3 perlop - Perl operators and precedence
7 Perl operators have the following associativity and precedence,
8 listed from highest precedence to lowest. Operators borrowed from
9 C keep the same precedence relationship with each other, even where
10 C's precedence is slightly screwy. (This makes learning Perl easier
11 for C folks.) With very few exceptions, these all operate on scalar
12 values only, not array values.
14 left terms and list operators (leftward)
18 right ! ~ \ and unary + and -
23 nonassoc named unary operators
24 nonassoc < > <= >= lt gt le ge
25 nonassoc == != <=> eq ne cmp
34 nonassoc list operators (rightward)
39 In the following sections, these operators are covered in precedence order.
41 Many operators can be overloaded for objects. See L<overload>.
45 =head2 Terms and List Operators (Leftward)
47 A TERM has the highest precedence in Perl. They include variables,
48 quote and quote-like operators, any expression in parentheses,
49 and any function whose arguments are parenthesized. Actually, there
50 aren't really functions in this sense, just list operators and unary
51 operators behaving as functions because you put parentheses around
52 the arguments. These are all documented in L<perlfunc>.
54 If any list operator (print(), etc.) or any unary operator (chdir(), etc.)
55 is followed by a left parenthesis as the next token, the operator and
56 arguments within parentheses are taken to be of highest precedence,
57 just like a normal function call.
59 In the absence of parentheses, the precedence of list operators such as
60 C<print>, C<sort>, or C<chmod> is either very high or very low depending on
61 whether you are looking at the left side or the right side of the operator.
64 @ary = (1, 3, sort 4, 2);
65 print @ary; # prints 1324
67 the commas on the right of the sort are evaluated before the sort,
68 but the commas on the left are evaluated after. In other words,
69 list operators tend to gobble up all arguments that follow, and
70 then act like a simple TERM with regard to the preceding expression.
71 Be careful with parentheses:
73 # These evaluate exit before doing the print:
74 print($foo, exit); # Obviously not what you want.
75 print $foo, exit; # Nor is this.
77 # These do the print before evaluating exit:
78 (print $foo), exit; # This is what you want.
79 print($foo), exit; # Or this.
80 print ($foo), exit; # Or even this.
84 print ($foo & 255) + 1, "\n";
86 probably doesn't do what you expect at first glance. See
87 L<Named Unary Operators> for more discussion of this.
89 Also parsed as terms are the C<do {}> and C<eval {}> constructs, as
90 well as subroutine and method calls, and the anonymous
91 constructors C<[]> and C<{}>.
93 See also L<Quote and Quote-like Operators> toward the end of this section,
94 as well as L<"I/O Operators">.
96 =head2 The Arrow Operator
98 "C<< -> >>" is an infix dereference operator, just as it is in C
99 and C++. If the right side is either a C<[...]>, C<{...}>, or a
100 C<(...)> subscript, then the left side must be either a hard or
101 symbolic reference to an array, a hash, or a subroutine respectively.
102 (Or technically speaking, a location capable of holding a hard
103 reference, if it's an array or hash reference being used for
104 assignment.) See L<perlreftut> and L<perlref>.
106 Otherwise, the right side is a method name or a simple scalar
107 variable containing either the method name or a subroutine reference,
108 and the left side must be either an object (a blessed reference)
109 or a class name (that is, a package name). See L<perlobj>.
111 =head2 Auto-increment and Auto-decrement
113 "++" and "--" work as in C. That is, if placed before a variable, they
114 increment or decrement the variable before returning the value, and if
115 placed after, increment or decrement the variable after returning the value.
117 The auto-increment operator has a little extra builtin magic to it. If
118 you increment a variable that is numeric, or that has ever been used in
119 a numeric context, you get a normal increment. If, however, the
120 variable has been used in only string contexts since it was set, and
121 has a value that is not the empty string and matches the pattern
122 C</^[a-zA-Z]*[0-9]*\z/>, the increment is done as a string, preserving each
123 character within its range, with carry:
125 print ++($foo = '99'); # prints '100'
126 print ++($foo = 'a0'); # prints 'a1'
127 print ++($foo = 'Az'); # prints 'Ba'
128 print ++($foo = 'zz'); # prints 'aaa'
130 The auto-decrement operator is not magical.
132 =head2 Exponentiation
134 Binary "**" is the exponentiation operator. It binds even more
135 tightly than unary minus, so -2**4 is -(2**4), not (-2)**4. (This is
136 implemented using C's pow(3) function, which actually works on doubles
139 =head2 Symbolic Unary Operators
141 Unary "!" performs logical negation, i.e., "not". See also C<not> for a lower
142 precedence version of this.
144 Unary "-" performs arithmetic negation if the operand is numeric. If
145 the operand is an identifier, a string consisting of a minus sign
146 concatenated with the identifier is returned. Otherwise, if the string
147 starts with a plus or minus, a string starting with the opposite sign
148 is returned. One effect of these rules is that C<-bareword> is equivalent
151 Unary "~" performs bitwise negation, i.e., 1's complement. For
152 example, C<0666 & ~027> is 0640. (See also L<Integer Arithmetic> and
153 L<Bitwise String Operators>.) Note that the width of the result is
154 platform-dependent: ~0 is 32 bits wide on a 32-bit platform, but 64
155 bits wide on a 64-bit platform, so if you are expecting a certain bit
156 width, remember use the & operator to mask off the excess bits.
158 Unary "+" has no effect whatsoever, even on strings. It is useful
159 syntactically for separating a function name from a parenthesized expression
160 that would otherwise be interpreted as the complete list of function
161 arguments. (See examples above under L<Terms and List Operators (Leftward)>.)
163 Unary "\" creates a reference to whatever follows it. See L<perlreftut>
164 and L<perlref>. Do not confuse this behavior with the behavior of
165 backslash within a string, although both forms do convey the notion
166 of protecting the next thing from interpolation.
168 =head2 Binding Operators
170 Binary "=~" binds a scalar expression to a pattern match. Certain operations
171 search or modify the string $_ by default. This operator makes that kind
172 of operation work on some other string. The right argument is a search
173 pattern, substitution, or transliteration. The left argument is what is
174 supposed to be searched, substituted, or transliterated instead of the default
175 $_. When used in scalar context, the return value generally indicates the
176 success of the operation. Behavior in list context depends on the particular
177 operator. See L</"Regexp Quote-Like Operators"> for details.
179 If the right argument is an expression rather than a search pattern,
180 substitution, or transliteration, it is interpreted as a search pattern at run
181 time. This can be less efficient than an explicit search, because the
182 pattern must be compiled every time the expression is evaluated.
184 Binary "!~" is just like "=~" except the return value is negated in
187 =head2 Multiplicative Operators
189 Binary "*" multiplies two numbers.
191 Binary "/" divides two numbers.
193 Binary "%" computes the modulus of two numbers. Given integer
194 operands C<$a> and C<$b>: If C<$b> is positive, then C<$a % $b> is
195 C<$a> minus the largest multiple of C<$b> that is not greater than
196 C<$a>. If C<$b> is negative, then C<$a % $b> is C<$a> minus the
197 smallest multiple of C<$b> that is not less than C<$a> (i.e. the
198 result will be less than or equal to zero).
199 Note than when C<use integer> is in scope, "%" gives you direct access
200 to the modulus operator as implemented by your C compiler. This
201 operator is not as well defined for negative operands, but it will
204 Binary "x" is the repetition operator. In scalar context or if the left
205 operand is not enclosed in parentheses, it returns a string consisting
206 of the left operand repeated the number of times specified by the right
207 operand. In list context, if the left operand is enclosed in
208 parentheses, it repeats the list.
210 print '-' x 80; # print row of dashes
212 print "\t" x ($tab/8), ' ' x ($tab%8); # tab over
214 @ones = (1) x 80; # a list of 80 1's
215 @ones = (5) x @ones; # set all elements to 5
218 =head2 Additive Operators
220 Binary "+" returns the sum of two numbers.
222 Binary "-" returns the difference of two numbers.
224 Binary "." concatenates two strings.
226 =head2 Shift Operators
228 Binary "<<" returns the value of its left argument shifted left by the
229 number of bits specified by the right argument. Arguments should be
230 integers. (See also L<Integer Arithmetic>.)
232 Binary ">>" returns the value of its left argument shifted right by
233 the number of bits specified by the right argument. Arguments should
234 be integers. (See also L<Integer Arithmetic>.)
236 Note that both "<<" and ">>" in Perl are implemented directly using
237 "<<" and ">>" in C. If C<use integer> (see L<Integer Arithmetic>) is
238 in force then signed C integers are used, else unsigned C integers are
239 used. Either way, the implementation isn't going to generate results
240 larger than the size of the integer type Perl was built with (32 bits
243 The result of overflowing the range of the integers is undefined
244 because it is undefined also in C. In other words, using 32-bit
245 integers, C<< 1 << 32 >> is undefined. Shifting by a negative number
246 of bits is also undefined.
248 =head2 Named Unary Operators
250 The various named unary operators are treated as functions with one
251 argument, with optional parentheses. These include the filetest
252 operators, like C<-f>, C<-M>, etc. See L<perlfunc>.
254 If any list operator (print(), etc.) or any unary operator (chdir(), etc.)
255 is followed by a left parenthesis as the next token, the operator and
256 arguments within parentheses are taken to be of highest precedence,
257 just like a normal function call. For example,
258 because named unary operators are higher precedence than ||:
260 chdir $foo || die; # (chdir $foo) || die
261 chdir($foo) || die; # (chdir $foo) || die
262 chdir ($foo) || die; # (chdir $foo) || die
263 chdir +($foo) || die; # (chdir $foo) || die
265 but, because * is higher precedence than named operators:
267 chdir $foo * 20; # chdir ($foo * 20)
268 chdir($foo) * 20; # (chdir $foo) * 20
269 chdir ($foo) * 20; # (chdir $foo) * 20
270 chdir +($foo) * 20; # chdir ($foo * 20)
272 rand 10 * 20; # rand (10 * 20)
273 rand(10) * 20; # (rand 10) * 20
274 rand (10) * 20; # (rand 10) * 20
275 rand +(10) * 20; # rand (10 * 20)
277 See also L<"Terms and List Operators (Leftward)">.
279 =head2 Relational Operators
281 Binary "<" returns true if the left argument is numerically less than
284 Binary ">" returns true if the left argument is numerically greater
285 than the right argument.
287 Binary "<=" returns true if the left argument is numerically less than
288 or equal to the right argument.
290 Binary ">=" returns true if the left argument is numerically greater
291 than or equal to the right argument.
293 Binary "lt" returns true if the left argument is stringwise less than
296 Binary "gt" returns true if the left argument is stringwise greater
297 than the right argument.
299 Binary "le" returns true if the left argument is stringwise less than
300 or equal to the right argument.
302 Binary "ge" returns true if the left argument is stringwise greater
303 than or equal to the right argument.
305 =head2 Equality Operators
307 Binary "==" returns true if the left argument is numerically equal to
310 Binary "!=" returns true if the left argument is numerically not equal
311 to the right argument.
313 Binary "<=>" returns -1, 0, or 1 depending on whether the left
314 argument is numerically less than, equal to, or greater than the right
315 argument. If your platform supports NaNs (not-a-numbers) as numeric
316 values, using them with "<=>" returns undef. NaN is not "<", "==", ">",
317 "<=" or ">=" anything (even NaN), so those 5 return false. NaN != NaN
318 returns true, as does NaN != anything else. If your platform doesn't
319 support NaNs then NaN is just a string with numeric value 0.
321 perl -le '$a = NaN; print "No NaN support here" if $a == $a'
322 perl -le '$a = NaN; print "NaN support here" if $a != $a'
324 Binary "eq" returns true if the left argument is stringwise equal to
327 Binary "ne" returns true if the left argument is stringwise not equal
328 to the right argument.
330 Binary "cmp" returns -1, 0, or 1 depending on whether the left
331 argument is stringwise less than, equal to, or greater than the right
334 "lt", "le", "ge", "gt" and "cmp" use the collation (sort) order specified
335 by the current locale if C<use locale> is in effect. See L<perllocale>.
339 Binary "&" returns its operators ANDed together bit by bit.
340 (See also L<Integer Arithmetic> and L<Bitwise String Operators>.)
342 =head2 Bitwise Or and Exclusive Or
344 Binary "|" returns its operators ORed together bit by bit.
345 (See also L<Integer Arithmetic> and L<Bitwise String Operators>.)
347 Binary "^" returns its operators XORed together bit by bit.
348 (See also L<Integer Arithmetic> and L<Bitwise String Operators>.)
350 =head2 C-style Logical And
352 Binary "&&" performs a short-circuit logical AND operation. That is,
353 if the left operand is false, the right operand is not even evaluated.
354 Scalar or list context propagates down to the right operand if it
357 =head2 C-style Logical Or
359 Binary "||" performs a short-circuit logical OR operation. That is,
360 if the left operand is true, the right operand is not even evaluated.
361 Scalar or list context propagates down to the right operand if it
364 The C<||> and C<&&> operators differ from C's in that, rather than returning
365 0 or 1, they return the last value evaluated. Thus, a reasonably portable
366 way to find out the home directory (assuming it's not "0") might be:
368 $home = $ENV{'HOME'} || $ENV{'LOGDIR'} ||
369 (getpwuid($<))[7] || die "You're homeless!\n";
371 In particular, this means that you shouldn't use this
372 for selecting between two aggregates for assignment:
374 @a = @b || @c; # this is wrong
375 @a = scalar(@b) || @c; # really meant this
376 @a = @b ? @b : @c; # this works fine, though
378 As more readable alternatives to C<&&> and C<||> when used for
379 control flow, Perl provides C<and> and C<or> operators (see below).
380 The short-circuit behavior is identical. The precedence of "and" and
381 "or" is much lower, however, so that you can safely use them after a
382 list operator without the need for parentheses:
384 unlink "alpha", "beta", "gamma"
385 or gripe(), next LINE;
387 With the C-style operators that would have been written like this:
389 unlink("alpha", "beta", "gamma")
390 || (gripe(), next LINE);
392 Using "or" for assignment is unlikely to do what you want; see below.
394 =head2 Range Operators
396 Binary ".." is the range operator, which is really two different
397 operators depending on the context. In list context, it returns an
398 list of values counting (up by ones) from the left value to the right
399 value. If the left value is greater than the right value then it
400 returns the empty array. The range operator is useful for writing
401 C<foreach (1..10)> loops and for doing slice operations on arrays. In
402 the current implementation, no temporary array is created when the
403 range operator is used as the expression in C<foreach> loops, but older
404 versions of Perl might burn a lot of memory when you write something
407 for (1 .. 1_000_000) {
411 The range operator also works on strings, using the magical auto-increment,
414 In scalar context, ".." returns a boolean value. The operator is
415 bistable, like a flip-flop, and emulates the line-range (comma) operator
416 of B<sed>, B<awk>, and various editors. Each ".." operator maintains its
417 own boolean state. It is false as long as its left operand is false.
418 Once the left operand is true, the range operator stays true until the
419 right operand is true, I<AFTER> which the range operator becomes false
420 again. It doesn't become false till the next time the range operator is
421 evaluated. It can test the right operand and become false on the same
422 evaluation it became true (as in B<awk>), but it still returns true once.
423 If you don't want it to test the right operand till the next
424 evaluation, as in B<sed>, just use three dots ("...") instead of
425 two. In all other regards, "..." behaves just like ".." does.
427 The right operand is not evaluated while the operator is in the
428 "false" state, and the left operand is not evaluated while the
429 operator is in the "true" state. The precedence is a little lower
430 than || and &&. The value returned is either the empty string for
431 false, or a sequence number (beginning with 1) for true. The
432 sequence number is reset for each range encountered. The final
433 sequence number in a range has the string "E0" appended to it, which
434 doesn't affect its numeric value, but gives you something to search
435 for if you want to exclude the endpoint. You can exclude the
436 beginning point by waiting for the sequence number to be greater
437 than 1. If either operand of scalar ".." is a constant expression,
438 that operand is implicitly compared to the C<$.> variable, the
439 current line number. Examples:
441 As a scalar operator:
443 if (101 .. 200) { print; } # print 2nd hundred lines
444 next line if (1 .. /^$/); # skip header lines
445 s/^/> / if (/^$/ .. eof()); # quote body
447 # parse mail messages
449 $in_header = 1 .. /^$/;
450 $in_body = /^$/ .. eof();
451 # do something based on those
453 close ARGV if eof; # reset $. each file
458 for (101 .. 200) { print; } # print $_ 100 times
459 @foo = @foo[0 .. $#foo]; # an expensive no-op
460 @foo = @foo[$#foo-4 .. $#foo]; # slice last 5 items
462 The range operator (in list context) makes use of the magical
463 auto-increment algorithm if the operands are strings. You
466 @alphabet = ('A' .. 'Z');
468 to get all normal letters of the English alphabet, or
470 $hexdigit = (0 .. 9, 'a' .. 'f')[$num & 15];
472 to get a hexadecimal digit, or
474 @z2 = ('01' .. '31'); print $z2[$mday];
476 to get dates with leading zeros. If the final value specified is not
477 in the sequence that the magical increment would produce, the sequence
478 goes until the next value would be longer than the final value
481 =head2 Conditional Operator
483 Ternary "?:" is the conditional operator, just as in C. It works much
484 like an if-then-else. If the argument before the ? is true, the
485 argument before the : is returned, otherwise the argument after the :
486 is returned. For example:
488 printf "I have %d dog%s.\n", $n,
489 ($n == 1) ? '' : "s";
491 Scalar or list context propagates downward into the 2nd
492 or 3rd argument, whichever is selected.
494 $a = $ok ? $b : $c; # get a scalar
495 @a = $ok ? @b : @c; # get an array
496 $a = $ok ? @b : @c; # oops, that's just a count!
498 The operator may be assigned to if both the 2nd and 3rd arguments are
499 legal lvalues (meaning that you can assign to them):
501 ($a_or_b ? $a : $b) = $c;
503 Because this operator produces an assignable result, using assignments
504 without parentheses will get you in trouble. For example, this:
506 $a % 2 ? $a += 10 : $a += 2
510 (($a % 2) ? ($a += 10) : $a) += 2
514 ($a % 2) ? ($a += 10) : ($a += 2)
516 That should probably be written more simply as:
518 $a += ($a % 2) ? 10 : 2;
520 =head2 Assignment Operators
522 "=" is the ordinary assignment operator.
524 Assignment operators work as in C. That is,
532 although without duplicating any side effects that dereferencing the lvalue
533 might trigger, such as from tie(). Other assignment operators work similarly.
534 The following are recognized:
541 Although these are grouped by family, they all have the precedence
544 Unlike in C, the scalar assignment operator produces a valid lvalue.
545 Modifying an assignment is equivalent to doing the assignment and
546 then modifying the variable that was assigned to. This is useful
547 for modifying a copy of something, like this:
549 ($tmp = $global) =~ tr [A-Z] [a-z];
560 Similarly, a list assignment in list context produces the list of
561 lvalues assigned to, and a list assignment in scalar context returns
562 the number of elements produced by the expression on the right hand
563 side of the assignment.
565 =head2 Comma Operator
567 Binary "," is the comma operator. In scalar context it evaluates
568 its left argument, throws that value away, then evaluates its right
569 argument and returns that value. This is just like C's comma operator.
571 In list context, it's just the list argument separator, and inserts
572 both its arguments into the list.
574 The => digraph is mostly just a synonym for the comma operator. It's useful for
575 documenting arguments that come in pairs. As of release 5.001, it also forces
576 any word to the left of it to be interpreted as a string.
578 =head2 List Operators (Rightward)
580 On the right side of a list operator, it has very low precedence,
581 such that it controls all comma-separated expressions found there.
582 The only operators with lower precedence are the logical operators
583 "and", "or", and "not", which may be used to evaluate calls to list
584 operators without the need for extra parentheses:
586 open HANDLE, "filename"
587 or die "Can't open: $!\n";
589 See also discussion of list operators in L<Terms and List Operators (Leftward)>.
593 Unary "not" returns the logical negation of the expression to its right.
594 It's the equivalent of "!" except for the very low precedence.
598 Binary "and" returns the logical conjunction of the two surrounding
599 expressions. It's equivalent to && except for the very low
600 precedence. This means that it short-circuits: i.e., the right
601 expression is evaluated only if the left expression is true.
603 =head2 Logical or and Exclusive Or
605 Binary "or" returns the logical disjunction of the two surrounding
606 expressions. It's equivalent to || except for the very low precedence.
607 This makes it useful for control flow
609 print FH $data or die "Can't write to FH: $!";
611 This means that it short-circuits: i.e., the right expression is evaluated
612 only if the left expression is false. Due to its precedence, you should
613 probably avoid using this for assignment, only for control flow.
615 $a = $b or $c; # bug: this is wrong
616 ($a = $b) or $c; # really means this
617 $a = $b || $c; # better written this way
619 However, when it's a list-context assignment and you're trying to use
620 "||" for control flow, you probably need "or" so that the assignment
621 takes higher precedence.
623 @info = stat($file) || die; # oops, scalar sense of stat!
624 @info = stat($file) or die; # better, now @info gets its due
626 Then again, you could always use parentheses.
628 Binary "xor" returns the exclusive-OR of the two surrounding expressions.
629 It cannot short circuit, of course.
631 =head2 C Operators Missing From Perl
633 Here is what C has that Perl doesn't:
639 Address-of operator. (But see the "\" operator for taking a reference.)
643 Dereference-address operator. (Perl's prefix dereferencing
644 operators are typed: $, @, %, and &.)
648 Type-casting operator.
652 =head2 Quote and Quote-like Operators
654 While we usually think of quotes as literal values, in Perl they
655 function as operators, providing various kinds of interpolating and
656 pattern matching capabilities. Perl provides customary quote characters
657 for these behaviors, but also provides a way for you to choose your
658 quote character for any of them. In the following table, a C<{}> represents
659 any pair of delimiters you choose.
661 Customary Generic Meaning Interpolates
666 // m{} Pattern match yes*
668 s{}{} Substitution yes*
669 tr{}{} Transliteration no (but see below)
672 * unless the delimiter is ''.
674 Non-bracketing delimiters use the same character fore and aft, but the four
675 sorts of brackets (round, angle, square, curly) will all nest, which means
684 Note, however, that this does not always work for quoting Perl code:
686 $s = q{ if($a eq "}") ... }; # WRONG
688 is a syntax error. The C<Text::Balanced> module (from CPAN, and
689 starting from Perl 5.8 part of the standard distribution) is able
692 There can be whitespace between the operator and the quoting
693 characters, except when C<#> is being used as the quoting character.
694 C<q#foo#> is parsed as the string C<foo>, while C<q #foo#> is the
695 operator C<q> followed by a comment. Its argument will be taken
696 from the next line. This allows you to write:
698 s {foo} # Replace foo
701 The following escape sequences are available in constructs that interpolate
702 and in transliterations.
709 \a alarm (bell) (BEL)
711 \033 octal char (ESC)
713 \x{263a} wide hex char (SMILEY)
714 \c[ control char (ESC)
715 \N{name} named Unicode character
717 The following escape sequences are available in constructs that interpolate
718 but not in transliterations.
720 \l lowercase next char
721 \u uppercase next char
724 \E end case modification
725 \Q quote non-word characters till \E
727 If C<use locale> is in effect, the case map used by C<\l>, C<\L>,
728 C<\u> and C<\U> is taken from the current locale. See L<perllocale>.
729 If Unicode (for example, C<\N{}> or wide hex characters of 0x100 or
730 beyond) is being used, the case map used by C<\l>, C<\L>, C<\u> and
731 C<\U> is as defined by Unicode. For documentation of C<\N{name}>,
734 All systems use the virtual C<"\n"> to represent a line terminator,
735 called a "newline". There is no such thing as an unvarying, physical
736 newline character. It is only an illusion that the operating system,
737 device drivers, C libraries, and Perl all conspire to preserve. Not all
738 systems read C<"\r"> as ASCII CR and C<"\n"> as ASCII LF. For example,
739 on a Mac, these are reversed, and on systems without line terminator,
740 printing C<"\n"> may emit no actual data. In general, use C<"\n"> when
741 you mean a "newline" for your system, but use the literal ASCII when you
742 need an exact character. For example, most networking protocols expect
743 and prefer a CR+LF (C<"\015\012"> or C<"\cM\cJ">) for line terminators,
744 and although they often accept just C<"\012">, they seldom tolerate just
745 C<"\015">. If you get in the habit of using C<"\n"> for networking,
746 you may be burned some day.
748 For constructs that do interpolate, variables beginning with "C<$>"
749 or "C<@>" are interpolated. Subscripted variables such as C<$a[3]> or
750 C<$href->{key}[0]> are also interpolated, as are array and hash slices.
751 But method calls such as C<$obj->meth> are not.
753 Interpolating an array or slice interpolates the elements in order,
754 separated by the value of C<$">, so is equivalent to interpolating
755 C<join $", @array>. "Punctuation" arrays such as C<@+> are only
756 interpolated if the name is enclosed in braces C<@{+}>.
758 You cannot include a literal C<$> or C<@> within a C<\Q> sequence.
759 An unescaped C<$> or C<@> interpolates the corresponding variable,
760 while escaping will cause the literal string C<\$> to be inserted.
761 You'll need to write something like C<m/\Quser\E\@\Qhost/>.
763 Patterns are subject to an additional level of interpretation as a
764 regular expression. This is done as a second pass, after variables are
765 interpolated, so that regular expressions may be incorporated into the
766 pattern from the variables. If this is not what you want, use C<\Q> to
767 interpolate a variable literally.
769 Apart from the behavior described above, Perl does not expand
770 multiple levels of interpolation. In particular, contrary to the
771 expectations of shell programmers, back-quotes do I<NOT> interpolate
772 within double quotes, nor do single quotes impede evaluation of
773 variables when used within double quotes.
775 =head2 Regexp Quote-Like Operators
777 Here are the quote-like operators that apply to pattern
778 matching and related activities.
784 This is just like the C</pattern/> search, except that it matches only
785 once between calls to the reset() operator. This is a useful
786 optimization when you want to see only the first occurrence of
787 something in each file of a set of files, for instance. Only C<??>
788 patterns local to the current package are reset.
792 # blank line between header and body
795 reset if eof; # clear ?? status for next file
798 This usage is vaguely deprecated, which means it just might possibly
799 be removed in some distant future version of Perl, perhaps somewhere
800 around the year 2168.
802 =item m/PATTERN/cgimosx
804 =item /PATTERN/cgimosx
806 Searches a string for a pattern match, and in scalar context returns
807 true if it succeeds, false if it fails. If no string is specified
808 via the C<=~> or C<!~> operator, the $_ string is searched. (The
809 string specified with C<=~> need not be an lvalue--it may be the
810 result of an expression evaluation, but remember the C<=~> binds
811 rather tightly.) See also L<perlre>. See L<perllocale> for
812 discussion of additional considerations that apply when C<use locale>
817 c Do not reset search position on a failed match when /g is in effect.
818 g Match globally, i.e., find all occurrences.
819 i Do case-insensitive pattern matching.
820 m Treat string as multiple lines.
821 o Compile pattern only once.
822 s Treat string as single line.
823 x Use extended regular expressions.
825 If "/" is the delimiter then the initial C<m> is optional. With the C<m>
826 you can use any pair of non-alphanumeric, non-whitespace characters
827 as delimiters. This is particularly useful for matching path names
828 that contain "/", to avoid LTS (leaning toothpick syndrome). If "?" is
829 the delimiter, then the match-only-once rule of C<?PATTERN?> applies.
830 If "'" is the delimiter, no interpolation is performed on the PATTERN.
832 PATTERN may contain variables, which will be interpolated (and the
833 pattern recompiled) every time the pattern search is evaluated, except
834 for when the delimiter is a single quote. (Note that C<$(>, C<$)>, and
835 C<$|> are not interpolated because they look like end-of-string tests.)
836 If you want such a pattern to be compiled only once, add a C</o> after
837 the trailing delimiter. This avoids expensive run-time recompilations,
838 and is useful when the value you are interpolating won't change over
839 the life of the script. However, mentioning C</o> constitutes a promise
840 that you won't change the variables in the pattern. If you change them,
841 Perl won't even notice. See also L<"qr/STRING/imosx">.
843 If the PATTERN evaluates to the empty string, the last
844 I<successfully> matched regular expression is used instead. In this
845 case, only the C<g> and C<c> flags on the empty pattern is honoured -
846 the other flags are taken from the original pattern. If no match has
847 previously succeeded, this will (silently) act instead as a genuine
848 empty pattern (which will always match).
850 If the C</g> option is not used, C<m//> in list context returns a
851 list consisting of the subexpressions matched by the parentheses in the
852 pattern, i.e., (C<$1>, C<$2>, C<$3>...). (Note that here C<$1> etc. are
853 also set, and that this differs from Perl 4's behavior.) When there are
854 no parentheses in the pattern, the return value is the list C<(1)> for
855 success. With or without parentheses, an empty list is returned upon
860 open(TTY, '/dev/tty');
861 <TTY> =~ /^y/i && foo(); # do foo if desired
863 if (/Version: *([0-9.]*)/) { $version = $1; }
865 next if m#^/usr/spool/uucp#;
870 print if /$arg/o; # compile only once
873 if (($F1, $F2, $Etc) = ($foo =~ /^(\S+)\s+(\S+)\s*(.*)/))
875 This last example splits $foo into the first two words and the
876 remainder of the line, and assigns those three fields to $F1, $F2, and
877 $Etc. The conditional is true if any variables were assigned, i.e., if
880 The C</g> modifier specifies global pattern matching--that is,
881 matching as many times as possible within the string. How it behaves
882 depends on the context. In list context, it returns a list of the
883 substrings matched by any capturing parentheses in the regular
884 expression. If there are no parentheses, it returns a list of all
885 the matched strings, as if there were parentheses around the whole
888 In scalar context, each execution of C<m//g> finds the next match,
889 returning true if it matches, and false if there is no further match.
890 The position after the last match can be read or set using the pos()
891 function; see L<perlfunc/pos>. A failed match normally resets the
892 search position to the beginning of the string, but you can avoid that
893 by adding the C</c> modifier (e.g. C<m//gc>). Modifying the target
894 string also resets the search position.
896 You can intermix C<m//g> matches with C<m/\G.../g>, where C<\G> is a
897 zero-width assertion that matches the exact position where the previous
898 C<m//g>, if any, left off. Without the C</g> modifier, the C<\G> assertion
899 still anchors at pos(), but the match is of course only attempted once.
900 Using C<\G> without C</g> on a target string that has not previously had a
901 C</g> match applied to it is the same as using the C<\A> assertion to match
902 the beginning of the string. Note also that, currently, C<\G> is only
903 properly supported when anchored at the very beginning of the pattern.
908 ($one,$five,$fifteen) = (`uptime` =~ /(\d+\.\d+)/g);
912 while (defined($paragraph = <>)) {
913 while ($paragraph =~ /[a-z]['")]*[.!?]+['")]*\s/g) {
917 print "$sentences\n";
919 # using m//gc with \G
923 print $1 while /(o)/gc; print "', pos=", pos, "\n";
925 print $1 if /\G(q)/gc; print "', pos=", pos, "\n";
927 print $1 while /(p)/gc; print "', pos=", pos, "\n";
929 print "Final: '$1', pos=",pos,"\n" if /\G(.)/;
931 The last example should print:
941 Notice that the final match matched C<q> instead of C<p>, which a match
942 without the C<\G> anchor would have done. Also note that the final match
943 did not update C<pos> -- C<pos> is only updated on a C</g> match. If the
944 final match did indeed match C<p>, it's a good bet that you're running an
945 older (pre-5.6.0) Perl.
947 A useful idiom for C<lex>-like scanners is C</\G.../gc>. You can
948 combine several regexps like this to process a string part-by-part,
949 doing different actions depending on which regexp matched. Each
950 regexp tries to match where the previous one leaves off.
953 $url = new URI::URL "http://www/"; die if $url eq "xXx";
957 print(" digits"), redo LOOP if /\G\d+\b[,.;]?\s*/gc;
958 print(" lowercase"), redo LOOP if /\G[a-z]+\b[,.;]?\s*/gc;
959 print(" UPPERCASE"), redo LOOP if /\G[A-Z]+\b[,.;]?\s*/gc;
960 print(" Capitalized"), redo LOOP if /\G[A-Z][a-z]+\b[,.;]?\s*/gc;
961 print(" MiXeD"), redo LOOP if /\G[A-Za-z]+\b[,.;]?\s*/gc;
962 print(" alphanumeric"), redo LOOP if /\G[A-Za-z0-9]+\b[,.;]?\s*/gc;
963 print(" line-noise"), redo LOOP if /\G[^A-Za-z0-9]+/gc;
964 print ". That's all!\n";
967 Here is the output (split into several lines):
969 line-noise lowercase line-noise lowercase UPPERCASE line-noise
970 UPPERCASE line-noise lowercase line-noise lowercase line-noise
971 lowercase lowercase line-noise lowercase lowercase line-noise
972 MiXeD line-noise. That's all!
978 A single-quoted, literal string. A backslash represents a backslash
979 unless followed by the delimiter or another backslash, in which case
980 the delimiter or backslash is interpolated.
982 $foo = q!I said, "You said, 'She said it.'"!;
983 $bar = q('This is it.');
984 $baz = '\n'; # a two-character string
990 A double-quoted, interpolated string.
993 (*** The previous line contains the naughty word "$1".\n)
994 if /\b(tcl|java|python)\b/i; # :-)
995 $baz = "\n"; # a one-character string
997 =item qr/STRING/imosx
999 This operator quotes (and possibly compiles) its I<STRING> as a regular
1000 expression. I<STRING> is interpolated the same way as I<PATTERN>
1001 in C<m/PATTERN/>. If "'" is used as the delimiter, no interpolation
1002 is done. Returns a Perl value which may be used instead of the
1003 corresponding C</STRING/imosx> expression.
1007 $rex = qr/my.STRING/is;
1014 The result may be used as a subpattern in a match:
1017 $string =~ /foo${re}bar/; # can be interpolated in other patterns
1018 $string =~ $re; # or used standalone
1019 $string =~ /$re/; # or this way
1021 Since Perl may compile the pattern at the moment of execution of qr()
1022 operator, using qr() may have speed advantages in some situations,
1023 notably if the result of qr() is used standalone:
1026 my $patterns = shift;
1027 my @compiled = map qr/$_/i, @$patterns;
1030 foreach my $pat (@compiled) {
1031 $success = 1, last if /$pat/;
1037 Precompilation of the pattern into an internal representation at
1038 the moment of qr() avoids a need to recompile the pattern every
1039 time a match C</$pat/> is attempted. (Perl has many other internal
1040 optimizations, but none would be triggered in the above example if
1041 we did not use qr() operator.)
1045 i Do case-insensitive pattern matching.
1046 m Treat string as multiple lines.
1047 o Compile pattern only once.
1048 s Treat string as single line.
1049 x Use extended regular expressions.
1051 See L<perlre> for additional information on valid syntax for STRING, and
1052 for a detailed look at the semantics of regular expressions.
1058 A string which is (possibly) interpolated and then executed as a
1059 system command with C</bin/sh> or its equivalent. Shell wildcards,
1060 pipes, and redirections will be honored. The collected standard
1061 output of the command is returned; standard error is unaffected. In
1062 scalar context, it comes back as a single (potentially multi-line)
1063 string, or undef if the command failed. In list context, returns a
1064 list of lines (however you've defined lines with $/ or
1065 $INPUT_RECORD_SEPARATOR), or an empty list if the command failed.
1067 Because backticks do not affect standard error, use shell file descriptor
1068 syntax (assuming the shell supports this) if you care to address this.
1069 To capture a command's STDERR and STDOUT together:
1071 $output = `cmd 2>&1`;
1073 To capture a command's STDOUT but discard its STDERR:
1075 $output = `cmd 2>/dev/null`;
1077 To capture a command's STDERR but discard its STDOUT (ordering is
1080 $output = `cmd 2>&1 1>/dev/null`;
1082 To exchange a command's STDOUT and STDERR in order to capture the STDERR
1083 but leave its STDOUT to come out the old STDERR:
1085 $output = `cmd 3>&1 1>&2 2>&3 3>&-`;
1087 To read both a command's STDOUT and its STDERR separately, it's easiest
1088 and safest to redirect them separately to files, and then read from those
1089 files when the program is done:
1091 system("program args 1>/tmp/program.stdout 2>/tmp/program.stderr");
1093 Using single-quote as a delimiter protects the command from Perl's
1094 double-quote interpolation, passing it on to the shell instead:
1096 $perl_info = qx(ps $$); # that's Perl's $$
1097 $shell_info = qx'ps $$'; # that's the new shell's $$
1099 How that string gets evaluated is entirely subject to the command
1100 interpreter on your system. On most platforms, you will have to protect
1101 shell metacharacters if you want them treated literally. This is in
1102 practice difficult to do, as it's unclear how to escape which characters.
1103 See L<perlsec> for a clean and safe example of a manual fork() and exec()
1104 to emulate backticks safely.
1106 On some platforms (notably DOS-like ones), the shell may not be
1107 capable of dealing with multiline commands, so putting newlines in
1108 the string may not get you what you want. You may be able to evaluate
1109 multiple commands in a single line by separating them with the command
1110 separator character, if your shell supports that (e.g. C<;> on many Unix
1111 shells; C<&> on the Windows NT C<cmd> shell).
1113 Beginning with v5.6.0, Perl will attempt to flush all files opened for
1114 output before starting the child process, but this may not be supported
1115 on some platforms (see L<perlport>). To be safe, you may need to set
1116 C<$|> ($AUTOFLUSH in English) or call the C<autoflush()> method of
1117 C<IO::Handle> on any open handles.
1119 Beware that some command shells may place restrictions on the length
1120 of the command line. You must ensure your strings don't exceed this
1121 limit after any necessary interpolations. See the platform-specific
1122 release notes for more details about your particular environment.
1124 Using this operator can lead to programs that are difficult to port,
1125 because the shell commands called vary between systems, and may in
1126 fact not be present at all. As one example, the C<type> command under
1127 the POSIX shell is very different from the C<type> command under DOS.
1128 That doesn't mean you should go out of your way to avoid backticks
1129 when they're the right way to get something done. Perl was made to be
1130 a glue language, and one of the things it glues together is commands.
1131 Just understand what you're getting yourself into.
1133 See L<"I/O Operators"> for more discussion.
1137 Evaluates to a list of the words extracted out of STRING, using embedded
1138 whitespace as the word delimiters. It can be understood as being roughly
1141 split(' ', q/STRING/);
1143 the difference being that it generates a real list at compile time. So
1148 is semantically equivalent to the list:
1152 Some frequently seen examples:
1154 use POSIX qw( setlocale localeconv )
1155 @EXPORT = qw( foo bar baz );
1157 A common mistake is to try to separate the words with comma or to
1158 put comments into a multi-line C<qw>-string. For this reason, the
1159 C<use warnings> pragma and the B<-w> switch (that is, the C<$^W> variable)
1160 produces warnings if the STRING contains the "," or the "#" character.
1162 =item s/PATTERN/REPLACEMENT/egimosx
1164 Searches a string for a pattern, and if found, replaces that pattern
1165 with the replacement text and returns the number of substitutions
1166 made. Otherwise it returns false (specifically, the empty string).
1168 If no string is specified via the C<=~> or C<!~> operator, the C<$_>
1169 variable is searched and modified. (The string specified with C<=~> must
1170 be scalar variable, an array element, a hash element, or an assignment
1171 to one of those, i.e., an lvalue.)
1173 If the delimiter chosen is a single quote, no interpolation is
1174 done on either the PATTERN or the REPLACEMENT. Otherwise, if the
1175 PATTERN contains a $ that looks like a variable rather than an
1176 end-of-string test, the variable will be interpolated into the pattern
1177 at run-time. If you want the pattern compiled only once the first time
1178 the variable is interpolated, use the C</o> option. If the pattern
1179 evaluates to the empty string, the last successfully executed regular
1180 expression is used instead. See L<perlre> for further explanation on these.
1181 See L<perllocale> for discussion of additional considerations that apply
1182 when C<use locale> is in effect.
1186 e Evaluate the right side as an expression.
1187 g Replace globally, i.e., all occurrences.
1188 i Do case-insensitive pattern matching.
1189 m Treat string as multiple lines.
1190 o Compile pattern only once.
1191 s Treat string as single line.
1192 x Use extended regular expressions.
1194 Any non-alphanumeric, non-whitespace delimiter may replace the
1195 slashes. If single quotes are used, no interpretation is done on the
1196 replacement string (the C</e> modifier overrides this, however). Unlike
1197 Perl 4, Perl 5 treats backticks as normal delimiters; the replacement
1198 text is not evaluated as a command. If the
1199 PATTERN is delimited by bracketing quotes, the REPLACEMENT has its own
1200 pair of quotes, which may or may not be bracketing quotes, e.g.,
1201 C<s(foo)(bar)> or C<< s<foo>/bar/ >>. A C</e> will cause the
1202 replacement portion to be treated as a full-fledged Perl expression
1203 and evaluated right then and there. It is, however, syntax checked at
1204 compile-time. A second C<e> modifier will cause the replacement portion
1205 to be C<eval>ed before being run as a Perl expression.
1209 s/\bgreen\b/mauve/g; # don't change wintergreen
1211 $path =~ s|/usr/bin|/usr/local/bin|;
1213 s/Login: $foo/Login: $bar/; # run-time pattern
1215 ($foo = $bar) =~ s/this/that/; # copy first, then change
1217 $count = ($paragraph =~ s/Mister\b/Mr./g); # get change-count
1220 s/\d+/$&*2/e; # yields 'abc246xyz'
1221 s/\d+/sprintf("%5d",$&)/e; # yields 'abc 246xyz'
1222 s/\w/$& x 2/eg; # yields 'aabbcc 224466xxyyzz'
1224 s/%(.)/$percent{$1}/g; # change percent escapes; no /e
1225 s/%(.)/$percent{$1} || $&/ge; # expr now, so /e
1226 s/^=(\w+)/&pod($1)/ge; # use function call
1228 # expand variables in $_, but dynamics only, using
1229 # symbolic dereferencing
1232 # Add one to the value of any numbers in the string
1235 # This will expand any embedded scalar variable
1236 # (including lexicals) in $_ : First $1 is interpolated
1237 # to the variable name, and then evaluated
1240 # Delete (most) C comments.
1242 /\* # Match the opening delimiter.
1243 .*? # Match a minimal number of characters.
1244 \*/ # Match the closing delimiter.
1247 s/^\s*(.*?)\s*$/$1/; # trim white space in $_, expensively
1249 for ($variable) { # trim white space in $variable, cheap
1254 s/([^ ]*) *([^ ]*)/$2 $1/; # reverse 1st two fields
1256 Note the use of $ instead of \ in the last example. Unlike
1257 B<sed>, we use the \<I<digit>> form in only the left hand side.
1258 Anywhere else it's $<I<digit>>.
1260 Occasionally, you can't use just a C</g> to get all the changes
1261 to occur that you might want. Here are two common cases:
1263 # put commas in the right places in an integer
1264 1 while s/(\d)(\d\d\d)(?!\d)/$1,$2/g;
1266 # expand tabs to 8-column spacing
1267 1 while s/\t+/' ' x (length($&)*8 - length($`)%8)/e;
1269 =item tr/SEARCHLIST/REPLACEMENTLIST/cds
1271 =item y/SEARCHLIST/REPLACEMENTLIST/cds
1273 Transliterates all occurrences of the characters found in the search list
1274 with the corresponding character in the replacement list. It returns
1275 the number of characters replaced or deleted. If no string is
1276 specified via the =~ or !~ operator, the $_ string is transliterated. (The
1277 string specified with =~ must be a scalar variable, an array element, a
1278 hash element, or an assignment to one of those, i.e., an lvalue.)
1280 A character range may be specified with a hyphen, so C<tr/A-J/0-9/>
1281 does the same replacement as C<tr/ACEGIBDFHJ/0246813579/>.
1282 For B<sed> devotees, C<y> is provided as a synonym for C<tr>. If the
1283 SEARCHLIST is delimited by bracketing quotes, the REPLACEMENTLIST has
1284 its own pair of quotes, which may or may not be bracketing quotes,
1285 e.g., C<tr[A-Z][a-z]> or C<tr(+\-*/)/ABCD/>.
1287 Note that C<tr> does B<not> do regular expression character classes
1288 such as C<\d> or C<[:lower:]>. The <tr> operator is not equivalent to
1289 the tr(1) utility. If you want to map strings between lower/upper
1290 cases, see L<perlfunc/lc> and L<perlfunc/uc>, and in general consider
1291 using the C<s> operator if you need regular expressions.
1293 Note also that the whole range idea is rather unportable between
1294 character sets--and even within character sets they may cause results
1295 you probably didn't expect. A sound principle is to use only ranges
1296 that begin from and end at either alphabets of equal case (a-e, A-E),
1297 or digits (0-4). Anything else is unsafe. If in doubt, spell out the
1298 character sets in full.
1302 c Complement the SEARCHLIST.
1303 d Delete found but unreplaced characters.
1304 s Squash duplicate replaced characters.
1306 If the C</c> modifier is specified, the SEARCHLIST character set
1307 is complemented. If the C</d> modifier is specified, any characters
1308 specified by SEARCHLIST not found in REPLACEMENTLIST are deleted.
1309 (Note that this is slightly more flexible than the behavior of some
1310 B<tr> programs, which delete anything they find in the SEARCHLIST,
1311 period.) If the C</s> modifier is specified, sequences of characters
1312 that were transliterated to the same character are squashed down
1313 to a single instance of the character.
1315 If the C</d> modifier is used, the REPLACEMENTLIST is always interpreted
1316 exactly as specified. Otherwise, if the REPLACEMENTLIST is shorter
1317 than the SEARCHLIST, the final character is replicated till it is long
1318 enough. If the REPLACEMENTLIST is empty, the SEARCHLIST is replicated.
1319 This latter is useful for counting characters in a class or for
1320 squashing character sequences in a class.
1324 $ARGV[1] =~ tr/A-Z/a-z/; # canonicalize to lower case
1326 $cnt = tr/*/*/; # count the stars in $_
1328 $cnt = $sky =~ tr/*/*/; # count the stars in $sky
1330 $cnt = tr/0-9//; # count the digits in $_
1332 tr/a-zA-Z//s; # bookkeeper -> bokeper
1334 ($HOST = $host) =~ tr/a-z/A-Z/;
1336 tr/a-zA-Z/ /cs; # change non-alphas to single space
1339 [\000-\177]; # delete 8th bit
1341 If multiple transliterations are given for a character, only the
1346 will transliterate any A to X.
1348 Because the transliteration table is built at compile time, neither
1349 the SEARCHLIST nor the REPLACEMENTLIST are subjected to double quote
1350 interpolation. That means that if you want to use variables, you
1353 eval "tr/$oldlist/$newlist/";
1356 eval "tr/$oldlist/$newlist/, 1" or die $@;
1360 A line-oriented form of quoting is based on the shell "here-document"
1361 syntax. Following a C<< << >> you specify a string to terminate
1362 the quoted material, and all lines following the current line down to
1363 the terminating string are the value of the item. The terminating
1364 string may be either an identifier (a word), or some quoted text. If
1365 quoted, the type of quotes you use determines the treatment of the
1366 text, just as in regular quoting. An unquoted identifier works like
1367 double quotes. There must be no space between the C<< << >> and
1368 the identifier, unless the identifier is quoted. (If you put a space it
1369 will be treated as a null identifier, which is valid, and matches the first
1370 empty line.) The terminating string must appear by itself (unquoted and
1371 with no surrounding whitespace) on the terminating line.
1374 The price is $Price.
1377 print << "EOF"; # same as above
1378 The price is $Price.
1381 print << `EOC`; # execute commands
1386 print <<"foo", <<"bar"; # you can stack them
1392 myfunc(<< "THIS", 23, <<'THAT');
1399 Just don't forget that you have to put a semicolon on the end
1400 to finish the statement, as Perl doesn't know you're not going to
1408 If you want your here-docs to be indented with the
1409 rest of the code, you'll need to remove leading whitespace
1410 from each line manually:
1412 ($quote = <<'FINIS') =~ s/^\s+//gm;
1413 The Road goes ever on and on,
1414 down from the door where it began.
1417 If you use a here-doc within a delimited construct, such as in C<s///eg>,
1418 the quoted material must come on the lines following the final delimiter.
1433 If the terminating identifier is on the last line of the program, you
1434 must be sure there is a newline after it; otherwise, Perl will give the
1435 warning B<Can't find string terminator "END" anywhere before EOF...>.
1437 Additionally, the quoting rules for the identifier are not related to
1438 Perl's quoting rules -- C<q()>, C<qq()>, and the like are not supported
1439 in place of C<''> and C<"">, and the only interpolation is for backslashing
1440 the quoting character:
1442 print << "abc\"def";
1446 Finally, quoted strings cannot span multiple lines. The general rule is
1447 that the identifier must be a string literal. Stick with that, and you
1452 =head2 Gory details of parsing quoted constructs
1454 When presented with something that might have several different
1455 interpretations, Perl uses the B<DWIM> (that's "Do What I Mean")
1456 principle to pick the most probable interpretation. This strategy
1457 is so successful that Perl programmers often do not suspect the
1458 ambivalence of what they write. But from time to time, Perl's
1459 notions differ substantially from what the author honestly meant.
1461 This section hopes to clarify how Perl handles quoted constructs.
1462 Although the most common reason to learn this is to unravel labyrinthine
1463 regular expressions, because the initial steps of parsing are the
1464 same for all quoting operators, they are all discussed together.
1466 The most important Perl parsing rule is the first one discussed
1467 below: when processing a quoted construct, Perl first finds the end
1468 of that construct, then interprets its contents. If you understand
1469 this rule, you may skip the rest of this section on the first
1470 reading. The other rules are likely to contradict the user's
1471 expectations much less frequently than this first one.
1473 Some passes discussed below are performed concurrently, but because
1474 their results are the same, we consider them individually. For different
1475 quoting constructs, Perl performs different numbers of passes, from
1476 one to five, but these passes are always performed in the same order.
1480 =item Finding the end
1482 The first pass is finding the end of the quoted construct, whether
1483 it be a multicharacter delimiter C<"\nEOF\n"> in the C<<<EOF>
1484 construct, a C</> that terminates a C<qq//> construct, a C<]> which
1485 terminates C<qq[]> construct, or a C<< > >> which terminates a
1486 fileglob started with C<< < >>.
1488 When searching for single-character non-pairing delimiters, such
1489 as C</>, combinations of C<\\> and C<\/> are skipped. However,
1490 when searching for single-character pairing delimiter like C<[>,
1491 combinations of C<\\>, C<\]>, and C<\[> are all skipped, and nested
1492 C<[>, C<]> are skipped as well. When searching for multicharacter
1493 delimiters, nothing is skipped.
1495 For constructs with three-part delimiters (C<s///>, C<y///>, and
1496 C<tr///>), the search is repeated once more.
1498 During this search no attention is paid to the semantics of the construct.
1501 "$hash{"$foo/$bar"}"
1506 bar # NOT a comment, this slash / terminated m//!
1509 do not form legal quoted expressions. The quoted part ends on the
1510 first C<"> and C</>, and the rest happens to be a syntax error.
1511 Because the slash that terminated C<m//> was followed by a C<SPACE>,
1512 the example above is not C<m//x>, but rather C<m//> with no C</x>
1513 modifier. So the embedded C<#> is interpreted as a literal C<#>.
1515 =item Removal of backslashes before delimiters
1517 During the second pass, text between the starting and ending
1518 delimiters is copied to a safe location, and the C<\> is removed
1519 from combinations consisting of C<\> and delimiter--or delimiters,
1520 meaning both starting and ending delimiters will should these differ.
1521 This removal does not happen for multi-character delimiters.
1522 Note that the combination C<\\> is left intact, just as it was.
1524 Starting from this step no information about the delimiters is
1529 The next step is interpolation in the text obtained, which is now
1530 delimiter-independent. There are four different cases.
1534 =item C<<<'EOF'>, C<m''>, C<s'''>, C<tr///>, C<y///>
1536 No interpolation is performed.
1540 The only interpolation is removal of C<\> from pairs C<\\>.
1542 =item C<"">, C<``>, C<qq//>, C<qx//>, C<< <file*glob> >>
1544 C<\Q>, C<\U>, C<\u>, C<\L>, C<\l> (possibly paired with C<\E>) are
1545 converted to corresponding Perl constructs. Thus, C<"$foo\Qbaz$bar">
1546 is converted to C<$foo . (quotemeta("baz" . $bar))> internally.
1547 The other combinations are replaced with appropriate expansions.
1549 Let it be stressed that I<whatever falls between C<\Q> and C<\E>>
1550 is interpolated in the usual way. Something like C<"\Q\\E"> has
1551 no C<\E> inside. instead, it has C<\Q>, C<\\>, and C<E>, so the
1552 result is the same as for C<"\\\\E">. As a general rule, backslashes
1553 between C<\Q> and C<\E> may lead to counterintuitive results. So,
1554 C<"\Q\t\E"> is converted to C<quotemeta("\t")>, which is the same
1555 as C<"\\\t"> (since TAB is not alphanumeric). Note also that:
1560 may be closer to the conjectural I<intention> of the writer of C<"\Q\t\E">.
1562 Interpolated scalars and arrays are converted internally to the C<join> and
1563 C<.> catenation operations. Thus, C<"$foo XXX '@arr'"> becomes:
1565 $foo . " XXX '" . (join $", @arr) . "'";
1567 All operations above are performed simultaneously, left to right.
1569 Because the result of C<"\Q STRING \E"> has all metacharacters
1570 quoted, there is no way to insert a literal C<$> or C<@> inside a
1571 C<\Q\E> pair. If protected by C<\>, C<$> will be quoted to became
1572 C<"\\\$">; if not, it is interpreted as the start of an interpolated
1575 Note also that the interpolation code needs to make a decision on
1576 where the interpolated scalar ends. For instance, whether
1577 C<< "a $b -> {c}" >> really means:
1579 "a " . $b . " -> {c}";
1585 Most of the time, the longest possible text that does not include
1586 spaces between components and which contains matching braces or
1587 brackets. because the outcome may be determined by voting based
1588 on heuristic estimators, the result is not strictly predictable.
1589 Fortunately, it's usually correct for ambiguous cases.
1591 =item C<?RE?>, C</RE/>, C<m/RE/>, C<s/RE/foo/>,
1593 Processing of C<\Q>, C<\U>, C<\u>, C<\L>, C<\l>, and interpolation
1594 happens (almost) as with C<qq//> constructs, but the substitution
1595 of C<\> followed by RE-special chars (including C<\>) is not
1596 performed. Moreover, inside C<(?{BLOCK})>, C<(?# comment )>, and
1597 a C<#>-comment in a C<//x>-regular expression, no processing is
1598 performed whatsoever. This is the first step at which the presence
1599 of the C<//x> modifier is relevant.
1601 Interpolation has several quirks: C<$|>, C<$(>, and C<$)> are not
1602 interpolated, and constructs C<$var[SOMETHING]> are voted (by several
1603 different estimators) to be either an array element or C<$var>
1604 followed by an RE alternative. This is where the notation
1605 C<${arr[$bar]}> comes handy: C</${arr[0-9]}/> is interpreted as
1606 array element C<-9>, not as a regular expression from the variable
1607 C<$arr> followed by a digit, which would be the interpretation of
1608 C</$arr[0-9]/>. Since voting among different estimators may occur,
1609 the result is not predictable.
1611 It is at this step that C<\1> is begrudgingly converted to C<$1> in
1612 the replacement text of C<s///> to correct the incorrigible
1613 I<sed> hackers who haven't picked up the saner idiom yet. A warning
1614 is emitted if the C<use warnings> pragma or the B<-w> command-line flag
1615 (that is, the C<$^W> variable) was set.
1617 The lack of processing of C<\\> creates specific restrictions on
1618 the post-processed text. If the delimiter is C</>, one cannot get
1619 the combination C<\/> into the result of this step. C</> will
1620 finish the regular expression, C<\/> will be stripped to C</> on
1621 the previous step, and C<\\/> will be left as is. Because C</> is
1622 equivalent to C<\/> inside a regular expression, this does not
1623 matter unless the delimiter happens to be character special to the
1624 RE engine, such as in C<s*foo*bar*>, C<m[foo]>, or C<?foo?>; or an
1625 alphanumeric char, as in:
1629 In the RE above, which is intentionally obfuscated for illustration, the
1630 delimiter is C<m>, the modifier is C<mx>, and after backslash-removal the
1631 RE is the same as for C<m/ ^ a s* b /mx>). There's more than one
1632 reason you're encouraged to restrict your delimiters to non-alphanumeric,
1633 non-whitespace choices.
1637 This step is the last one for all constructs except regular expressions,
1638 which are processed further.
1640 =item Interpolation of regular expressions
1642 Previous steps were performed during the compilation of Perl code,
1643 but this one happens at run time--although it may be optimized to
1644 be calculated at compile time if appropriate. After preprocessing
1645 described above, and possibly after evaluation if catenation,
1646 joining, casing translation, or metaquoting are involved, the
1647 resulting I<string> is passed to the RE engine for compilation.
1649 Whatever happens in the RE engine might be better discussed in L<perlre>,
1650 but for the sake of continuity, we shall do so here.
1652 This is another step where the presence of the C<//x> modifier is
1653 relevant. The RE engine scans the string from left to right and
1654 converts it to a finite automaton.
1656 Backslashed characters are either replaced with corresponding
1657 literal strings (as with C<\{>), or else they generate special nodes
1658 in the finite automaton (as with C<\b>). Characters special to the
1659 RE engine (such as C<|>) generate corresponding nodes or groups of
1660 nodes. C<(?#...)> comments are ignored. All the rest is either
1661 converted to literal strings to match, or else is ignored (as is
1662 whitespace and C<#>-style comments if C<//x> is present).
1664 Parsing of the bracketed character class construct, C<[...]>, is
1665 rather different than the rule used for the rest of the pattern.
1666 The terminator of this construct is found using the same rules as
1667 for finding the terminator of a C<{}>-delimited construct, the only
1668 exception being that C<]> immediately following C<[> is treated as
1669 though preceded by a backslash. Similarly, the terminator of
1670 C<(?{...})> is found using the same rules as for finding the
1671 terminator of a C<{}>-delimited construct.
1673 It is possible to inspect both the string given to RE engine and the
1674 resulting finite automaton. See the arguments C<debug>/C<debugcolor>
1675 in the C<use L<re>> pragma, as well as Perl's B<-Dr> command-line
1676 switch documented in L<perlrun/"Command Switches">.
1678 =item Optimization of regular expressions
1680 This step is listed for completeness only. Since it does not change
1681 semantics, details of this step are not documented and are subject
1682 to change without notice. This step is performed over the finite
1683 automaton that was generated during the previous pass.
1685 It is at this stage that C<split()> silently optimizes C</^/> to
1690 =head2 I/O Operators
1692 There are several I/O operators you should know about.
1694 A string enclosed by backticks (grave accents) first undergoes
1695 double-quote interpolation. It is then interpreted as an external
1696 command, and the output of that command is the value of the
1697 backtick string, like in a shell. In scalar context, a single string
1698 consisting of all output is returned. In list context, a list of
1699 values is returned, one per line of output. (You can set C<$/> to use
1700 a different line terminator.) The command is executed each time the
1701 pseudo-literal is evaluated. The status value of the command is
1702 returned in C<$?> (see L<perlvar> for the interpretation of C<$?>).
1703 Unlike in B<csh>, no translation is done on the return data--newlines
1704 remain newlines. Unlike in any of the shells, single quotes do not
1705 hide variable names in the command from interpretation. To pass a
1706 literal dollar-sign through to the shell you need to hide it with a
1707 backslash. The generalized form of backticks is C<qx//>. (Because
1708 backticks always undergo shell expansion as well, see L<perlsec> for
1711 In scalar context, evaluating a filehandle in angle brackets yields
1712 the next line from that file (the newline, if any, included), or
1713 C<undef> at end-of-file or on error. When C<$/> is set to C<undef>
1714 (sometimes known as file-slurp mode) and the file is empty, it
1715 returns C<''> the first time, followed by C<undef> subsequently.
1717 Ordinarily you must assign the returned value to a variable, but
1718 there is one situation where an automatic assignment happens. If
1719 and only if the input symbol is the only thing inside the conditional
1720 of a C<while> statement (even if disguised as a C<for(;;)> loop),
1721 the value is automatically assigned to the global variable $_,
1722 destroying whatever was there previously. (This may seem like an
1723 odd thing to you, but you'll use the construct in almost every Perl
1724 script you write.) The $_ variable is not implicitly localized.
1725 You'll have to put a C<local $_;> before the loop if you want that
1728 The following lines are equivalent:
1730 while (defined($_ = <STDIN>)) { print; }
1731 while ($_ = <STDIN>) { print; }
1732 while (<STDIN>) { print; }
1733 for (;<STDIN>;) { print; }
1734 print while defined($_ = <STDIN>);
1735 print while ($_ = <STDIN>);
1736 print while <STDIN>;
1738 This also behaves similarly, but avoids $_ :
1740 while (my $line = <STDIN>) { print $line }
1742 In these loop constructs, the assigned value (whether assignment
1743 is automatic or explicit) is then tested to see whether it is
1744 defined. The defined test avoids problems where line has a string
1745 value that would be treated as false by Perl, for example a "" or
1746 a "0" with no trailing newline. If you really mean for such values
1747 to terminate the loop, they should be tested for explicitly:
1749 while (($_ = <STDIN>) ne '0') { ... }
1750 while (<STDIN>) { last unless $_; ... }
1752 In other boolean contexts, C<< <I<filehandle>> >> without an
1753 explicit C<defined> test or comparison elicit a warning if the
1754 C<use warnings> pragma or the B<-w>
1755 command-line switch (the C<$^W> variable) is in effect.
1757 The filehandles STDIN, STDOUT, and STDERR are predefined. (The
1758 filehandles C<stdin>, C<stdout>, and C<stderr> will also work except
1759 in packages, where they would be interpreted as local identifiers
1760 rather than global.) Additional filehandles may be created with
1761 the open() function, amongst others. See L<perlopentut> and
1762 L<perlfunc/open> for details on this.
1764 If a <FILEHANDLE> is used in a context that is looking for
1765 a list, a list comprising all input lines is returned, one line per
1766 list element. It's easy to grow to a rather large data space this
1767 way, so use with care.
1769 <FILEHANDLE> may also be spelled C<readline(*FILEHANDLE)>.
1770 See L<perlfunc/readline>.
1772 The null filehandle <> is special: it can be used to emulate the
1773 behavior of B<sed> and B<awk>. Input from <> comes either from
1774 standard input, or from each file listed on the command line. Here's
1775 how it works: the first time <> is evaluated, the @ARGV array is
1776 checked, and if it is empty, C<$ARGV[0]> is set to "-", which when opened
1777 gives you standard input. The @ARGV array is then processed as a list
1778 of filenames. The loop
1781 ... # code for each line
1784 is equivalent to the following Perl-like pseudo code:
1786 unshift(@ARGV, '-') unless @ARGV;
1787 while ($ARGV = shift) {
1790 ... # code for each line
1794 except that it isn't so cumbersome to say, and will actually work.
1795 It really does shift the @ARGV array and put the current filename
1796 into the $ARGV variable. It also uses filehandle I<ARGV>
1797 internally--<> is just a synonym for <ARGV>, which
1798 is magical. (The pseudo code above doesn't work because it treats
1799 <ARGV> as non-magical.)
1801 You can modify @ARGV before the first <> as long as the array ends up
1802 containing the list of filenames you really want. Line numbers (C<$.>)
1803 continue as though the input were one big happy file. See the example
1804 in L<perlfunc/eof> for how to reset line numbers on each file.
1806 If you want to set @ARGV to your own list of files, go right ahead.
1807 This sets @ARGV to all plain text files if no @ARGV was given:
1809 @ARGV = grep { -f && -T } glob('*') unless @ARGV;
1811 You can even set them to pipe commands. For example, this automatically
1812 filters compressed arguments through B<gzip>:
1814 @ARGV = map { /\.(gz|Z)$/ ? "gzip -dc < $_ |" : $_ } @ARGV;
1816 If you want to pass switches into your script, you can use one of the
1817 Getopts modules or put a loop on the front like this:
1819 while ($_ = $ARGV[0], /^-/) {
1822 if (/^-D(.*)/) { $debug = $1 }
1823 if (/^-v/) { $verbose++ }
1824 # ... # other switches
1828 # ... # code for each line
1831 The <> symbol will return C<undef> for end-of-file only once.
1832 If you call it again after this, it will assume you are processing another
1833 @ARGV list, and if you haven't set @ARGV, will read input from STDIN.
1835 If what the angle brackets contain is a simple scalar variable (e.g.,
1836 <$foo>), then that variable contains the name of the
1837 filehandle to input from, or its typeglob, or a reference to the
1843 If what's within the angle brackets is neither a filehandle nor a simple
1844 scalar variable containing a filehandle name, typeglob, or typeglob
1845 reference, it is interpreted as a filename pattern to be globbed, and
1846 either a list of filenames or the next filename in the list is returned,
1847 depending on context. This distinction is determined on syntactic
1848 grounds alone. That means C<< <$x> >> is always a readline() from
1849 an indirect handle, but C<< <$hash{key}> >> is always a glob().
1850 That's because $x is a simple scalar variable, but C<$hash{key}> is
1851 not--it's a hash element.
1853 One level of double-quote interpretation is done first, but you can't
1854 say C<< <$foo> >> because that's an indirect filehandle as explained
1855 in the previous paragraph. (In older versions of Perl, programmers
1856 would insert curly brackets to force interpretation as a filename glob:
1857 C<< <${foo}> >>. These days, it's considered cleaner to call the
1858 internal function directly as C<glob($foo)>, which is probably the right
1859 way to have done it in the first place.) For example:
1865 is roughly equivalent to:
1867 open(FOO, "echo *.c | tr -s ' \t\r\f' '\\012\\012\\012\\012'|");
1873 except that the globbing is actually done internally using the standard
1874 C<File::Glob> extension. Of course, the shortest way to do the above is:
1878 A (file)glob evaluates its (embedded) argument only when it is
1879 starting a new list. All values must be read before it will start
1880 over. In list context, this isn't important because you automatically
1881 get them all anyway. However, in scalar context the operator returns
1882 the next value each time it's called, or C<undef> when the list has
1883 run out. As with filehandle reads, an automatic C<defined> is
1884 generated when the glob occurs in the test part of a C<while>,
1885 because legal glob returns (e.g. a file called F<0>) would otherwise
1886 terminate the loop. Again, C<undef> is returned only once. So if
1887 you're expecting a single value from a glob, it is much better to
1890 ($file) = <blurch*>;
1896 because the latter will alternate between returning a filename and
1899 If you're trying to do variable interpolation, it's definitely better
1900 to use the glob() function, because the older notation can cause people
1901 to become confused with the indirect filehandle notation.
1903 @files = glob("$dir/*.[ch]");
1904 @files = glob($files[$i]);
1906 =head2 Constant Folding
1908 Like C, Perl does a certain amount of expression evaluation at
1909 compile time whenever it determines that all arguments to an
1910 operator are static and have no side effects. In particular, string
1911 concatenation happens at compile time between literals that don't do
1912 variable substitution. Backslash interpolation also happens at
1913 compile time. You can say
1915 'Now is the time for all' . "\n" .
1916 'good men to come to.'
1918 and this all reduces to one string internally. Likewise, if
1921 foreach $file (@filenames) {
1922 if (-s $file > 5 + 100 * 2**16) { }
1925 the compiler will precompute the number which that expression
1926 represents so that the interpreter won't have to.
1928 =head2 Bitwise String Operators
1930 Bitstrings of any size may be manipulated by the bitwise operators
1933 If the operands to a binary bitwise op are strings of different
1934 sizes, B<|> and B<^> ops act as though the shorter operand had
1935 additional zero bits on the right, while the B<&> op acts as though
1936 the longer operand were truncated to the length of the shorter.
1937 The granularity for such extension or truncation is one or more
1940 # ASCII-based examples
1941 print "j p \n" ^ " a h"; # prints "JAPH\n"
1942 print "JA" | " ph\n"; # prints "japh\n"
1943 print "japh\nJunk" & '_____'; # prints "JAPH\n";
1944 print 'p N$' ^ " E<H\n"; # prints "Perl\n";
1946 If you are intending to manipulate bitstrings, be certain that
1947 you're supplying bitstrings: If an operand is a number, that will imply
1948 a B<numeric> bitwise operation. You may explicitly show which type of
1949 operation you intend by using C<""> or C<0+>, as in the examples below.
1951 $foo = 150 | 105 ; # yields 255 (0x96 | 0x69 is 0xFF)
1952 $foo = '150' | 105 ; # yields 255
1953 $foo = 150 | '105'; # yields 255
1954 $foo = '150' | '105'; # yields string '155' (under ASCII)
1956 $baz = 0+$foo & 0+$bar; # both ops explicitly numeric
1957 $biz = "$foo" ^ "$bar"; # both ops explicitly stringy
1959 See L<perlfunc/vec> for information on how to manipulate individual bits
1962 =head2 Integer Arithmetic
1964 By default, Perl assumes that it must do most of its arithmetic in
1965 floating point. But by saying
1969 you may tell the compiler that it's okay to use integer operations
1970 (if it feels like it) from here to the end of the enclosing BLOCK.
1971 An inner BLOCK may countermand this by saying
1975 which lasts until the end of that BLOCK. Note that this doesn't
1976 mean everything is only an integer, merely that Perl may use integer
1977 operations if it is so inclined. For example, even under C<use
1978 integer>, if you take the C<sqrt(2)>, you'll still get C<1.4142135623731>
1981 Used on numbers, the bitwise operators ("&", "|", "^", "~", "<<",
1982 and ">>") always produce integral results. (But see also
1983 L<Bitwise String Operators>.) However, C<use integer> still has meaning for
1984 them. By default, their results are interpreted as unsigned integers, but
1985 if C<use integer> is in effect, their results are interpreted
1986 as signed integers. For example, C<~0> usually evaluates to a large
1987 integral value. However, C<use integer; ~0> is C<-1> on twos-complement
1990 =head2 Floating-point Arithmetic
1992 While C<use integer> provides integer-only arithmetic, there is no
1993 analogous mechanism to provide automatic rounding or truncation to a
1994 certain number of decimal places. For rounding to a certain number
1995 of digits, sprintf() or printf() is usually the easiest route.
1998 Floating-point numbers are only approximations to what a mathematician
1999 would call real numbers. There are infinitely more reals than floats,
2000 so some corners must be cut. For example:
2002 printf "%.20g\n", 123456789123456789;
2003 # produces 123456789123456784
2005 Testing for exact equality of floating-point equality or inequality is
2006 not a good idea. Here's a (relatively expensive) work-around to compare
2007 whether two floating-point numbers are equal to a particular number of
2008 decimal places. See Knuth, volume II, for a more robust treatment of
2012 my ($X, $Y, $POINTS) = @_;
2014 $tX = sprintf("%.${POINTS}g", $X);
2015 $tY = sprintf("%.${POINTS}g", $Y);
2019 The POSIX module (part of the standard perl distribution) implements
2020 ceil(), floor(), and other mathematical and trigonometric functions.
2021 The Math::Complex module (part of the standard perl distribution)
2022 defines mathematical functions that work on both the reals and the
2023 imaginary numbers. Math::Complex not as efficient as POSIX, but
2024 POSIX can't work with complex numbers.
2026 Rounding in financial applications can have serious implications, and
2027 the rounding method used should be specified precisely. In these
2028 cases, it probably pays not to trust whichever system rounding is
2029 being used by Perl, but to instead implement the rounding function you
2032 =head2 Bigger Numbers
2034 The standard Math::BigInt and Math::BigFloat modules provide
2035 variable-precision arithmetic and overloaded operators, although
2036 they're currently pretty slow. At the cost of some space and
2037 considerable speed, they avoid the normal pitfalls associated with
2038 limited-precision representations.
2041 $x = Math::BigInt->new('123456789123456789');
2044 # prints +15241578780673678515622620750190521
2046 There are several modules that let you calculate with (bound only by
2047 memory and cpu-time) unlimited or fixed precision. There are also
2048 some non-standard modules that provide faster implementations via
2049 external C libraries.
2051 Here is a short, but incomplete summary:
2053 Math::Fraction big, unlimited fractions like 9973 / 12967
2054 Math::String treat string sequences like numbers
2055 Math::FixedPrecision calculate with a fixed precision
2056 Math::Currency for currency calculations
2057 Bit::Vector manipulate bit vectors fast (uses C)
2058 Math::BigIntFast Bit::Vector wrapper for big numbers
2059 Math::Pari provides access to the Pari C library
2060 Math::BigInteger uses an external C library
2061 Math::Cephes uses external Cephes C library (no big numbers)
2062 Math::Cephes::Fraction fractions via the Cephes library
2063 Math::GMP another one using an external C library