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.
12 @B::ISA = qw(Exporter);
14 # walkoptree_slow comes from B.pm (you are there),
15 # walkoptree comes from B.xs
20 @B::EXPORT_OK = qw(minus_c ppname save_BEGINs
21 class peekop cast_I32 cstring cchar hash threadsv_names
22 main_root main_start main_cv svref_2object opnumber
23 sub_generation amagic_generation perlstring
24 walkoptree_slow walkoptree walkoptree_exec walksymtable
25 parents comppadlist sv_undef compile_stats timing_info
26 begin_av init_av check_av end_av regex_padav dowarn
27 defstash curstash warnhook diehook inc_gv @optype
30 push @B::EXPORT_OK, qw(unitcheck_av) if $] > 5.009;
32 # All the above in this BEGIN, because our BOOT code needs $VERSION set,
33 # and will append to @EXPORT_OK. And we need to run the BOOT code before
34 # we see OPf_KIDS below.
39 @B::SV::ISA = 'B::OBJECT';
40 @B::NULL::ISA = 'B::SV';
41 @B::PV::ISA = 'B::SV';
42 @B::IV::ISA = 'B::SV';
43 @B::NV::ISA = 'B::SV';
44 # RV is eliminated with 5.11.0, but effectively is a specialisation of IV now.
45 @B::RV::ISA = $] >= 5.011 ? 'B::IV' : 'B::SV';
46 @B::PVIV::ISA = qw(B::PV B::IV);
47 @B::PVNV::ISA = qw(B::PVIV B::NV);
48 @B::PVMG::ISA = 'B::PVNV';
49 @B::REGEXP::ISA = 'B::PVMG' if $] >= 5.011;
50 # Change in the inheritance hierarchy post 5.9.0
51 @B::PVLV::ISA = $] > 5.009 ? 'B::GV' : 'B::PVMG';
52 # BM is eliminated post 5.9.5, but effectively is a specialisation of GV now.
53 @B::BM::ISA = $] > 5.009005 ? 'B::GV' : 'B::PVMG';
54 @B::AV::ISA = 'B::PVMG';
55 @B::GV::ISA = 'B::PVMG';
56 @B::HV::ISA = 'B::PVMG';
57 @B::CV::ISA = 'B::PVMG';
58 @B::IO::ISA = 'B::PVMG';
59 @B::FM::ISA = 'B::CV';
61 @B::OP::ISA = 'B::OBJECT';
62 @B::UNOP::ISA = 'B::OP';
63 @B::BINOP::ISA = 'B::UNOP';
64 @B::LOGOP::ISA = 'B::UNOP';
65 @B::LISTOP::ISA = 'B::BINOP';
66 @B::SVOP::ISA = 'B::OP';
67 @B::PADOP::ISA = 'B::OP';
68 @B::PVOP::ISA = 'B::OP';
69 @B::LOOP::ISA = 'B::LISTOP';
70 @B::PMOP::ISA = 'B::LISTOP';
71 @B::COP::ISA = 'B::OP';
73 @B::SPECIAL::ISA = 'B::OBJECT';
75 @B::optype = qw(OP UNOP BINOP LOGOP LISTOP PMOP SVOP PADOP PVOP LOOP COP);
76 # bytecode.pl contained the following comment:
77 # Nullsv *must* come first in the following so that the condition
78 # ($$sv == 0) can continue to be used to test (sv == Nullsv).
79 @B::specialsv_name = qw(Nullsv &PL_sv_undef &PL_sv_yes &PL_sv_no
80 (SV*)pWARN_ALL (SV*)pWARN_NONE (SV*)pWARN_STD);
83 # Stop "-w" from complaining about the lack of a real B::OBJECT class
88 my $name = (shift())->NAME;
90 # The regex below corresponds to the isCONTROLVAR macro
93 $name =~ s/^([\cA-\cZ\c\\c[\c]\c?\c_\c^])/"^".
94 chr( utf8::unicode_to_native( 64 ^ ord($1) ))/e;
96 # When we say unicode_to_native we really mean ascii_to_native,
97 # which matters iff this is a non-ASCII platform (EBCDIC).
102 sub B::IV::int_value {
104 return (($self->FLAGS() & SVf_IVisUV()) ? $self->UVX : $self->IV);
107 sub B::NULL::as_string() {""}
108 *B::IV::as_string = \&B::IV::int_value;
109 *B::PV::as_string = \&B::PV::PV;
116 my ($class, $value) = @_;
118 walkoptree_debug($value);
128 sub parents { \@parents }
133 return sprintf("%s (0x%x) %s", class($op), $$op, $op->name);
136 sub walkoptree_slow {
137 my($op, $method, $level) = @_;
138 $op_count++; # just for statistics
140 warn(sprintf("walkoptree: %d. %s\n", $level, peekop($op))) if $debug;
141 $op->$method($level) if $op->can($method);
142 if ($$op && ($op->flags & OPf_KIDS)) {
144 unshift(@parents, $op);
145 for ($kid = $op->first; $$kid; $kid = $kid->sibling) {
146 walkoptree_slow($kid, $method, $level + 1);
150 if (class($op) eq 'PMOP'
151 && ref($op->pmreplroot)
152 && ${$op->pmreplroot}
153 && $op->pmreplroot->isa( 'B::OP' ))
155 unshift(@parents, $op);
156 walkoptree_slow($op->pmreplroot, $method, $level + 1);
162 return "Total number of OPs processed: $op_count\n";
166 my ($sec, $min, $hr) = localtime;
167 my ($user, $sys) = times;
168 sprintf("%02d:%02d:%02d user=$user sys=$sys",
169 $hr, $min, $sec, $user, $sys);
179 my ($obj, $value) = @_;
180 # warn(sprintf("savesym: sym_%x => %s\n", $$obj, $value)); # debug
181 $symtable{sprintf("sym_%x", $$obj)} = $value;
186 return $symtable{sprintf("sym_%x", $$obj)};
189 sub walkoptree_exec {
190 my ($op, $method, $level) = @_;
193 my $prefix = " " x $level;
194 for (; $$op; $op = $op->next) {
197 print $prefix, "goto $sym\n";
200 savesym($op, sprintf("%s (0x%lx)", class($op), $$op));
201 $op->$method($level);
204 /^(d?or(assign)?|and(assign)?|mapwhile|grepwhile|entertry|range|cond_expr)$/)
206 print $prefix, uc($1), " => {\n";
207 walkoptree_exec($op->other, $method, $level + 1);
208 print $prefix, "}\n";
209 } elsif ($ppname eq "match" || $ppname eq "subst") {
210 my $pmreplstart = $op->pmreplstart;
212 print $prefix, "PMREPLSTART => {\n";
213 walkoptree_exec($pmreplstart, $method, $level + 1);
214 print $prefix, "}\n";
216 } elsif ($ppname eq "substcont") {
217 print $prefix, "SUBSTCONT => {\n";
218 walkoptree_exec($op->other->pmreplstart, $method, $level + 1);
219 print $prefix, "}\n";
221 } elsif ($ppname eq "enterloop") {
222 print $prefix, "REDO => {\n";
223 walkoptree_exec($op->redoop, $method, $level + 1);
224 print $prefix, "}\n", $prefix, "NEXT => {\n";
225 walkoptree_exec($op->nextop, $method, $level + 1);
226 print $prefix, "}\n", $prefix, "LAST => {\n";
227 walkoptree_exec($op->lastop, $method, $level + 1);
228 print $prefix, "}\n";
229 } elsif ($ppname eq "subst") {
230 my $replstart = $op->pmreplstart;
232 print $prefix, "SUBST => {\n";
233 walkoptree_exec($replstart, $method, $level + 1);
234 print $prefix, "}\n";
241 my ($symref, $method, $recurse, $prefix) = @_;
246 $prefix = '' unless defined $prefix;
247 while (($sym, $ref) = each %$symref) {
248 $fullname = "*main::".$prefix.$sym;
250 $sym = $prefix . $sym;
251 if (svref_2object(\*$sym)->NAME ne "main::" && $sym ne "<none>::" && &$recurse($sym)) {
252 walksymtable(\%$fullname, $method, $recurse, $sym);
255 svref_2object(\*$fullname)->$method();
266 my ($class, $section, $symtable, $default) = @_;
267 $output_fh ||= FileHandle->new_tmpfile;
268 my $obj = bless [-1, $section, $symtable, $default], $class;
269 $sections{$section} = $obj;
274 my ($class, $section) = @_;
275 return $sections{$section};
280 while (defined($_ = shift)) {
281 print $output_fh "$section->[1]\t$_\n";
288 return $section->[0];
293 return $section->[1];
298 return $section->[2];
303 return $section->[3];
307 my ($section, $fh, $format) = @_;
308 my $name = $section->name;
309 my $sym = $section->symtable || {};
310 my $default = $section->default;
312 seek($output_fh, 0, 0);
313 while (<$output_fh>) {
318 exists($sym->{$1}) ? $sym->{$1} : $default;
320 printf $fh $format, $_;
332 B - The Perl Compiler Backend
340 The C<B> module supplies classes which allow a Perl program to delve
341 into its own innards. It is the module used to implement the
342 "backends" of the Perl compiler. Usage of the compiler does not
343 require knowledge of this module: see the F<O> module for the
344 user-visible part. The C<B> module is of use to those who want to
345 write new compiler backends. This documentation assumes that the
346 reader knows a fair amount about perl's internals including such
347 things as SVs, OPs and the internal symbol table and syntax tree
352 The C<B> module contains a set of utility functions for querying the
353 current state of the Perl interpreter; typically these functions
354 return objects from the B::SV and B::OP classes, or their derived
355 classes. These classes in turn define methods for querying the
356 resulting objects about their own internal state.
358 =head1 Utility Functions
360 The C<B> module exports a variety of functions: some are simple
361 utility functions, others provide a Perl program with a way to
362 get an initial "handle" on an internal object.
364 =head2 Functions Returning C<B::SV>, C<B::AV>, C<B::HV>, and C<B::CV> objects
366 For descriptions of the class hierarchy of these objects and the
367 methods that can be called on them, see below, L<"OVERVIEW OF
368 CLASSES"> and L<"SV-RELATED CLASSES">.
374 Returns the SV object corresponding to the C variable C<sv_undef>.
378 Returns the SV object corresponding to the C variable C<sv_yes>.
382 Returns the SV object corresponding to the C variable C<sv_no>.
384 =item svref_2object(SVREF)
386 Takes a reference to any Perl value, and turns the referred-to value
387 into an object in the appropriate B::OP-derived or B::SV-derived
388 class. Apart from functions such as C<main_root>, this is the primary
389 way to get an initial "handle" on an internal perl data structure
390 which can then be followed with the other access methods.
392 The returned object will only be valid as long as the underlying OPs
393 and SVs continue to exist. Do not attempt to use the object after the
394 underlying structures are freed.
396 =item amagic_generation
398 Returns the SV object corresponding to the C variable C<amagic_generation>.
402 Returns the AV object (i.e. in class B::AV) representing INIT blocks.
406 Returns the AV object (i.e. in class B::AV) representing CHECK blocks.
410 Returns the AV object (i.e. in class B::AV) representing UNITCHECK blocks.
414 Returns the AV object (i.e. in class B::AV) representing BEGIN blocks.
418 Returns the AV object (i.e. in class B::AV) representing END blocks.
422 Returns the AV object (i.e. in class B::AV) of the global comppadlist.
426 Only when perl was compiled with ithreads.
430 Return the (faked) CV corresponding to the main part of the Perl
435 =head2 Functions for Examining the Symbol Table
439 =item walksymtable(SYMREF, METHOD, RECURSE, PREFIX)
441 Walk the symbol table starting at SYMREF and call METHOD on each
442 symbol (a B::GV object) visited. When the walk reaches package
443 symbols (such as "Foo::") it invokes RECURSE, passing in the symbol
444 name, and only recurses into the package if that sub returns true.
446 PREFIX is the name of the SYMREF you're walking.
450 # Walk CGI's symbol table calling print_subs on each symbol.
451 # Recurse only into CGI::Util::
452 walksymtable(\%CGI::, 'print_subs', sub { $_[0] eq 'CGI::Util::' },
455 print_subs() is a B::GV method you have declared. Also see L<"B::GV
460 =head2 Functions Returning C<B::OP> objects or for walking op trees
462 For descriptions of the class hierarchy of these objects and the
463 methods that can be called on them, see below, L<"OVERVIEW OF
464 CLASSES"> and L<"OP-RELATED CLASSES">.
470 Returns the root op (i.e. an object in the appropriate B::OP-derived
471 class) of the main part of the Perl program.
475 Returns the starting op of the main part of the Perl program.
477 =item walkoptree(OP, METHOD)
479 Does a tree-walk of the syntax tree based at OP and calls METHOD on
480 each op it visits. Each node is visited before its children. If
481 C<walkoptree_debug> (see below) has been called to turn debugging on then
482 the method C<walkoptree_debug> is called on each op before METHOD is
485 =item walkoptree_debug(DEBUG)
487 Returns the current debugging flag for C<walkoptree>. If the optional
488 DEBUG argument is non-zero, it sets the debugging flag to that. See
489 the description of C<walkoptree> above for what the debugging flag
494 =head2 Miscellaneous Utility Functions
500 Return the PP function name (e.g. "pp_add") of op number OPNUM.
504 Returns a string in the form "0x..." representing the value of the
505 internal hash function used by perl on string STR.
509 Casts I to the internal I32 type used by that perl.
513 Does the equivalent of the C<-c> command-line option. Obviously, this
514 is only useful in a BEGIN block or else the flag is set too late.
518 Returns a double-quote-surrounded escaped version of STR which can
519 be used as a string in C source code.
521 =item perlstring(STR)
523 Returns a double-quote-surrounded escaped version of STR which can
524 be used as a string in Perl source code.
528 Returns the class of an object without the part of the classname
529 preceding the first C<"::">. This is used to turn C<"B::UNOP"> into
530 C<"UNOP"> for example.
534 In a perl compiled for threads, this returns a list of the special
535 per-thread threadsv variables.
539 =head2 Exported utility variabiles
545 my $op_type = $optype[$op_type_num];
547 A simple mapping of the op type number to its type (like 'COP' or 'BINOP').
549 =item @specialsv_name
551 my $sv_name = $specialsv_name[$sv_index];
553 Certain SV types are considered 'special'. They're represented by
554 B::SPECIAL and are referred to by a number from the specialsv_list.
555 This array maps that number back to the name of the SV (like 'Nullsv'
561 =head1 OVERVIEW OF CLASSES
563 The C structures used by Perl's internals to hold SV and OP
564 information (PVIV, AV, HV, ..., OP, SVOP, UNOP, ...) are modelled on a
565 class hierarchy and the C<B> module gives access to them via a true
566 object hierarchy. Structure fields which point to other objects
567 (whether types of SV or types of OP) are represented by the C<B>
568 module as Perl objects of the appropriate class.
570 The bulk of the C<B> module is the methods for accessing fields of
573 Note that all access is read-only. You cannot modify the internals by
574 using this module. Also, note that the B::OP and B::SV objects created
575 by this module are only valid for as long as the underlying objects
576 exist; their creation doesn't increase the reference counts of the
577 underlying objects. Trying to access the fields of a freed object will
578 give incomprehensible results, or worse.
580 =head2 SV-RELATED CLASSES
582 B::IV, B::NV, B::RV, B::PV, B::PVIV, B::PVNV, B::PVMG, B::BM (5.9.5 and
583 earlier), B::PVLV, B::AV, B::HV, B::CV, B::GV, B::FM, B::IO. These classes
584 correspond in the obvious way to the underlying C structures of similar names.
585 The inheritance hierarchy mimics the underlying C "inheritance". For the
586 5.10.x branch, (I<ie> 5.10.0, 5.10.1 I<etc>) this is:
590 +------------+------------+------------+
592 B::PV B::IV B::NV B::RV
604 +-----+-----+-----+-----+
606 B::AV B::GV B::HV B::CV B::IO
611 For 5.9.0 and earlier, PVLV is a direct subclass of PVMG, and BM is still
612 present as a distinct type, so the base of this diagram is
619 +------+-----+-----+-----+-----+-----+
621 B::PVLV B::BM B::AV B::GV B::HV B::CV B::IO
626 For 5.11.0 and later, B::RV is abolished, and IVs can be used to store
627 references, and a new type B::REGEXP is introduced, giving this structure:
631 +------------+------------+
645 +-------+-------+---+---+-------+-------+
647 B::AV B::GV B::HV B::CV B::IO B::REGEXP
653 Access methods correspond to the underlying C macros for field access,
654 usually with the leading "class indication" prefix removed (Sv, Av,
655 Hv, ...). The leading prefix is only left in cases where its removal
656 would cause a clash in method name. For example, C<GvREFCNT> stays
657 as-is since its abbreviation would clash with the "superclass" method
658 C<REFCNT> (corresponding to the C function C<SvREFCNT>).
670 Returns a reference to the regular scalar corresponding to this
671 B::SV object. In other words, this method is the inverse operation
672 to the svref_2object() subroutine. This scalar and other data it points
673 at should be considered read-only: modifying them is neither safe nor
674 guaranteed to have a sensible effect.
684 Returns the value of the IV, I<interpreted as
685 a signed integer>. This will be misleading
686 if C<FLAGS & SVf_IVisUV>. Perhaps you want the
687 C<int_value> method instead?
695 This method returns the value of the IV as an integer.
696 It differs from C<IV> in that it returns the correct
697 value regardless of whether it's stored signed or
730 This method is the one you usually want. It constructs a
731 string using the length and offset information in the struct:
732 for ordinary scalars it will return the string that you'd see
733 from Perl, even if it contains null characters.
737 Same as B::RV::RV, except that it will die() if the PV isn't
742 This method is less often useful. It assumes that the string
743 stored in the struct is null-terminated, and disregards the
746 It is the appropriate method to use if you need to get the name
747 of a lexical variable from a padname array. Lexical variable names
748 are always stored with a null terminator, and the length field
749 (SvCUR) is overloaded for other purposes and can't be relied on here.
753 =head2 B::PVMG Methods
763 =head2 B::MAGIC Methods
771 Only valid on r-magic, returns the string that generated the regexp.
781 Will die() if called on r-magic.
787 Only valid on r-magic, returns the integer value of the REGEX stored
792 =head2 B::PVLV Methods
826 This method returns TRUE if the GP field of the GV is NULL.
832 This method returns the name of the glob, but if the first
833 character of the name is a control character, then it converts
834 it to ^X first, so that *^G would return "^G" rather than "\cG".
836 It's useful if you want to print out the name of a variable.
837 If you restrict yourself to globs which exist at compile-time
838 then the result ought to be unambiguous, because code like
839 C<${"^G"} = 1> is compiled as two ops - a constant string and
840 a dereference (rv2gv) - so that the glob is created at runtime.
842 If you're working with globs at runtime, and need to disambiguate
843 *^G from *{"^G"}, then you should use the raw NAME method.
907 Takes one arguments ( 'stdin' | 'stdout' | 'stderr' ) and returns true
908 if the IoIFP of the object is equal to the handle whose name was
909 passed as argument ( i.e. $io->IsSTD('stderr') is true if
910 IoIFP($io) == PerlIO_stdin() ).
926 Like C<ARRAY>, but takes an index as an argument to get only one element,
927 rather than a list of all of them.
931 This method is deprecated if running under Perl 5.8, and is no longer present
932 if running under Perl 5.9
936 This method returns the AV specific flags. In Perl 5.9 these are now stored
937 in with the main SV flags, so this method is no longer present.
967 For constant subroutines, returns the constant SV returned by the subroutine.
993 This method is not present if running under Perl 5.9, as the PMROOT
994 information is no longer stored directly in the hash.
998 =head2 OP-RELATED CLASSES
1000 C<B::OP>, C<B::UNOP>, C<B::BINOP>, C<B::LOGOP>, C<B::LISTOP>, C<B::PMOP>,
1001 C<B::SVOP>, C<B::PADOP>, C<B::PVOP>, C<B::LOOP>, C<B::COP>.
1003 These classes correspond in the obvious way to the underlying C
1004 structures of similar names. The inheritance hierarchy mimics the
1005 underlying C "inheritance":
1009 +---------------+--------+--------+-------+
1011 B::UNOP B::SVOP B::PADOP B::COP B::PVOP
1022 Access methods correspond to the underlying C structre field names,
1023 with the leading "class indication" prefix (C<"op_">) removed.
1025 =head2 B::OP Methods
1027 These methods get the values of similarly named fields within the OP
1028 data structure. See top of C<op.h> for more info.
1038 This returns the op name as a string (e.g. "add", "rv2av").
1042 This returns the function name as a string (e.g. "PL_ppaddr[OP_ADD]",
1043 "PL_ppaddr[OP_RV2AV]").
1047 This returns the op description from the global C PL_op_desc array
1048 (e.g. "addition" "array deref").
1064 =head2 B::UNOP METHOD
1072 =head2 B::BINOP METHOD
1080 =head2 B::LOGOP METHOD
1088 =head2 B::LISTOP METHOD
1096 =head2 B::PMOP Methods
1106 Only up to Perl 5.9.4
1118 Only when perl was compiled with ithreads.
1122 =head2 B::SVOP METHOD
1132 =head2 B::PADOP METHOD
1140 =head2 B::PVOP METHOD
1148 =head2 B::LOOP Methods
1160 =head2 B::COP Methods
1191 Malcolm Beattie, C<mbeattie@sable.ox.ac.uk>