8 Scope::Upper - Act on upper scopes.
23 L</reap>, L</localize>, L</localize_elem>, L</localize_delete> and L</WORDS> :
28 reap localize localize_elem localize_delete
33 my ($class, $name) = @_;
35 localize '$tag' => bless({ name => $name }, $class) => UP;
37 reap { print Scope->tag->name, ": end\n" } UP;
40 # Get the tag stored in the caller namespace
43 my $pkg = __PACKAGE__;
44 $pkg = caller $l++ while $pkg eq __PACKAGE__;
50 sub name { shift->{name} }
52 # Locally capture warnings and reprint them with the name prefixed
54 localize_elem '%SIG', '__WARN__' => sub {
55 print Scope->tag->name, ': ', @_;
61 for (reverse 0 .. $#INC) {
62 # First UP is the for loop, second is the sub boundary
63 localize_delete '@INC', $_ => UP UP;
72 Scope->new("top"); # initializes $UserLand::tag
76 my $one = 1 + undef; # prints "top: Use of uninitialized value..."
81 print $@; # prints "Can't locate Cwd.pm in @INC
82 } # (@INC contains:) at..."
84 require Cwd; # loads Cwd.pm
87 } # prints "top: done"
89 L</unwind> and L</want_at> :
93 use Scope::Upper qw<unwind want_at :words>;
96 my @result = shift->();
97 my $cx = SUB UP; # Point to the sub above this one
98 unwind +(want_at($cx) ? @result : scalar @result) => $cx;
105 my @things = qw<a b c>;
106 return @things; # returns to try() and then outside zap()
112 my @stuff = zap(); # @stuff contains qw<a b c>
113 my $stuff = zap(); # $stuff contains 3
119 use Scope::Upper qw<uplevel CALLER>;
127 my $sub = (caller 0)[3];
128 print "$_[0] from $sub()";
132 target('hello'); # "hello from Uplevel::target()"
134 L</uid> and L</validate_uid> :
136 use Scope::Upper qw<uid validate_uid>;
143 if ($uid eq uid(UP)) { # yes
146 if (validate_uid($uid)) { # yes
152 if (validate_uid($uid)) { # no
158 This module lets you defer actions I<at run-time> that will take place when the control flow returns into an upper scope.
165 hook an upper scope end with L</reap> ;
169 localize variables, array/hash values or deletions of elements in higher contexts with respectively L</localize>, L</localize_elem> and L</localize_delete> ;
173 return values immediately to an upper level with L</unwind> and L</yield>, and know which context was in use then with L</want_at> ;
177 execute a subroutine in the setting of an upper subroutine stack frame with L</uplevel> ;
181 uniquely identify contextes with L</uid> and L</validate_uid>.
187 In all those functions, C<$context> refers to the target scope.
189 You have to use one or a combination of L</WORDS> to build the C<$context> passed to these functions.
190 This is needed in order to ensure that the module still works when your program is ran in the debugger.
191 The only thing you can assume is that it is an I<absolute> indicator of the frame, which means that you can safely store it at some point and use it when needed, and it will still denote the original scope.
197 XSLoader::load(__PACKAGE__, $VERSION);
203 reap { ... } $context;
204 &reap($callback, $context);
206 Adds a destructor that calls C<$callback> (in void context) when the upper scope represented by C<$context> ends.
210 localize $what, $value;
211 localize $what, $value, $context;
213 Introduces a C<local> delayed to the time of first return into the upper scope denoted by C<$context>.
220 A glob, in which case C<$value> can either be a glob or a reference.
221 L</localize> follows then the same syntax as C<local *x = $value>.
222 For example, if C<$value> is a scalar reference, then the C<SCALAR> slot of the glob will be set to C<$$value> - just like C<local *x = \1> sets C<$x> to C<1>.
226 A string beginning with a sigil, representing the symbol to localize and to assign to.
227 If the sigil is C<'$'>, L</localize> follows the same syntax as C<local $x = $value>, i.e. C<$value> isn't dereferenced.
230 localize '$x', \'foo' => HERE;
232 will set C<$x> to a reference to the string C<'foo'>.
233 Other sigils (C<'@'>, C<'%'>, C<'&'> and C<'*'>) require C<$value> to be a reference of the corresponding type.
235 When the symbol is given by a string, it is resolved when the actual localization takes place and not when L</localize> is called.
236 Thus, if the symbol name is not qualified, it will refer to the variable in the package where the localization actually takes place and not in the one where the L</localize> call was compiled.
241 sub new { localize '$tag', $_[0] => UP }
252 will localize C<$Tool::tag> and not C<$Scope::tag>.
253 If you want the other behaviour, you just have to specify C<$what> as a glob or a qualified name.
255 Note that if C<$what> is a string denoting a variable that wasn't declared beforehand, the relevant slot will be vivified as needed and won't be deleted from the glob when the localization ends.
256 This situation never arises with C<local> because it only compiles when the localized variable is already declared.
257 Although I believe it shouldn't be a problem as glob slots definedness is pretty much an implementation detail, this behaviour may change in the future if proved harmful.
261 =head2 C<localize_elem>
263 localize_elem $what, $key, $value;
264 localize_elem $what, $key, $value, $context;
266 Introduces a C<local $what[$key] = $value> or C<local $what{$key} = $value> delayed to the time of first return into the upper scope denoted by C<$context>.
267 Unlike L</localize>, C<$what> must be a string and the type of localization is inferred from its sigil.
268 The two only valid types are array and hash ; for anything besides those, L</localize_elem> will throw an exception.
269 C<$key> is either an array index or a hash key, depending of which kind of variable you localize.
271 If C<$what> is a string pointing to an undeclared variable, the variable will be vivified as soon as the localization occurs and emptied when it ends, although it will still exist in its glob.
273 =head2 C<localize_delete>
275 localize_delete $what, $key;
276 localize_delete $what, $key, $context;
278 Introduces the deletion of a variable or an array/hash element delayed to the time of first return into the upper scope denoted by C<$context>.
285 A glob, in which case C<$key> is ignored and the call is equivalent to C<local *x>.
289 A string beginning with C<'@'> or C<'%'>, for which the call is equivalent to respectiveley C<local $a[$key]; delete $a[$key]> and C<local $h{$key}; delete $h{$key}>.
293 A string beginning with C<'&'>, which more or less does C<undef &func> in the upper scope.
294 It's actually more powerful, as C<&func> won't even C<exists> anymore.
302 unwind @values, $context;
304 Returns C<@values> I<from> the subroutine, eval or format context pointed by or just above C<$context>, and immediately restart the program flow at this point - thus effectively returning C<@values> to an upper scope.
305 If C<@values> is empty, then the C<$context> parameter is optional and defaults to the current context (making the call equivalent to a bare C<return;>) ; otherwise it is mandatory.
307 The upper context isn't coerced onto C<@values>, which is hence always evaluated in list context.
311 my @a = ('a' .. 'z');
316 will set C<$num> to C<'z'>.
317 You can use L</want_at> to handle these cases.
322 yield @values, $context;
324 Returns C<@values> I<from> the context pointed by or just above C<$context>, and immediately restart the program flow at this point.
325 If C<@values> is empty, then the C<$context> parameter is optional and defaults to the current context ; otherwise it is mandatory.
327 L</yield> differs from L</unwind> in that it can target I<any> upper scope (besides a C<s///e> substitution context) and not necessarily a sub, an eval or a format.
328 Hence you can use it to return values from a C<do> or a C<map> block :
332 eval { require Time::HiRes } or yield time() => HERE;
337 yield if $seen{$_}++; # returns the empty list from the block
341 Like for L</unwind>, the upper context isn't coerced onto C<@values>.
346 my $want = want_at $context;
348 Like C<wantarray>, but for the subroutine/eval/format at or just above C<$context>.
350 The previous example can then be "corrected" :
353 my @a = ('a' .. 'z');
354 unwind +(want_at(HERE) ? @a : scalar @a) => HERE;
358 will rightfully set C<$num> to C<26>.
362 my @ret = uplevel { ...; return @ret };
363 my @ret = uplevel { my @args = @_; ...; return @ret } @args, $context;
364 my @ret = &uplevel($callback, @args, $context);
366 Executes the code reference C<$callback> with arguments C<@args> as if it were located at the subroutine stack frame pointed by C<$context>, effectively fooling C<caller> and C<die> into believing that the call actually happened higher in the stack.
367 The code is executed in the context of the C<uplevel> call, and what it returns is returned as-is by C<uplevel>.
379 my @inverses = target(1, 2, 4); # @inverses contains (0, 0.5, 0.25)
380 my $count = target(1, 2, 4); # $count is 3
382 Note that if C<@args> is empty, then the C<$context> parameter is optional and defaults to the current context ; otherwise it is mandatory.
384 L<Sub::Uplevel> also implements a pure-Perl version of C<uplevel>.
385 Both are identical, with the following caveats :
391 The L<Sub::Uplevel> implementation of C<uplevel> may execute a code reference in the context of B<any> upper stack frame.
392 The L<Scope::Upper> version can only uplevel to a B<subroutine> stack frame, and will croak if you try to target an C<eval> or a format.
396 Exceptions thrown from the code called by this version of C<uplevel> will not be caught by C<eval> blocks between the target frame and the uplevel call, while they will for L<Sub::Uplevel>'s version.
404 uplevel { die 'wut' } CALLER(2); # for Scope::Upper
405 # uplevel(3, sub { die 'wut' }) # for Sub::Uplevel
408 print "inner block: $@";
412 print "outer block: $@";
414 will print "inner block: wut..." with L<Sub::Uplevel> and "outer block: wut..." with L<Scope::Upper>.
418 L<Sub::Uplevel> globally overrides the Perl keyword C<caller>, while L<Scope::Upper> does not.
422 A simple wrapper lets you mimic the interface of L<Sub::Uplevel/uplevel> :
429 my $cxt = Scope::Upper::CALLER($frame);
430 &Scope::Upper::uplevel($code => @_ => $cxt);
433 Albeit the three exceptions listed above, it passes all the tests of L<Sub::Uplevel>.
438 my $uid = uid $context;
440 Returns an unique identifier (UID) for the context (or dynamic scope) pointed by C<$context>, or for the current context if C<$context> is omitted.
441 This UID will only be valid for the life time of the context it represents, and another UID will be generated next time the same scope is executed.
447 if ($uid eq uid()) { # yes, this is the same context
451 if ($uid eq uid()) { # no, we are one scope below
454 if ($uid eq uid(UP)) { # yes, UP points to the same scope as $uid
460 # $uid is now invalid
463 if ($uid eq uid()) { # no, this is another block
468 For example, each loop iteration gets its own UID :
477 # %uids has 5 entries
479 The UIDs are not guaranteed to be numbers, so you must use the C<eq> operator to compare them.
481 To check whether a given UID is valid, you can use the L</validate_uid> function.
483 =head2 C<validate_uid>
485 my $is_valid = validate_uid $uid;
487 Returns true if and only if C<$uid> is the UID of a currently valid context (that is, it designates a scope that is higher than the current one in the call stack).
493 if (validate_uid($uid)) { # yes
497 if (validate_uid($uid)) { # yes
503 if (validate_uid($uid)) { # no
509 =head2 C<SU_THREADSAFE>
511 True iff the module could have been built when thread-safety features.
519 my $top_context = TOP;
521 Returns the context that currently represents the highest scope.
525 my $current_context = HERE;
527 The context of the current scope.
529 =head2 Getting a context from a context
531 For any of those functions, C<$from> is expected to be a context.
532 When omitted, it defaults to the the current context.
536 my $upper_context = UP;
537 my $upper_context = UP $from;
539 The context of the scope just above C<$from>.
543 my $sub_context = SUB;
544 my $sub_context = SUB $from;
546 The context of the closest subroutine above C<$from>.
547 Note that C<$from> is returned if it is already a subroutine context ; hence C<SUB SUB == SUB>.
551 my $eval_context = EVAL;
552 my $eval_context = EVAL $from;
554 The context of the closest eval above C<$from>.
555 Note that C<$from> is returned if it is already an eval context ; hence C<EVAL EVAL == EVAL>.
557 =head2 Getting a context from a level
559 Here, C<$level> should denote a number of scopes above the current one.
560 When omitted, it defaults to C<0> and those functions return the same context as L</HERE>.
565 my $context = SCOPE $level;
567 The C<$level>-th upper context, regardless of its type.
571 my $context = CALLER;
572 my $context = CALLER $level;
574 The context of the C<$level>-th upper subroutine/eval/format.
575 It kind of corresponds to the context represented by C<caller $level>, but while e.g. C<caller 0> refers to the caller context, C<CALLER 0> will refer to the top scope in the current context.
579 Where L</reap> fires depending on the C<$cxt> :
585 reap \&cleanup => $cxt;
587 } # $cxt = SCOPE(0) = HERE
589 }->(); # $cxt = SCOPE(1) = UP = SUB = CALLER(0)
591 }; # $cxt = SCOPE(2) = UP UP = UP SUB = EVAL = CALLER(1)
593 }->(); # $cxt = SCOPE(3) = SUB UP SUB = SUB EVAL = CALLER(2)
596 Where L</localize>, L</localize_elem> and L</localize_delete> act depending on the C<$cxt> :
602 localize '$x' => 1 => $cxt;
603 # $cxt = SCOPE(0) = HERE
606 # $cxt = SCOPE(1) = UP = SUB = CALLER(0)
609 # $cxt = SCOPE(2) = UP UP = UP SUB = EVAL = CALLER(1)
612 # $cxt = SCOPE(3) = SUB UP SUB = SUB EVAL = CALLER(2)
615 # $cxt = SCOPE(4), UP SUB UP SUB = UP SUB EVAL = UP CALLER(2) = TOP
618 Where L</unwind>, L</yield>, L</want_at> and L</uplevel> point to depending on the C<$cxt>:
624 unwind @things => $cxt; # or yield @things => $cxt
625 # or uplevel { ... } $cxt
629 }->(); # $cxt = SCOPE(0) = SCOPE(1) = HERE = UP = SUB = CALLER(0)
631 }; # $cxt = SCOPE(2) = UP UP = UP SUB = EVAL = CALLER(1) (*)
633 }->(); # $cxt = SCOPE(3) = SUB UP SUB = SUB EVAL = CALLER(2)
636 # (*) Note that uplevel() will croak if you pass that scope frame,
637 # because it cannot target eval scopes.
641 The functions L</reap>, L</localize>, L</localize_elem>, L</localize_delete>, L</unwind>, L</yield>, L</want_at> and L</uplevel> are only exported on request, either individually or by the tags C<':funcs'> and C<':all'>.
643 The constant L</SU_THREADSAFE> is also only exported on request, individually or by the tags C<':consts'> and C<':all'>.
645 Same goes for the words L</TOP>, L</HERE>, L</UP>, L</SUB>, L</EVAL>, L</SCOPE> and L</CALLER> that are only exported on request, individually or by the tags C<':words'> and C<':all'>.
649 use base qw<Exporter>;
655 localize localize_elem localize_delete
661 words => [ qw<TOP HERE UP SUB EVAL SCOPE CALLER> ],
662 consts => [ qw<SU_THREADSAFE> ],
664 our @EXPORT_OK = map { @$_ } values %EXPORT_TAGS;
665 $EXPORT_TAGS{'all'} = [ @EXPORT_OK ];
669 Be careful that local variables are restored in the reverse order in which they were localized.
670 Consider those examples:
674 reap sub { print $x } => HERE;
682 reap sub { $x = 2 } => HERE;
687 The first case is "solved" by moving the C<local> before the C<reap>, and the second by using L</localize> instead of L</reap>.
689 The effects of L</reap>, L</localize> and L</localize_elem> can't cross C<BEGIN> blocks, hence calling those functions in C<import> is deemed to be useless.
690 This is an hopeless case because C<BEGIN> blocks are executed once while localizing constructs should do their job at each run.
691 However, it's possible to hook the end of the current scope compilation with L<B::Hooks::EndOfScope>.
693 Some rare oddities may still happen when running inside the debugger.
694 It may help to use a perl higher than 5.8.9 or 5.10.0, as they contain some context-related fixes.
696 Calling C<goto> to replace an L</uplevel>'d code frame does not work :
702 for a C<perl> older than the 5.8 series ;
706 for a C<DEBUGGING> C<perl> run with debugging flags set (as in C<perl -D ...>) ;
710 when the runloop callback is replaced by another module.
714 In those three cases, L</uplevel> will look for a C<goto &sub> statement in its callback and, if there is one, throw an exception before executing the code.
716 Moreover, in order to handle C<goto> statements properly, L</uplevel> currently has to suffer a run-time overhead proportional to the size of the the callback in every case (with a small ratio), and proportional to the size of B<all> the code executed as the result of the L</uplevel> call (including subroutine calls inside the callback) when a C<goto> statement is found in the L</uplevel> callback.
717 Despite this shortcoming, this XS version of L</uplevel> should still run way faster than the pure-Perl version from L<Sub::Uplevel>.
724 This module may happen to build with a C++ compiler as well, but don't rely on it, as no guarantee is made in this regard.
726 L<XSLoader> (core since perl 5.006).
730 L<perlfunc/local>, L<perlsub/"Temporary Values via local()">.
732 L<Alias>, L<Hook::Scope>, L<Scope::Guard>, L<Guard>.
736 L<Continuation::Escape> is a thin wrapper around L<Scope::Upper> that gives you a continuation passing style interface to L</unwind>.
737 It's easier to use, but it requires you to have control over the scope where you want to return.
743 Vincent Pit, C<< <perl at profvince.com> >>, L<http://www.profvince.com>.
745 You can contact me by mail or on C<irc.perl.org> (vincent).
749 Please report any bugs or feature requests to C<bug-scope-upper at rt.cpan.org>, or through the web interface at L<http://rt.cpan.org/NoAuth/ReportBug.html?Queue=Scope-Upper>.
750 I will be notified, and then you'll automatically be notified of progress on your bug as I make changes.
754 You can find documentation for this module with the perldoc command.
758 Tests code coverage report is available at L<http://www.profvince.com/perl/cover/Scope-Upper>.
760 =head1 ACKNOWLEDGEMENTS
762 Inspired by Ricardo Signes.
764 Thanks to Shawn M. Moore for motivation.
766 =head1 COPYRIGHT & LICENSE
768 Copyright 2008,2009,2010,2011,2012 Vincent Pit, all rights reserved.
770 This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
774 1; # End of Scope::Upper