Speed up lib/unicore/mktables by 45%
[p5sagit/p5-mst-13.2.git] / ext / B / B / Concise.pm
CommitLineData
c99ca59a 1package B::Concise;
c27ea44e 2# Copyright (C) 2000-2003 Stephen McCamant. All rights reserved.
c99ca59a 3# This program is free software; you can redistribute and/or modify it
4# under the same terms as Perl itself.
5
8ec8fbef 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.
78ad9108 11
8ec8fbef 12use strict; # use #2
13use warnings; # uses #3 and #4, since warnings uses Carp
78ad9108 14
8ec8fbef 15use Exporter (); # use #5
16
b4ec42b6 17our $VERSION = "0.65";
78ad9108 18our @ISA = qw(Exporter);
cc02ea56 19our @EXPORT_OK = qw( set_style set_style_standard add_callback
20 concise_subref concise_cv concise_main
21 add_style walk_output compile reset_sequence );
22our %EXPORT_TAGS =
23 ( io => [qw( walk_output compile reset_sequence )],
24 style => [qw( add_style set_style_standard )],
25 cb => [qw( add_callback )],
26 mech => [qw( concise_subref concise_cv concise_main )], );
78ad9108 27
8ec8fbef 28# use #6
c99ca59a 29use B qw(class ppname main_start main_root main_cv cstring svref_2object
6a077020 30 SVf_IOK SVf_NOK SVf_POK SVf_IVisUV SVf_FAKE OPf_KIDS OPf_SPECIAL
31 CVf_ANON);
c99ca59a 32
f95e3c3c 33my %style =
c99ca59a 34 ("terse" =>
c3caa09d 35 ["(?(#label =>\n)?)(*( )*)#class (#addr) #name (?([#targ])?) "
36 . "#svclass~(?((#svaddr))?)~#svval~(?(label \"#coplabel\")?)\n",
c99ca59a 37 "(*( )*)goto #class (#addr)\n",
38 "#class pp_#name"],
39 "concise" =>
40 ["#hyphseq2 (*( (x( ;)x))*)<#classsym> "
cc02ea56 41 . "#exname#arg(?([#targarglife])?)~#flags(?(/#private)?)(x(;~->#next)x)\n"
42 , " (*( )*) goto #seq\n",
c99ca59a 43 "(?(<#seq>)?)#exname#arg(?([#targarglife])?)"],
44 "linenoise" =>
45 ["(x(;(*( )*))x)#noise#arg(?([#targarg])?)(x( ;\n)x)",
46 "gt_#seq ",
47 "(?(#seq)?)#noise#arg(?([#targarg])?)"],
48 "debug" =>
49 ["#class (#addr)\n\top_next\t\t#nextaddr\n\top_sibling\t#sibaddr\n\t"
7252851f 50 . "op_ppaddr\tPL_ppaddr[OP_#NAME]\n\top_type\t\t#typenum\n" .
51 ($] > 5.009 ? '' : "\top_seq\t\t#seqnum\n")
2814eb74 52 . "\top_flags\t#flagval\n\top_private\t#privval\n"
c99ca59a 53 . "(?(\top_first\t#firstaddr\n)?)(?(\top_last\t\t#lastaddr\n)?)"
54 . "(?(\top_sv\t\t#svaddr\n)?)",
55 " GOTO #addr\n",
56 "#addr"],
57 "env" => [$ENV{B_CONCISE_FORMAT}, $ENV{B_CONCISE_GOTO_FORMAT},
58 $ENV{B_CONCISE_TREE_FORMAT}],
59 );
60
724aa791 61# Renderings, ie how Concise prints, is controlled by these vars
62# primary:
63our $stylename; # selects current style from %style
64my $order = "basic"; # how optree is walked & printed: basic, exec, tree
65
66# rendering mechanics:
67# these 'formats' are the line-rendering templates
68# they're updated from %style when $stylename changes
69my ($format, $gotofmt, $treefmt);
70
71# lesser players:
72my $base = 36; # how <sequence#> is displayed
73my $big_endian = 1; # more <sequence#> display
74my $tree_style = 0; # tree-order details
75my $banner = 1; # print banner before optree is traversed
cc02ea56 76my $do_main = 0; # force printing of main routine
724aa791 77
cc02ea56 78# another factor: can affect all styles!
724aa791 79our @callbacks; # allow external management
80
81set_style_standard("concise");
82
c99ca59a 83my $curcv;
c27ea44e 84my $cop_seq_base;
78ad9108 85
86sub set_style {
87 ($format, $gotofmt, $treefmt) = @_;
724aa791 88 #warn "set_style: deprecated, use set_style_standard instead\n"; # someday
f95e3c3c 89 die "expecting 3 style-format args\n" unless @_ == 3;
90}
91
92sub add_style {
93 my ($newstyle,@args) = @_;
94 die "style '$newstyle' already exists, choose a new name\n"
95 if exists $style{$newstyle};
96 die "expecting 3 style-format args\n" unless @args == 3;
97 $style{$newstyle} = [@args];
724aa791 98 $stylename = $newstyle; # update rendering state
78ad9108 99}
100
31b49ad4 101sub set_style_standard {
724aa791 102 ($stylename) = @_; # update rendering state
f95e3c3c 103 die "err: style '$stylename' unknown\n" unless exists $style{$stylename};
104 set_style(@{$style{$stylename}});
31b49ad4 105}
106
78ad9108 107sub add_callback {
108 push @callbacks, @_;
109}
c99ca59a 110
f95e3c3c 111# output handle, used with all Concise-output printing
cc02ea56 112our $walkHandle; # public for your convenience
113BEGIN { $walkHandle = \*STDOUT }
f95e3c3c 114
115sub walk_output { # updates $walkHandle
116 my $handle = shift;
cc02ea56 117 return $walkHandle unless $handle; # allow use as accessor
118
f95e3c3c 119 if (ref $handle eq 'SCALAR') {
2ce64696 120 require Config;
121 die "no perlio in this build, can't call walk_output (\\\$scalar)\n"
122 unless $Config::Config{useperlio};
f95e3c3c 123 # in 5.8+, open(FILEHANDLE,MODE,REFERENCE) writes to string
2ce64696 124 open my $tmp, '>', $handle; # but cant re-set existing STDOUT
f95e3c3c 125 $walkHandle = $tmp; # so use my $tmp as intermediate var
cc02ea56 126 return $walkHandle;
f95e3c3c 127 }
cc02ea56 128 my $iotype = ref $handle;
f95e3c3c 129 die "expecting argument/object that can print\n"
cc02ea56 130 unless $iotype eq 'GLOB' or $iotype and $handle->can('print');
131 $walkHandle = $handle;
f95e3c3c 132}
133
8ec8fbef 134sub concise_subref {
f95e3c3c 135 my($order, $coderef) = @_;
136 my $codeobj = svref_2object($coderef);
cc02ea56 137
138 return concise_stashref(@_)
139 unless ref $codeobj eq 'B::CV';
f95e3c3c 140 concise_cv_obj($order, $codeobj);
8ec8fbef 141}
142
cc02ea56 143sub concise_stashref {
144 my($order, $h) = @_;
145 foreach my $k (sort keys %$h) {
146 local *s = $h->{$k};
147 my $coderef = *s{CODE} or next;
148 reset_sequence();
149 print "FUNC: ", *s, "\n";
150 my $codeobj = svref_2object($coderef);
151 next unless ref $codeobj eq 'B::CV';
152 eval { concise_cv_obj($order, $codeobj) }
153 or warn "err $@ on $codeobj";
154 }
155}
156
8ec8fbef 157# This should have been called concise_subref, but it was exported
158# under this name in versions before 0.56
159sub concise_cv { concise_subref(@_); }
160
161sub concise_cv_obj {
162 my ($order, $cv) = @_;
c99ca59a 163 $curcv = $cv;
f95e3c3c 164 die "err: coderef has no START\n" if class($cv->START) eq "NULL";
c27ea44e 165 sequence($cv->START);
c99ca59a 166 if ($order eq "exec") {
167 walk_exec($cv->START);
168 } elsif ($order eq "basic") {
169 walk_topdown($cv->ROOT, sub { $_[0]->concise($_[1]) }, 0);
170 } else {
f95e3c3c 171 print $walkHandle tree($cv->ROOT, 0);
c99ca59a 172 }
173}
174
31b49ad4 175sub concise_main {
176 my($order) = @_;
177 sequence(main_start);
178 $curcv = main_cv;
179 if ($order eq "exec") {
180 return if class(main_start) eq "NULL";
181 walk_exec(main_start);
182 } elsif ($order eq "tree") {
183 return if class(main_root) eq "NULL";
f95e3c3c 184 print $walkHandle tree(main_root, 0);
31b49ad4 185 } elsif ($order eq "basic") {
186 return if class(main_root) eq "NULL";
187 walk_topdown(main_root,
188 sub { $_[0]->concise($_[1]) }, 0);
189 }
190}
191
8ec8fbef 192sub concise_specials {
193 my($name, $order, @cv_s) = @_;
194 my $i = 1;
195 if ($name eq "BEGIN") {
196 splice(@cv_s, 0, 7); # skip 7 BEGIN blocks in this file
197 } elsif ($name eq "CHECK") {
198 pop @cv_s; # skip the CHECK block that calls us
199 }
f95e3c3c 200 for my $cv (@cv_s) {
201 print $walkHandle "$name $i:\n";
8ec8fbef 202 $i++;
203 concise_cv_obj($order, $cv);
204 }
205}
206
c99ca59a 207my $start_sym = "\e(0"; # "\cN" sometimes also works
208my $end_sym = "\e(B"; # "\cO" respectively
209
f95e3c3c 210my @tree_decorations =
c99ca59a 211 ([" ", "--", "+-", "|-", "| ", "`-", "-", 1],
212 [" ", "-", "+", "+", "|", "`", "", 0],
213 [" ", map("$start_sym$_$end_sym", "qq", "wq", "tq", "x ", "mq", "q"), 1],
214 [" ", map("$start_sym$_$end_sym", "q", "w", "t", "x", "m"), "", 0],
215 );
78ad9108 216
cc02ea56 217
218sub compileOpts {
219 # set rendering state from options and args
c99ca59a 220 my @options = grep(/^-/, @_);
221 my @args = grep(!/^-/, @_);
c99ca59a 222 for my $o (@options) {
cc02ea56 223 # mode/order
c99ca59a 224 if ($o eq "-basic") {
225 $order = "basic";
226 } elsif ($o eq "-exec") {
227 $order = "exec";
228 } elsif ($o eq "-tree") {
229 $order = "tree";
cc02ea56 230 }
231 # tree-specific
232 elsif ($o eq "-compact") {
c99ca59a 233 $tree_style |= 1;
234 } elsif ($o eq "-loose") {
235 $tree_style &= ~1;
236 } elsif ($o eq "-vt") {
237 $tree_style |= 2;
238 } elsif ($o eq "-ascii") {
239 $tree_style &= ~2;
cc02ea56 240 }
241 # sequence numbering
242 elsif ($o =~ /^-base(\d+)$/) {
c99ca59a 243 $base = $1;
244 } elsif ($o eq "-bigendian") {
245 $big_endian = 1;
246 } elsif ($o eq "-littleendian") {
247 $big_endian = 0;
cc02ea56 248 }
249 elsif ($o eq "-nobanner") {
724aa791 250 $banner = 0;
cc02ea56 251 } elsif ($o eq "-banner") {
252 $banner = 1;
253 }
254 elsif ($o eq "-main") {
255 $do_main = 1;
256 } elsif ($o eq "-nomain") {
257 $do_main = 0;
724aa791 258 }
cc02ea56 259 # line-style options
724aa791 260 elsif (exists $style{substr($o, 1)}) {
f95e3c3c 261 $stylename = substr($o, 1);
724aa791 262 set_style_standard($stylename);
c99ca59a 263 } else {
264 warn "Option $o unrecognized";
265 }
266 }
cc02ea56 267 return (@args);
268}
269
270sub compile {
271 my (@args) = compileOpts(@_);
c27ea44e 272 return sub {
cc02ea56 273 my @newargs = compileOpts(@_); # accept new rendering options
274 warn "disregarding non-options: @newargs\n" if @newargs;
275
276 for my $objname (@args) {
59910b6d 277 next unless $objname; # skip null args to avoid noisy responses
278
cc02ea56 279 if ($objname eq "BEGIN") {
280 concise_specials("BEGIN", $order,
281 B::begin_av->isa("B::AV") ?
282 B::begin_av->ARRAY : ());
283 } elsif ($objname eq "INIT") {
284 concise_specials("INIT", $order,
285 B::init_av->isa("B::AV") ?
286 B::init_av->ARRAY : ());
287 } elsif ($objname eq "CHECK") {
288 concise_specials("CHECK", $order,
289 B::check_av->isa("B::AV") ?
290 B::check_av->ARRAY : ());
291 } elsif ($objname eq "END") {
292 concise_specials("END", $order,
293 B::end_av->isa("B::AV") ?
294 B::end_av->ARRAY : ());
295 }
296 else {
297 # convert function names to subrefs
298 my $objref;
299 if (ref $objname) {
300 print $walkHandle "B::Concise::compile($objname)\n"
301 if $banner;
302 $objref = $objname;
8ec8fbef 303 } else {
cc02ea56 304 $objname = "main::" . $objname unless $objname =~ /::/;
305 print $walkHandle "$objname:\n";
306 no strict 'refs';
307 die "err: unknown function ($objname)\n"
308 unless *{$objname}{CODE};
309 $objref = \&$objname;
8ec8fbef 310 }
cc02ea56 311 concise_subref($order, $objref);
c99ca59a 312 }
313 }
c27ea44e 314 if (!@args or $do_main) {
f95e3c3c 315 print $walkHandle "main program:\n" if $do_main;
31b49ad4 316 concise_main($order);
c99ca59a 317 }
cc02ea56 318 return @args; # something
c99ca59a 319 }
320}
321
322my %labels;
724aa791 323my $lastnext; # remembers op-chain, used to insert gotos
c99ca59a 324
325my %opclass = ('OP' => "0", 'UNOP' => "1", 'BINOP' => "2", 'LOGOP' => "|",
326 'LISTOP' => "@", 'PMOP' => "/", 'SVOP' => "\$", 'GVOP' => "*",
051f02e9 327 'PVOP' => '"', 'LOOP' => "{", 'COP' => ";", 'PADOP' => "#");
c99ca59a 328
8ec8fbef 329no warnings 'qw'; # "Possible attempt to put comments..."; use #7
35fc55f1 330my @linenoise =
331 qw'# () sc ( @? 1 $* gv *{ m$ m@ m% m? p/ *$ $ $# & a& pt \\ s\\ rf bl
c99ca59a 332 ` *? <> ?? ?/ r/ c/ // qr s/ /c y/ = @= C sC Cp sp df un BM po +1 +I
333 -1 -I 1+ I+ 1- I- ** * i* / i/ %$ i% x + i+ - i- . " << >> < i<
334 > i> <= i, >= i. == i= != i! <? i? s< s> s, s. s= s! s? b& b^ b| -0 -i
335 ! ~ a2 si cs rd sr e^ lg sq in %x %o ab le ss ve ix ri sf FL od ch cy
336 uf lf uc lc qm @ [f [ @[ eh vl ky dl ex % ${ @{ uk pk st jn ) )[ a@
337 a% sl +] -] [- [+ so rv GS GW MS MW .. f. .f && || ^^ ?: &= |= -> s{ s}
338 v} ca wa di rs ;; ; ;d }{ { } {} f{ it {l l} rt }l }n }r dm }g }e ^o
339 ^c ^| ^# um bm t~ u~ ~d DB db ^s se ^g ^r {w }w pf pr ^O ^K ^R ^W ^d ^v
340 ^e ^t ^k t. fc ic fl .s .p .b .c .l .a .h g1 s1 g2 s2 ?. l? -R -W -X -r
341 -w -x -e -o -O -z -s -M -A -C -S -c -b -f -d -p -l -u -g -k -t -T -B cd
342 co cr u. cm ut r. l@ s@ r@ mD uD oD rD tD sD wD cD f$ w$ p$ sh e$ k$ g3
343 g4 s4 g5 s5 T@ C@ L@ G@ A@ S@ Hg Hc Hr Hw Mg Mc Ms Mr Sg Sc So rq do {e
344 e} {t t} g6 G6 6e g7 G7 7e g8 G8 8e g9 G9 9e 6s 7s 8s 9s 6E 7E 8E 9E Pn
c27ea44e 345 Pu GP SP EP Gn Gg GG SG EG g0 c$ lk t$ ;s n> // /= CO';
c99ca59a 346
347my $chars = "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
348
19e169bf 349sub op_flags { # common flags (see BASOP.op_flags in op.h)
c99ca59a 350 my($x) = @_;
351 my(@v);
352 push @v, "v" if ($x & 3) == 1;
353 push @v, "s" if ($x & 3) == 2;
354 push @v, "l" if ($x & 3) == 3;
355 push @v, "K" if $x & 4;
356 push @v, "P" if $x & 8;
357 push @v, "R" if $x & 16;
358 push @v, "M" if $x & 32;
359 push @v, "S" if $x & 64;
360 push @v, "*" if $x & 128;
361 return join("", @v);
362}
363
364sub base_n {
365 my $x = shift;
366 return "-" . base_n(-$x) if $x < 0;
367 my $str = "";
368 do { $str .= substr($chars, $x % $base, 1) } while $x = int($x / $base);
369 $str = reverse $str if $big_endian;
370 return $str;
371}
372
c27ea44e 373my %sequence_num;
374my $seq_max = 1;
375
f95e3c3c 376sub reset_sequence {
377 # reset the sequence
378 %sequence_num = ();
379 $seq_max = 1;
cc02ea56 380 $lastnext = 0;
f95e3c3c 381}
382
c27ea44e 383sub seq {
384 my($op) = @_;
385 return "-" if not exists $sequence_num{$$op};
386 return base_n($sequence_num{$$op});
387}
c99ca59a 388
389sub walk_topdown {
390 my($op, $sub, $level) = @_;
391 $sub->($op, $level);
392 if ($op->flags & OPf_KIDS) {
393 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
394 walk_topdown($kid, $sub, $level + 1);
395 }
396 }
c6e79e55 397 if (class($op) eq "PMOP") {
398 my $maybe_root = $op->pmreplroot;
399 if (ref($maybe_root) and $maybe_root->isa("B::OP")) {
400 # It really is the root of the replacement, not something
401 # else stored here for lack of space elsewhere
402 walk_topdown($maybe_root, $sub, $level + 1);
403 }
c99ca59a 404 }
405}
406
407sub walklines {
408 my($ar, $level) = @_;
409 for my $l (@$ar) {
410 if (ref($l) eq "ARRAY") {
411 walklines($l, $level + 1);
412 } else {
413 $l->concise($level);
414 }
415 }
416}
417
418sub walk_exec {
419 my($top, $level) = @_;
420 my %opsseen;
421 my @lines;
422 my @todo = ([$top, \@lines]);
423 while (@todo and my($op, $targ) = @{shift @todo}) {
424 for (; $$op; $op = $op->next) {
425 last if $opsseen{$$op}++;
426 push @$targ, $op;
427 my $name = $op->name;
62e36f8a 428 if (class($op) eq "LOGOP") {
c99ca59a 429 my $ar = [];
430 push @$targ, $ar;
431 push @todo, [$op->other, $ar];
432 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
433 my $ar = [];
434 push @$targ, $ar;
435 push @todo, [$op->pmreplstart, $ar];
436 } elsif ($name =~ /^enter(loop|iter)$/) {
7252851f 437 if ($] > 5.009) {
438 $labels{${$op->nextop}} = "NEXT";
439 $labels{${$op->lastop}} = "LAST";
440 $labels{${$op->redoop}} = "REDO";
441 } else {
442 $labels{$op->nextop->seq} = "NEXT";
443 $labels{$op->lastop->seq} = "LAST";
444 $labels{$op->redoop->seq} = "REDO";
445 }
c99ca59a 446 }
447 }
448 }
449 walklines(\@lines, 0);
450}
451
c27ea44e 452# The structure of this routine is purposely modeled after op.c's peep()
453sub sequence {
454 my($op) = @_;
455 my $oldop = 0;
456 return if class($op) eq "NULL" or exists $sequence_num{$$op};
457 for (; $$op; $op = $op->next) {
458 last if exists $sequence_num{$$op};
459 my $name = $op->name;
460 if ($name =~ /^(null|scalar|lineseq|scope)$/) {
461 next if $oldop and $ {$op->next};
462 } else {
463 $sequence_num{$$op} = $seq_max++;
464 if (class($op) eq "LOGOP") {
465 my $other = $op->other;
466 $other = $other->next while $other->name eq "null";
467 sequence($other);
468 } elsif (class($op) eq "LOOP") {
469 my $redoop = $op->redoop;
470 $redoop = $redoop->next while $redoop->name eq "null";
471 sequence($redoop);
472 my $nextop = $op->nextop;
473 $nextop = $nextop->next while $nextop->name eq "null";
474 sequence($nextop);
475 my $lastop = $op->lastop;
476 $lastop = $lastop->next while $lastop->name eq "null";
477 sequence($lastop);
478 } elsif ($name eq "subst" and $ {$op->pmreplstart}) {
479 my $replstart = $op->pmreplstart;
480 $replstart = $replstart->next while $replstart->name eq "null";
481 sequence($replstart);
482 }
483 }
484 $oldop = $op;
485 }
486}
487
724aa791 488sub fmt_line { # generate text-line for op.
cc02ea56 489 my($hr, $op, $text, $level) = @_;
490
491 $_->($hr, $op, \$text, \$level, $stylename) for @callbacks;
492
724aa791 493 return '' if $hr->{SKIP}; # suppress line if a callback said so
cc02ea56 494 return '' if $hr->{goto} and $hr->{goto} eq '-'; # no goto nowhere
f95e3c3c 495
cc02ea56 496 # spec: (?(text1#varText2)?)
c99ca59a 497 $text =~ s/\(\?\(([^\#]*?)\#(\w+)([^\#]*?)\)\?\)/
f95e3c3c 498 $hr->{$2} ? $1.$hr->{$2}.$3 : ""/eg;
499
cc02ea56 500 # spec: (x(exec_text;basic_text)x)
c99ca59a 501 $text =~ s/\(x\((.*?);(.*?)\)x\)/$order eq "exec" ? $1 : $2/egs;
cc02ea56 502
503 # spec: (*(text)*)
c99ca59a 504 $text =~ s/\(\*\(([^;]*?)\)\*\)/$1 x $level/egs;
cc02ea56 505
506 # spec: (*(text1;text2)*)
c99ca59a 507 $text =~ s/\(\*\((.*?);(.*?)\)\*\)/$1 x ($level - 1) . $2 x ($level>0)/egs;
cc02ea56 508
509 # convert #Var to tag=>val form: Var\t#var
510 $text =~ s/\#([A-Z][a-z]+)(\d+)?/\t\u$1\t\L#$1$2/gs;
511
512 # spec: #varN
724aa791 513 $text =~ s/\#([a-zA-Z]+)(\d+)/sprintf("%-$2s", $hr->{$1})/eg;
514
cc02ea56 515 $text =~ s/\#([a-zA-Z]+)/$hr->{$1}/eg; # populate #var's
516 $text =~ s/[ \t]*~+[ \t]*/ /g; # squeeze tildes
f95e3c3c 517 chomp $text;
518 return "$text\n" if $text ne "";
519 return $text; # suppress empty lines
c99ca59a 520}
521
19e169bf 522our %priv; # used to display each opcode's BASEOP.op_private values
523
c99ca59a 524$priv{$_}{128} = "LVINTRO"
525 for ("pos", "substr", "vec", "threadsv", "gvsv", "rv2sv", "rv2hv", "rv2gv",
526 "rv2av", "rv2arylen", "aelem", "helem", "aslice", "hslice", "padsv",
241416b8 527 "padav", "padhv", "enteriter");
c99ca59a 528$priv{$_}{64} = "REFC" for ("leave", "leavesub", "leavesublv", "leavewrite");
529$priv{"aassign"}{64} = "COMMON";
4ac6efe6 530$priv{"aassign"}{32} = "PHASH" if $] < 5.009;
c99ca59a 531$priv{"sassign"}{64} = "BKWARD";
532$priv{$_}{64} = "RTIME" for ("match", "subst", "substcont");
533@{$priv{"trans"}}{1,2,4,8,16,64} = ("<UTF", ">UTF", "IDENT", "SQUASH", "DEL",
534 "COMPL", "GROWS");
535$priv{"repeat"}{64} = "DOLIST";
536$priv{"leaveloop"}{64} = "CONT";
537@{$priv{$_}}{32,64,96} = ("DREFAV", "DREFHV", "DREFSV")
314d4778 538 for (qw(rv2gv rv2sv padsv aelem helem));
d4797c1d 539@{$priv{"entersub"}}{16,32,64} = ("DBG","TARG","NOMOD");
c99ca59a 540@{$priv{$_}}{4,8,128} = ("INARGS","AMPER","NO()") for ("entersub", "rv2cv");
541$priv{"gv"}{32} = "EARLYCV";
542$priv{"aelem"}{16} = $priv{"helem"}{16} = "LVDEFER";
241416b8 543$priv{$_}{16} = "OURINTR" for ("gvsv", "rv2sv", "rv2av", "rv2hv", "r2gv",
544 "enteriter");
c99ca59a 545$priv{$_}{16} = "TARGMY"
546 for (map(($_,"s$_"),"chop", "chomp"),
547 map(($_,"i_$_"), "postinc", "postdec", "multiply", "divide", "modulo",
548 "add", "subtract", "negate"), "pow", "concat", "stringify",
549 "left_shift", "right_shift", "bit_and", "bit_xor", "bit_or",
550 "complement", "atan2", "sin", "cos", "rand", "exp", "log", "sqrt",
551 "int", "hex", "oct", "abs", "length", "index", "rindex", "sprintf",
552 "ord", "chr", "crypt", "quotemeta", "join", "push", "unshift", "flock",
553 "chdir", "chown", "chroot", "unlink", "chmod", "utime", "rename",
554 "link", "symlink", "mkdir", "rmdir", "wait", "waitpid", "system",
555 "exec", "kill", "getppid", "getpgrp", "setpgrp", "getpriority",
556 "setpriority", "time", "sleep");
ef3e5ea9 557$priv{$_}{4} = "REVERSED" for ("enteriter", "iter");
d4797c1d 558@{$priv{"const"}}{4,8,16,32,64,128} = ("SHORT","STRICT","ENTERED",'$[',"BARE","WARN");
c99ca59a 559$priv{"flip"}{64} = $priv{"flop"}{64} = "LINENUM";
560$priv{"list"}{64} = "GUESSED";
561$priv{"delete"}{64} = "SLICE";
562$priv{"exists"}{64} = "SUB";
563$priv{$_}{64} = "LOCALE"
564 for ("sort", "prtf", "sprintf", "slt", "sle", "seq", "sne", "sgt", "sge",
565 "scmp", "lc", "uc", "lcfirst", "ucfirst");
6c3fb703 566@{$priv{"sort"}}{1,2,4,8,16} = ("NUM", "INT", "REV", "INPLACE","DESC");
c99ca59a 567$priv{"threadsv"}{64} = "SVREFd";
c27ea44e 568@{$priv{$_}}{16,32,64,128} = ("INBIN","INCR","OUTBIN","OUTCR")
569 for ("open", "backtick");
c99ca59a 570$priv{"exit"}{128} = "VMS";
feaeca78 571$priv{$_}{2} = "FTACCESS"
572 for ("ftrread", "ftrwrite", "ftrexec", "fteread", "ftewrite", "fteexec");
32454ac8 573if ($] >= 5.009) {
574 # Stacked filetests are post 5.8.x
575 $priv{$_}{4} = "FTSTACKED"
576 for ("ftrread", "ftrwrite", "ftrexec", "fteread", "ftewrite", "fteexec",
577 "ftis", "fteowned", "ftrowned", "ftzero", "ftsize", "ftmtime",
578 "ftatime", "ftctime", "ftsock", "ftchr", "ftblk", "ftfile", "ftdir",
579 "ftpipe", "ftlink", "ftsuid", "ftsgid", "ftsvtx", "fttty", "fttext",
580 "ftbinary");
581 # Lexical $_ is post 5.8.x
582 $priv{$_}{2} = "GREPLEX"
583 for ("mapwhile", "mapstart", "grepwhile", "grepstart");
584}
c99ca59a 585
586sub private_flags {
587 my($name, $x) = @_;
588 my @s;
589 for my $flag (128, 96, 64, 32, 16, 8, 4, 2, 1) {
590 if ($priv{$name}{$flag} and $x & $flag and $x >= $flag) {
591 $x -= $flag;
592 push @s, $priv{$name}{$flag};
593 }
594 }
595 push @s, $x if $x;
596 return join(",", @s);
597}
598
c27ea44e 599sub concise_sv {
600 my($sv, $hr) = @_;
601 $hr->{svclass} = class($sv);
31b49ad4 602 $hr->{svclass} = "UV"
603 if $hr->{svclass} eq "IV" and $sv->FLAGS & SVf_IVisUV;
c27ea44e 604 $hr->{svaddr} = sprintf("%#x", $$sv);
605 if ($hr->{svclass} eq "GV") {
606 my $gv = $sv;
607 my $stash = $gv->STASH->NAME;
608 if ($stash eq "main") {
609 $stash = "";
610 } else {
611 $stash = $stash . "::";
612 }
613 $hr->{svval} = "*$stash" . $gv->SAFENAME;
614 return "*$stash" . $gv->SAFENAME;
615 } else {
616 while (class($sv) eq "RV") {
617 $hr->{svval} .= "\\";
618 $sv = $sv->RV;
619 }
620 if (class($sv) eq "SPECIAL") {
40b5b14f 621 $hr->{svval} .= ["Null", "sv_undef", "sv_yes", "sv_no"]->[$$sv];
c27ea44e 622 } elsif ($sv->FLAGS & SVf_NOK) {
40b5b14f 623 $hr->{svval} .= $sv->NV;
c27ea44e 624 } elsif ($sv->FLAGS & SVf_IOK) {
31b49ad4 625 $hr->{svval} .= $sv->int_value;
c27ea44e 626 } elsif ($sv->FLAGS & SVf_POK) {
40b5b14f 627 $hr->{svval} .= cstring($sv->PV);
31b49ad4 628 } elsif (class($sv) eq "HV") {
629 $hr->{svval} .= 'HASH';
c27ea44e 630 }
cc02ea56 631
632 $hr->{svval} = 'undef' unless defined $hr->{svval};
633 my $out = $hr->{svclass};
634 return $out .= " $hr->{svval}" ;
c27ea44e 635 }
636}
637
c99ca59a 638sub concise_op {
639 my ($op, $level, $format) = @_;
640 my %h;
641 $h{exname} = $h{name} = $op->name;
642 $h{NAME} = uc $h{name};
643 $h{class} = class($op);
644 $h{extarg} = $h{targ} = $op->targ;
645 $h{extarg} = "" unless $h{extarg};
646 if ($h{name} eq "null" and $h{targ}) {
8ec8fbef 647 # targ holds the old type
c99ca59a 648 $h{exname} = "ex-" . substr(ppname($h{targ}), 3);
649 $h{extarg} = "";
8ec8fbef 650 } elsif ($op->name =~ /^leave(sub(lv)?|write)?$/) {
651 # targ potentially holds a reference count
652 if ($op->private & 64) {
653 my $refs = "ref" . ($h{targ} != 1 ? "s" : "");
654 $h{targarglife} = $h{targarg} = "$h{targ} $refs";
655 }
c99ca59a 656 } elsif ($h{targ}) {
657 my $padname = (($curcv->PADLIST->ARRAY)[0]->ARRAY)[$h{targ}];
658 if (defined $padname and class($padname) ne "SPECIAL") {
0b40bd6d 659 $h{targarg} = $padname->PVX;
127212b2 660 if ($padname->FLAGS & SVf_FAKE) {
4ac6efe6 661 if ($] < 5.009) {
662 $h{targarglife} = "$h{targarg}:FAKE";
663 } else {
664 # These changes relate to the jumbo closure fix.
665 # See changes 19939 and 20005
666 my $fake = '';
667 $fake .= 'a' if $padname->IVX & 1; # PAD_FAKELEX_ANON
668 $fake .= 'm' if $padname->IVX & 2; # PAD_FAKELEX_MULTI
669 $fake .= ':' . $padname->NVX if $curcv->CvFLAGS & CVf_ANON;
670 $h{targarglife} = "$h{targarg}:FAKE:$fake";
671 }
127212b2 672 }
673 else {
674 my $intro = $padname->NVX - $cop_seq_base;
675 my $finish = int($padname->IVX) - $cop_seq_base;
676 $finish = "end" if $finish == 999999999 - $cop_seq_base;
677 $h{targarglife} = "$h{targarg}:$intro,$finish";
678 }
c99ca59a 679 } else {
680 $h{targarglife} = $h{targarg} = "t" . $h{targ};
681 }
682 }
683 $h{arg} = "";
684 $h{svclass} = $h{svaddr} = $h{svval} = "";
685 if ($h{class} eq "PMOP") {
686 my $precomp = $op->precomp;
7a9b44b9 687 if (defined $precomp) {
c27ea44e 688 $precomp = cstring($precomp); # Escape literal control sequences
689 $precomp = "/$precomp/";
690 } else {
691 $precomp = "";
7a9b44b9 692 }
b2a3cfdd 693 my $pmreplroot = $op->pmreplroot;
34a48b4b 694 my $pmreplstart;
c6e79e55 695 if (ref($pmreplroot) eq "B::GV") {
b2a3cfdd 696 # with C<@stash_array = split(/pat/, str);>,
c6e79e55 697 # *stash_array is stored in /pat/'s pmreplroot.
b2a3cfdd 698 $h{arg} = "($precomp => \@" . $pmreplroot->NAME . ")";
c6e79e55 699 } elsif (!ref($pmreplroot) and $pmreplroot) {
700 # same as the last case, except the value is actually a
701 # pad offset for where the GV is kept (this happens under
702 # ithreads)
703 my $gv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$pmreplroot];
704 $h{arg} = "($precomp => \@" . $gv->NAME . ")";
b2a3cfdd 705 } elsif ($ {$op->pmreplstart}) {
c99ca59a 706 undef $lastnext;
707 $pmreplstart = "replstart->" . seq($op->pmreplstart);
708 $h{arg} = "(" . join(" ", $precomp, $pmreplstart) . ")";
709 } else {
710 $h{arg} = "($precomp)";
711 }
712 } elsif ($h{class} eq "PVOP" and $h{name} ne "trans") {
713 $h{arg} = '("' . $op->pv . '")';
714 $h{svval} = '"' . $op->pv . '"';
715 } elsif ($h{class} eq "COP") {
716 my $label = $op->label;
c3caa09d 717 $h{coplabel} = $label;
c99ca59a 718 $label = $label ? "$label: " : "";
719 my $loc = $op->file;
720 $loc =~ s[.*/][];
721 $loc .= ":" . $op->line;
722 my($stash, $cseq) = ($op->stash->NAME, $op->cop_seq - $cop_seq_base);
723 my $arybase = $op->arybase;
724 $arybase = $arybase ? ' $[=' . $arybase : "";
725 $h{arg} = "($label$stash $cseq $loc$arybase)";
726 } elsif ($h{class} eq "LOOP") {
727 $h{arg} = "(next->" . seq($op->nextop) . " last->" . seq($op->lastop)
728 . " redo->" . seq($op->redoop) . ")";
729 } elsif ($h{class} eq "LOGOP") {
730 undef $lastnext;
731 $h{arg} = "(other->" . seq($op->other) . ")";
732 } elsif ($h{class} eq "SVOP") {
6a077020 733 unless ($h{name} eq 'aelemfast' and $op->flags & OPf_SPECIAL) {
734 if (! ${$op->sv}) {
735 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->targ];
736 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
737 $h{targarglife} = $h{targarg} = "";
738 } else {
739 $h{arg} = "(" . concise_sv($op->sv, \%h) . ")";
740 }
c99ca59a 741 }
31b49ad4 742 } elsif ($h{class} eq "PADOP") {
743 my $sv = (($curcv->PADLIST->ARRAY)[1]->ARRAY)[$op->padix];
744 $h{arg} = "[" . concise_sv($sv, \%h) . "]";
c99ca59a 745 }
746 $h{seq} = $h{hyphseq} = seq($op);
747 $h{seq} = "" if $h{seq} eq "-";
7252851f 748 if ($] > 5.009) {
749 $h{opt} = $op->opt;
750 $h{static} = $op->static;
751 $h{label} = $labels{$$op};
752 } else {
753 $h{seqnum} = $op->seq;
754 $h{label} = $labels{$op->seq};
755 }
c99ca59a 756 $h{next} = $op->next;
757 $h{next} = (class($h{next}) eq "NULL") ? "(end)" : seq($h{next});
758 $h{nextaddr} = sprintf("%#x", $ {$op->next});
759 $h{sibaddr} = sprintf("%#x", $ {$op->sibling});
760 $h{firstaddr} = sprintf("%#x", $ {$op->first}) if $op->can("first");
761 $h{lastaddr} = sprintf("%#x", $ {$op->last}) if $op->can("last");
762
763 $h{classsym} = $opclass{$h{class}};
764 $h{flagval} = $op->flags;
765 $h{flags} = op_flags($op->flags);
766 $h{privval} = $op->private;
767 $h{private} = private_flags($h{name}, $op->private);
768 $h{addr} = sprintf("%#x", $$op);
c99ca59a 769 $h{typenum} = $op->type;
770 $h{noise} = $linenoise[$op->type];
f95e3c3c 771
cc02ea56 772 return fmt_line(\%h, $op, $format, $level);
c99ca59a 773}
774
775sub B::OP::concise {
776 my($op, $level) = @_;
777 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
724aa791 778 # insert a 'goto' line
cc02ea56 779 my $synth = {"seq" => seq($lastnext), "class" => class($lastnext),
780 "addr" => sprintf("%#x", $$lastnext),
781 "goto" => seq($lastnext), # simplify goto '-' removal
782 };
783 print $walkHandle fmt_line($synth, $op, $gotofmt, $level+1);
c99ca59a 784 }
785 $lastnext = $op->next;
f95e3c3c 786 print $walkHandle concise_op($op, $level, $format);
c99ca59a 787}
788
31b49ad4 789# B::OP::terse (see Terse.pm) now just calls this
790sub b_terse {
791 my($op, $level) = @_;
792
793 # This isn't necessarily right, but there's no easy way to get
794 # from an OP to the right CV. This is a limitation of the
795 # ->terse() interface style, and there isn't much to do about
796 # it. In particular, we can die in concise_op if the main pad
797 # isn't long enough, or has the wrong kind of entries, compared to
798 # the pad a sub was compiled with. The fix for that would be to
799 # make a backwards compatible "terse" format that never even
800 # looked at the pad, just like the old B::Terse. I don't think
801 # that's worth the effort, though.
802 $curcv = main_cv unless $curcv;
803
804 if ($order eq "exec" and $lastnext and $$lastnext != $$op) {
724aa791 805 # insert a 'goto'
31b49ad4 806 my $h = {"seq" => seq($lastnext), "class" => class($lastnext),
807 "addr" => sprintf("%#x", $$lastnext)};
cc02ea56 808 print # $walkHandle
809 fmt_line($h, $op, $style{"terse"}[1], $level+1);
31b49ad4 810 }
811 $lastnext = $op->next;
cc02ea56 812 print # $walkHandle
813 concise_op($op, $level, $style{"terse"}[0]);
31b49ad4 814}
815
c99ca59a 816sub tree {
817 my $op = shift;
818 my $level = shift;
819 my $style = $tree_decorations[$tree_style];
820 my($space, $single, $kids, $kid, $nokid, $last, $lead, $size) = @$style;
821 my $name = concise_op($op, $level, $treefmt);
822 if (not $op->flags & OPf_KIDS) {
823 return $name . "\n";
824 }
825 my @lines;
826 for (my $kid = $op->first; $$kid; $kid = $kid->sibling) {
827 push @lines, tree($kid, $level+1);
828 }
829 my $i;
830 for ($i = $#lines; substr($lines[$i], 0, 1) eq " "; $i--) {
831 $lines[$i] = $space . $lines[$i];
832 }
833 if ($i > 0) {
834 $lines[$i] = $last . $lines[$i];
835 while ($i-- > 1) {
836 if (substr($lines[$i], 0, 1) eq " ") {
837 $lines[$i] = $nokid . $lines[$i];
838 } else {
f95e3c3c 839 $lines[$i] = $kid . $lines[$i];
c99ca59a 840 }
841 }
842 $lines[$i] = $kids . $lines[$i];
843 } else {
844 $lines[0] = $single . $lines[0];
845 }
846 return("$name$lead" . shift @lines,
847 map(" " x (length($name)+$size) . $_, @lines));
848}
849
213a1a26 850# *** Warning: fragile kludge ahead ***
851# Because the B::* modules run in the same interpreter as the code
2814eb74 852# they're compiling, their presence tends to distort the view we have of
853# the code we're looking at. In particular, perl gives sequence numbers
854# to COPs. If the program we're looking at were run on its own, this
855# would start at 1. Because all of B::Concise and all the modules it
856# uses are compiled first, though, by the time we get to the user's
857# program the sequence number is already pretty high, which could be
858# distracting if you're trying to tell OPs apart. Therefore we'd like to
859# subtract an offset from all the sequence numbers we display, to
860# restore the simpler view of the world. The trick is to know what that
861# offset will be, when we're still compiling B::Concise! If we
213a1a26 862# hardcoded a value, it would have to change every time B::Concise or
2814eb74 863# other modules we use do. To help a little, what we do here is compile
864# a little code at the end of the module, and compute the base sequence
865# number for the user's program as being a small offset later, so all we
866# have to worry about are changes in the offset.
7252851f 867
868# [For 5.8.x and earlier perl is generating sequence numbers for all ops,
869# and using them to reference labels]
870
871
213a1a26 872# When you say "perl -MO=Concise -e '$a'", the output should look like:
873
874# 4 <@> leave[t1] vKP/REFC ->(end)
875# 1 <0> enter ->2
876 #^ smallest OP sequence number should be 1
877# 2 <;> nextstate(main 1 -e:1) v ->3
878 # ^ smallest COP sequence number should be 1
879# - <1> ex-rv2sv vK/1 ->4
880# 3 <$> gvsv(*a) s ->4
881
c27ea44e 882# If the second of the marked numbers there isn't 1, it means you need
883# to update the corresponding magic number in the next line.
884# Remember, this needs to stay the last things in the module.
e69a2255 885
c27ea44e 886# Why is this different for MacOS? Does it matter?
8ec8fbef 887my $cop_seq_mnum = $^O eq 'MacOS' ? 12 : 11;
e69a2255 888$cop_seq_base = svref_2object(eval 'sub{0;}')->START->cop_seq + $cop_seq_mnum;
c99ca59a 889
8901;
891
892__END__
893
894=head1 NAME
895
896B::Concise - Walk Perl syntax tree, printing concise info about ops
897
898=head1 SYNOPSIS
899
900 perl -MO=Concise[,OPTIONS] foo.pl
901
78ad9108 902 use B::Concise qw(set_style add_callback);
903
c99ca59a 904=head1 DESCRIPTION
905
906This compiler backend prints the internal OPs of a Perl program's syntax
907tree in one of several space-efficient text formats suitable for debugging
908the inner workings of perl or other compiler backends. It can print OPs in
909the order they appear in the OP tree, in the order they will execute, or
910in a text approximation to their tree structure, and the format of the
911information displyed is customizable. Its function is similar to that of
912perl's B<-Dx> debugging flag or the B<B::Terse> module, but it is more
913sophisticated and flexible.
914
f8a679e6 915=head1 EXAMPLE
916
19e169bf 917Here's an example of 2 outputs (aka 'renderings'), using the
918-exec and -basic (i.e. default) formatting conventions on the same code
919snippet.
920
921 % perl -MO=Concise,-exec -e '$a = $b + 42'
922 1 <0> enter
923 2 <;> nextstate(main 1 -e:1) v
924 3 <#> gvsv[*b] s
925 4 <$> const[IV 42] s
926 * 5 <2> add[t3] sK/2
927 6 <#> gvsv[*a] s
928 7 <2> sassign vKS/2
929 8 <@> leave[1 ref] vKP/REFC
930
931Each line corresponds to an opcode. The opcode marked with '*' is used
932in a few examples below.
933
934The 1st column is the op's sequence number, starting at 1, and is
935displayed in base 36 by default. This rendering is in -exec (i.e.
936execution) order.
937
938The symbol between angle brackets indicates the op's type, for
939example; <2> is a BINOP, <@> a LISTOP, and <#> is a PADOP, which is
940used in threaded perls. (see L</"OP class abbreviations">).
941
942The opname, as in B<'add[t1]'>, which may be followed by op-specific
943information in parentheses or brackets (ex B<'[t1]'>).
944
945The op-flags (ex B<'sK/2'>) follow, and are described in (L</"OP flags
946abbreviations">).
f8a679e6 947
948 % perl -MO=Concise -e '$a = $b + 42'
8ec8fbef 949 8 <@> leave[1 ref] vKP/REFC ->(end)
f8a679e6 950 1 <0> enter ->2
951 2 <;> nextstate(main 1 -e:1) v ->3
952 7 <2> sassign vKS/2 ->8
19e169bf 953 * 5 <2> add[t1] sK/2 ->6
f8a679e6 954 - <1> ex-rv2sv sK/1 ->4
955 3 <$> gvsv(*b) s ->4
956 4 <$> const(IV 42) s ->5
957 - <1> ex-rv2sv sKRM*/1 ->7
958 6 <$> gvsv(*a) s ->7
959
19e169bf 960The default rendering is top-down, so they're not in execution order.
961This form reflects the way the stack is used to parse and evaluate
962expressions; the add operates on the two terms below it in the tree.
f8a679e6 963
19e169bf 964Nullops appear as C<ex-opname>, where I<opname> is an op that has been
965optimized away by perl. They're displayed with a sequence-number of
966'-', because they are not executed (they don't appear in previous
967example), they're printed here because they reflect the parse.
f8a679e6 968
19e169bf 969The arrow points to the sequence number of the next op; they're not
970displayed in -exec mode, for obvious reasons.
f8a679e6 971
19e169bf 972Note that because this rendering was done on a non-threaded perl, the
973PADOPs in the previous examples are now SVOPs, and some (but not all)
974of the square brackets have been replaced by round ones. This is a
975subtle feature to provide some visual distinction between renderings
976on threaded and un-threaded perls.
f8a679e6 977
f8a679e6 978
c99ca59a 979=head1 OPTIONS
980
981Arguments that don't start with a hyphen are taken to be the names of
8ec8fbef 982subroutines to print the OPs of; if no such functions are specified,
983the main body of the program (outside any subroutines, and not
19e169bf 984including use'd or require'd files) is rendered. Passing C<BEGIN>,
8ec8fbef 985C<CHECK>, C<INIT>, or C<END> will cause all of the corresponding
986special blocks to be printed.
c99ca59a 987
724aa791 988Options affect how things are rendered (ie printed). They're presented
989here by their visual effect, 1st being strongest. They're grouped
990according to how they interrelate; within each group the options are
991mutually exclusive (unless otherwise stated).
992
993=head2 Options for Opcode Ordering
994
995These options control the 'vertical display' of opcodes. The display
996'order' is also called 'mode' elsewhere in this document.
997
c99ca59a 998=over 4
999
1000=item B<-basic>
1001
1002Print OPs in the order they appear in the OP tree (a preorder
1003traversal, starting at the root). The indentation of each OP shows its
19e169bf 1004level in the tree, and the '->' at the end of the line indicates the
1005next opcode in execution order. This mode is the default, so the flag
1006is included simply for completeness.
c99ca59a 1007
1008=item B<-exec>
1009
1010Print OPs in the order they would normally execute (for the majority
1011of constructs this is a postorder traversal of the tree, ending at the
1012root). In most cases the OP that usually follows a given OP will
1013appear directly below it; alternate paths are shown by indentation. In
1014cases like loops when control jumps out of a linear path, a 'goto'
1015line is generated.
1016
1017=item B<-tree>
1018
1019Print OPs in a text approximation of a tree, with the root of the tree
1020at the left and 'left-to-right' order of children transformed into
1021'top-to-bottom'. Because this mode grows both to the right and down,
1022it isn't suitable for large programs (unless you have a very wide
1023terminal).
1024
724aa791 1025=back
1026
1027=head2 Options for Line-Style
1028
1029These options select the line-style (or just style) used to render
1030each opcode, and dictates what info is actually printed into each line.
1031
1032=over 4
1033
1034=item B<-concise>
1035
1036Use the author's favorite set of formatting conventions. This is the
1037default, of course.
1038
1039=item B<-terse>
1040
1041Use formatting conventions that emulate the output of B<B::Terse>. The
1042basic mode is almost indistinguishable from the real B<B::Terse>, and the
1043exec mode looks very similar, but is in a more logical order and lacks
1044curly brackets. B<B::Terse> doesn't have a tree mode, so the tree mode
1045is only vaguely reminiscent of B<B::Terse>.
1046
1047=item B<-linenoise>
1048
1049Use formatting conventions in which the name of each OP, rather than being
1050written out in full, is represented by a one- or two-character abbreviation.
1051This is mainly a joke.
1052
1053=item B<-debug>
1054
1055Use formatting conventions reminiscent of B<B::Debug>; these aren't
1056very concise at all.
1057
1058=item B<-env>
1059
1060Use formatting conventions read from the environment variables
1061C<B_CONCISE_FORMAT>, C<B_CONCISE_GOTO_FORMAT>, and C<B_CONCISE_TREE_FORMAT>.
1062
1063=back
1064
1065=head2 Options for tree-specific formatting
1066
1067=over 4
1068
c99ca59a 1069=item B<-compact>
1070
1071Use a tree format in which the minimum amount of space is used for the
1072lines connecting nodes (one character in most cases). This squeezes out
1073a few precious columns of screen real estate.
1074
1075=item B<-loose>
1076
1077Use a tree format that uses longer edges to separate OP nodes. This format
1078tends to look better than the compact one, especially in ASCII, and is
1079the default.
1080
1081=item B<-vt>
1082
1083Use tree connecting characters drawn from the VT100 line-drawing set.
1084This looks better if your terminal supports it.
1085
1086=item B<-ascii>
1087
1088Draw the tree with standard ASCII characters like C<+> and C<|>. These don't
1089look as clean as the VT100 characters, but they'll work with almost any
1090terminal (or the horizontal scrolling mode of less(1)) and are suitable
1091for text documentation or email. This is the default.
1092
724aa791 1093=back
c99ca59a 1094
724aa791 1095These are pairwise exclusive, i.e. compact or loose, vt or ascii.
1096
1097=head2 Options controlling sequence numbering
1098
1099=over 4
c99ca59a 1100
1101=item B<-base>I<n>
1102
1103Print OP sequence numbers in base I<n>. If I<n> is greater than 10, the
1104digit for 11 will be 'a', and so on. If I<n> is greater than 36, the digit
1105for 37 will be 'A', and so on until 62. Values greater than 62 are not
1106currently supported. The default is 36.
1107
1108=item B<-bigendian>
1109
1110Print sequence numbers with the most significant digit first. This is the
1111usual convention for Arabic numerals, and the default.
1112
1113=item B<-littleendian>
1114
724aa791 1115Print seqence numbers with the least significant digit first. This is
1116obviously mutually exclusive with bigendian.
c99ca59a 1117
724aa791 1118=back
c99ca59a 1119
724aa791 1120=head2 Other options
c99ca59a 1121
cc02ea56 1122These are pairwise exclusive.
1123
724aa791 1124=over 4
c99ca59a 1125
724aa791 1126=item B<-main>
c99ca59a 1127
724aa791 1128Include the main program in the output, even if subroutines were also
cc02ea56 1129specified. This rendering is normally suppressed when a subroutine
1130name or reference is given.
1131
1132=item B<-nomain>
1133
1134This restores the default behavior after you've changed it with '-main'
1135(it's not normally needed). If no subroutine name/ref is given, main is
1136rendered, regardless of this flag.
1137
1138=item B<-nobanner>
1139
1140Renderings usually include a banner line identifying the function name
1141or stringified subref. This suppresses the printing of the banner.
1142
1143TBC: Remove the stringified coderef; while it provides a 'cookie' for
1144each function rendered, the cookies used should be 1,2,3.. not a
1145random hex-address. It also complicates string comparison of two
1146different trees.
c99ca59a 1147
724aa791 1148=item B<-banner>
c99ca59a 1149
cc02ea56 1150restores default banner behavior.
1151
1152=item B<-banneris> => subref
1153
1154TBC: a hookpoint (and an option to set it) for a user-supplied
1155function to produce a banner appropriate for users needs. It's not
1156ideal, because the rendering-state variables, which are a natural
1157candidate for use in concise.t, are unavailable to the user.
c99ca59a 1158
724aa791 1159=back
c99ca59a 1160
724aa791 1161=head2 Option Stickiness
c99ca59a 1162
724aa791 1163If you invoke Concise more than once in a program, you should know that
1164the options are 'sticky'. This means that the options you provide in
1165the first call will be remembered for the 2nd call, unless you
1166re-specify or change them.
c99ca59a 1167
cc02ea56 1168=head1 ABBREVIATIONS
1169
1170The concise style uses symbols to convey maximum info with minimal
1171clutter (like hex addresses). With just a little practice, you can
1172start to see the flowers, not just the branches, in the trees.
1173
1174=head2 OP class abbreviations
1175
1176These symbols appear before the op-name, and indicate the
1177B:: namespace that represents the ops in your Perl code.
1178
1179 0 OP (aka BASEOP) An OP with no children
1180 1 UNOP An OP with one child
1181 2 BINOP An OP with two children
1182 | LOGOP A control branch OP
1183 @ LISTOP An OP that could have lots of children
1184 / PMOP An OP with a regular expression
1185 $ SVOP An OP with an SV
1186 " PVOP An OP with a string
1187 { LOOP An OP that holds pointers for a loop
1188 ; COP An OP that marks the start of a statement
1189 # PADOP An OP with a GV on the pad
1190
1191=head2 OP flags abbreviations
1192
19e169bf 1193OP flags are either public or private. The public flags alter the
1194behavior of each opcode in consistent ways, and are represented by 0
1195or more single characters.
cc02ea56 1196
1197 v OPf_WANT_VOID Want nothing (void context)
1198 s OPf_WANT_SCALAR Want single value (scalar context)
1199 l OPf_WANT_LIST Want list of any length (list context)
19e169bf 1200 Want is unknown
cc02ea56 1201 K OPf_KIDS There is a firstborn child.
1202 P OPf_PARENS This operator was parenthesized.
1203 (Or block needs explicit scope entry.)
1204 R OPf_REF Certified reference.
1205 (Return container, not containee).
1206 M OPf_MOD Will modify (lvalue).
1207 S OPf_STACKED Some arg is arriving on the stack.
1208 * OPf_SPECIAL Do something weird for this op (see op.h)
1209
19e169bf 1210Private flags, if any are set for an opcode, are displayed after a '/'
1211
1212 8 <@> leave[1 ref] vKP/REFC ->(end)
1213 7 <2> sassign vKS/2 ->8
1214
1215They're opcode specific, and occur less often than the public ones, so
1216they're represented by short mnemonics instead of single-chars; see
1217L<op.h> for gory details, or try this quick 2-liner:
1218
1219 $> perl -MB::Concise -de 1
1220 DB<1> |x \%B::Concise::priv
1221
c99ca59a 1222=head1 FORMATTING SPECIFICATIONS
1223
724aa791 1224For each line-style ('concise', 'terse', 'linenoise', etc.) there are
12253 format-specs which control how OPs are rendered.
1226
1227The first is the 'default' format, which is used in both basic and exec
1228modes to print all opcodes. The 2nd, goto-format, is used in exec
1229mode when branches are encountered. They're not real opcodes, and are
1230inserted to look like a closing curly brace. The tree-format is tree
1231specific.
1232
cc02ea56 1233When a line is rendered, the correct format-spec is copied and scanned
1234for the following items; data is substituted in, and other
1235manipulations like basic indenting are done, for each opcode rendered.
1236
1237There are 3 kinds of items that may be populated; special patterns,
1238#vars, and literal text, which is copied verbatim. (Yes, it's a set
1239of s///g steps.)
1240
1241=head2 Special Patterns
1242
1243These items are the primitives used to perform indenting, and to
1244select text from amongst alternatives.
c99ca59a 1245
1246=over 4
1247
1248=item B<(x(>I<exec_text>B<;>I<basic_text>B<)x)>
1249
1250Generates I<exec_text> in exec mode, or I<basic_text> in basic mode.
1251
1252=item B<(*(>I<text>B<)*)>
1253
1254Generates one copy of I<text> for each indentation level.
1255
1256=item B<(*(>I<text1>B<;>I<text2>B<)*)>
1257
1258Generates one fewer copies of I<text1> than the indentation level, followed
1259by one copy of I<text2> if the indentation level is more than 0.
1260
1261=item B<(?(>I<text1>B<#>I<var>I<Text2>B<)?)>
1262
1263If the value of I<var> is true (not empty or zero), generates the
1264value of I<var> surrounded by I<text1> and I<Text2>, otherwise
1265nothing.
1266
cc02ea56 1267=item B<~>
1268
1269Any number of tildes and surrounding whitespace will be collapsed to
1270a single space.
1271
1272=back
1273
1274=head2 # Variables
1275
1276These #vars represent opcode properties that you may want as part of
1277your rendering. The '#' is intended as a private sigil; a #var's
1278value is interpolated into the style-line, much like "read $this".
1279
1280These vars take 3 forms:
1281
1282=over 4
1283
c99ca59a 1284=item B<#>I<var>
1285
cc02ea56 1286A property named 'var' is assumed to exist for the opcodes, and is
1287interpolated into the rendering.
c99ca59a 1288
1289=item B<#>I<var>I<N>
1290
cc02ea56 1291Generates the value of I<var>, left justified to fill I<N> spaces.
1292Note that this means while you can have properties 'foo' and 'foo2',
1293you cannot render 'foo2', but you could with 'foo2a'. You would be
1294wise not to rely on this behavior going forward ;-)
c99ca59a 1295
cc02ea56 1296=item B<#>I<Var>
c99ca59a 1297
cc02ea56 1298This ucfirst form of #var generates a tag-value form of itself for
1299display; it converts '#Var' into a 'Var => #var' style, which is then
1300handled as described above. (Imp-note: #Vars cannot be used for
1301conditional-fills, because the => #var transform is done after the check
1302for #Var's value).
c99ca59a 1303
1304=back
1305
cc02ea56 1306The following variables are 'defined' by B::Concise; when they are
1307used in a style, their respective values are plugged into the
1308rendering of each opcode.
1309
1310Only some of these are used by the standard styles, the others are
1311provided for you to delve into optree mechanics, should you wish to
1312add a new style (see L</add_style> below) that uses them. You can
1313also add new ones using L<add_callback>.
c99ca59a 1314
1315=over 4
1316
1317=item B<#addr>
1318
cc02ea56 1319The address of the OP, in hexadecimal.
c99ca59a 1320
1321=item B<#arg>
1322
1323The OP-specific information of the OP (such as the SV for an SVOP, the
cc02ea56 1324non-local exit pointers for a LOOP, etc.) enclosed in parentheses.
c99ca59a 1325
1326=item B<#class>
1327
1328The B-determined class of the OP, in all caps.
1329
f8a679e6 1330=item B<#classsym>
c99ca59a 1331
1332A single symbol abbreviating the class of the OP.
1333
c3caa09d 1334=item B<#coplabel>
1335
1336The label of the statement or block the OP is the start of, if any.
1337
c99ca59a 1338=item B<#exname>
1339
1340The name of the OP, or 'ex-foo' if the OP is a null that used to be a foo.
1341
1342=item B<#extarg>
1343
1344The target of the OP, or nothing for a nulled OP.
1345
1346=item B<#firstaddr>
1347
19e169bf 1348The address of the OP's first child, in hexadecimal.
c99ca59a 1349
1350=item B<#flags>
1351
1352The OP's flags, abbreviated as a series of symbols.
1353
1354=item B<#flagval>
1355
1356The numeric value of the OP's flags.
1357
f8a679e6 1358=item B<#hyphseq>
c99ca59a 1359
1360The sequence number of the OP, or a hyphen if it doesn't have one.
1361
1362=item B<#label>
1363
1364'NEXT', 'LAST', or 'REDO' if the OP is a target of one of those in exec
1365mode, or empty otherwise.
1366
1367=item B<#lastaddr>
1368
19e169bf 1369The address of the OP's last child, in hexadecimal.
c99ca59a 1370
1371=item B<#name>
1372
1373The OP's name.
1374
1375=item B<#NAME>
1376
1377The OP's name, in all caps.
1378
1379=item B<#next>
1380
1381The sequence number of the OP's next OP.
1382
1383=item B<#nextaddr>
1384
19e169bf 1385The address of the OP's next OP, in hexadecimal.
c99ca59a 1386
1387=item B<#noise>
1388
c27ea44e 1389A one- or two-character abbreviation for the OP's name.
c99ca59a 1390
1391=item B<#private>
1392
1393The OP's private flags, rendered with abbreviated names if possible.
1394
1395=item B<#privval>
1396
1397The numeric value of the OP's private flags.
1398
1399=item B<#seq>
1400
2814eb74 1401The sequence number of the OP. Note that this is a sequence number
1402generated by B::Concise.
c99ca59a 1403
7252851f 1404=item B<#seqnum>
1405
14065.8.x and earlier only. 5.9 and later do not provide this.
1407
1408The real sequence number of the OP, as a regular number and not adjusted
1409to be relative to the start of the real program. (This will generally be
1410a fairly large number because all of B<B::Concise> is compiled before
1411your program is).
1412
2814eb74 1413=item B<#opt>
c99ca59a 1414
2814eb74 1415Whether or not the op has been optimised by the peephole optimiser.
1416
7252851f 1417Only available in 5.9 and later.
1418
2814eb74 1419=item B<#static>
1420
1421Whether or not the op is statically defined. This flag is used by the
1422B::C compiler backend and indicates that the op should not be freed.
c99ca59a 1423
7252851f 1424Only available in 5.9 and later.
1425
c99ca59a 1426=item B<#sibaddr>
1427
19e169bf 1428The address of the OP's next youngest sibling, in hexadecimal.
c99ca59a 1429
1430=item B<#svaddr>
1431
19e169bf 1432The address of the OP's SV, if it has an SV, in hexadecimal.
c99ca59a 1433
1434=item B<#svclass>
1435
1436The class of the OP's SV, if it has one, in all caps (e.g., 'IV').
1437
1438=item B<#svval>
1439
1440The value of the OP's SV, if it has one, in a short human-readable format.
1441
1442=item B<#targ>
1443
1444The numeric value of the OP's targ.
1445
1446=item B<#targarg>
1447
1448The name of the variable the OP's targ refers to, if any, otherwise the
1449letter t followed by the OP's targ in decimal.
1450
1451=item B<#targarglife>
1452
1453Same as B<#targarg>, but followed by the COP sequence numbers that delimit
1454the variable's lifetime (or 'end' for a variable in an open scope) for a
1455variable.
1456
1457=item B<#typenum>
1458
1459The numeric value of the OP's type, in decimal.
1460
1461=back
1462
78ad9108 1463=head1 Using B::Concise outside of the O framework
1464
cc02ea56 1465The common (and original) usage of B::Concise was for command-line
1466renderings of simple code, as given in EXAMPLE. But you can also use
1467B<B::Concise> from your code, and call compile() directly, and
724aa791 1468repeatedly. By doing so, you can avoid the compile-time only
cc02ea56 1469operation of O.pm, and even use the debugger to step through
1470B::Concise::compile() itself.
f95e3c3c 1471
cc02ea56 1472Once you're doing this, you may alter Concise output by adding new
1473rendering styles, and by optionally adding callback routines which
1474populate new variables, if such were referenced from those (just
1475added) styles.
f95e3c3c 1476
724aa791 1477=head2 Example: Altering Concise Renderings
78ad9108 1478
1479 use B::Concise qw(set_style add_callback);
cc02ea56 1480 add_style($yourStyleName => $defaultfmt, $gotofmt, $treefmt);
78ad9108 1481 add_callback
f95e3c3c 1482 ( sub {
1483 my ($h, $op, $format, $level, $stylename) = @_;
78ad9108 1484 $h->{variable} = some_func($op);
cc02ea56 1485 });
1486 $walker = B::Concise::compile(@options,@subnames,@subrefs);
1487 $walker->();
78ad9108 1488
f95e3c3c 1489=head2 set_style()
1490
724aa791 1491B<set_style> accepts 3 arguments, and updates the three format-specs
1492comprising a line-style (basic-exec, goto, tree). It has one minor
1493drawback though; it doesn't register the style under a new name. This
1494can become an issue if you render more than once and switch styles.
1495Thus you may prefer to use add_style() and/or set_style_standard()
1496instead.
1497
1498=head2 set_style_standard($name)
1499
1500This restores one of the standard line-styles: C<terse>, C<concise>,
1501C<linenoise>, C<debug>, C<env>, into effect. It also accepts style
1502names previously defined with add_style().
f95e3c3c 1503
1504=head2 add_style()
78ad9108 1505
f95e3c3c 1506This subroutine accepts a new style name and three style arguments as
1507above, and creates, registers, and selects the newly named style. It is
1508an error to re-add a style; call set_style_standard() to switch between
1509several styles.
1510
f95e3c3c 1511=head2 add_callback()
1512
19e169bf 1513If your newly minted styles refer to any new #variables, you'll need
1514to define a callback subroutine that will populate (or modify) those
1515variables. They are then available for use in the style you've
1516chosen.
f95e3c3c 1517
1518The callbacks are called for each opcode visited by Concise, in the
1519same order as they are added. Each subroutine is passed five
1520parameters.
1521
1522 1. A hashref, containing the variable names and values which are
1523 populated into the report-line for the op
1524 2. the op, as a B<B::OP> object
1525 3. a reference to the format string
1526 4. the formatting (indent) level
1527 5. the selected stylename
78ad9108 1528
1529To define your own variables, simply add them to the hash, or change
1530existing values if you need to. The level and format are passed in as
1531references to scalars, but it is unlikely that they will need to be
1532changed or even used.
1533
724aa791 1534=head2 Running B::Concise::compile()
f95e3c3c 1535
1536B<compile> accepts options as described above in L</OPTIONS>, and
1537arguments, which are either coderefs, or subroutine names.
1538
cc02ea56 1539It constructs and returns a $treewalker coderef, which when invoked,
1540traverses, or walks, and renders the optrees of the given arguments to
1541STDOUT. You can reuse this, and can change the rendering style used
1542each time; thereafter the coderef renders in the new style.
f95e3c3c 1543
1544B<walk_output> lets you change the print destination from STDOUT to
19e169bf 1545another open filehandle, or into a string passed as a ref (unless
1546you've built perl with -Uuseperlio).
f95e3c3c 1547
cc02ea56 1548 my $walker = B::Concise::compile('-terse','aFuncName', \&aSubRef); # 1
f95e3c3c 1549 walk_output(\my $buf);
cc02ea56 1550 $walker->(); # 1 renders -terse
1551 set_style_standard('concise'); # 2
1552 $walker->(); # 2 renders -concise
1553 $walker->(@new); # 3 renders whatever
1554 print "3 different renderings: terse, concise, and @new: $buf\n";
1555
1556When $walker is called, it traverses the subroutines supplied when it
1557was created, and renders them using the current style. You can change
1558the style afterwards in several different ways:
1559
1560 1. call C<compile>, altering style or mode/order
1561 2. call C<set_style_standard>
1562 3. call $walker, passing @new options
1563
1564Passing new options to the $walker is the easiest way to change
1565amongst any pre-defined styles (the ones you add are automatically
1566recognized as options), and is the only way to alter rendering order
1567without calling compile again. Note however that rendering state is
1568still shared amongst multiple $walker objects, so they must still be
1569used in a coordinated manner.
f95e3c3c 1570
1571=head2 B::Concise::reset_sequence()
1572
1573This function (not exported) lets you reset the sequence numbers (note
1574that they're numbered arbitrarily, their goal being to be human
1575readable). Its purpose is mostly to support testing, i.e. to compare
1576the concise output from two identical anonymous subroutines (but
1577different instances). Without the reset, B::Concise, seeing that
1578they're separate optrees, generates different sequence numbers in
1579the output.
1580
1581=head2 Errors
1582
1583All detected errors, (invalid arguments, internal errors, etc.) are
1584resolved with a die($message). Use an eval if you wish to catch these
1585errors and continue processing.
31b49ad4 1586
724aa791 1587In particular, B<compile> will die if you've asked for a non-existent
1588function-name, a non-existent coderef, or a non-CODE reference.
78ad9108 1589
c99ca59a 1590=head1 AUTHOR
1591
31b49ad4 1592Stephen McCamant, E<lt>smcc@CSUA.Berkeley.EDUE<gt>.
c99ca59a 1593
1594=cut