3 perlref - Perl references and nested data structures
7 Before release 5 of Perl it was difficult to represent complex data
8 structures, because all references had to be symbolic, and even that was
9 difficult to do when you wanted to refer to a variable rather than a
10 symbol table entry. Perl not only makes it easier to use symbolic
11 references to variables, but lets you have "hard" references to any piece
12 of data. Any scalar may hold a hard reference. Because arrays and hashes
13 contain scalars, you can now easily build arrays of arrays, arrays of
14 hashes, hashes of arrays, arrays of hashes of functions, and so on.
16 Hard references are smart--they keep track of reference counts for you,
17 automatically freeing the thing referred to when its reference count
18 goes to zero. (Note: The reference counts for values in self-referential
19 or cyclic data structures may not go to zero without a little help; see
20 L<perlobj/"Two-Phased Garbage Collection"> for a detailed explanation.
21 If that thing happens to be an object, the object is
22 destructed. See L<perlobj> for more about objects. (In a sense,
23 everything in Perl is an object, but we usually reserve the word for
24 references to objects that have been officially "blessed" into a class package.)
27 A symbolic reference contains the name of a variable, just as a
28 symbolic link in the filesystem contains merely the name of a file.
29 The C<*glob> notation is a kind of symbolic reference. Hard references
30 are more like hard links in the file system: merely another way
31 at getting at the same underlying object, irrespective of its name.
33 "Hard" references are easy to use in Perl. There is just one
34 overriding principle: Perl does no implicit referencing or
35 dereferencing. When a scalar is holding a reference, it always behaves
36 as a scalar. It doesn't magically start being an array or a hash
37 unless you tell it so explicitly by dereferencing it.
39 References can be constructed several ways.
45 By using the backslash operator on a variable, subroutine, or value.
46 (This works much like the & (address-of) operator works in C.) Note
47 that this typically creates I<ANOTHER> reference to a variable, because
48 there's already a reference to the variable in the symbol table. But
49 the symbol table reference might go away, and you'll still have the
50 reference that the backslash returned. Here are some examples:
58 It isn't possible to create a true reference to an IO handle (filehandle or
59 dirhandle) using the backslash operator. See the explanation of the
60 *foo{THING} syntax below. (However, you're apt to find Perl code
61 out there using globrefs as though they were IO handles, which is
62 grandfathered into continued functioning.)
66 A reference to an anonymous array can be constructed using square
69 $arrayref = [1, 2, ['a', 'b', 'c']];
71 Here we've constructed a reference to an anonymous array of three elements
72 whose final element is itself reference to another anonymous array of three
73 elements. (The multidimensional syntax described later can be used to
74 access this. For example, after the above, C<$arrayref-E<gt>[2][1]> would have
77 Note that taking a reference to an enumerated list is not the same
78 as using square brackets--instead it's the same as creating
81 @list = (\$a, \@b, \%c);
82 @list = \($a, @b, %c); # same thing!
84 As a special case, C<\(@foo)> returns a list of references to the contents
85 of C<@foo>, not a reference to C<@foo> itself. Likewise for C<%foo>.
89 A reference to an anonymous hash can be constructed using curly
97 Anonymous hash and array constructors can be intermixed freely to
98 produce as complicated a structure as you want. The multidimensional
99 syntax described below works for these too. The values above are
100 literals, but variables and expressions would work just as well, because
101 assignment operators in Perl (even within local() or my()) are executable
102 statements, not compile-time declarations.
104 Because curly brackets (braces) are used for several other things
105 including BLOCKs, you may occasionally have to disambiguate braces at the
106 beginning of a statement by putting a C<+> or a C<return> in front so
107 that Perl realizes the opening brace isn't starting a BLOCK. The economy and
108 mnemonic value of using curlies is deemed worth this occasional extra
111 For example, if you wanted a function to make a new hash and return a
112 reference to it, you have these options:
114 sub hashem { { @_ } } # silently wrong
115 sub hashem { +{ @_ } } # ok
116 sub hashem { return { @_ } } # ok
120 A reference to an anonymous subroutine can be constructed by using
121 C<sub> without a subname:
123 $coderef = sub { print "Boink!\n" };
125 Note the presence of the semicolon. Except for the fact that the code
126 inside isn't executed immediately, a C<sub {}> is not so much a
127 declaration as it is an operator, like C<do{}> or C<eval{}>. (However, no
128 matter how many times you execute that line (unless you're in an
129 C<eval("...")>), C<$coderef> will still have a reference to the I<SAME>
130 anonymous subroutine.)
132 Anonymous subroutines act as closures with respect to my() variables,
133 that is, variables visible lexically within the current scope. Closure
134 is a notion out of the Lisp world that says if you define an anonymous
135 function in a particular lexical context, it pretends to run in that
136 context even when it's called outside of the context.
138 In human terms, it's a funny way of passing arguments to a subroutine when
139 you define it as well as when you call it. It's useful for setting up
140 little bits of code to run later, such as callbacks. You can even
141 do object-oriented stuff with it, though Perl provides a different
142 mechanism to do that already--see L<perlobj>.
144 You can also think of closure as a way to write a subroutine template without
145 using eval. (In fact, in version 5.000, eval was the I<only> way to get
146 closures. You may wish to use "require 5.001" if you use closures.)
148 Here's a small example of how closures works:
152 return sub { my $y = shift; print "$x, $y!\n"; };
154 $h = newprint("Howdy");
155 $g = newprint("Greetings");
165 Greetings, earthlings!
167 Note particularly that $x continues to refer to the value passed into
168 newprint() I<despite> the fact that the "my $x" has seemingly gone out of
169 scope by the time the anonymous subroutine runs. That's what closure
172 This applies to only lexical variables, by the way. Dynamic variables
173 continue to work as they have always worked. Closure is not something
174 that most Perl programmers need trouble themselves about to begin with.
178 References are often returned by special subroutines called constructors.
179 Perl objects are just references to a special kind of object that happens to know
180 which package it's associated with. Constructors are just special
181 subroutines that know how to create that association. They do so by
182 starting with an ordinary reference, and it remains an ordinary reference
183 even while it's also being an object. Constructors are customarily
184 named new(), but don't have to be:
186 $objref = new Doggie (Tail => 'short', Ears => 'long');
190 References of the appropriate type can spring into existence if you
191 dereference them in a context that assumes they exist. Because we haven't
192 talked about dereferencing yet, we can't show you any examples yet.
196 A reference can be created by using a special syntax, lovingly known as
197 the *foo{THING} syntax. *foo{THING} returns a reference to the THING
198 slot in *foo (which is the symbol table entry which holds everything
201 $scalarref = *foo{SCALAR};
202 $arrayref = *ARGV{ARRAY};
203 $hashref = *ENV{HASH};
204 $coderef = *handler{CODE};
206 $globref = *foo{GLOB};
208 All of these are self-explanatory except for *foo{IO}. It returns the
209 IO handle, used for file handles (L<perlfunc/open>), sockets
210 (L<perlfunc/socket> and L<perlfunc/socketpair>), and directory handles
211 (L<perlfunc/opendir>). For compatibility with previous versions of
212 Perl, *foo{FILEHANDLE} is a synonym for *foo{IO}.
214 *foo{THING} returns undef if that particular THING hasn't been used yet,
215 except in the case of scalars. *foo{SCALAR} returns a reference to an
216 anonymous scalar if $foo hasn't been used yet. This might change in a
219 The use of *foo{IO} is the best way to pass bareword filehandles into or
220 out of subroutines, or to store them in larger data structures.
222 splutter(*STDOUT{IO});
225 print $fh "her um well a hmmm\n";
228 $rec = get_rec(*STDIN{IO});
234 Beware, though, that you can't do this with a routine which is going to
235 open the filehandle for you, because *HANDLE{IO} will be undef if HANDLE
236 hasn't been used yet. Use \*HANDLE for that sort of thing instead.
238 Using \*HANDLE (or *HANDLE) is another way to use and store non-bareword
239 filehandles (before perl version 5.002 it was the only way). The two
240 methods are largely interchangeable, you can do
243 $rec = get_rec(\*STDIN);
245 with the above subroutine definitions.
249 That's it for creating references. By now you're probably dying to
250 know how to use references to get back to your long-lost data. There
251 are several basic methods.
257 Anywhere you'd put an identifier (or chain of identifiers) as part
258 of a variable or subroutine name, you can replace the identifier with
259 a simple scalar variable containing a reference of the correct type:
262 push(@$arrayref, $filename);
263 $$arrayref[0] = "January";
264 $$hashref{"KEY"} = "VALUE";
266 print $globref "output\n";
268 It's important to understand that we are specifically I<NOT> dereferencing
269 C<$arrayref[0]> or C<$hashref{"KEY"}> there. The dereference of the
270 scalar variable happens I<BEFORE> it does any key lookups. Anything more
271 complicated than a simple scalar variable must use methods 2 or 3 below.
272 However, a "simple scalar" includes an identifier that itself uses method
273 1 recursively. Therefore, the following prints "howdy".
275 $refrefref = \\\"howdy";
280 Anywhere you'd put an identifier (or chain of identifiers) as part of a
281 variable or subroutine name, you can replace the identifier with a
282 BLOCK returning a reference of the correct type. In other words, the
283 previous examples could be written like this:
285 $bar = ${$scalarref};
286 push(@{$arrayref}, $filename);
287 ${$arrayref}[0] = "January";
288 ${$hashref}{"KEY"} = "VALUE";
290 $globref->print("output\n"); # iff IO::Handle is loaded
292 Admittedly, it's a little silly to use the curlies in this case, but
293 the BLOCK can contain any arbitrary expression, in particular,
294 subscripted expressions:
296 &{ $dispatch{$index} }(1,2,3); # call correct routine
298 Because of being able to omit the curlies for the simple case of C<$$x>,
299 people often make the mistake of viewing the dereferencing symbols as
300 proper operators, and wonder about their precedence. If they were,
301 though, you could use parentheses instead of braces. That's not the case.
302 Consider the difference below; case 0 is a short-hand version of case 1,
305 $$hashref{"KEY"} = "VALUE"; # CASE 0
306 ${$hashref}{"KEY"} = "VALUE"; # CASE 1
307 ${$hashref{"KEY"}} = "VALUE"; # CASE 2
308 ${$hashref->{"KEY"}} = "VALUE"; # CASE 3
310 Case 2 is also deceptive in that you're accessing a variable
311 called %hashref, not dereferencing through $hashref to the hash
312 it's presumably referencing. That would be case 3.
316 The case of individual array elements arises often enough that it gets
317 cumbersome to use method 2. As a form of syntactic sugar, the two
318 lines like that above can be written:
320 $arrayref->[0] = "January";
321 $hashref->{"KEY"} = "VALUE";
323 The left side of the array can be any expression returning a reference,
324 including a previous dereference. Note that C<$array[$x]> is I<NOT> the
325 same thing as C<$array-E<gt>[$x]> here:
327 $array[$x]->{"foo"}->[0] = "January";
329 This is one of the cases we mentioned earlier in which references could
330 spring into existence when in an lvalue context. Before this
331 statement, C<$array[$x]> may have been undefined. If so, it's
332 automatically defined with a hash reference so that we can look up
333 C<{"foo"}> in it. Likewise C<$array[$x]-E<gt>{"foo"}> will automatically get
334 defined with an array reference so that we can look up C<[0]> in it.
336 One more thing here. The arrow is optional I<BETWEEN> brackets
337 subscripts, so you can shrink the above down to
339 $array[$x]{"foo"}[0] = "January";
341 Which, in the degenerate case of using only ordinary arrays, gives you
342 multidimensional arrays just like C's:
344 $score[$x][$y][$z] += 42;
346 Well, okay, not entirely like C's arrays, actually. C doesn't know how
347 to grow its arrays on demand. Perl does.
351 If a reference happens to be a reference to an object, then there are
352 probably methods to access the things referred to, and you should probably
353 stick to those methods unless you're in the class package that defines the
354 object's methods. In other words, be nice, and don't violate the object's
355 encapsulation without a very good reason. Perl does not enforce
356 encapsulation. We are not totalitarians here. We do expect some basic
361 The ref() operator may be used to determine what type of thing the
362 reference is pointing to. See L<perlfunc>.
364 The bless() operator may be used to associate a reference with a package
365 functioning as an object class. See L<perlobj>.
367 A typeglob may be dereferenced the same way a reference can, because
368 the dereference syntax always indicates the kind of reference desired.
369 So C<${*foo}> and C<${\$foo}> both indicate the same scalar variable.
371 Here's a trick for interpolating a subroutine call into a string:
373 print "My sub returned @{[mysub(1,2,3)]} that time.\n";
375 The way it works is that when the C<@{...}> is seen in the double-quoted
376 string, it's evaluated as a block. The block creates a reference to an
377 anonymous array containing the results of the call to C<mysub(1,2,3)>. So
378 the whole block returns a reference to an array, which is then
379 dereferenced by C<@{...}> and stuck into the double-quoted string. This
380 chicanery is also useful for arbitrary expressions:
382 print "That yields @{[$n + 5]} widgets\n";
384 =head2 Symbolic references
386 We said that references spring into existence as necessary if they are
387 undefined, but we didn't say what happens if a value used as a
388 reference is already defined, but I<ISN'T> a hard reference. If you
389 use it as a reference in this case, it'll be treated as a symbolic
390 reference. That is, the value of the scalar is taken to be the I<NAME>
391 of a variable, rather than a direct link to a (possibly) anonymous
394 People frequently expect it to work like this. So it does.
397 $$name = 1; # Sets $foo
398 ${$name} = 2; # Sets $foo
399 ${$name x 2} = 3; # Sets $foofoo
400 $name->[0] = 4; # Sets $foo[0]
401 @$name = (); # Clears @foo
402 &$name(); # Calls &foo() (as in Perl 4)
404 ${"${pack}::$name"} = 5; # Sets $THAT::foo without eval
406 This is very powerful, and slightly dangerous, in that it's possible
407 to intend (with the utmost sincerity) to use a hard reference, and
408 accidentally use a symbolic reference instead. To protect against
413 and then only hard references will be allowed for the rest of the enclosing
414 block. An inner block may countermand that with
418 Only package variables are visible to symbolic references. Lexical
419 variables (declared with my()) aren't in a symbol table, and thus are
420 invisible to this mechanism. For example:
429 This will still print 10, not 20. Remember that local() affects package
430 variables, which are all "global" to the package.
432 =head2 Not-so-symbolic references
434 A new feature contributing to readability in perl version 5.001 is that the
435 brackets around a symbolic reference behave more like quotes, just as they
436 always have within a string. That is,
441 has always meant to print "pop on over", despite the fact that push is
442 a reserved word. This has been generalized to work the same outside
445 print ${push} . "over";
449 print ${ push } . "over";
451 will have the same effect. (This would have been a syntax error in
452 Perl 5.000, though Perl 4 allowed it in the spaceless form.) Note that this
453 construct is I<not> considered to be a symbolic reference when you're
457 ${ bareword }; # Okay, means $bareword.
458 ${ "bareword" }; # Error, symbolic reference.
460 Similarly, because of all the subscripting that is done using single
461 words, we've applied the same rule to any bareword that is used for
462 subscripting a hash. So now, instead of writing
464 $array{ "aaa" }{ "bbb" }{ "ccc" }
468 $array{ aaa }{ bbb }{ ccc }
470 and not worry about whether the subscripts are reserved words. In the
471 rare event that you do wish to do something like
475 you can force interpretation as a reserved word by adding anything that
476 makes it more than a bareword:
482 The B<-w> switch will warn you if it interprets a reserved word as a string.
483 But it will no longer warn you about using lowercase words, because the
484 string is effectively quoted.
488 You may not (usefully) use a reference as the key to a hash. It will be
489 converted into a string:
493 If you try to dereference the key, it won't do a hard dereference, and
494 you won't accomplish what you're attempting. You might want to do something
500 And then at least you can use the values(), which will be
501 real refs, instead of the keys(), which won't.
505 Besides the obvious documents, source code can be instructive.
506 Some rather pathological examples of the use of references can be found
507 in the F<t/op/ref.t> regression test in the Perl source directory.
509 See also L<perldsc> and L<perllol> for how to use references to create
510 complex data structures, and L<perlobj> for how to use them to create