3 # Copyright (c) 1996, 1997, 1998 Malcolm Beattie
5 # You may distribute under the terms of either the GNU General Public
6 # License or the Artistic License, as specified in the README file.
10 our $VERSION = '1.23';
16 # walkoptree_slow comes from B.pm (you are there),
17 # walkoptree comes from B.xs
18 @EXPORT_OK = qw(minus_c ppname save_BEGINs
19 class peekop cast_I32 cstring cchar hash threadsv_names
20 main_root main_start main_cv svref_2object opnumber
21 sub_generation amagic_generation perlstring
22 walkoptree_slow walkoptree walkoptree_exec walksymtable
23 parents comppadlist sv_undef compile_stats timing_info
24 begin_av init_av check_av end_av regex_padav dowarn defstash
25 curstash warnhook diehook inc_gv @optype @specialsv_name
27 push @EXPORT_OK, qw(unitcheck_av) if $] > 5.009;
31 @B::SV::ISA = 'B::OBJECT';
32 @B::NULL::ISA = 'B::SV';
33 @B::PV::ISA = 'B::SV';
34 @B::IV::ISA = 'B::SV';
35 @B::NV::ISA = 'B::SV';
36 # RV is eliminated with 5.11.0, but effectively is a specialisation of IV now.
37 @B::RV::ISA = $] >= 5.011 ? 'B::IV' : 'B::SV';
38 @B::PVIV::ISA = qw(B::PV B::IV);
39 @B::PVNV::ISA = qw(B::PVIV B::NV);
40 @B::PVMG::ISA = 'B::PVNV';
41 @B::REGEXP::ISA = 'B::PVMG' if $] >= 5.011;
42 # Change in the inheritance hierarchy post 5.9.0
43 @B::PVLV::ISA = $] > 5.009 ? 'B::GV' : 'B::PVMG';
44 # BM is eliminated post 5.9.5, but effectively is a specialisation of GV now.
45 @B::BM::ISA = $] > 5.009005 ? 'B::GV' : 'B::PVMG';
46 @B::AV::ISA = 'B::PVMG';
47 @B::GV::ISA = 'B::PVMG';
48 @B::HV::ISA = 'B::PVMG';
49 @B::CV::ISA = 'B::PVMG';
50 @B::IO::ISA = 'B::PVMG';
51 @B::FM::ISA = 'B::CV';
53 @B::OP::ISA = 'B::OBJECT';
54 @B::UNOP::ISA = 'B::OP';
55 @B::BINOP::ISA = 'B::UNOP';
56 @B::LOGOP::ISA = 'B::UNOP';
57 @B::LISTOP::ISA = 'B::BINOP';
58 @B::SVOP::ISA = 'B::OP';
59 @B::PADOP::ISA = 'B::OP';
60 @B::PVOP::ISA = 'B::OP';
61 @B::LOOP::ISA = 'B::LISTOP';
62 @B::PMOP::ISA = 'B::LISTOP';
63 @B::COP::ISA = 'B::OP';
65 @B::SPECIAL::ISA = 'B::OBJECT';
67 @B::optype = qw(OP UNOP BINOP LOGOP LISTOP PMOP SVOP PADOP PVOP LOOP COP);
68 # bytecode.pl contained the following comment:
69 # Nullsv *must* come first in the following so that the condition
70 # ($$sv == 0) can continue to be used to test (sv == Nullsv).
71 @B::specialsv_name = qw(Nullsv &PL_sv_undef &PL_sv_yes &PL_sv_no
72 (SV*)pWARN_ALL (SV*)pWARN_NONE (SV*)pWARN_STD);
75 # Stop "-w" from complaining about the lack of a real B::OBJECT class
80 my $name = (shift())->NAME;
82 # The regex below corresponds to the isCONTROLVAR macro
85 $name =~ s/^([\cA-\cZ\c\\c[\c]\c?\c_\c^])/"^".
86 chr( utf8::unicode_to_native( 64 ^ ord($1) ))/e;
88 # When we say unicode_to_native we really mean ascii_to_native,
89 # which matters iff this is a non-ASCII platform (EBCDIC).
94 sub B::IV::int_value {
96 return (($self->FLAGS() & SVf_IVisUV()) ? $self->UVX : $self->IV);
99 sub B::NULL::as_string() {""}
100 sub B::IV::as_string() {goto &B::IV::int_value}
101 sub B::PV::as_string() {goto &B::PV::PV}
108 my ($class, $value) = @_;
110 walkoptree_debug($value);
120 sub parents { \@parents }
125 return sprintf("%s (0x%x) %s", class($op), $$op, $op->name);
128 sub walkoptree_slow {
129 my($op, $method, $level) = @_;
130 $op_count++; # just for statistics
132 warn(sprintf("walkoptree: %d. %s\n", $level, peekop($op))) if $debug;
133 $op->$method($level) if $op->can($method);
134 if ($$op && ($op->flags & OPf_KIDS)) {
136 unshift(@parents, $op);
137 for ($kid = $op->first; $$kid; $kid = $kid->sibling) {
138 walkoptree_slow($kid, $method, $level + 1);
142 if (class($op) eq 'PMOP'
143 && ref($op->pmreplroot)
144 && ${$op->pmreplroot}
145 && $op->pmreplroot->isa( 'B::OP' ))
147 unshift(@parents, $op);
148 walkoptree_slow($op->pmreplroot, $method, $level + 1);
154 return "Total number of OPs processed: $op_count\n";
158 my ($sec, $min, $hr) = localtime;
159 my ($user, $sys) = times;
160 sprintf("%02d:%02d:%02d user=$user sys=$sys",
161 $hr, $min, $sec, $user, $sys);
171 my ($obj, $value) = @_;
172 # warn(sprintf("savesym: sym_%x => %s\n", $$obj, $value)); # debug
173 $symtable{sprintf("sym_%x", $$obj)} = $value;
178 return $symtable{sprintf("sym_%x", $$obj)};
181 sub walkoptree_exec {
182 my ($op, $method, $level) = @_;
185 my $prefix = " " x $level;
186 for (; $$op; $op = $op->next) {
189 print $prefix, "goto $sym\n";
192 savesym($op, sprintf("%s (0x%lx)", class($op), $$op));
193 $op->$method($level);
196 /^(d?or(assign)?|and(assign)?|mapwhile|grepwhile|entertry|range|cond_expr)$/)
198 print $prefix, uc($1), " => {\n";
199 walkoptree_exec($op->other, $method, $level + 1);
200 print $prefix, "}\n";
201 } elsif ($ppname eq "match" || $ppname eq "subst") {
202 my $pmreplstart = $op->pmreplstart;
204 print $prefix, "PMREPLSTART => {\n";
205 walkoptree_exec($pmreplstart, $method, $level + 1);
206 print $prefix, "}\n";
208 } elsif ($ppname eq "substcont") {
209 print $prefix, "SUBSTCONT => {\n";
210 walkoptree_exec($op->other->pmreplstart, $method, $level + 1);
211 print $prefix, "}\n";
213 } elsif ($ppname eq "enterloop") {
214 print $prefix, "REDO => {\n";
215 walkoptree_exec($op->redoop, $method, $level + 1);
216 print $prefix, "}\n", $prefix, "NEXT => {\n";
217 walkoptree_exec($op->nextop, $method, $level + 1);
218 print $prefix, "}\n", $prefix, "LAST => {\n";
219 walkoptree_exec($op->lastop, $method, $level + 1);
220 print $prefix, "}\n";
221 } elsif ($ppname eq "subst") {
222 my $replstart = $op->pmreplstart;
224 print $prefix, "SUBST => {\n";
225 walkoptree_exec($replstart, $method, $level + 1);
226 print $prefix, "}\n";
233 my ($symref, $method, $recurse, $prefix) = @_;
238 $prefix = '' unless defined $prefix;
239 while (($sym, $ref) = each %$symref) {
240 $fullname = "*main::".$prefix.$sym;
242 $sym = $prefix . $sym;
243 if (svref_2object(\*$sym)->NAME ne "main::" && $sym ne "<none>::" && &$recurse($sym)) {
244 walksymtable(\%$fullname, $method, $recurse, $sym);
247 svref_2object(\*$fullname)->$method();
258 my ($class, $section, $symtable, $default) = @_;
259 $output_fh ||= FileHandle->new_tmpfile;
260 my $obj = bless [-1, $section, $symtable, $default], $class;
261 $sections{$section} = $obj;
266 my ($class, $section) = @_;
267 return $sections{$section};
272 while (defined($_ = shift)) {
273 print $output_fh "$section->[1]\t$_\n";
280 return $section->[0];
285 return $section->[1];
290 return $section->[2];
295 return $section->[3];
299 my ($section, $fh, $format) = @_;
300 my $name = $section->name;
301 my $sym = $section->symtable || {};
302 my $default = $section->default;
304 seek($output_fh, 0, 0);
305 while (<$output_fh>) {
310 exists($sym->{$1}) ? $sym->{$1} : $default;
312 printf $fh $format, $_;
326 B - The Perl Compiler Backend
334 The C<B> module supplies classes which allow a Perl program to delve
335 into its own innards. It is the module used to implement the
336 "backends" of the Perl compiler. Usage of the compiler does not
337 require knowledge of this module: see the F<O> module for the
338 user-visible part. The C<B> module is of use to those who want to
339 write new compiler backends. This documentation assumes that the
340 reader knows a fair amount about perl's internals including such
341 things as SVs, OPs and the internal symbol table and syntax tree
346 The C<B> module contains a set of utility functions for querying the
347 current state of the Perl interpreter; typically these functions
348 return objects from the B::SV and B::OP classes, or their derived
349 classes. These classes in turn define methods for querying the
350 resulting objects about their own internal state.
352 =head1 Utility Functions
354 The C<B> module exports a variety of functions: some are simple
355 utility functions, others provide a Perl program with a way to
356 get an initial "handle" on an internal object.
358 =head2 Functions Returning C<B::SV>, C<B::AV>, C<B::HV>, and C<B::CV> objects
360 For descriptions of the class hierarchy of these objects and the
361 methods that can be called on them, see below, L<"OVERVIEW OF
362 CLASSES"> and L<"SV-RELATED CLASSES">.
368 Returns the SV object corresponding to the C variable C<sv_undef>.
372 Returns the SV object corresponding to the C variable C<sv_yes>.
376 Returns the SV object corresponding to the C variable C<sv_no>.
378 =item svref_2object(SVREF)
380 Takes a reference to any Perl value, and turns the referred-to value
381 into an object in the appropriate B::OP-derived or B::SV-derived
382 class. Apart from functions such as C<main_root>, this is the primary
383 way to get an initial "handle" on an internal perl data structure
384 which can then be followed with the other access methods.
386 The returned object will only be valid as long as the underlying OPs
387 and SVs continue to exist. Do not attempt to use the object after the
388 underlying structures are freed.
390 =item amagic_generation
392 Returns the SV object corresponding to the C variable C<amagic_generation>.
396 Returns the AV object (i.e. in class B::AV) representing INIT blocks.
400 Returns the AV object (i.e. in class B::AV) representing CHECK blocks.
404 Returns the AV object (i.e. in class B::AV) representing UNITCHECK blocks.
408 Returns the AV object (i.e. in class B::AV) representing BEGIN blocks.
412 Returns the AV object (i.e. in class B::AV) representing END blocks.
416 Returns the AV object (i.e. in class B::AV) of the global comppadlist.
420 Only when perl was compiled with ithreads.
424 Return the (faked) CV corresponding to the main part of the Perl
429 =head2 Functions for Examining the Symbol Table
433 =item walksymtable(SYMREF, METHOD, RECURSE, PREFIX)
435 Walk the symbol table starting at SYMREF and call METHOD on each
436 symbol (a B::GV object) visited. When the walk reaches package
437 symbols (such as "Foo::") it invokes RECURSE, passing in the symbol
438 name, and only recurses into the package if that sub returns true.
440 PREFIX is the name of the SYMREF you're walking.
444 # Walk CGI's symbol table calling print_subs on each symbol.
445 # Recurse only into CGI::Util::
446 walksymtable(\%CGI::, 'print_subs', sub { $_[0] eq 'CGI::Util::' },
449 print_subs() is a B::GV method you have declared. Also see L<"B::GV
454 =head2 Functions Returning C<B::OP> objects or for walking op trees
456 For descriptions of the class hierarchy of these objects and the
457 methods that can be called on them, see below, L<"OVERVIEW OF
458 CLASSES"> and L<"OP-RELATED CLASSES">.
464 Returns the root op (i.e. an object in the appropriate B::OP-derived
465 class) of the main part of the Perl program.
469 Returns the starting op of the main part of the Perl program.
471 =item walkoptree(OP, METHOD)
473 Does a tree-walk of the syntax tree based at OP and calls METHOD on
474 each op it visits. Each node is visited before its children. If
475 C<walkoptree_debug> (see below) has been called to turn debugging on then
476 the method C<walkoptree_debug> is called on each op before METHOD is
479 =item walkoptree_debug(DEBUG)
481 Returns the current debugging flag for C<walkoptree>. If the optional
482 DEBUG argument is non-zero, it sets the debugging flag to that. See
483 the description of C<walkoptree> above for what the debugging flag
488 =head2 Miscellaneous Utility Functions
494 Return the PP function name (e.g. "pp_add") of op number OPNUM.
498 Returns a string in the form "0x..." representing the value of the
499 internal hash function used by perl on string STR.
503 Casts I to the internal I32 type used by that perl.
507 Does the equivalent of the C<-c> command-line option. Obviously, this
508 is only useful in a BEGIN block or else the flag is set too late.
512 Returns a double-quote-surrounded escaped version of STR which can
513 be used as a string in C source code.
515 =item perlstring(STR)
517 Returns a double-quote-surrounded escaped version of STR which can
518 be used as a string in Perl source code.
522 Returns the class of an object without the part of the classname
523 preceding the first C<"::">. This is used to turn C<"B::UNOP"> into
524 C<"UNOP"> for example.
528 In a perl compiled for threads, this returns a list of the special
529 per-thread threadsv variables.
533 =head2 Exported utility variabiles
539 my $op_type = $optype[$op_type_num];
541 A simple mapping of the op type number to its type (like 'COP' or 'BINOP').
543 =item @specialsv_name
545 my $sv_name = $specialsv_name[$sv_index];
547 Certain SV types are considered 'special'. They're represented by
548 B::SPECIAL and are referred to by a number from the specialsv_list.
549 This array maps that number back to the name of the SV (like 'Nullsv'
555 =head1 OVERVIEW OF CLASSES
557 The C structures used by Perl's internals to hold SV and OP
558 information (PVIV, AV, HV, ..., OP, SVOP, UNOP, ...) are modelled on a
559 class hierarchy and the C<B> module gives access to them via a true
560 object hierarchy. Structure fields which point to other objects
561 (whether types of SV or types of OP) are represented by the C<B>
562 module as Perl objects of the appropriate class.
564 The bulk of the C<B> module is the methods for accessing fields of
567 Note that all access is read-only. You cannot modify the internals by
568 using this module. Also, note that the B::OP and B::SV objects created
569 by this module are only valid for as long as the underlying objects
570 exist; their creation doesn't increase the reference counts of the
571 underlying objects. Trying to access the fields of a freed object will
572 give incomprehensible results, or worse.
574 =head2 SV-RELATED CLASSES
576 B::IV, B::NV, B::RV, B::PV, B::PVIV, B::PVNV, B::PVMG, B::BM (5.9.5 and
577 earlier), B::PVLV, B::AV, B::HV, B::CV, B::GV, B::FM, B::IO. These classes
578 correspond in the obvious way to the underlying C structures of similar names.
579 The inheritance hierarchy mimics the underlying C "inheritance". For the
580 5.10.x branch, (I<ie> 5.10.0, 5.10.1 I<etc>) this is:
584 +------------+------------+------------+
586 B::PV B::IV B::NV B::RV
598 +-----+-----+-----+-----+
600 B::AV B::GV B::HV B::CV B::IO
605 For 5.9.0 and earlier, PVLV is a direct subclass of PVMG, and BM is still
606 present as a distinct type, so the base of this diagram is
613 +------+-----+-----+-----+-----+-----+
615 B::PVLV B::BM B::AV B::GV B::HV B::CV B::IO
620 For 5.11.0 and later, B::RV is abolished, and IVs can be used to store
621 references, and a new type B::REGEXP is introduced, giving this structure:
625 +------------+------------+
639 +-------+-------+---+---+-------+-------+
641 B::AV B::GV B::HV B::CV B::IO B::REGEXP
647 Access methods correspond to the underlying C macros for field access,
648 usually with the leading "class indication" prefix removed (Sv, Av,
649 Hv, ...). The leading prefix is only left in cases where its removal
650 would cause a clash in method name. For example, C<GvREFCNT> stays
651 as-is since its abbreviation would clash with the "superclass" method
652 C<REFCNT> (corresponding to the C function C<SvREFCNT>).
664 Returns a reference to the regular scalar corresponding to this
665 B::SV object. In other words, this method is the inverse operation
666 to the svref_2object() subroutine. This scalar and other data it points
667 at should be considered read-only: modifying them is neither safe nor
668 guaranteed to have a sensible effect.
678 Returns the value of the IV, I<interpreted as
679 a signed integer>. This will be misleading
680 if C<FLAGS & SVf_IVisUV>. Perhaps you want the
681 C<int_value> method instead?
689 This method returns the value of the IV as an integer.
690 It differs from C<IV> in that it returns the correct
691 value regardless of whether it's stored signed or
724 This method is the one you usually want. It constructs a
725 string using the length and offset information in the struct:
726 for ordinary scalars it will return the string that you'd see
727 from Perl, even if it contains null characters.
731 Same as B::RV::RV, except that it will die() if the PV isn't
736 This method is less often useful. It assumes that the string
737 stored in the struct is null-terminated, and disregards the
740 It is the appropriate method to use if you need to get the name
741 of a lexical variable from a padname array. Lexical variable names
742 are always stored with a null terminator, and the length field
743 (SvCUR) is overloaded for other purposes and can't be relied on here.
747 =head2 B::PVMG Methods
757 =head2 B::MAGIC Methods
765 Only valid on r-magic, returns the string that generated the regexp.
775 Will die() if called on r-magic.
781 Only valid on r-magic, returns the integer value of the REGEX stored
786 =head2 B::PVLV Methods
820 This method returns TRUE if the GP field of the GV is NULL.
826 This method returns the name of the glob, but if the first
827 character of the name is a control character, then it converts
828 it to ^X first, so that *^G would return "^G" rather than "\cG".
830 It's useful if you want to print out the name of a variable.
831 If you restrict yourself to globs which exist at compile-time
832 then the result ought to be unambiguous, because code like
833 C<${"^G"} = 1> is compiled as two ops - a constant string and
834 a dereference (rv2gv) - so that the glob is created at runtime.
836 If you're working with globs at runtime, and need to disambiguate
837 *^G from *{"^G"}, then you should use the raw NAME method.
901 Takes one arguments ( 'stdin' | 'stdout' | 'stderr' ) and returns true
902 if the IoIFP of the object is equal to the handle whose name was
903 passed as argument ( i.e. $io->IsSTD('stderr') is true if
904 IoIFP($io) == PerlIO_stdin() ).
920 Like C<ARRAY>, but takes an index as an argument to get only one element,
921 rather than a list of all of them.
925 This method is deprecated if running under Perl 5.8, and is no longer present
926 if running under Perl 5.9
930 This method returns the AV specific flags. In Perl 5.9 these are now stored
931 in with the main SV flags, so this method is no longer present.
961 For constant subroutines, returns the constant SV returned by the subroutine.
987 This method is not present if running under Perl 5.9, as the PMROOT
988 information is no longer stored directly in the hash.
992 =head2 OP-RELATED CLASSES
994 C<B::OP>, C<B::UNOP>, C<B::BINOP>, C<B::LOGOP>, C<B::LISTOP>, C<B::PMOP>,
995 C<B::SVOP>, C<B::PADOP>, C<B::PVOP>, C<B::LOOP>, C<B::COP>.
997 These classes correspond in the obvious way to the underlying C
998 structures of similar names. The inheritance hierarchy mimics the
999 underlying C "inheritance":
1003 +---------------+--------+--------+-------+
1005 B::UNOP B::SVOP B::PADOP B::COP B::PVOP
1016 Access methods correspond to the underlying C structre field names,
1017 with the leading "class indication" prefix (C<"op_">) removed.
1019 =head2 B::OP Methods
1021 These methods get the values of similarly named fields within the OP
1022 data structure. See top of C<op.h> for more info.
1032 This returns the op name as a string (e.g. "add", "rv2av").
1036 This returns the function name as a string (e.g. "PL_ppaddr[OP_ADD]",
1037 "PL_ppaddr[OP_RV2AV]").
1041 This returns the op description from the global C PL_op_desc array
1042 (e.g. "addition" "array deref").
1058 =head2 B::UNOP METHOD
1066 =head2 B::BINOP METHOD
1074 =head2 B::LOGOP METHOD
1082 =head2 B::LISTOP METHOD
1090 =head2 B::PMOP Methods
1100 Only up to Perl 5.9.4
1114 Only when perl was compiled with ithreads.
1118 =head2 B::SVOP METHOD
1128 =head2 B::PADOP METHOD
1136 =head2 B::PVOP METHOD
1144 =head2 B::LOOP Methods
1156 =head2 B::COP Methods
1187 Malcolm Beattie, C<mbeattie@sable.ox.ac.uk>