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 $VERSION = "0.76";
18 our @ISA = qw(Exporter);
19 our @EXPORT_OK = qw( set_style set_style_standard add_callback
20 concise_subref concise_cv concise_main
21 add_style walk_output compile reset_sequence );
23 ( io => [qw( walk_output compile reset_sequence )],
24 style => [qw( add_style set_style_standard )],
25 cb => [qw( add_callback )],
26 mech => [qw( concise_subref concise_cv concise_main )], );
29 use B qw(class ppname main_start main_root main_cv cstring svref_2object
30 SVf_IOK SVf_NOK SVf_POK SVf_IVisUV SVf_FAKE OPf_KIDS OPf_SPECIAL
31 CVf_ANON PAD_FAKELEX_ANON PAD_FAKELEX_MULTI SVf_ROK);
35 ["(?(#label =>\n)?)(*( )*)#class (#addr) #name (?([#targ])?) "
36 . "#svclass~(?((#svaddr))?)~#svval~(?(label \"#coplabel\")?)\n",
37 "(*( )*)goto #class (#addr)\n",
40 ["#hyphseq2 (*( (x( ;)x))*)<#classsym> #exname#arg(?([#targarglife])?)"
41 . "~#flags(?(/#private)?)(?(:#hints)?)(x(;~->#next)x)\n"
42 , " (*( )*) goto #seq\n",
43 "(?(<#seq>)?)#exname#arg(?([#targarglife])?)"],
45 ["(x(;(*( )*))x)#noise#arg(?([#targarg])?)(x( ;\n)x)",
47 "(?(#seq)?)#noise#arg(?([#targarg])?)"],
49 ["#class (#addr)\n\top_next\t\t#nextaddr\n\top_sibling\t#sibaddr\n\t"
50 . "op_ppaddr\tPL_ppaddr[OP_#NAME]\n\top_type\t\t#typenum\n" .
51 ($] > 5.009 ? '' : "\top_seq\t\t#seqnum\n")
52 . "\top_flags\t#flagval\n\top_private\t#privval\t#hintsval\n"
53 . "(?(\top_first\t#firstaddr\n)?)(?(\top_last\t\t#lastaddr\n)?)"
54 . "(?(\top_sv\t\t#svaddr\n)?)",
57 "env" => [$ENV{B_CONCISE_FORMAT}, $ENV{B_CONCISE_GOTO_FORMAT},
58 $ENV{B_CONCISE_TREE_FORMAT}],
61 # Renderings, ie how Concise prints, is controlled by these vars
63 our $stylename; # selects current style from %style
64 my $order = "basic"; # how optree is walked & printed: basic, exec, tree
66 # rendering mechanics:
67 # these 'formats' are the line-rendering templates
68 # they're updated from %style when $stylename changes
69 my ($format, $gotofmt, $treefmt);
72 my $base = 36; # how <sequence#> is displayed
73 my $big_endian = 1; # more <sequence#> display
74 my $tree_style = 0; # tree-order details
75 my $banner = 1; # print banner before optree is traversed
76 my $do_main = 0; # force printing of main routine
77 my $show_src; # show source code
79 # another factor: can affect all styles!
80 our @callbacks; # allow external management
82 set_style_standard("concise");
88 ($format, $gotofmt, $treefmt) = @_;
89 #warn "set_style: deprecated, use set_style_standard instead\n"; # someday
90 die "expecting 3 style-format args\n" unless @_ == 3;
94 my ($newstyle,@args) = @_;
95 die "style '$newstyle' already exists, choose a new name\n"
96 if exists $style{$newstyle};
97 die "expecting 3 style-format args\n" unless @args == 3;
98 $style{$newstyle} = [@args];
99 $stylename = $newstyle; # update rendering state
102 sub set_style_standard {
103 ($stylename) = @_; # update rendering state
104 die "err: style '$stylename' unknown\n" unless exists $style{$stylename};
105 set_style(@{$style{$stylename}});
112 # output handle, used with all Concise-output printing
113 our $walkHandle; # public for your convenience
114 BEGIN { $walkHandle = \*STDOUT }
116 sub walk_output { # updates $walkHandle
118 return $walkHandle unless $handle; # allow use as accessor
120 if (ref $handle eq 'SCALAR') {
122 die "no perlio in this build, can't call walk_output (\\\$scalar)\n"
123 unless $Config::Config{useperlio};
124 # in 5.8+, open(FILEHANDLE,MODE,REFERENCE) writes to string
125 open my $tmp, '>', $handle; # but cant re-set existing STDOUT
126 $walkHandle = $tmp; # so use my $tmp as intermediate var
129 my $iotype = ref $handle;
130 die "expecting argument/object that can print\n"
131 unless $iotype eq 'GLOB' or $iotype and $handle->can('print');
132 $walkHandle = $handle;
136 my($order, $coderef, $name) = @_;
137 my $codeobj = svref_2object($coderef);
139 return concise_stashref(@_)
140 unless ref $codeobj eq 'B::CV';
141 concise_cv_obj($order, $codeobj, $name);
144 sub concise_stashref {
147 foreach my $k (sort keys %$h) {
148 next unless defined $h->{$k};
150 my $coderef = *s{CODE} or next;
152 print "FUNC: ", *s, "\n";
153 my $codeobj = svref_2object($coderef);
154 next unless ref $codeobj eq 'B::CV';
155 eval { concise_cv_obj($order, $codeobj, $k) };
156 warn "err $@ on $codeobj" if $@;
160 # This should have been called concise_subref, but it was exported
161 # under this name in versions before 0.56
162 *concise_cv = \&concise_subref;
165 my ($order, $cv, $name) = @_;
166 # name is either a string, or a CODE ref (copy of $cv arg??)
170 if (ref($cv->XSUBANY) =~ /B::(\w+)/) {
171 print $walkHandle "$name is a constant sub, optimized to a $1\n";
175 print $walkHandle "$name is XS code\n";
178 if (class($cv->START) eq "NULL") {
180 if (ref $name eq 'CODE') {
181 print $walkHandle "coderef $name has no START\n";
183 elsif (exists &$name) {
184 print $walkHandle "$name exists in stash, but has no START\n";
187 print $walkHandle "$name not in symbol table\n";
191 sequence($cv->START);
192 if ($order eq "exec") {
193 walk_exec($cv->START);
195 elsif ($order eq "basic") {
196 # walk_topdown($cv->ROOT, sub { $_[0]->concise($_[1]) }, 0);
197 my $root = $cv->ROOT;
198 unless (ref $root eq 'B::NULL') {
199 walk_topdown($root, sub { $_[0]->concise($_[1]) }, 0);
201 print $walkHandle "B::NULL encountered doing ROOT on $cv. avoiding disaster\n";
204 print $walkHandle tree($cv->ROOT, 0);
210 sequence(main_start);
212 if ($order eq "exec") {
213 return if class(main_start) eq "NULL";
214 walk_exec(main_start);
215 } elsif ($order eq "tree") {
216 return if class(main_root) eq "NULL";
217 print $walkHandle tree(main_root, 0);
218 } elsif ($order eq "basic") {
219 return if class(main_root) eq "NULL";
220 walk_topdown(main_root,
221 sub { $_[0]->concise($_[1]) }, 0);
225 sub concise_specials {
226 my($name, $order, @cv_s) = @_;
228 if ($name eq "BEGIN") {
229 splice(@cv_s, 0, 8); # skip 7 BEGIN blocks in this file. NOW 8 ??
230 } elsif ($name eq "CHECK") {
231 pop @cv_s; # skip the CHECK block that calls us
234 print $walkHandle "$name $i:\n";
236 concise_cv_obj($order, $cv, $name);
240 my $start_sym = "\e(0"; # "\cN" sometimes also works
241 my $end_sym = "\e(B"; # "\cO" respectively
243 my @tree_decorations =
244 ([" ", "--", "+-", "|-", "| ", "`-", "-", 1],
245 [" ", "-", "+", "+", "|", "`", "", 0],
246 [" ", map("$start_sym$_$end_sym", "qq", "wq", "tq", "x ", "mq", "q"), 1],
247 [" ", map("$start_sym$_$end_sym", "q", "w", "t", "x", "m"), "", 0],
250 my @render_packs; # collect -stash=<packages>
253 # set rendering state from options and args
256 @options = grep(/^-/, @_);
257 @args = grep(!/^-/, @_);
259 for my $o (@options) {
261 if ($o eq "-basic") {
263 } elsif ($o eq "-exec") {
265 } elsif ($o eq "-tree") {
269 elsif ($o eq "-compact") {
271 } elsif ($o eq "-loose") {
273 } elsif ($o eq "-vt") {
275 } elsif ($o eq "-ascii") {
279 elsif ($o =~ /^-base(\d+)$/) {
281 } elsif ($o eq "-bigendian") {
283 } elsif ($o eq "-littleendian") {
286 # miscellaneous, presentation
287 elsif ($o eq "-nobanner") {
289 } elsif ($o eq "-banner") {
292 elsif ($o eq "-main") {
294 } elsif ($o eq "-nomain") {
296 } elsif ($o eq "-src") {
299 elsif ($o =~ /^-stash=(.*)/) {
302 if (!defined %{$pkg.'::'}) {
306 if (!$Config::Config{usedl}
307 && keys %{$pkg.'::'} == 1
308 && $pkg->can('bootstrap')) {
309 # It is something that we're staticly linked to, but hasn't
314 push @render_packs, $pkg;
317 elsif (exists $style{substr($o, 1)}) {
318 $stylename = substr($o, 1);
319 set_style_standard($stylename);
321 warn "Option $o unrecognized";
328 my (@args) = compileOpts(@_);
330 my @newargs = compileOpts(@_); # accept new rendering options
331 warn "disregarding non-options: @newargs\n" if @newargs;
333 for my $objname (@args) {
334 next unless $objname; # skip null args to avoid noisy responses
336 if ($objname eq "BEGIN") {
337 concise_specials("BEGIN", $order,
338 B::begin_av->isa("B::AV") ?
339 B::begin_av->ARRAY : ());
340 } elsif ($objname eq "INIT") {
341 concise_specials("INIT", $order,
342 B::init_av->isa("B::AV") ?
343 B::init_av->ARRAY : ());
344 } elsif ($objname eq "CHECK") {
345 concise_specials("CHECK", $order,
346 B::check_av->isa("B::AV") ?
347 B::check_av->ARRAY : ());
348 } elsif ($objname eq "UNITCHECK") {
349 concise_specials("UNITCHECK", $order,
350 B::unitcheck_av->isa("B::AV") ?
351 B::unitcheck_av->ARRAY : ());
352 } elsif ($objname eq "END") {
353 concise_specials("END", $order,
354 B::end_av->isa("B::AV") ?
355 B::end_av->ARRAY : ());
358 # convert function names to subrefs
361 print $walkHandle "B::Concise::compile($objname)\n"
365 $objname = "main::" . $objname unless $objname =~ /::/;
366 print $walkHandle "$objname:\n";
368 unless (exists &$objname) {
369 print $walkHandle "err: unknown function ($objname)\n";
372 $objref = \&$objname;
374 concise_subref($order, $objref, $objname);
377 for my $pkg (@render_packs) {
379 concise_stashref($order, \%{$pkg.'::'});
382 if (!@args or $do_main or @render_packs) {
383 print $walkHandle "main program:\n" if $do_main;
384 concise_main($order);
386 return @args; # something
391 my $lastnext; # remembers op-chain, used to insert gotos
393 my %opclass = ('OP' => "0", 'UNOP' => "1", 'BINOP' => "2", 'LOGOP' => "|",
394 'LISTOP' => "@", 'PMOP' => "/", 'SVOP' => "\$", 'GVOP' => "*",
395 'PVOP' => '"', 'LOOP' => "{", 'COP' => ";", 'PADOP' => "#");
397 no warnings 'qw'; # "Possible attempt to put comments..."; use #7
399 qw'# () sc ( @? 1 $* gv *{ m$ m@ m% m? p/ *$ $ $# & a& pt \\ s\\ rf bl
400 ` *? <> ?? ?/ r/ c/ // qr s/ /c y/ = @= C sC Cp sp df un BM po +1 +I
401 -1 -I 1+ I+ 1- I- ** * i* / i/ %$ i% x + i+ - i- . " << >> < i<
402 > i> <= i, >= i. == i= != i! <? i? s< s> s, s. s= s! s? b& b^ b| -0 -i
403 ! ~ a2 si cs rd sr e^ lg sq in %x %o ab le ss ve ix ri sf FL od ch cy
404 uf lf uc lc qm @ [f [ @[ eh vl ky dl ex % ${ @{ uk pk st jn ) )[ a@
405 a% sl +] -] [- [+ so rv GS GW MS MW .. f. .f && || ^^ ?: &= |= -> s{ s}
406 v} ca wa di rs ;; ; ;d }{ { } {} f{ it {l l} rt }l }n }r dm }g }e ^o
407 ^c ^| ^# um bm t~ u~ ~d DB db ^s se ^g ^r {w }w pf pr ^O ^K ^R ^W ^d ^v
408 ^e ^t ^k t. fc ic fl .s .p .b .c .l .a .h g1 s1 g2 s2 ?. l? -R -W -X -r
409 -w -x -e -o -O -z -s -M -A -C -S -c -b -f -d -p -l -u -g -k -t -T -B cd
410 co cr u. cm ut r. l@ s@ r@ mD uD oD rD tD sD wD cD f$ w$ p$ sh e$ k$ g3
411 g4 s4 g5 s5 T@ C@ L@ G@ A@ S@ Hg Hc Hr Hw Mg Mc Ms Mr Sg Sc So rq do {e
412 e} {t t} g6 G6 6e g7 G7 7e g8 G8 8e g9 G9 9e 6s 7s 8s 9s 6E 7E 8E 9E Pn
413 Pu GP SP EP Gn Gg GG SG EG g0 c$ lk t$ ;s n> // /= CO';
415 my $chars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
417 sub op_flags { # common flags (see BASOP.op_flags in op.h)
420 push @v, "v" if ($x & 3) == 1;
421 push @v, "s" if ($x & 3) == 2;
422 push @v, "l" if ($x & 3) == 3;
423 push @v, "K" if $x & 4;
424 push @v, "P" if $x & 8;
425 push @v, "R" if $x & 16;
426 push @v, "M" if $x & 32;
427 push @v, "S" if $x & 64;
428 push @v, "*" if $x & 128;
434 return "-" . base_n(-$x) if $x < 0;
436 do { $str .= substr($chars, $x % $base, 1) } while $x = int($x / $base);
437 $str = reverse $str if $big_endian;
453 return "-" if not exists $sequence_num{$$op};
454 return base_n($sequence_num{$$op});
458 my($op, $sub, $level) = @_;
460 if ($op->flags & OPf_KIDS) {
461 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
462 walk_topdown($kid, $sub, $level + 1);
465 elsif (class($op) eq "PMOP") {
466 my $maybe_root = $op->pmreplroot;
467 if (ref($maybe_root) and $maybe_root->isa("B::OP")) {
468 # It really is the root of the replacement, not something
469 # else stored here for lack of space elsewhere
470 walk_topdown($maybe_root, $sub, $level + 1);
476 my($ar, $level) = @_;
478 if (ref($l) eq "ARRAY") {
479 walklines($l, $level + 1);
487 my($top, $level) = @_;
490 my @todo = ([$top, \@lines]);
491 while (@todo and my($op, $targ) = @{shift @todo}) {
492 for (; $$op; $op = $op->next) {
493 last if $opsseen{$$op}++;
495 my $name = $op->name;
496 if (class($op) eq "LOGOP") {
499 push @todo, [$op->other, $ar];
500 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
503 push @todo, [$op->pmreplstart, $ar];
504 } elsif ($name =~ /^enter(loop|iter)$/) {
506 $labels{${$op->nextop}} = "NEXT";
507 $labels{${$op->lastop}} = "LAST";
508 $labels{${$op->redoop}} = "REDO";
510 $labels{$op->nextop->seq} = "NEXT";
511 $labels{$op->lastop->seq} = "LAST";
512 $labels{$op->redoop->seq} = "REDO";
517 walklines(\@lines, 0);
520 # The structure of this routine is purposely modeled after op.c's peep()
524 return if class($op) eq "NULL" or exists $sequence_num{$$op};
525 for (; $$op; $op = $op->next) {
526 last if exists $sequence_num{$$op};
527 my $name = $op->name;
528 if ($name =~ /^(null|scalar|lineseq|scope)$/) {
529 next if $oldop and $ {$op->next};
531 $sequence_num{$$op} = $seq_max++;
532 if (class($op) eq "LOGOP") {
533 my $other = $op->other;
534 $other = $other->next while $other->name eq "null";
536 } elsif (class($op) eq "LOOP") {
537 my $redoop = $op->redoop;
538 $redoop = $redoop->next while $redoop->name eq "null";
540 my $nextop = $op->nextop;
541 $nextop = $nextop->next while $nextop->name eq "null";
543 my $lastop = $op->lastop;
544 $lastop = $lastop->next while $lastop->name eq "null";
546 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
547 my $replstart = $op->pmreplstart;
548 $replstart = $replstart->next while $replstart->name eq "null";
549 sequence($replstart);
556 sub fmt_line { # generate text-line for op.
557 my($hr, $op, $text, $level) = @_;
559 $_->($hr, $op, \$text, \$level, $stylename) for @callbacks;
561 return '' if $hr->{SKIP}; # suppress line if a callback said so
562 return '' if $hr->{goto} and $hr->{goto} eq '-'; # no goto nowhere
564 # spec: (?(text1#varText2)?)
565 $text =~ s/\(\?\(([^\#]*?)\#(\w+)([^\#]*?)\)\?\)/
566 $hr->{$2} ? $1.$hr->{$2}.$3 : ""/eg;
568 # spec: (x(exec_text;basic_text)x)
569 $text =~ s/\(x\((.*?);(.*?)\)x\)/$order eq "exec" ? $1 : $2/egs;
572 $text =~ s/\(\*\(([^;]*?)\)\*\)/$1 x $level/egs;
574 # spec: (*(text1;text2)*)
575 $text =~ s/\(\*\((.*?);(.*?)\)\*\)/$1 x ($level - 1) . $2 x ($level>0)/egs;
577 # convert #Var to tag=>val form: Var\t#var
578 $text =~ s/\#([A-Z][a-z]+)(\d+)?/\t\u$1\t\L#$1$2/gs;
581 $text =~ s/\#([a-zA-Z]+)(\d+)/sprintf("%-$2s", $hr->{$1})/eg;
583 $text =~ s/\#([a-zA-Z]+)/$hr->{$1}/eg; # populate #var's
584 $text =~ s/[ \t]*~+[ \t]*/ /g; # squeeze tildes
586 $text = "# $hr->{src}\n$text" if $show_src and $hr->{src};
589 return "$text\n" if $text ne "";
590 return $text; # suppress empty lines
593 our %priv; # used to display each opcode's BASEOP.op_private values
595 $priv{$_}{128} = "LVINTRO"
596 for ("pos", "substr", "vec", "threadsv", "gvsv", "rv2sv", "rv2hv", "rv2gv",
597 "rv2av", "rv2arylen", "aelem", "helem", "aslice", "hslice", "padsv",
598 "padav", "padhv", "enteriter");
599 $priv{$_}{64} = "REFC" for ("leave", "leavesub", "leavesublv", "leavewrite");
600 $priv{"aassign"}{64} = "COMMON";
601 $priv{"aassign"}{32} = $] < 5.009 ? "PHASH" : "STATE";
602 $priv{"sassign"}{32} = "STATE";
603 $priv{"sassign"}{64} = "BKWARD";
604 $priv{$_}{64} = "RTIME" for ("match", "subst", "substcont", "qr");
605 @{$priv{"trans"}}{1,2,4,8,16,64} = ("<UTF", ">UTF", "IDENT", "SQUASH", "DEL",
607 $priv{"repeat"}{64} = "DOLIST";
608 $priv{"leaveloop"}{64} = "CONT";
609 @{$priv{$_}}{32,64,96} = ("DREFAV", "DREFHV", "DREFSV")
610 for (qw(rv2gv rv2sv padsv aelem helem));
611 $priv{$_}{16} = "STATE" for ("padav", "padhv", "padsv");
612 @{$priv{"entersub"}}{16,32,64} = ("DBG","TARG","NOMOD");
613 @{$priv{$_}}{4,8,128} = ("INARGS","AMPER","NO()") for ("entersub", "rv2cv");
614 $priv{"gv"}{32} = "EARLYCV";
615 $priv{"aelem"}{16} = $priv{"helem"}{16} = "LVDEFER";
616 $priv{$_}{16} = "OURINTR" for ("gvsv", "rv2sv", "rv2av", "rv2hv", "r2gv",
618 $priv{$_}{16} = "TARGMY"
619 for (map(($_,"s$_"),"chop", "chomp"),
620 map(($_,"i_$_"), "postinc", "postdec", "multiply", "divide", "modulo",
621 "add", "subtract", "negate"), "pow", "concat", "stringify",
622 "left_shift", "right_shift", "bit_and", "bit_xor", "bit_or",
623 "complement", "atan2", "sin", "cos", "rand", "exp", "log", "sqrt",
624 "int", "hex", "oct", "abs", "length", "index", "rindex", "sprintf",
625 "ord", "chr", "crypt", "quotemeta", "join", "push", "unshift", "flock",
626 "chdir", "chown", "chroot", "unlink", "chmod", "utime", "rename",
627 "link", "symlink", "mkdir", "rmdir", "wait", "waitpid", "system",
628 "exec", "kill", "getppid", "getpgrp", "setpgrp", "getpriority",
629 "setpriority", "time", "sleep");
630 $priv{$_}{4} = "REVERSED" for ("enteriter", "iter");
631 @{$priv{"const"}}{4,8,16,32,64,128} = ("SHORT","STRICT","ENTERED",'$[',"BARE","WARN");
632 $priv{"flip"}{64} = $priv{"flop"}{64} = "LINENUM";
633 $priv{"list"}{64} = "GUESSED";
634 $priv{"delete"}{64} = "SLICE";
635 $priv{"exists"}{64} = "SUB";
636 @{$priv{"sort"}}{1,2,4,8,16,32,64} = ("NUM", "INT", "REV", "INPLACE","DESC","QSORT","STABLE");
637 $priv{"threadsv"}{64} = "SVREFd";
638 @{$priv{$_}}{16,32,64,128} = ("INBIN","INCR","OUTBIN","OUTCR")
639 for ("open", "backtick");
640 $priv{"exit"}{128} = "VMS";
641 $priv{$_}{2} = "FTACCESS"
642 for ("ftrread", "ftrwrite", "ftrexec", "fteread", "ftewrite", "fteexec");
643 $priv{"entereval"}{2} = "HAS_HH";
645 # Stacked filetests are post 5.8.x
646 $priv{$_}{4} = "FTSTACKED"
647 for ("ftrread", "ftrwrite", "ftrexec", "fteread", "ftewrite", "fteexec",
648 "ftis", "fteowned", "ftrowned", "ftzero", "ftsize", "ftmtime",
649 "ftatime", "ftctime", "ftsock", "ftchr", "ftblk", "ftfile", "ftdir",
650 "ftpipe", "ftlink", "ftsuid", "ftsgid", "ftsvtx", "fttty", "fttext",
652 # Lexical $_ is post 5.8.x
653 $priv{$_}{2} = "GREPLEX"
654 for ("mapwhile", "mapstart", "grepwhile", "grepstart");
657 our %hints; # used to display each COP's op_hints values
659 # strict refs, subs, vars
660 @hints{2,512,1024} = ('$', '&', '*');
661 # integers, locale, bytes, arybase
662 @hints{1,4,8,16,32} = ('i', 'l', 'b', '[');
663 # block scope, localise %^H, $^OPEN (in), $^OPEN (out)
664 @hints{256,131072,262144,524288} = ('{','%','<','>');
665 # overload new integer, float, binary, string, re
666 @hints{4096,8192,16384,32768,65536} = ('I', 'F', 'B', 'S', 'R');
668 @hints{1048576,2097152} = ('T', 'E');
669 # filetest access, UTF-8
670 @hints{4194304,8388608} = ('X', 'U');
675 for my $flag (sort {$b <=> $a} keys %$hash) {
676 if ($hash->{$flag} and $x & $flag and $x >= $flag) {
678 push @s, $hash->{$flag};
682 return join(",", @s);
687 _flags($priv{$name}, $x);
696 my($sv, $hr, $preferpv) = @_;
697 $hr->{svclass} = class($sv);
698 $hr->{svclass} = "UV"
699 if $hr->{svclass} eq "IV" and $sv->FLAGS & SVf_IVisUV;
700 Carp::cluck("bad concise_sv: $sv") unless $sv and $$sv;
701 $hr->{svaddr} = sprintf("%#x", $$sv);
702 if ($hr->{svclass} eq "GV" && $sv->isGV_with_GP()) {
704 my $stash = $gv->STASH->NAME; if ($stash eq "main") {
707 $stash = $stash . "::";
709 $hr->{svval} = "*$stash" . $gv->SAFENAME;
710 return "*$stash" . $gv->SAFENAME;
713 while (class($sv) eq "IV" && $sv->FLAGS & SVf_ROK) {
714 $hr->{svval} .= "\\";
718 while (class($sv) eq "RV") {
719 $hr->{svval} .= "\\";
723 if (class($sv) eq "SPECIAL") {
724 $hr->{svval} .= ["Null", "sv_undef", "sv_yes", "sv_no"]->[$$sv];
725 } elsif ($preferpv && $sv->FLAGS & SVf_POK) {
726 $hr->{svval} .= cstring($sv->PV);
727 } elsif ($sv->FLAGS & SVf_NOK) {
728 $hr->{svval} .= $sv->NV;
729 } elsif ($sv->FLAGS & SVf_IOK) {
730 $hr->{svval} .= $sv->int_value;
731 } elsif ($sv->FLAGS & SVf_POK) {
732 $hr->{svval} .= cstring($sv->PV);
733 } elsif (class($sv) eq "HV") {
734 $hr->{svval} .= 'HASH';
737 $hr->{svval} = 'undef' unless defined $hr->{svval};
738 my $out = $hr->{svclass};
739 return $out .= " $hr->{svval}" ;
747 if ($fullnm eq '-e') {
748 $srclines{$fullnm} = [ $fullnm, "-src not supported for -e" ];
751 open (my $fh, '<', $fullnm)
752 or warn "# $fullnm: $!, (chdirs not supported by this feature yet)\n"
756 unshift @l, $fullnm; # like @{_<$fullnm} in debug, array starts at 1
757 $srclines{$fullnm} = \@l;
761 my ($op, $level, $format) = @_;
763 $h{exname} = $h{name} = $op->name;
764 $h{NAME} = uc $h{name};
765 $h{class} = class($op);
766 $h{extarg} = $h{targ} = $op->targ;
767 $h{extarg} = "" unless $h{extarg};
768 if ($h{name} eq "null" and $h{targ}) {
769 # targ holds the old type
770 $h{exname} = "ex-" . substr(ppname($h{targ}), 3);
772 } elsif ($op->name =~ /^leave(sub(lv)?|write)?$/) {
773 # targ potentially holds a reference count
774 if ($op->private & 64) {
775 my $refs = "ref" . ($h{targ} != 1 ? "s" : "");
776 $h{targarglife} = $h{targarg} = "$h{targ} $refs";
779 my $padname = (($curcv->PADLIST->ARRAY)[0]->ARRAY)[$h{targ}];
780 if (defined $padname and class($padname) ne "SPECIAL") {
781 $h{targarg} = $padname->PVX;
782 if ($padname->FLAGS & SVf_FAKE) {
784 $h{targarglife} = "$h{targarg}:FAKE";
786 # These changes relate to the jumbo closure fix.
787 # See changes 19939 and 20005
790 if $padname->PARENT_FAKELEX_FLAGS & PAD_FAKELEX_ANON;
792 if $padname->PARENT_FAKELEX_FLAGS & PAD_FAKELEX_MULTI;
793 $fake .= ':' . $padname->PARENT_PAD_INDEX
794 if $curcv->CvFLAGS & CVf_ANON;
795 $h{targarglife} = "$h{targarg}:FAKE:$fake";
799 my $intro = $padname->COP_SEQ_RANGE_LOW - $cop_seq_base;
800 my $finish = int($padname->COP_SEQ_RANGE_HIGH) - $cop_seq_base;
801 $finish = "end" if $finish == 999999999 - $cop_seq_base;
802 $h{targarglife} = "$h{targarg}:$intro,$finish";
805 $h{targarglife} = $h{targarg} = "t" . $h{targ};
809 $h{svclass} = $h{svaddr} = $h{svval} = "";
810 if ($h{class} eq "PMOP") {
811 my $precomp = $op->precomp;
812 if (defined $precomp) {
813 $precomp = cstring($precomp); # Escape literal control sequences
814 $precomp = "/$precomp/";
818 my $pmreplroot = $op->pmreplroot;
820 if (ref($pmreplroot) eq "B::GV") {
821 # with C<@stash_array = split(/pat/, str);>,
822 # *stash_array is stored in /pat/'s pmreplroot.
823 $h{arg} = "($precomp => \@" . $pmreplroot->NAME . ")";
824 } elsif (!ref($pmreplroot) and $pmreplroot) {
825 # same as the last case, except the value is actually a
826 # pad offset for where the GV is kept (this happens under
828 my $gv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$pmreplroot];
829 $h{arg} = "($precomp => \@" . $gv->NAME . ")";
830 } elsif ($ {$op->pmreplstart}) {
832 $pmreplstart = "replstart->" . seq($op->pmreplstart);
833 $h{arg} = "(" . join(" ", $precomp, $pmreplstart) . ")";
835 $h{arg} = "($precomp)";
837 } elsif ($h{class} eq "PVOP" and $h{name} ne "trans") {
838 $h{arg} = '("' . $op->pv . '")';
839 $h{svval} = '"' . $op->pv . '"';
840 } elsif ($h{class} eq "COP") {
841 my $label = $op->label;
842 $h{coplabel} = $label;
843 $label = $label ? "$label: " : "";
849 my($stash, $cseq) = ($op->stash->NAME, $op->cop_seq - $cop_seq_base);
850 my $arybase = $op->arybase;
851 $arybase = $arybase ? ' $[=' . $arybase : "";
852 $h{arg} = "($label$stash $cseq $loc$arybase)";
854 fill_srclines($pathnm) unless exists $srclines{$pathnm};
855 # Would love to retain Jim's use of // but this code needs to be
857 my $line = $srclines{$pathnm}[$ln];
858 $line = "-src unavailable under -e" unless defined $line;
859 $h{src} = "$ln: $line";
861 } elsif ($h{class} eq "LOOP") {
862 $h{arg} = "(next->" . seq($op->nextop) . " last->" . seq($op->lastop)
863 . " redo->" . seq($op->redoop) . ")";
864 } elsif ($h{class} eq "LOGOP") {
866 $h{arg} = "(other->" . seq($op->other) . ")";
868 elsif ($h{class} eq "SVOP" or $h{class} eq "PADOP") {
869 unless ($h{name} eq 'aelemfast' and $op->flags & OPf_SPECIAL) {
870 my $idx = ($h{class} eq "SVOP") ? $op->targ : $op->padix;
871 my $preferpv = $h{name} eq "method_named";
872 if ($h{class} eq "PADOP" or !${$op->sv}) {
873 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$idx];
874 $h{arg} = "[" . concise_sv($sv, \%h, $preferpv) . "]";
875 $h{targarglife} = $h{targarg} = "";
877 $h{arg} = "(" . concise_sv($op->sv, \%h, $preferpv) . ")";
881 $h{seq} = $h{hyphseq} = seq($op);
882 $h{seq} = "" if $h{seq} eq "-";
885 $h{label} = $labels{$$op};
887 $h{seqnum} = $op->seq;
888 $h{label} = $labels{$op->seq};
890 $h{next} = $op->next;
891 $h{next} = (class($h{next}) eq "NULL") ? "(end)" : seq($h{next});
892 $h{nextaddr} = sprintf("%#x", $ {$op->next});
893 $h{sibaddr} = sprintf("%#x", $ {$op->sibling});
894 $h{firstaddr} = sprintf("%#x", $ {$op->first}) if $op->can("first");
895 $h{lastaddr} = sprintf("%#x", $ {$op->last}) if $op->can("last");
897 $h{classsym} = $opclass{$h{class}};
898 $h{flagval} = $op->flags;
899 $h{flags} = op_flags($op->flags);
900 $h{privval} = $op->private;
901 $h{private} = private_flags($h{name}, $op->private);
902 if ($op->can("hints")) {
903 $h{hintsval} = $op->hints;
904 $h{hints} = hints_flags($h{hintsval});
906 $h{hintsval} = $h{hints} = '';
908 $h{addr} = sprintf("%#x", $$op);
909 $h{typenum} = $op->type;
910 $h{noise} = $linenoise[$op->type];
912 return fmt_line(\%h, $op, $format, $level);
916 my($op, $level) = @_;
917 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
918 # insert a 'goto' line
919 my $synth = {"seq" => seq($lastnext), "class" => class($lastnext),
920 "addr" => sprintf("%#x", $$lastnext),
921 "goto" => seq($lastnext), # simplify goto '-' removal
923 print $walkHandle fmt_line($synth, $op, $gotofmt, $level+1);
925 $lastnext = $op->next;
926 print $walkHandle concise_op($op, $level, $format);
929 # B::OP::terse (see Terse.pm) now just calls this
931 my($op, $level) = @_;
933 # This isn't necessarily right, but there's no easy way to get
934 # from an OP to the right CV. This is a limitation of the
935 # ->terse() interface style, and there isn't much to do about
936 # it. In particular, we can die in concise_op if the main pad
937 # isn't long enough, or has the wrong kind of entries, compared to
938 # the pad a sub was compiled with. The fix for that would be to
939 # make a backwards compatible "terse" format that never even
940 # looked at the pad, just like the old B::Terse. I don't think
941 # that's worth the effort, though.
942 $curcv = main_cv unless $curcv;
944 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
946 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
947 "addr" => sprintf("%#x", $$lastnext)};
949 fmt_line($h, $op, $style{"terse"}[1], $level+1);
951 $lastnext = $op->next;
953 concise_op($op, $level, $style{"terse"}[0]);
959 my $style = $tree_decorations[$tree_style];
960 my($space, $single, $kids, $kid, $nokid, $last, $lead, $size) = @$style;
961 my $name = concise_op($op, $level, $treefmt);
962 if (not $op->flags & OPf_KIDS) {
966 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
967 push @lines, tree($kid, $level+1);
970 for ($i = $#lines; substr($lines[$i], 0, 1) eq " "; $i--) {
971 $lines[$i] = $space . $lines[$i];
974 $lines[$i] = $last . $lines[$i];
976 if (substr($lines[$i], 0, 1) eq " ") {
977 $lines[$i] = $nokid . $lines[$i];
979 $lines[$i] = $kid . $lines[$i];
982 $lines[$i] = $kids . $lines[$i];
984 $lines[0] = $single . $lines[0];
986 return("$name$lead" . shift @lines,
987 map(" " x (length($name)+$size) . $_, @lines));
990 # *** Warning: fragile kludge ahead ***
991 # Because the B::* modules run in the same interpreter as the code
992 # they're compiling, their presence tends to distort the view we have of
993 # the code we're looking at. In particular, perl gives sequence numbers
994 # to COPs. If the program we're looking at were run on its own, this
995 # would start at 1. Because all of B::Concise and all the modules it
996 # uses are compiled first, though, by the time we get to the user's
997 # program the sequence number is already pretty high, which could be
998 # distracting if you're trying to tell OPs apart. Therefore we'd like to
999 # subtract an offset from all the sequence numbers we display, to
1000 # restore the simpler view of the world. The trick is to know what that
1001 # offset will be, when we're still compiling B::Concise! If we
1002 # hardcoded a value, it would have to change every time B::Concise or
1003 # other modules we use do. To help a little, what we do here is compile
1004 # a little code at the end of the module, and compute the base sequence
1005 # number for the user's program as being a small offset later, so all we
1006 # have to worry about are changes in the offset.
1008 # [For 5.8.x and earlier perl is generating sequence numbers for all ops,
1009 # and using them to reference labels]
1012 # When you say "perl -MO=Concise -e '$a'", the output should look like:
1014 # 4 <@> leave[t1] vKP/REFC ->(end)
1016 #^ smallest OP sequence number should be 1
1017 # 2 <;> nextstate(main 1 -e:1) v ->3
1018 # ^ smallest COP sequence number should be 1
1019 # - <1> ex-rv2sv vK/1 ->4
1020 # 3 <$> gvsv(*a) s ->4
1022 # If the second of the marked numbers there isn't 1, it means you need
1023 # to update the corresponding magic number in the next line.
1024 # Remember, this needs to stay the last things in the module.
1026 # Why is this different for MacOS? Does it matter?
1027 my $cop_seq_mnum = $^O eq 'MacOS' ? 12 : 11;
1028 $cop_seq_base = svref_2object(eval 'sub{0;}')->START->cop_seq + $cop_seq_mnum;
1036 B::Concise - Walk Perl syntax tree, printing concise info about ops
1040 perl -MO=Concise[,OPTIONS] foo.pl
1042 use B::Concise qw(set_style add_callback);
1046 This compiler backend prints the internal OPs of a Perl program's syntax
1047 tree in one of several space-efficient text formats suitable for debugging
1048 the inner workings of perl or other compiler backends. It can print OPs in
1049 the order they appear in the OP tree, in the order they will execute, or
1050 in a text approximation to their tree structure, and the format of the
1051 information displayed is customizable. Its function is similar to that of
1052 perl's B<-Dx> debugging flag or the B<B::Terse> module, but it is more
1053 sophisticated and flexible.
1057 Here's two outputs (or 'renderings'), using the -exec and -basic
1058 (i.e. default) formatting conventions on the same code snippet.
1060 % perl -MO=Concise,-exec -e '$a = $b + 42'
1062 2 <;> nextstate(main 1 -e:1) v
1064 4 <$> const[IV 42] s
1065 * 5 <2> add[t3] sK/2
1068 8 <@> leave[1 ref] vKP/REFC
1070 In this -exec rendering, each opcode is executed in the order shown.
1071 The add opcode, marked with '*', is discussed in more detail.
1073 The 1st column is the op's sequence number, starting at 1, and is
1074 displayed in base 36 by default. Here they're purely linear; the
1075 sequences are very helpful when looking at code with loops and
1078 The symbol between angle brackets indicates the op's type, for
1079 example; <2> is a BINOP, <@> a LISTOP, and <#> is a PADOP, which is
1080 used in threaded perls. (see L</"OP class abbreviations">).
1082 The opname, as in B<'add[t1]'>, may be followed by op-specific
1083 information in parentheses or brackets (ex B<'[t1]'>).
1085 The op-flags (ex B<'sK/2'>) are described in (L</"OP flags
1088 % perl -MO=Concise -e '$a = $b + 42'
1089 8 <@> leave[1 ref] vKP/REFC ->(end)
1091 2 <;> nextstate(main 1 -e:1) v ->3
1092 7 <2> sassign vKS/2 ->8
1093 * 5 <2> add[t1] sK/2 ->6
1094 - <1> ex-rv2sv sK/1 ->4
1095 3 <$> gvsv(*b) s ->4
1096 4 <$> const(IV 42) s ->5
1097 - <1> ex-rv2sv sKRM*/1 ->7
1098 6 <$> gvsv(*a) s ->7
1100 The default rendering is top-down, so they're not in execution order.
1101 This form reflects the way the stack is used to parse and evaluate
1102 expressions; the add operates on the two terms below it in the tree.
1104 Nullops appear as C<ex-opname>, where I<opname> is an op that has been
1105 optimized away by perl. They're displayed with a sequence-number of
1106 '-', because they are not executed (they don't appear in previous
1107 example), they're printed here because they reflect the parse.
1109 The arrow points to the sequence number of the next op; they're not
1110 displayed in -exec mode, for obvious reasons.
1112 Note that because this rendering was done on a non-threaded perl, the
1113 PADOPs in the previous examples are now SVOPs, and some (but not all)
1114 of the square brackets have been replaced by round ones. This is a
1115 subtle feature to provide some visual distinction between renderings
1116 on threaded and un-threaded perls.
1121 Arguments that don't start with a hyphen are taken to be the names of
1122 subroutines to render; if no such functions are specified, the main
1123 body of the program (outside any subroutines, and not including use'd
1124 or require'd files) is rendered. Passing C<BEGIN>, C<UNITCHECK>,
1125 C<CHECK>, C<INIT>, or C<END> will cause all of the corresponding
1126 special blocks to be printed. Arguments must follow options.
1128 Options affect how things are rendered (ie printed). They're presented
1129 here by their visual effect, 1st being strongest. They're grouped
1130 according to how they interrelate; within each group the options are
1131 mutually exclusive (unless otherwise stated).
1133 =head2 Options for Opcode Ordering
1135 These options control the 'vertical display' of opcodes. The display
1136 'order' is also called 'mode' elsewhere in this document.
1142 Print OPs in the order they appear in the OP tree (a preorder
1143 traversal, starting at the root). The indentation of each OP shows its
1144 level in the tree, and the '->' at the end of the line indicates the
1145 next opcode in execution order. This mode is the default, so the flag
1146 is included simply for completeness.
1150 Print OPs in the order they would normally execute (for the majority
1151 of constructs this is a postorder traversal of the tree, ending at the
1152 root). In most cases the OP that usually follows a given OP will
1153 appear directly below it; alternate paths are shown by indentation. In
1154 cases like loops when control jumps out of a linear path, a 'goto'
1159 Print OPs in a text approximation of a tree, with the root of the tree
1160 at the left and 'left-to-right' order of children transformed into
1161 'top-to-bottom'. Because this mode grows both to the right and down,
1162 it isn't suitable for large programs (unless you have a very wide
1167 =head2 Options for Line-Style
1169 These options select the line-style (or just style) used to render
1170 each opcode, and dictates what info is actually printed into each line.
1176 Use the author's favorite set of formatting conventions. This is the
1181 Use formatting conventions that emulate the output of B<B::Terse>. The
1182 basic mode is almost indistinguishable from the real B<B::Terse>, and the
1183 exec mode looks very similar, but is in a more logical order and lacks
1184 curly brackets. B<B::Terse> doesn't have a tree mode, so the tree mode
1185 is only vaguely reminiscent of B<B::Terse>.
1189 Use formatting conventions in which the name of each OP, rather than being
1190 written out in full, is represented by a one- or two-character abbreviation.
1191 This is mainly a joke.
1195 Use formatting conventions reminiscent of B<B::Debug>; these aren't
1196 very concise at all.
1200 Use formatting conventions read from the environment variables
1201 C<B_CONCISE_FORMAT>, C<B_CONCISE_GOTO_FORMAT>, and C<B_CONCISE_TREE_FORMAT>.
1205 =head2 Options for tree-specific formatting
1211 Use a tree format in which the minimum amount of space is used for the
1212 lines connecting nodes (one character in most cases). This squeezes out
1213 a few precious columns of screen real estate.
1217 Use a tree format that uses longer edges to separate OP nodes. This format
1218 tends to look better than the compact one, especially in ASCII, and is
1223 Use tree connecting characters drawn from the VT100 line-drawing set.
1224 This looks better if your terminal supports it.
1228 Draw the tree with standard ASCII characters like C<+> and C<|>. These don't
1229 look as clean as the VT100 characters, but they'll work with almost any
1230 terminal (or the horizontal scrolling mode of less(1)) and are suitable
1231 for text documentation or email. This is the default.
1235 These are pairwise exclusive, i.e. compact or loose, vt or ascii.
1237 =head2 Options controlling sequence numbering
1243 Print OP sequence numbers in base I<n>. If I<n> is greater than 10, the
1244 digit for 11 will be 'a', and so on. If I<n> is greater than 36, the digit
1245 for 37 will be 'A', and so on until 62. Values greater than 62 are not
1246 currently supported. The default is 36.
1250 Print sequence numbers with the most significant digit first. This is the
1251 usual convention for Arabic numerals, and the default.
1253 =item B<-littleendian>
1255 Print seqence numbers with the least significant digit first. This is
1256 obviously mutually exclusive with bigendian.
1260 =head2 Other options
1266 With this option, the rendering of each statement (starting with the
1267 nextstate OP) will be preceded by the 1st line of source code that
1268 generates it. For example:
1272 2 <;> nextstate(main 1 junk.pl:1) v:{
1273 3 <0> padsv[$i:1,10] vM/LVINTRO
1274 # 3: for $i (0..9) {
1275 4 <;> nextstate(main 3 junk.pl:3) v:{
1279 8 <{> enteriter(next->j last->m redo->9)[$i:1,10] lKS
1281 l <|> and(other->9) vK/1
1283 9 <;> nextstate(main 2 junk.pl:4) v
1285 b <$> const[PV "line "] s
1290 =item B<-stash="somepackage">
1292 With this, "somepackage" will be required, then the stash is
1293 inspected, and each function is rendered.
1297 The following options are pairwise exclusive.
1303 Include the main program in the output, even if subroutines were also
1304 specified. This rendering is normally suppressed when a subroutine
1305 name or reference is given.
1309 This restores the default behavior after you've changed it with '-main'
1310 (it's not normally needed). If no subroutine name/ref is given, main is
1311 rendered, regardless of this flag.
1315 Renderings usually include a banner line identifying the function name
1316 or stringified subref. This suppresses the printing of the banner.
1318 TBC: Remove the stringified coderef; while it provides a 'cookie' for
1319 each function rendered, the cookies used should be 1,2,3.. not a
1320 random hex-address. It also complicates string comparison of two
1325 restores default banner behavior.
1327 =item B<-banneris> => subref
1329 TBC: a hookpoint (and an option to set it) for a user-supplied
1330 function to produce a banner appropriate for users needs. It's not
1331 ideal, because the rendering-state variables, which are a natural
1332 candidate for use in concise.t, are unavailable to the user.
1336 =head2 Option Stickiness
1338 If you invoke Concise more than once in a program, you should know that
1339 the options are 'sticky'. This means that the options you provide in
1340 the first call will be remembered for the 2nd call, unless you
1341 re-specify or change them.
1343 =head1 ABBREVIATIONS
1345 The concise style uses symbols to convey maximum info with minimal
1346 clutter (like hex addresses). With just a little practice, you can
1347 start to see the flowers, not just the branches, in the trees.
1349 =head2 OP class abbreviations
1351 These symbols appear before the op-name, and indicate the
1352 B:: namespace that represents the ops in your Perl code.
1354 0 OP (aka BASEOP) An OP with no children
1355 1 UNOP An OP with one child
1356 2 BINOP An OP with two children
1357 | LOGOP A control branch OP
1358 @ LISTOP An OP that could have lots of children
1359 / PMOP An OP with a regular expression
1360 $ SVOP An OP with an SV
1361 " PVOP An OP with a string
1362 { LOOP An OP that holds pointers for a loop
1363 ; COP An OP that marks the start of a statement
1364 # PADOP An OP with a GV on the pad
1366 =head2 OP flags abbreviations
1368 OP flags are either public or private. The public flags alter the
1369 behavior of each opcode in consistent ways, and are represented by 0
1370 or more single characters.
1372 v OPf_WANT_VOID Want nothing (void context)
1373 s OPf_WANT_SCALAR Want single value (scalar context)
1374 l OPf_WANT_LIST Want list of any length (list context)
1376 K OPf_KIDS There is a firstborn child.
1377 P OPf_PARENS This operator was parenthesized.
1378 (Or block needs explicit scope entry.)
1379 R OPf_REF Certified reference.
1380 (Return container, not containee).
1381 M OPf_MOD Will modify (lvalue).
1382 S OPf_STACKED Some arg is arriving on the stack.
1383 * OPf_SPECIAL Do something weird for this op (see op.h)
1385 Private flags, if any are set for an opcode, are displayed after a '/'
1387 8 <@> leave[1 ref] vKP/REFC ->(end)
1388 7 <2> sassign vKS/2 ->8
1390 They're opcode specific, and occur less often than the public ones, so
1391 they're represented by short mnemonics instead of single-chars; see
1392 F<op.h> for gory details, or try this quick 2-liner:
1394 $> perl -MB::Concise -de 1
1395 DB<1> |x \%B::Concise::priv
1397 =head1 FORMATTING SPECIFICATIONS
1399 For each line-style ('concise', 'terse', 'linenoise', etc.) there are
1400 3 format-specs which control how OPs are rendered.
1402 The first is the 'default' format, which is used in both basic and exec
1403 modes to print all opcodes. The 2nd, goto-format, is used in exec
1404 mode when branches are encountered. They're not real opcodes, and are
1405 inserted to look like a closing curly brace. The tree-format is tree
1408 When a line is rendered, the correct format-spec is copied and scanned
1409 for the following items; data is substituted in, and other
1410 manipulations like basic indenting are done, for each opcode rendered.
1412 There are 3 kinds of items that may be populated; special patterns,
1413 #vars, and literal text, which is copied verbatim. (Yes, it's a set
1416 =head2 Special Patterns
1418 These items are the primitives used to perform indenting, and to
1419 select text from amongst alternatives.
1423 =item B<(x(>I<exec_text>B<;>I<basic_text>B<)x)>
1425 Generates I<exec_text> in exec mode, or I<basic_text> in basic mode.
1427 =item B<(*(>I<text>B<)*)>
1429 Generates one copy of I<text> for each indentation level.
1431 =item B<(*(>I<text1>B<;>I<text2>B<)*)>
1433 Generates one fewer copies of I<text1> than the indentation level, followed
1434 by one copy of I<text2> if the indentation level is more than 0.
1436 =item B<(?(>I<text1>B<#>I<var>I<Text2>B<)?)>
1438 If the value of I<var> is true (not empty or zero), generates the
1439 value of I<var> surrounded by I<text1> and I<Text2>, otherwise
1444 Any number of tildes and surrounding whitespace will be collapsed to
1451 These #vars represent opcode properties that you may want as part of
1452 your rendering. The '#' is intended as a private sigil; a #var's
1453 value is interpolated into the style-line, much like "read $this".
1455 These vars take 3 forms:
1461 A property named 'var' is assumed to exist for the opcodes, and is
1462 interpolated into the rendering.
1464 =item B<#>I<var>I<N>
1466 Generates the value of I<var>, left justified to fill I<N> spaces.
1467 Note that this means while you can have properties 'foo' and 'foo2',
1468 you cannot render 'foo2', but you could with 'foo2a'. You would be
1469 wise not to rely on this behavior going forward ;-)
1473 This ucfirst form of #var generates a tag-value form of itself for
1474 display; it converts '#Var' into a 'Var => #var' style, which is then
1475 handled as described above. (Imp-note: #Vars cannot be used for
1476 conditional-fills, because the => #var transform is done after the check
1481 The following variables are 'defined' by B::Concise; when they are
1482 used in a style, their respective values are plugged into the
1483 rendering of each opcode.
1485 Only some of these are used by the standard styles, the others are
1486 provided for you to delve into optree mechanics, should you wish to
1487 add a new style (see L</add_style> below) that uses them. You can
1488 also add new ones using L</add_callback>.
1494 The address of the OP, in hexadecimal.
1498 The OP-specific information of the OP (such as the SV for an SVOP, the
1499 non-local exit pointers for a LOOP, etc.) enclosed in parentheses.
1503 The B-determined class of the OP, in all caps.
1507 A single symbol abbreviating the class of the OP.
1511 The label of the statement or block the OP is the start of, if any.
1515 The name of the OP, or 'ex-foo' if the OP is a null that used to be a foo.
1519 The target of the OP, or nothing for a nulled OP.
1523 The address of the OP's first child, in hexadecimal.
1527 The OP's flags, abbreviated as a series of symbols.
1531 The numeric value of the OP's flags.
1535 The COP's hint flags, rendered with abbreviated names if possible. An empty
1536 string if this is not a COP. Here are the symbols used:
1561 The numeric value of the COP's hint flags, or an empty string if this is not
1566 The sequence number of the OP, or a hyphen if it doesn't have one.
1570 'NEXT', 'LAST', or 'REDO' if the OP is a target of one of those in exec
1571 mode, or empty otherwise.
1575 The address of the OP's last child, in hexadecimal.
1583 The OP's name, in all caps.
1587 The sequence number of the OP's next OP.
1591 The address of the OP's next OP, in hexadecimal.
1595 A one- or two-character abbreviation for the OP's name.
1599 The OP's private flags, rendered with abbreviated names if possible.
1603 The numeric value of the OP's private flags.
1607 The sequence number of the OP. Note that this is a sequence number
1608 generated by B::Concise.
1612 5.8.x and earlier only. 5.9 and later do not provide this.
1614 The real sequence number of the OP, as a regular number and not adjusted
1615 to be relative to the start of the real program. (This will generally be
1616 a fairly large number because all of B<B::Concise> is compiled before
1621 Whether or not the op has been optimised by the peephole optimiser.
1623 Only available in 5.9 and later.
1627 The address of the OP's next youngest sibling, in hexadecimal.
1631 The address of the OP's SV, if it has an SV, in hexadecimal.
1635 The class of the OP's SV, if it has one, in all caps (e.g., 'IV').
1639 The value of the OP's SV, if it has one, in a short human-readable format.
1643 The numeric value of the OP's targ.
1647 The name of the variable the OP's targ refers to, if any, otherwise the
1648 letter t followed by the OP's targ in decimal.
1650 =item B<#targarglife>
1652 Same as B<#targarg>, but followed by the COP sequence numbers that delimit
1653 the variable's lifetime (or 'end' for a variable in an open scope) for a
1658 The numeric value of the OP's type, in decimal.
1662 =head1 One-Liner Command tips
1666 =item perl -MO=Concise,bar foo.pl
1668 Renders only bar() from foo.pl. To see main, drop the ',bar'. To see
1671 =item perl -MDigest::MD5=md5 -MO=Concise,md5 -e1
1673 Identifies md5 as an XS function. The export is needed so that BC can
1676 =item perl -MPOSIX -MO=Concise,_POSIX_ARG_MAX -e1
1678 Identifies _POSIX_ARG_MAX as a constant sub, optimized to an IV.
1679 Although POSIX isn't entirely consistent across platforms, this is
1680 likely to be present in virtually all of them.
1682 =item perl -MPOSIX -MO=Concise,a -e 'print _POSIX_SAVED_IDS'
1684 This renders a print statement, which includes a call to the function.
1685 It's identical to rendering a file with a use call and that single
1686 statement, except for the filename which appears in the nextstate ops.
1688 =item perl -MPOSIX -MO=Concise,a -e 'sub a{_POSIX_SAVED_IDS}'
1690 This is B<very> similar to previous, only the first two ops differ. This
1691 subroutine rendering is more representative, insofar as a single main
1692 program will have many subs.
1694 =item perl -MB::Concise -e 'B::Concise::compile("-exec","-src", \%B::Concise::)->()'
1696 This renders all functions in the B::Concise package with the source
1697 lines. It eschews the O framework so that the stashref can be passed
1698 directly to B::Concise::compile(). See -stash option for a more
1699 convenient way to render a package.
1703 =head1 Using B::Concise outside of the O framework
1705 The common (and original) usage of B::Concise was for command-line
1706 renderings of simple code, as given in EXAMPLE. But you can also use
1707 B<B::Concise> from your code, and call compile() directly, and
1708 repeatedly. By doing so, you can avoid the compile-time only
1709 operation of O.pm, and even use the debugger to step through
1710 B::Concise::compile() itself.
1712 Once you're doing this, you may alter Concise output by adding new
1713 rendering styles, and by optionally adding callback routines which
1714 populate new variables, if such were referenced from those (just
1717 =head2 Example: Altering Concise Renderings
1719 use B::Concise qw(set_style add_callback);
1720 add_style($yourStyleName => $defaultfmt, $gotofmt, $treefmt);
1723 my ($h, $op, $format, $level, $stylename) = @_;
1724 $h->{variable} = some_func($op);
1726 $walker = B::Concise::compile(@options,@subnames,@subrefs);
1731 B<set_style> accepts 3 arguments, and updates the three format-specs
1732 comprising a line-style (basic-exec, goto, tree). It has one minor
1733 drawback though; it doesn't register the style under a new name. This
1734 can become an issue if you render more than once and switch styles.
1735 Thus you may prefer to use add_style() and/or set_style_standard()
1738 =head2 set_style_standard($name)
1740 This restores one of the standard line-styles: C<terse>, C<concise>,
1741 C<linenoise>, C<debug>, C<env>, into effect. It also accepts style
1742 names previously defined with add_style().
1746 This subroutine accepts a new style name and three style arguments as
1747 above, and creates, registers, and selects the newly named style. It is
1748 an error to re-add a style; call set_style_standard() to switch between
1751 =head2 add_callback ()
1753 If your newly minted styles refer to any new #variables, you'll need
1754 to define a callback subroutine that will populate (or modify) those
1755 variables. They are then available for use in the style you've
1758 The callbacks are called for each opcode visited by Concise, in the
1759 same order as they are added. Each subroutine is passed five
1762 1. A hashref, containing the variable names and values which are
1763 populated into the report-line for the op
1764 2. the op, as a B<B::OP> object
1765 3. a reference to the format string
1766 4. the formatting (indent) level
1767 5. the selected stylename
1769 To define your own variables, simply add them to the hash, or change
1770 existing values if you need to. The level and format are passed in as
1771 references to scalars, but it is unlikely that they will need to be
1772 changed or even used.
1774 =head2 Running B::Concise::compile()
1776 B<compile> accepts options as described above in L</OPTIONS>, and
1777 arguments, which are either coderefs, or subroutine names.
1779 It constructs and returns a $treewalker coderef, which when invoked,
1780 traverses, or walks, and renders the optrees of the given arguments to
1781 STDOUT. You can reuse this, and can change the rendering style used
1782 each time; thereafter the coderef renders in the new style.
1784 B<walk_output> lets you change the print destination from STDOUT to
1785 another open filehandle, or into a string passed as a ref (unless
1786 you've built perl with -Uuseperlio).
1788 my $walker = B::Concise::compile('-terse','aFuncName', \&aSubRef); # 1
1789 walk_output(\my $buf);
1790 $walker->(); # 1 renders -terse
1791 set_style_standard('concise'); # 2
1792 $walker->(); # 2 renders -concise
1793 $walker->(@new); # 3 renders whatever
1794 print "3 different renderings: terse, concise, and @new: $buf\n";
1796 When $walker is called, it traverses the subroutines supplied when it
1797 was created, and renders them using the current style. You can change
1798 the style afterwards in several different ways:
1800 1. call C<compile>, altering style or mode/order
1801 2. call C<set_style_standard>
1802 3. call $walker, passing @new options
1804 Passing new options to the $walker is the easiest way to change
1805 amongst any pre-defined styles (the ones you add are automatically
1806 recognized as options), and is the only way to alter rendering order
1807 without calling compile again. Note however that rendering state is
1808 still shared amongst multiple $walker objects, so they must still be
1809 used in a coordinated manner.
1811 =head2 B::Concise::reset_sequence()
1813 This function (not exported) lets you reset the sequence numbers (note
1814 that they're numbered arbitrarily, their goal being to be human
1815 readable). Its purpose is mostly to support testing, i.e. to compare
1816 the concise output from two identical anonymous subroutines (but
1817 different instances). Without the reset, B::Concise, seeing that
1818 they're separate optrees, generates different sequence numbers in
1823 Errors in rendering (non-existent function-name, non-existent coderef)
1824 are written to the STDOUT, or wherever you've set it via
1827 Errors using the various *style* calls, and bad args to walk_output(),
1828 result in die(). Use an eval if you wish to catch these errors and
1829 continue processing.
1833 Stephen McCamant, E<lt>smcc@CSUA.Berkeley.EDUE<gt>.