+ I+ 1- I- ** * i* / i/ %$ i% x + i+ - i- . " << >> < i<
409 > i> <= i, >= i. == i= != i! <? i? s< s> s, s. s= s! s? b& b^ b| -0 -i
410 ! ~ a2 si cs rd sr e^ lg sq in %x %o ab le ss ve ix ri sf FL od ch cy
411 uf lf uc lc qm @ [f [ @[ eh vl ky dl ex % ${ @{ uk pk st jn ) )[ a@
412 a% sl +] -] [- [+ so rv GS GW MS MW .. f. .f && || ^^ ?: &= |= -> s{ s}
413 v} ca wa di rs ;; ; ;d }{ { } {} f{ it {l l} rt }l }n }r dm }g }e ^o
414 ^c ^| ^# um bm t~ u~ ~d DB db ^s se ^g ^r {w }w pf pr ^O ^K ^R ^W ^d ^v
415 ^e ^t ^k t. fc ic fl .s .p .b .c .l .a .h g1 s1 g2 s2 ?. l? -R -W -X -r
416 -w -x -e -o -O -z -s -M -A -C -S -c -b -f -d -p -l -u -g -k -t -T -B cd
417 co cr u. cm ut r. l@ s@ r@ mD uD oD rD tD sD wD cD f$ w$ p$ sh e$ k$ g3
418 g4 s4 g5 s5 T@ C@ L@ G@ A@ S@ Hg Hc Hr Hw Mg Mc Ms Mr Sg Sc So rq do {e
419 e} {t t} g6 G6 6e g7 G7 7e g8 G8 8e g9 G9 9e 6s 7s 8s 9s 6E 7E 8E 9E Pn
420 Pu GP SP EP Gn Gg GG SG EG g0 c$ lk t$ ;s n> // /= CO';
422 my $chars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
424 sub op_flags { # common flags (see BASOP.op_flags in op.h)
427 push @v, "v" if ($x & 3) == 1;
428 push @v, "s" if ($x & 3) == 2;
429 push @v, "l" if ($x & 3) == 3;
430 push @v, "K" if $x & 4;
431 push @v, "P" if $x & 8;
432 push @v, "R" if $x & 16;
433 push @v, "M" if $x & 32;
434 push @v, "S" if $x & 64;
435 push @v, "*" if $x & 128;
441 return "-" . base_n(-$x) if $x < 0;
443 do { $str .= substr($chars, $x % $base, 1) } while $x = int($x / $base);
444 $str = reverse $str if $big_endian;
460 return "-" if not exists $sequence_num{$$op};
461 return base_n($sequence_num{$$op});
465 my($op, $sub, $level) = @_;
467 if ($op->flags & OPf_KIDS) {
468 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
469 walk_topdown($kid, $sub, $level + 1);
472 if (class($op) eq "PMOP") {
473 my $maybe_root = $op->pmreplroot;
474 if (ref($maybe_root) and $maybe_root->isa("B::OP")) {
475 # It really is the root of the replacement, not something
476 # else stored here for lack of space elsewhere
477 walk_topdown($maybe_root, $sub, $level + 1);
483 my($ar, $level) = @_;
485 if (ref($l) eq "ARRAY") {
486 walklines($l, $level + 1);
494 my($top, $level) = @_;
497 my @todo = ([$top, \@lines]);
498 while (@todo and my($op, $targ) = @{shift @todo}) {
499 for (; $$op; $op = $op->next) {
500 last if $opsseen{$$op}++;
502 my $name = $op->name;
503 if (class($op) eq "LOGOP") {
506 push @todo, [$op->other, $ar];
507 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
510 push @todo, [$op->pmreplstart, $ar];
511 } elsif ($name =~ /^enter(loop|iter)$/) {
512 $labels{${$op->nextop}} = "NEXT";
513 $labels{${$op->lastop}} = "LAST";
514 $labels{${$op->redoop}} = "REDO";
518 walklines(\@lines, 0);
521 # The structure of this routine is purposely modeled after op.c's peep()
525 return if class($op) eq "NULL" or exists $sequence_num{$$op};
526 for (; $$op; $op = $op->next) {
527 last if exists $sequence_num{$$op};
528 my $name = $op->name;
529 if ($name =~ /^(null|scalar|lineseq|scope)$/) {
530 next if $oldop and $ {$op->next};
532 $sequence_num{$$op} = $seq_max++;
533 if (class($op) eq "LOGOP") {
534 my $other = $op->other;
535 $other = $other->next while $other->name eq "null";
537 } elsif (class($op) eq "LOOP") {
538 my $redoop = $op->redoop;
539 $redoop = $redoop->next while $redoop->name eq "null";
541 my $nextop = $op->nextop;
542 $nextop = $nextop->next while $nextop->name eq "null";
544 my $lastop = $op->lastop;
545 $lastop = $lastop->next while $lastop->name eq "null";
547 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
548 my $replstart = $op->pmreplstart;
549 $replstart = $replstart->next while $replstart->name eq "null";
550 sequence($replstart);
557 sub fmt_line { # generate text-line for op.
558 my($hr, $op, $text, $level) = @_;
560 $_->($hr, $op, \$text, \$level, $stylename) for @callbacks;
562 return '' if $hr->{SKIP}; # suppress line if a callback said so
563 return '' if $hr->{goto} and $hr->{goto} eq '-'; # no goto nowhere
565 # spec: (?(text1#varText2)?)
566 $text =~ s/\(\?\(([^\#]*?)\#(\w+)([^\#]*?)\)\?\)/
567 $hr->{$2} ? $1.$hr->{$2}.$3 : ""/eg;
569 # spec: (x(exec_text;basic_text)x)
570 $text =~ s/\(x\((.*?);(.*?)\)x\)/$order eq "exec" ? $1 : $2/egs;
573 $text =~ s/\(\*\(([^;]*?)\)\*\)/$1 x $level/egs;
575 # spec: (*(text1;text2)*)
576 $text =~ s/\(\*\((.*?);(.*?)\)\*\)/$1 x ($level - 1) . $2 x ($level>0)/egs;
578 # convert #Var to tag=>val form: Var\t#var
579 $text =~ s/\#([A-Z][a-z]+)(\d+)?/\t\u$1\t\L#$1$2/gs;
582 $text =~ s/\#([a-zA-Z]+)(\d+)/sprintf("%-$2s", $hr->{$1})/eg;
584 $text =~ s/\#([a-zA-Z]+)/$hr->{$1}/eg; # populate #var's
585 $text =~ s/[ \t]*~+[ \t]*/ /g; # squeeze tildes
587 $text = "# $hr->{src}\n$text" if $show_src and $hr->{src};
590 return "$text\n" if $text ne "" and $order ne "tree";
591 return $text; # suppress empty lines
594 our %priv; # used to display each opcode's BASEOP.op_private values
596 $priv{$_}{128} = "LVINTRO"
597 for qw(pos substr vec threadsv gvsv rv2sv rv2hv rv2gv rv2av rv2arylen
598 aelem helem aslice hslice padsv padav padhv enteriter entersub
600 $priv{$_}{64} = "REFC" for qw(leave leavesub leavesublv leavewrite);
601 @{$priv{aassign}}{32,64} = qw(STATE COMMON);
602 @{$priv{sassign}}{32,64,128} = qw(STATE BKWARD CV2GV);
603 $priv{$_}{64} = "RTIME" for qw(match subst substcont qr);
604 @{$priv{$_}}{1,2,4,8,16,64} = qw(<UTF >UTF IDENT SQUASH DEL COMPL GROWS)
605 for qw(trans transr);
606 $priv{repeat}{64} = "DOLIST";
607 $priv{leaveloop}{64} = "CONT";
608 $priv{$_}{4} = "DREFed" for qw(rv2sv rv2av rv2hv);
609 @{$priv{$_}}{32,64,96} = qw(DREFAV DREFHV DREFSV)
610 for qw(rv2gv rv2sv padsv aelem helem);
611 $priv{$_}{16} = "STATE" for qw(padav padhv padsv);
612 @{$priv{rv2gv}}{4,16} = qw(NOINIT FAKE);
613 @{$priv{entersub}}{1,4,16,32,64} = qw(INARGS TARG DBG DEREF);
614 @{$priv{rv2cv}}{1,8,128} = qw(CONST AMPER NO());
615 $priv{gv}{32} = "EARLYCV";
616 $priv{$_}{16} = "LVDEFER" for qw(aelem helem);
617 $priv{$_}{16} = "OURINTR" for qw(gvsv rv2sv rv2av rv2hv r2gv enteriter);
618 $priv{$_}{8} = "LVSUB"
619 for qw(rv2av rv2gv rv2hv padav padhv aelem helem aslice hslice
620 av2arylen keys rkeys substr pos vec);
621 @{$priv{$_}}{32,64} = qw(BOOL BOOL?) for qw(rv2hv padhv);
622 $priv{substr}{16} = "REPL1ST";
623 $priv{$_}{16} = "TARGMY"
624 for map(($_,"s$_"), qw(chop chomp)),
625 map(($_,"i_$_"), qw(postinc postdec multiply divide modulo add
627 qw(pow concat stringify left_shift right_shift bit_and bit_xor
628 bit_or complement atan2 sin cos rand exp log sqrt int hex oct
629 abs length index rindex sprintf ord chr crypt quotemeta join
630 push unshift flock chdir chown chroot unlink chmod utime rename
631 link symlink mkdir rmdir wait waitpid system exec kill getppid
632 getpgrp setpgrp getpriority setpriority time sleep);
633 $priv{$_}{4} = "REVERSED" for qw(enteriter iter);
634 @{$priv{const}}{2,4,8,16,64,128} = qw(NOVER SHORT STRICT ENTERED BARE FOLD);
635 $priv{$_}{64} = "LINENUM" for qw(flip flop);
636 $priv{list}{64} = "GUESSED";
637 $priv{delete}{64} = "SLICE";
638 $priv{exists}{64} = "SUB";
639 @{$priv{sort}}{1,2,4,8,16,32,64} = qw(NUM INT REV INPLACE DESC QSORT STABLE);
640 $priv{reverse}{8} = "INPLACE";
641 $priv{threadsv}{64} = "SVREFd";
642 @{$priv{$_}}{16,32,64,128} = qw(INBIN INCR OUTBIN OUTCR)
643 for qw(open backtick);
644 $priv{exit}{128} = "VMS";
645 $priv{$_}{2} = "FTACCESS"
646 for qw(ftrread ftrwrite ftrexec fteread ftewrite fteexec);
647 @{$priv{entereval}}{2,4,8,16} = qw(HAS_HH UNI BYTES COPHH);
648 @{$priv{$_}}{4,8,16} = qw(FTSTACKED FTSTACKING FTAFTERt)
649 for qw(ftrread ftrwrite ftrexec fteread ftewrite fteexec ftis fteowned
650 ftrowned ftzero ftsize ftmtime ftatime ftctime ftsock ftchr
651 ftblk ftfile ftdir ftpipe ftlink ftsuid ftsgid ftsvtx fttty
653 $priv{$_}{2} = "GREPLEX"
654 for qw(mapwhile mapstart grepwhile grepstart);
655 $priv{$_}{128} = "+1" for qw(caller wantarray runcv);
656 @{$priv{coreargs}}{1,2,64,128} = qw(DREF1 DREF2 $MOD MARK);
657 $priv{$_}{128} = "UTF" for qw(last redo next goto dump);
658 $priv{split}{128} = "IMPLIM";
660 our %hints; # used to display each COP's op_hints values
662 # strict refs, subs, vars
663 @hints{2,512,1024,32,64,128} = ('$', '&', '*', 'x$', 'x&', 'x*');
664 # integers, locale, bytes
665 @hints{1,4,8,16} = ('i', 'l', 'b');
666 # block scope, localise %^H, $^OPEN (in), $^OPEN (out)
667 @hints{256,131072,262144,524288} = ('{','%','<','>');
668 # overload new integer, float, binary, string, re
669 @hints{4096,8192,16384,32768,65536} = ('I', 'F', 'B', 'S', 'R');
671 @hints{1048576,2097152} = ('T', 'E');
672 # filetest access, UTF-8
673 @hints{4194304,8388608} = ('X', 'U');
678 for my $flag (sort {$b <=> $a} keys %$hash) {
679 if ($hash->{$flag} and $x & $flag and $x >= $flag) {
681 push @s, $hash->{$flag};
685 return join(",", @s);
690 _flags($priv{$name}, $x);
699 my($sv, $hr, $preferpv) = @_;
700 $hr->{svclass} = class($sv);
701 $hr->{svclass} = "UV"
702 if $hr->{svclass} eq "IV" and $sv->FLAGS & SVf_IVisUV;
703 Carp::cluck("bad concise_sv: $sv") unless $sv and $$sv;
704 $hr->{svaddr} = sprintf("%#x", $$sv);
705 if ($hr->{svclass} eq "GV" && $sv->isGV_with_GP()) {
707 my $stash = $gv->STASH;
708 if (class($stash) eq "SPECIAL") {
712 $stash = $stash->NAME;
714 if ($stash eq "main") {
717 $stash = $stash . "::";
719 $hr->{svval} = "*$stash" . $gv->SAFENAME;
720 return "*$stash" . $gv->SAFENAME;
723 while (class($sv) eq "IV" && $sv->FLAGS & SVf_ROK) {
724 $hr->{svval} .= "\\";
728 while (class($sv) eq "RV") {
729 $hr->{svval} .= "\\";
733 if (class($sv) eq "SPECIAL") {
734 $hr->{svval} .= ["Null", "sv_undef", "sv_yes", "sv_no"]->[$$sv];
736 && ($sv->FLAGS & SVf_POK || class($sv) eq "REGEXP")) {
737 $hr->{svval} .= cstring($sv->PV);
738 } elsif ($sv->FLAGS & SVf_NOK) {
739 $hr->{svval} .= $sv->NV;
740 } elsif ($sv->FLAGS & SVf_IOK) {
741 $hr->{svval} .= $sv->int_value;
742 } elsif ($sv->FLAGS & SVf_POK || class($sv) eq "REGEXP") {
743 $hr->{svval} .= cstring($sv->PV);
744 } elsif (class($sv) eq "HV") {
745 $hr->{svval} .= 'HASH';
748 $hr->{svval} = 'undef' unless defined $hr->{svval};
749 my $out = $hr->{svclass};
750 return $out .= " $hr->{svval}" ;
758 if ($fullnm eq '-e') {
759 $srclines{$fullnm} = [ $fullnm, "-src not supported for -e" ];
762 open (my $fh, '<', $fullnm)
763 or warn "# $fullnm: $!, (chdirs not supported by this feature yet)\n"
767 unshift @l, $fullnm; # like @{_<$fullnm} in debug, array starts at 1
768 $srclines{$fullnm} = \@l;
772 my ($op, $level, $format) = @_;
774 $h{exname} = $h{name} = $op->name;
775 $h{NAME} = uc $h{name};
776 $h{class} = class($op);
777 $h{extarg} = $h{targ} = $op->targ;
778 $h{extarg} = "" unless $h{extarg};
779 if ($h{name} eq "null" and $h{targ}) {
780 # targ holds the old type
781 $h{exname} = "ex-" . substr(ppname($h{targ}), 3);
783 } elsif ($op->name =~ /^leave(sub(lv)?|write)?$/) {
784 # targ potentially holds a reference count
785 if ($op->private & 64) {
786 my $refs = "ref" . ($h{targ} != 1 ? "s" : "");
787 $h{targarglife} = $h{targarg} = "$h{targ} $refs";
790 my $count = $h{name} eq 'padrange' ? ($op->private & 127) : 1;
791 my (@targarg, @targarglife);
792 for my $i (0..$count-1) {
793 my ($targarg, $targarglife);
794 my $padname = (($curcv->PADLIST->ARRAY)[0]->ARRAY)[$h{targ}+$i];
795 if (defined $padname and class($padname) ne "SPECIAL") {
796 $targarg = $padname->PVX;
797 if ($padname->FLAGS & SVf_FAKE) {
798 # These changes relate to the jumbo closure fix.
799 # See changes 19939 and 20005
802 if $padname->PARENT_FAKELEX_FLAGS & PAD_FAKELEX_ANON;
804 if $padname->PARENT_FAKELEX_FLAGS & PAD_FAKELEX_MULTI;
805 $fake .= ':' . $padname->PARENT_PAD_INDEX
806 if $curcv->CvFLAGS & CVf_ANON;
807 $targarglife = "$targarg:FAKE:$fake";
810 my $intro = $padname->COP_SEQ_RANGE_LOW - $cop_seq_base;
811 my $finish = int($padname->COP_SEQ_RANGE_HIGH) - $cop_seq_base;
812 $finish = "end" if $finish == 999999999 - $cop_seq_base;
813 $targarglife = "$targarg:$intro,$finish";
816 $targarglife = $targarg = "t" . ($h{targ}+$i);
818 push @targarg, $targarg;
819 push @targarglife, $targarglife;
821 $h{targarg} = join '; ', @targarg;
822 $h{targarglife} = join '; ', @targarglife;
825 $h{svclass} = $h{svaddr} = $h{svval} = "";
826 if ($h{class} eq "PMOP") {
828 my $precomp = $op->precomp;
829 if (defined $precomp) {
830 $precomp = cstring($precomp); # Escape literal control sequences
831 $precomp = "/$precomp/";
835 if ($op->name eq 'subst') {
836 if (class($op->pmreplstart) ne "NULL") {
838 $extra = " replstart->" . seq($op->pmreplstart);
841 elsif ($op->name eq 'pushre') {
842 # with C<@stash_array = split(/pat/, str);>,
843 # *stash_array is stored in /pat/'s pmreplroot.
844 my $gv = $op->pmreplroot;
846 # threaded: the value is actually a pad offset for where
847 # the GV is kept (op_pmtargetoff)
849 $gv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$gv]->NAME;
853 # unthreaded: its a GV (if it exists)
854 $gv = (ref($gv) eq "B::GV") ? $gv->NAME : undef;
856 $extra = " => \@$gv" if $gv;
858 $h{arg} = "($precomp$extra)";
859 } elsif ($h{class} eq "PVOP" and $h{name} !~ '^transr?\z') {
860 $h{arg} = '("' . $op->pv . '")';
861 $h{svval} = '"' . $op->pv . '"';
862 } elsif ($h{class} eq "COP") {
863 my $label = $op->label;
864 $h{coplabel} = $label;
865 $label = $label ? "$label: " : "";
871 my($stash, $cseq) = ($op->stash->NAME, $op->cop_seq - $cop_seq_base);
872 $h{arg} = "($label$stash $cseq $loc)";
874 fill_srclines($pathnm) unless exists $srclines{$pathnm};
875 # Would love to retain Jim's use of // but this code needs to be
877 my $line = $srclines{$pathnm}[$ln];
878 $line = "-src unavailable under -e" unless defined $line;
879 $h{src} = "$ln: $line";
881 } elsif ($h{class} eq "LOOP") {
882 $h{arg} = "(next->" . seq($op->nextop) . " last->" . seq($op->lastop)
883 . " redo->" . seq($op->redoop) . ")";
884 } elsif ($h{class} eq "LOGOP") {
886 $h{arg} = "(other->" . seq($op->other) . ")";
888 elsif ($h{class} eq "SVOP" or $h{class} eq "PADOP") {
889 unless ($h{name} eq 'aelemfast' and $op->flags & OPf_SPECIAL) {
890 my $idx = ($h{class} eq "SVOP") ? $op->targ : $op->padix;
891 my $preferpv = $h{name} eq "method_named";
892 if ($h{class} eq "PADOP" or !${$op->sv}) {
893 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$idx];
894 $h{arg} = "[" . concise_sv($sv, \%h, $preferpv) . "]";
895 $h{targarglife} = $h{targarg} = "";
897 $h{arg} = "(" . concise_sv($op->sv, \%h, $preferpv) . ")";
901 $h{seq} = $h{hyphseq} = seq($op);
902 $h{seq} = "" if $h{seq} eq "-";
904 $h{label} = $labels{$$op};
905 $h{next} = $op->next;
906 $h{next} = (class($h{next}) eq "NULL") ? "(end)" : seq($h{next});
907 $h{nextaddr} = sprintf("%#x", $ {$op->next});
908 $h{sibaddr} = sprintf("%#x", $ {$op->sibling});
909 $h{firstaddr} = sprintf("%#x", $ {$op->first}) if $op->can("first");
910 $h{lastaddr} = sprintf("%#x", $ {$op->last}) if $op->can("last");
912 $h{classsym} = $opclass{$h{class}};
913 $h{flagval} = $op->flags;
914 $h{flags} = op_flags($op->flags);
915 $h{privval} = $op->private;
916 $h{private} = private_flags($h{name}, $op->private);
917 if ($op->can("hints")) {
918 $h{hintsval} = $op->hints;
919 $h{hints} = hints_flags($h{hintsval});
921 $h{hintsval} = $h{hints} = '';
923 $h{addr} = sprintf("%#x", $$op);
924 $h{typenum} = $op->type;
925 $h{noise} = $linenoise[$op->type];
927 return fmt_line(\%h, $op, $format, $level);
931 my($op, $level) = @_;
932 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
933 # insert a 'goto' line
934 my $synth = {"seq" => seq($lastnext), "class" => class($lastnext),
935 "addr" => sprintf("%#x", $$lastnext),
936 "goto" => seq($lastnext), # simplify goto '-' removal
938 print $walkHandle fmt_line($synth, $op, $gotofmt, $level+1);
940 $lastnext = $op->next;
941 print $walkHandle concise_op($op, $level, $format);
944 # B::OP::terse (see Terse.pm) now just calls this
946 my($op, $level) = @_;
948 # This isn't necessarily right, but there's no easy way to get
949 # from an OP to the right CV. This is a limitation of the
950 # ->terse() interface style, and there isn't much to do about
951 # it. In particular, we can die in concise_op if the main pad
952 # isn't long enough, or has the wrong kind of entries, compared to
953 # the pad a sub was compiled with. The fix for that would be to
954 # make a backwards compatible "terse" format that never even
955 # looked at the pad, just like the old B::Terse. I don't think
956 # that's worth the effort, though.
957 $curcv = main_cv unless $curcv;
959 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
961 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
962 "addr" => sprintf("%#x", $$lastnext)};
964 fmt_line($h, $op, $style{"terse"}[1], $level+1);
966 $lastnext = $op->next;
968 concise_op($op, $level, $style{"terse"}[0]);
974 my $style = $tree_decorations[$tree_style];
975 my($space, $single, $kids, $kid, $nokid, $last, $lead, $size) = @$style;
976 my $name = concise_op($op, $level, $treefmt);
977 if (not $op->flags & OPf_KIDS) {
981 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
982 push @lines, tree($kid, $level+1);
985 for ($i = $#lines; substr($lines[$i], 0, 1) eq " "; $i--) {
986 $lines[$i] = $space . $lines[$i];
989 $lines[$i] = $last . $lines[$i];
991 if (substr($lines[$i], 0, 1) eq " ") {
992 $lines[$i] = $nokid . $lines[$i];
994 $lines[$i] = $kid . $lines[$i];
997 $lines[$i] = $kids . $lines[$i];
999 $lines[0] = $single . $lines[0];
1001 return("$name$lead" . shift @lines,
1002 map(" " x (length($name)+$size) . $_, @lines));
1005 # *** Warning: fragile kludge ahead ***
1006 # Because the B::* modules run in the same interpreter as the code
1007 # they're compiling, their presence tends to distort the view we have of
1008 # the code we're looking at. In particular, perl gives sequence numbers
1009 # to COPs. If the program we're looking at were run on its own, this
1010 # would start at 1. Because all of B::Concise and all the modules it
1011 # uses are compiled first, though, by the time we get to the user's
1012 # program the sequence number is already pretty high, which could be
1013 # distracting if you're trying to tell OPs apart. Therefore we'd like to
1014 # subtract an offset from all the sequence numbers we display, to
1015 # restore the simpler view of the world. The trick is to know what that
1016 # offset will be, when we're still compiling B::Concise! If we
1017 # hardcoded a value, it would have to change every time B::Concise or
1018 # other modules we use do. To help a little, what we do here is compile
1019 # a little code at the end of the module, and compute the base sequence
1020 # number for the user's program as being a small offset later, so all we
1021 # have to worry about are changes in the offset.
1023 # [For 5.8.x and earlier perl is generating sequence numbers for all ops,
1024 # and using them to reference labels]
1027 # When you say "perl -MO=Concise -e '$a'", the output should look like:
1029 # 4 <@> leave[t1] vKP/REFC ->(end)
1031 #^ smallest OP sequence number should be 1
1032 # 2 <;> nextstate(main 1 -e:1) v ->3
1033 # ^ smallest COP sequence number should be 1
1034 # - <1> ex-rv2sv vK/1 ->4
1035 # 3 <$> gvsv(*a) s ->4
1037 # If the second of the marked numbers there isn't 1, it means you need
1038 # to update the corresponding magic number in the next line.
1039 # Remember, this needs to stay the last things in the module.
1041 # Why is this different for MacOS? Does it matter?
1042 my $cop_seq_mnum = $^O eq 'MacOS' ? 12 : 11;
1043 $cop_seq_base = svref_2object(eval 'sub{0;}')->START->cop_seq + $cop_seq_mnum;
1051 B::Concise - Walk Perl syntax tree, printing concise info about ops
1055 perl -MO=Concise[,OPTIONS] foo.pl
1057 use B::Concise qw(set_style add_callback);
1061 This compiler backend prints the internal OPs of a Perl program's syntax
1062 tree in one of several space-efficient text formats suitable for debugging
1063 the inner workings of perl or other compiler backends. It can print OPs in
1064 the order they appear in the OP tree, in the order they will execute, or
1065 in a text approximation to their tree structure, and the format of the
1066 information displayed is customizable. Its function is similar to that of
1067 perl's B<-Dx> debugging flag or the B<B::Terse> module, but it is more
1068 sophisticated and flexible.
1072 Here's two outputs (or 'renderings'), using the -exec and -basic
1073 (i.e. default) formatting conventions on the same code snippet.
1075 % perl -MO=Concise,-exec -e '$a = $b + 42'
1077 2 <;> nextstate(main 1 -e:1) v
1079 4 <$> const[IV 42] s
1080 * 5 <2> add[t3] sK/2
1083 8 <@> leave[1 ref] vKP/REFC
1085 In this -exec rendering, each opcode is executed in the order shown.
1086 The add opcode, marked with '*', is discussed in more detail.
1088 The 1st column is the op's sequence number, starting at 1, and is
1089 displayed in base 36 by default. Here they're purely linear; the
1090 sequences are very helpful when looking at code with loops and
1093 The symbol between angle brackets indicates the op's type, for
1094 example; <2> is a BINOP, <@> a LISTOP, and <#> is a PADOP, which is
1095 used in threaded perls. (see L</"OP class abbreviations">).
1097 The opname, as in B<'add[t1]'>, may be followed by op-specific
1098 information in parentheses or brackets (ex B<'[t1]'>).
1100 The op-flags (ex B<'sK/2'>) are described in (L</"OP flags
1103 % perl -MO=Concise -e '$a = $b + 42'
1104 8 <@> leave[1 ref] vKP/REFC ->(end)
1106 2 <;> nextstate(main 1 -e:1) v ->3
1107 7 <2> sassign vKS/2 ->8
1108 * 5 <2> add[t1] sK/2 ->6
1109 - <1> ex-rv2sv sK/1 ->4
1110 3 <$> gvsv(*b) s ->4
1111 4 <$> const(IV 42) s ->5
1112 - <1> ex-rv2sv sKRM*/1 ->7
1113 6 <$> gvsv(*a) s ->7
1115 The default rendering is top-down, so they're not in execution order.
1116 This form reflects the way the stack is used to parse and evaluate
1117 expressions; the add operates on the two terms below it in the tree.
1119 Nullops appear as C<ex-opname>, where I<opname> is an op that has been
1120 optimized away by perl. They're displayed with a sequence-number of
1121 '-', because they are not executed (they don't appear in previous
1122 example), they're printed here because they reflect the parse.
1124 The arrow points to the sequence number of the next op; they're not
1125 displayed in -exec mode, for obvious reasons.
1127 Note that because this rendering was done on a non-threaded perl, the
1128 PADOPs in the previous examples are now SVOPs, and some (but not all)
1129 of the square brackets have been replaced by round ones. This is a
1130 subtle feature to provide some visual distinction between renderings
1131 on threaded and un-threaded perls.
1136 Arguments that don't start with a hyphen are taken to be the names of
1137 subroutines or formats to render; if no
1138 such functions are specified, the main
1139 body of the program (outside any subroutines, and not including use'd
1140 or require'd files) is rendered. Passing C<BEGIN>, C<UNITCHECK>,
1141 C<CHECK>, C<INIT>, or C<END> will cause all of the corresponding
1142 special blocks to be printed. Arguments must follow options.
1144 Options affect how things are rendered (ie printed). They're presented
1145 here by their visual effect, 1st being strongest. They're grouped
1146 according to how they interrelate; within each group the options are
1147 mutually exclusive (unless otherwise stated).
1149 =head2 Options for Opcode Ordering
1151 These options control the 'vertical display' of opcodes. The display
1152 'order' is also called 'mode' elsewhere in this document.
1158 Print OPs in the order they appear in the OP tree (a preorder
1159 traversal, starting at the root). The indentation of each OP shows its
1160 level in the tree, and the '->' at the end of the line indicates the
1161 next opcode in execution order. This mode is the default, so the flag
1162 is included simply for completeness.
1166 Print OPs in the order they would normally execute (for the majority
1167 of constructs this is a postorder traversal of the tree, ending at the
1168 root). In most cases the OP that usually follows a given OP will
1169 appear directly below it; alternate paths are shown by indentation. In
1170 cases like loops when control jumps out of a linear path, a 'goto'
1175 Print OPs in a text approximation of a tree, with the root of the tree
1176 at the left and 'left-to-right' order of children transformed into
1177 'top-to-bottom'. Because this mode grows both to the right and down,
1178 it isn't suitable for large programs (unless you have a very wide
1183 =head2 Options for Line-Style
1185 These options select the line-style (or just style) used to render
1186 each opcode, and dictates what info is actually printed into each line.
1192 Use the author's favorite set of formatting conventions. This is the
1197 Use formatting conventions that emulate the output of B<B::Terse>. The
1198 basic mode is almost indistinguishable from the real B<B::Terse>, and the
1199 exec mode looks very similar, but is in a more logical order and lacks
1200 curly brackets. B<B::Terse> doesn't have a tree mode, so the tree mode
1201 is only vaguely reminiscent of B<B::Terse>.
1205 Use formatting conventions in which the name of each OP, rather than being
1206 written out in full, is represented by a one- or two-character abbreviation.
1207 This is mainly a joke.
1211 Use formatting conventions reminiscent of B<B::Debug>; these aren't
1212 very concise at all.
1216 Use formatting conventions read from the environment variables
1217 C<B_CONCISE_FORMAT>, C<B_CONCISE_GOTO_FORMAT>, and C<B_CONCISE_TREE_FORMAT>.
1221 =head2 Options for tree-specific formatting
1227 Use a tree format in which the minimum amount of space is used for the
1228 lines connecting nodes (one character in most cases). This squeezes out
1229 a few precious columns of screen real estate.
1233 Use a tree format that uses longer edges to separate OP nodes. This format
1234 tends to look better than the compact one, especially in ASCII, and is
1239 Use tree connecting characters drawn from the VT100 line-drawing set.
1240 This looks better if your terminal supports it.
1244 Draw the tree with standard ASCII characters like C<+> and C<|>. These don't
1245 look as clean as the VT100 characters, but they'll work with almost any
1246 terminal (or the horizontal scrolling mode of less(1)) and are suitable
1247 for text documentation or email. This is the default.
1251 These are pairwise exclusive, i.e. compact or loose, vt or ascii.
1253 =head2 Options controlling sequence numbering
1259 Print OP sequence numbers in base I<n>. If I<n> is greater than 10, the
1260 digit for 11 will be 'a', and so on. If I<n> is greater than 36, the digit
1261 for 37 will be 'A', and so on until 62. Values greater than 62 are not
1262 currently supported. The default is 36.
1266 Print sequence numbers with the most significant digit first. This is the
1267 usual convention for Arabic numerals, and the default.
1269 =item B<-littleendian>
1271 Print sequence numbers with the least significant digit first. This is
1272 obviously mutually exclusive with bigendian.
1276 =head2 Other options
1282 With this option, the rendering of each statement (starting with the
1283 nextstate OP) will be preceded by the 1st line of source code that
1284 generates it. For example:
1288 2 <;> nextstate(main 1 junk.pl:1) v:{
1289 3 <0> padsv[$i:1,10] vM/LVINTRO
1290 # 3: for $i (0..9) {
1291 4 <;> nextstate(main 3 junk.pl:3) v:{
1295 8 <{> enteriter(next->j last->m redo->9)[$i:1,10] lKS
1297 l <|> and(other->9) vK/1
1299 9 <;> nextstate(main 2 junk.pl:4) v
1301 b <$> const[PV "line "] s
1306 =item B<-stash="somepackage">
1308 With this, "somepackage" will be required, then the stash is
1309 inspected, and each function is rendered.
1313 The following options are pairwise exclusive.
1319 Include the main program in the output, even if subroutines were also
1320 specified. This rendering is normally suppressed when a subroutine
1321 name or reference is given.
1325 This restores the default behavior after you've changed it with '-main'
1326 (it's not normally needed). If no subroutine name/ref is given, main is
1327 rendered, regardless of this flag.
1331 Renderings usually include a banner line identifying the function name
1332 or stringified subref. This suppresses the printing of the banner.
1334 TBC: Remove the stringified coderef; while it provides a 'cookie' for
1335 each function rendered, the cookies used should be 1,2,3.. not a
1336 random hex-address. It also complicates string comparison of two
1341 restores default banner behavior.
1343 =item B<-banneris> => subref
1345 TBC: a hookpoint (and an option to set it) for a user-supplied
1346 function to produce a banner appropriate for users needs. It's not
1347 ideal, because the rendering-state variables, which are a natural
1348 candidate for use in concise.t, are unavailable to the user.
1352 =head2 Option Stickiness
1354 If you invoke Concise more than once in a program, you should know that
1355 the options are 'sticky'. This means that the options you provide in
1356 the first call will be remembered for the 2nd call, unless you
1357 re-specify or change them.
1359 =head1 ABBREVIATIONS
1361 The concise style uses symbols to convey maximum info with minimal
1362 clutter (like hex addresses). With just a little practice, you can
1363 start to see the flowers, not just the branches, in the trees.
1365 =head2 OP class abbreviations
1367 These symbols appear before the op-name, and indicate the
1368 B:: namespace that represents the ops in your Perl code.
1370 0 OP (aka BASEOP) An OP with no children
1371 1 UNOP An OP with one child
1372 2 BINOP An OP with two children
1373 | LOGOP A control branch OP
1374 @ LISTOP An OP that could have lots of children
1375 / PMOP An OP with a regular expression
1376 $ SVOP An OP with an SV
1377 " PVOP An OP with a string
1378 { LOOP An OP that holds pointers for a loop
1379 ; COP An OP that marks the start of a statement
1380 # PADOP An OP with a GV on the pad
1382 =head2 OP flags abbreviations
1384 OP flags are either public or private. The public flags alter the
1385 behavior of each opcode in consistent ways, and are represented by 0
1386 or more single characters.
1388 v OPf_WANT_VOID Want nothing (void context)
1389 s OPf_WANT_SCALAR Want single value (scalar context)
1390 l OPf_WANT_LIST Want list of any length (list context)
1392 K OPf_KIDS There is a firstborn child.
1393 P OPf_PARENS This operator was parenthesized.
1394 (Or block needs explicit scope entry.)
1395 R OPf_REF Certified reference.
1396 (Return container, not containee).
1397 M OPf_MOD Will modify (lvalue).
1398 S OPf_STACKED Some arg is arriving on the stack.
1399 * OPf_SPECIAL Do something weird for this op (see op.h)
1401 Private flags, if any are set for an opcode, are displayed after a '/'
1403 8 <@> leave[1 ref] vKP/REFC ->(end)
1404 7 <2> sassign vKS/2 ->8
1406 They're opcode specific, and occur less often than the public ones, so
1407 they're represented by short mnemonics instead of single-chars; see
1408 F<op.h> for gory details, or try this quick 2-liner:
1410 $> perl -MB::Concise -de 1
1411 DB<1> |x \%B::Concise::priv
1413 =head1 FORMATTING SPECIFICATIONS
1415 For each line-style ('concise', 'terse', 'linenoise', etc.) there are
1416 3 format-specs which control how OPs are rendered.
1418 The first is the 'default' format, which is used in both basic and exec
1419 modes to print all opcodes. The 2nd, goto-format, is used in exec
1420 mode when branches are encountered. They're not real opcodes, and are
1421 inserted to look like a closing curly brace. The tree-format is tree
1424 When a line is rendered, the correct format-spec is copied and scanned
1425 for the following items; data is substituted in, and other
1426 manipulations like basic indenting are done, for each opcode rendered.
1428 There are 3 kinds of items that may be populated; special patterns,
1429 #vars, and literal text, which is copied verbatim. (Yes, it's a set
1432 =head2 Special Patterns
1434 These items are the primitives used to perform indenting, and to
1435 select text from amongst alternatives.
1439 =item B<(x(>I<exec_text>B<;>I<basic_text>B<)x)>
1441 Generates I<exec_text> in exec mode, or I<basic_text> in basic mode.
1443 =item B<(*(>I<text>B<)*)>
1445 Generates one copy of I<text> for each indentation level.
1447 =item B<(*(>I<text1>B<;>I<text2>B<)*)>
1449 Generates one fewer copies of I<text1> than the indentation level, followed
1450 by one copy of I<text2> if the indentation level is more than 0.
1452 =item B<(?(>I<text1>B<#>I<var>I<Text2>B<)?)>
1454 If the value of I<var> is true (not empty or zero), generates the
1455 value of I<var> surrounded by I<text1> and I<Text2>, otherwise
1460 Any number of tildes and surrounding whitespace will be collapsed to
1467 These #vars represent opcode properties that you may want as part of
1468 your rendering. The '#' is intended as a private sigil; a #var's
1469 value is interpolated into the style-line, much like "read $this".
1471 These vars take 3 forms:
1477 A property named 'var' is assumed to exist for the opcodes, and is
1478 interpolated into the rendering.
1480 =item B<#>I<var>I<N>
1482 Generates the value of I<var>, left justified to fill I<N> spaces.
1483 Note that this means while you can have properties 'foo' and 'foo2',
1484 you cannot render 'foo2', but you could with 'foo2a'. You would be
1485 wise not to rely on this behavior going forward ;-)
1489 This ucfirst form of #var generates a tag-value form of itself for
1490 display; it converts '#Var' into a 'Var => #var' style, which is then
1491 handled as described above. (Imp-note: #Vars cannot be used for
1492 conditional-fills, because the => #var transform is done after the check
1497 The following variables are 'defined' by B::Concise; when they are
1498 used in a style, their respective values are plugged into the
1499 rendering of each opcode.
1501 Only some of these are used by the standard styles, the others are
1502 provided for you to delve into optree mechanics, should you wish to
1503 add a new style (see L</add_style> below) that uses them. You can
1504 also add new ones using L</add_callback>.
1510 The address of the OP, in hexadecimal.
1514 The OP-specific information of the OP (such as the SV for an SVOP, the
1515 non-local exit pointers for a LOOP, etc.) enclosed in parentheses.
1519 The B-determined class of the OP, in all caps.
1523 A single symbol abbreviating the class of the OP.
1527 The label of the statement or block the OP is the start of, if any.
1531 The name of the OP, or 'ex-foo' if the OP is a null that used to be a foo.
1535 The target of the OP, or nothing for a nulled OP.
1539 The address of the OP's first child, in hexadecimal.
1543 The OP's flags, abbreviated as a series of symbols.
1547 The numeric value of the OP's flags.
1551 The COP's hint flags, rendered with abbreviated names if possible. An empty
1552 string if this is not a COP. Here are the symbols used:
1557 x$ explicit use/no strict refs
1558 x& explicit use/no strict subs
1559 x* explicit use/no strict vars
1579 The numeric value of the COP's hint flags, or an empty string if this is not
1584 The sequence number of the OP, or a hyphen if it doesn't have one.
1588 'NEXT', 'LAST', or 'REDO' if the OP is a target of one of those in exec
1589 mode, or empty otherwise.
1593 The address of the OP's last child, in hexadecimal.
1601 The OP's name, in all caps.
1605 The sequence number of the OP's next OP.
1609 The address of the OP's next OP, in hexadecimal.
1613 A one- or two-character abbreviation for the OP's name.
1617 The OP's private flags, rendered with abbreviated names if possible.
1621 The numeric value of the OP's private flags.
1625 The sequence number of the OP. Note that this is a sequence number
1626 generated by B::Concise.
1630 5.8.x and earlier only. 5.9 and later do not provide this.
1632 The real sequence number of the OP, as a regular number and not adjusted
1633 to be relative to the start of the real program. (This will generally be
1634 a fairly large number because all of B<B::Concise> is compiled before
1639 Whether or not the op has been optimized by the peephole optimizer.
1641 Only available in 5.9 and later.
1645 The address of the OP's next youngest sibling, in hexadecimal.
1649 The address of the OP's SV, if it has an SV, in hexadecimal.
1653 The class of the OP's SV, if it has one, in all caps (e.g., 'IV').
1657 The value of the OP's SV, if it has one, in a short human-readable format.
1661 The numeric value of the OP's targ.
1665 The name of the variable the OP's targ refers to, if any, otherwise the
1666 letter t followed by the OP's targ in decimal.
1668 =item B<#targarglife>
1670 Same as B<#targarg>, but followed by the COP sequence numbers that delimit
1671 the variable's lifetime (or 'end' for a variable in an open scope) for a
1676 The numeric value of the OP's type, in decimal.
1680 =head1 One-Liner Command tips
1684 =item perl -MO=Concise,bar foo.pl
1686 Renders only bar() from foo.pl. To see main, drop the ',bar'. To see
1689 =item perl -MDigest::MD5=md5 -MO=Concise,md5 -e1
1691 Identifies md5 as an XS function. The export is needed so that BC can
1694 =item perl -MPOSIX -MO=Concise,_POSIX_ARG_MAX -e1
1696 Identifies _POSIX_ARG_MAX as a constant sub, optimized to an IV.
1697 Although POSIX isn't entirely consistent across platforms, this is
1698 likely to be present in virtually all of them.
1700 =item perl -MPOSIX -MO=Concise,a -e 'print _POSIX_SAVED_IDS'
1702 This renders a print statement, which includes a call to the function.
1703 It's identical to rendering a file with a use call and that single
1704 statement, except for the filename which appears in the nextstate ops.
1706 =item perl -MPOSIX -MO=Concise,a -e 'sub a{_POSIX_SAVED_IDS}'
1708 This is B<very> similar to previous, only the first two ops differ. This
1709 subroutine rendering is more representative, insofar as a single main
1710 program will have many subs.
1712 =item perl -MB::Concise -e 'B::Concise::compile("-exec","-src", \%B::Concise::)->()'
1714 This renders all functions in the B::Concise package with the source
1715 lines. It eschews the O framework so that the stashref can be passed
1716 directly to B::Concise::compile(). See -stash option for a more
1717 convenient way to render a package.
1721 =head1 Using B::Concise outside of the O framework
1723 The common (and original) usage of B::Concise was for command-line
1724 renderings of simple code, as given in EXAMPLE. But you can also use
1725 B<B::Concise> from your code, and call compile() directly, and
1726 repeatedly. By doing so, you can avoid the compile-time only
1727 operation of O.pm, and even use the debugger to step through
1728 B::Concise::compile() itself.
1730 Once you're doing this, you may alter Concise output by adding new
1731 rendering styles, and by optionally adding callback routines which
1732 populate new variables, if such were referenced from those (just
1735 =head2 Example: Altering Concise Renderings
1737 use B::Concise qw(set_style add_callback);
1738 add_style($yourStyleName => $defaultfmt, $gotofmt, $treefmt);
1741 my ($h, $op, $format, $level, $stylename) = @_;
1742 $h->{variable} = some_func($op);
1744 $walker = B::Concise::compile(@options,@subnames,@subrefs);
1749 B<set_style> accepts 3 arguments, and updates the three format-specs
1750 comprising a line-style (basic-exec, goto, tree). It has one minor
1751 drawback though; it doesn't register the style under a new name. This
1752 can become an issue if you render more than once and switch styles.
1753 Thus you may prefer to use add_style() and/or set_style_standard()
1756 =head2 set_style_standard($name)
1758 This restores one of the standard line-styles: C<terse>, C<concise>,
1759 C<linenoise>, C<debug>, C<env>, into effect. It also accepts style
1760 names previously defined with add_style().
1764 This subroutine accepts a new style name and three style arguments as
1765 above, and creates, registers, and selects the newly named style. It is
1766 an error to re-add a style; call set_style_standard() to switch between
1769 =head2 add_callback ()
1771 If your newly minted styles refer to any new #variables, you'll need
1772 to define a callback subroutine that will populate (or modify) those
1773 variables. They are then available for use in the style you've
1776 The callbacks are called for each opcode visited by Concise, in the
1777 same order as they are added. Each subroutine is passed five
1780 1. A hashref, containing the variable names and values which are
1781 populated into the report-line for the op
1782 2. the op, as a B<B::OP> object
1783 3. a reference to the format string
1784 4. the formatting (indent) level
1785 5. the selected stylename
1787 To define your own variables, simply add them to the hash, or change
1788 existing values if you need to. The level and format are passed in as
1789 references to scalars, but it is unlikely that they will need to be
1790 changed or even used.
1792 =head2 Running B::Concise::compile()
1794 B<compile> accepts options as described above in L</OPTIONS>, and
1795 arguments, which are either coderefs, or subroutine names.
1797 It constructs and returns a $treewalker coderef, which when invoked,
1798 traverses, or walks, and renders the optrees of the given arguments to
1799 STDOUT. You can reuse this, and can change the rendering style used
1800 each time; thereafter the coderef renders in the new style.
1802 B<walk_output> lets you change the print destination from STDOUT to
1803 another open filehandle, or into a string passed as a ref (unless
1804 you've built perl with -Uuseperlio).
1806 my $walker = B::Concise::compile('-terse','aFuncName', \&aSubRef); # 1
1807 walk_output(\my $buf);
1808 $walker->(); # 1 renders -terse
1809 set_style_standard('concise'); # 2
1810 $walker->(); # 2 renders -concise
1811 $walker->(@new); # 3 renders whatever
1812 print "3 different renderings: terse, concise, and @new: $buf\n";
1814 When $walker is called, it traverses the subroutines supplied when it
1815 was created, and renders them using the current style. You can change
1816 the style afterwards in several different ways:
1818 1. call C<compile>, altering style or mode/order
1819 2. call C<set_style_standard>
1820 3. call $walker, passing @new options
1822 Passing new options to the $walker is the easiest way to change
1823 amongst any pre-defined styles (the ones you add are automatically
1824 recognized as options), and is the only way to alter rendering order
1825 without calling compile again. Note however that rendering state is
1826 still shared amongst multiple $walker objects, so they must still be
1827 used in a coordinated manner.
1829 =head2 B::Concise::reset_sequence()
1831 This function (not exported) lets you reset the sequence numbers (note
1832 that they're numbered arbitrarily, their goal being to be human
1833 readable). Its purpose is mostly to support testing, i.e. to compare
1834 the concise output from two identical anonymous subroutines (but
1835 different instances). Without the reset, B::Concise, seeing that
1836 they're separate optrees, generates different sequence numbers in
1841 Errors in rendering (non-existent function-name, non-existent coderef)
1842 are written to the STDOUT, or wherever you've set it via
1845 Errors using the various *style* calls, and bad args to walk_output(),
1846 result in die(). Use an eval if you wish to catch these errors and
1847 continue processing.
1851 Stephen McCamant, E<lt>smcc@CSUA.Berkeley.EDUE<gt>.