1PERLOP(1) Perl Programmers Reference Guide PERLOP(1)
2
6 perlop - Perl operators and precedence
7
9 Operator Precedence and Associativity
10 Operator precedence and associativity work in Perl more or less like
11 they do in mathematics.
12 Operator precedence means some operators are evaluated before others.
14 For example, in "2 + 4 * 5", the multiplication has higher precedence
15 so "4 * 5" is evaluated first yielding "2 + 20 == 22" and not "6 * 5 ==
16 30".
17 Operator associativity defines what happens if a sequence of the same
19 operators is used one after another: whether the evaluator will
20 evaluate the left operations first or the right. For example, in "8 -
21 4 - 2", subtraction is left associative so Perl evaluates the
22 expression left to right. "8 - 4" is evaluated first making the
23 expression "4 - 2 == 2" and not "8 - 2 == 6".
24 Perl operators have the following associativity and precedence, listed
26 from highest precedence to lowest. Operators borrowed from C keep the
27 same precedence relationship with each other, even where C's precedence
28 is slightly screwy. (This makes learning Perl easier for C folks.)
29 With very few exceptions, these all operate on scalar values only, not
30 array values.
31 left terms and list operators (leftward)
33 left ->
34 nonassoc ++ --
35 right **
36 right ! ~ \ and unary + and -
37 left =~ !~
38 left * / % x
39 left + - .
40 left << >>
41 nonassoc named unary operators
42 nonassoc < > <= >= lt gt le ge
43 nonassoc == != <=> eq ne cmp ~~
44 left &
45 left | ^
46 left &&
47 left || //
48 nonassoc .. ...
49 right ?:
50 right = += -= *= etc.
51 left , =>
52 nonassoc list operators (rightward)
53 right not
54 left and
55 left or xor
56 In the following sections, these operators are covered in precedence
58 order.
59 Many operators can be overloaded for objects. See overload.
61 Terms and List Operators (Leftward)
63 A TERM has the highest precedence in Perl. They include variables,
64 quote and quote-like operators, any expression in parentheses, and any
65 function whose arguments are parenthesized. Actually, there aren't
66 really functions in this sense, just list operators and unary operators
67 behaving as functions because you put parentheses around the arguments.
68 These are all documented in perlfunc.
69 If any list operator (print(), etc.) or any unary operator (chdir(),
71 etc.) is followed by a left parenthesis as the next token, the
72 operator and arguments within parentheses are taken to be of highest
73 precedence, just like a normal function call.
74 In the absence of parentheses, the precedence of list operators such as
76 "print", "sort", or "chmod" is either very high or very low depending
77 on whether you are looking at the left side or the right side of the
78 operator. For example, in
79 @ary = (1, 3, sort 4, 2);
81 print @ary; # prints 1324
82 the commas on the right of the sort are evaluated before the sort, but
84 the commas on the left are evaluated after. In other words, list
85 operators tend to gobble up all arguments that follow, and then act
86 like a simple TERM with regard to the preceding expression. Be careful
87 with parentheses:
88 # These evaluate exit before doing the print:
90 print($foo, exit); # Obviously not what you want.
91 print $foo, exit; # Nor is this.
92 # These do the print before evaluating exit:
94 (print $foo), exit; # This is what you want.
95 print($foo), exit; # Or this.
96 print ($foo), exit; # Or even this.
97 Also note that
99 print ($foo & 255) + 1, "\n";
101 probably doesn't do what you expect at first glance. The parentheses
103 enclose the argument list for "print" which is evaluated (printing the
104 result of "$foo & 255"). Then one is added to the return value of
105 "print" (usually 1). The result is something like this:
106 1 + 1, "\n"; # Obviously not what you meant.
108 To do what you meant properly, you must write:
110 print(($foo & 255) + 1, "\n");
112 See "Named Unary Operators" for more discussion of this.
114 Also parsed as terms are the "do {}" and "eval {}" constructs, as well
116 as subroutine and method calls, and the anonymous constructors "[]" and
117 "{}".
118 See also "Quote and Quote-like Operators" toward the end of this
120 section, as well as "I/O Operators".
121 The Arrow Operator
123 ""->"" is an infix dereference operator, just as it is in C and C++.
124 If the right side is either a "[...]", "{...}", or a "(...)" subscript,
125 then the left side must be either a hard or symbolic reference to an
126 array, a hash, or a subroutine respectively. (Or technically speaking,
127 a location capable of holding a hard reference, if it's an array or
128 hash reference being used for assignment.) See perlreftut and perlref.
129 Otherwise, the right side is a method name or a simple scalar variable
131 containing either the method name or a subroutine reference, and the
132 left side must be either an object (a blessed reference) or a class
133 name (that is, a package name). See perlobj.
134 Auto-increment and Auto-decrement
136 "++" and "--" work as in C. That is, if placed before a variable, they
137 increment or decrement the variable by one before returning the value,
138 and if placed after, increment or decrement after returning the value.
139 $i = 0; $j = 0;
141 print $i++; # prints 0
142 print ++$j; # prints 1
143 Note that just as in C, Perl doesn't define when the variable is
145 incremented or decremented. You just know it will be done sometime
146 before or after the value is returned. This also means that modifying a
147 variable twice in the same statement will lead to undefined behaviour.
148 Avoid statements like:
149 $i = $i ++;
151 print ++ $i + $i ++;
152 Perl will not guarantee what the result of the above statements is.
154 The auto-increment operator has a little extra builtin magic to it. If
156 you increment a variable that is numeric, or that has ever been used in
157 a numeric context, you get a normal increment. If, however, the
158 variable has been used in only string contexts since it was set, and
159 has a value that is not the empty string and matches the pattern
160 "/^[a-zA-Z]*[0-9]*\z/", the increment is done as a string, preserving
161 each character within its range, with carry:
162 print ++($foo = '99'); # prints '100'
164 print ++($foo = 'a0'); # prints 'a1'
165 print ++($foo = 'Az'); # prints 'Ba'
166 print ++($foo = 'zz'); # prints 'aaa'
167 "undef" is always treated as numeric, and in particular is changed to 0
169 before incrementing (so that a post-increment of an undef value will
170 return 0 rather than "undef").
171 The auto-decrement operator is not magical.
173 Exponentiation
175 Binary "**" is the exponentiation operator. It binds even more tightly
176 than unary minus, so -2**4 is -(2**4), not (-2)**4. (This is
177 implemented using C's pow(3) function, which actually works on doubles
178 internally.)
179 Symbolic Unary Operators
181 Unary "!" performs logical negation, i.e., "not". See also "not" for a
182 lower precedence version of this.
183 Unary "-" performs arithmetic negation if the operand is numeric. If
185 the operand is an identifier, a string consisting of a minus sign
186 concatenated with the identifier is returned. Otherwise, if the string
187 starts with a plus or minus, a string starting with the opposite sign
188 is returned. One effect of these rules is that -bareword is equivalent
189 to the string "-bareword". If, however, the string begins with a non-
190 alphabetic character (excluding "+" or "-"), Perl will attempt to
191 convert the string to a numeric and the arithmetic negation is
192 performed. If the string cannot be cleanly converted to a numeric, Perl
193 will give the warning Argument "the string" isn't numeric in negation
194 (-) at ....
195 Unary "~" performs bitwise negation, i.e., 1's complement. For
197 example, "0666 & ~027" is 0640. (See also "Integer Arithmetic" and
198 "Bitwise String Operators".) Note that the width of the result is
199 platform-dependent: ~0 is 32 bits wide on a 32-bit platform, but 64
200 bits wide on a 64-bit platform, so if you are expecting a certain bit
201 width, remember to use the & operator to mask off the excess bits.
202 Unary "+" has no effect whatsoever, even on strings. It is useful
204 syntactically for separating a function name from a parenthesized
205 expression that would otherwise be interpreted as the complete list of
206 function arguments. (See examples above under "Terms and List
207 Operators (Leftward)".)
208 Unary "\" creates a reference to whatever follows it. See perlreftut
210 and perlref. Do not confuse this behavior with the behavior of
211 backslash within a string, although both forms do convey the notion of
212 protecting the next thing from interpolation.
213 Binding Operators
215 Binary "=~" binds a scalar expression to a pattern match. Certain
216 operations search or modify the string $_ by default. This operator
217 makes that kind of operation work on some other string. The right
218 argument is a search pattern, substitution, or transliteration. The
219 left argument is what is supposed to be searched, substituted, or
220 transliterated instead of the default $_. When used in scalar context,
221 the return value generally indicates the success of the operation.
222 Behavior in list context depends on the particular operator. See
223 "Regexp Quote-Like Operators" for details and perlretut for examples
224 using these operators.
225 If the right argument is an expression rather than a search pattern,
227 substitution, or transliteration, it is interpreted as a search pattern
228 at run time. Note that this means that its contents will be
229 interpolated twice, so
230 '\\' =~ q'\\';
232 is not ok, as the regex engine will end up trying to compile the
234 pattern "\", which it will consider a syntax error.
235 Binary "!~" is just like "=~" except the return value is negated in the
237 logical sense.
238 Multiplicative Operators
240 Binary "*" multiplies two numbers.
241 Binary "/" divides two numbers.
243 Binary "%" is the modulo operator, which computes the division
245 remainder of its first argument with respect to its second argument.
246 Given integer operands $a and $b: If $b is positive, then "$a % $b" is
247 $a minus the largest multiple of $b less than or equal to $a. If $b is
248 negative, then "$a % $b" is $a minus the smallest multiple of $b that
249 is not less than $a (i.e. the result will be less than or equal to
250 zero). If the operands $a and $b are floating point values and the
251 absolute value of $b (that is "abs($b)") is less than "(UV_MAX + 1)",
252 only the integer portion of $a and $b will be used in the operation
253 (Note: here "UV_MAX" means the maximum of the unsigned integer type).
254 If the absolute value of the right operand ("abs($b)") is greater than
255 or equal to "(UV_MAX + 1)", "%" computes the floating-point remainder
256 $r in the equation "($r = $a - $i*$b)" where $i is a certain integer
257 that makes $r have the same sign as the right operand $b (not as the
258 left operand $a like C function "fmod()") and the absolute value less
259 than that of $b. Note that when "use integer" is in scope, "%" gives
260 you direct access to the modulo operator as implemented by your C
261 compiler. This operator is not as well defined for negative operands,
262 but it will execute faster.
263 Binary "x" is the repetition operator. In scalar context or if the
265 left operand is not enclosed in parentheses, it returns a string
266 consisting of the left operand repeated the number of times specified
267 by the right operand. In list context, if the left operand is enclosed
268 in parentheses or is a list formed by "qw/STRING/", it repeats the
269 list. If the right operand is zero or negative, it returns an empty
270 string or an empty list, depending on the context.
271 print '-' x 80; # print row of dashes
273 print "\t" x ($tab/8), ' ' x ($tab%8); # tab over
275 @ones = (1) x 80; # a list of 80 1's
277 @ones = (5) x @ones; # set all elements to 5
278 Additive Operators
280 Binary "+" returns the sum of two numbers.
281 Binary "-" returns the difference of two numbers.
283 Binary "." concatenates two strings.
285 Shift Operators
287 Binary "<<" returns the value of its left argument shifted left by the
288 number of bits specified by the right argument. Arguments should be
289 integers. (See also "Integer Arithmetic".)
290 Binary ">>" returns the value of its left argument shifted right by the
292 number of bits specified by the right argument. Arguments should be
293 integers. (See also "Integer Arithmetic".)
294 Note that both "<<" and ">>" in Perl are implemented directly using
296 "<<" and ">>" in C. If "use integer" (see "Integer Arithmetic") is in
297 force then signed C integers are used, else unsigned C integers are
298 used. Either way, the implementation isn't going to generate results
299 larger than the size of the integer type Perl was built with (32 bits
300 or 64 bits).
301 The result of overflowing the range of the integers is undefined
303 because it is undefined also in C. In other words, using 32-bit
304 integers, "1 << 32" is undefined. Shifting by a negative number of
305 bits is also undefined.
306 Named Unary Operators
308 The various named unary operators are treated as functions with one
309 argument, with optional parentheses.
310 If any list operator (print(), etc.) or any unary operator (chdir(),
312 etc.) is followed by a left parenthesis as the next token, the
313 operator and arguments within parentheses are taken to be of highest
314 precedence, just like a normal function call. For example, because
315 named unary operators are higher precedence than ||:
316 chdir $foo || die; # (chdir $foo) || die
318 chdir($foo) || die; # (chdir $foo) || die
319 chdir ($foo) || die; # (chdir $foo) || die
320 chdir +($foo) || die; # (chdir $foo) || die
321 but, because * is higher precedence than named operators:
323 chdir $foo * 20; # chdir ($foo * 20)
325 chdir($foo) * 20; # (chdir $foo) * 20
326 chdir ($foo) * 20; # (chdir $foo) * 20
327 chdir +($foo) * 20; # chdir ($foo * 20)
328 rand 10 * 20; # rand (10 * 20)
330 rand(10) * 20; # (rand 10) * 20
331 rand (10) * 20; # (rand 10) * 20
332 rand +(10) * 20; # rand (10 * 20)
333 Regarding precedence, the filetest operators, like "-f", "-M", etc. are
335 treated like named unary operators, but they don't follow this
336 functional parenthesis rule. That means, for example, that
337 "-f($file).".bak"" is equivalent to "-f "$file.bak"".
338 See also "Terms and List Operators (Leftward)".
340 Relational Operators
342 Binary "<" returns true if the left argument is numerically less than
343 the right argument.
344 Binary ">" returns true if the left argument is numerically greater
346 than the right argument.
347 Binary "<=" returns true if the left argument is numerically less than
349 or equal to the right argument.
350 Binary ">=" returns true if the left argument is numerically greater
352 than or equal to the right argument.
353 Binary "lt" returns true if the left argument is stringwise less than
355 the right argument.
356 Binary "gt" returns true if the left argument is stringwise greater
358 than the right argument.
359 Binary "le" returns true if the left argument is stringwise less than
361 or equal to the right argument.
362 Binary "ge" returns true if the left argument is stringwise greater
364 than or equal to the right argument.
365 Equality Operators
367 Binary "==" returns true if the left argument is numerically equal to
368 the right argument.
369 Binary "!=" returns true if the left argument is numerically not equal
371 to the right argument.
372 Binary "<=>" returns -1, 0, or 1 depending on whether the left argument
374 is numerically less than, equal to, or greater than the right argument.
375 If your platform supports NaNs (not-a-numbers) as numeric values, using
376 them with "<=>" returns undef. NaN is not "<", "==", ">", "<=" or ">="
377 anything (even NaN), so those 5 return false. NaN != NaN returns true,
378 as does NaN != anything else. If your platform doesn't support NaNs
379 then NaN is just a string with numeric value 0.
380 perl -le '$a = "NaN"; print "No NaN support here" if $a == $a'
382 perl -le '$a = "NaN"; print "NaN support here" if $a != $a'
383 Binary "eq" returns true if the left argument is stringwise equal to
385 the right argument.
386 Binary "ne" returns true if the left argument is stringwise not equal
388 to the right argument.
389 Binary "cmp" returns -1, 0, or 1 depending on whether the left argument
391 is stringwise less than, equal to, or greater than the right argument.
392 Binary "~~" does a smart match between its arguments. Smart matching is
394 described in "Smart matching in detail" in perlsyn.
395 "lt", "le", "ge", "gt" and "cmp" use the collation (sort) order
397 specified by the current locale if "use locale" is in effect. See
398 perllocale.
399 Bitwise And
401 Binary "&" returns its operands ANDed together bit by bit. (See also
402 "Integer Arithmetic" and "Bitwise String Operators".)
403 Note that "&" has lower priority than relational operators, so for
405 example the brackets are essential in a test like
406 print "Even\n" if ($x & 1) == 0;
408 Bitwise Or and Exclusive Or
410 Binary "|" returns its operands ORed together bit by bit. (See also
411 "Integer Arithmetic" and "Bitwise String Operators".)
412 Binary "^" returns its operands XORed together bit by bit. (See also
414 "Integer Arithmetic" and "Bitwise String Operators".)
415 Note that "|" and "^" have lower priority than relational operators, so
417 for example the brackets are essential in a test like
418 print "false\n" if (8 | 2) != 10;
420 C-style Logical And
422 Binary "&&" performs a short-circuit logical AND operation. That is,
423 if the left operand is false, the right operand is not even evaluated.
424 Scalar or list context propagates down to the right operand if it is
425 evaluated.
426 C-style Logical Or
428 Binary "||" performs a short-circuit logical OR operation. That is, if
429 the left operand is true, the right operand is not even evaluated.
430 Scalar or list context propagates down to the right operand if it is
431 evaluated.
432 C-style Logical Defined-Or
434 Although it has no direct equivalent in C, Perl's "//" operator is
435 related to its C-style or. In fact, it's exactly the same as "||",
436 except that it tests the left hand side's definedness instead of its
437 truth. Thus, "$a // $b" is similar to "defined($a) || $b" (except that
438 it returns the value of $a rather than the value of "defined($a)") and
439 is exactly equivalent to "defined($a) ? $a : $b". This is very useful
440 for providing default values for variables. If you actually want to
441 test if at least one of $a and $b is defined, use "defined($a // $b)".
442 The "||", "//" and "&&" operators return the last value evaluated
444 (unlike C's "||" and "&&", which return 0 or 1). Thus, a reasonably
445 portable way to find out the home directory might be:
446 $home = $ENV{'HOME'} // $ENV{'LOGDIR'} //
448 (getpwuid($<))[7] // die "You're homeless!\n";
449 In particular, this means that you shouldn't use this for selecting
451 between two aggregates for assignment:
452 @a = @b || @c; # this is wrong
454 @a = scalar(@b) || @c; # really meant this
455 @a = @b ? @b : @c; # this works fine, though
456 As more readable alternatives to "&&" and "||" when used for control
458 flow, Perl provides the "and" and "or" operators (see below). The
459 short-circuit behavior is identical. The precedence of "and" and "or"
460 is much lower, however, so that you can safely use them after a list
461 operator without the need for parentheses:
462 unlink "alpha", "beta", "gamma"
464 or gripe(), next LINE;
465 With the C-style operators that would have been written like this:
467 unlink("alpha", "beta", "gamma")
469 || (gripe(), next LINE);
470 Using "or" for assignment is unlikely to do what you want; see below.
472 Range Operators
474 Binary ".." is the range operator, which is really two different
475 operators depending on the context. In list context, it returns a list
476 of values counting (up by ones) from the left value to the right value.
477 If the left value is greater than the right value then it returns the
478 empty list. The range operator is useful for writing "foreach (1..10)"
479 loops and for doing slice operations on arrays. In the current
480 implementation, no temporary array is created when the range operator
481 is used as the expression in "foreach" loops, but older versions of
482 Perl might burn a lot of memory when you write something like this:
483 for (1 .. 1_000_000) {
485 # code
486 }
487 The range operator also works on strings, using the magical auto-
489 increment, see below.
490 In scalar context, ".." returns a boolean value. The operator is
492 bistable, like a flip-flop, and emulates the line-range (comma)
493 operator of sed, awk, and various editors. Each ".." operator maintains
494 its own boolean state, even across calls to a subroutine that contains
495 it. It is false as long as its left operand is false. Once the left
496 operand is true, the range operator stays true until the right operand
497 is true, AFTER which the range operator becomes false again. It
498 doesn't become false till the next time the range operator is
499 evaluated. It can test the right operand and become false on the same
500 evaluation it became true (as in awk), but it still returns true once.
501 If you don't want it to test the right operand until the next
502 evaluation, as in sed, just use three dots ("...") instead of two. In
503 all other regards, "..." behaves just like ".." does.
504 The right operand is not evaluated while the operator is in the "false"
506 state, and the left operand is not evaluated while the operator is in
507 the "true" state. The precedence is a little lower than || and &&.
508 The value returned is either the empty string for false, or a sequence
509 number (beginning with 1) for true. The sequence number is reset for
510 each range encountered. The final sequence number in a range has the
511 string "E0" appended to it, which doesn't affect its numeric value, but
512 gives you something to search for if you want to exclude the endpoint.
513 You can exclude the beginning point by waiting for the sequence number
514 to be greater than 1.
515 If either operand of scalar ".." is a constant expression, that operand
517 is considered true if it is equal ("==") to the current input line
518 number (the $. variable).
519 To be pedantic, the comparison is actually "int(EXPR) == int(EXPR)",
521 but that is only an issue if you use a floating point expression; when
522 implicitly using $. as described in the previous paragraph, the
523 comparison is "int(EXPR) == int($.)" which is only an issue when $. is
524 set to a floating point value and you are not reading from a file.
525 Furthermore, "span" .. "spat" or "2.18 .. 3.14" will not do what you
526 want in scalar context because each of the operands are evaluated using
527 their integer representation.
528 Examples:
530 As a scalar operator:
532 if (101 .. 200) { print; } # print 2nd hundred lines, short for
534 # if ($. == 101 .. $. == 200) { print; }
535 next LINE if (1 .. /^$/); # skip header lines, short for
537 # next LINE if ($. == 1 .. /^$/);
538 # (typically in a loop labeled LINE)
539 s/^/> / if (/^$/ .. eof()); # quote body
541 # parse mail messages
543 while (<>) {
544 $in_header = 1 .. /^$/;
545 $in_body = /^$/ .. eof;
546 if ($in_header) {
547 # do something
548 } else { # in body
549 # do something else
550 }
551 } continue {
552 close ARGV if eof; # reset $. each file
553 }
554 Here's a simple example to illustrate the difference between the two
556 range operators:
557 @lines = (" - Foo",
559 "01 - Bar",
560 "1 - Baz",
561 " - Quux");
562 foreach (@lines) {
564 if (/0/ .. /1/) {
565 print "$_\n";
566 }
567 }
568 This program will print only the line containing "Bar". If the range
570 operator is changed to "...", it will also print the "Baz" line.
571 And now some examples as a list operator:
573 for (101 .. 200) { print; } # print $_ 100 times
575 @foo = @foo[0 .. $#foo]; # an expensive no-op
576 @foo = @foo[$#foo-4 .. $#foo]; # slice last 5 items
577 The range operator (in list context) makes use of the magical auto-
579 increment algorithm if the operands are strings. You can say
580 @alphabet = ('A' .. 'Z');
582 to get all normal letters of the English alphabet, or
584 $hexdigit = (0 .. 9, 'a' .. 'f')[$num & 15];
586 to get a hexadecimal digit, or
588 @z2 = ('01' .. '31'); print $z2[$mday];
590 to get dates with leading zeros.
592 If the final value specified is not in the sequence that the magical
594 increment would produce, the sequence goes until the next value would
595 be longer than the final value specified.
596 If the initial value specified isn't part of a magical increment
598 sequence (that is, a non-empty string matching "/^[a-zA-Z]*[0-9]*\z/"),
599 only the initial value will be returned. So the following will only
600 return an alpha:
601 use charnames 'greek';
603 my @greek_small = ("\N{alpha}" .. "\N{omega}");
604 To get lower-case greek letters, use this instead:
606 my @greek_small = map { chr } ( ord("\N{alpha}") .. ord("\N{omega}") );
608 Because each operand is evaluated in integer form, "2.18 .. 3.14" will
610 return two elements in list context.
611 @list = (2.18 .. 3.14); # same as @list = (2 .. 3);
613 Conditional Operator
615 Ternary "?:" is the conditional operator, just as in C. It works much
616 like an if-then-else. If the argument before the ? is true, the
617 argument before the : is returned, otherwise the argument after the :
618 is returned. For example:
619 printf "I have %d dog%s.\n", $n,
621 ($n == 1) ? '' : "s";
622 Scalar or list context propagates downward into the 2nd or 3rd
624 argument, whichever is selected.
625 $a = $ok ? $b : $c; # get a scalar
627 @a = $ok ? @b : @c; # get an array
628 $a = $ok ? @b : @c; # oops, that's just a count!
629 The operator may be assigned to if both the 2nd and 3rd arguments are
631 legal lvalues (meaning that you can assign to them):
632 ($a_or_b ? $a : $b) = $c;
634 Because this operator produces an assignable result, using assignments
636 without parentheses will get you in trouble. For example, this:
637 $a % 2 ? $a += 10 : $a += 2
639 Really means this:
641 (($a % 2) ? ($a += 10) : $a) += 2
643 Rather than this:
645 ($a % 2) ? ($a += 10) : ($a += 2)
647 That should probably be written more simply as:
649 $a += ($a % 2) ? 10 : 2;
651 Assignment Operators
653 "=" is the ordinary assignment operator.
654 Assignment operators work as in C. That is,
656 $a += 2;
658 is equivalent to
660 $a = $a + 2;
662 although without duplicating any side effects that dereferencing the
664 lvalue might trigger, such as from tie(). Other assignment operators
665 work similarly. The following are recognized:
666 **= += *= &= <<= &&=
668 -= /= |= >>= ||=
669 .= %= ^= //=
670 x=
671 Although these are grouped by family, they all have the precedence of
673 assignment.
674 Unlike in C, the scalar assignment operator produces a valid lvalue.
676 Modifying an assignment is equivalent to doing the assignment and then
677 modifying the variable that was assigned to. This is useful for
678 modifying a copy of something, like this:
679 ($tmp = $global) =~ tr [A-Z] [a-z];
681 Likewise,
683 ($a += 2) *= 3;
685 is equivalent to
687 $a += 2;
689 $a *= 3;
690 Similarly, a list assignment in list context produces the list of
692 lvalues assigned to, and a list assignment in scalar context returns
693 the number of elements produced by the expression on the right hand
694 side of the assignment.
695 Comma Operator
697 Binary "," is the comma operator. In scalar context it evaluates its
698 left argument, throws that value away, then evaluates its right
699 argument and returns that value. This is just like C's comma operator.
700 In list context, it's just the list argument separator, and inserts
702 both its arguments into the list. These arguments are also evaluated
703 from left to right.
704 The "=>" operator is a synonym for the comma except that it causes its
706 left operand to be interpreted as a string if it begins with a letter
707 or underscore and is composed only of letters, digits and underscores.
708 This includes operands that might otherwise be interpreted as
709 operators, constants, single number v-strings or function calls. If in
710 doubt about this behaviour, the left operand can be quoted explicitly.
711 Otherwise, the "=>" operator behaves exactly as the comma operator or
713 list argument separator, according to context.
714 For example:
716 use constant FOO => "something";
718 my %h = ( FOO => 23 );
720 is equivalent to:
722 my %h = ("FOO", 23);
724 It is NOT:
726 my %h = ("something", 23);
728 The "=>" operator is helpful in documenting the correspondence between
730 keys and values in hashes, and other paired elements in lists.
731 %hash = ( $key => $value );
733 login( $username => $password );
734 Yada Yada Operator
736 The yada yada operator (noted "...") is a placeholder for code. Perl
737 parses it without error, but when you try to execute a yada yada, it
738 throws an exception with the text "Unimplemented":
739 sub unimplemented { ... }
741 eval { unimplemented() };
743 if( $@ eq 'Unimplemented' ) {
744 print "I found the yada yada!\n";
745 }
746 You can only use the yada yada to stand in for a complete statement.
748 These examples of the yada yada work:
749 { ... }
751 sub foo { ... }
753 ...;
755 eval { ... };
757 sub foo {
759 my( $self ) = shift;
760 ...;
762 }
763 do { my $n; ...; print 'Hurrah!' };
765 The yada yada cannot stand in for an expression that is part of a
767 larger statement since the "..." is also the three-dot version of the
768 range operator (see "Range Operators"). These examples of the yada yada
769 are still syntax errors:
770 print ...;
772 open my($fh), '>', '/dev/passwd' or ...;
774 if( $condition && ... ) { print "Hello\n" };
776 There are some cases where Perl can't immediately tell the difference
778 between an expression and a statement. For instance, the syntax for a
779 block and an anonymous hash reference constructor look the same unless
780 there's something in the braces that give Perl a hint. The yada yada is
781 a syntax error if Perl doesn't guess that the "{ ... }" is a block. In
782 that case, it doesn't think the "..." is the yada yada because it's
783 expecting an expression instead of a statement:
784 my @transformed = map { ... } @input; # syntax error
786 You can use a ";" inside your block to denote that the "{ ... }" is a
788 block and not a hash reference constructor. Now the yada yada works:
789 my @transformed = map {; ... } @input; # ; disambiguates
791 my @transformed = map { ...; } @input; # ; disambiguates
793 List Operators (Rightward)
795 On the right side of a list operator, it has very low precedence, such
796 that it controls all comma-separated expressions found there. The only
797 operators with lower precedence are the logical operators "and", "or",
798 and "not", which may be used to evaluate calls to list operators
799 without the need for extra parentheses:
800 open HANDLE, "filename"
802 or die "Can't open: $!\n";
803 See also discussion of list operators in "Terms and List Operators
805 (Leftward)".
806 Logical Not
808 Unary "not" returns the logical negation of the expression to its
809 right. It's the equivalent of "!" except for the very low precedence.
810 Logical And
812 Binary "and" returns the logical conjunction of the two surrounding
813 expressions. It's equivalent to && except for the very low precedence.
814 This means that it short-circuits: i.e., the right expression is
815 evaluated only if the left expression is true.
816 Logical or, Defined or, and Exclusive Or
818 Binary "or" returns the logical disjunction of the two surrounding
819 expressions. It's equivalent to || except for the very low precedence.
820 This makes it useful for control flow
821 print FH $data or die "Can't write to FH: $!";
823 This means that it short-circuits: i.e., the right expression is
825 evaluated only if the left expression is false. Due to its precedence,
826 you should probably avoid using this for assignment, only for control
827 flow.
828 $a = $b or $c; # bug: this is wrong
830 ($a = $b) or $c; # really means this
831 $a = $b || $c; # better written this way
832 However, when it's a list-context assignment and you're trying to use
834 "||" for control flow, you probably need "or" so that the assignment
835 takes higher precedence.
836 @info = stat($file) || die; # oops, scalar sense of stat!
838 @info = stat($file) or die; # better, now @info gets its due
839 Then again, you could always use parentheses.
841 Binary "xor" returns the exclusive-OR of the two surrounding
843 expressions. It cannot short circuit, of course.
844 C Operators Missing From Perl
846 Here is what C has that Perl doesn't:
847 unary & Address-of operator. (But see the "\" operator for taking a
849 reference.)
850 unary * Dereference-address operator. (Perl's prefix dereferencing
852 operators are typed: $, @, %, and &.)
853 (TYPE) Type-casting operator.
855 Quote and Quote-like Operators
857 While we usually think of quotes as literal values, in Perl they
858 function as operators, providing various kinds of interpolating and
859 pattern matching capabilities. Perl provides customary quote
860 characters for these behaviors, but also provides a way for you to
861 choose your quote character for any of them. In the following table, a
862 "{}" represents any pair of delimiters you choose.
863 Customary Generic Meaning Interpolates
865 '' q{} Literal no
866 "" qq{} Literal yes
867 `` qx{} Command yes*
868 qw{} Word list no
869 // m{} Pattern match yes*
870 qr{} Pattern yes*
871 s{}{} Substitution yes*
872 tr{}{} Transliteration no (but see below)
873 <<EOF here-doc yes*
874 * unless the delimiter is ''.
876 Non-bracketing delimiters use the same character fore and aft, but the
878 four sorts of brackets (round, angle, square, curly) will all nest,
879 which means that
880 q{foo{bar}baz}
882 is the same as
884 'foo{bar}baz'
886 Note, however, that this does not always work for quoting Perl code:
888 $s = q{ if($a eq "}") ... }; # WRONG
890 is a syntax error. The "Text::Balanced" module (from CPAN, and starting
892 from Perl 5.8 part of the standard distribution) is able to do this
893 properly.
894 There can be whitespace between the operator and the quoting
896 characters, except when "#" is being used as the quoting character.
897 "q#foo#" is parsed as the string "foo", while "q #foo#" is the operator
898 "q" followed by a comment. Its argument will be taken from the next
899 line. This allows you to write:
900 s {foo} # Replace foo
902 {bar} # with bar.
903 The following escape sequences are available in constructs that
905 interpolate and in transliterations.
906 \t tab (HT, TAB)
908 \n newline (NL)
909 \r return (CR)
910 \f form feed (FF)
911 \b backspace (BS)
912 \a alarm (bell) (BEL)
913 \e escape (ESC)
914 \033 octal char (example: ESC)
915 \x1b hex char (example: ESC)
916 \x{263a} wide hex char (example: SMILEY)
917 \c[ control char (example: ESC)
918 \N{name} named Unicode character
919 \N{U+263D} Unicode character (example: FIRST QUARTER MOON)
920 The character following "\c" is mapped to some other character by
922 converting letters to upper case and then (on ASCII systems) by
923 inverting the 7th bit (0x40). The most interesting range is from '@' to
924 '_' (0x40 through 0x5F), resulting in a control character from 0x00
925 through 0x1F. A '?' maps to the DEL character. On EBCDIC systems only
926 '@', the letters, '[', '\', ']', '^', '_' and '?' will work, resulting
927 in 0x00 through 0x1F and 0x7F.
928 "\N{U+wide hex char}" means the Unicode character whose Unicode ordinal
930 number is wide hex char. For documentation of "\N{name}", see
931 charnames.
932 NOTE: Unlike C and other languages, Perl has no "\v" escape sequence
934 for the vertical tab (VT - ASCII 11), but you may use "\ck" or "\x0b".
935 ("\v" does have meaning in regular expression patterns in Perl, see
936 perlre.)
937 The following escape sequences are available in constructs that
939 interpolate, but not in transliterations.
940 \l lowercase next char
942 \u uppercase next char
943 \L lowercase till \E
944 \U uppercase till \E
945 \E end case modification
946 \Q quote non-word characters till \E
947 If "use locale" is in effect, the case map used by "\l", "\L", "\u" and
949 "\U" is taken from the current locale. See perllocale. If Unicode
950 (for example, "\N{}" or wide hex characters of 0x100 or beyond) is
951 being used, the case map used by "\l", "\L", "\u" and "\U" is as
952 defined by Unicode.
953 All systems use the virtual "\n" to represent a line terminator, called
955 a "newline". There is no such thing as an unvarying, physical newline
956 character. It is only an illusion that the operating system, device
957 drivers, C libraries, and Perl all conspire to preserve. Not all
958 systems read "\r" as ASCII CR and "\n" as ASCII LF. For example, on a
959 Mac, these are reversed, and on systems without line terminator,
960 printing "\n" may emit no actual data. In general, use "\n" when you
961 mean a "newline" for your system, but use the literal ASCII when you
962 need an exact character. For example, most networking protocols expect
963 and prefer a CR+LF ("\015\012" or "\cM\cJ") for line terminators, and
964 although they often accept just "\012", they seldom tolerate just
965 "\015". If you get in the habit of using "\n" for networking, you may
966 be burned some day.
967 For constructs that do interpolate, variables beginning with ""$"" or
969 ""@"" are interpolated. Subscripted variables such as $a[3] or
970 "$href->{key}[0]" are also interpolated, as are array and hash slices.
971 But method calls such as "$obj->meth" are not.
972 Interpolating an array or slice interpolates the elements in order,
974 separated by the value of $", so is equivalent to interpolating "join
975 $", @array". "Punctuation" arrays such as "@*" are only interpolated
976 if the name is enclosed in braces "@{*}", but special arrays @_, "@+",
977 and "@-" are interpolated, even without braces.
978 You cannot include a literal "$" or "@" within a "\Q" sequence. An
980 unescaped "$" or "@" interpolates the corresponding variable, while
981 escaping will cause the literal string "\$" to be inserted. You'll
982 need to write something like "m/\Quser\E\@\Qhost/".
983 Patterns are subject to an additional level of interpretation as a
985 regular expression. This is done as a second pass, after variables are
986 interpolated, so that regular expressions may be incorporated into the
987 pattern from the variables. If this is not what you want, use "\Q" to
988 interpolate a variable literally.
989 Apart from the behavior described above, Perl does not expand multiple
991 levels of interpolation. In particular, contrary to the expectations
992 of shell programmers, back-quotes do NOT interpolate within double
993 quotes, nor do single quotes impede evaluation of variables when used
994 within double quotes.
995 Regexp Quote-Like Operators
997 Here are the quote-like operators that apply to pattern matching and
998 related activities.
999 qr/STRING/msixpo
1001 This operator quotes (and possibly compiles) its STRING as a
1002 regular expression. STRING is interpolated the same way as
1003 PATTERN in "m/PATTERN/". If "'" is used as the delimiter, no
1004 interpolation is done. Returns a Perl value which may be used
1005 instead of the corresponding "/STRING/msixpo" expression. The
1006 returned value is a normalized version of the original pattern.
1007 It magically differs from a string containing the same
1008 characters: "ref(qr/x/)" returns "Regexp", even though
1009 dereferencing the result returns undef.
1010 For example,
1012 $rex = qr/my.STRING/is;
1014 print $rex; # prints (?si-xm:my.STRING)
1015 s/$rex/foo/;
1016 is equivalent to
1018 s/my.STRING/foo/is;
1020 The result may be used as a subpattern in a match:
1022 $re = qr/$pattern/;
1024 $string =~ /foo${re}bar/; # can be interpolated in other patterns
1025 $string =~ $re; # or used standalone
1026 $string =~ /$re/; # or this way
1027 Since Perl may compile the pattern at the moment of execution
1029 of qr() operator, using qr() may have speed advantages in some
1030 situations, notably if the result of qr() is used standalone:
1031 sub match {
1033 my $patterns = shift;
1034 my @compiled = map qr/$_/i, @$patterns;
1035 grep {
1036 my $success = 0;
1037 foreach my $pat (@compiled) {
1038 $success = 1, last if /$pat/;
1039 }
1040 $success;
1041 } @_;
1042 }
1043 Precompilation of the pattern into an internal representation
1045 at the moment of qr() avoids a need to recompile the pattern
1046 every time a match "/$pat/" is attempted. (Perl has many other
1047 internal optimizations, but none would be triggered in the
1048 above example if we did not use qr() operator.)
1049 Options are:
1051 m Treat string as multiple lines.
1053 s Treat string as single line. (Make . match a newline)
1054 i Do case-insensitive pattern matching.
1055 x Use extended regular expressions.
1056 p When matching preserve a copy of the matched string so
1057 that ${^PREMATCH}, ${^MATCH}, ${^POSTMATCH} will be defined.
1058 o Compile pattern only once.
1059 If a precompiled pattern is embedded in a larger pattern then
1061 the effect of 'msixp' will be propagated appropriately. The
1062 effect of the 'o' modifier has is not propagated, being
1063 restricted to those patterns explicitly using it.
1064 See perlre for additional information on valid syntax for
1066 STRING, and for a detailed look at the semantics of regular
1067 expressions.
1068 m/PATTERN/msixpogc
1070 /PATTERN/msixpogc
1071 Searches a string for a pattern match, and in scalar context
1072 returns true if it succeeds, false if it fails. If no string
1073 is specified via the "=~" or "!~" operator, the $_ string is
1074 searched. (The string specified with "=~" need not be an
1075 lvalue--it may be the result of an expression evaluation, but
1076 remember the "=~" binds rather tightly.) See also perlre. See
1077 perllocale for discussion of additional considerations that
1078 apply when "use locale" is in effect.
1079 Options are as described in "qr//"; in addition, the following
1081 match process modifiers are available:
1082 g Match globally, i.e., find all occurrences.
1084 c Do not reset search position on a failed match when /g is in effect.
1085 If "/" is the delimiter then the initial "m" is optional. With
1087 the "m" you can use any pair of non-whitespace characters as
1088 delimiters. This is particularly useful for matching path
1089 names that contain "/", to avoid LTS (leaning toothpick
1090 syndrome). If "?" is the delimiter, then the match-only-once
1091 rule of "?PATTERN?" applies. If "'" is the delimiter, no
1092 interpolation is performed on the PATTERN. When using a
1093 character valid in an identifier, whitespace is required after
1094 the "m".
1095 PATTERN may contain variables, which will be interpolated (and
1097 the pattern recompiled) every time the pattern search is
1098 evaluated, except for when the delimiter is a single quote.
1099 (Note that $(, $), and $| are not interpolated because they
1100 look like end-of-string tests.) If you want such a pattern to
1101 be compiled only once, add a "/o" after the trailing delimiter.
1102 This avoids expensive run-time recompilations, and is useful
1103 when the value you are interpolating won't change over the life
1104 of the script. However, mentioning "/o" constitutes a promise
1105 that you won't change the variables in the pattern. If you
1106 change them, Perl won't even notice. See also "STRING/msixpo""
1107 in "qr.
1108 The empty pattern //
1110 If the PATTERN evaluates to the empty string, the last
1111 successfully matched regular expression is used instead. In
1112 this case, only the "g" and "c" flags on the empty pattern is
1113 honoured - the other flags are taken from the original pattern.
1114 If no match has previously succeeded, this will (silently) act
1115 instead as a genuine empty pattern (which will always match).
1116 Note that it's possible to confuse Perl into thinking "//" (the
1118 empty regex) is really "//" (the defined-or operator). Perl is
1119 usually pretty good about this, but some pathological cases
1120 might trigger this, such as "$a///" (is that "($a) / (//)" or
1121 "$a // /"?) and "print $fh //" ("print $fh(//" or "print($fh
1122 //"?). In all of these examples, Perl will assume you meant
1123 defined-or. If you meant the empty regex, just use parentheses
1124 or spaces to disambiguate, or even prefix the empty regex with
1125 an "m" (so "//" becomes "m//").
1126 Matching in list context
1128 If the "/g" option is not used, "m//" in list context returns a
1129 list consisting of the subexpressions matched by the
1130 parentheses in the pattern, i.e., ($1, $2, $3...). (Note that
1131 here $1 etc. are also set, and that this differs from Perl 4's
1132 behavior.) When there are no parentheses in the pattern, the
1133 return value is the list "(1)" for success. With or without
1134 parentheses, an empty list is returned upon failure.
1135 Examples:
1137 open(TTY, '/dev/tty');
1139 <TTY> =~ /^y/i && foo(); # do foo if desired
1140 if (/Version: *([0-9.]*)/) { $version = $1; }
1142 next if m#^/usr/spool/uucp#;
1144 # poor man's grep
1146 $arg = shift;
1147 while (<>) {
1148 print if /$arg/o; # compile only once
1149 }
1150 if (($F1, $F2, $Etc) = ($foo =~ /^(\S+)\s+(\S+)\s*(.*)/))
1152 This last example splits $foo into the first two words and the
1154 remainder of the line, and assigns those three fields to $F1,
1155 $F2, and $Etc. The conditional is true if any variables were
1156 assigned, i.e., if the pattern matched.
1157 The "/g" modifier specifies global pattern matching--that is,
1159 matching as many times as possible within the string. How it
1160 behaves depends on the context. In list context, it returns a
1161 list of the substrings matched by any capturing parentheses in
1162 the regular expression. If there are no parentheses, it
1163 returns a list of all the matched strings, as if there were
1164 parentheses around the whole pattern.
1165 In scalar context, each execution of "m//g" finds the next
1167 match, returning true if it matches, and false if there is no
1168 further match. The position after the last match can be read
1169 or set using the pos() function; see "pos" in perlfunc. A
1170 failed match normally resets the search position to the
1171 beginning of the string, but you can avoid that by adding the
1172 "/c" modifier (e.g. "m//gc"). Modifying the target string also
1173 resets the search position.
1174 \G assertion
1176 You can intermix "m//g" matches with "m/\G.../g", where "\G" is
1177 a zero-width assertion that matches the exact position where
1178 the previous "m//g", if any, left off. Without the "/g"
1179 modifier, the "\G" assertion still anchors at pos(), but the
1180 match is of course only attempted once. Using "\G" without
1181 "/g" on a target string that has not previously had a "/g"
1182 match applied to it is the same as using the "\A" assertion to
1183 match the beginning of the string. Note also that, currently,
1184 "\G" is only properly supported when anchored at the very
1185 beginning of the pattern.
1186 Examples:
1188 # list context
1190 ($one,$five,$fifteen) = (`uptime` =~ /(\d+\.\d+)/g);
1191 # scalar context
1193 $/ = "";
1194 while (defined($paragraph = <>)) {
1195 while ($paragraph =~ /[a-z]['")]*[.!?]+['")]*\s/g) {
1196 $sentences++;
1197 }
1198 }
1199 print "$sentences\n";
1200 # using m//gc with \G
1202 $_ = "ppooqppqq";
1203 while ($i++ < 2) {
1204 print "1: '";
1205 print $1 while /(o)/gc; print "', pos=", pos, "\n";
1206 print "2: '";
1207 print $1 if /\G(q)/gc; print "', pos=", pos, "\n";
1208 print "3: '";
1209 print $1 while /(p)/gc; print "', pos=", pos, "\n";
1210 }
1211 print "Final: '$1', pos=",pos,"\n" if /\G(.)/;
1212 The last example should print:
1214 1: 'oo', pos=4
1216 2: 'q', pos=5
1217 3: 'pp', pos=7
1218 1: '', pos=7
1219 2: 'q', pos=8
1220 3: '', pos=8
1221 Final: 'q', pos=8
1222 Notice that the final match matched "q" instead of "p", which a
1224 match without the "\G" anchor would have done. Also note that
1225 the final match did not update "pos". "pos" is only updated on
1226 a "/g" match. If the final match did indeed match "p", it's a
1227 good bet that you're running an older (pre-5.6.0) Perl.
1228 A useful idiom for "lex"-like scanners is "/\G.../gc". You can
1230 combine several regexps like this to process a string part-by-
1231 part, doing different actions depending on which regexp
1232 matched. Each regexp tries to match where the previous one
1233 leaves off.
1234 $_ = <<'EOL';
1236 $url = URI::URL->new( "http://example.com/" ); die if $url eq "xXx";
1237 EOL
1238 LOOP:
1239 {
1240 print(" digits"), redo LOOP if /\G\d+\b[,.;]?\s*/gc;
1241 print(" lowercase"), redo LOOP if /\G[a-z]+\b[,.;]?\s*/gc;
1242 print(" UPPERCASE"), redo LOOP if /\G[A-Z]+\b[,.;]?\s*/gc;
1243 print(" Capitalized"), redo LOOP if /\G[A-Z][a-z]+\b[,.;]?\s*/gc;
1244 print(" MiXeD"), redo LOOP if /\G[A-Za-z]+\b[,.;]?\s*/gc;
1245 print(" alphanumeric"), redo LOOP if /\G[A-Za-z0-9]+\b[,.;]?\s*/gc;
1246 print(" line-noise"), redo LOOP if /\G[^A-Za-z0-9]+/gc;
1247 print ". That's all!\n";
1248 }
1249 Here is the output (split into several lines):
1251 line-noise lowercase line-noise lowercase UPPERCASE line-noise
1253 UPPERCASE line-noise lowercase line-noise lowercase line-noise
1254 lowercase lowercase line-noise lowercase lowercase line-noise
1255 MiXeD line-noise. That's all!
1256 ?PATTERN?
1258 This is just like the "/pattern/" search, except that it
1259 matches only once between calls to the reset() operator. This
1260 is a useful optimization when you want to see only the first
1261 occurrence of something in each file of a set of files, for
1262 instance. Only "??" patterns local to the current package are
1263 reset.
1264 while (<>) {
1266 if (?^$?) {
1267 # blank line between header and body
1268 }
1269 } continue {
1270 reset if eof; # clear ?? status for next file
1271 }
1272 This usage is vaguely deprecated, which means it just might
1274 possibly be removed in some distant future version of Perl,
1275 perhaps somewhere around the year 2168.
1276 s/PATTERN/REPLACEMENT/msixpogce
1278 Searches a string for a pattern, and if found, replaces that
1279 pattern with the replacement text and returns the number of
1280 substitutions made. Otherwise it returns false (specifically,
1281 the empty string).
1282 If no string is specified via the "=~" or "!~" operator, the $_
1284 variable is searched and modified. (The string specified with
1285 "=~" must be scalar variable, an array element, a hash element,
1286 or an assignment to one of those, i.e., an lvalue.)
1287 If the delimiter chosen is a single quote, no interpolation is
1289 done on either the PATTERN or the REPLACEMENT. Otherwise, if
1290 the PATTERN contains a $ that looks like a variable rather than
1291 an end-of-string test, the variable will be interpolated into
1292 the pattern at run-time. If you want the pattern compiled only
1293 once the first time the variable is interpolated, use the "/o"
1294 option. If the pattern evaluates to the empty string, the last
1295 successfully executed regular expression is used instead. See
1296 perlre for further explanation on these. See perllocale for
1297 discussion of additional considerations that apply when "use
1298 locale" is in effect.
1299 Options are as with m// with the addition of the following
1301 replacement specific options:
1302 e Evaluate the right side as an expression.
1304 ee Evaluate the right side as a string then eval the result
1305 Any non-whitespace delimiter may replace the slashes. Add
1307 space after the "s" when using a character allowed in
1308 identifiers. If single quotes are used, no interpretation is
1309 done on the replacement string (the "/e" modifier overrides
1310 this, however). Unlike Perl 4, Perl 5 treats backticks as
1311 normal delimiters; the replacement text is not evaluated as a
1312 command. If the PATTERN is delimited by bracketing quotes, the
1313 REPLACEMENT has its own pair of quotes, which may or may not be
1314 bracketing quotes, e.g., "s(foo)(bar)" or "s<foo>/bar/". A
1315 "/e" will cause the replacement portion to be treated as a
1316 full-fledged Perl expression and evaluated right then and
1317 there. It is, however, syntax checked at compile-time. A
1318 second "e" modifier will cause the replacement portion to be
1319 "eval"ed before being run as a Perl expression.
1320 Examples:
1322 s/\bgreen\b/mauve/g; # don't change wintergreen
1324 $path =~ s|/usr/bin|/usr/local/bin|;
1326 s/Login: $foo/Login: $bar/; # run-time pattern
1328 ($foo = $bar) =~ s/this/that/; # copy first, then change
1330 $count = ($paragraph =~ s/Mister\b/Mr./g); # get change-count
1332 $_ = 'abc123xyz';
1334 s/\d+/$&*2/e; # yields 'abc246xyz'
1335 s/\d+/sprintf("%5d",$&)/e; # yields 'abc 246xyz'
1336 s/\w/$& x 2/eg; # yields 'aabbcc 224466xxyyzz'
1337 s/%(.)/$percent{$1}/g; # change percent escapes; no /e
1339 s/%(.)/$percent{$1} || $&/ge; # expr now, so /e
1340 s/^=(\w+)/pod($1)/ge; # use function call
1341 # expand variables in $_, but dynamics only, using
1343 # symbolic dereferencing
1344 s/\$(\w+)/${$1}/g;
1345 # Add one to the value of any numbers in the string
1347 s/(\d+)/1 + $1/eg;
1348 # This will expand any embedded scalar variable
1350 # (including lexicals) in $_ : First $1 is interpolated
1351 # to the variable name, and then evaluated
1352 s/(\$\w+)/$1/eeg;
1353 # Delete (most) C comments.
1355 $program =~ s {
1356 /\* # Match the opening delimiter.
1357 .*? # Match a minimal number of characters.
1358 \*/ # Match the closing delimiter.
1359 } []gsx;
1360 s/^\s*(.*?)\s*$/$1/; # trim whitespace in $_, expensively
1362 for ($variable) { # trim whitespace in $variable, cheap
1364 s/^\s+//;
1365 s/\s+$//;
1366 }
1367 s/([^ ]*) *([^ ]*)/$2 $1/; # reverse 1st two fields
1369 Note the use of $ instead of \ in the last example. Unlike
1371 sed, we use the \<digit> form in only the left hand side.
1372 Anywhere else it's $<digit>.
1373 Occasionally, you can't use just a "/g" to get all the changes
1375 to occur that you might want. Here are two common cases:
1376 # put commas in the right places in an integer
1378 1 while s/(\d)(\d\d\d)(?!\d)/$1,$2/g;
1379 # expand tabs to 8-column spacing
1381 1 while s/\t+/' ' x (length($&)*8 - length($`)%8)/e;
1382 Quote-Like Operators
1384 q/STRING/
1385 'STRING'
1386 A single-quoted, literal string. A backslash represents a
1387 backslash unless followed by the delimiter or another backslash, in
1388 which case the delimiter or backslash is interpolated.
1389 $foo = q!I said, "You said, 'She said it.'"!;
1391 $bar = q('This is it.');
1392 $baz = '\n'; # a two-character string
1393 qq/STRING/
1395 "STRING"
1396 A double-quoted, interpolated string.
1397 $_ .= qq
1399 (*** The previous line contains the naughty word "$1".\n)
1400 if /\b(tcl|java|python)\b/i; # :-)
1401 $baz = "\n"; # a one-character string
1402 qx/STRING/
1404 `STRING`
1405 A string which is (possibly) interpolated and then executed as a
1406 system command with "/bin/sh" or its equivalent. Shell wildcards,
1407 pipes, and redirections will be honored. The collected standard
1408 output of the command is returned; standard error is unaffected.
1409 In scalar context, it comes back as a single (potentially multi-
1410 line) string, or undef if the command failed. In list context,
1411 returns a list of lines (however you've defined lines with $/ or
1412 $INPUT_RECORD_SEPARATOR), or an empty list if the command failed.
1413 Because backticks do not affect standard error, use shell file
1415 descriptor syntax (assuming the shell supports this) if you care to
1416 address this. To capture a command's STDERR and STDOUT together:
1417 $output = `cmd 2>&1`;
1419 To capture a command's STDOUT but discard its STDERR:
1421 $output = `cmd 2>/dev/null`;
1423 To capture a command's STDERR but discard its STDOUT (ordering is
1425 important here):
1426 $output = `cmd 2>&1 1>/dev/null`;
1428 To exchange a command's STDOUT and STDERR in order to capture the
1430 STDERR but leave its STDOUT to come out the old STDERR:
1431 $output = `cmd 3>&1 1>&2 2>&3 3>&-`;
1433 To read both a command's STDOUT and its STDERR separately, it's
1435 easiest to redirect them separately to files, and then read from
1436 those files when the program is done:
1437 system("program args 1>program.stdout 2>program.stderr");
1439 The STDIN filehandle used by the command is inherited from Perl's
1441 STDIN. For example:
1442 open BLAM, "blam" || die "Can't open: $!";
1444 open STDIN, "<&BLAM";
1445 print `sort`;
1446 will print the sorted contents of the file "blam".
1448 Using single-quote as a delimiter protects the command from Perl's
1450 double-quote interpolation, passing it on to the shell instead:
1451 $perl_info = qx(ps $$); # that's Perl's $$
1453 $shell_info = qx'ps $$'; # that's the new shell's $$
1454 How that string gets evaluated is entirely subject to the command
1456 interpreter on your system. On most platforms, you will have to
1457 protect shell metacharacters if you want them treated literally.
1458 This is in practice difficult to do, as it's unclear how to escape
1459 which characters. See perlsec for a clean and safe example of a
1460 manual fork() and exec() to emulate backticks safely.
1461 On some platforms (notably DOS-like ones), the shell may not be
1463 capable of dealing with multiline commands, so putting newlines in
1464 the string may not get you what you want. You may be able to
1465 evaluate multiple commands in a single line by separating them with
1466 the command separator character, if your shell supports that (e.g.
1467 ";" on many Unix shells; "&" on the Windows NT "cmd" shell).
1468 Beginning with v5.6.0, Perl will attempt to flush all files opened
1470 for output before starting the child process, but this may not be
1471 supported on some platforms (see perlport). To be safe, you may
1472 need to set $| ($AUTOFLUSH in English) or call the "autoflush()"
1473 method of "IO::Handle" on any open handles.
1474 Beware that some command shells may place restrictions on the
1476 length of the command line. You must ensure your strings don't
1477 exceed this limit after any necessary interpolations. See the
1478 platform-specific release notes for more details about your
1479 particular environment.
1480 Using this operator can lead to programs that are difficult to
1482 port, because the shell commands called vary between systems, and
1483 may in fact not be present at all. As one example, the "type"
1484 command under the POSIX shell is very different from the "type"
1485 command under DOS. That doesn't mean you should go out of your way
1486 to avoid backticks when they're the right way to get something
1487 done. Perl was made to be a glue language, and one of the things
1488 it glues together is commands. Just understand what you're getting
1489 yourself into.
1490 See "I/O Operators" for more discussion.
1492 qw/STRING/
1494 Evaluates to a list of the words extracted out of STRING, using
1495 embedded whitespace as the word delimiters. It can be understood
1496 as being roughly equivalent to:
1497 split(' ', q/STRING/);
1499 the differences being that it generates a real list at compile
1501 time, and in scalar context it returns the last element in the
1502 list. So this expression:
1503 qw(foo bar baz)
1505 is semantically equivalent to the list:
1507 'foo', 'bar', 'baz'
1509 Some frequently seen examples:
1511 use POSIX qw( setlocale localeconv )
1513 @EXPORT = qw( foo bar baz );
1514 A common mistake is to try to separate the words with comma or to
1516 put comments into a multi-line "qw"-string. For this reason, the
1517 "use warnings" pragma and the -w switch (that is, the $^W variable)
1518 produces warnings if the STRING contains the "," or the "#"
1519 character.
1520 tr/SEARCHLIST/REPLACEMENTLIST/cds
1522 y/SEARCHLIST/REPLACEMENTLIST/cds
1523 Transliterates all occurrences of the characters found in the
1524 search list with the corresponding character in the replacement
1525 list. It returns the number of characters replaced or deleted. If
1526 no string is specified via the =~ or !~ operator, the $_ string is
1527 transliterated. (The string specified with =~ must be a scalar
1528 variable, an array element, a hash element, or an assignment to one
1529 of those, i.e., an lvalue.)
1530 A character range may be specified with a hyphen, so "tr/A-J/0-9/"
1532 does the same replacement as "tr/ACEGIBDFHJ/0246813579/". For sed
1533 devotees, "y" is provided as a synonym for "tr". If the SEARCHLIST
1534 is delimited by bracketing quotes, the REPLACEMENTLIST has its own
1535 pair of quotes, which may or may not be bracketing quotes, e.g.,
1536 "tr[A-Z][a-z]" or "tr(+\-*/)/ABCD/".
1537 Note that "tr" does not do regular expression character classes
1539 such as "\d" or "[:lower:]". The "tr" operator is not equivalent
1540 to the tr(1) utility. If you want to map strings between
1541 lower/upper cases, see "lc" in perlfunc and "uc" in perlfunc, and
1542 in general consider using the "s" operator if you need regular
1543 expressions.
1544 Note also that the whole range idea is rather unportable between
1546 character sets--and even within character sets they may cause
1547 results you probably didn't expect. A sound principle is to use
1548 only ranges that begin from and end at either alphabets of equal
1549 case (a-e, A-E), or digits (0-4). Anything else is unsafe. If in
1550 doubt, spell out the character sets in full.
1551 Options:
1553 c Complement the SEARCHLIST.
1555 d Delete found but unreplaced characters.
1556 s Squash duplicate replaced characters.
1557 If the "/c" modifier is specified, the SEARCHLIST character set is
1559 complemented. If the "/d" modifier is specified, any characters
1560 specified by SEARCHLIST not found in REPLACEMENTLIST are deleted.
1561 (Note that this is slightly more flexible than the behavior of some
1562 tr programs, which delete anything they find in the SEARCHLIST,
1563 period.) If the "/s" modifier is specified, sequences of characters
1564 that were transliterated to the same character are squashed down to
1565 a single instance of the character.
1566 If the "/d" modifier is used, the REPLACEMENTLIST is always
1568 interpreted exactly as specified. Otherwise, if the
1569 REPLACEMENTLIST is shorter than the SEARCHLIST, the final character
1570 is replicated till it is long enough. If the REPLACEMENTLIST is
1571 empty, the SEARCHLIST is replicated. This latter is useful for
1572 counting characters in a class or for squashing character sequences
1573 in a class.
1574 Examples:
1576 $ARGV[1] =~ tr/A-Z/a-z/; # canonicalize to lower case
1578 $cnt = tr/*/*/; # count the stars in $_
1580 $cnt = $sky =~ tr/*/*/; # count the stars in $sky
1582 $cnt = tr/0-9//; # count the digits in $_
1584 tr/a-zA-Z//s; # bookkeeper -> bokeper
1586 ($HOST = $host) =~ tr/a-z/A-Z/;
1588 tr/a-zA-Z/ /cs; # change non-alphas to single space
1590 tr [\200-\377]
1592 [\000-\177]; # delete 8th bit
1593 If multiple transliterations are given for a character, only the
1595 first one is used:
1596 tr/AAA/XYZ/
1598 will transliterate any A to X.
1600 Because the transliteration table is built at compile time, neither
1602 the SEARCHLIST nor the REPLACEMENTLIST are subjected to double
1603 quote interpolation. That means that if you want to use variables,
1604 you must use an eval():
1605 eval "tr/$oldlist/$newlist/";
1607 die $@ if $@;
1608 eval "tr/$oldlist/$newlist/, 1" or die $@;
1610 <<EOF
1612 A line-oriented form of quoting is based on the shell "here-
1613 document" syntax. Following a "<<" you specify a string to
1614 terminate the quoted material, and all lines following the current
1615 line down to the terminating string are the value of the item.
1616 The terminating string may be either an identifier (a word), or
1618 some quoted text. An unquoted identifier works like double quotes.
1619 There may not be a space between the "<<" and the identifier,
1620 unless the identifier is explicitly quoted. (If you put a space it
1621 will be treated as a null identifier, which is valid, and matches
1622 the first empty line.) The terminating string must appear by
1623 itself (unquoted and with no surrounding whitespace) on the
1624 terminating line.
1625 If the terminating string is quoted, the type of quotes used
1627 determine the treatment of the text.
1628 Double Quotes
1630 Double quotes indicate that the text will be interpolated using
1631 exactly the same rules as normal double quoted strings.
1632 print <<EOF;
1634 The price is $Price.
1635 EOF
1636 print << "EOF"; # same as above
1638 The price is $Price.
1639 EOF
1640 Single Quotes
1642 Single quotes indicate the text is to be treated literally with
1643 no interpolation of its content. This is similar to single
1644 quoted strings except that backslashes have no special meaning,
1645 with "\\" being treated as two backslashes and not one as they
1646 would in every other quoting construct.
1647 This is the only form of quoting in perl where there is no need
1649 to worry about escaping content, something that code generators
1650 can and do make good use of.
1651 Backticks
1653 The content of the here doc is treated just as it would be if
1654 the string were embedded in backticks. Thus the content is
1655 interpolated as though it were double quoted and then executed
1656 via the shell, with the results of the execution returned.
1657 print << `EOC`; # execute command and get results
1659 echo hi there
1660 EOC
1661 It is possible to stack multiple here-docs in a row:
1663 print <<"foo", <<"bar"; # you can stack them
1665 I said foo.
1666 foo
1667 I said bar.
1668 bar
1669 myfunc(<< "THIS", 23, <<'THAT');
1671 Here's a line
1672 or two.
1673 THIS
1674 and here's another.
1675 THAT
1676 Just don't forget that you have to put a semicolon on the end to
1678 finish the statement, as Perl doesn't know you're not going to try
1679 to do this:
1680 print <<ABC
1682 179231
1683 ABC
1684 + 20;
1685 If you want to remove the line terminator from your here-docs, use
1687 "chomp()".
1688 chomp($string = <<'END');
1690 This is a string.
1691 END
1692 If you want your here-docs to be indented with the rest of the
1694 code, you'll need to remove leading whitespace from each line
1695 manually:
1696 ($quote = <<'FINIS') =~ s/^\s+//gm;
1698 The Road goes ever on and on,
1699 down from the door where it began.
1700 FINIS
1701 If you use a here-doc within a delimited construct, such as in
1703 "s///eg", the quoted material must come on the lines following the
1704 final delimiter. So instead of
1705 s/this/<<E . 'that'
1707 the other
1708 E
1709 . 'more '/eg;
1710 you have to write
1712 s/this/<<E . 'that'
1714 . 'more '/eg;
1715 the other
1716 E
1717 If the terminating identifier is on the last line of the program,
1719 you must be sure there is a newline after it; otherwise, Perl will
1720 give the warning Can't find string terminator "END" anywhere before
1721 EOF....
1722 Additionally, the quoting rules for the end of string identifier
1724 are not related to Perl's quoting rules. "q()", "qq()", and the
1725 like are not supported in place of '' and "", and the only
1726 interpolation is for backslashing the quoting character:
1727 print << "abc\"def";
1729 testing...
1730 abc"def
1731 Finally, quoted strings cannot span multiple lines. The general
1733 rule is that the identifier must be a string literal. Stick with
1734 that, and you should be safe.
1735 Gory details of parsing quoted constructs
1737 When presented with something that might have several different
1738 interpretations, Perl uses the DWIM (that's "Do What I Mean") principle
1739 to pick the most probable interpretation. This strategy is so
1740 successful that Perl programmers often do not suspect the ambivalence
1741 of what they write. But from time to time, Perl's notions differ
1742 substantially from what the author honestly meant.
1743 This section hopes to clarify how Perl handles quoted constructs.
1745 Although the most common reason to learn this is to unravel
1746 labyrinthine regular expressions, because the initial steps of parsing
1747 are the same for all quoting operators, they are all discussed
1748 together.
1749 The most important Perl parsing rule is the first one discussed below:
1751 when processing a quoted construct, Perl first finds the end of that
1752 construct, then interprets its contents. If you understand this rule,
1753 you may skip the rest of this section on the first reading. The other
1754 rules are likely to contradict the user's expectations much less
1755 frequently than this first one.
1756 Some passes discussed below are performed concurrently, but because
1758 their results are the same, we consider them individually. For
1759 different quoting constructs, Perl performs different numbers of
1760 passes, from one to four, but these passes are always performed in the
1761 same order.
1762 Finding the end
1764 The first pass is finding the end of the quoted construct, where
1765 the information about the delimiters is used in parsing. During
1766 this search, text between the starting and ending delimiters is
1767 copied to a safe location. The text copied gets delimiter-
1768 independent.
1769 If the construct is a here-doc, the ending delimiter is a line that
1771 has a terminating string as the content. Therefore "<<EOF" is
1772 terminated by "EOF" immediately followed by "\n" and starting from
1773 the first column of the terminating line. When searching for the
1774 terminating line of a here-doc, nothing is skipped. In other words,
1775 lines after the here-doc syntax are compared with the terminating
1776 string line by line.
1777 For the constructs except here-docs, single characters are used as
1779 starting and ending delimiters. If the starting delimiter is an
1780 opening punctuation (that is "(", "[", "{", or "<"), the ending
1781 delimiter is the corresponding closing punctuation (that is ")",
1782 "]", "}", or ">"). If the starting delimiter is an unpaired
1783 character like "/" or a closing punctuation, the ending delimiter
1784 is same as the starting delimiter. Therefore a "/" terminates a
1785 "qq//" construct, while a "]" terminates "qq[]" and "qq]]"
1786 constructs.
1787 When searching for single-character delimiters, escaped delimiters
1789 and "\\" are skipped. For example, while searching for terminating
1790 "/", combinations of "\\" and "\/" are skipped. If the delimiters
1791 are bracketing, nested pairs are also skipped. For example, while
1792 searching for closing "]" paired with the opening "[", combinations
1793 of "\\", "\]", and "\[" are all skipped, and nested "[" and "]" are
1794 skipped as well. However, when backslashes are used as the
1795 delimiters (like "qq\\" and "tr\\\"), nothing is skipped. During
1796 the search for the end, backslashes that escape delimiters are
1797 removed (exactly speaking, they are not copied to the safe
1798 location).
1799 For constructs with three-part delimiters ("s///", "y///", and
1801 "tr///"), the search is repeated once more. If the first delimiter
1802 is not an opening punctuation, three delimiters must be same such
1803 as "s!!!" and "tr)))", in which case the second delimiter
1804 terminates the left part and starts the right part at once. If the
1805 left part is delimited by bracketing punctuations (that is "()",
1806 "[]", "{}", or "<>"), the right part needs another pair of
1807 delimiters such as "s(){}" and "tr[]//". In these cases,
1808 whitespaces and comments are allowed between both parts, though the
1809 comment must follow at least one whitespace; otherwise a character
1810 expected as the start of the comment may be regarded as the
1811 starting delimiter of the right part.
1812 During this search no attention is paid to the semantics of the
1814 construct. Thus:
1815 "$hash{"$foo/$bar"}"
1817 or:
1819 m/
1821 bar # NOT a comment, this slash / terminated m//!
1822 /x
1823 do not form legal quoted expressions. The quoted part ends on the
1825 first """ and "/", and the rest happens to be a syntax error.
1826 Because the slash that terminated "m//" was followed by a "SPACE",
1827 the example above is not "m//x", but rather "m//" with no "/x"
1828 modifier. So the embedded "#" is interpreted as a literal "#".
1829 Also no attention is paid to "\c\" (multichar control char syntax)
1831 during this search. Thus the second "\" in "qq/\c\/" is interpreted
1832 as a part of "\/", and the following "/" is not recognized as a
1833 delimiter. Instead, use "\034" or "\x1c" at the end of quoted
1834 constructs.
1835 Interpolation
1837 The next step is interpolation in the text obtained, which is now
1838 delimiter-independent. There are multiple cases.
1839 "<<'EOF'"
1841 No interpolation is performed. Note that the combination "\\"
1842 is left intact, since escaped delimiters are not available for
1843 here-docs.
1844 "m''", the pattern of "s'''"
1846 No interpolation is performed at this stage. Any backslashed
1847 sequences including "\\" are treated at the stage to "parsing
1848 regular expressions".
1849 '', "q//", "tr'''", "y'''", the replacement of "s'''"
1851 The only interpolation is removal of "\" from pairs of "\\".
1852 Therefore "-" in "tr'''" and "y'''" is treated literally as a
1853 hyphen and no character range is available. "\1" in the
1854 replacement of "s'''" does not work as $1.
1855 "tr///", "y///"
1857 No variable interpolation occurs. String modifying
1858 combinations for case and quoting such as "\Q", "\U", and "\E"
1859 are not recognized. The other escape sequences such as "\200"
1860 and "\t" and backslashed characters such as "\\" and "\-" are
1861 converted to appropriate literals. The character "-" is
1862 treated specially and therefore "\-" is treated as a literal
1863 "-".
1864 "", "``", "qq//", "qx//", "<file*glob>", "<<"EOF""
1866 "\Q", "\U", "\u", "\L", "\l" (possibly paired with "\E") are
1867 converted to corresponding Perl constructs. Thus,
1868 "$foo\Qbaz$bar" is converted to "$foo . (quotemeta("baz" .
1869 $bar))" internally. The other escape sequences such as "\200"
1870 and "\t" and backslashed characters such as "\\" and "\-" are
1871 replaced with appropriate expansions.
1872 Let it be stressed that whatever falls between "\Q" and "\E" is
1874 interpolated in the usual way. Something like "\Q\\E" has no
1875 "\E" inside. instead, it has "\Q", "\\", and "E", so the
1876 result is the same as for "\\\\E". As a general rule,
1877 backslashes between "\Q" and "\E" may lead to counterintuitive
1878 results. So, "\Q\t\E" is converted to "quotemeta("\t")", which
1879 is the same as "\\\t" (since TAB is not alphanumeric). Note
1880 also that:
1881 $str = '\t';
1883 return "\Q$str";
1884 may be closer to the conjectural intention of the writer of
1886 "\Q\t\E".
1887 Interpolated scalars and arrays are converted internally to the
1889 "join" and "." catenation operations. Thus, "$foo XXX '@arr'"
1890 becomes:
1891 $foo . " XXX '" . (join $", @arr) . "'";
1893 All operations above are performed simultaneously, left to
1895 right.
1896 Because the result of "\Q STRING \E" has all metacharacters
1898 quoted, there is no way to insert a literal "$" or "@" inside a
1899 "\Q\E" pair. If protected by "\", "$" will be quoted to became
1900 "\\\$"; if not, it is interpreted as the start of an
1901 interpolated scalar.
1902 Note also that the interpolation code needs to make a decision
1904 on where the interpolated scalar ends. For instance, whether
1905 "a $b -> {c}" really means:
1906 "a " . $b . " -> {c}";
1908 or:
1910 "a " . $b -> {c};
1912 Most of the time, the longest possible text that does not
1914 include spaces between components and which contains matching
1915 braces or brackets. because the outcome may be determined by
1916 voting based on heuristic estimators, the result is not
1917 strictly predictable. Fortunately, it's usually correct for
1918 ambiguous cases.
1919 the replacement of "s///"
1921 Processing of "\Q", "\U", "\u", "\L", "\l", and interpolation
1922 happens as with "qq//" constructs.
1923 It is at this step that "\1" is begrudgingly converted to $1 in
1925 the replacement text of "s///", in order to correct the
1926 incorrigible sed hackers who haven't picked up the saner idiom
1927 yet. A warning is emitted if the "use warnings" pragma or the
1928 -w command-line flag (that is, the $^W variable) was set.
1929 "RE" in "?RE?", "/RE/", "m/RE/", "s/RE/foo/",
1931 Processing of "\Q", "\U", "\u", "\L", "\l", "\E", and
1932 interpolation happens (almost) as with "qq//" constructs.
1933 Processing of "\N{...}" is also done here, and compiled into an
1935 intermediate form for the regex compiler. (This is because, as
1936 mentioned below, the regex compilation may be done at execution
1937 time, and "\N{...}" is a compile-time construct.)
1938 However any other combinations of "\" followed by a character
1940 are not substituted but only skipped, in order to parse them as
1941 regular expressions at the following step. As "\c" is skipped
1942 at this step, "@" of "\c@" in RE is possibly treated as an
1943 array symbol (for example @foo), even though the same text in
1944 "qq//" gives interpolation of "\c@".
1945 Moreover, inside "(?{BLOCK})", "(?# comment )", and a
1947 "#"-comment in a "//x"-regular expression, no processing is
1948 performed whatsoever. This is the first step at which the
1949 presence of the "//x" modifier is relevant.
1950 Interpolation in patterns has several quirks: $|, $(, $), "@+"
1952 and "@-" are not interpolated, and constructs $var[SOMETHING]
1953 are voted (by several different estimators) to be either an
1954 array element or $var followed by an RE alternative. This is
1955 where the notation "${arr[$bar]}" comes handy: "/${arr[0-9]}/"
1956 is interpreted as array element "-9", not as a regular
1957 expression from the variable $arr followed by a digit, which
1958 would be the interpretation of "/$arr[0-9]/". Since voting
1959 among different estimators may occur, the result is not
1960 predictable.
1961 The lack of processing of "\\" creates specific restrictions on
1963 the post-processed text. If the delimiter is "/", one cannot
1964 get the combination "\/" into the result of this step. "/"
1965 will finish the regular expression, "\/" will be stripped to
1966 "/" on the previous step, and "\\/" will be left as is.
1967 Because "/" is equivalent to "\/" inside a regular expression,
1968 this does not matter unless the delimiter happens to be
1969 character special to the RE engine, such as in "s*foo*bar*",
1970 "m[foo]", or "?foo?"; or an alphanumeric char, as in:
1971 m m ^ a \s* b mmx;
1973 In the RE above, which is intentionally obfuscated for
1975 illustration, the delimiter is "m", the modifier is "mx", and
1976 after delimiter-removal the RE is the same as for "m/ ^ a \s* b
1977 /mx". There's more than one reason you're encouraged to
1978 restrict your delimiters to non-alphanumeric, non-whitespace
1979 choices.
1980 This step is the last one for all constructs except regular
1982 expressions, which are processed further.
1983 parsing regular expressions
1985 Previous steps were performed during the compilation of Perl code,
1986 but this one happens at run time, although it may be optimized to
1987 be calculated at compile time if appropriate. After preprocessing
1988 described above, and possibly after evaluation if concatenation,
1989 joining, casing translation, or metaquoting are involved, the
1990 resulting string is passed to the RE engine for compilation.
1991 Whatever happens in the RE engine might be better discussed in
1993 perlre, but for the sake of continuity, we shall do so here.
1994 This is another step where the presence of the "//x" modifier is
1996 relevant. The RE engine scans the string from left to right and
1997 converts it to a finite automaton.
1998 Backslashed characters are either replaced with corresponding
2000 literal strings (as with "\{"), or else they generate special nodes
2001 in the finite automaton (as with "\b"). Characters special to the
2002 RE engine (such as "|") generate corresponding nodes or groups of
2003 nodes. "(?#...)" comments are ignored. All the rest is either
2004 converted to literal strings to match, or else is ignored (as is
2005 whitespace and "#"-style comments if "//x" is present).
2006 Parsing of the bracketed character class construct, "[...]", is
2008 rather different than the rule used for the rest of the pattern.
2009 The terminator of this construct is found using the same rules as
2010 for finding the terminator of a "{}"-delimited construct, the only
2011 exception being that "]" immediately following "[" is treated as
2012 though preceded by a backslash. Similarly, the terminator of
2013 "(?{...})" is found using the same rules as for finding the
2014 terminator of a "{}"-delimited construct.
2015 It is possible to inspect both the string given to RE engine and
2017 the resulting finite automaton. See the arguments
2018 "debug"/"debugcolor" in the "use re" pragma, as well as Perl's -Dr
2019 command-line switch documented in "Command Switches" in perlrun.
2020 Optimization of regular expressions
2022 This step is listed for completeness only. Since it does not
2023 change semantics, details of this step are not documented and are
2024 subject to change without notice. This step is performed over the
2025 finite automaton that was generated during the previous pass.
2026 It is at this stage that "split()" silently optimizes "/^/" to mean
2028 "/^/m".
2029 I/O Operators
2031 There are several I/O operators you should know about.
2032 A string enclosed by backticks (grave accents) first undergoes double-
2034 quote interpolation. It is then interpreted as an external command,
2035 and the output of that command is the value of the backtick string,
2036 like in a shell. In scalar context, a single string consisting of all
2037 output is returned. In list context, a list of values is returned, one
2038 per line of output. (You can set $/ to use a different line
2039 terminator.) The command is executed each time the pseudo-literal is
2040 evaluated. The status value of the command is returned in $? (see
2041 perlvar for the interpretation of $?). Unlike in csh, no translation
2042 is done on the return data--newlines remain newlines. Unlike in any of
2043 the shells, single quotes do not hide variable names in the command
2044 from interpretation. To pass a literal dollar-sign through to the
2045 shell you need to hide it with a backslash. The generalized form of
2046 backticks is "qx//". (Because backticks always undergo shell expansion
2047 as well, see perlsec for security concerns.)
2048 In scalar context, evaluating a filehandle in angle brackets yields the
2050 next line from that file (the newline, if any, included), or "undef" at
2051 end-of-file or on error. When $/ is set to "undef" (sometimes known as
2052 file-slurp mode) and the file is empty, it returns '' the first time,
2053 followed by "undef" subsequently.
2054 Ordinarily you must assign the returned value to a variable, but there
2056 is one situation where an automatic assignment happens. If and only if
2057 the input symbol is the only thing inside the conditional of a "while"
2058 statement (even if disguised as a "for(;;)" loop), the value is
2059 automatically assigned to the global variable $_, destroying whatever
2060 was there previously. (This may seem like an odd thing to you, but
2061 you'll use the construct in almost every Perl script you write.) The
2062 $_ variable is not implicitly localized. You'll have to put a "local
2063 $_;" before the loop if you want that to happen.
2064 The following lines are equivalent:
2066 while (defined($_ = <STDIN>)) { print; }
2068 while ($_ = <STDIN>) { print; }
2069 while (<STDIN>) { print; }
2070 for (;<STDIN>;) { print; }
2071 print while defined($_ = <STDIN>);
2072 print while ($_ = <STDIN>);
2073 print while <STDIN>;
2074 This also behaves similarly, but avoids $_ :
2076 while (my $line = <STDIN>) { print $line }
2078 In these loop constructs, the assigned value (whether assignment is
2080 automatic or explicit) is then tested to see whether it is defined.
2081 The defined test avoids problems where line has a string value that
2082 would be treated as false by Perl, for example a "" or a "0" with no
2083 trailing newline. If you really mean for such values to terminate the
2084 loop, they should be tested for explicitly:
2085 while (($_ = <STDIN>) ne '0') { ... }
2087 while (<STDIN>) { last unless $_; ... }
2088 In other boolean contexts, "<filehandle>" without an explicit "defined"
2090 test or comparison elicits a warning if the "use warnings" pragma or
2091 the -w command-line switch (the $^W variable) is in effect.
2092 The filehandles STDIN, STDOUT, and STDERR are predefined. (The
2094 filehandles "stdin", "stdout", and "stderr" will also work except in
2095 packages, where they would be interpreted as local identifiers rather
2096 than global.) Additional filehandles may be created with the open()
2097 function, amongst others. See perlopentut and "open" in perlfunc for
2098 details on this.
2099 If a <FILEHANDLE> is used in a context that is looking for a list, a
2101 list comprising all input lines is returned, one line per list element.
2102 It's easy to grow to a rather large data space this way, so use with
2103 care.
2104 <FILEHANDLE> may also be spelled "readline(*FILEHANDLE)". See
2106 "readline" in perlfunc.
2107 The null filehandle <> is special: it can be used to emulate the
2109 behavior of sed and awk. Input from <> comes either from standard
2110 input, or from each file listed on the command line. Here's how it
2111 works: the first time <> is evaluated, the @ARGV array is checked, and
2112 if it is empty, $ARGV[0] is set to "-", which when opened gives you
2113 standard input. The @ARGV array is then processed as a list of
2114 filenames. The loop
2115 while (<>) {
2117 ... # code for each line
2118 }
2119 is equivalent to the following Perl-like pseudo code:
2121 unshift(@ARGV, '-') unless @ARGV;
2123 while ($ARGV = shift) {
2124 open(ARGV, $ARGV);
2125 while (<ARGV>) {
2126 ... # code for each line
2127 }
2128 }
2129 except that it isn't so cumbersome to say, and will actually work. It
2131 really does shift the @ARGV array and put the current filename into the
2132 $ARGV variable. It also uses filehandle ARGV internally. <> is just a
2133 synonym for <ARGV>, which is magical. (The pseudo code above doesn't
2134 work because it treats <ARGV> as non-magical.)
2135 Since the null filehandle uses the two argument form of "open" in
2137 perlfunc it interprets special characters, so if you have a script like
2138 this:
2139 while (<>) {
2141 print;
2142 }
2143 and call it with "perl dangerous.pl 'rm -rfv *|'", it actually opens a
2145 pipe, executes the "rm" command and reads "rm"'s output from that pipe.
2146 If you want all items in @ARGV to be interpreted as file names, you can
2147 use the module "ARGV::readonly" from CPAN.
2148 You can modify @ARGV before the first <> as long as the array ends up
2150 containing the list of filenames you really want. Line numbers ($.)
2151 continue as though the input were one big happy file. See the example
2152 in "eof" in perlfunc for how to reset line numbers on each file.
2153 If you want to set @ARGV to your own list of files, go right ahead.
2155 This sets @ARGV to all plain text files if no @ARGV was given:
2156 @ARGV = grep { -f && -T } glob('*') unless @ARGV;
2158 You can even set them to pipe commands. For example, this
2160 automatically filters compressed arguments through gzip:
2161 @ARGV = map { /\.(gz|Z)$/ ? "gzip -dc < $_ |" : $_ } @ARGV;
2163 If you want to pass switches into your script, you can use one of the
2165 Getopts modules or put a loop on the front like this:
2166 while ($_ = $ARGV[0], /^-/) {
2168 shift;
2169 last if /^--$/;
2170 if (/^-D(.*)/) { $debug = $1 }
2171 if (/^-v/) { $verbose++ }
2172 # ... # other switches
2173 }
2174 while (<>) {
2176 # ... # code for each line
2177 }
2178 The <> symbol will return "undef" for end-of-file only once. If you
2180 call it again after this, it will assume you are processing another
2181 @ARGV list, and if you haven't set @ARGV, will read input from STDIN.
2182 If what the angle brackets contain is a simple scalar variable (e.g.,
2184 <$foo>), then that variable contains the name of the filehandle to
2185 input from, or its typeglob, or a reference to the same. For example:
2186 $fh = \*STDIN;
2188 $line = <$fh>;
2189 If what's within the angle brackets is neither a filehandle nor a
2191 simple scalar variable containing a filehandle name, typeglob, or
2192 typeglob reference, it is interpreted as a filename pattern to be
2193 globbed, and either a list of filenames or the next filename in the
2194 list is returned, depending on context. This distinction is determined
2195 on syntactic grounds alone. That means "<$x>" is always a readline()
2196 from an indirect handle, but "<$hash{key}>" is always a glob(). That's
2197 because $x is a simple scalar variable, but $hash{key} is not--it's a
2198 hash element. Even "<$x >" (note the extra space) is treated as
2199 "glob("$x ")", not "readline($x)".
2200 One level of double-quote interpretation is done first, but you can't
2202 say "<$foo>" because that's an indirect filehandle as explained in the
2203 previous paragraph. (In older versions of Perl, programmers would
2204 insert curly brackets to force interpretation as a filename glob:
2205 "<${foo}>". These days, it's considered cleaner to call the internal
2206 function directly as "glob($foo)", which is probably the right way to
2207 have done it in the first place.) For example:
2208 while (<*.c>) {
2210 chmod 0644, $_;
2211 }
2212 is roughly equivalent to:
2214 open(FOO, "echo *.c | tr -s ' \t\r\f' '\\012\\012\\012\\012'|");
2216 while (<FOO>) {
2217 chomp;
2218 chmod 0644, $_;
2219 }
2220 except that the globbing is actually done internally using the standard
2222 "File::Glob" extension. Of course, the shortest way to do the above
2223 is:
2224 chmod 0644, <*.c>;
2226 A (file)glob evaluates its (embedded) argument only when it is starting
2228 a new list. All values must be read before it will start over. In
2229 list context, this isn't important because you automatically get them
2230 all anyway. However, in scalar context the operator returns the next
2231 value each time it's called, or "undef" when the list has run out. As
2232 with filehandle reads, an automatic "defined" is generated when the
2233 glob occurs in the test part of a "while", because legal glob returns
2234 (e.g. a file called 0) would otherwise terminate the loop. Again,
2235 "undef" is returned only once. So if you're expecting a single value
2236 from a glob, it is much better to say
2237 ($file) = <blurch*>;
2239 than
2241 $file = <blurch*>;
2243 because the latter will alternate between returning a filename and
2245 returning false.
2246 If you're trying to do variable interpolation, it's definitely better
2248 to use the glob() function, because the older notation can cause people
2249 to become confused with the indirect filehandle notation.
2250 @files = glob("$dir/*.[ch]");
2252 @files = glob($files[$i]);
2253 Constant Folding
2255 Like C, Perl does a certain amount of expression evaluation at compile
2256 time whenever it determines that all arguments to an operator are
2257 static and have no side effects. In particular, string concatenation
2258 happens at compile time between literals that don't do variable
2259 substitution. Backslash interpolation also happens at compile time.
2260 You can say
2261 'Now is the time for all' . "\n" .
2263 'good men to come to.'
2264 and this all reduces to one string internally. Likewise, if you say
2266 foreach $file (@filenames) {
2268 if (-s $file > 5 + 100 * 2**16) { }
2269 }
2270 the compiler will precompute the number which that expression
2272 represents so that the interpreter won't have to.
2273 No-ops
2275 Perl doesn't officially have a no-op operator, but the bare constants 0
2276 and 1 are special-cased to not produce a warning in a void context, so
2277 you can for example safely do
2278 1 while foo();
2280 Bitwise String Operators
2282 Bitstrings of any size may be manipulated by the bitwise operators ("~
2283 | & ^").
2284 If the operands to a binary bitwise op are strings of different sizes,
2286 | and ^ ops act as though the shorter operand had additional zero bits
2287 on the right, while the & op acts as though the longer operand were
2288 truncated to the length of the shorter. The granularity for such
2289 extension or truncation is one or more bytes.
2290 # ASCII-based examples
2292 print "j p \n" ^ " a h"; # prints "JAPH\n"
2293 print "JA" | " ph\n"; # prints "japh\n"
2294 print "japh\nJunk" & '_____'; # prints "JAPH\n";
2295 print 'p N$' ^ " E<H\n"; # prints "Perl\n";
2296 If you are intending to manipulate bitstrings, be certain that you're
2298 supplying bitstrings: If an operand is a number, that will imply a
2299 numeric bitwise operation. You may explicitly show which type of
2300 operation you intend by using "" or "0+", as in the examples below.
2301 $foo = 150 | 105; # yields 255 (0x96 | 0x69 is 0xFF)
2303 $foo = '150' | 105; # yields 255
2304 $foo = 150 | '105'; # yields 255
2305 $foo = '150' | '105'; # yields string '155' (under ASCII)
2306 $baz = 0+$foo & 0+$bar; # both ops explicitly numeric
2308 $biz = "$foo" ^ "$bar"; # both ops explicitly stringy
2309 See "vec" in perlfunc for information on how to manipulate individual
2311 bits in a bit vector.
2312 Integer Arithmetic
2314 By default, Perl assumes that it must do most of its arithmetic in
2315 floating point. But by saying
2316 use integer;
2318 you may tell the compiler that it's okay to use integer operations (if
2320 it feels like it) from here to the end of the enclosing BLOCK. An
2321 inner BLOCK may countermand this by saying
2322 no integer;
2324 which lasts until the end of that BLOCK. Note that this doesn't mean
2326 everything is only an integer, merely that Perl may use integer
2327 operations if it is so inclined. For example, even under "use
2328 integer", if you take the sqrt(2), you'll still get 1.4142135623731 or
2329 so.
2330 Used on numbers, the bitwise operators ("&", "|", "^", "~", "<<", and
2332 ">>") always produce integral results. (But see also "Bitwise String
2333 Operators".) However, "use integer" still has meaning for them. By
2334 default, their results are interpreted as unsigned integers, but if
2335 "use integer" is in effect, their results are interpreted as signed
2336 integers. For example, "~0" usually evaluates to a large integral
2337 value. However, "use integer; ~0" is "-1" on two's-complement
2338 machines.
2339 Floating-point Arithmetic
2341 While "use integer" provides integer-only arithmetic, there is no
2342 analogous mechanism to provide automatic rounding or truncation to a
2343 certain number of decimal places. For rounding to a certain number of
2344 digits, sprintf() or printf() is usually the easiest route. See
2345 perlfaq4.
2346 Floating-point numbers are only approximations to what a mathematician
2348 would call real numbers. There are infinitely more reals than floats,
2349 so some corners must be cut. For example:
2350 printf "%.20g\n", 123456789123456789;
2352 # produces 123456789123456784
2353 Testing for exact floating-point equality or inequality is not a good
2355 idea. Here's a (relatively expensive) work-around to compare whether
2356 two floating-point numbers are equal to a particular number of decimal
2357 places. See Knuth, volume II, for a more robust treatment of this
2358 topic.
2359 sub fp_equal {
2361 my ($X, $Y, $POINTS) = @_;
2362 my ($tX, $tY);
2363 $tX = sprintf("%.${POINTS}g", $X);
2364 $tY = sprintf("%.${POINTS}g", $Y);
2365 return $tX eq $tY;
2366 }
2367 The POSIX module (part of the standard perl distribution) implements
2369 ceil(), floor(), and other mathematical and trigonometric functions.
2370 The Math::Complex module (part of the standard perl distribution)
2371 defines mathematical functions that work on both the reals and the
2372 imaginary numbers. Math::Complex not as efficient as POSIX, but POSIX
2373 can't work with complex numbers.
2374 Rounding in financial applications can have serious implications, and
2376 the rounding method used should be specified precisely. In these
2377 cases, it probably pays not to trust whichever system rounding is being
2378 used by Perl, but to instead implement the rounding function you need
2379 yourself.
2380 Bigger Numbers
2382 The standard Math::BigInt and Math::BigFloat modules provide variable-
2383 precision arithmetic and overloaded operators, although they're
2384 currently pretty slow. At the cost of some space and considerable
2385 speed, they avoid the normal pitfalls associated with limited-precision
2386 representations.
2387 use Math::BigInt;
2389 $x = Math::BigInt->new('123456789123456789');
2390 print $x * $x;
2391 # prints +15241578780673678515622620750190521
2393 There are several modules that let you calculate with (bound only by
2395 memory and cpu-time) unlimited or fixed precision. There are also some
2396 non-standard modules that provide faster implementations via external C
2397 libraries.
2398 Here is a short, but incomplete summary:
2400 Math::Fraction big, unlimited fractions like 9973 / 12967
2402 Math::String treat string sequences like numbers
2403 Math::FixedPrecision calculate with a fixed precision
2404 Math::Currency for currency calculations
2405 Bit::Vector manipulate bit vectors fast (uses C)
2406 Math::BigIntFast Bit::Vector wrapper for big numbers
2407 Math::Pari provides access to the Pari C library
2408 Math::BigInteger uses an external C library
2409 Math::Cephes uses external Cephes C library (no big numbers)
2410 Math::Cephes::Fraction fractions via the Cephes library
2411 Math::GMP another one using an external C library
2412 Choose wisely.
2414perl v5.12.4 2011-06-07 PERLOP(1)