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.
11 our $VERSION = "0.55";
12 our @ISA = qw(Exporter);
13 our @EXPORT_OK = qw(set_style set_style_standard add_callback
14 concise_cv concise_main);
16 use B qw(class ppname main_start main_root main_cv cstring svref_2object
17 SVf_IOK SVf_NOK SVf_POK SVf_IVisUV OPf_KIDS);
21 ["(?(#label =>\n)?)(*( )*)#class (#addr) #name (?([#targ])?) "
22 . "#svclass~(?((#svaddr))?)~#svval~(?(label \"#coplabel\")?)\n",
23 "(*( )*)goto #class (#addr)\n",
26 ["#hyphseq2 (*( (x( ;)x))*)<#classsym> "
27 . "#exname#arg(?([#targarglife])?)~#flags(?(/#private)?)(x(;~->#next)x)\n",
28 " (*( )*) goto #seq\n",
29 "(?(<#seq>)?)#exname#arg(?([#targarglife])?)"],
31 ["(x(;(*( )*))x)#noise#arg(?([#targarg])?)(x( ;\n)x)",
33 "(?(#seq)?)#noise#arg(?([#targarg])?)"],
35 ["#class (#addr)\n\top_next\t\t#nextaddr\n\top_sibling\t#sibaddr\n\t"
36 . "op_ppaddr\tPL_ppaddr[OP_#NAME]\n\top_type\t\t#typenum\n\top_seq\t\t"
37 . "#seqnum\n\top_flags\t#flagval\n\top_private\t#privval\n"
38 . "(?(\top_first\t#firstaddr\n)?)(?(\top_last\t\t#lastaddr\n)?)"
39 . "(?(\top_sv\t\t#svaddr\n)?)",
42 "env" => [$ENV{B_CONCISE_FORMAT}, $ENV{B_CONCISE_GOTO_FORMAT},
43 $ENV{B_CONCISE_TREE_FORMAT}],
46 my($format, $gotofmt, $treefmt);
52 ($format, $gotofmt, $treefmt) = @_;
55 sub set_style_standard {
57 set_style(@{$style{$name}});
65 my ($order, $cvref) = @_;
66 my $cv = svref_2object($cvref);
69 if ($order eq "exec") {
70 walk_exec($cv->START);
71 } elsif ($order eq "basic") {
72 walk_topdown($cv->ROOT, sub { $_[0]->concise($_[1]) }, 0);
74 print tree($cv->ROOT, 0)
82 if ($order eq "exec") {
83 return if class(main_start) eq "NULL";
84 walk_exec(main_start);
85 } elsif ($order eq "tree") {
86 return if class(main_root) eq "NULL";
87 print tree(main_root, 0);
88 } elsif ($order eq "basic") {
89 return if class(main_root) eq "NULL";
90 walk_topdown(main_root,
91 sub { $_[0]->concise($_[1]) }, 0);
95 my $start_sym = "\e(0"; # "\cN" sometimes also works
96 my $end_sym = "\e(B"; # "\cO" respectively
98 my @tree_decorations =
99 ([" ", "--", "+-", "|-", "| ", "`-", "-", 1],
100 [" ", "-", "+", "+", "|", "`", "", 0],
101 [" ", map("$start_sym$_$end_sym", "qq", "wq", "tq", "x ", "mq", "q"), 1],
102 [" ", map("$start_sym$_$end_sym", "q", "w", "t", "x", "m"), "", 0],
111 set_style_standard("concise");
114 my @options = grep(/^-/, @_);
115 my @args = grep(!/^-/, @_);
117 for my $o (@options) {
118 if ($o eq "-basic") {
120 } elsif ($o eq "-exec") {
122 } elsif ($o eq "-tree") {
124 } elsif ($o eq "-compact") {
126 } elsif ($o eq "-loose") {
128 } elsif ($o eq "-vt") {
130 } elsif ($o eq "-ascii") {
132 } elsif ($o eq "-main") {
134 } elsif ($o =~ /^-base(\d+)$/) {
136 } elsif ($o eq "-bigendian") {
138 } elsif ($o eq "-littleendian") {
140 } elsif (exists $style{substr($o, 1)}) {
141 set_style(@{$style{substr($o, 1)}});
143 warn "Option $o unrecognized";
148 for my $objname (@args) {
149 $objname = "main::" . $objname unless $objname =~ /::/;
151 eval "concise_cv(\$order, \\&$objname)";
152 die "concise_cv($order, \\&$objname) failed: $@" if $@;
155 if (!@args or $do_main) {
156 print "main program:\n" if $do_main;
157 concise_main($order);
165 my %opclass = ('OP' => "0", 'UNOP' => "1", 'BINOP' => "2", 'LOGOP' => "|",
166 'LISTOP' => "@", 'PMOP' => "/", 'SVOP' => "\$", 'GVOP' => "*",
167 'PVOP' => '"', 'LOOP' => "{", 'COP' => ";", 'PADOP' => "#");
169 no warnings 'qw'; # "Possible attempt to put comments..."
171 qw'# () sc ( @? 1 $* gv *{ m$ m@ m% m? p/ *$ $ $# & a& pt \\ s\\ rf bl
172 ` *? <> ?? ?/ r/ c/ // qr s/ /c y/ = @= C sC Cp sp df un BM po +1 +I
173 -1 -I 1+ I+ 1- I- ** * i* / i/ %$ i% x + i+ - i- . " << >> < i<
174 > i> <= i, >= i. == i= != i! <? i? s< s> s, s. s= s! s? b& b^ b| -0 -i
175 ! ~ a2 si cs rd sr e^ lg sq in %x %o ab le ss ve ix ri sf FL od ch cy
176 uf lf uc lc qm @ [f [ @[ eh vl ky dl ex % ${ @{ uk pk st jn ) )[ a@
177 a% sl +] -] [- [+ so rv GS GW MS MW .. f. .f && || ^^ ?: &= |= -> s{ s}
178 v} ca wa di rs ;; ; ;d }{ { } {} f{ it {l l} rt }l }n }r dm }g }e ^o
179 ^c ^| ^# um bm t~ u~ ~d DB db ^s se ^g ^r {w }w pf pr ^O ^K ^R ^W ^d ^v
180 ^e ^t ^k t. fc ic fl .s .p .b .c .l .a .h g1 s1 g2 s2 ?. l? -R -W -X -r
181 -w -x -e -o -O -z -s -M -A -C -S -c -b -f -d -p -l -u -g -k -t -T -B cd
182 co cr u. cm ut r. l@ s@ r@ mD uD oD rD tD sD wD cD f$ w$ p$ sh e$ k$ g3
183 g4 s4 g5 s5 T@ C@ L@ G@ A@ S@ Hg Hc Hr Hw Mg Mc Ms Mr Sg Sc So rq do {e
184 e} {t t} g6 G6 6e g7 G7 7e g8 G8 8e g9 G9 9e 6s 7s 8s 9s 6E 7E 8E 9E Pn
185 Pu GP SP EP Gn Gg GG SG EG g0 c$ lk t$ ;s n> // /= CO';
187 my $chars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
192 push @v, "v" if ($x & 3) == 1;
193 push @v, "s" if ($x & 3) == 2;
194 push @v, "l" if ($x & 3) == 3;
195 push @v, "K" if $x & 4;
196 push @v, "P" if $x & 8;
197 push @v, "R" if $x & 16;
198 push @v, "M" if $x & 32;
199 push @v, "S" if $x & 64;
200 push @v, "*" if $x & 128;
206 return "-" . base_n(-$x) if $x < 0;
208 do { $str .= substr($chars, $x % $base, 1) } while $x = int($x / $base);
209 $str = reverse $str if $big_endian;
218 return "-" if not exists $sequence_num{$$op};
219 return base_n($sequence_num{$$op});
223 my($op, $sub, $level) = @_;
225 if ($op->flags & OPf_KIDS) {
226 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
227 walk_topdown($kid, $sub, $level + 1);
230 if (class($op) eq "PMOP" and $op->pmreplroot and $ {$op->pmreplroot}
231 and $op->pmreplroot->isa("B::OP")) {
232 walk_topdown($op->pmreplroot, $sub, $level + 1);
237 my($ar, $level) = @_;
239 if (ref($l) eq "ARRAY") {
240 walklines($l, $level + 1);
248 my($top, $level) = @_;
251 my @todo = ([$top, \@lines]);
252 while (@todo and my($op, $targ) = @{shift @todo}) {
253 for (; $$op; $op = $op->next) {
254 last if $opsseen{$$op}++;
256 my $name = $op->name;
257 if (class($op) eq "LOGOP") {
260 push @todo, [$op->other, $ar];
261 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
264 push @todo, [$op->pmreplstart, $ar];
265 } elsif ($name =~ /^enter(loop|iter)$/) {
266 $labels{$op->nextop->seq} = "NEXT";
267 $labels{$op->lastop->seq} = "LAST";
268 $labels{$op->redoop->seq} = "REDO";
272 walklines(\@lines, 0);
275 # The structure of this routine is purposely modeled after op.c's peep()
279 return if class($op) eq "NULL" or exists $sequence_num{$$op};
280 for (; $$op; $op = $op->next) {
281 last if exists $sequence_num{$$op};
282 my $name = $op->name;
283 if ($name =~ /^(null|scalar|lineseq|scope)$/) {
284 next if $oldop and $ {$op->next};
286 $sequence_num{$$op} = $seq_max++;
287 if (class($op) eq "LOGOP") {
288 my $other = $op->other;
289 $other = $other->next while $other->name eq "null";
291 } elsif (class($op) eq "LOOP") {
292 my $redoop = $op->redoop;
293 $redoop = $redoop->next while $redoop->name eq "null";
295 my $nextop = $op->nextop;
296 $nextop = $nextop->next while $nextop->name eq "null";
298 my $lastop = $op->lastop;
299 $lastop = $lastop->next while $lastop->name eq "null";
301 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
302 my $replstart = $op->pmreplstart;
303 $replstart = $replstart->next while $replstart->name eq "null";
304 sequence($replstart);
312 my($hr, $fmt, $level) = @_;
314 $text =~ s/\(\?\(([^\#]*?)\#(\w+)([^\#]*?)\)\?\)/
315 $hr->{$2} ? $1.$hr->{$2}.$3 : ""/eg;
316 $text =~ s/\(x\((.*?);(.*?)\)x\)/$order eq "exec" ? $1 : $2/egs;
317 $text =~ s/\(\*\(([^;]*?)\)\*\)/$1 x $level/egs;
318 $text =~ s/\(\*\((.*?);(.*?)\)\*\)/$1 x ($level - 1) . $2 x ($level>0)/egs;
319 $text =~ s/#([a-zA-Z]+)(\d+)/sprintf("%-$2s", $hr->{$1})/eg;
320 $text =~ s/#([a-zA-Z]+)/$hr->{$1}/eg;
321 $text =~ s/[ \t]*~+[ \t]*/ /g;
326 $priv{$_}{128} = "LVINTRO"
327 for ("pos", "substr", "vec", "threadsv", "gvsv", "rv2sv", "rv2hv", "rv2gv",
328 "rv2av", "rv2arylen", "aelem", "helem", "aslice", "hslice", "padsv",
330 $priv{$_}{64} = "REFC" for ("leave", "leavesub", "leavesublv", "leavewrite");
331 $priv{"aassign"}{64} = "COMMON";
332 $priv{"sassign"}{64} = "BKWARD";
333 $priv{$_}{64} = "RTIME" for ("match", "subst", "substcont");
334 @{$priv{"trans"}}{1,2,4,8,16,64} = ("<UTF", ">UTF", "IDENT", "SQUASH", "DEL",
336 $priv{"repeat"}{64} = "DOLIST";
337 $priv{"leaveloop"}{64} = "CONT";
338 @{$priv{$_}}{32,64,96} = ("DREFAV", "DREFHV", "DREFSV")
339 for ("entersub", map("rv2${_}v", "a", "s", "h", "g"), "aelem", "helem");
340 $priv{"entersub"}{16} = "DBG";
341 $priv{"entersub"}{32} = "TARG";
342 @{$priv{$_}}{4,8,128} = ("INARGS","AMPER","NO()") for ("entersub", "rv2cv");
343 $priv{"gv"}{32} = "EARLYCV";
344 $priv{"aelem"}{16} = $priv{"helem"}{16} = "LVDEFER";
345 $priv{$_}{16} = "OURINTR" for ("gvsv", "rv2sv", "rv2av", "rv2hv", "r2gv");
346 $priv{$_}{16} = "TARGMY"
347 for (map(($_,"s$_"),"chop", "chomp"),
348 map(($_,"i_$_"), "postinc", "postdec", "multiply", "divide", "modulo",
349 "add", "subtract", "negate"), "pow", "concat", "stringify",
350 "left_shift", "right_shift", "bit_and", "bit_xor", "bit_or",
351 "complement", "atan2", "sin", "cos", "rand", "exp", "log", "sqrt",
352 "int", "hex", "oct", "abs", "length", "index", "rindex", "sprintf",
353 "ord", "chr", "crypt", "quotemeta", "join", "push", "unshift", "flock",
354 "chdir", "chown", "chroot", "unlink", "chmod", "utime", "rename",
355 "link", "symlink", "mkdir", "rmdir", "wait", "waitpid", "system",
356 "exec", "kill", "getppid", "getpgrp", "setpgrp", "getpriority",
357 "setpriority", "time", "sleep");
358 @{$priv{"const"}}{8,16,32,64,128} = ("STRICT","ENTERED", '$[', "BARE", "WARN");
359 $priv{"flip"}{64} = $priv{"flop"}{64} = "LINENUM";
360 $priv{"list"}{64} = "GUESSED";
361 $priv{"delete"}{64} = "SLICE";
362 $priv{"exists"}{64} = "SUB";
363 $priv{$_}{64} = "LOCALE"
364 for ("sort", "prtf", "sprintf", "slt", "sle", "seq", "sne", "sgt", "sge",
365 "scmp", "lc", "uc", "lcfirst", "ucfirst");
366 @{$priv{"sort"}}{1,2,4} = ("NUM", "INT", "REV");
367 $priv{"threadsv"}{64} = "SVREFd";
368 @{$priv{$_}}{16,32,64,128} = ("INBIN","INCR","OUTBIN","OUTCR")
369 for ("open", "backtick");
370 $priv{"exit"}{128} = "VMS";
375 for my $flag (128, 96, 64, 32, 16, 8, 4, 2, 1) {
376 if ($priv{$name}{$flag} and $x & $flag and $x >= $flag) {
378 push @s, $priv{$name}{$flag};
382 return join(",", @s);
387 $hr->{svclass} = class($sv);
388 $hr->{svclass} = "UV"
389 if $hr->{svclass} eq "IV" and $sv->FLAGS & SVf_IVisUV;
390 $hr->{svaddr} = sprintf("%#x", $$sv);
391 if ($hr->{svclass} eq "GV") {
393 my $stash = $gv->STASH->NAME;
394 if ($stash eq "main") {
397 $stash = $stash . "::";
399 $hr->{svval} = "*$stash" . $gv->SAFENAME;
400 return "*$stash" . $gv->SAFENAME;
402 while (class($sv) eq "RV") {
403 $hr->{svval} .= "\\";
406 if (class($sv) eq "SPECIAL") {
407 $hr->{svval} .= ["Null", "sv_undef", "sv_yes", "sv_no"]->[$$sv];
408 } elsif ($sv->FLAGS & SVf_NOK) {
409 $hr->{svval} .= $sv->NV;
410 } elsif ($sv->FLAGS & SVf_IOK) {
411 $hr->{svval} .= $sv->int_value;
412 } elsif ($sv->FLAGS & SVf_POK) {
413 $hr->{svval} .= cstring($sv->PV);
414 } elsif (class($sv) eq "HV") {
415 $hr->{svval} .= 'HASH';
417 return $hr->{svclass} . " " . $hr->{svval};
422 my ($op, $level, $format) = @_;
424 $h{exname} = $h{name} = $op->name;
425 $h{NAME} = uc $h{name};
426 $h{class} = class($op);
427 $h{extarg} = $h{targ} = $op->targ;
428 $h{extarg} = "" unless $h{extarg};
429 if ($h{name} eq "null" and $h{targ}) {
430 $h{exname} = "ex-" . substr(ppname($h{targ}), 3);
433 my $padname = (($curcv->PADLIST->ARRAY)[0]->ARRAY)[$h{targ}];
434 if (defined $padname and class($padname) ne "SPECIAL") {
435 $h{targarg} = $padname->PVX;
436 my $intro = $padname->NVX - $cop_seq_base;
437 my $finish = int($padname->IVX) - $cop_seq_base;
438 $finish = "end" if $finish == 999999999 - $cop_seq_base;
439 $h{targarglife} = "$h{targarg}:$intro,$finish";
441 $h{targarglife} = $h{targarg} = "t" . $h{targ};
445 $h{svclass} = $h{svaddr} = $h{svval} = "";
446 if ($h{class} eq "PMOP") {
447 my $precomp = $op->precomp;
448 if (defined $precomp) {
449 $precomp = cstring($precomp); # Escape literal control sequences
450 $precomp = "/$precomp/";
454 my $pmreplroot = $op->pmreplroot;
456 if ($pmreplroot && $$pmreplroot && $pmreplroot->isa("B::GV")) {
457 # with C<@stash_array = split(/pat/, str);>,
458 # *stash_array is stored in pmreplroot.
459 $h{arg} = "($precomp => \@" . $pmreplroot->NAME . ")";
460 } elsif ($ {$op->pmreplstart}) {
462 $pmreplstart = "replstart->" . seq($op->pmreplstart);
463 $h{arg} = "(" . join(" ", $precomp, $pmreplstart) . ")";
465 $h{arg} = "($precomp)";
467 } elsif ($h{class} eq "PVOP" and $h{name} ne "trans") {
468 $h{arg} = '("' . $op->pv . '")';
469 $h{svval} = '"' . $op->pv . '"';
470 } elsif ($h{class} eq "COP") {
471 my $label = $op->label;
472 $h{coplabel} = $label;
473 $label = $label ? "$label: " : "";
476 $loc .= ":" . $op->line;
477 my($stash, $cseq) = ($op->stash->NAME, $op->cop_seq - $cop_seq_base);
478 my $arybase = $op->arybase;
479 $arybase = $arybase ? ' $[=' . $arybase : "";
480 $h{arg} = "($label$stash $cseq $loc$arybase)";
481 } elsif ($h{class} eq "LOOP") {
482 $h{arg} = "(next->" . seq($op->nextop) . " last->" . seq($op->lastop)
483 . " redo->" . seq($op->redoop) . ")";
484 } elsif ($h{class} eq "LOGOP") {
486 $h{arg} = "(other->" . seq($op->other) . ")";
487 } elsif ($h{class} eq "SVOP") {
489 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->targ];
490 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
491 $h{targarglife} = $h{targarg} = "";
493 $h{arg} = "(" . concise_sv($op->sv, \%h) . ")";
495 } elsif ($h{class} eq "PADOP") {
496 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->padix];
497 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
499 $h{seq} = $h{hyphseq} = seq($op);
500 $h{seq} = "" if $h{seq} eq "-";
501 $h{seqnum} = $op->seq;
502 $h{next} = $op->next;
503 $h{next} = (class($h{next}) eq "NULL") ? "(end)" : seq($h{next});
504 $h{nextaddr} = sprintf("%#x", $ {$op->next});
505 $h{sibaddr} = sprintf("%#x", $ {$op->sibling});
506 $h{firstaddr} = sprintf("%#x", $ {$op->first}) if $op->can("first");
507 $h{lastaddr} = sprintf("%#x", $ {$op->last}) if $op->can("last");
509 $h{classsym} = $opclass{$h{class}};
510 $h{flagval} = $op->flags;
511 $h{flags} = op_flags($op->flags);
512 $h{privval} = $op->private;
513 $h{private} = private_flags($h{name}, $op->private);
514 $h{addr} = sprintf("%#x", $$op);
515 $h{label} = $labels{$op->seq};
516 $h{typenum} = $op->type;
517 $h{noise} = $linenoise[$op->type];
518 $_->(\%h, $op, \$format, \$level) for @callbacks;
519 return fmt_line(\%h, $format, $level);
523 my($op, $level) = @_;
524 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
525 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
526 "addr" => sprintf("%#x", $$lastnext)};
527 print fmt_line($h, $gotofmt, $level+1);
529 $lastnext = $op->next;
530 print concise_op($op, $level, $format);
533 # B::OP::terse (see Terse.pm) now just calls this
535 my($op, $level) = @_;
537 # This isn't necessarily right, but there's no easy way to get
538 # from an OP to the right CV. This is a limitation of the
539 # ->terse() interface style, and there isn't much to do about
540 # it. In particular, we can die in concise_op if the main pad
541 # isn't long enough, or has the wrong kind of entries, compared to
542 # the pad a sub was compiled with. The fix for that would be to
543 # make a backwards compatible "terse" format that never even
544 # looked at the pad, just like the old B::Terse. I don't think
545 # that's worth the effort, though.
546 $curcv = main_cv unless $curcv;
548 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
549 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
550 "addr" => sprintf("%#x", $$lastnext)};
551 print fmt_line($h, $style{"terse"}[1], $level+1);
553 $lastnext = $op->next;
554 print concise_op($op, $level, $style{"terse"}[0]);
560 my $style = $tree_decorations[$tree_style];
561 my($space, $single, $kids, $kid, $nokid, $last, $lead, $size) = @$style;
562 my $name = concise_op($op, $level, $treefmt);
563 if (not $op->flags & OPf_KIDS) {
567 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
568 push @lines, tree($kid, $level+1);
571 for ($i = $#lines; substr($lines[$i], 0, 1) eq " "; $i--) {
572 $lines[$i] = $space . $lines[$i];
575 $lines[$i] = $last . $lines[$i];
577 if (substr($lines[$i], 0, 1) eq " ") {
578 $lines[$i] = $nokid . $lines[$i];
580 $lines[$i] = $kid . $lines[$i];
583 $lines[$i] = $kids . $lines[$i];
585 $lines[0] = $single . $lines[0];
587 return("$name$lead" . shift @lines,
588 map(" " x (length($name)+$size) . $_, @lines));
591 # *** Warning: fragile kludge ahead ***
592 # Because the B::* modules run in the same interpreter as the code
593 # they're compiling, their presence tends to distort the view we have
594 # of the code we're looking at. In particular, perl gives sequence
595 # numbers to both OPs in general and COPs in particular. If the
596 # program we're looking at were run on its own, these numbers would
597 # start at 1. Because all of B::Concise and all the modules it uses
598 # are compiled first, though, by the time we get to the user's program
599 # the sequence numbers are alreay at pretty high numbers, which would
600 # be distracting if you're trying to tell OPs apart. Therefore we'd
601 # like to subtract an offset from all the sequence numbers we display,
602 # to restore the simpler view of the world. The trick is to know what
603 # that offset will be, when we're still compiling B::Concise! If we
604 # hardcoded a value, it would have to change every time B::Concise or
605 # other modules we use do. To help a little, what we do here is
606 # compile a little code at the end of the module, and compute the base
607 # sequence number for the user's program as being a small offset
608 # later, so all we have to worry about are changes in the offset.
609 # (Note that we now only play this game with COP sequence numbers. OP
610 # sequence numbers aren't used to refer to OPs from a distance, and
611 # they don't have much significance, so we just generate our own
612 # sequence numbers which are easier to control. This way we also don't
613 # stand in the way of a possible future removal of OP sequence
616 # When you say "perl -MO=Concise -e '$a'", the output should look like:
618 # 4 <@> leave[t1] vKP/REFC ->(end)
620 #^ smallest OP sequence number should be 1
621 # 2 <;> nextstate(main 1 -e:1) v ->3
622 # ^ smallest COP sequence number should be 1
623 # - <1> ex-rv2sv vK/1 ->4
624 # 3 <$> gvsv(*a) s ->4
626 # If the second of the marked numbers there isn't 1, it means you need
627 # to update the corresponding magic number in the next line.
628 # Remember, this needs to stay the last things in the module.
630 # Why is this different for MacOS? Does it matter?
631 my $cop_seq_mnum = $^O eq 'MacOS' ? 10 : 9;
632 $cop_seq_base = svref_2object(eval 'sub{0;}')->START->cop_seq + $cop_seq_mnum;
640 B::Concise - Walk Perl syntax tree, printing concise info about ops
644 perl -MO=Concise[,OPTIONS] foo.pl
646 use B::Concise qw(set_style add_callback);
650 This compiler backend prints the internal OPs of a Perl program's syntax
651 tree in one of several space-efficient text formats suitable for debugging
652 the inner workings of perl or other compiler backends. It can print OPs in
653 the order they appear in the OP tree, in the order they will execute, or
654 in a text approximation to their tree structure, and the format of the
655 information displyed is customizable. Its function is similar to that of
656 perl's B<-Dx> debugging flag or the B<B::Terse> module, but it is more
657 sophisticated and flexible.
661 Here's is a short example of output, using the default formatting
664 % perl -MO=Concise -e '$a = $b + 42'
665 8 <@> leave[t1] vKP/REFC ->(end)
667 2 <;> nextstate(main 1 -e:1) v ->3
668 7 <2> sassign vKS/2 ->8
669 5 <2> add[t1] sK/2 ->6
670 - <1> ex-rv2sv sK/1 ->4
672 4 <$> const(IV 42) s ->5
673 - <1> ex-rv2sv sKRM*/1 ->7
676 Each line corresponds to an operator. Null ops appear as C<ex-opname>,
677 where I<opname> is the op that has been optimized away by perl.
679 The number on the first row indicates the op's sequence number. It's
680 given in base 36 by default.
682 The symbol between angle brackets indicates the op's type : for example,
683 <2> is a BINOP, <@> a LISTOP, etc. (see L</"OP class abbreviations">).
685 The opname may be followed by op-specific information in parentheses
686 (e.g. C<gvsv(*b)>), and by targ information in brackets (e.g.
689 Next come the op flags. The common flags are listed below
690 (L</"OP flags abbreviations">). The private flags follow, separated
691 by a slash. For example, C<vKP/REFC> means that the leave op has
692 public flags OPf_WANT_VOID, OPf_KIDS, and OPf_PARENS, and the private
695 Finally an arrow points to the sequence number of the next op.
699 Arguments that don't start with a hyphen are taken to be the names of
700 subroutines to print the OPs of; if no such functions are specified, the
701 main body of the program (outside any subroutines, and not including use'd
702 or require'd files) is printed.
708 Print OPs in the order they appear in the OP tree (a preorder
709 traversal, starting at the root). The indentation of each OP shows its
710 level in the tree. This mode is the default, so the flag is included
711 simply for completeness.
715 Print OPs in the order they would normally execute (for the majority
716 of constructs this is a postorder traversal of the tree, ending at the
717 root). In most cases the OP that usually follows a given OP will
718 appear directly below it; alternate paths are shown by indentation. In
719 cases like loops when control jumps out of a linear path, a 'goto'
724 Print OPs in a text approximation of a tree, with the root of the tree
725 at the left and 'left-to-right' order of children transformed into
726 'top-to-bottom'. Because this mode grows both to the right and down,
727 it isn't suitable for large programs (unless you have a very wide
732 Use a tree format in which the minimum amount of space is used for the
733 lines connecting nodes (one character in most cases). This squeezes out
734 a few precious columns of screen real estate.
738 Use a tree format that uses longer edges to separate OP nodes. This format
739 tends to look better than the compact one, especially in ASCII, and is
744 Use tree connecting characters drawn from the VT100 line-drawing set.
745 This looks better if your terminal supports it.
749 Draw the tree with standard ASCII characters like C<+> and C<|>. These don't
750 look as clean as the VT100 characters, but they'll work with almost any
751 terminal (or the horizontal scrolling mode of less(1)) and are suitable
752 for text documentation or email. This is the default.
756 Include the main program in the output, even if subroutines were also
761 Print OP sequence numbers in base I<n>. If I<n> is greater than 10, the
762 digit for 11 will be 'a', and so on. If I<n> is greater than 36, the digit
763 for 37 will be 'A', and so on until 62. Values greater than 62 are not
764 currently supported. The default is 36.
768 Print sequence numbers with the most significant digit first. This is the
769 usual convention for Arabic numerals, and the default.
771 =item B<-littleendian>
773 Print seqence numbers with the least significant digit first.
777 Use the author's favorite set of formatting conventions. This is the
782 Use formatting conventions that emulate the output of B<B::Terse>. The
783 basic mode is almost indistinguishable from the real B<B::Terse>, and the
784 exec mode looks very similar, but is in a more logical order and lacks
785 curly brackets. B<B::Terse> doesn't have a tree mode, so the tree mode
786 is only vaguely reminiscient of B<B::Terse>.
790 Use formatting conventions in which the name of each OP, rather than being
791 written out in full, is represented by a one- or two-character abbreviation.
792 This is mainly a joke.
796 Use formatting conventions reminiscient of B<B::Debug>; these aren't
801 Use formatting conventions read from the environment variables
802 C<B_CONCISE_FORMAT>, C<B_CONCISE_GOTO_FORMAT>, and C<B_CONCISE_TREE_FORMAT>.
806 =head1 FORMATTING SPECIFICATIONS
808 For each general style ('concise', 'terse', 'linenoise', etc.) there are
809 three specifications: one of how OPs should appear in the basic or exec
810 modes, one of how 'goto' lines should appear (these occur in the exec
811 mode only), and one of how nodes should appear in tree mode. Each has the
812 same format, described below. Any text that doesn't match a special
813 pattern is copied verbatim.
817 =item B<(x(>I<exec_text>B<;>I<basic_text>B<)x)>
819 Generates I<exec_text> in exec mode, or I<basic_text> in basic mode.
821 =item B<(*(>I<text>B<)*)>
823 Generates one copy of I<text> for each indentation level.
825 =item B<(*(>I<text1>B<;>I<text2>B<)*)>
827 Generates one fewer copies of I<text1> than the indentation level, followed
828 by one copy of I<text2> if the indentation level is more than 0.
830 =item B<(?(>I<text1>B<#>I<var>I<Text2>B<)?)>
832 If the value of I<var> is true (not empty or zero), generates the
833 value of I<var> surrounded by I<text1> and I<Text2>, otherwise
838 Generates the value of the variable I<var>.
842 Generates the value of I<var>, left jutified to fill I<N> spaces.
846 Any number of tildes and surrounding whitespace will be collapsed to
851 The following variables are recognized:
857 The address of the OP, in hexidecimal.
861 The OP-specific information of the OP (such as the SV for an SVOP, the
862 non-local exit pointers for a LOOP, etc.) enclosed in paretheses.
866 The B-determined class of the OP, in all caps.
870 A single symbol abbreviating the class of the OP.
874 The label of the statement or block the OP is the start of, if any.
878 The name of the OP, or 'ex-foo' if the OP is a null that used to be a foo.
882 The target of the OP, or nothing for a nulled OP.
886 The address of the OP's first child, in hexidecimal.
890 The OP's flags, abbreviated as a series of symbols.
894 The numeric value of the OP's flags.
898 The sequence number of the OP, or a hyphen if it doesn't have one.
902 'NEXT', 'LAST', or 'REDO' if the OP is a target of one of those in exec
903 mode, or empty otherwise.
907 The address of the OP's last child, in hexidecimal.
915 The OP's name, in all caps.
919 The sequence number of the OP's next OP.
923 The address of the OP's next OP, in hexidecimal.
927 A one- or two-character abbreviation for the OP's name.
931 The OP's private flags, rendered with abbreviated names if possible.
935 The numeric value of the OP's private flags.
939 The sequence number of the OP. Note that this is now a sequence number
940 generated by B::Concise, rather than the real op_seq value (for which
945 The real sequence number of the OP, as a regular number and not adjusted
946 to be relative to the start of the real program. (This will generally be
947 a fairly large number because all of B<B::Concise> is compiled before
952 The address of the OP's next youngest sibling, in hexidecimal.
956 The address of the OP's SV, if it has an SV, in hexidecimal.
960 The class of the OP's SV, if it has one, in all caps (e.g., 'IV').
964 The value of the OP's SV, if it has one, in a short human-readable format.
968 The numeric value of the OP's targ.
972 The name of the variable the OP's targ refers to, if any, otherwise the
973 letter t followed by the OP's targ in decimal.
975 =item B<#targarglife>
977 Same as B<#targarg>, but followed by the COP sequence numbers that delimit
978 the variable's lifetime (or 'end' for a variable in an open scope) for a
983 The numeric value of the OP's type, in decimal.
989 =head2 OP flags abbreviations
991 v OPf_WANT_VOID Want nothing (void context)
992 s OPf_WANT_SCALAR Want single value (scalar context)
993 l OPf_WANT_LIST Want list of any length (list context)
994 K OPf_KIDS There is a firstborn child.
995 P OPf_PARENS This operator was parenthesized.
996 (Or block needs explicit scope entry.)
997 R OPf_REF Certified reference.
998 (Return container, not containee).
999 M OPf_MOD Will modify (lvalue).
1000 S OPf_STACKED Some arg is arriving on the stack.
1001 * OPf_SPECIAL Do something weird for this op (see op.h)
1003 =head2 OP class abbreviations
1005 0 OP (aka BASEOP) An OP with no children
1006 1 UNOP An OP with one child
1007 2 BINOP An OP with two children
1008 | LOGOP A control branch OP
1009 @ LISTOP An OP that could have lots of children
1010 / PMOP An OP with a regular expression
1011 $ SVOP An OP with an SV
1012 " PVOP An OP with a string
1013 { LOOP An OP that holds pointers for a loop
1014 ; COP An OP that marks the start of a statement
1015 # PADOP An OP with a GV on the pad
1017 =head1 Using B::Concise outside of the O framework
1019 It is possible to extend B<B::Concise> by using it outside of the B<O>
1020 framework and providing new styles and new variables.
1022 use B::Concise qw(set_style add_callback);
1023 set_style($format, $gotofmt, $treefmt);
1028 my ($h, $op, $level, $format) = @_;
1029 $h->{variable} = some_func($op);
1032 B::Concise::compile(@options)->();
1034 You can specify a style by calling the B<set_style> subroutine. If you
1035 have a new variable in your style, or you want to change the value of an
1036 existing variable, you will need to add a callback to specify the value
1039 This is done by calling B<add_callback> passing references to any
1040 callback subroutines. The subroutines are called in the same order as
1041 they are added. Each subroutine is passed four parameters. These are a
1042 reference to a hash, the keys of which are the names of the variables
1043 and the values of which are their values, the op, the level and the
1046 To define your own variables, simply add them to the hash, or change
1047 existing values if you need to. The level and format are passed in as
1048 references to scalars, but it is unlikely that they will need to be
1049 changed or even used.
1051 To switch back to one of the standard styles like C<concise> or
1052 C<terse>, use C<set_style_standard>.
1054 To see the output, call the subroutine returned by B<compile> in the
1055 same way that B<O> does.
1059 Stephen McCamant, E<lt>smcc@CSUA.Berkeley.EDUE<gt>.