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 $VERSION = "0.57";
18 our @ISA = qw(Exporter);
19 our @EXPORT_OK = qw(set_style set_style_standard add_callback
20 concise_subref concise_cv concise_main);
23 use B qw(class ppname main_start main_root main_cv cstring svref_2object
24 SVf_IOK SVf_NOK SVf_POK SVf_IVisUV SVf_FAKE OPf_KIDS CVf_ANON);
28 ["(?(#label =>\n)?)(*( )*)#class (#addr) #name (?([#targ])?) "
29 . "#svclass~(?((#svaddr))?)~#svval~(?(label \"#coplabel\")?)\n",
30 "(*( )*)goto #class (#addr)\n",
33 ["#hyphseq2 (*( (x( ;)x))*)<#classsym> "
34 . "#exname#arg(?([#targarglife])?)~#flags(?(/#private)?)(x(;~->#next)x)\n",
35 " (*( )*) goto #seq\n",
36 "(?(<#seq>)?)#exname#arg(?([#targarglife])?)"],
38 ["(x(;(*( )*))x)#noise#arg(?([#targarg])?)(x( ;\n)x)",
40 "(?(#seq)?)#noise#arg(?([#targarg])?)"],
42 ["#class (#addr)\n\top_next\t\t#nextaddr\n\top_sibling\t#sibaddr\n\t"
43 . "op_ppaddr\tPL_ppaddr[OP_#NAME]\n\top_type\t\t#typenum\n\top_seq\t\t"
44 . "#seqnum\n\top_flags\t#flagval\n\top_private\t#privval\n"
45 . "(?(\top_first\t#firstaddr\n)?)(?(\top_last\t\t#lastaddr\n)?)"
46 . "(?(\top_sv\t\t#svaddr\n)?)",
49 "env" => [$ENV{B_CONCISE_FORMAT}, $ENV{B_CONCISE_GOTO_FORMAT},
50 $ENV{B_CONCISE_TREE_FORMAT}],
53 my($format, $gotofmt, $treefmt);
59 ($format, $gotofmt, $treefmt) = @_;
62 sub set_style_standard {
64 set_style(@{$style{$name}});
72 my($order, $subref) = @_;
73 concise_cv_obj($order, svref_2object($subref));
76 # This should have been called concise_subref, but it was exported
77 # under this name in versions before 0.56
78 sub concise_cv { concise_subref(@_); }
81 my ($order, $cv) = @_;
84 if ($order eq "exec") {
85 walk_exec($cv->START);
86 } elsif ($order eq "basic") {
87 walk_topdown($cv->ROOT, sub { $_[0]->concise($_[1]) }, 0);
89 print tree($cv->ROOT, 0)
97 if ($order eq "exec") {
98 return if class(main_start) eq "NULL";
99 walk_exec(main_start);
100 } elsif ($order eq "tree") {
101 return if class(main_root) eq "NULL";
102 print tree(main_root, 0);
103 } elsif ($order eq "basic") {
104 return if class(main_root) eq "NULL";
105 walk_topdown(main_root,
106 sub { $_[0]->concise($_[1]) }, 0);
110 sub concise_specials {
111 my($name, $order, @cv_s) = @_;
113 if ($name eq "BEGIN") {
114 splice(@cv_s, 0, 7); # skip 7 BEGIN blocks in this file
115 } elsif ($name eq "CHECK") {
116 pop @cv_s; # skip the CHECK block that calls us
121 concise_cv_obj($order, $cv);
125 my $start_sym = "\e(0"; # "\cN" sometimes also works
126 my $end_sym = "\e(B"; # "\cO" respectively
128 my @tree_decorations =
129 ([" ", "--", "+-", "|-", "| ", "`-", "-", 1],
130 [" ", "-", "+", "+", "|", "`", "", 0],
131 [" ", map("$start_sym$_$end_sym", "qq", "wq", "tq", "x ", "mq", "q"), 1],
132 [" ", map("$start_sym$_$end_sym", "q", "w", "t", "x", "m"), "", 0],
141 set_style_standard("concise");
144 my @options = grep(/^-/, @_);
145 my @args = grep(!/^-/, @_);
147 for my $o (@options) {
148 if ($o eq "-basic") {
150 } elsif ($o eq "-exec") {
152 } elsif ($o eq "-tree") {
154 } elsif ($o eq "-compact") {
156 } elsif ($o eq "-loose") {
158 } elsif ($o eq "-vt") {
160 } elsif ($o eq "-ascii") {
162 } elsif ($o eq "-main") {
164 } elsif ($o =~ /^-base(\d+)$/) {
166 } elsif ($o eq "-bigendian") {
168 } elsif ($o eq "-littleendian") {
170 } elsif (exists $style{substr($o, 1)}) {
171 set_style(@{$style{substr($o, 1)}});
173 warn "Option $o unrecognized";
178 for my $objname (@args) {
179 if ($objname eq "BEGIN") {
180 concise_specials("BEGIN", $order,
181 B::begin_av->isa("B::AV") ?
182 B::begin_av->ARRAY : ());
183 } elsif ($objname eq "INIT") {
184 concise_specials("INIT", $order,
185 B::init_av->isa("B::AV") ?
186 B::init_av->ARRAY : ());
187 } elsif ($objname eq "CHECK") {
188 concise_specials("CHECK", $order,
189 B::check_av->isa("B::AV") ?
190 B::check_av->ARRAY : ());
191 } elsif ($objname eq "END") {
192 concise_specials("END", $order,
193 B::end_av->isa("B::AV") ?
194 B::end_av->ARRAY : ());
196 $objname = "main::" . $objname unless $objname =~ /::/;
198 eval "concise_subref(\$order, \\&$objname)";
199 die "concise_subref($order, \\&$objname) failed: $@" if $@;
203 if (!@args or $do_main) {
204 print "main program:\n" if $do_main;
205 concise_main($order);
213 my %opclass = ('OP' => "0", 'UNOP' => "1", 'BINOP' => "2", 'LOGOP' => "|",
214 'LISTOP' => "@", 'PMOP' => "/", 'SVOP' => "\$", 'GVOP' => "*",
215 'PVOP' => '"', 'LOOP' => "{", 'COP' => ";", 'PADOP' => "#");
217 no warnings 'qw'; # "Possible attempt to put comments..."; use #7
219 qw'# () sc ( @? 1 $* gv *{ m$ m@ m% m? p/ *$ $ $# & a& pt \\ s\\ rf bl
220 ` *? <> ?? ?/ r/ c/ // qr s/ /c y/ = @= C sC Cp sp df un BM po +1 +I
221 -1 -I 1+ I+ 1- I- ** * i* / i/ %$ i% x + i+ - i- . " << >> < i<
222 > i> <= i, >= i. == i= != i! <? i? s< s> s, s. s= s! s? b& b^ b| -0 -i
223 ! ~ a2 si cs rd sr e^ lg sq in %x %o ab le ss ve ix ri sf FL od ch cy
224 uf lf uc lc qm @ [f [ @[ eh vl ky dl ex % ${ @{ uk pk st jn ) )[ a@
225 a% sl +] -] [- [+ so rv GS GW MS MW .. f. .f && || ^^ ?: &= |= -> s{ s}
226 v} ca wa di rs ;; ; ;d }{ { } {} f{ it {l l} rt }l }n }r dm }g }e ^o
227 ^c ^| ^# um bm t~ u~ ~d DB db ^s se ^g ^r {w }w pf pr ^O ^K ^R ^W ^d ^v
228 ^e ^t ^k t. fc ic fl .s .p .b .c .l .a .h g1 s1 g2 s2 ?. l? -R -W -X -r
229 -w -x -e -o -O -z -s -M -A -C -S -c -b -f -d -p -l -u -g -k -t -T -B cd
230 co cr u. cm ut r. l@ s@ r@ mD uD oD rD tD sD wD cD f$ w$ p$ sh e$ k$ g3
231 g4 s4 g5 s5 T@ C@ L@ G@ A@ S@ Hg Hc Hr Hw Mg Mc Ms Mr Sg Sc So rq do {e
232 e} {t t} g6 G6 6e g7 G7 7e g8 G8 8e g9 G9 9e 6s 7s 8s 9s 6E 7E 8E 9E Pn
233 Pu GP SP EP Gn Gg GG SG EG g0 c$ lk t$ ;s n> // /= CO';
235 my $chars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
240 push @v, "v" if ($x & 3) == 1;
241 push @v, "s" if ($x & 3) == 2;
242 push @v, "l" if ($x & 3) == 3;
243 push @v, "K" if $x & 4;
244 push @v, "P" if $x & 8;
245 push @v, "R" if $x & 16;
246 push @v, "M" if $x & 32;
247 push @v, "S" if $x & 64;
248 push @v, "*" if $x & 128;
254 return "-" . base_n(-$x) if $x < 0;
256 do { $str .= substr($chars, $x % $base, 1) } while $x = int($x / $base);
257 $str = reverse $str if $big_endian;
266 return "-" if not exists $sequence_num{$$op};
267 return base_n($sequence_num{$$op});
271 my($op, $sub, $level) = @_;
273 if ($op->flags & OPf_KIDS) {
274 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
275 walk_topdown($kid, $sub, $level + 1);
278 if (class($op) eq "PMOP") {
279 my $maybe_root = $op->pmreplroot;
280 if (ref($maybe_root) and $maybe_root->isa("B::OP")) {
281 # It really is the root of the replacement, not something
282 # else stored here for lack of space elsewhere
283 walk_topdown($maybe_root, $sub, $level + 1);
289 my($ar, $level) = @_;
291 if (ref($l) eq "ARRAY") {
292 walklines($l, $level + 1);
300 my($top, $level) = @_;
303 my @todo = ([$top, \@lines]);
304 while (@todo and my($op, $targ) = @{shift @todo}) {
305 for (; $$op; $op = $op->next) {
306 last if $opsseen{$$op}++;
308 my $name = $op->name;
309 if (class($op) eq "LOGOP") {
312 push @todo, [$op->other, $ar];
313 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
316 push @todo, [$op->pmreplstart, $ar];
317 } elsif ($name =~ /^enter(loop|iter)$/) {
318 $labels{$op->nextop->seq} = "NEXT";
319 $labels{$op->lastop->seq} = "LAST";
320 $labels{$op->redoop->seq} = "REDO";
324 walklines(\@lines, 0);
327 # The structure of this routine is purposely modeled after op.c's peep()
331 return if class($op) eq "NULL" or exists $sequence_num{$$op};
332 for (; $$op; $op = $op->next) {
333 last if exists $sequence_num{$$op};
334 my $name = $op->name;
335 if ($name =~ /^(null|scalar|lineseq|scope)$/) {
336 next if $oldop and $ {$op->next};
338 $sequence_num{$$op} = $seq_max++;
339 if (class($op) eq "LOGOP") {
340 my $other = $op->other;
341 $other = $other->next while $other->name eq "null";
343 } elsif (class($op) eq "LOOP") {
344 my $redoop = $op->redoop;
345 $redoop = $redoop->next while $redoop->name eq "null";
347 my $nextop = $op->nextop;
348 $nextop = $nextop->next while $nextop->name eq "null";
350 my $lastop = $op->lastop;
351 $lastop = $lastop->next while $lastop->name eq "null";
353 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
354 my $replstart = $op->pmreplstart;
355 $replstart = $replstart->next while $replstart->name eq "null";
356 sequence($replstart);
364 my($hr, $fmt, $level) = @_;
366 $text =~ s/\(\?\(([^\#]*?)\#(\w+)([^\#]*?)\)\?\)/
367 $hr->{$2} ? $1.$hr->{$2}.$3 : ""/eg;
368 $text =~ s/\(x\((.*?);(.*?)\)x\)/$order eq "exec" ? $1 : $2/egs;
369 $text =~ s/\(\*\(([^;]*?)\)\*\)/$1 x $level/egs;
370 $text =~ s/\(\*\((.*?);(.*?)\)\*\)/$1 x ($level - 1) . $2 x ($level>0)/egs;
371 $text =~ s/#([a-zA-Z]+)(\d+)/sprintf("%-$2s", $hr->{$1})/eg;
372 $text =~ s/#([a-zA-Z]+)/$hr->{$1}/eg;
373 $text =~ s/[ \t]*~+[ \t]*/ /g;
378 $priv{$_}{128} = "LVINTRO"
379 for ("pos", "substr", "vec", "threadsv", "gvsv", "rv2sv", "rv2hv", "rv2gv",
380 "rv2av", "rv2arylen", "aelem", "helem", "aslice", "hslice", "padsv",
381 "padav", "padhv", "enteriter");
382 $priv{$_}{64} = "REFC" for ("leave", "leavesub", "leavesublv", "leavewrite");
383 $priv{"aassign"}{64} = "COMMON";
384 $priv{"sassign"}{64} = "BKWARD";
385 $priv{$_}{64} = "RTIME" for ("match", "subst", "substcont");
386 @{$priv{"trans"}}{1,2,4,8,16,64} = ("<UTF", ">UTF", "IDENT", "SQUASH", "DEL",
388 $priv{"repeat"}{64} = "DOLIST";
389 $priv{"leaveloop"}{64} = "CONT";
390 @{$priv{$_}}{32,64,96} = ("DREFAV", "DREFHV", "DREFSV")
391 for ("entersub", map("rv2${_}v", "a", "s", "h", "g"), "aelem", "helem");
392 $priv{"entersub"}{16} = "DBG";
393 $priv{"entersub"}{32} = "TARG";
394 @{$priv{$_}}{4,8,128} = ("INARGS","AMPER","NO()") for ("entersub", "rv2cv");
395 $priv{"gv"}{32} = "EARLYCV";
396 $priv{"aelem"}{16} = $priv{"helem"}{16} = "LVDEFER";
397 $priv{$_}{16} = "OURINTR" for ("gvsv", "rv2sv", "rv2av", "rv2hv", "r2gv",
399 $priv{$_}{16} = "TARGMY"
400 for (map(($_,"s$_"),"chop", "chomp"),
401 map(($_,"i_$_"), "postinc", "postdec", "multiply", "divide", "modulo",
402 "add", "subtract", "negate"), "pow", "concat", "stringify",
403 "left_shift", "right_shift", "bit_and", "bit_xor", "bit_or",
404 "complement", "atan2", "sin", "cos", "rand", "exp", "log", "sqrt",
405 "int", "hex", "oct", "abs", "length", "index", "rindex", "sprintf",
406 "ord", "chr", "crypt", "quotemeta", "join", "push", "unshift", "flock",
407 "chdir", "chown", "chroot", "unlink", "chmod", "utime", "rename",
408 "link", "symlink", "mkdir", "rmdir", "wait", "waitpid", "system",
409 "exec", "kill", "getppid", "getpgrp", "setpgrp", "getpriority",
410 "setpriority", "time", "sleep");
411 @{$priv{"const"}}{8,16,32,64,128} = ("STRICT","ENTERED", '$[', "BARE", "WARN");
412 $priv{"flip"}{64} = $priv{"flop"}{64} = "LINENUM";
413 $priv{"list"}{64} = "GUESSED";
414 $priv{"delete"}{64} = "SLICE";
415 $priv{"exists"}{64} = "SUB";
416 $priv{$_}{64} = "LOCALE"
417 for ("sort", "prtf", "sprintf", "slt", "sle", "seq", "sne", "sgt", "sge",
418 "scmp", "lc", "uc", "lcfirst", "ucfirst");
419 @{$priv{"sort"}}{1,2,4} = ("NUM", "INT", "REV");
420 $priv{"threadsv"}{64} = "SVREFd";
421 @{$priv{$_}}{16,32,64,128} = ("INBIN","INCR","OUTBIN","OUTCR")
422 for ("open", "backtick");
423 $priv{"exit"}{128} = "VMS";
424 $priv{$_}{2} = "FTACCESS"
425 for ("ftrread", "ftrwrite", "ftrexec", "fteread", "ftewrite", "fteexec");
430 for my $flag (128, 96, 64, 32, 16, 8, 4, 2, 1) {
431 if ($priv{$name}{$flag} and $x & $flag and $x >= $flag) {
433 push @s, $priv{$name}{$flag};
437 return join(",", @s);
442 $hr->{svclass} = class($sv);
443 $hr->{svclass} = "UV"
444 if $hr->{svclass} eq "IV" and $sv->FLAGS & SVf_IVisUV;
445 $hr->{svaddr} = sprintf("%#x", $$sv);
446 if ($hr->{svclass} eq "GV") {
448 my $stash = $gv->STASH->NAME;
449 if ($stash eq "main") {
452 $stash = $stash . "::";
454 $hr->{svval} = "*$stash" . $gv->SAFENAME;
455 return "*$stash" . $gv->SAFENAME;
457 while (class($sv) eq "RV") {
458 $hr->{svval} .= "\\";
461 if (class($sv) eq "SPECIAL") {
462 $hr->{svval} .= ["Null", "sv_undef", "sv_yes", "sv_no"]->[$$sv];
463 } elsif ($sv->FLAGS & SVf_NOK) {
464 $hr->{svval} .= $sv->NV;
465 } elsif ($sv->FLAGS & SVf_IOK) {
466 $hr->{svval} .= $sv->int_value;
467 } elsif ($sv->FLAGS & SVf_POK) {
468 $hr->{svval} .= cstring($sv->PV);
469 } elsif (class($sv) eq "HV") {
470 $hr->{svval} .= 'HASH';
472 return $hr->{svclass} . " " . $hr->{svval};
477 my ($op, $level, $format) = @_;
479 $h{exname} = $h{name} = $op->name;
480 $h{NAME} = uc $h{name};
481 $h{class} = class($op);
482 $h{extarg} = $h{targ} = $op->targ;
483 $h{extarg} = "" unless $h{extarg};
484 if ($h{name} eq "null" and $h{targ}) {
485 # targ holds the old type
486 $h{exname} = "ex-" . substr(ppname($h{targ}), 3);
488 } elsif ($op->name =~ /^leave(sub(lv)?|write)?$/) {
489 # targ potentially holds a reference count
490 if ($op->private & 64) {
491 my $refs = "ref" . ($h{targ} != 1 ? "s" : "");
492 $h{targarglife} = $h{targarg} = "$h{targ} $refs";
495 my $padname = (($curcv->PADLIST->ARRAY)[0]->ARRAY)[$h{targ}];
496 if (defined $padname and class($padname) ne "SPECIAL") {
497 $h{targarg} = $padname->PVX;
498 if ($padname->FLAGS & SVf_FAKE) {
500 $fake .= 'a' if $padname->IVX & 1; # PAD_FAKELEX_ANON
501 $fake .= 'm' if $padname->IVX & 2; # PAD_FAKELEX_MULTI
502 $fake .= ':' . $padname->NVX if $curcv->CvFLAGS & CVf_ANON;
503 $h{targarglife} = "$h{targarg}:FAKE:$fake";
506 my $intro = $padname->NVX - $cop_seq_base;
507 my $finish = int($padname->IVX) - $cop_seq_base;
508 $finish = "end" if $finish == 999999999 - $cop_seq_base;
509 $h{targarglife} = "$h{targarg}:$intro,$finish";
512 $h{targarglife} = $h{targarg} = "t" . $h{targ};
516 $h{svclass} = $h{svaddr} = $h{svval} = "";
517 if ($h{class} eq "PMOP") {
518 my $precomp = $op->precomp;
519 if (defined $precomp) {
520 $precomp = cstring($precomp); # Escape literal control sequences
521 $precomp = "/$precomp/";
525 my $pmreplroot = $op->pmreplroot;
527 if (ref($pmreplroot) eq "B::GV") {
528 # with C<@stash_array = split(/pat/, str);>,
529 # *stash_array is stored in /pat/'s pmreplroot.
530 $h{arg} = "($precomp => \@" . $pmreplroot->NAME . ")";
531 } elsif (!ref($pmreplroot) and $pmreplroot) {
532 # same as the last case, except the value is actually a
533 # pad offset for where the GV is kept (this happens under
535 my $gv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$pmreplroot];
536 $h{arg} = "($precomp => \@" . $gv->NAME . ")";
537 } elsif ($ {$op->pmreplstart}) {
539 $pmreplstart = "replstart->" . seq($op->pmreplstart);
540 $h{arg} = "(" . join(" ", $precomp, $pmreplstart) . ")";
542 $h{arg} = "($precomp)";
544 } elsif ($h{class} eq "PVOP" and $h{name} ne "trans") {
545 $h{arg} = '("' . $op->pv . '")';
546 $h{svval} = '"' . $op->pv . '"';
547 } elsif ($h{class} eq "COP") {
548 my $label = $op->label;
549 $h{coplabel} = $label;
550 $label = $label ? "$label: " : "";
553 $loc .= ":" . $op->line;
554 my($stash, $cseq) = ($op->stash->NAME, $op->cop_seq - $cop_seq_base);
555 my $arybase = $op->arybase;
556 $arybase = $arybase ? ' $[=' . $arybase : "";
557 $h{arg} = "($label$stash $cseq $loc$arybase)";
558 } elsif ($h{class} eq "LOOP") {
559 $h{arg} = "(next->" . seq($op->nextop) . " last->" . seq($op->lastop)
560 . " redo->" . seq($op->redoop) . ")";
561 } elsif ($h{class} eq "LOGOP") {
563 $h{arg} = "(other->" . seq($op->other) . ")";
564 } elsif ($h{class} eq "SVOP") {
566 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->targ];
567 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
568 $h{targarglife} = $h{targarg} = "";
570 $h{arg} = "(" . concise_sv($op->sv, \%h) . ")";
572 } elsif ($h{class} eq "PADOP") {
573 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->padix];
574 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
576 $h{seq} = $h{hyphseq} = seq($op);
577 $h{seq} = "" if $h{seq} eq "-";
578 $h{seqnum} = $op->seq;
579 $h{next} = $op->next;
580 $h{next} = (class($h{next}) eq "NULL") ? "(end)" : seq($h{next});
581 $h{nextaddr} = sprintf("%#x", $ {$op->next});
582 $h{sibaddr} = sprintf("%#x", $ {$op->sibling});
583 $h{firstaddr} = sprintf("%#x", $ {$op->first}) if $op->can("first");
584 $h{lastaddr} = sprintf("%#x", $ {$op->last}) if $op->can("last");
586 $h{classsym} = $opclass{$h{class}};
587 $h{flagval} = $op->flags;
588 $h{flags} = op_flags($op->flags);
589 $h{privval} = $op->private;
590 $h{private} = private_flags($h{name}, $op->private);
591 $h{addr} = sprintf("%#x", $$op);
592 $h{label} = $labels{$op->seq};
593 $h{typenum} = $op->type;
594 $h{noise} = $linenoise[$op->type];
595 $_->(\%h, $op, \$format, \$level) for @callbacks;
596 return fmt_line(\%h, $format, $level);
600 my($op, $level) = @_;
601 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
602 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
603 "addr" => sprintf("%#x", $$lastnext)};
604 print fmt_line($h, $gotofmt, $level+1);
606 $lastnext = $op->next;
607 print concise_op($op, $level, $format);
610 # B::OP::terse (see Terse.pm) now just calls this
612 my($op, $level) = @_;
614 # This isn't necessarily right, but there's no easy way to get
615 # from an OP to the right CV. This is a limitation of the
616 # ->terse() interface style, and there isn't much to do about
617 # it. In particular, we can die in concise_op if the main pad
618 # isn't long enough, or has the wrong kind of entries, compared to
619 # the pad a sub was compiled with. The fix for that would be to
620 # make a backwards compatible "terse" format that never even
621 # looked at the pad, just like the old B::Terse. I don't think
622 # that's worth the effort, though.
623 $curcv = main_cv unless $curcv;
625 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
626 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
627 "addr" => sprintf("%#x", $$lastnext)};
628 print fmt_line($h, $style{"terse"}[1], $level+1);
630 $lastnext = $op->next;
631 print concise_op($op, $level, $style{"terse"}[0]);
637 my $style = $tree_decorations[$tree_style];
638 my($space, $single, $kids, $kid, $nokid, $last, $lead, $size) = @$style;
639 my $name = concise_op($op, $level, $treefmt);
640 if (not $op->flags & OPf_KIDS) {
644 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
645 push @lines, tree($kid, $level+1);
648 for ($i = $#lines; substr($lines[$i], 0, 1) eq " "; $i--) {
649 $lines[$i] = $space . $lines[$i];
652 $lines[$i] = $last . $lines[$i];
654 if (substr($lines[$i], 0, 1) eq " ") {
655 $lines[$i] = $nokid . $lines[$i];
657 $lines[$i] = $kid . $lines[$i];
660 $lines[$i] = $kids . $lines[$i];
662 $lines[0] = $single . $lines[0];
664 return("$name$lead" . shift @lines,
665 map(" " x (length($name)+$size) . $_, @lines));
668 # *** Warning: fragile kludge ahead ***
669 # Because the B::* modules run in the same interpreter as the code
670 # they're compiling, their presence tends to distort the view we have
671 # of the code we're looking at. In particular, perl gives sequence
672 # numbers to both OPs in general and COPs in particular. If the
673 # program we're looking at were run on its own, these numbers would
674 # start at 1. Because all of B::Concise and all the modules it uses
675 # are compiled first, though, by the time we get to the user's program
676 # the sequence numbers are alreay at pretty high numbers, which would
677 # be distracting if you're trying to tell OPs apart. Therefore we'd
678 # like to subtract an offset from all the sequence numbers we display,
679 # to restore the simpler view of the world. The trick is to know what
680 # that offset will be, when we're still compiling B::Concise! If we
681 # hardcoded a value, it would have to change every time B::Concise or
682 # other modules we use do. To help a little, what we do here is
683 # compile a little code at the end of the module, and compute the base
684 # sequence number for the user's program as being a small offset
685 # later, so all we have to worry about are changes in the offset.
686 # (Note that we now only play this game with COP sequence numbers. OP
687 # sequence numbers aren't used to refer to OPs from a distance, and
688 # they don't have much significance, so we just generate our own
689 # sequence numbers which are easier to control. This way we also don't
690 # stand in the way of a possible future removal of OP sequence
693 # When you say "perl -MO=Concise -e '$a'", the output should look like:
695 # 4 <@> leave[t1] vKP/REFC ->(end)
697 #^ smallest OP sequence number should be 1
698 # 2 <;> nextstate(main 1 -e:1) v ->3
699 # ^ smallest COP sequence number should be 1
700 # - <1> ex-rv2sv vK/1 ->4
701 # 3 <$> gvsv(*a) s ->4
703 # If the second of the marked numbers there isn't 1, it means you need
704 # to update the corresponding magic number in the next line.
705 # Remember, this needs to stay the last things in the module.
707 # Why is this different for MacOS? Does it matter?
708 my $cop_seq_mnum = $^O eq 'MacOS' ? 12 : 11;
709 $cop_seq_base = svref_2object(eval 'sub{0;}')->START->cop_seq + $cop_seq_mnum;
717 B::Concise - Walk Perl syntax tree, printing concise info about ops
721 perl -MO=Concise[,OPTIONS] foo.pl
723 use B::Concise qw(set_style add_callback);
727 This compiler backend prints the internal OPs of a Perl program's syntax
728 tree in one of several space-efficient text formats suitable for debugging
729 the inner workings of perl or other compiler backends. It can print OPs in
730 the order they appear in the OP tree, in the order they will execute, or
731 in a text approximation to their tree structure, and the format of the
732 information displyed is customizable. Its function is similar to that of
733 perl's B<-Dx> debugging flag or the B<B::Terse> module, but it is more
734 sophisticated and flexible.
738 Here's is a short example of output, using the default formatting
741 % perl -MO=Concise -e '$a = $b + 42'
742 8 <@> leave[1 ref] vKP/REFC ->(end)
744 2 <;> nextstate(main 1 -e:1) v ->3
745 7 <2> sassign vKS/2 ->8
746 5 <2> add[t1] sK/2 ->6
747 - <1> ex-rv2sv sK/1 ->4
749 4 <$> const(IV 42) s ->5
750 - <1> ex-rv2sv sKRM*/1 ->7
753 Each line corresponds to an operator. Null ops appear as C<ex-opname>,
754 where I<opname> is the op that has been optimized away by perl.
756 The number on the first row indicates the op's sequence number. It's
757 given in base 36 by default.
759 The symbol between angle brackets indicates the op's type : for example,
760 <2> is a BINOP, <@> a LISTOP, etc. (see L</"OP class abbreviations">).
762 The opname may be followed by op-specific information in parentheses
763 (e.g. C<gvsv(*b)>), and by targ information in brackets (e.g.
766 Next come the op flags. The common flags are listed below
767 (L</"OP flags abbreviations">). The private flags follow, separated
768 by a slash. For example, C<vKP/REFC> means that the leave op has
769 public flags OPf_WANT_VOID, OPf_KIDS, and OPf_PARENS, and the private
772 Finally an arrow points to the sequence number of the next op.
776 Arguments that don't start with a hyphen are taken to be the names of
777 subroutines to print the OPs of; if no such functions are specified,
778 the main body of the program (outside any subroutines, and not
779 including use'd or require'd files) is printed. Passing C<BEGIN>,
780 C<CHECK>, C<INIT>, or C<END> will cause all of the corresponding
781 special blocks to be printed.
787 Print OPs in the order they appear in the OP tree (a preorder
788 traversal, starting at the root). The indentation of each OP shows its
789 level in the tree. This mode is the default, so the flag is included
790 simply for completeness.
794 Print OPs in the order they would normally execute (for the majority
795 of constructs this is a postorder traversal of the tree, ending at the
796 root). In most cases the OP that usually follows a given OP will
797 appear directly below it; alternate paths are shown by indentation. In
798 cases like loops when control jumps out of a linear path, a 'goto'
803 Print OPs in a text approximation of a tree, with the root of the tree
804 at the left and 'left-to-right' order of children transformed into
805 'top-to-bottom'. Because this mode grows both to the right and down,
806 it isn't suitable for large programs (unless you have a very wide
811 Use a tree format in which the minimum amount of space is used for the
812 lines connecting nodes (one character in most cases). This squeezes out
813 a few precious columns of screen real estate.
817 Use a tree format that uses longer edges to separate OP nodes. This format
818 tends to look better than the compact one, especially in ASCII, and is
823 Use tree connecting characters drawn from the VT100 line-drawing set.
824 This looks better if your terminal supports it.
828 Draw the tree with standard ASCII characters like C<+> and C<|>. These don't
829 look as clean as the VT100 characters, but they'll work with almost any
830 terminal (or the horizontal scrolling mode of less(1)) and are suitable
831 for text documentation or email. This is the default.
835 Include the main program in the output, even if subroutines were also
840 Print OP sequence numbers in base I<n>. If I<n> is greater than 10, the
841 digit for 11 will be 'a', and so on. If I<n> is greater than 36, the digit
842 for 37 will be 'A', and so on until 62. Values greater than 62 are not
843 currently supported. The default is 36.
847 Print sequence numbers with the most significant digit first. This is the
848 usual convention for Arabic numerals, and the default.
850 =item B<-littleendian>
852 Print seqence numbers with the least significant digit first.
856 Use the author's favorite set of formatting conventions. This is the
861 Use formatting conventions that emulate the output of B<B::Terse>. The
862 basic mode is almost indistinguishable from the real B<B::Terse>, and the
863 exec mode looks very similar, but is in a more logical order and lacks
864 curly brackets. B<B::Terse> doesn't have a tree mode, so the tree mode
865 is only vaguely reminiscient of B<B::Terse>.
869 Use formatting conventions in which the name of each OP, rather than being
870 written out in full, is represented by a one- or two-character abbreviation.
871 This is mainly a joke.
875 Use formatting conventions reminiscient of B<B::Debug>; these aren't
880 Use formatting conventions read from the environment variables
881 C<B_CONCISE_FORMAT>, C<B_CONCISE_GOTO_FORMAT>, and C<B_CONCISE_TREE_FORMAT>.
885 =head1 FORMATTING SPECIFICATIONS
887 For each general style ('concise', 'terse', 'linenoise', etc.) there are
888 three specifications: one of how OPs should appear in the basic or exec
889 modes, one of how 'goto' lines should appear (these occur in the exec
890 mode only), and one of how nodes should appear in tree mode. Each has the
891 same format, described below. Any text that doesn't match a special
892 pattern is copied verbatim.
896 =item B<(x(>I<exec_text>B<;>I<basic_text>B<)x)>
898 Generates I<exec_text> in exec mode, or I<basic_text> in basic mode.
900 =item B<(*(>I<text>B<)*)>
902 Generates one copy of I<text> for each indentation level.
904 =item B<(*(>I<text1>B<;>I<text2>B<)*)>
906 Generates one fewer copies of I<text1> than the indentation level, followed
907 by one copy of I<text2> if the indentation level is more than 0.
909 =item B<(?(>I<text1>B<#>I<var>I<Text2>B<)?)>
911 If the value of I<var> is true (not empty or zero), generates the
912 value of I<var> surrounded by I<text1> and I<Text2>, otherwise
917 Generates the value of the variable I<var>.
921 Generates the value of I<var>, left jutified to fill I<N> spaces.
925 Any number of tildes and surrounding whitespace will be collapsed to
930 The following variables are recognized:
936 The address of the OP, in hexidecimal.
940 The OP-specific information of the OP (such as the SV for an SVOP, the
941 non-local exit pointers for a LOOP, etc.) enclosed in paretheses.
945 The B-determined class of the OP, in all caps.
949 A single symbol abbreviating the class of the OP.
953 The label of the statement or block the OP is the start of, if any.
957 The name of the OP, or 'ex-foo' if the OP is a null that used to be a foo.
961 The target of the OP, or nothing for a nulled OP.
965 The address of the OP's first child, in hexidecimal.
969 The OP's flags, abbreviated as a series of symbols.
973 The numeric value of the OP's flags.
977 The sequence number of the OP, or a hyphen if it doesn't have one.
981 'NEXT', 'LAST', or 'REDO' if the OP is a target of one of those in exec
982 mode, or empty otherwise.
986 The address of the OP's last child, in hexidecimal.
994 The OP's name, in all caps.
998 The sequence number of the OP's next OP.
1002 The address of the OP's next OP, in hexidecimal.
1006 A one- or two-character abbreviation for the OP's name.
1010 The OP's private flags, rendered with abbreviated names if possible.
1014 The numeric value of the OP's private flags.
1018 The sequence number of the OP. Note that this is now a sequence number
1019 generated by B::Concise, rather than the real op_seq value (for which
1024 The real sequence number of the OP, as a regular number and not adjusted
1025 to be relative to the start of the real program. (This will generally be
1026 a fairly large number because all of B<B::Concise> is compiled before
1031 The address of the OP's next youngest sibling, in hexidecimal.
1035 The address of the OP's SV, if it has an SV, in hexidecimal.
1039 The class of the OP's SV, if it has one, in all caps (e.g., 'IV').
1043 The value of the OP's SV, if it has one, in a short human-readable format.
1047 The numeric value of the OP's targ.
1051 The name of the variable the OP's targ refers to, if any, otherwise the
1052 letter t followed by the OP's targ in decimal.
1054 =item B<#targarglife>
1056 Same as B<#targarg>, but followed by the COP sequence numbers that delimit
1057 the variable's lifetime (or 'end' for a variable in an open scope) for a
1062 The numeric value of the OP's type, in decimal.
1066 =head1 ABBREVIATIONS
1068 =head2 OP flags abbreviations
1070 v OPf_WANT_VOID Want nothing (void context)
1071 s OPf_WANT_SCALAR Want single value (scalar context)
1072 l OPf_WANT_LIST Want list of any length (list context)
1073 K OPf_KIDS There is a firstborn child.
1074 P OPf_PARENS This operator was parenthesized.
1075 (Or block needs explicit scope entry.)
1076 R OPf_REF Certified reference.
1077 (Return container, not containee).
1078 M OPf_MOD Will modify (lvalue).
1079 S OPf_STACKED Some arg is arriving on the stack.
1080 * OPf_SPECIAL Do something weird for this op (see op.h)
1082 =head2 OP class abbreviations
1084 0 OP (aka BASEOP) An OP with no children
1085 1 UNOP An OP with one child
1086 2 BINOP An OP with two children
1087 | LOGOP A control branch OP
1088 @ LISTOP An OP that could have lots of children
1089 / PMOP An OP with a regular expression
1090 $ SVOP An OP with an SV
1091 " PVOP An OP with a string
1092 { LOOP An OP that holds pointers for a loop
1093 ; COP An OP that marks the start of a statement
1094 # PADOP An OP with a GV on the pad
1096 =head1 Using B::Concise outside of the O framework
1098 It is possible to extend B<B::Concise> by using it outside of the B<O>
1099 framework and providing new styles and new variables.
1101 use B::Concise qw(set_style add_callback);
1102 set_style($format, $gotofmt, $treefmt);
1107 my ($h, $op, $level, $format) = @_;
1108 $h->{variable} = some_func($op);
1111 B::Concise::compile(@options)->();
1113 You can specify a style by calling the B<set_style> subroutine. If you
1114 have a new variable in your style, or you want to change the value of an
1115 existing variable, you will need to add a callback to specify the value
1118 This is done by calling B<add_callback> passing references to any
1119 callback subroutines. The subroutines are called in the same order as
1120 they are added. Each subroutine is passed four parameters. These are a
1121 reference to a hash, the keys of which are the names of the variables
1122 and the values of which are their values, the op, the level and the
1125 To define your own variables, simply add them to the hash, or change
1126 existing values if you need to. The level and format are passed in as
1127 references to scalars, but it is unlikely that they will need to be
1128 changed or even used.
1130 To switch back to one of the standard styles like C<concise> or
1131 C<terse>, use C<set_style_standard>.
1133 To see the output, call the subroutine returned by B<compile> in the
1134 same way that B<O> does.
1138 Stephen McCamant, E<lt>smcc@CSUA.Berkeley.EDUE<gt>.