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 $VERSION = "0.59";
18 our @ISA = qw(Exporter);
19 our @EXPORT_OK = qw(set_style set_style_standard add_callback
20 concise_subref concise_cv concise_main);
23 use B qw(class ppname main_start main_root main_cv cstring svref_2object
24 SVf_IOK SVf_NOK SVf_POK SVf_IVisUV SVf_FAKE OPf_KIDS CVf_ANON);
28 ["(?(#label =>\n)?)(*( )*)#class (#addr) #name (?([#targ])?) "
29 . "#svclass~(?((#svaddr))?)~#svval~(?(label \"#coplabel\")?)\n",
30 "(*( )*)goto #class (#addr)\n",
33 ["#hyphseq2 (*( (x( ;)x))*)<#classsym> "
34 . "#exname#arg(?([#targarglife])?)~#flags(?(/#private)?)(x(;~->#next)x)\n",
35 " (*( )*) goto #seq\n",
36 "(?(<#seq>)?)#exname#arg(?([#targarglife])?)"],
38 ["(x(;(*( )*))x)#noise#arg(?([#targarg])?)(x( ;\n)x)",
40 "(?(#seq)?)#noise#arg(?([#targarg])?)"],
42 ["#class (#addr)\n\top_next\t\t#nextaddr\n\top_sibling\t#sibaddr\n\t"
43 . "op_ppaddr\tPL_ppaddr[OP_#NAME]\n\top_type\t\t#typenum\n"
44 . "\top_flags\t#flagval\n\top_private\t#privval\n"
45 . "(?(\top_first\t#firstaddr\n)?)(?(\top_last\t\t#lastaddr\n)?)"
46 . "(?(\top_sv\t\t#svaddr\n)?)",
49 "env" => [$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{$$op}++;
308 my $name = $op->name;
309 if (class($op) eq "LOGOP") {
312 push @todo, [$op->other, $ar];
313 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
316 push @todo, [$op->pmreplstart, $ar];
317 } elsif ($name =~ /^enter(loop|iter)$/) {
318 $labels{${$op->nextop}} = "NEXT";
319 $labels{${$op->lastop}} = "LAST";
320 $labels{${$op->redoop}} = "REDO";
324 walklines(\@lines, 0);
327 # The structure of this routine is purposely modeled after op.c's peep()
331 return if class($op) eq "NULL" or exists $sequence_num{$$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"}}{1,2,4,8} = ("NUM", "INT", "REV", "INPLACE");
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");
426 $priv{$_}{4} = "FTSTACKED"
427 for ("ftrread", "ftrwrite", "ftrexec", "fteread", "ftewrite", "fteexec",
428 "ftis", "fteowned", "ftrowned", "ftzero", "ftsize", "ftmtime",
429 "ftatime", "ftctime", "ftsock", "ftchr", "ftblk", "ftfile", "ftdir",
430 "ftpipe", "ftlink", "ftsuid", "ftsgid", "ftsvtx", "fttty", "fttext",
432 $priv{$_}{2} = "GREPLEX"
433 for ("mapwhile", "mapstart", "grepwhile", "grepstart");
438 for my $flag (128, 96, 64, 32, 16, 8, 4, 2, 1) {
439 if ($priv{$name}{$flag} and $x & $flag and $x >= $flag) {
441 push @s, $priv{$name}{$flag};
445 return join(",", @s);
450 $hr->{svclass} = class($sv);
451 $hr->{svclass} = "UV"
452 if $hr->{svclass} eq "IV" and $sv->FLAGS & SVf_IVisUV;
453 $hr->{svaddr} = sprintf("%#x", $$sv);
454 if ($hr->{svclass} eq "GV") {
456 my $stash = $gv->STASH->NAME;
457 if ($stash eq "main") {
460 $stash = $stash . "::";
462 $hr->{svval} = "*$stash" . $gv->SAFENAME;
463 return "*$stash" . $gv->SAFENAME;
465 while (class($sv) eq "RV") {
466 $hr->{svval} .= "\\";
469 if (class($sv) eq "SPECIAL") {
470 $hr->{svval} .= ["Null", "sv_undef", "sv_yes", "sv_no"]->[$$sv];
471 } elsif ($sv->FLAGS & SVf_NOK) {
472 $hr->{svval} .= $sv->NV;
473 } elsif ($sv->FLAGS & SVf_IOK) {
474 $hr->{svval} .= $sv->int_value;
475 } elsif ($sv->FLAGS & SVf_POK) {
476 $hr->{svval} .= cstring($sv->PV);
477 } elsif (class($sv) eq "HV") {
478 $hr->{svval} .= 'HASH';
480 return $hr->{svclass} . " " . $hr->{svval};
485 my ($op, $level, $format) = @_;
487 $h{exname} = $h{name} = $op->name;
488 $h{NAME} = uc $h{name};
489 $h{class} = class($op);
490 $h{extarg} = $h{targ} = $op->targ;
491 $h{extarg} = "" unless $h{extarg};
492 if ($h{name} eq "null" and $h{targ}) {
493 # targ holds the old type
494 $h{exname} = "ex-" . substr(ppname($h{targ}), 3);
496 } elsif ($op->name =~ /^leave(sub(lv)?|write)?$/) {
497 # targ potentially holds a reference count
498 if ($op->private & 64) {
499 my $refs = "ref" . ($h{targ} != 1 ? "s" : "");
500 $h{targarglife} = $h{targarg} = "$h{targ} $refs";
503 my $padname = (($curcv->PADLIST->ARRAY)[0]->ARRAY)[$h{targ}];
504 if (defined $padname and class($padname) ne "SPECIAL") {
505 $h{targarg} = $padname->PVX;
506 if ($padname->FLAGS & SVf_FAKE) {
508 $fake .= 'a' if $padname->IVX & 1; # PAD_FAKELEX_ANON
509 $fake .= 'm' if $padname->IVX & 2; # PAD_FAKELEX_MULTI
510 $fake .= ':' . $padname->NVX if $curcv->CvFLAGS & CVf_ANON;
511 $h{targarglife} = "$h{targarg}:FAKE:$fake";
514 my $intro = $padname->NVX - $cop_seq_base;
515 my $finish = int($padname->IVX) - $cop_seq_base;
516 $finish = "end" if $finish == 999999999 - $cop_seq_base;
517 $h{targarglife} = "$h{targarg}:$intro,$finish";
520 $h{targarglife} = $h{targarg} = "t" . $h{targ};
524 $h{svclass} = $h{svaddr} = $h{svval} = "";
525 if ($h{class} eq "PMOP") {
526 my $precomp = $op->precomp;
527 if (defined $precomp) {
528 $precomp = cstring($precomp); # Escape literal control sequences
529 $precomp = "/$precomp/";
533 my $pmreplroot = $op->pmreplroot;
535 if (ref($pmreplroot) eq "B::GV") {
536 # with C<@stash_array = split(/pat/, str);>,
537 # *stash_array is stored in /pat/'s pmreplroot.
538 $h{arg} = "($precomp => \@" . $pmreplroot->NAME . ")";
539 } elsif (!ref($pmreplroot) and $pmreplroot) {
540 # same as the last case, except the value is actually a
541 # pad offset for where the GV is kept (this happens under
543 my $gv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$pmreplroot];
544 $h{arg} = "($precomp => \@" . $gv->NAME . ")";
545 } elsif ($ {$op->pmreplstart}) {
547 $pmreplstart = "replstart->" . seq($op->pmreplstart);
548 $h{arg} = "(" . join(" ", $precomp, $pmreplstart) . ")";
550 $h{arg} = "($precomp)";
552 } elsif ($h{class} eq "PVOP" and $h{name} ne "trans") {
553 $h{arg} = '("' . $op->pv . '")';
554 $h{svval} = '"' . $op->pv . '"';
555 } elsif ($h{class} eq "COP") {
556 my $label = $op->label;
557 $h{coplabel} = $label;
558 $label = $label ? "$label: " : "";
561 $loc .= ":" . $op->line;
562 my($stash, $cseq) = ($op->stash->NAME, $op->cop_seq - $cop_seq_base);
563 my $arybase = $op->arybase;
564 $arybase = $arybase ? ' $[=' . $arybase : "";
565 $h{arg} = "($label$stash $cseq $loc$arybase)";
566 } elsif ($h{class} eq "LOOP") {
567 $h{arg} = "(next->" . seq($op->nextop) . " last->" . seq($op->lastop)
568 . " redo->" . seq($op->redoop) . ")";
569 } elsif ($h{class} eq "LOGOP") {
571 $h{arg} = "(other->" . seq($op->other) . ")";
572 } elsif ($h{class} eq "SVOP") {
574 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->targ];
575 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
576 $h{targarglife} = $h{targarg} = "";
578 $h{arg} = "(" . concise_sv($op->sv, \%h) . ")";
580 } elsif ($h{class} eq "PADOP") {
581 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->padix];
582 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
584 $h{seq} = $h{hyphseq} = seq($op);
585 $h{seq} = "" if $h{seq} eq "-";
587 $h{static} = $op->static;
588 $h{next} = $op->next;
589 $h{next} = (class($h{next}) eq "NULL") ? "(end)" : seq($h{next});
590 $h{nextaddr} = sprintf("%#x", $ {$op->next});
591 $h{sibaddr} = sprintf("%#x", $ {$op->sibling});
592 $h{firstaddr} = sprintf("%#x", $ {$op->first}) if $op->can("first");
593 $h{lastaddr} = sprintf("%#x", $ {$op->last}) if $op->can("last");
595 $h{classsym} = $opclass{$h{class}};
596 $h{flagval} = $op->flags;
597 $h{flags} = op_flags($op->flags);
598 $h{privval} = $op->private;
599 $h{private} = private_flags($h{name}, $op->private);
600 $h{addr} = sprintf("%#x", $$op);
601 $h{label} = $labels{$$op};
602 $h{typenum} = $op->type;
603 $h{noise} = $linenoise[$op->type];
604 $_->(\%h, $op, \$format, \$level) for @callbacks;
605 return fmt_line(\%h, $format, $level);
609 my($op, $level) = @_;
610 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
611 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
612 "addr" => sprintf("%#x", $$lastnext)};
613 print fmt_line($h, $gotofmt, $level+1);
615 $lastnext = $op->next;
616 print concise_op($op, $level, $format);
619 # B::OP::terse (see Terse.pm) now just calls this
621 my($op, $level) = @_;
623 # This isn't necessarily right, but there's no easy way to get
624 # from an OP to the right CV. This is a limitation of the
625 # ->terse() interface style, and there isn't much to do about
626 # it. In particular, we can die in concise_op if the main pad
627 # isn't long enough, or has the wrong kind of entries, compared to
628 # the pad a sub was compiled with. The fix for that would be to
629 # make a backwards compatible "terse" format that never even
630 # looked at the pad, just like the old B::Terse. I don't think
631 # that's worth the effort, though.
632 $curcv = main_cv unless $curcv;
634 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
635 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
636 "addr" => sprintf("%#x", $$lastnext)};
637 print fmt_line($h, $style{"terse"}[1], $level+1);
639 $lastnext = $op->next;
640 print concise_op($op, $level, $style{"terse"}[0]);
646 my $style = $tree_decorations[$tree_style];
647 my($space, $single, $kids, $kid, $nokid, $last, $lead, $size) = @$style;
648 my $name = concise_op($op, $level, $treefmt);
649 if (not $op->flags & OPf_KIDS) {
653 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
654 push @lines, tree($kid, $level+1);
657 for ($i = $#lines; substr($lines[$i], 0, 1) eq " "; $i--) {
658 $lines[$i] = $space . $lines[$i];
661 $lines[$i] = $last . $lines[$i];
663 if (substr($lines[$i], 0, 1) eq " ") {
664 $lines[$i] = $nokid . $lines[$i];
666 $lines[$i] = $kid . $lines[$i];
669 $lines[$i] = $kids . $lines[$i];
671 $lines[0] = $single . $lines[0];
673 return("$name$lead" . shift @lines,
674 map(" " x (length($name)+$size) . $_, @lines));
677 # *** Warning: fragile kludge ahead ***
678 # Because the B::* modules run in the same interpreter as the code
679 # they're compiling, their presence tends to distort the view we have of
680 # the code we're looking at. In particular, perl gives sequence numbers
681 # to COPs. If the program we're looking at were run on its own, this
682 # would start at 1. Because all of B::Concise and all the modules it
683 # uses are compiled first, though, by the time we get to the user's
684 # program the sequence number is already pretty high, which could be
685 # distracting if you're trying to tell OPs apart. Therefore we'd like to
686 # subtract an offset from all the sequence numbers we display, to
687 # restore the simpler view of the world. The trick is to know what that
688 # offset will be, when we're still compiling B::Concise! If we
689 # hardcoded a value, it would have to change every time B::Concise or
690 # other modules we use do. To help a little, what we do here is compile
691 # a little code at the end of the module, and compute the base sequence
692 # number for the user's program as being a small offset later, so all we
693 # have to worry about are changes in the offset.
695 # When you say "perl -MO=Concise -e '$a'", the output should look like:
697 # 4 <@> leave[t1] vKP/REFC ->(end)
699 #^ smallest OP sequence number should be 1
700 # 2 <;> nextstate(main 1 -e:1) v ->3
701 # ^ smallest COP sequence number should be 1
702 # - <1> ex-rv2sv vK/1 ->4
703 # 3 <$> gvsv(*a) s ->4
705 # If the second of the marked numbers there isn't 1, it means you need
706 # to update the corresponding magic number in the next line.
707 # Remember, this needs to stay the last things in the module.
709 # Why is this different for MacOS? Does it matter?
710 my $cop_seq_mnum = $^O eq 'MacOS' ? 12 : 11;
711 $cop_seq_base = svref_2object(eval 'sub{0;}')->START->cop_seq + $cop_seq_mnum;
719 B::Concise - Walk Perl syntax tree, printing concise info about ops
723 perl -MO=Concise[,OPTIONS] foo.pl
725 use B::Concise qw(set_style add_callback);
729 This compiler backend prints the internal OPs of a Perl program's syntax
730 tree in one of several space-efficient text formats suitable for debugging
731 the inner workings of perl or other compiler backends. It can print OPs in
732 the order they appear in the OP tree, in the order they will execute, or
733 in a text approximation to their tree structure, and the format of the
734 information displyed is customizable. Its function is similar to that of
735 perl's B<-Dx> debugging flag or the B<B::Terse> module, but it is more
736 sophisticated and flexible.
740 Here's is a short example of output, using the default formatting
743 % perl -MO=Concise -e '$a = $b + 42'
744 8 <@> leave[1 ref] vKP/REFC ->(end)
746 2 <;> nextstate(main 1 -e:1) v ->3
747 7 <2> sassign vKS/2 ->8
748 5 <2> add[t1] sK/2 ->6
749 - <1> ex-rv2sv sK/1 ->4
751 4 <$> const(IV 42) s ->5
752 - <1> ex-rv2sv sKRM*/1 ->7
755 Each line corresponds to an operator. Null ops appear as C<ex-opname>,
756 where I<opname> is the op that has been optimized away by perl.
758 The number on the first row indicates the op's sequence number. It's
759 given in base 36 by default.
761 The symbol between angle brackets indicates the op's type : for example,
762 <2> is a BINOP, <@> a LISTOP, etc. (see L</"OP class abbreviations">).
764 The opname may be followed by op-specific information in parentheses
765 (e.g. C<gvsv(*b)>), and by targ information in brackets (e.g.
768 Next come the op flags. The common flags are listed below
769 (L</"OP flags abbreviations">). The private flags follow, separated
770 by a slash. For example, C<vKP/REFC> means that the leave op has
771 public flags OPf_WANT_VOID, OPf_KIDS, and OPf_PARENS, and the private
774 Finally an arrow points to the sequence number of the next op.
778 Arguments that don't start with a hyphen are taken to be the names of
779 subroutines to print the OPs of; if no such functions are specified,
780 the main body of the program (outside any subroutines, and not
781 including use'd or require'd files) is printed. Passing C<BEGIN>,
782 C<CHECK>, C<INIT>, or C<END> will cause all of the corresponding
783 special blocks to be printed.
789 Print OPs in the order they appear in the OP tree (a preorder
790 traversal, starting at the root). The indentation of each OP shows its
791 level in the tree. This mode is the default, so the flag is included
792 simply for completeness.
796 Print OPs in the order they would normally execute (for the majority
797 of constructs this is a postorder traversal of the tree, ending at the
798 root). In most cases the OP that usually follows a given OP will
799 appear directly below it; alternate paths are shown by indentation. In
800 cases like loops when control jumps out of a linear path, a 'goto'
805 Print OPs in a text approximation of a tree, with the root of the tree
806 at the left and 'left-to-right' order of children transformed into
807 'top-to-bottom'. Because this mode grows both to the right and down,
808 it isn't suitable for large programs (unless you have a very wide
813 Use a tree format in which the minimum amount of space is used for the
814 lines connecting nodes (one character in most cases). This squeezes out
815 a few precious columns of screen real estate.
819 Use a tree format that uses longer edges to separate OP nodes. This format
820 tends to look better than the compact one, especially in ASCII, and is
825 Use tree connecting characters drawn from the VT100 line-drawing set.
826 This looks better if your terminal supports it.
830 Draw the tree with standard ASCII characters like C<+> and C<|>. These don't
831 look as clean as the VT100 characters, but they'll work with almost any
832 terminal (or the horizontal scrolling mode of less(1)) and are suitable
833 for text documentation or email. This is the default.
837 Include the main program in the output, even if subroutines were also
842 Print OP sequence numbers in base I<n>. If I<n> is greater than 10, the
843 digit for 11 will be 'a', and so on. If I<n> is greater than 36, the digit
844 for 37 will be 'A', and so on until 62. Values greater than 62 are not
845 currently supported. The default is 36.
849 Print sequence numbers with the most significant digit first. This is the
850 usual convention for Arabic numerals, and the default.
852 =item B<-littleendian>
854 Print seqence numbers with the least significant digit first.
858 Use the author's favorite set of formatting conventions. This is the
863 Use formatting conventions that emulate the output of B<B::Terse>. The
864 basic mode is almost indistinguishable from the real B<B::Terse>, and the
865 exec mode looks very similar, but is in a more logical order and lacks
866 curly brackets. B<B::Terse> doesn't have a tree mode, so the tree mode
867 is only vaguely reminiscient of B<B::Terse>.
871 Use formatting conventions in which the name of each OP, rather than being
872 written out in full, is represented by a one- or two-character abbreviation.
873 This is mainly a joke.
877 Use formatting conventions reminiscient of B<B::Debug>; these aren't
882 Use formatting conventions read from the environment variables
883 C<B_CONCISE_FORMAT>, C<B_CONCISE_GOTO_FORMAT>, and C<B_CONCISE_TREE_FORMAT>.
887 =head1 FORMATTING SPECIFICATIONS
889 For each general style ('concise', 'terse', 'linenoise', etc.) there are
890 three specifications: one of how OPs should appear in the basic or exec
891 modes, one of how 'goto' lines should appear (these occur in the exec
892 mode only), and one of how nodes should appear in tree mode. Each has the
893 same format, described below. Any text that doesn't match a special
894 pattern is copied verbatim.
898 =item B<(x(>I<exec_text>B<;>I<basic_text>B<)x)>
900 Generates I<exec_text> in exec mode, or I<basic_text> in basic mode.
902 =item B<(*(>I<text>B<)*)>
904 Generates one copy of I<text> for each indentation level.
906 =item B<(*(>I<text1>B<;>I<text2>B<)*)>
908 Generates one fewer copies of I<text1> than the indentation level, followed
909 by one copy of I<text2> if the indentation level is more than 0.
911 =item B<(?(>I<text1>B<#>I<var>I<Text2>B<)?)>
913 If the value of I<var> is true (not empty or zero), generates the
914 value of I<var> surrounded by I<text1> and I<Text2>, otherwise
919 Generates the value of the variable I<var>.
923 Generates the value of I<var>, left jutified to fill I<N> spaces.
927 Any number of tildes and surrounding whitespace will be collapsed to
932 The following variables are recognized:
938 The address of the OP, in hexidecimal.
942 The OP-specific information of the OP (such as the SV for an SVOP, the
943 non-local exit pointers for a LOOP, etc.) enclosed in paretheses.
947 The B-determined class of the OP, in all caps.
951 A single symbol abbreviating the class of the OP.
955 The label of the statement or block the OP is the start of, if any.
959 The name of the OP, or 'ex-foo' if the OP is a null that used to be a foo.
963 The target of the OP, or nothing for a nulled OP.
967 The address of the OP's first child, in hexidecimal.
971 The OP's flags, abbreviated as a series of symbols.
975 The numeric value of the OP's flags.
979 The sequence number of the OP, or a hyphen if it doesn't have one.
983 'NEXT', 'LAST', or 'REDO' if the OP is a target of one of those in exec
984 mode, or empty otherwise.
988 The address of the OP's last child, in hexidecimal.
996 The OP's name, in all caps.
1000 The sequence number of the OP's next OP.
1004 The address of the OP's next OP, in hexidecimal.
1008 A one- or two-character abbreviation for the OP's name.
1012 The OP's private flags, rendered with abbreviated names if possible.
1016 The numeric value of the OP's private flags.
1020 The sequence number of the OP. Note that this is a sequence number
1021 generated by B::Concise.
1025 Whether or not the op has been optimised by the peephole optimiser.
1029 Whether or not the op is statically defined. This flag is used by the
1030 B::C compiler backend and indicates that the op should not be freed.
1034 The address of the OP's next youngest sibling, in hexidecimal.
1038 The address of the OP's SV, if it has an SV, in hexidecimal.
1042 The class of the OP's SV, if it has one, in all caps (e.g., 'IV').
1046 The value of the OP's SV, if it has one, in a short human-readable format.
1050 The numeric value of the OP's targ.
1054 The name of the variable the OP's targ refers to, if any, otherwise the
1055 letter t followed by the OP's targ in decimal.
1057 =item B<#targarglife>
1059 Same as B<#targarg>, but followed by the COP sequence numbers that delimit
1060 the variable's lifetime (or 'end' for a variable in an open scope) for a
1065 The numeric value of the OP's type, in decimal.
1069 =head1 ABBREVIATIONS
1071 =head2 OP flags abbreviations
1073 v OPf_WANT_VOID Want nothing (void context)
1074 s OPf_WANT_SCALAR Want single value (scalar context)
1075 l OPf_WANT_LIST Want list of any length (list context)
1076 K OPf_KIDS There is a firstborn child.
1077 P OPf_PARENS This operator was parenthesized.
1078 (Or block needs explicit scope entry.)
1079 R OPf_REF Certified reference.
1080 (Return container, not containee).
1081 M OPf_MOD Will modify (lvalue).
1082 S OPf_STACKED Some arg is arriving on the stack.
1083 * OPf_SPECIAL Do something weird for this op (see op.h)
1085 =head2 OP class abbreviations
1087 0 OP (aka BASEOP) An OP with no children
1088 1 UNOP An OP with one child
1089 2 BINOP An OP with two children
1090 | LOGOP A control branch OP
1091 @ LISTOP An OP that could have lots of children
1092 / PMOP An OP with a regular expression
1093 $ SVOP An OP with an SV
1094 " PVOP An OP with a string
1095 { LOOP An OP that holds pointers for a loop
1096 ; COP An OP that marks the start of a statement
1097 # PADOP An OP with a GV on the pad
1099 =head1 Using B::Concise outside of the O framework
1101 It is possible to extend B<B::Concise> by using it outside of the B<O>
1102 framework and providing new styles and new variables.
1104 use B::Concise qw(set_style add_callback);
1105 set_style($format, $gotofmt, $treefmt);
1110 my ($h, $op, $level, $format) = @_;
1111 $h->{variable} = some_func($op);
1114 B::Concise::compile(@options)->();
1116 You can specify a style by calling the B<set_style> subroutine. If you
1117 have a new variable in your style, or you want to change the value of an
1118 existing variable, you will need to add a callback to specify the value
1121 This is done by calling B<add_callback> passing references to any
1122 callback subroutines. The subroutines are called in the same order as
1123 they are added. Each subroutine is passed four parameters. These are a
1124 reference to a hash, the keys of which are the names of the variables
1125 and the values of which are their values, the op, the level and the
1128 To define your own variables, simply add them to the hash, or change
1129 existing values if you need to. The level and format are passed in as
1130 references to scalars, but it is unlikely that they will need to be
1131 changed or even used.
1133 To switch back to one of the standard styles like C<concise> or
1134 C<terse>, use C<set_style_standard>.
1136 To see the output, call the subroutine returned by B<compile> in the
1137 same way that B<O> does.
1141 Stephen McCamant, E<lt>smcc@CSUA.Berkeley.EDUE<gt>.