| PPI documentation | Contained in the PPI distribution. |
PPI::Structure - The base class for Perl braced structures
PPI::Structure
isa PPI::Node
isa PPI::Element
PPI::Structure is the root class for all Perl bracing structures. This
covers all forms of [ ... ] , { ... } , and ( ... ) brace
types, and includes cases where only one half of the pair exist.
The class PPI::Structure itself is full abstract and no objects of that type should actually exist in the tree.
A PPI::Structure has an unusual existance. Unlike a PPI::Document or PPI::Statement, which both simply contain other elements, a structure both contains and consists of content.
That is, the brace tokens are not considered to be "children" of the structure, but are part of it.
In practice, this will mean that while the ->elements and ->tokens methods (and related) will return a list with the brace tokens at either end, the ->children method explicitly will not return the brace.
Excluding the transient PPI::Structure::Unknown that exists briefly inside the parser, there are eight types of structure.
This covers all round braces used for function arguments, in foreach
loops, literal lists, and braces used for precedence-ordering purposes.
Although not used for the foreach loop list, this is used for
the special case of the round-brace three-part semicolon-seperated for
loop expression (the traditional C style for loop).
This is for the expression being matched in switch statements.
This is for the matching expression in "when" statements.
This round-brace structure covers boolean conditional braces, such as
for if and while blocks.
This curly-brace and common structure is used for all form of code
blocks. This includes those for if, do and similar, as well
as grep, map, sort, sub and (labelled or anonymous)
scoping blocks.
This class covers brace structures used for the construction of
anonymous ARRAY and HASH references.
This class covers square-braces and curly-braces used after a
-> pointer to access the subscript of an ARRAY or HASH.
PPI::Structure itself has very few methods. Most of the time, you will be
working with the more generic PPI::Element or PPI::Node methods, or one
of the methods that are subclass-specific.
For lack of better terminology (like "open" and "close") that has not already in use for some other more important purpose, the two individual braces for the structure are known within PPI as the "start" and "finish" braces (at least for method purposes).
The start method returns the start brace for the structure (i.e. the
opening brace).
Returns the brace as a PPI::Token::Structure or undef if the
structure does not have a starting brace.
Under normal parsing circumstances this should never occur, but may happen due to manipulation of the PDOM tree.
The finish method returns the finish brace for the structure (i.e. the
closing brace).
Returns the brace as a PPI::Token::Structure or undef if the
structure does not have a finishing brace. This can be quite common if
the document is not complete (for example, from an editor where the user
may be halfway through typeing a subroutine).
The braces method is a utility method which returns the brace type,
regardless of whether has both braces defined, or just the starting
brace, or just the ending brace.
Returns on of the three strings '[]', '{}', or '()', or undef
on error (primarily not having a start brace, as mentioned above).
The complete method is a convenience method that returns true if
the both braces are defined for the structure, or false if only one
brace is defined.
Unlike the top level complete method which checks for completeness
in depth, the structure complete method ONLY confirms completeness
for the braces, and does not recurse downwards.
See the support section in the main module.
Adam Kennedy <adamk@cpan.org>
Copyright 2001 - 2011 Adam Kennedy.
This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
The full text of the license can be found in the LICENSE file included with this module.
| PPI documentation | Contained in the PPI distribution. |
package PPI::Structure;
use strict; use Scalar::Util (); use Params::Util qw{_INSTANCE}; use PPI::Node (); use PPI::Exception (); use vars qw{$VERSION @ISA *_PARENT}; BEGIN { $VERSION = '1.215'; @ISA = 'PPI::Node'; *_PARENT = *PPI::Element::_PARENT; } use PPI::Structure::Block (); use PPI::Structure::Condition (); use PPI::Structure::Constructor (); use PPI::Structure::For (); use PPI::Structure::Given (); use PPI::Structure::List (); use PPI::Structure::Subscript (); use PPI::Structure::Unknown (); use PPI::Structure::When (); ##################################################################### # Constructor sub new { my $class = shift; my $Token = PPI::Token::__LEXER__opens($_[0]) ? shift : return undef; # Create the object my $self = bless { children => [], start => $Token, }, $class; # Set the start braces parent link Scalar::Util::weaken( $_PARENT{Scalar::Util::refaddr $Token} = $self ); $self; } ##################################################################### # PPI::Structure API methods
sub start { $_[0]->{start} }
sub finish { $_[0]->{finish} }
sub braces { my $self = $_[0]->{start} ? shift : return undef; return { '[' => '[]', '(' => '()', '{' => '{}', }->{ $self->{start}->{content} }; }
sub complete { !! ($_[0]->{start} and $_[0]->{finish}); } ##################################################################### # PPI::Node overloaded methods # For us, the "elements" concept includes the brace tokens sub elements { my $self = shift; if ( wantarray ) { # Return a list in array context return ( $self->{start} || (), @{$self->{children}}, $self->{finish} || () ); } else { # Return the number of elements in scalar context. # This is memory-cheaper than creating another big array return scalar(@{$self->{children}}) + ($self->{start} ? 1 : 0) + ($self->{finish} ? 1 : 0); } } # For us, the first element is probably the opening brace sub first_element { # Technically, if we have no children and no opening brace, # then the first element is the closing brace. $_[0]->{start} or $_[0]->{children}->[0] or $_[0]->{finish}; } # For us, the last element is probably the closing brace sub last_element { # Technically, if we have no children and no closing brace, # then the last element is the opening brace $_[0]->{finish} or $_[0]->{children}->[-1] or $_[0]->{start}; } # Location is same as the start token, if any sub location { my $self = shift; my $first = $self->first_element or return undef; $first->location; } ##################################################################### # PPI::Element overloaded methods # Get the full set of tokens, including start and finish sub tokens { my $self = shift; my @tokens = ( $self->{start} || (), $self->SUPER::tokens(@_), $self->{finish} || (), ); @tokens; } # Like the token method ->content, get our merged contents. # This will recurse downwards through everything ### Reimplement this using List::Utils stuff sub content { my $self = shift; my $content = $self->{start} ? $self->{start}->content : ''; foreach my $child ( @{$self->{children}} ) { $content .= $child->content; } $content .= $self->{finish}->content if $self->{finish}; $content; } # Is the structure completed sub _complete { !! ( defined $_[0]->{finish} ); } # You can insert either another structure, or a token sub insert_before { my $self = shift; my $Element = _INSTANCE(shift, 'PPI::Element') or return undef; if ( $Element->isa('PPI::Structure') ) { return $self->__insert_before($Element); } elsif ( $Element->isa('PPI::Token') ) { return $self->__insert_before($Element); } ''; } # As above, you can insert either another structure, or a token sub insert_after { my $self = shift; my $Element = _INSTANCE(shift, 'PPI::Element') or return undef; if ( $Element->isa('PPI::Structure') ) { return $self->__insert_after($Element); } elsif ( $Element->isa('PPI::Token') ) { return $self->__insert_after($Element); } ''; } 1;