7 A Perl script consists of a sequence of declarations and statements.
8 The sequence of statements is executed just once, unlike in B<sed>
9 and B<awk> scripts, where the sequence of statements is executed
10 for each input line. While this means that you must explicitly
11 loop over the lines of your input file (or files), it also means
12 you have much more control over which files and which lines you look at.
13 (Actually, I'm lying--it is possible to do an implicit loop with
14 either the B<-n> or B<-p> switch. It's just not the mandatory
15 default like it is in B<sed> and B<awk>.)
17 Perl is, for the most part, a free-form language. (The only exception
18 to this is format declarations, for obvious reasons.) Text from a
19 C<"#"> character until the end of the line is a comment, and is
20 ignored. If you attempt to use C</* */> C-style comments, it will be
21 interpreted either as division or pattern matching, depending on the
22 context, and C++ C<//> comments just look like a null regular
23 expression, so don't do that.
27 The only things you need to declare in Perl are report formats
28 and subroutines--and even undefined subroutines can be handled
29 through AUTOLOAD. A variable holds the undefined value (C<undef>)
30 until it has been assigned a defined value, which is anything
31 other than C<undef>. When used as a number, C<undef> is treated
32 as C<0>; when used as a string, it is treated the empty string,
33 C<"">; and when used as a reference that isn't being assigned
34 to, it is treated as an error. If you enable warnings, you'll
35 be notified of an uninitialized value whenever you treat C<undef>
36 as a string or a number. Well, usually. Boolean contexts, such as:
41 are exempt from warnings (because they care about truth rather than
42 definedness). Operators such as C<++>, C<-->, C<+=>,
43 C<-=>, and C<.=>, that operate on undefined left values such as:
48 are also always exempt from such warnings.
50 A declaration can be put anywhere a statement can, but has no effect on
51 the execution of the primary sequence of statements--declarations all
52 take effect at compile time. Typically all the declarations are put at
53 the beginning or the end of the script. However, if you're using
54 lexically-scoped private variables created with C<my()>, you'll
56 your format or subroutine definition is within the same block scope
57 as the my if you expect to be able to access those private variables.
59 Declaring a subroutine allows a subroutine name to be used as if it were a
60 list operator from that point forward in the program. You can declare a
61 subroutine without defining it by saying C<sub name>, thus:
64 $me = myname $0 or die "can't get myname";
66 Note that myname() functions as a list operator, not as a unary operator;
67 so be careful to use C<or> instead of C<||> in this case. However, if
68 you were to declare the subroutine as C<sub myname ($)>, then
69 C<myname> would function as a unary operator, so either C<or> or
72 Subroutines declarations can also be loaded up with the C<require> statement
73 or both loaded and imported into your namespace with a C<use> statement.
74 See L<perlmod> for details on this.
76 A statement sequence may contain declarations of lexically-scoped
77 variables, but apart from declaring a variable name, the declaration acts
78 like an ordinary statement, and is elaborated within the sequence of
79 statements as if it were an ordinary statement. That means it actually
80 has both compile-time and run-time effects.
82 =head2 Simple statements
84 The only kind of simple statement is an expression evaluated for its
85 side effects. Every simple statement must be terminated with a
86 semicolon, unless it is the final statement in a block, in which case
87 the semicolon is optional. (A semicolon is still encouraged there if the
88 block takes up more than one line, because you may eventually add another line.)
89 Note that there are some operators like C<eval {}> and C<do {}> that look
90 like compound statements, but aren't (they're just TERMs in an expression),
91 and thus need an explicit termination if used as the last item in a statement.
93 Any simple statement may optionally be followed by a I<SINGLE> modifier,
94 just before the terminating semicolon (or block ending). The possible
103 The C<if> and C<unless> modifiers have the expected semantics,
104 presuming you're a speaker of English. The C<foreach> modifier is an
105 iterator: For each value in EXPR, it aliases C<$_> to the value and
106 executes the statement. The C<while> and C<until> modifiers have the
107 usual "C<while> loop" semantics (conditional evaluated first), except
108 when applied to a C<do>-BLOCK (or to the deprecated C<do>-SUBROUTINE
109 statement), in which case the block executes once before the
110 conditional is evaluated. This is so that you can write loops like:
115 } until $line eq ".\n";
117 See L<perlfunc/do>. Note also that the loop control statements described
118 later will I<NOT> work in this construct, because modifiers don't take
119 loop labels. Sorry. You can always put another block inside of it
120 (for C<next>) or around it (for C<last>) to do that sort of thing.
121 For C<next>, just double the braces:
128 For C<last>, you have to be more elaborate:
137 B<NOTE:> The behaviour of a C<my> statement modified with a statement
138 modifier conditional or loop construct (e.g. C<my $x if ...>) is
139 B<undefined>. The value of the C<my> variable may be C<undef>, any
140 previously assigned value, or possibly anything else. Don't rely on
141 it. Future versions of perl might do something different from the
142 version of perl you try it out on. Here be dragons.
144 =head2 Compound statements
146 In Perl, a sequence of statements that defines a scope is called a block.
147 Sometimes a block is delimited by the file containing it (in the case
148 of a required file, or the program as a whole), and sometimes a block
149 is delimited by the extent of a string (in the case of an eval).
151 But generally, a block is delimited by curly brackets, also known as braces.
152 We will call this syntactic construct a BLOCK.
154 The following compound statements may be used to control flow:
157 if (EXPR) BLOCK else BLOCK
158 if (EXPR) BLOCK elsif (EXPR) BLOCK ... else BLOCK
159 LABEL while (EXPR) BLOCK
160 LABEL while (EXPR) BLOCK continue BLOCK
161 LABEL for (EXPR; EXPR; EXPR) BLOCK
162 LABEL foreach VAR (LIST) BLOCK
163 LABEL foreach VAR (LIST) BLOCK continue BLOCK
164 LABEL BLOCK continue BLOCK
166 Note that, unlike C and Pascal, these are defined in terms of BLOCKs,
167 not statements. This means that the curly brackets are I<required>--no
168 dangling statements allowed. If you want to write conditionals without
169 curly brackets there are several other ways to do it. The following
170 all do the same thing:
172 if (!open(FOO)) { die "Can't open $FOO: $!"; }
173 die "Can't open $FOO: $!" unless open(FOO);
174 open(FOO) or die "Can't open $FOO: $!"; # FOO or bust!
175 open(FOO) ? 'hi mom' : die "Can't open $FOO: $!";
176 # a bit exotic, that last one
178 The C<if> statement is straightforward. Because BLOCKs are always
179 bounded by curly brackets, there is never any ambiguity about which
180 C<if> an C<else> goes with. If you use C<unless> in place of C<if>,
181 the sense of the test is reversed.
183 The C<while> statement executes the block as long as the expression is
184 true (does not evaluate to the null string C<""> or C<0> or C<"0">).
185 The LABEL is optional, and if present, consists of an identifier followed
186 by a colon. The LABEL identifies the loop for the loop control
187 statements C<next>, C<last>, and C<redo>.
188 If the LABEL is omitted, the loop control statement
189 refers to the innermost enclosing loop. This may include dynamically
190 looking back your call-stack at run time to find the LABEL. Such
191 desperate behavior triggers a warning if you use the C<use warnings>
192 pragma or the B<-w> flag.
194 If there is a C<continue> BLOCK, it is always executed just before the
195 conditional is about to be evaluated again, just like the third part of a
196 C<for> loop in C. Thus it can be used to increment a loop variable, even
197 when the loop has been continued via the C<next> statement (which is
198 similar to the C C<continue> statement).
202 The C<next> command is like the C<continue> statement in C; it starts
203 the next iteration of the loop:
205 LINE: while (<STDIN>) {
206 next LINE if /^#/; # discard comments
210 The C<last> command is like the C<break> statement in C (as used in
211 loops); it immediately exits the loop in question. The
212 C<continue> block, if any, is not executed:
214 LINE: while (<STDIN>) {
215 last LINE if /^$/; # exit when done with header
219 The C<redo> command restarts the loop block without evaluating the
220 conditional again. The C<continue> block, if any, is I<not> executed.
221 This command is normally used by programs that want to lie to themselves
222 about what was just input.
224 For example, when processing a file like F</etc/termcap>.
225 If your input lines might end in backslashes to indicate continuation, you
226 want to skip ahead and get the next record.
237 which is Perl short-hand for the more explicitly written version:
239 LINE: while (defined($line = <ARGV>)) {
241 if ($line =~ s/\\$//) {
243 redo LINE unless eof(); # not eof(ARGV)!
248 Note that if there were a C<continue> block on the above code, it would
249 get executed only on lines discarded by the regex (since redo skips the
250 continue block). A continue block is often used to reset line counters
251 or C<?pat?> one-time matches:
253 # inspired by :1,$g/fred/s//WILMA/
255 ?(fred)? && s//WILMA $1 WILMA/;
256 ?(barney)? && s//BETTY $1 BETTY/;
257 ?(homer)? && s//MARGE $1 MARGE/;
259 print "$ARGV $.: $_";
260 close ARGV if eof(); # reset $.
261 reset if eof(); # reset ?pat?
264 If the word C<while> is replaced by the word C<until>, the sense of the
265 test is reversed, but the conditional is still tested before the first
268 The loop control statements don't work in an C<if> or C<unless>, since
269 they aren't loops. You can double the braces to make them such, though.
273 next if /barney/; # same effect as "last", but doesn't document as well
277 This is caused by the fact that a block by itself acts as a loop that
278 executes once, see L<"Basic BLOCKs and Switch Statements">.
280 The form C<while/if BLOCK BLOCK>, available in Perl 4, is no longer
281 available. Replace any occurrence of C<if BLOCK> by C<if (do BLOCK)>.
285 Perl's C-style C<for> loop works like the corresponding C<while> loop;
286 that means that this:
288 for ($i = 1; $i < 10; $i++) {
301 There is one minor difference: if variables are declared with C<my>
302 in the initialization section of the C<for>, the lexical scope of
303 those variables is exactly the C<for> loop (the body of the loop
304 and the control sections).
306 Besides the normal array index looping, C<for> can lend itself
307 to many other interesting applications. Here's one that avoids the
308 problem you get into if you explicitly test for end-of-file on
309 an interactive file descriptor causing your program to appear to
312 $on_a_tty = -t STDIN && -t STDOUT;
313 sub prompt { print "yes? " if $on_a_tty }
314 for ( prompt(); <STDIN>; prompt() ) {
320 The C<foreach> loop iterates over a normal list value and sets the
321 variable VAR to be each element of the list in turn. If the variable
322 is preceded with the keyword C<my>, then it is lexically scoped, and
323 is therefore visible only within the loop. Otherwise, the variable is
324 implicitly local to the loop and regains its former value upon exiting
325 the loop. If the variable was previously declared with C<my>, it uses
326 that variable instead of the global one, but it's still localized to
327 the loop. This implicit localisation occurs I<only> in a C<foreach>
330 The C<foreach> keyword is actually a synonym for the C<for> keyword, so
331 you can use C<foreach> for readability or C<for> for brevity. (Or because
332 the Bourne shell is more familiar to you than I<csh>, so writing C<for>
333 comes more naturally.) If VAR is omitted, C<$_> is set to each value.
335 If any element of LIST is an lvalue, you can modify it by modifying
336 VAR inside the loop. Conversely, if any element of LIST is NOT an
337 lvalue, any attempt to modify that element will fail. In other words,
338 the C<foreach> loop index variable is an implicit alias for each item
339 in the list that you're looping over.
341 If any part of LIST is an array, C<foreach> will get very confused if
342 you add or remove elements within the loop body, for example with
343 C<splice>. So don't do that.
345 C<foreach> probably won't do what you expect if VAR is a tied or other
346 special variable. Don't do that either.
350 for (@ary) { s/foo/bar/ }
352 for my $elem (@elements) {
356 for $count (10,9,8,7,6,5,4,3,2,1,'BOOM') {
357 print $count, "\n"; sleep(1);
360 for (1..15) { print "Merry Christmas\n"; }
362 foreach $item (split(/:[\\\n:]*/, $ENV{TERMCAP})) {
363 print "Item: $item\n";
366 Here's how a C programmer might code up a particular algorithm in Perl:
368 for (my $i = 0; $i < @ary1; $i++) {
369 for (my $j = 0; $j < @ary2; $j++) {
370 if ($ary1[$i] > $ary2[$j]) {
371 last; # can't go to outer :-(
373 $ary1[$i] += $ary2[$j];
375 # this is where that last takes me
378 Whereas here's how a Perl programmer more comfortable with the idiom might
381 OUTER: for my $wid (@ary1) {
382 INNER: for my $jet (@ary2) {
383 next OUTER if $wid > $jet;
388 See how much easier this is? It's cleaner, safer, and faster. It's
389 cleaner because it's less noisy. It's safer because if code gets added
390 between the inner and outer loops later on, the new code won't be
391 accidentally executed. The C<next> explicitly iterates the other loop
392 rather than merely terminating the inner one. And it's faster because
393 Perl executes a C<foreach> statement more rapidly than it would the
394 equivalent C<for> loop.
396 =head2 Basic BLOCKs and Switch Statements
398 A BLOCK by itself (labeled or not) is semantically equivalent to a
399 loop that executes once. Thus you can use any of the loop control
400 statements in it to leave or restart the block. (Note that this is
401 I<NOT> true in C<eval{}>, C<sub{}>, or contrary to popular belief
402 C<do{}> blocks, which do I<NOT> count as loops.) The C<continue>
405 The BLOCK construct is particularly nice for doing case
409 if (/^abc/) { $abc = 1; last SWITCH; }
410 if (/^def/) { $def = 1; last SWITCH; }
411 if (/^xyz/) { $xyz = 1; last SWITCH; }
415 There is no official C<switch> statement in Perl, because there are
416 already several ways to write the equivalent.
418 However, starting from Perl 5.8 to get switch and case one can use
419 the Switch extension and say:
423 after which one has switch and case. It is not as fast as it could be
424 because it's not really part of the language (it's done using source
425 filters) but it is available, and it's very flexible.
427 In addition to the above BLOCK construct, you could write
430 $abc = 1, last SWITCH if /^abc/;
431 $def = 1, last SWITCH if /^def/;
432 $xyz = 1, last SWITCH if /^xyz/;
436 (That's actually not as strange as it looks once you realize that you can
437 use loop control "operators" within an expression, That's just the normal
443 /^abc/ && do { $abc = 1; last SWITCH; };
444 /^def/ && do { $def = 1; last SWITCH; };
445 /^xyz/ && do { $xyz = 1; last SWITCH; };
449 or formatted so it stands out more as a "proper" C<switch> statement:
472 /^abc/ and $abc = 1, last SWITCH;
473 /^def/ and $def = 1, last SWITCH;
474 /^xyz/ and $xyz = 1, last SWITCH;
489 A common idiom for a C<switch> statement is to use C<foreach>'s aliasing to make
490 a temporary assignment to C<$_> for convenient matching:
492 SWITCH: for ($where) {
493 /In Card Names/ && do { push @flags, '-e'; last; };
494 /Anywhere/ && do { push @flags, '-h'; last; };
495 /In Rulings/ && do { last; };
496 die "unknown value for form variable where: `$where'";
499 Another interesting approach to a switch statement is arrange
500 for a C<do> block to return the proper value:
503 if ($flag & O_RDONLY) { "r" } # XXX: isn't this 0?
504 elsif ($flag & O_WRONLY) { ($flag & O_APPEND) ? "a" : "w" }
505 elsif ($flag & O_RDWR) {
506 if ($flag & O_CREAT) { "w+" }
507 else { ($flag & O_APPEND) ? "a+" : "r+" }
514 ($flags & O_WRONLY) ? "write-only" :
515 ($flags & O_RDWR) ? "read-write" :
519 Or if you are certain that all the C<&&> clauses are true, you can use
520 something like this, which "switches" on the value of the
521 C<HTTP_USER_AGENT> environment variable.
524 # pick out jargon file page based on browser
525 $dir = 'http://www.wins.uva.nl/~mes/jargon';
526 for ($ENV{HTTP_USER_AGENT}) {
527 $page = /Mac/ && 'm/Macintrash.html'
528 || /Win(dows )?NT/ && 'e/evilandrude.html'
529 || /Win|MSIE|WebTV/ && 'm/MicroslothWindows.html'
530 || /Linux/ && 'l/Linux.html'
531 || /HP-UX/ && 'h/HP-SUX.html'
532 || /SunOS/ && 's/ScumOS.html'
533 || 'a/AppendixB.html';
535 print "Location: $dir/$page\015\012\015\012";
537 That kind of switch statement only works when you know the C<&&> clauses
538 will be true. If you don't, the previous C<?:> example should be used.
540 You might also consider writing a hash of subroutine references
541 instead of synthesizing a C<switch> statement.
545 Although not for the faint of heart, Perl does support a C<goto>
546 statement. There are three forms: C<goto>-LABEL, C<goto>-EXPR, and
547 C<goto>-&NAME. A loop's LABEL is not actually a valid target for
548 a C<goto>; it's just the name of the loop.
550 The C<goto>-LABEL form finds the statement labeled with LABEL and resumes
551 execution there. It may not be used to go into any construct that
552 requires initialization, such as a subroutine or a C<foreach> loop. It
553 also can't be used to go into a construct that is optimized away. It
554 can be used to go almost anywhere else within the dynamic scope,
555 including out of subroutines, but it's usually better to use some other
556 construct such as C<last> or C<die>. The author of Perl has never felt the
557 need to use this form of C<goto> (in Perl, that is--C is another matter).
559 The C<goto>-EXPR form expects a label name, whose scope will be resolved
560 dynamically. This allows for computed C<goto>s per FORTRAN, but isn't
561 necessarily recommended if you're optimizing for maintainability:
563 goto(("FOO", "BAR", "GLARCH")[$i]);
565 The C<goto>-&NAME form is highly magical, and substitutes a call to the
566 named subroutine for the currently running subroutine. This is used by
567 C<AUTOLOAD()> subroutines that wish to load another subroutine and then
568 pretend that the other subroutine had been called in the first place
569 (except that any modifications to C<@_> in the current subroutine are
570 propagated to the other subroutine.) After the C<goto>, not even C<caller()>
571 will be able to tell that this routine was called first.
573 In almost all cases like this, it's usually a far, far better idea to use the
574 structured control flow mechanisms of C<next>, C<last>, or C<redo> instead of
575 resorting to a C<goto>. For certain applications, the catch and throw pair of
576 C<eval{}> and die() for exception processing can also be a prudent approach.
578 =head2 PODs: Embedded Documentation
580 Perl has a mechanism for intermixing documentation with source code.
581 While it's expecting the beginning of a new statement, if the compiler
582 encounters a line that begins with an equal sign and a word, like this
584 =head1 Here There Be Pods!
586 Then that text and all remaining text up through and including a line
587 beginning with C<=cut> will be ignored. The format of the intervening
588 text is described in L<perlpod>.
590 This allows you to intermix your source code
591 and your documentation text freely, as in
595 The snazzle() function will behave in the most spectacular
596 form that you can possibly imagine, not even excepting
597 cybernetic pyrotechnics.
599 =cut back to the compiler, nuff of this pod stuff!
606 Note that pod translators should look at only paragraphs beginning
607 with a pod directive (it makes parsing easier), whereas the compiler
608 actually knows to look for pod escapes even in the middle of a
609 paragraph. This means that the following secret stuff will be
610 ignored by both the compiler and the translators.
614 warn "Neither POD nor CODE!?"
618 You probably shouldn't rely upon the C<warn()> being podded out forever.
619 Not all pod translators are well-behaved in this regard, and perhaps
620 the compiler will become pickier.
622 One may also use pod directives to quickly comment out a section
625 =head2 Plain Old Comments (Not!)
627 Much like the C preprocessor, Perl can process line directives. Using
628 this, one can control Perl's idea of filenames and line numbers in
629 error or warning messages (especially for strings that are processed
630 with C<eval()>). The syntax for this mechanism is the same as for most
631 C preprocessors: it matches the regular expression
632 C</^#\s*line\s+(\d+)\s*(?:\s"([^"]+)")?\s*$/> with C<$1> being the line
633 number for the next line, and C<$2> being the optional filename
634 (specified within quotes).
636 There is a fairly obvious gotcha included with the line directive:
637 Debuggers and profilers will only show the last source line to appear
638 at a particular line number in a given file. Care should be taken not
639 to cause line number collisions in code you'd like to debug later.
641 Here are some examples that you should be able to type into your command
646 # the `#' on the previous line must be the first char on line
649 foo at bzzzt line 201.
653 eval qq[\n#line 2001 ""\ndie 'foo']; print $@;
658 eval qq[\n#line 200 "foo bar"\ndie 'foo']; print $@;
660 foo at foo bar line 200.
664 eval "\n#line " . __LINE__ . ' "' . __FILE__ ."\"\ndie 'foo'";
667 foo at goop line 345.