{ "content": [ { "type": "text", "text": "# perlop (man)\n\n## NAME\n\nperlop - Perl operators and precedence\n\n## DESCRIPTION\n\nIn Perl, the operator determines what operation is performed, independent of the type of the\noperands. For example \"$x+$y\" is always a numeric addition, and if $x or $y do not contain\nnumbers, an attempt is made to convert them to numbers first.\n\n## Sections\n\n- **NAME**\n- **DESCRIPTION**\n\nUse structuredContent.sections for detailed options, examples, and full documentation.\n" } ], "structuredContent": { "command": "perlop", "section": "", "mode": "man", "summary": "perlop - Perl operators and precedence", "synopsis": null, "tldr_summary": null, "tldr_examples": [], "tldr_source": null, "flags": [], "examples": [], "see_also": [], "section_outline": [ { "name": "NAME", "lines": 2, "subsections": [] }, { "name": "DESCRIPTION", "lines": 3145, "subsections": [] } ], "sections": { "NAME": { "content": "perlop - Perl operators and precedence\n", "subsections": [] }, "DESCRIPTION": { "content": "In Perl, the operator determines what operation is performed, independent of the type of the\noperands. For example \"$x+$y\" is always a numeric addition, and if $x or $y do not contain\nnumbers, an attempt is made to convert them to numbers first.\n\nThis is in contrast to many other dynamic languages, where the operation is determined by the\ntype of the first argument. It also means that Perl has two versions of some operators, one\nfor numeric and one for string comparison. For example \"$x==$y\" compares two numbers for\nequality, and \"$xeq$y\" compares two strings.\n\nThere are a few exceptions though: \"x\" can be either string repetition or list repetition,\ndepending on the type of the left operand, and \"&\", \"|\", \"^\" and \"~\" can be either string or\nnumeric bit operations.\n\nOperator Precedence and Associativity\nOperator precedence and associativity work in Perl more or less like they do in mathematics.\n\nOperator precedence means some operators group more tightly than others. For example, in \"2\n+ 4 * 5\", the multiplication has higher precedence, so \"4 * 5\" is grouped together as the\nright-hand operand of the addition, rather than \"2 + 4\" being grouped together as the left-\nhand operand of the multiplication. It is as if the expression were written \"2 + (4 * 5)\",\nnot \"(2 + 4) * 5\". So the expression yields \"2 + 20 == 22\", rather than \"6 * 5 == 30\".\n\nOperator associativity defines what happens if a sequence of the same operators is used one\nafter another: usually that they will be grouped at the left or the right. For example, in \"9\n- 3 - 2\", subtraction is left associative, so \"9 - 3\" is grouped together as the left-hand\noperand of the second subtraction, rather than \"3 - 2\" being grouped together as the right-\nhand operand of the first subtraction. It is as if the expression were written \"(9 - 3) - 2\",\nnot \"9 - (3 - 2)\". So the expression yields \"6 - 2 == 4\", rather than \"9 - 1 == 8\".\n\nFor simple operators that evaluate all their operands and then combine the values in some\nway, precedence and associativity (and parentheses) imply some ordering requirements on those\ncombining operations. For example, in \"2 + 4 * 5\", the grouping implied by precedence means\nthat the multiplication of 4 and 5 must be performed before the addition of 2 and 20, simply\nbecause the result of that multiplication is required as one of the operands of the addition.\nBut the order of operations is not fully determined by this: in \"2 * 2 + 4 * 5\" both\nmultiplications must be performed before the addition, but the grouping does not say anything\nabout the order in which the two multiplications are performed. In fact Perl has a general\nrule that the operands of an operator are evaluated in left-to-right order. A few operators\nsuch as \"&&=\" have special evaluation rules that can result in an operand not being evaluated\nat all; in general, the top-level operator in an expression has control of operand\nevaluation.\n\nSome comparison operators, as their associativity, chain with some operators of the same\nprecedence (but never with operators of different precedence). This chaining means that each\ncomparison is performed on the two arguments surrounding it, with each interior argument\ntaking part in two comparisons, and the comparison results are implicitly ANDed. Thus\n\"$x<$y<=$z\" behaves exactly like \"$x<$y&&$y<=$z\", assuming that \"$y\" is as simple a scalar as\nit looks. The ANDing short-circuits just like \"&&\" does, stopping the sequence of\ncomparisons as soon as one yields false.\n\nIn a chained comparison, each argument expression is evaluated at most once, even if it takes\npart in two comparisons, but the result of the evaluation is fetched for each comparison.\n(It is not evaluated at all if the short-circuiting means that it's not required for any\ncomparisons.) This matters if the computation of an interior argument is expensive or non-\ndeterministic. For example,\n\nif($x < expensivesub() <= $z) { ...\n\nis not entirely like\n\nif($x < expensivesub() && expensivesub() <= $z) { ...\n\nbut instead closer to\n\nmy $tmp = expensivesub();\nif($x < $tmp && $tmp <= $z) { ...\n\nin that the subroutine is only called once. However, it's not exactly like this latter code\neither, because the chained comparison doesn't actually involve any temporary variable (named\nor otherwise): there is no assignment. This doesn't make much difference where the\nexpression is a call to an ordinary subroutine, but matters more with an lvalue subroutine,\nor if the argument expression yields some unusual kind of scalar by other means. For\nexample, if the argument expression yields a tied scalar, then the expression is evaluated to\nproduce that scalar at most once, but the value of that scalar may be fetched up to twice,\nonce for each comparison in which it is actually used.\n\nIn this example, the expression is evaluated only once, and the tied scalar (the result of\nthe expression) is fetched for each comparison that uses it.\n\nif ($x < $tiedscalar < $z) { ...\n\nIn the next example, the expression is evaluated only once, and the tied scalar is fetched\nonce as part of the operation within the expression. The result of that operation is fetched\nfor each comparison, which normally doesn't matter unless that expression result is also\nmagical due to operator overloading.\n\nif ($x < $tiedscalar + 42 < $z) { ...\n\nSome operators are instead non-associative, meaning that it is a syntax error to use a\nsequence of those operators of the same precedence. For example, \"$x..$y..$z\" is an error.\n\nPerl operators have the following associativity and precedence, listed from highest\nprecedence to lowest. Operators borrowed from C keep the same precedence relationship with\neach other, even where C's precedence is slightly screwy. (This makes learning Perl easier\nfor C folks.) With very few exceptions, these all operate on scalar values only, not array\nvalues.\n\nleft terms and list operators (leftward)\nleft ->\nnonassoc ++ --\nright\nright ! ~ ~. \\ and unary + and -\nleft =~ !~\nleft * / % x\nleft + - .\nleft << >>\nnonassoc named unary operators\nnonassoc isa\nchained < > <= >= lt gt le ge\nchain/na == != eq ne <=> cmp ~~\nleft & &.\nleft | |. ^ ^.\nleft &&\nleft || //\nnonassoc .. ...\nright ?:\nright = += -= *= etc. goto last next redo dump\nleft , =>\nnonassoc list operators (rightward)\nright not\nleft and\nleft or xor\n\nIn the following sections, these operators are covered in detail, in the same order in which\nthey appear in the table above.\n\nMany operators can be overloaded for objects. See overload.\n\nTerms and List Operators (Leftward)\nA TERM has the highest precedence in Perl. They include variables, quote and quote-like\noperators, any expression in parentheses, and any function whose arguments are parenthesized.\nActually, there aren't really functions in this sense, just list operators and unary\noperators behaving as functions because you put parentheses around the arguments. These are\nall documented in perlfunc.\n\nIf any list operator (\"print()\", etc.) or any unary operator (\"chdir()\", etc.) is followed\nby a left parenthesis as the next token, the operator and arguments within parentheses are\ntaken to be of highest precedence, just like a normal function call.\n\nIn the absence of parentheses, the precedence of list operators such as \"print\", \"sort\", or\n\"chmod\" is either very high or very low depending on whether you are looking at the left side\nor the right side of the operator. For example, in\n\n@ary = (1, 3, sort 4, 2);\nprint @ary; # prints 1324\n\nthe commas on the right of the \"sort\" are evaluated before the \"sort\", but the commas on the\nleft are evaluated after. In other words, list operators tend to gobble up all arguments\nthat follow, and then act like a simple TERM with regard to the preceding expression. Be\ncareful with parentheses:\n\n# These evaluate exit before doing the print:\nprint($foo, exit); # Obviously not what you want.\nprint $foo, exit; # Nor is this.\n\n# These do the print before evaluating exit:\n(print $foo), exit; # This is what you want.\nprint($foo), exit; # Or this.\nprint ($foo), exit; # Or even this.\n\nAlso note that\n\nprint ($foo & 255) + 1, \"\\n\";\n\nprobably doesn't do what you expect at first glance. The parentheses enclose the argument\nlist for \"print\" which is evaluated (printing the result of \"$foo&255\"). Then one is added\nto the return value of \"print\" (usually 1). The result is something like this:\n\n1 + 1, \"\\n\"; # Obviously not what you meant.\n\nTo do what you meant properly, you must write:\n\nprint(($foo & 255) + 1, \"\\n\");\n\nSee \"Named Unary Operators\" for more discussion of this.\n\nAlso parsed as terms are the \"do{}\" and \"eval{}\" constructs, as well as subroutine and method\ncalls, and the anonymous constructors \"[]\" and \"{}\".\n\nSee also \"Quote and Quote-like Operators\" toward the end of this section, as well as \"I/O\nOperators\".\n\nThe Arrow Operator\n\"\"->\"\" is an infix dereference operator, just as it is in C and C++. If the right side is\neither a \"[...]\", \"{...}\", or a \"(...)\" subscript, then the left side must be either a hard\nor symbolic reference to an array, a hash, or a subroutine respectively. (Or technically\nspeaking, a location capable of holding a hard reference, if it's an array or hash reference\nbeing used for assignment.) See perlreftut and perlref.\n\nOtherwise, the right side is a method name or a simple scalar variable containing either the\nmethod name or a subroutine reference, and the left side must be either an object (a blessed\nreference) or a class name (that is, a package name). See perlobj.\n\nThe dereferencing cases (as opposed to method-calling cases) are somewhat extended by the\n\"postderef\" feature. For the details of that feature, consult \"Postfix Dereference Syntax\"\nin perlref.\n\nAuto-increment and Auto-decrement\n\"++\" and \"--\" work as in C. That is, if placed before a variable, they increment or\ndecrement the variable by one before returning the value, and if placed after, increment or\ndecrement after returning the value.\n\n$i = 0; $j = 0;\nprint $i++; # prints 0\nprint ++$j; # prints 1\n\nNote that just as in C, Perl doesn't define when the variable is incremented or decremented.\nYou just know it will be done sometime before or after the value is returned. This also\nmeans that modifying a variable twice in the same statement will lead to undefined behavior.\nAvoid statements like:\n\n$i = $i ++;\nprint ++ $i + $i ++;\n\nPerl will not guarantee what the result of the above statements is.\n\nThe auto-increment operator has a little extra builtin magic to it. If you increment a\nvariable that is numeric, or that has ever been used in a numeric context, you get a normal\nincrement. If, however, the variable has been used in only string contexts since it was set,\nand has a value that is not the empty string and matches the pattern \"/^[a-zA-Z]*[0-9]*\\z/\",\nthe increment is done as a string, preserving each character within its range, with carry:\n\nprint ++($foo = \"99\"); # prints \"100\"\nprint ++($foo = \"a0\"); # prints \"a1\"\nprint ++($foo = \"Az\"); # prints \"Ba\"\nprint ++($foo = \"zz\"); # prints \"aaa\"\n\n\"undef\" is always treated as numeric, and in particular is changed to 0 before incrementing\n(so that a post-increment of an undef value will return 0 rather than \"undef\").\n\nThe auto-decrement operator is not magical.\n\nExponentiation\nBinary \"\" is the exponentiation operator. It binds even more tightly than unary minus, so\n\"-24\" is \"-(24)\", not \"(-2)4\". (This is implemented using C's pow(3) function, which\nactually works on doubles internally.)\n\nNote that certain exponentiation expressions are ill-defined: these include \"00\", \"1Inf\",\nand \"Inf0\". Do not expect any particular results from these special cases, the results are\nplatform-dependent.\n\nSymbolic Unary Operators\nUnary \"!\" performs logical negation, that is, \"not\". See also \"not\" for a lower precedence\nversion of this.\n\nUnary \"-\" performs arithmetic negation if the operand is numeric, including any string that\nlooks like a number. If the operand is an identifier, a string consisting of a minus sign\nconcatenated with the identifier is returned. Otherwise, if the string starts with a plus or\nminus, a string starting with the opposite sign is returned. One effect of these rules is\nthat \"-bareword\" is equivalent to the string \"-bareword\". If, however, the string begins\nwith a non-alphabetic character (excluding \"+\" or \"-\"), Perl will attempt to convert the\nstring to a numeric, and the arithmetic negation is performed. If the string cannot be\ncleanly converted to a numeric, Perl will give the warning Argument \"the string\" isn't\nnumeric in negation (-) at ....\n\nUnary \"~\" performs bitwise negation, that is, 1's complement. For example, \"0666&~027\" is\n0640. (See also \"Integer Arithmetic\" and \"Bitwise String Operators\".) Note that the width\nof the result is platform-dependent: \"~0\" is 32 bits wide on a 32-bit platform, but 64 bits\nwide on a 64-bit platform, so if you are expecting a certain bit width, remember to use the\n\"&\" operator to mask off the excess bits.\n\nStarting in Perl 5.28, it is a fatal error to try to complement a string containing a\ncharacter with an ordinal value above 255.\n\nIf the \"bitwise\" feature is enabled via \"usefeature'bitwise'\" or \"use v5.28\", then unary \"~\"\nalways treats its argument as a number, and an alternate form of the operator, \"~.\", always\ntreats its argument as a string. So \"~0\" and \"~\"0\"\" will both give 232-1 on 32-bit\nplatforms, whereas \"~.0\" and \"~.\"0\"\" will both yield \"\\xff\". Until Perl 5.28, this feature\nproduced a warning in the \"experimental::bitwise\" category.\n\nUnary \"+\" has no effect whatsoever, even on strings. It is useful syntactically for\nseparating a function name from a parenthesized expression that would otherwise be\ninterpreted as the complete list of function arguments. (See examples above under \"Terms and\nList Operators (Leftward)\".)\n\nUnary \"\\\" creates references. If its operand is a single sigilled thing, it creates a\nreference to that object. If its operand is a parenthesised list, then it creates references\nto the things mentioned in the list. Otherwise it puts its operand in list context, and\ncreates a list of references to the scalars in the list provided by the operand. See\nperlreftut and perlref. Do not confuse this behavior with the behavior of backslash within a\nstring, although both forms do convey the notion of protecting the next thing from\ninterpolation.\n\nBinding Operators\nBinary \"=~\" binds a scalar expression to a pattern match. Certain operations search or\nmodify the string $ by default. This operator makes that kind of operation work on some\nother string. The right argument is a search pattern, substitution, or transliteration. The\nleft argument is what is supposed to be searched, substituted, or transliterated instead of\nthe default $. When used in scalar context, the return value generally indicates the\nsuccess of the operation. The exceptions are substitution (\"s///\") and transliteration\n(\"y///\") with the \"/r\" (non-destructive) option, which cause the return value to be the\nresult of the substitution. Behavior in list context depends on the particular operator.\nSee \"Regexp Quote-Like Operators\" for details and perlretut for examples using these\noperators.\n\nIf the right argument is an expression rather than a search pattern, substitution, or\ntransliteration, it is interpreted as a search pattern at run time. Note that this means\nthat its contents will be interpolated twice, so\n\n'\\\\' =~ q'\\\\';\n\nis not ok, as the regex engine will end up trying to compile the pattern \"\\\", which it will\nconsider a syntax error.\n\nBinary \"!~\" is just like \"=~\" except the return value is negated in the logical sense.\n\nBinary \"!~\" with a non-destructive substitution (\"s///r\") or transliteration (\"y///r\") is a\nsyntax error.\n\nMultiplicative Operators\nBinary \"*\" multiplies two numbers.\n\nBinary \"/\" divides two numbers.\n\nBinary \"%\" is the modulo operator, which computes the division remainder of its first\nargument with respect to its second argument. Given integer operands $m and $n: If $n is\npositive, then \"$m%$n\" is $m minus the largest multiple of $n less than or equal to $m. If\n$n is negative, then \"$m%$n\" is $m minus the smallest multiple of $n that is not less than $m\n(that is, the result will be less than or equal to zero). If the operands $m and $n are\nfloating point values and the absolute value of $n (that is \"abs($n)\") is less than\n\"(UVMAX+1)\", only the integer portion of $m and $n will be used in the operation (Note: here\n\"UVMAX\" means the maximum of the unsigned integer type). If the absolute value of the right\noperand (\"abs($n)\") is greater than or equal to \"(UVMAX+1)\", \"%\" computes the floating-point\nremainder $r in the equation \"($r=$m-$i*$n)\" where $i is a certain integer that makes $r have\nthe same sign as the right operand $n (not as the left operand $m like C function \"fmod()\")\nand the absolute value less than that of $n. Note that when \"useinteger\" is in scope, \"%\"\ngives you direct access to the modulo operator as implemented by your C compiler. This\noperator is not as well defined for negative operands, but it will execute faster.\n\nBinary \"x\" is the repetition operator. In scalar context, or if the left operand is neither\nenclosed in parentheses nor a \"qw//\" list, it performs a string repetition. In that case it\nsupplies scalar context to the left operand, and returns a string consisting of the left\noperand string repeated the number of times specified by the right operand. If the \"x\" is in\nlist context, and the left operand is either enclosed in parentheses or a \"qw//\" list, it\nperforms a list repetition. In that case it supplies list context to the left operand, and\nreturns a list consisting of the left operand list repeated the number of times specified by\nthe right operand. If the right operand is zero or negative (raising a warning on negative),\nit returns an empty string or an empty list, depending on the context.\n\nprint '-' x 80; # print row of dashes\n\nprint \"\\t\" x ($tab/8), ' ' x ($tab%8); # tab over\n\n@ones = (1) x 80; # a list of 80 1's\n@ones = (5) x @ones; # set all elements to 5\n\nAdditive Operators\nBinary \"+\" returns the sum of two numbers.\n\nBinary \"-\" returns the difference of two numbers.\n\nBinary \".\" concatenates two strings.\n\nShift Operators\nBinary \"<<\" returns the value of its left argument shifted left by the number of bits\nspecified by the right argument. Arguments should be integers. (See also \"Integer\nArithmetic\".)\n\nBinary \">>\" returns the value of its left argument shifted right by the number of bits\nspecified by the right argument. Arguments should be integers. (See also \"Integer\nArithmetic\".)\n\nIf \"useinteger\" (see \"Integer Arithmetic\") is in force then signed C integers are used\n(arithmetic shift), otherwise unsigned C integers are used (logical shift), even for negative\nshiftees. In arithmetic right shift the sign bit is replicated on the left, in logical shift\nzero bits come in from the left.\n\nEither way, the implementation isn't going to generate results larger than the size of the\ninteger type Perl was built with (32 bits or 64 bits).\n\nShifting by negative number of bits means the reverse shift: left shift becomes right shift,\nright shift becomes left shift. This is unlike in C, where negative shift is undefined.\n\nShifting by more bits than the size of the integers means most of the time zero (all bits\nfall off), except that under \"useinteger\" right overshifting a negative shiftee results in\n-1. This is unlike in C, where shifting by too many bits is undefined. A common C behavior\nis \"shift by modulo wordbits\", so that for example\n\n1 >> 64 == 1 >> (64 % 64) == 1 >> 0 == 1 # Common C behavior.\n\nbut that is completely accidental.\n\nIf you get tired of being subject to your platform's native integers, the \"usebigint\" pragma\nneatly sidesteps the issue altogether:\n\nprint 20 << 20; # 20971520\nprint 20 << 40; # 5120 on 32-bit machines,\n# 21990232555520 on 64-bit machines\nuse bigint;\nprint 20 << 100; # 25353012004564588029934064107520\n\nNamed Unary Operators\nThe various named unary operators are treated as functions with one argument, with optional\nparentheses.\n\nIf any list operator (\"print()\", etc.) or any unary operator (\"chdir()\", etc.) is followed\nby a left parenthesis as the next token, the operator and arguments within parentheses are\ntaken to be of highest precedence, just like a normal function call. For example, because\nnamed unary operators are higher precedence than \"||\":\n\nchdir $foo || die; # (chdir $foo) || die\nchdir($foo) || die; # (chdir $foo) || die\nchdir ($foo) || die; # (chdir $foo) || die\nchdir +($foo) || die; # (chdir $foo) || die\n\nbut, because \"*\" is higher precedence than named operators:\n\nchdir $foo * 20; # chdir ($foo * 20)\nchdir($foo) * 20; # (chdir $foo) * 20\nchdir ($foo) * 20; # (chdir $foo) * 20\nchdir +($foo) * 20; # chdir ($foo * 20)\n\nrand 10 * 20; # rand (10 * 20)\nrand(10) * 20; # (rand 10) * 20\nrand (10) * 20; # (rand 10) * 20\nrand +(10) * 20; # rand (10 * 20)\n\nRegarding precedence, the filetest operators, like \"-f\", \"-M\", etc. are treated like named\nunary operators, but they don't follow this functional parenthesis rule. That means, for\nexample, that \"-f($file).\".bak\"\" is equivalent to \"-f\"$file.bak\"\".\n\nSee also \"Terms and List Operators (Leftward)\".\n\nRelational Operators\nPerl operators that return true or false generally return values that can be safely used as\nnumbers. For example, the relational operators in this section and the equality operators in\nthe next one return 1 for true and a special version of the defined empty string, \"\", which\ncounts as a zero but is exempt from warnings about improper numeric conversions, just as\n\"0buttrue\" is.\n\nBinary \"<\" returns true if the left argument is numerically less than the right argument.\n\nBinary \">\" returns true if the left argument is numerically greater than the right argument.\n\nBinary \"<=\" returns true if the left argument is numerically less than or equal to the right\nargument.\n\nBinary \">=\" returns true if the left argument is numerically greater than or equal to the\nright argument.\n\nBinary \"lt\" returns true if the left argument is stringwise less than the right argument.\n\nBinary \"gt\" returns true if the left argument is stringwise greater than the right argument.\n\nBinary \"le\" returns true if the left argument is stringwise less than or equal to the right\nargument.\n\nBinary \"ge\" returns true if the left argument is stringwise greater than or equal to the\nright argument.\n\nA sequence of relational operators, such as \"$x<$y<=$z\", performs chained comparisons, in the\nmanner described above in the section \"Operator Precedence and Associativity\". Beware that\nthey do not chain with equality operators, which have lower precedence.\n\nEquality Operators\nBinary \"==\" returns true if the left argument is numerically equal to the right argument.\n\nBinary \"!=\" returns true if the left argument is numerically not equal to the right argument.\n\nBinary \"eq\" returns true if the left argument is stringwise equal to the right argument.\n\nBinary \"ne\" returns true if the left argument is stringwise not equal to the right argument.\n\nA sequence of the above equality operators, such as \"$x==$y==$z\", performs chained\ncomparisons, in the manner described above in the section \"Operator Precedence and\nAssociativity\". Beware that they do not chain with relational operators, which have higher\nprecedence.\n\nBinary \"<=>\" returns -1, 0, or 1 depending on whether the left argument is numerically less\nthan, equal to, or greater than the right argument. If your platform supports \"NaN\"'s (not-\na-numbers) as numeric values, using them with \"<=>\" returns undef. \"NaN\" is not \"<\", \"==\",\n\">\", \"<=\" or \">=\" anything (even \"NaN\"), so those 5 return false. \"NaN!=NaN\" returns true,\nas does \"NaN!=\"anythingelse. If your platform doesn't support \"NaN\"'s then \"NaN\" is just a\nstring with numeric value 0.\n\n$ perl -le '$x = \"NaN\"; print \"No NaN support here\" if $x == $x'\n$ perl -le '$x = \"NaN\"; print \"NaN support here\" if $x != $x'\n\n(Note that the bigint, bigrat, and bignum pragmas all support \"NaN\".)\n\nBinary \"cmp\" returns -1, 0, or 1 depending on whether the left argument is stringwise less\nthan, equal to, or greater than the right argument.\n\nBinary \"~~\" does a smartmatch between its arguments. Smart matching is described in the next\nsection.\n\nThe two-sided ordering operators \"<=>\" and \"cmp\", and the smartmatch operator \"~~\", are non-\nassociative with respect to each other and with respect to the equality operators of the same\nprecedence.\n\n\"lt\", \"le\", \"ge\", \"gt\" and \"cmp\" use the collation (sort) order specified by the current\n\"LCCOLLATE\" locale if a \"uselocale\" form that includes collation is in effect. See\nperllocale. Do not mix these with Unicode, only use them with legacy 8-bit locale encodings.\nThe standard \"Unicode::Collate\" and \"Unicode::Collate::Locale\" modules offer much more\npowerful solutions to collation issues.\n\nFor case-insensitive comparisons, look at the \"fc\" in perlfunc case-folding function,\navailable in Perl v5.16 or later:\n\nif ( fc($x) eq fc($y) ) { ... }\n\nClass Instance Operator\nBinary \"isa\" evaluates to true when the left argument is an object instance of the class (or\na subclass derived from that class) given by the right argument. If the left argument is not\ndefined, not a blessed object instance, nor does not derive from the class given by the right\nargument, the operator evaluates as false. The right argument may give the class either as a\nbareword or a scalar expression that yields a string class name:\n\nif( $obj isa Some::Class ) { ... }\n\nif( $obj isa \"Different::Class\" ) { ... }\nif( $obj isa $nameofclass ) { ... }\n\nThis is an experimental feature and is available from Perl 5.31.6 when enabled by \"use\nfeature 'isa'\". It emits a warning in the \"experimental::isa\" category.\n\nSmartmatch Operator\nFirst available in Perl 5.10.1 (the 5.10.0 version behaved differently), binary \"~~\" does a\n\"smartmatch\" between its arguments. This is mostly used implicitly in the \"when\" construct\ndescribed in perlsyn, although not all \"when\" clauses call the smartmatch operator. Unique\namong all of Perl's operators, the smartmatch operator can recurse. The smartmatch operator\nis experimental and its behavior is subject to change.\n\nIt is also unique in that all other Perl operators impose a context (usually string or\nnumeric context) on their operands, autoconverting those operands to those imposed contexts.\nIn contrast, smartmatch infers contexts from the actual types of its operands and uses that\ntype information to select a suitable comparison mechanism.\n\nThe \"~~\" operator compares its operands \"polymorphically\", determining how to compare them\naccording to their actual types (numeric, string, array, hash, etc.). Like the equality\noperators with which it shares the same precedence, \"~~\" returns 1 for true and \"\" for false.\nIt is often best read aloud as \"in\", \"inside of\", or \"is contained in\", because the left\noperand is often looked for inside the right operand. That makes the order of the operands\nto the smartmatch operand often opposite that of the regular match operator. In other words,\nthe \"smaller\" thing is usually placed in the left operand and the larger one in the right.\n\nThe behavior of a smartmatch depends on what type of things its arguments are, as determined\nby the following table. The first row of the table whose types apply determines the\nsmartmatch behavior. Because what actually happens is mostly determined by the type of the\nsecond operand, the table is sorted on the right operand instead of on the left.\n\nLeft Right Description and pseudocode\n===============================================================\nAny undef check whether Any is undefined\nlike: !defined Any\n\nAny Object invoke ~~ overloading on Object, or die\n\nRight operand is an ARRAY:\n\nLeft Right Description and pseudocode\n===============================================================\nARRAY1 ARRAY2 recurse on paired elements of ARRAY1 and ARRAY2[2]\nlike: (ARRAY1[0] ~~ ARRAY2[0])\n&& (ARRAY1[1] ~~ ARRAY2[1]) && ...\nHASH ARRAY any ARRAY elements exist as HASH keys\nlike: grep { exists HASH->{$} } ARRAY\nRegexp ARRAY any ARRAY elements pattern match Regexp\nlike: grep { /Regexp/ } ARRAY\nundef ARRAY undef in ARRAY\nlike: grep { !defined } ARRAY\nAny ARRAY smartmatch each ARRAY element[3]\nlike: grep { Any ~~ $ } ARRAY\n\nRight operand is a HASH:\n\nLeft Right Description and pseudocode\n===============================================================\nHASH1 HASH2 all same keys in both HASHes\nlike: keys HASH1 ==\ngrep { exists HASH2->{$} } keys HASH1\nARRAY HASH any ARRAY elements exist as HASH keys\nlike: grep { exists HASH->{$} } ARRAY\nRegexp HASH any HASH keys pattern match Regexp\nlike: grep { /Regexp/ } keys HASH\nundef HASH always false (undef can't be a key)\nlike: 0 == 1\nAny HASH HASH key existence\nlike: exists HASH->{Any}\n\nRight operand is CODE:\n\nLeft Right Description and pseudocode\n===============================================================\nARRAY CODE sub returns true on all ARRAY elements[1]\nlike: !grep { !CODE->($) } ARRAY\nHASH CODE sub returns true on all HASH keys[1]\nlike: !grep { !CODE->($) } keys HASH\nAny CODE sub passed Any returns true\nlike: CODE->(Any)\n\nRight operand is a Regexp:\n\nLeft Right Description and pseudocode\n===============================================================\nARRAY Regexp any ARRAY elements match Regexp\nlike: grep { /Regexp/ } ARRAY\nHASH Regexp any HASH keys match Regexp\nlike: grep { /Regexp/ } keys HASH\nAny Regexp pattern match\nlike: Any =~ /Regexp/\n\nOther:\n\nLeft Right Description and pseudocode\n===============================================================\nObject Any invoke ~~ overloading on Object,\nor fall back to...\n\nAny Num numeric equality\nlike: Any == Num\nNum nummy[4] numeric equality\nlike: Num == nummy\nundef Any check whether undefined\nlike: !defined(Any)\nAny Any string equality\nlike: Any eq Any\n\nNotes:\n\n1. Empty hashes or arrays match.\n2. That is, each element smartmatches the element of the same index in the other array.[3]\n3. If a circular reference is found, fall back to referential equality.\n4. Either an actual number, or a string that looks like one.\n\nThe smartmatch implicitly dereferences any non-blessed hash or array reference, so the \"HASH\"\nand \"ARRAY\" entries apply in those cases. For blessed references, the \"Object\" entries\napply. Smartmatches involving hashes only consider hash keys, never hash values.\n\nThe \"like\" code entry is not always an exact rendition. For example, the smartmatch operator\nshort-circuits whenever possible, but \"grep\" does not. Also, \"grep\" in scalar context\nreturns the number of matches, but \"~~\" returns only true or false.\n\nUnlike most operators, the smartmatch operator knows to treat \"undef\" specially:\n\nuse v5.10.1;\n@array = (1, 2, 3, undef, 4, 5);\nsay \"some elements undefined\" if undef ~~ @array;\n\nEach operand is considered in a modified scalar context, the modification being that array\nand hash variables are passed by reference to the operator, which implicitly dereferences\nthem. Both elements of each pair are the same:\n\nuse v5.10.1;\n\nmy %hash = (red => 1, blue => 2, green => 3,\norange => 4, yellow => 5, purple => 6,\nblack => 7, grey => 8, white => 9);\n\nmy @array = qw(red blue green);\n\nsay \"some array elements in hash keys\" if @array ~~ %hash;\nsay \"some array elements in hash keys\" if \\@array ~~ \\%hash;\n\nsay \"red in array\" if \"red\" ~~ @array;\nsay \"red in array\" if \"red\" ~~ \\@array;\n\nsay \"some keys end in e\" if /e$/ ~~ %hash;\nsay \"some keys end in e\" if /e$/ ~~ \\%hash;\n\nTwo arrays smartmatch if each element in the first array smartmatches (that is, is \"in\") the\ncorresponding element in the second array, recursively.\n\nuse v5.10.1;\nmy @little = qw(red blue green);\nmy @bigger = (\"red\", \"blue\", [ \"orange\", \"green\" ] );\nif (@little ~~ @bigger) { # true!\nsay \"little is contained in bigger\";\n}\n\nBecause the smartmatch operator recurses on nested arrays, this will still report that \"red\"\nis in the array.\n\nuse v5.10.1;\nmy @array = qw(red blue green);\nmy $nestedarray = [[[[[[[ @array ]]]]]]];\nsay \"red in array\" if \"red\" ~~ $nestedarray;\n\nIf two arrays smartmatch each other, then they are deep copies of each others' values, as\nthis example reports:\n\nuse v5.12.0;\nmy @a = (0, 1, 2, [3, [4, 5], 6], 7);\nmy @b = (0, 1, 2, [3, [4, 5], 6], 7);\n\nif (@a ~~ @b && @b ~~ @a) {\nsay \"a and b are deep copies of each other\";\n}\nelsif (@a ~~ @b) {\nsay \"a smartmatches in b\";\n}\nelsif (@b ~~ @a) {\nsay \"b smartmatches in a\";\n}\nelse {\nsay \"a and b don't smartmatch each other at all\";\n}\n\nIf you were to set \"$b[3]=4\", then instead of reporting that \"a and b are deep copies of each\nother\", it now reports that \"b smartmatches in a\". That's because the corresponding position\nin @a contains an array that (eventually) has a 4 in it.\n\nSmartmatching one hash against another reports whether both contain the same keys, no more\nand no less. This could be used to see whether two records have the same field names,\nwithout caring what values those fields might have. For example:\n\nuse v5.10.1;\nsub makedogtag {\nstate $REQUIREDFIELDS = { name=>1, rank=>1, serialnum=>1 };\n\nmy ($class, $initfields) = @;\n\ndie \"Must supply (only) name, rank, and serial number\"\nunless $initfields ~~ $REQUIREDFIELDS;\n\n...\n}\n\nHowever, this only does what you mean if $initfields is indeed a hash reference. The\ncondition \"$initfields ~~ $REQUIREDFIELDS\" also allows the strings \"name\", \"rank\",\n\"serialnum\" as well as any array reference that contains \"name\" or \"rank\" or \"serialnum\"\nanywhere to pass through.\n\nThe smartmatch operator is most often used as the implicit operator of a \"when\" clause. See\nthe section on \"Switch Statements\" in perlsyn.\n\nSmartmatching of Objects\n\nTo avoid relying on an object's underlying representation, if the smartmatch's right operand\nis an object that doesn't overload \"~~\", it raises the exception \"\"Smartmatching a\nnon-overloaded object breaks encapsulation\"\". That's because one has no business digging\naround to see whether something is \"in\" an object. These are all illegal on objects without\na \"~~\" overload:\n\n%hash ~~ $object\n42 ~~ $object\n\"fred\" ~~ $object\n\nHowever, you can change the way an object is smartmatched by overloading the \"~~\" operator.\nThis is allowed to extend the usual smartmatch semantics. For objects that do have an \"~~\"\noverload, see overload.\n\nUsing an object as the left operand is allowed, although not very useful. Smartmatching\nrules take precedence over overloading, so even if the object in the left operand has\nsmartmatch overloading, this will be ignored. A left operand that is a non-overloaded object\nfalls back on a string or numeric comparison of whatever the \"ref\" operator returns. That\nmeans that\n\n$object ~~ X\n\ndoes not invoke the overload method with \"X\" as an argument. Instead the above table is\nconsulted as normal, and based on the type of \"X\", overloading may or may not be invoked.\nFor simple strings or numbers, \"in\" becomes equivalent to this:\n\n$object ~~ $number ref($object) == $number\n$object ~~ $string ref($object) eq $string\n\nFor example, this reports that the handle smells IOish (but please don't really do this!):\n\nuse IO::Handle;\nmy $fh = IO::Handle->new();\nif ($fh ~~ /\\bIO\\b/) {\nsay \"handle smells IOish\";\n}\n\nThat's because it treats $fh as a string like \"IO::Handle=GLOB(0x8039e0)\", then pattern\nmatches against that.\n\nBitwise And\nBinary \"&\" returns its operands ANDed together bit by bit. Although no warning is currently\nraised, the result is not well defined when this operation is performed on operands that\naren't either numbers (see \"Integer Arithmetic\") nor bitstrings (see \"Bitwise String\nOperators\").\n\nNote that \"&\" has lower priority than relational operators, so for example the parentheses\nare essential in a test like\n\nprint \"Even\\n\" if ($x & 1) == 0;\n\nIf the \"bitwise\" feature is enabled via \"usefeature'bitwise'\" or \"use v5.28\", then this\noperator always treats its operands as numbers. Before Perl 5.28 this feature produced a\nwarning in the \"experimental::bitwise\" category.\n\nBitwise Or and Exclusive Or\nBinary \"|\" returns its operands ORed together bit by bit.\n\nBinary \"^\" returns its operands XORed together bit by bit.\n\nAlthough no warning is currently raised, the results are not well defined when these\noperations are performed on operands that aren't either numbers (see \"Integer Arithmetic\")\nnor bitstrings (see \"Bitwise String Operators\").\n\nNote that \"|\" and \"^\" have lower priority than relational operators, so for example the\nparentheses are essential in a test like\n\nprint \"false\\n\" if (8 | 2) != 10;\n\nIf the \"bitwise\" feature is enabled via \"usefeature'bitwise'\" or \"use v5.28\", then this\noperator always treats its operands as numbers. Before Perl 5.28. this feature produced a\nwarning in the \"experimental::bitwise\" category.\n\nC-style Logical And\nBinary \"&&\" performs a short-circuit logical AND operation. That is, if the left operand is\nfalse, the right operand is not even evaluated. Scalar or list context propagates down to\nthe right operand if it is evaluated.\n\nC-style Logical Or\nBinary \"||\" performs a short-circuit logical OR operation. That is, if the left operand is\ntrue, the right operand is not even evaluated. Scalar or list context propagates down to the\nright operand if it is evaluated.\n\nLogical Defined-Or\nAlthough it has no direct equivalent in C, Perl's \"//\" operator is related to its C-style\n\"or\". In fact, it's exactly the same as \"||\", except that it tests the left hand side's\ndefinedness instead of its truth. Thus, \"EXPR1//EXPR2\" returns the value of \"EXPR1\" if it's\ndefined, otherwise, the value of \"EXPR2\" is returned. (\"EXPR1\" is evaluated in scalar\ncontext, \"EXPR2\" in the context of \"//\" itself). Usually, this is the same result as\n\"defined(EXPR1)?EXPR1:EXPR2\" (except that the ternary-operator form can be used as a lvalue,\nwhile \"EXPR1//EXPR2\" cannot). This is very useful for providing default values for\nvariables. If you actually want to test if at least one of $x and $y is defined, use\n\"defined($x//$y)\".\n\nThe \"||\", \"//\" and \"&&\" operators return the last value evaluated (unlike C's \"||\" and \"&&\",\nwhich return 0 or 1). Thus, a reasonably portable way to find out the home directory might\nbe:\n\n$home = $ENV{HOME}\n// $ENV{LOGDIR}\n// (getpwuid($<))[7]\n// die \"You're homeless!\\n\";\n\nIn particular, this means that you shouldn't use this for selecting between two aggregates\nfor assignment:\n\n@a = @b || @c; # This doesn't do the right thing\n@a = scalar(@b) || @c; # because it really means this.\n@a = @b ? @b : @c; # This works fine, though.\n\nAs alternatives to \"&&\" and \"||\" when used for control flow, Perl provides the \"and\" and \"or\"\noperators (see below). The short-circuit behavior is identical. The precedence of \"and\" and\n\"or\" is much lower, however, so that you can safely use them after a list operator without\nthe need for parentheses:\n\nunlink \"alpha\", \"beta\", \"gamma\"\nor gripe(), next LINE;\n\nWith the C-style operators that would have been written like this:\n\nunlink(\"alpha\", \"beta\", \"gamma\")\n|| (gripe(), next LINE);\n\nIt would be even more readable to write that this way:\n\nunless(unlink(\"alpha\", \"beta\", \"gamma\")) {\ngripe();\nnext LINE;\n}\n\nUsing \"or\" for assignment is unlikely to do what you want; see below.\n\nRange Operators\nBinary \"..\" is the range operator, which is really two different operators depending on the\ncontext. In list context, it returns a list of values counting (up by ones) from the left\nvalue to the right value. If the left value is greater than the right value then it returns\nthe empty list. The range operator is useful for writing \"foreach(1..10)\" loops and for\ndoing slice operations on arrays. In the current implementation, no temporary array is\ncreated when the range operator is used as the expression in \"foreach\" loops, but older\nversions of Perl might burn a lot of memory when you write something like this:\n\nfor (1 .. 1000000) {\n# code\n}\n\nThe range operator also works on strings, using the magical auto-increment, see below.\n\nIn scalar context, \"..\" returns a boolean value. The operator is bistable, like a flip-flop,\nand emulates the line-range (comma) operator of sed, awk, and various editors. Each \"..\"\noperator maintains its own boolean state, even across calls to a subroutine that contains it.\nIt is false as long as its left operand is false. Once the left operand is true, the range\noperator stays true until the right operand is true, AFTER which the range operator becomes\nfalse again. It doesn't become false till the next time the range operator is evaluated. It\ncan test the right operand and become false on the same evaluation it became true (as in\nawk), but it still returns true once. If you don't want it to test the right operand until\nthe next evaluation, as in sed, just use three dots (\"...\") instead of two. In all other\nregards, \"...\" behaves just like \"..\" does.\n\nThe right operand is not evaluated while the operator is in the \"false\" state, and the left\noperand is not evaluated while the operator is in the \"true\" state. The precedence is a\nlittle lower than || and &&. The value returned is either the empty string for false, or a\nsequence number (beginning with 1) for true. The sequence number is reset for each range\nencountered. The final sequence number in a range has the string \"E0\" appended to it, which\ndoesn't affect its numeric value, but gives you something to search for if you want to\nexclude the endpoint. You can exclude the beginning point by waiting for the sequence number\nto be greater than 1.\n\nIf either operand of scalar \"..\" is a constant expression, that operand is considered true if\nit is equal (\"==\") to the current input line number (the $. variable).\n\nTo be pedantic, the comparison is actually \"int(EXPR)==int(EXPR)\", but that is only an issue\nif you use a floating point expression; when implicitly using $. as described in the previous\nparagraph, the comparison is \"int(EXPR)==int($.)\" which is only an issue when $. is set to a\nfloating point value and you are not reading from a file. Furthermore, \"span\"..\"spat\" or\n\"2.18..3.14\" will not do what you want in scalar context because each of the operands are\nevaluated using their integer representation.\n\nExamples:\n\nAs a scalar operator:\n\nif (101 .. 200) { print; } # print 2nd hundred lines, short for\n# if ($. == 101 .. $. == 200) { print; }\n\nnext LINE if (1 .. /^$/); # skip header lines, short for\n# next LINE if ($. == 1 .. /^$/);\n# (typically in a loop labeled LINE)\n\ns/^/> / if (/^$/ .. eof()); # quote body\n\n# parse mail messages\nwhile (<>) {\n$inheader = 1 .. /^$/;\n$inbody = /^$/ .. eof;\nif ($inheader) {\n# do something\n} else { # in body\n# do something else\n}\n} continue {\nclose ARGV if eof; # reset $. each file\n}\n\nHere's a simple example to illustrate the difference between the two range operators:\n\n@lines = (\" - Foo\",\n\"01 - Bar\",\n\"1 - Baz\",\n\" - Quux\");\n\nforeach (@lines) {\nif (/0/ .. /1/) {\nprint \"$\\n\";\n}\n}\n\nThis program will print only the line containing \"Bar\". If the range operator is changed to\n\"...\", it will also print the \"Baz\" line.\n\nAnd now some examples as a list operator:\n\nfor (101 .. 200) { print } # print $ 100 times\n@foo = @foo[0 .. $#foo]; # an expensive no-op\n@foo = @foo[$#foo-4 .. $#foo]; # slice last 5 items\n\nBecause each operand is evaluated in integer form, \"2.18..3.14\" will return two elements in\nlist context.\n\n@list = (2.18 .. 3.14); # same as @list = (2 .. 3);\n\nThe range operator in list context can make use of the magical auto-increment algorithm if\nboth operands are strings, subject to the following rules:\n\no With one exception (below), if both strings look like numbers to Perl, the magic\nincrement will not be applied, and the strings will be treated as numbers (more\nspecifically, integers) instead.\n\nFor example, \"-2\"..\"2\" is the same as \"-2..2\", and \"2.18\"..\"3.14\" produces \"2, 3\".\n\no The exception to the above rule is when the left-hand string begins with 0 and is longer\nthan one character, in this case the magic increment will be applied, even though strings\nlike \"01\" would normally look like a number to Perl.\n\nFor example, \"01\"..\"04\" produces \"01\", \"02\", \"03\", \"04\", and \"00\"..\"-1\" produces \"00\"\nthrough \"99\" - this may seem surprising, but see the following rules for why it works\nthis way. To get dates with leading zeros, you can say:\n\n@z2 = (\"01\" .. \"31\");\nprint $z2[$mday];\n\nIf you want to force strings to be interpreted as numbers, you could say\n\n@numbers = ( 0+$first .. 0+$last );\n\nNote: In Perl versions 5.30 and below, any string on the left-hand side beginning with\n\"0\", including the string \"0\" itself, would cause the magic string increment behavior.\nThis means that on these Perl versions, \"0\"..\"-1\" would produce \"0\" through \"99\", which\nwas inconsistent with \"0..-1\", which produces the empty list. This also means that\n\"0\"..\"9\" now produces a list of integers instead of a list of strings.\n\no If the initial value specified isn't part of a magical increment sequence (that is, a\nnon-empty string matching \"/^[a-zA-Z]*[0-9]*\\z/\"), only the initial value will be\nreturned.\n\nFor example, \"ax\"..\"az\" produces \"ax\", \"ay\", \"az\", but \"*x\"..\"az\" produces only \"*x\".\n\no For other initial values that are strings that do follow the rules of the magical\nincrement, the corresponding sequence will be returned.\n\nFor example, you can say\n\n@alphabet = (\"A\" .. \"Z\");\n\nto get all normal letters of the English alphabet, or\n\n$hexdigit = (0 .. 9, \"a\" .. \"f\")[$num & 15];\n\nto get a hexadecimal digit.\n\no If the final value specified is not in the sequence that the magical increment would\nproduce, the sequence goes until the next value would be longer than the final value\nspecified. If the length of the final string is shorter than the first, the empty list is\nreturned.\n\nFor example, \"a\"..\"--\" is the same as \"a\"..\"zz\", \"0\"..\"xx\" produces \"0\" through \"99\", and\n\"aaa\"..\"--\" returns the empty list.\n\nAs of Perl 5.26, the list-context range operator on strings works as expected in the scope of\n\"usefeature'unicodestrings\". In previous versions, and outside the scope of that feature, it\nexhibits \"The \"Unicode Bug\"\" in perlunicode: its behavior depends on the internal encoding of\nthe range endpoint.\n\nBecause the magical increment only works on non-empty strings matching\n\"/^[a-zA-Z]*[0-9]*\\z/\", the following will only return an alpha:\n\nuse charnames \"greek\";\nmy @greeksmall = (\"\\N{alpha}\" .. \"\\N{omega}\");\n\nTo get the 25 traditional lowercase Greek letters, including both sigmas, you could use this\ninstead:\n\nuse charnames \"greek\";\nmy @greeksmall = map { chr } ( ord(\"\\N{alpha}\")\n..\nord(\"\\N{omega}\")\n);\n\nHowever, because there are many other lowercase Greek characters than just those, to match\nlowercase Greek characters in a regular expression, you could use the pattern\n\"/(?:(?=\\p{Greek})\\p{Lower})+/\" (or the experimental feature \"/(?[\\p{Greek}&\\p{Lower}])+/\").\n\nConditional Operator\nTernary \"?:\" is the conditional operator, just as in C. It works much like an if-then-else.\nIf the argument before the \"?\" is true, the argument before the \":\" is returned, otherwise\nthe argument after the \":\" is returned. For example:\n\nprintf \"I have %d dog%s.\\n\", $n,\n($n == 1) ? \"\" : \"s\";\n\nScalar or list context propagates downward into the 2nd or 3rd argument, whichever is\nselected.\n\n$x = $ok ? $y : $z; # get a scalar\n@x = $ok ? @y : @z; # get an array\n$x = $ok ? @y : @z; # oops, that's just a count!\n\nThe operator may be assigned to if both the 2nd and 3rd arguments are legal lvalues (meaning\nthat you can assign to them):\n\n($xory ? $x : $y) = $z;\n\nBecause this operator produces an assignable result, using assignments without parentheses\nwill get you in trouble. For example, this:\n\n$x % 2 ? $x += 10 : $x += 2\n\nReally means this:\n\n(($x % 2) ? ($x += 10) : $x) += 2\n\nRather than this:\n\n($x % 2) ? ($x += 10) : ($x += 2)\n\nThat should probably be written more simply as:\n\n$x += ($x % 2) ? 10 : 2;\n\nAssignment Operators\n\"=\" is the ordinary assignment operator.\n\nAssignment operators work as in C. That is,\n\n$x += 2;\n\nis equivalent to\n\n$x = $x + 2;\n\nalthough without duplicating any side effects that dereferencing the lvalue might trigger,\nsuch as from \"tie()\". Other assignment operators work similarly. The following are\nrecognized:\n\n= += *= &= &.= <<= &&=\n-= /= |= |.= >>= ||=\n.= %= ^= ^.= //=\nx=\n\nAlthough these are grouped by family, they all have the precedence of assignment. These\ncombined assignment operators can only operate on scalars, whereas the ordinary assignment\noperator can assign to arrays, hashes, lists and even references. (See \"Context\" and \"List\nvalue constructors\" in perldata, and \"Assigning to References\" in perlref.)\n\nUnlike in C, the scalar assignment operator produces a valid lvalue. Modifying an assignment\nis equivalent to doing the assignment and then modifying the variable that was assigned to.\nThis is useful for modifying a copy of something, like this:\n\n($tmp = $global) =~ tr/13579/24680/;\n\nAlthough as of 5.14, that can be also be accomplished this way:\n\nuse v5.14;\n$tmp = ($global =~ tr/13579/24680/r);\n\nLikewise,\n\n($x += 2) *= 3;\n\nis equivalent to\n\n$x += 2;\n$x *= 3;\n\nSimilarly, a list assignment in list context produces the list of lvalues assigned to, and a\nlist assignment in scalar context returns the number of elements produced by the expression\non the right hand side of the assignment.\n\nThe three dotted bitwise assignment operators (\"&.=\" \"|.=\" \"^.=\") are new in Perl 5.22. See\n\"Bitwise String Operators\".\n\nComma Operator\nBinary \",\" is the comma operator. In scalar context it evaluates its left argument, throws\nthat value away, then evaluates its right argument and returns that value. This is just like\nC's comma operator.\n\nIn list context, it's just the list argument separator, and inserts both its arguments into\nthe list. These arguments are also evaluated from left to right.\n\nThe \"=>\" operator (sometimes pronounced \"fat comma\") is a synonym for the comma except that\nit causes a word on its left to be interpreted as a string if it begins with a letter or\nunderscore and is composed only of letters, digits and underscores. This includes operands\nthat might otherwise be interpreted as operators, constants, single number v-strings or\nfunction calls. If in doubt about this behavior, the left operand can be quoted explicitly.\n\nOtherwise, the \"=>\" operator behaves exactly as the comma operator or list argument\nseparator, according to context.\n\nFor example:\n\nuse constant FOO => \"something\";\n\nmy %h = ( FOO => 23 );\n\nis equivalent to:\n\nmy %h = (\"FOO\", 23);\n\nIt is NOT:\n\nmy %h = (\"something\", 23);\n\nThe \"=>\" operator is helpful in documenting the correspondence between keys and values in\nhashes, and other paired elements in lists.\n\n%hash = ( $key => $value );\nlogin( $username => $password );\n\nThe special quoting behavior ignores precedence, and hence may apply to part of the left\noperand:\n\nprint time.shift => \"bbb\";\n\nThat example prints something like \"1314363215shiftbbb\", because the \"=>\" implicitly quotes\nthe \"shift\" immediately on its left, ignoring the fact that \"time.shift\" is the entire left\noperand.\n\nList Operators (Rightward)\nOn the right side of a list operator, the comma has very low precedence, such that it\ncontrols all comma-separated expressions found there. The only operators with lower\nprecedence are the logical operators \"and\", \"or\", and \"not\", which may be used to evaluate\ncalls to list operators without the need for parentheses:\n\nopen HANDLE, \"< :encoding(UTF-8)\", \"filename\"\nor die \"Can't open: $!\\n\";\n\nHowever, some people find that code harder to read than writing it with parentheses:\n\nopen(HANDLE, \"< :encoding(UTF-8)\", \"filename\")\nor die \"Can't open: $!\\n\";\n\nin which case you might as well just use the more customary \"||\" operator:\n\nopen(HANDLE, \"< :encoding(UTF-8)\", \"filename\")\n|| die \"Can't open: $!\\n\";\n\nSee also discussion of list operators in \"Terms and List Operators (Leftward)\".\n\nLogical Not\nUnary \"not\" returns the logical negation of the expression to its right. It's the equivalent\nof \"!\" except for the very low precedence.\n\nLogical And\nBinary \"and\" returns the logical conjunction of the two surrounding expressions. It's\nequivalent to \"&&\" except for the very low precedence. This means that it short-circuits:\nthe right expression is evaluated only if the left expression is true.\n\nLogical or and Exclusive Or\nBinary \"or\" returns the logical disjunction of the two surrounding expressions. It's\nequivalent to \"||\" except for the very low precedence. This makes it useful for control\nflow:\n\nprint FH $data or die \"Can't write to FH: $!\";\n\nThis means that it short-circuits: the right expression is evaluated only if the left\nexpression is false. Due to its precedence, you must be careful to avoid using it as\nreplacement for the \"||\" operator. It usually works out better for flow control than in\nassignments:\n\n$x = $y or $z; # bug: this is wrong\n($x = $y) or $z; # really means this\n$x = $y || $z; # better written this way\n\nHowever, when it's a list-context assignment and you're trying to use \"||\" for control flow,\nyou probably need \"or\" so that the assignment takes higher precedence.\n\n@info = stat($file) || die; # oops, scalar sense of stat!\n@info = stat($file) or die; # better, now @info gets its due\n\nThen again, you could always use parentheses.\n\nBinary \"xor\" returns the exclusive-OR of the two surrounding expressions. It cannot short-\ncircuit (of course).\n\nThere is no low precedence operator for defined-OR.\n\nC Operators Missing From Perl\nHere is what C has that Perl doesn't:\n\nunary & Address-of operator. (But see the \"\\\" operator for taking a reference.)\n\nunary * Dereference-address operator. (Perl's prefix dereferencing operators are typed: \"$\",\n\"@\", \"%\", and \"&\".)\n\n(TYPE) Type-casting operator.\n\nQuote and Quote-like Operators\nWhile we usually think of quotes as literal values, in Perl they function as operators,\nproviding various kinds of interpolating and pattern matching capabilities. Perl provides\ncustomary quote characters for these behaviors, but also provides a way for you to choose\nyour quote character for any of them. In the following table, a \"{}\" represents any pair of\ndelimiters you choose.\n\nCustomary Generic Meaning Interpolates\n'' q{} Literal no\n\"\" qq{} Literal yes\n`` qx{} Command yes*\nqw{} Word list no\n// m{} Pattern match yes*\nqr{} Pattern yes*\ns{}{} Substitution yes*\ntr{}{} Transliteration no (but see below)\ny{}{} Transliteration no (but see below)\n<{key}[0]\" are also interpolated, as are array\nand hash slices. But method calls such as \"$obj->meth\" are not.\n\nInterpolating an array or slice interpolates the elements in order, separated by the value of\n$\", so is equivalent to interpolating \"join$\",@array\". \"Punctuation\" arrays such as \"@*\" are\nusually interpolated only if the name is enclosed in braces \"@{*}\", but the arrays @, \"@+\",\nand \"@-\" are interpolated even without braces.\n\nFor double-quoted strings, the quoting from \"\\Q\" is applied after interpolation and escapes\nare processed.\n\n\"abc\\Qfoo\\tbar$s\\Exyz\"\n\nis equivalent to\n\n\"abc\" . quotemeta(\"foo\\tbar$s\") . \"xyz\"\n\nFor the pattern of regex operators (\"qr//\", \"m//\" and \"s///\"), the quoting from \"\\Q\" is\napplied after interpolation is processed, but before escapes are processed. This allows the\npattern to match literally (except for \"$\" and \"@\"). For example, the following matches:\n\n'\\s\\t' =~ /\\Q\\s\\t/\n\nBecause \"$\" or \"@\" trigger interpolation, you'll need to use something like\n\"/\\Quser\\E\\@\\Qhost/\" to match them literally.\n\nPatterns are subject to an additional level of interpretation as a regular expression. This\nis done as a second pass, after variables are interpolated, so that regular expressions may\nbe incorporated into the pattern from the variables. If this is not what you want, use \"\\Q\"\nto interpolate a variable literally.\n\nApart from the behavior described above, Perl does not expand multiple levels of\ninterpolation. In particular, contrary to the expectations of shell programmers, back-quotes\ndo NOT interpolate within double quotes, nor do single quotes impede evaluation of variables\nwhen used within double quotes.\n\nRegexp Quote-Like Operators\nHere are the quote-like operators that apply to pattern matching and related activities.\n\n\"qr/STRING/msixpodualn\"\nThis operator quotes (and possibly compiles) its STRING as a regular expression.\nSTRING is interpolated the same way as PATTERN in \"m/PATTERN/\". If \"'\" is used as\nthe delimiter, no variable interpolation is done. Returns a Perl value which may be\nused instead of the corresponding \"/STRING/msixpodualn\" expression. The returned\nvalue is a normalized version of the original pattern. It magically differs from a\nstring containing the same characters: \"ref(qr/x/)\" returns \"Regexp\"; however,\ndereferencing it is not well defined (you currently get the normalized version of the\noriginal pattern, but this may change).\n\nFor example,\n\n$rex = qr/my.STRING/is;\nprint $rex; # prints (?si-xm:my.STRING)\ns/$rex/foo/;\n\nis equivalent to\n\ns/my.STRING/foo/is;\n\nThe result may be used as a subpattern in a match:\n\n$re = qr/$pattern/;\n$string =~ /foo${re}bar/; # can be interpolated in other\n# patterns\n$string =~ $re; # or used standalone\n$string =~ /$re/; # or this way\n\nSince Perl may compile the pattern at the moment of execution of the \"qr()\" operator,\nusing \"qr()\" may have speed advantages in some situations, notably if the result of\n\"qr()\" is used standalone:\n\nsub match {\nmy $patterns = shift;\nmy @compiled = map qr/$/i, @$patterns;\ngrep {\nmy $success = 0;\nforeach my $pat (@compiled) {\n$success = 1, last if /$pat/;\n}\n$success;\n} @;\n}\n\nPrecompilation of the pattern into an internal representation at the moment of \"qr()\"\navoids the need to recompile the pattern every time a match \"/$pat/\" is attempted.\n(Perl has many other internal optimizations, but none would be triggered in the above\nexample if we did not use \"qr()\" operator.)\n\nOptions (specified by the following modifiers) are:\n\nm Treat string as multiple lines.\ns Treat string as single line. (Make . match a newline)\ni Do case-insensitive pattern matching.\nx Use extended regular expressions; specifying two\nx's means \\t and the SPACE character are ignored within\nsquare-bracketed character classes\np When matching preserve a copy of the matched string so\nthat ${^PREMATCH}, ${^MATCH}, ${^POSTMATCH} will be\ndefined (ignored starting in v5.20) as these are always\ndefined starting in that release\no Compile pattern only once.\na ASCII-restrict: Use ASCII for \\d, \\s, \\w and [[:posix:]]\ncharacter classes; specifying two a's adds the further\nrestriction that no ASCII character will match a\nnon-ASCII one under /i.\nl Use the current run-time locale's rules.\nu Use Unicode rules.\nd Use Unicode or native charset, as in 5.12 and earlier.\nn Non-capture mode. Don't let () fill in $1, $2, etc...\n\nIf a precompiled pattern is embedded in a larger pattern then the effect of\n\"msixpluadn\" will be propagated appropriately. The effect that the \"/o\" modifier has\nis not propagated, being restricted to those patterns explicitly using it.\n\nThe \"/a\", \"/d\", \"/l\", and \"/u\" modifiers (added in Perl 5.14) control the character\nset rules, but \"/a\" is the only one you are likely to want to specify explicitly; the\nother three are selected automatically by various pragmas.\n\nSee perlre for additional information on valid syntax for STRING, and for a detailed\nlook at the semantics of regular expressions. In particular, all modifiers except\nthe largely obsolete \"/o\" are further explained in \"Modifiers\" in perlre. \"/o\" is\ndescribed in the next section.\n\n\"m/PATTERN/msixpodualngc\"\n\"/PATTERN/msixpodualngc\"\nSearches a string for a pattern match, and in scalar context returns true if it\nsucceeds, false if it fails. If no string is specified via the \"=~\" or \"!~\"\noperator, the $ string is searched. (The string specified with \"=~\" need not be an\nlvalue--it may be the result of an expression evaluation, but remember the \"=~\" binds\nrather tightly.) See also perlre.\n\nOptions are as described in \"qr//\" above; in addition, the following match process\nmodifiers are available:\n\ng Match globally, i.e., find all occurrences.\nc Do not reset search position on a failed match when /g is\nin effect.\n\nIf \"/\" is the delimiter then the initial \"m\" is optional. With the \"m\" you can use\nany pair of non-whitespace (ASCII) characters as delimiters. This is particularly\nuseful for matching path names that contain \"/\", to avoid LTS (leaning toothpick\nsyndrome). If \"?\" is the delimiter, then a match-only-once rule applies, described\nin \"m?PATTERN?\" below. If \"'\" (single quote) is the delimiter, no variable\ninterpolation is performed on the PATTERN. When using a delimiter character valid in\nan identifier, whitespace is required after the \"m\".\n\nPATTERN may contain variables, which will be interpolated every time the pattern\nsearch is evaluated, except for when the delimiter is a single quote. (Note that $(,\n$), and $| are not interpolated because they look like end-of-string tests.) Perl\nwill not recompile the pattern unless an interpolated variable that it contains\nchanges. You can force Perl to skip the test and never recompile by adding a \"/o\"\n(which stands for \"once\") after the trailing delimiter. Once upon a time, Perl would\nrecompile regular expressions unnecessarily, and this modifier was useful to tell it\nnot to do so, in the interests of speed. But now, the only reasons to use \"/o\" are\none of:\n\n1. The variables are thousands of characters long and you know that they don't\nchange, and you need to wring out the last little bit of speed by having Perl\nskip testing for that. (There is a maintenance penalty for doing this, as\nmentioning \"/o\" constitutes a promise that you won't change the variables in the\npattern. If you do change them, Perl won't even notice.)\n\n2. you want the pattern to use the initial values of the variables regardless of\nwhether they change or not. (But there are saner ways of accomplishing this than\nusing \"/o\".)\n\n3. If the pattern contains embedded code, such as\n\nuse re 'eval';\n$code = 'foo(?{ $x })';\n/$code/\n\nthen perl will recompile each time, even though the pattern string hasn't\nchanged, to ensure that the current value of $x is seen each time. Use \"/o\" if\nyou want to avoid this.\n\nThe bottom line is that using \"/o\" is almost never a good idea.\n\nThe empty pattern \"//\"\nIf the PATTERN evaluates to the empty string, the last successfully matched regular\nexpression is used instead. In this case, only the \"g\" and \"c\" flags on the empty\npattern are honored; the other flags are taken from the original pattern. If no\nmatch has previously succeeded, this will (silently) act instead as a genuine empty\npattern (which will always match).\n\nNote that it's possible to confuse Perl into thinking \"//\" (the empty regex) is\nreally \"//\" (the defined-or operator). Perl is usually pretty good about this, but\nsome pathological cases might trigger this, such as \"$x///\" (is that \"($x)/(//)\" or\n\"$x///\"?) and \"print$fh//\" (\"print$fh(//\" or \"print($fh//\"?). In all of these\nexamples, Perl will assume you meant defined-or. If you meant the empty regex, just\nuse parentheses or spaces to disambiguate, or even prefix the empty regex with an \"m\"\n(so \"//\" becomes \"m//\").\n\nMatching in list context\nIf the \"/g\" option is not used, \"m//\" in list context returns a list consisting of\nthe subexpressions matched by the parentheses in the pattern, that is, ($1, $2,\n$3...) (Note that here $1 etc. are also set). When there are no parentheses in the\npattern, the return value is the list \"(1)\" for success. With or without\nparentheses, an empty list is returned upon failure.\n\nExamples:\n\nopen(TTY, \"+ =~ /^y/i && foo(); # do foo if desired\n\nif (/Version: *([0-9.]*)/) { $version = $1; }\n\nnext if m#^/usr/spool/uucp#;\n\n# poor man's grep\n$arg = shift;\nwhile (<>) {\nprint if /$arg/o; # compile only once (no longer needed!)\n}\n\nif (($F1, $F2, $Etc) = ($foo =~ /^(\\S+)\\s+(\\S+)\\s*(.*)/))\n\nThis last example splits $foo into the first two words and the remainder of the line,\nand assigns those three fields to $F1, $F2, and $Etc. The conditional is true if any\nvariables were assigned; that is, if the pattern matched.\n\nThe \"/g\" modifier specifies global pattern matching--that is, matching as many times\nas possible within the string. How it behaves depends on the context. In list\ncontext, it returns a list of the substrings matched by any capturing parentheses in\nthe regular expression. If there are no parentheses, it returns a list of all the\nmatched strings, as if there were parentheses around the whole pattern.\n\nIn scalar context, each execution of \"m//g\" finds the next match, returning true if\nit matches, and false if there is no further match. The position after the last\nmatch can be read or set using the \"pos()\" function; see \"pos\" in perlfunc. A failed\nmatch normally resets the search position to the beginning of the string, but you can\navoid that by adding the \"/c\" modifier (for example, \"m//gc\"). Modifying the target\nstring also resets the search position.\n\n\"\\G assertion\"\nYou can intermix \"m//g\" matches with \"m/\\G.../g\", where \"\\G\" is a zero-width\nassertion that matches the exact position where the previous \"m//g\", if any, left\noff. Without the \"/g\" modifier, the \"\\G\" assertion still anchors at \"pos()\" as it\nwas at the start of the operation (see \"pos\" in perlfunc), but the match is of course\nonly attempted once. Using \"\\G\" without \"/g\" on a target string that has not\npreviously had a \"/g\" match applied to it is the same as using the \"\\A\" assertion to\nmatch the beginning of the string. Note also that, currently, \"\\G\" is only properly\nsupported when anchored at the very beginning of the pattern.\n\nExamples:\n\n# list context\n($one,$five,$fifteen) = (`uptime` =~ /(\\d+\\.\\d+)/g);\n\n# scalar context\nlocal $/ = \"\";\nwhile ($paragraph = <>) {\nwhile ($paragraph =~ /\\p{Ll}['\")]*[.!?]+['\")]*\\s/g) {\n$sentences++;\n}\n}\nsay $sentences;\n\nHere's another way to check for sentences in a paragraph:\n\nmy $sentencerx = qr{\n(?: (?<= ^ ) | (?<= \\s ) ) # after start-of-string or\n# whitespace\n\\p{Lu} # capital letter\n.*? # a bunch of anything\n(?<= \\S ) # that ends in non-\n# whitespace\n(?) {\nsay \"NEW PARAGRAPH\";\nmy $count = 0;\nwhile ($paragraph =~ /($sentencerx)/g) {\nprintf \"\\tgot sentence %d: <%s>\\n\", ++$count, $1;\n}\n}\n\nHere's how to use \"m//gc\" with \"\\G\":\n\n$ = \"ppooqppqq\";\nwhile ($i++ < 2) {\nprint \"1: '\";\nprint $1 while /(o)/gc; print \"', pos=\", pos, \"\\n\";\nprint \"2: '\";\nprint $1 if /\\G(q)/gc; print \"', pos=\", pos, \"\\n\";\nprint \"3: '\";\nprint $1 while /(p)/gc; print \"', pos=\", pos, \"\\n\";\n}\nprint \"Final: '$1', pos=\",pos,\"\\n\" if /\\G(.)/;\n\nThe last example should print:\n\n1: 'oo', pos=4\n2: 'q', pos=5\n3: 'pp', pos=7\n1: '', pos=7\n2: 'q', pos=8\n3: '', pos=8\nFinal: 'q', pos=8\n\nNotice that the final match matched \"q\" instead of \"p\", which a match without the\n\"\\G\" anchor would have done. Also note that the final match did not update \"pos\".\n\"pos\" is only updated on a \"/g\" match. If the final match did indeed match \"p\", it's\na good bet that you're running an ancient (pre-5.6.0) version of Perl.\n\nA useful idiom for \"lex\"-like scanners is \"/\\G.../gc\". You can combine several\nregexps like this to process a string part-by-part, doing different actions depending\non which regexp matched. Each regexp tries to match where the previous one leaves\noff.\n\n$ = <<'EOL';\n$url = URI::URL->new( \"http://example.com/\" );\ndie if $url eq \"xXx\";\nEOL\n\nLOOP: {\nprint(\" digits\"), redo LOOP if /\\G\\d+\\b[,.;]?\\s*/gc;\nprint(\" lowercase\"), redo LOOP\nif /\\G\\p{Ll}+\\b[,.;]?\\s*/gc;\nprint(\" UPPERCASE\"), redo LOOP\nif /\\G\\p{Lu}+\\b[,.;]?\\s*/gc;\nprint(\" Capitalized\"), redo LOOP\nif /\\G\\p{Lu}\\p{Ll}+\\b[,.;]?\\s*/gc;\nprint(\" MiXeD\"), redo LOOP if /\\G\\pL+\\b[,.;]?\\s*/gc;\nprint(\" alphanumeric\"), redo LOOP\nif /\\G[\\p{Alpha}\\pN]+\\b[,.;]?\\s*/gc;\nprint(\" line-noise\"), redo LOOP if /\\G\\W+/gc;\nprint \". That's all!\\n\";\n}\n\nHere is the output (split into several lines):\n\nline-noise lowercase line-noise UPPERCASE line-noise UPPERCASE\nline-noise lowercase line-noise lowercase line-noise lowercase\nlowercase line-noise lowercase lowercase line-noise lowercase\nlowercase line-noise MiXeD line-noise. That's all!\n\n\"m?PATTERN?msixpodualngc\"\nThis is just like the \"m/PATTERN/\" search, except that it matches only once between\ncalls to the \"reset()\" operator. This is a useful optimization when you want to see\nonly the first occurrence of something in each file of a set of files, for instance.\nOnly \"m??\" patterns local to the current package are reset.\n\nwhile (<>) {\nif (m?^$?) {\n# blank line between header and body\n}\n} continue {\nreset if eof; # clear m?? status for next file\n}\n\nAnother example switched the first \"latin1\" encoding it finds to \"utf8\" in a pod\nfile:\n\ns//utf8/ if m? ^ =encoding \\h+ \\K latin1 ?x;\n\nThe match-once behavior is controlled by the match delimiter being \"?\"; with any\nother delimiter this is the normal \"m//\" operator.\n\nIn the past, the leading \"m\" in \"m?PATTERN?\" was optional, but omitting it would\nproduce a deprecation warning. As of v5.22.0, omitting it produces a syntax error.\nIf you encounter this construct in older code, you can just add \"m\".\n\n\"s/PATTERN/REPLACEMENT/msixpodualngcer\"\nSearches a string for a pattern, and if found, replaces that pattern with the\nreplacement text and returns the number of substitutions made. Otherwise it returns\nfalse (a value that is both an empty string (\"\") and numeric zero (0) as described in\n\"Relational Operators\").\n\nIf the \"/r\" (non-destructive) option is used then it runs the substitution on a copy\nof the string and instead of returning the number of substitutions, it returns the\ncopy whether or not a substitution occurred. The original string is never changed\nwhen \"/r\" is used. The copy will always be a plain string, even if the input is an\nobject or a tied variable.\n\nIf no string is specified via the \"=~\" or \"!~\" operator, the $ variable is searched\nand modified. Unless the \"/r\" option is used, the string specified must be a scalar\nvariable, an array element, a hash element, or an assignment to one of those; that\nis, some sort of scalar lvalue.\n\nIf the delimiter chosen is a single quote, no variable interpolation is done on\neither the PATTERN or the REPLACEMENT. Otherwise, if the PATTERN contains a \"$\" that\nlooks like a variable rather than an end-of-string test, the variable will be\ninterpolated into the pattern at run-time. If you want the pattern compiled only\nonce the first time the variable is interpolated, use the \"/o\" option. If the\npattern evaluates to the empty string, the last successfully executed regular\nexpression is used instead. See perlre for further explanation on these.\n\nOptions are as with \"m//\" with the addition of the following replacement specific\noptions:\n\ne Evaluate the right side as an expression.\nee Evaluate the right side as a string then eval the\nresult.\nr Return substitution and leave the original string\nuntouched.\n\nAny non-whitespace delimiter may replace the slashes. Add space after the \"s\" when\nusing a character allowed in identifiers. If single quotes are used, no\ninterpretation is done on the replacement string (the \"/e\" modifier overrides this,\nhowever). Note that Perl treats backticks as normal delimiters; the replacement text\nis not evaluated as a command. If the PATTERN is delimited by bracketing quotes, the\nREPLACEMENT has its own pair of quotes, which may or may not be bracketing quotes,\nfor example, \"s(foo)(bar)\" or \"s/bar/\". A \"/e\" will cause the replacement\nportion to be treated as a full-fledged Perl expression and evaluated right then and\nthere. It is, however, syntax checked at compile-time. A second \"e\" modifier will\ncause the replacement portion to be \"eval\"ed before being run as a Perl expression.\n\nExamples:\n\ns/\\bgreen\\b/mauve/g; # don't change wintergreen\n\n$path =~ s|/usr/bin|/usr/local/bin|;\n\ns/Login: $foo/Login: $bar/; # run-time pattern\n\n($foo = $bar) =~ s/this/that/; # copy first, then\n# change\n($foo = \"$bar\") =~ s/this/that/; # convert to string,\n# copy, then change\n$foo = $bar =~ s/this/that/r; # Same as above using /r\n$foo = $bar =~ s/this/that/r\n=~ s/that/the other/r; # Chained substitutes\n# using /r\n@foo = map { s/this/that/r } @bar # /r is very useful in\n# maps\n\n$count = ($paragraph =~ s/Mister\\b/Mr./g); # get change-cnt\n\n$ = 'abc123xyz';\ns/\\d+/$&*2/e; # yields 'abc246xyz'\ns/\\d+/sprintf(\"%5d\",$&)/e; # yields 'abc 246xyz'\ns/\\w/$& x 2/eg; # yields 'aabbcc 224466xxyyzz'\n\ns/%(.)/$percent{$1}/g; # change percent escapes; no /e\ns/%(.)/$percent{$1} || $&/ge; # expr now, so /e\ns/^=(\\w+)/pod($1)/ge; # use function call\n\n$ = 'abc123xyz';\n$x = s/abc/def/r; # $x is 'def123xyz' and\n# $ remains 'abc123xyz'.\n\n# expand variables in $, but dynamics only, using\n# symbolic dereferencing\ns/\\$(\\w+)/${$1}/g;\n\n# Add one to the value of any numbers in the string\ns/(\\d+)/1 + $1/eg;\n\n# Titlecase words in the last 30 characters only\nsubstr($str, -30) =~ s/\\b(\\p{Alpha}+)\\b/\\u\\L$1/g;\n\n# This will expand any embedded scalar variable\n# (including lexicals) in $ : First $1 is interpolated\n# to the variable name, and then evaluated\ns/(\\$\\w+)/$1/eeg;\n\n# Delete (most) C comments.\n$program =~ s {\n/\\* # Match the opening delimiter.\n.*? # Match a minimal number of characters.\n\\*/ # Match the closing delimiter.\n} []gsx;\n\ns/^\\s*(.*?)\\s*$/$1/; # trim whitespace in $,\n# expensively\n\nfor ($variable) { # trim whitespace in $variable,\n# cheap\ns/^\\s+//;\ns/\\s+$//;\n}\n\ns/([^ ]*) *([^ ]*)/$2 $1/; # reverse 1st two fields\n\n$foo !~ s/A/a/g; # Lowercase all A's in $foo; return\n# 0 if any were found and changed;\n# otherwise return 1\n\nNote the use of \"$\" instead of \"\\\" in the last example. Unlike sed, we use the\n\\ form only in the left hand side. Anywhere else it's $.\n\nOccasionally, you can't use just a \"/g\" to get all the changes to occur that you\nmight want. Here are two common cases:\n\n# put commas in the right places in an integer\n1 while s/(\\d)(\\d\\d\\d)(?!\\d)/$1,$2/g;\n\n# expand tabs to 8-column spacing\n1 while s/\\t+/' ' x (length($&)*8 - length($`)%8)/e;\n\nWhile \"s///\" accepts the \"/c\" flag, it has no effect beyond producing a warning if\nwarnings are enabled.\n\nQuote-Like Operators\n\"q/STRING/\"\n'STRING'\nA single-quoted, literal string. A backslash represents a backslash unless followed by\nthe delimiter or another backslash, in which case the delimiter or backslash is\ninterpolated.\n\n$foo = q!I said, \"You said, 'She said it.'\"!;\n$bar = q('This is it.');\n$baz = '\\n'; # a two-character string\n\n\"qq/STRING/\"\n\"STRING\"\nA double-quoted, interpolated string.\n\n$ .= qq\n(* The previous line contains the naughty word \"$1\".\\n)\nif /\\b(tcl|java|python)\\b/i; # :-)\n$baz = \"\\n\"; # a one-character string\n\n\"qx/STRING/\"\n\"`STRING`\"\nA string which is (possibly) interpolated and then executed as a system command, via\n/bin/sh or its equivalent if required. Shell wildcards, pipes, and redirections will be\nhonored. Similarly to \"system\", if the string contains no shell metacharacters then it\nwill executed directly. The collected standard output of the command is returned;\nstandard error is unaffected. In scalar context, it comes back as a single (potentially\nmulti-line) string, or \"undef\" if the shell (or command) could not be started. In list\ncontext, returns a list of lines (however you've defined lines with $/ or\n$INPUTRECORDSEPARATOR), or an empty list if the shell (or command) could not be\nstarted.\n\nBecause backticks do not affect standard error, use shell file descriptor syntax\n(assuming the shell supports this) if you care to address this. To capture a command's\nSTDERR and STDOUT together:\n\n$output = `cmd 2>&1`;\n\nTo capture a command's STDOUT but discard its STDERR:\n\n$output = `cmd 2>/dev/null`;\n\nTo capture a command's STDERR but discard its STDOUT (ordering is important here):\n\n$output = `cmd 2>&1 1>/dev/null`;\n\nTo exchange a command's STDOUT and STDERR in order to capture the STDERR but leave its\nSTDOUT to come out the old STDERR:\n\n$output = `cmd 3>&1 1>&2 2>&3 3>&-`;\n\nTo read both a command's STDOUT and its STDERR separately, it's easiest to redirect them\nseparately to files, and then read from those files when the program is done:\n\nsystem(\"program args 1>program.stdout 2>program.stderr\");\n\nThe STDIN filehandle used by the command is inherited from Perl's STDIN. For example:\n\nopen(SPLAT, \"stuff\") || die \"can't open stuff: $!\";\nopen(STDIN, \"<&SPLAT\") || die \"can't dupe SPLAT: $!\";\nprint STDOUT `sort`;\n\nwill print the sorted contents of the file named \"stuff\".\n\nUsing single-quote as a delimiter protects the command from Perl's double-quote\ninterpolation, passing it on to the shell instead:\n\n$perlinfo = qx(ps $$); # that's Perl's $$\n$shellinfo = qx'ps $$'; # that's the new shell's $$\n\nHow that string gets evaluated is entirely subject to the command interpreter on your\nsystem. On most platforms, you will have to protect shell metacharacters if you want\nthem treated literally. This is in practice difficult to do, as it's unclear how to\nescape which characters. See perlsec for a clean and safe example of a manual \"fork()\"\nand \"exec()\" to emulate backticks safely.\n\nOn some platforms (notably DOS-like ones), the shell may not be capable of dealing with\nmultiline commands, so putting newlines in the string may not get you what you want. You\nmay be able to evaluate multiple commands in a single line by separating them with the\ncommand separator character, if your shell supports that (for example, \";\" on many Unix\nshells and \"&\" on the Windows NT \"cmd\" shell).\n\nPerl will attempt to flush all files opened for output before starting the child process,\nbut this may not be supported on some platforms (see perlport). To be safe, you may need\nto set $| ($AUTOFLUSH in \"English\") or call the \"autoflush()\" method of \"IO::Handle\" on\nany open handles.\n\nBeware that some command shells may place restrictions on the length of the command line.\nYou must ensure your strings don't exceed this limit after any necessary interpolations.\nSee the platform-specific release notes for more details about your particular\nenvironment.\n\nUsing this operator can lead to programs that are difficult to port, because the shell\ncommands called vary between systems, and may in fact not be present at all. As one\nexample, the \"type\" command under the POSIX shell is very different from the \"type\"\ncommand under DOS. That doesn't mean you should go out of your way to avoid backticks\nwhen they're the right way to get something done. Perl was made to be a glue language,\nand one of the things it glues together is commands. Just understand what you're getting\nyourself into.\n\nLike \"system\", backticks put the child process exit code in $?. If you'd like to\nmanually inspect failure, you can check all possible failure modes by inspecting $? like\nthis:\n\nif ($? == -1) {\nprint \"failed to execute: $!\\n\";\n}\nelsif ($? & 127) {\nprintf \"child died with signal %d, %s coredump\\n\",\n($? & 127), ($? & 128) ? 'with' : 'without';\n}\nelse {\nprintf \"child exited with value %d\\n\", $? >> 8;\n}\n\nUse the open pragma to control the I/O layers used when reading the output of the\ncommand, for example:\n\nuse open IN => \":encoding(UTF-8)\";\nmy $x = `cmd-producing-utf-8`;\n\n\"qx//\" can also be called like a function with \"readpipe\" in perlfunc.\n\nSee \"I/O Operators\" for more discussion.\n\n\"qw/STRING/\"\nEvaluates to a list of the words extracted out of STRING, using embedded whitespace as\nthe word delimiters. It can be understood as being roughly equivalent to:\n\nsplit(\" \", q/STRING/);\n\nthe differences being that it only splits on ASCII whitespace, generates a real list at\ncompile time, and in scalar context it returns the last element in the list. So this\nexpression:\n\nqw(foo bar baz)\n\nis semantically equivalent to the list:\n\n\"foo\", \"bar\", \"baz\"\n\nSome frequently seen examples:\n\nuse POSIX qw( setlocale localeconv )\n@EXPORT = qw( foo bar baz );\n\nA common mistake is to try to separate the words with commas or to put comments into a\nmulti-line \"qw\"-string. For this reason, the \"usewarnings\" pragma and the -w switch\n(that is, the $^W variable) produces warnings if the STRING contains the \",\" or the \"#\"\ncharacter.\n\n\"tr/SEARCHLIST/REPLACEMENTLIST/cdsr\"\n\"y/SEARCHLIST/REPLACEMENTLIST/cdsr\"\nTransliterates all occurrences of the characters found (or not found if the \"/c\" modifier\nis specified) in the search list with the positionally corresponding character in the\nreplacement list, possibly deleting some, depending on the modifiers specified. It\nreturns the number of characters replaced or deleted. If no string is specified via the\n\"=~\" or \"!~\" operator, the $ string is transliterated.\n\nFor sed devotees, \"y\" is provided as a synonym for \"tr\".\n\nIf the \"/r\" (non-destructive) option is present, a new copy of the string is made and its\ncharacters transliterated, and this copy is returned no matter whether it was modified or\nnot: the original string is always left unchanged. The new copy is always a plain\nstring, even if the input string is an object or a tied variable.\n\nUnless the \"/r\" option is used, the string specified with \"=~\" must be a scalar variable,\nan array element, a hash element, or an assignment to one of those; in other words, an\nlvalue.\n\nThe characters delimitting SEARCHLIST and REPLACEMENTLIST can be any printable character,\nnot just forward slashes. If they are single quotes (\"tr'SEARCHLIST'REPLACEMENTLIST'\"),\nthe only interpolation is removal of \"\\\" from pairs of \"\\\\\".\n\nOtherwise, a character range may be specified with a hyphen, so \"tr/A-J/0-9/\" does the\nsame replacement as \"tr/ACEGIBDFHJ/0246813579/\".\n\nIf the SEARCHLIST is delimited by bracketing quotes, the REPLACEMENTLIST must have its\nown pair of quotes, which may or may not be bracketing quotes; for example,\n\"tr[aeiouy][yuoiea]\" or \"tr(+\\-*/)/ABCD/\".\n\nCharacters may be literals, or (if the delimiters aren't single quotes) any of the escape\nsequences accepted in double-quoted strings. But there is never any variable\ninterpolation, so \"$\" and \"@\" are always treated as literals. A hyphen at the beginning\nor end, or preceded by a backslash is also always considered a literal. Escape sequence\ndetails are in the table near the beginning of this section.\n\nNote that \"tr\" does not do regular expression character classes such as \"\\d\" or \"\\pL\".\nThe \"tr\" operator is not equivalent to the tr(1) utility. \"tr[a-z][A-Z]\" will uppercase\nthe 26 letters \"a\" through \"z\", but for case changing not confined to ASCII, use \"lc\",\n\"uc\", \"lcfirst\", \"ucfirst\" (all documented in perlfunc), or the substitution operator\n\"s/PATTERN/REPLACEMENT/\" (with \"\\U\", \"\\u\", \"\\L\", and \"\\l\" string-interpolation escapes in\nthe REPLACEMENT portion).\n\nMost ranges are unportable between character sets, but certain ones signal Perl to do\nspecial handling to make them portable. There are two classes of portable ranges. The\nfirst are any subsets of the ranges \"A-Z\", \"a-z\", and \"0-9\", when expressed as literal\ncharacters.\n\ntr/h-k/H-K/\n\ncapitalizes the letters \"h\", \"i\", \"j\", and \"k\" and nothing else, no matter what the\nplatform's character set is. In contrast, all of\n\ntr/\\x68-\\x6B/\\x48-\\x4B/\ntr/h-\\x6B/H-\\x4B/\ntr/\\x68-k/\\x48-K/\n\ndo the same capitalizations as the previous example when run on ASCII platforms, but\nsomething completely different on EBCDIC ones.\n\nThe second class of portable ranges is invoked when one or both of the range's end points\nare expressed as \"\\N{...}\"\n\n$string =~ tr/\\N{U+20}-\\N{U+7E}//d;\n\nremoves from $string all the platform's characters which are equivalent to any of Unicode\nU+0020, U+0021, ... U+007D, U+007E. This is a portable range, and has the same effect on\nevery platform it is run on. In this example, these are the ASCII printable characters.\nSo after this is run, $string has only controls and characters which have no ASCII\nequivalents.\n\nBut, even for portable ranges, it is not generally obvious what is included without\nhaving to look things up in the manual. A sound principle is to use only ranges that\nboth begin from, and end at, either ASCII alphabetics of equal case (\"b-e\", \"B-E\"), or\ndigits (\"1-4\"). Anything else is unclear (and unportable unless \"\\N{...}\" is used). If\nin doubt, spell out the character sets in full.\n\nOptions:\n\nc Complement the SEARCHLIST.\nd Delete found but unreplaced characters.\nr Return the modified string and leave the original string\nuntouched.\ns Squash duplicate replaced characters.\n\nIf the \"/d\" modifier is specified, any characters specified by SEARCHLIST not found in\nREPLACEMENTLIST are deleted. (Note that this is slightly more flexible than the behavior\nof some tr programs, which delete anything they find in the SEARCHLIST, period.)\n\nIf the \"/s\" modifier is specified, sequences of characters, all in a row, that were\ntransliterated to the same character are squashed down to a single instance of that\ncharacter.\n\nmy $a = \"aaabbbca\";\n$a =~ tr/ab/dd/s; # $a now is \"dcd\"\n\nIf the \"/d\" modifier is used, the REPLACEMENTLIST is always interpreted exactly as\nspecified. Otherwise, if the REPLACEMENTLIST is shorter than the SEARCHLIST, the final\ncharacter, if any, is replicated until it is long enough. There won't be a final\ncharacter if and only if the REPLACEMENTLIST is empty, in which case REPLACEMENTLIST is\ncopied from SEARCHLIST. An empty REPLACEMENTLIST is useful for counting characters in\na class, or for squashing character sequences in a class.\n\ntr/abcd// tr/abcd/abcd/\ntr/abcd/AB/ tr/abcd/ABBB/\ntr/abcd//d s/[abcd]//g\ntr/abcd/AB/d (tr/ab/AB/ + s/[cd]//g) - but run together\n\nIf the \"/c\" modifier is specified, the characters to be transliterated are the ones NOT\nin SEARCHLIST, that is, it is complemented. If \"/d\" and/or \"/s\" are also specified, they\napply to the complemented SEARCHLIST. Recall, that if REPLACEMENTLIST is empty (except\nunder \"/d\") a copy of SEARCHLIST is used instead. That copy is made after complementing\nunder \"/c\". SEARCHLIST is sorted by code point order after complementing, and any\nREPLACEMENTLIST is applied to that sorted result. This means that under \"/c\", the order\nof the characters specified in SEARCHLIST is irrelevant. This can lead to different\nresults on EBCDIC systems if REPLACEMENTLIST contains more than one character, hence it\nis generally non-portable to use \"/c\" with such a REPLACEMENTLIST.\n\nAnother way of describing the operation is this: If \"/c\" is specified, the SEARCHLIST is\nsorted by code point order, then complemented. If REPLACEMENTLIST is empty and \"/d\" is\nnot specified, REPLACEMENTLIST is replaced by a copy of SEARCHLIST (as modified under\n\"/c\"), and these potentially modified lists are used as the basis for what follows. Any\ncharacter in the target string that isn't in SEARCHLIST is passed through unchanged.\nEvery other character in the target string is replaced by the character in\nREPLACEMENTLIST that positionally corresponds to its mate in SEARCHLIST, except that\nunder \"/s\", the 2nd and following characters are squeezed out in a sequence of characters\nin a row that all translate to the same character. If SEARCHLIST is longer than\nREPLACEMENTLIST, characters in the target string that match a character in SEARCHLIST\nthat doesn't have a correspondence in REPLACEMENTLIST are either deleted from the target\nstring if \"/d\" is specified; or replaced by the final character in REPLACEMENTLIST if\n\"/d\" isn't specified.\n\nSome examples:\n\n$ARGV[1] =~ tr/A-Z/a-z/; # canonicalize to lower case ASCII\n\n$cnt = tr/*/*/; # count the stars in $\n$cnt = tr/*//; # same thing\n\n$cnt = $sky =~ tr/*/*/; # count the stars in $sky\n$cnt = $sky =~ tr/*//; # same thing\n\n$cnt = $sky =~ tr/*//c; # count all the non-stars in $sky\n$cnt = $sky =~ tr/*/*/c; # same, but transliterate each non-star\n# into a star, leaving the already-stars\n# alone. Afterwards, everything in $sky\n# is a star.\n\n$cnt = tr/0-9//; # count the ASCII digits in $\n\ntr/a-zA-Z//s; # bookkeeper -> bokeper\ntr/o/o/s; # bookkeeper -> bokkeeper\ntr/oe/oe/s; # bookkeeper -> bokkeper\ntr/oe//s; # bookkeeper -> bokkeper\ntr/oe/o/s; # bookkeeper -> bokkopor\n\n($HOST = $host) =~ tr/a-z/A-Z/;\n$HOST = $host =~ tr/a-z/A-Z/r; # same thing\n\n$HOST = $host =~ tr/a-z/A-Z/r # chained with s///r\n=~ s/:/ -p/r;\n\ntr/a-zA-Z/ /cs; # change non-alphas to single space\n\n@stripped = map tr/a-zA-Z/ /csr, @original;\n# /r with map\n\ntr [\\200-\\377]\n[\\000-\\177]; # wickedly delete 8th bit\n\n$foo !~ tr/A/a/ # transliterate all the A's in $foo to 'a',\n# return 0 if any were found and changed.\n# Otherwise return 1\n\nIf multiple transliterations are given for a character, only the first one is used:\n\ntr/AAA/XYZ/\n\nwill transliterate any A to X.\n\nBecause the transliteration table is built at compile time, neither the SEARCHLIST nor\nthe REPLACEMENTLIST are subjected to double quote interpolation. That means that if you\nwant to use variables, you must use an \"eval()\":\n\neval \"tr/$oldlist/$newlist/\";\ndie $@ if $@;\n\neval \"tr/$oldlist/$newlist/, 1\" or die $@;\n\n\"<\"). If the starting delimiter is an unpaired character like \"/\" or a closing\npunctuation, the ending delimiter is the same as the starting delimiter. Therefore a \"/\"\nterminates a \"qq//\" construct, while a \"]\" terminates both \"qq[]\" and \"qq]]\" constructs.\n\nWhen searching for single-character delimiters, escaped delimiters and \"\\\\\" are skipped.\nFor example, while searching for terminating \"/\", combinations of \"\\\\\" and \"\\/\" are\nskipped. If the delimiters are bracketing, nested pairs are also skipped. For example,\nwhile searching for a closing \"]\" paired with the opening \"[\", combinations of \"\\\\\",\n\"\\]\", and \"\\[\" are all skipped, and nested \"[\" and \"]\" are skipped as well. However,\nwhen backslashes are used as the delimiters (like \"qq\\\\\" and \"tr\\\\\\\"), nothing is\nskipped. During the search for the end, backslashes that escape delimiters or other\nbackslashes are removed (exactly speaking, they are not copied to the safe location).\n\nFor constructs with three-part delimiters (\"s///\", \"y///\", and \"tr///\"), the search is\nrepeated once more. If the first delimiter is not an opening punctuation, the three\ndelimiters must be the same, such as \"s!!!\" and \"tr)))\", in which case the second\ndelimiter terminates the left part and starts the right part at once. If the left part\nis delimited by bracketing punctuation (that is \"()\", \"[]\", \"{}\", or \"<>\"), the right\npart needs another pair of delimiters such as \"s(){}\" and \"tr[]//\". In these cases,\nwhitespace and comments are allowed between the two parts, although the comment must\nfollow at least one whitespace character; otherwise a character expected as the start of\nthe comment may be regarded as the starting delimiter of the right part.\n\nDuring this search no attention is paid to the semantics of the construct. Thus:\n\n\"$hash{\"$foo/$bar\"}\"\n\nor:\n\nm/\nbar # NOT a comment, this slash / terminated m//!\n/x\n\ndo not form legal quoted expressions. The quoted part ends on the first \"\"\" and \"/\",\nand the rest happens to be a syntax error. Because the slash that terminated \"m//\" was\nfollowed by a \"SPACE\", the example above is not \"m//x\", but rather \"m//\" with no \"/x\"\nmodifier. So the embedded \"#\" is interpreted as a literal \"#\".\n\nAlso no attention is paid to \"\\c\\\" (multichar control char syntax) during this search.\nThus the second \"\\\" in \"qq/\\c\\/\" is interpreted as a part of \"\\/\", and the following \"/\"\nis not recognized as a delimiter. Instead, use \"\\034\" or \"\\x1c\" at the end of quoted\nconstructs.\n\nInterpolation\nThe next step is interpolation in the text obtained, which is now delimiter-independent.\nThere are multiple cases.\n\n\"<<'EOF'\"\nNo interpolation is performed. Note that the combination \"\\\\\" is left intact, since\nescaped delimiters are not available for here-docs.\n\n\"m''\", the pattern of \"s'''\"\nNo interpolation is performed at this stage. Any backslashed sequences including\n\"\\\\\" are treated at the stage to \"parsing regular expressions\".\n\n'', \"q//\", \"tr'''\", \"y'''\", the replacement of \"s'''\"\nThe only interpolation is removal of \"\\\" from pairs of \"\\\\\". Therefore \"-\" in\n\"tr'''\" and \"y'''\" is treated literally as a hyphen and no character range is\navailable. \"\\1\" in the replacement of \"s'''\" does not work as $1.\n\n\"tr///\", \"y///\"\nNo variable interpolation occurs. String modifying combinations for case and quoting\nsuch as \"\\Q\", \"\\U\", and \"\\E\" are not recognized. The other escape sequences such as\n\"\\200\" and \"\\t\" and backslashed characters such as \"\\\\\" and \"\\-\" are converted to\nappropriate literals. The character \"-\" is treated specially and therefore \"\\-\" is\ntreated as a literal \"-\".\n\n\"\", \"``\", \"qq//\", \"qx//\", \"\", \"<<\"EOF\"\"\n\"\\Q\", \"\\U\", \"\\u\", \"\\L\", \"\\l\", \"\\F\" (possibly paired with \"\\E\") are converted to\ncorresponding Perl constructs. Thus, \"$foo\\Qbaz$bar\" is converted to\n\"$foo.(quotemeta(\"baz\".$bar))\" internally. The other escape sequences such as \"\\200\"\nand \"\\t\" and backslashed characters such as \"\\\\\" and \"\\-\" are replaced with\nappropriate expansions.\n\nLet it be stressed that whatever falls between \"\\Q\" and \"\\E\" is interpolated in the\nusual way. Something like \"\\Q\\\\E\" has no \"\\E\" inside. Instead, it has \"\\Q\", \"\\\\\",\nand \"E\", so the result is the same as for \"\\\\\\\\E\". As a general rule, backslashes\nbetween \"\\Q\" and \"\\E\" may lead to counterintuitive results. So, \"\\Q\\t\\E\" is\nconverted to \"quotemeta(\"\\t\")\", which is the same as \"\\\\\\t\" (since TAB is not\nalphanumeric). Note also that:\n\n$str = '\\t';\nreturn \"\\Q$str\";\n\nmay be closer to the conjectural intention of the writer of \"\\Q\\t\\E\".\n\nInterpolated scalars and arrays are converted internally to the \"join\" and \".\"\ncatenation operations. Thus, \"$fooXXX'@arr'\" becomes:\n\n$foo . \" XXX '\" . (join $\", @arr) . \"'\";\n\nAll operations above are performed simultaneously, left to right.\n\nBecause the result of \"\\QSTRING\\E\" has all metacharacters quoted, there is no way to\ninsert a literal \"$\" or \"@\" inside a \"\\Q\\E\" pair. If protected by \"\\\", \"$\" will be\nquoted to become \"\\\\\\$\"; if not, it is interpreted as the start of an interpolated\nscalar.\n\nNote also that the interpolation code needs to make a decision on where the\ninterpolated scalar ends. For instance, whether \"a$x->{c}\" really means:\n\n\"a \" . $x . \" -> {c}\";\n\nor:\n\n\"a \" . $x -> {c};\n\nMost of the time, the longest possible text that does not include spaces between\ncomponents and which contains matching braces or brackets. because the outcome may\nbe determined by voting based on heuristic estimators, the result is not strictly\npredictable. Fortunately, it's usually correct for ambiguous cases.\n\nthe replacement of \"s///\"\nProcessing of \"\\Q\", \"\\U\", \"\\u\", \"\\L\", \"\\l\", \"\\F\" and interpolation happens as with\n\"qq//\" constructs.\n\nIt is at this step that \"\\1\" is begrudgingly converted to $1 in the replacement text\nof \"s///\", in order to correct the incorrigible sed hackers who haven't picked up the\nsaner idiom yet. A warning is emitted if the \"usewarnings\" pragma or the -w command-\nline flag (that is, the $^W variable) was set.\n\n\"RE\" in \"m?RE?\", \"/RE/\", \"m/RE/\", \"s/RE/foo/\",\nProcessing of \"\\Q\", \"\\U\", \"\\u\", \"\\L\", \"\\l\", \"\\F\", \"\\E\", and interpolation happens\n(almost) as with \"qq//\" constructs.\n\nProcessing of \"\\N{...}\" is also done here, and compiled into an intermediate form for\nthe regex compiler. (This is because, as mentioned below, the regex compilation may\nbe done at execution time, and \"\\N{...}\" is a compile-time construct.)\n\nHowever any other combinations of \"\\\" followed by a character are not substituted but\nonly skipped, in order to parse them as regular expressions at the following step.\nAs \"\\c\" is skipped at this step, \"@\" of \"\\c@\" in RE is possibly treated as an array\nsymbol (for example @foo), even though the same text in \"qq//\" gives interpolation of\n\"\\c@\".\n\nCode blocks such as \"(?{BLOCK})\" are handled by temporarily passing control back to\nthe perl parser, in a similar way that an interpolated array subscript expression\nsuch as \"foo$array[1+f(\"[xyz\")]bar\" would be.\n\nMoreover, inside \"(?{BLOCK})\", \"(?#comment)\", and a \"#\"-comment in a \"/x\"-regular\nexpression, no processing is performed whatsoever. This is the first step at which\nthe presence of the \"/x\" modifier is relevant.\n\nInterpolation in patterns has several quirks: $|, $(, $), \"@+\" and \"@-\" are not\ninterpolated, and constructs $var[SOMETHING] are voted (by several different\nestimators) to be either an array element or $var followed by an RE alternative.\nThis is where the notation \"${arr[$bar]}\" comes handy: \"/${arr[0-9]}/\" is interpreted\nas array element \"-9\", not as a regular expression from the variable $arr followed by\na digit, which would be the interpretation of \"/$arr[0-9]/\". Since voting among\ndifferent estimators may occur, the result is not predictable.\n\nThe lack of processing of \"\\\\\" creates specific restrictions on the post-processed\ntext. If the delimiter is \"/\", one cannot get the combination \"\\/\" into the result\nof this step. \"/\" will finish the regular expression, \"\\/\" will be stripped to \"/\"\non the previous step, and \"\\\\/\" will be left as is. Because \"/\" is equivalent to\n\"\\/\" inside a regular expression, this does not matter unless the delimiter happens\nto be character special to the RE engine, such as in \"s*foo*bar*\", \"m[foo]\", or\n\"m?foo?\"; or an alphanumeric char, as in:\n\nm m ^ a \\s* b mmx;\n\nIn the RE above, which is intentionally obfuscated for illustration, the delimiter is\n\"m\", the modifier is \"mx\", and after delimiter-removal the RE is the same as for\n\"m/^a\\s*b/mx\". There's more than one reason you're encouraged to restrict your\ndelimiters to non-alphanumeric, non-whitespace choices.\n\nThis step is the last one for all constructs except regular expressions, which are\nprocessed further.\n\nparsing regular expressions\nPrevious steps were performed during the compilation of Perl code, but this one happens\nat run time, although it may be optimized to be calculated at compile time if\nappropriate. After preprocessing described above, and possibly after evaluation if\nconcatenation, joining, casing translation, or metaquoting are involved, the resulting\nstring is passed to the RE engine for compilation.\n\nWhatever happens in the RE engine might be better discussed in perlre, but for the sake\nof continuity, we shall do so here.\n\nThis is another step where the presence of the \"/x\" modifier is relevant. The RE engine\nscans the string from left to right and converts it into a finite automaton.\n\nBackslashed characters are either replaced with corresponding literal strings (as with\n\"\\{\"), or else they generate special nodes in the finite automaton (as with \"\\b\").\nCharacters special to the RE engine (such as \"|\") generate corresponding nodes or groups\nof nodes. \"(?#...)\" comments are ignored. All the rest is either converted to literal\nstrings to match, or else is ignored (as is whitespace and \"#\"-style comments if \"/x\" is\npresent).\n\nParsing of the bracketed character class construct, \"[...]\", is rather different than the\nrule used for the rest of the pattern. The terminator of this construct is found using\nthe same rules as for finding the terminator of a \"{}\"-delimited construct, the only\nexception being that \"]\" immediately following \"[\" is treated as though preceded by a\nbackslash.\n\nThe terminator of runtime \"(?{...})\" is found by temporarily switching control to the\nperl parser, which should stop at the point where the logically balancing terminating \"}\"\nis found.\n\nIt is possible to inspect both the string given to RE engine and the resulting finite\nautomaton. See the arguments \"debug\"/\"debugcolor\" in the \"usere\" pragma, as well as\nPerl's -Dr command-line switch documented in \"Command Switches\" in perlrun.\n\nOptimization of regular expressions\nThis step is listed for completeness only. Since it does not change semantics, details\nof this step are not documented and are subject to change without notice. This step is\nperformed over the finite automaton that was generated during the previous pass.\n\nIt is at this stage that \"split()\" silently optimizes \"/^/\" to mean \"/^/m\".\n\nI/O Operators\nThere are several I/O operators you should know about.\n\nA string enclosed by backticks (grave accents) first undergoes double-quote interpolation.\nIt is then interpreted as an external command, and the output of that command is the value of\nthe backtick string, like in a shell. In scalar context, a single string consisting of all\noutput is returned. In list context, a list of values is returned, one per line of output.\n(You can set $/ to use a different line terminator.) The command is executed each time the\npseudo-literal is evaluated. The status value of the command is returned in $? (see perlvar\nfor the interpretation of $?). Unlike in csh, no translation is done on the return\ndata--newlines remain newlines. Unlike in any of the shells, single quotes do not hide\nvariable names in the command from interpretation. To pass a literal dollar-sign through to\nthe shell you need to hide it with a backslash. The generalized form of backticks is \"qx//\",\nor you can call the \"readpipe\" in perlfunc function. (Because backticks always undergo shell\nexpansion as well, see perlsec for security concerns.)\n\nIn scalar context, evaluating a filehandle in angle brackets yields the next line from that\nfile (the newline, if any, included), or \"undef\" at end-of-file or on error. When $/ is set\nto \"undef\" (sometimes known as file-slurp mode) and the file is empty, it returns '' the\nfirst time, followed by \"undef\" subsequently.\n\nOrdinarily you must assign the returned value to a variable, but there is one situation where\nan automatic assignment happens. If and only if the input symbol is the only thing inside\nthe conditional of a \"while\" statement (even if disguised as a \"for(;;)\" loop), the value is\nautomatically assigned to the global variable $, destroying whatever was there previously.\n(This may seem like an odd thing to you, but you'll use the construct in almost every Perl\nscript you write.) The $ variable is not implicitly localized. You'll have to put a\n\"local$;\" before the loop if you want that to happen. Furthermore, if the input symbol or\nan explicit assignment of the input symbol to a scalar is used as a \"while\"/\"for\" condition,\nthen the condition actually tests for definedness of the expression's value, not for its\nregular truth value.\n\nThus the following lines are equivalent:\n\nwhile (defined($ = )) { print; }\nwhile ($ = ) { print; }\nwhile () { print; }\nfor (;;) { print; }\nprint while defined($ = );\nprint while ($ = );\nprint while ;\n\nThis also behaves similarly, but assigns to a lexical variable instead of to $:\n\nwhile (my $line = ) { print $line }\n\nIn these loop constructs, the assigned value (whether assignment is automatic or explicit) is\nthen tested to see whether it is defined. The defined test avoids problems where the line\nhas a string value that would be treated as false by Perl; for example a \"\" or a \"0\" with no\ntrailing newline. If you really mean for such values to terminate the loop, they should be\ntested for explicitly:\n\nwhile (($ = ) ne '0') { ... }\nwhile () { last unless $; ... }\n\nIn other boolean contexts, \"\" without an explicit \"defined\" test or comparison\nelicits a warning if the \"usewarnings\" pragma or the -w command-line switch (the $^W\nvariable) is in effect.\n\nThe filehandles STDIN, STDOUT, and STDERR are predefined. (The filehandles \"stdin\",\n\"stdout\", and \"stderr\" will also work except in packages, where they would be interpreted as\nlocal identifiers rather than global.) Additional filehandles may be created with the\n\"open()\" function, amongst others. See perlopentut and \"open\" in perlfunc for details on\nthis.\n\nIf a \"\" is used in a context that is looking for a list, a list comprising all\ninput lines is returned, one line per list element. It's easy to grow to a rather large data\nspace this way, so use with care.\n\n\"\" may also be spelled \"readline(*FILEHANDLE)\". See \"readline\" in perlfunc.\n\nThe null filehandle \"<>\" (sometimes called the diamond operator) is special: it can be used\nto emulate the behavior of sed and awk, and any other Unix filter program that takes a list\nof filenames, doing the same to each line of input from all of them. Input from \"<>\" comes\neither from standard input, or from each file listed on the command line. Here's how it\nworks: the first time \"<>\" is evaluated, the @ARGV array is checked, and if it is empty,\n$ARGV[0] is set to \"-\", which when opened gives you standard input. The @ARGV array is then\nprocessed as a list of filenames. The loop\n\nwhile (<>) {\n... # code for each line\n}\n\nis equivalent to the following Perl-like pseudo code:\n\nunshift(@ARGV, '-') unless @ARGV;\nwhile ($ARGV = shift) {\nopen(ARGV, $ARGV);\nwhile () {\n... # code for each line\n}\n}\n\nexcept that it isn't so cumbersome to say, and will actually work. It really does shift the\n@ARGV array and put the current filename into the $ARGV variable. It also uses filehandle\nARGV internally. \"<>\" is just a synonym for \"\", which is magical. (The pseudo code\nabove doesn't work because it treats \"\" as non-magical.)\n\nSince the null filehandle uses the two argument form of \"open\" in perlfunc it interprets\nspecial characters, so if you have a script like this:\n\nwhile (<>) {\nprint;\n}\n\nand call it with \"perldangerous.pl'rm-rfv*|'\", it actually opens a pipe, executes the \"rm\"\ncommand and reads \"rm\"'s output from that pipe. If you want all items in @ARGV to be\ninterpreted as file names, you can use the module \"ARGV::readonly\" from CPAN, or use the\ndouble diamond bracket:\n\nwhile (<<>>) {\nprint;\n}\n\nUsing double angle brackets inside of a while causes the open to use the three argument form\n(with the second argument being \"<\"), so all arguments in \"ARGV\" are treated as literal\nfilenames (including \"-\"). (Note that for convenience, if you use \"<<>>\" and if @ARGV is\nempty, it will still read from the standard input.)\n\nYou can modify @ARGV before the first \"<>\" as long as the array ends up containing the list\nof filenames you really want. Line numbers ($.) continue as though the input were one big\nhappy file. See the example in \"eof\" in perlfunc for how to reset line numbers on each file.\n\nIf you want to set @ARGV to your own list of files, go right ahead. This sets @ARGV to all\nplain text files if no @ARGV was given:\n\n@ARGV = grep { -f && -T } glob('*') unless @ARGV;\n\nYou can even set them to pipe commands. For example, this automatically filters compressed\narguments through gzip:\n\n@ARGV = map { /\\.(gz|Z)$/ ? \"gzip -dc < $ |\" : $ } @ARGV;\n\nIf you want to pass switches into your script, you can use one of the \"Getopts\" modules or\nput a loop on the front like this:\n\nwhile ($ = $ARGV[0], /^-/) {\nshift;\nlast if /^--$/;\nif (/^-D(.*)/) { $debug = $1 }\nif (/^-v/) { $verbose++ }\n# ... # other switches\n}\n\nwhile (<>) {\n# ... # code for each line\n}\n\nThe \"<>\" symbol will return \"undef\" for end-of-file only once. If you call it again after\nthis, it will assume you are processing another @ARGV list, and if you haven't set @ARGV,\nwill read input from STDIN.\n\nIf what the angle brackets contain is a simple scalar variable (for example, $foo), then that\nvariable contains the name of the filehandle to input from, or its typeglob, or a reference\nto the same. For example:\n\n$fh = \\*STDIN;\n$line = <$fh>;\n\nIf what's within the angle brackets is neither a filehandle nor a simple scalar variable\ncontaining a filehandle name, typeglob, or typeglob reference, it is interpreted as a\nfilename pattern to be globbed, and either a list of filenames or the next filename in the\nlist is returned, depending on context. This distinction is determined on syntactic grounds\nalone. That means \"<$x>\" is always a \"readline()\" from an indirect handle, but\n\"<$hash{key}>\" is always a \"glob()\". That's because $x is a simple scalar variable, but\n$hash{key} is not--it's a hash element. Even \"<$x >\" (note the extra space) is treated as\n\"glob(\"$x \")\", not \"readline($x)\".\n\nOne level of double-quote interpretation is done first, but you can't say \"<$foo>\" because\nthat's an indirect filehandle as explained in the previous paragraph. (In older versions of\nPerl, programmers would insert curly brackets to force interpretation as a filename glob:\n\"<${foo}>\". These days, it's considered cleaner to call the internal function directly as\n\"glob($foo)\", which is probably the right way to have done it in the first place.) For\nexample:\n\nwhile (<*.c>) {\nchmod 0644, $;\n}\n\nis roughly equivalent to:\n\nopen(FOO, \"echo *.c | tr -s ' \\t\\r\\f' '\\\\012\\\\012\\\\012\\\\012'|\");\nwhile () {\nchomp;\nchmod 0644, $;\n}\n\nexcept that the globbing is actually done internally using the standard \"File::Glob\"\nextension. Of course, the shortest way to do the above is:\n\nchmod 0644, <*.c>;\n\nA (file)glob evaluates its (embedded) argument only when it is starting a new list. All\nvalues must be read before it will start over. In list context, this isn't important because\nyou automatically get them all anyway. However, in scalar context the operator returns the\nnext value each time it's called, or \"undef\" when the list has run out. As with filehandle\nreads, an automatic \"defined\" is generated when the glob occurs in the test part of a\n\"while\", because legal glob returns (for example, a file called 0) would otherwise terminate\nthe loop. Again, \"undef\" is returned only once. So if you're expecting a single value from\na glob, it is much better to say\n\n($file) = ;\n\nthan\n\n$file = ;\n\nbecause the latter will alternate between returning a filename and returning false.\n\nIf you're trying to do variable interpolation, it's definitely better to use the \"glob()\"\nfunction, because the older notation can cause people to become confused with the indirect\nfilehandle notation.\n\n@files = glob(\"$dir/*.[ch]\");\n@files = glob($files[$i]);\n\nIf an angle-bracket-based globbing expression is used as the condition of a \"while\" or \"for\"\nloop, then it will be implicitly assigned to $. If either a globbing expression or an\nexplicit assignment of a globbing expression to a scalar is used as a \"while\"/\"for\"\ncondition, then the condition actually tests for definedness of the expression's value, not\nfor its regular truth value.\n\nConstant Folding\nLike C, Perl does a certain amount of expression evaluation at compile time whenever it\ndetermines that all arguments to an operator are static and have no side effects. In\nparticular, string concatenation happens at compile time between literals that don't do\nvariable substitution. Backslash interpolation also happens at compile time. You can say\n\n'Now is the time for all'\n. \"\\n\"\n. 'good men to come to.'\n\nand this all reduces to one string internally. Likewise, if you say\n\nforeach $file (@filenames) {\nif (-s $file > 5 + 100 * 216) { }\n}\n\nthe compiler precomputes the number which that expression represents so that the interpreter\nwon't have to.\n\nNo-ops\nPerl doesn't officially have a no-op operator, but the bare constants 0 and 1 are special-\ncased not to produce a warning in void context, so you can for example safely do\n\n1 while foo();\n\nBitwise String Operators\nBitstrings of any size may be manipulated by the bitwise operators (\"~ | & ^\").\n\nIf the operands to a binary bitwise op are strings of different sizes, | and ^ ops act as\nthough the shorter operand had additional zero bits on the right, while the & op acts as\nthough the longer operand were truncated to the length of the shorter. The granularity for\nsuch extension or truncation is one or more bytes.\n\n# ASCII-based examples\nprint \"j p \\n\" ^ \" a h\"; # prints \"JAPH\\n\"\nprint \"JA\" | \" ph\\n\"; # prints \"japh\\n\"\nprint \"japh\\nJunk\" & ''; # prints \"JAPH\\n\";\nprint 'p N$' ^ \" E>\") always produce integral\nresults. (But see also \"Bitwise String Operators\".) However, \"useinteger\" still has meaning\nfor them. By default, their results are interpreted as unsigned integers, but if\n\"useinteger\" is in effect, their results are interpreted as signed integers. For example,\n\"~0\" usually evaluates to a large integral value. However, \"useinteger;~0\" is \"-1\" on two's-\ncomplement machines.\n\nFloating-point Arithmetic\nWhile \"useinteger\" provides integer-only arithmetic, there is no analogous mechanism to\nprovide automatic rounding or truncation to a certain number of decimal places. For rounding\nto a certain number of digits, \"sprintf()\" or \"printf()\" is usually the easiest route. See\nperlfaq4.\n\nFloating-point numbers are only approximations to what a mathematician would call real\nnumbers. There are infinitely more reals than floats, so some corners must be cut. For\nexample:\n\nprintf \"%.20g\\n\", 123456789123456789;\n# produces 123456789123456784\n\nTesting for exact floating-point equality or inequality is not a good idea. Here's a\n(relatively expensive) work-around to compare whether two floating-point numbers are equal to\na particular number of decimal places. See Knuth, volume II, for a more robust treatment of\nthis topic.\n\nsub fpequal {\nmy ($X, $Y, $POINTS) = @;\nmy ($tX, $tY);\n$tX = sprintf(\"%.${POINTS}g\", $X);\n$tY = sprintf(\"%.${POINTS}g\", $Y);\nreturn $tX eq $tY;\n}\n\nThe POSIX module (part of the standard perl distribution) implements \"ceil()\", \"floor()\", and\nother mathematical and trigonometric functions. The \"Math::Complex\" module (part of the\nstandard perl distribution) defines mathematical functions that work on both the reals and\nthe imaginary numbers. \"Math::Complex\" is not as efficient as POSIX, but POSIX can't work\nwith complex numbers.\n\nRounding in financial applications can have serious implications, and the rounding method\nused should be specified precisely. In these cases, it probably pays not to trust whichever\nsystem rounding is being used by Perl, but to instead implement the rounding function you\nneed yourself.\n\nBigger Numbers\nThe standard \"Math::BigInt\", \"Math::BigRat\", and \"Math::BigFloat\" modules, along with the\n\"bignum\", \"bigint\", and \"bigrat\" pragmas, provide variable-precision arithmetic and\noverloaded operators, although they're currently pretty slow. At the cost of some space and\nconsiderable speed, they avoid the normal pitfalls associated with limited-precision\nrepresentations.\n\nuse 5.010;\nuse bigint; # easy interface to Math::BigInt\n$x = 123456789123456789;\nsay $x * $x;\n+15241578780673678515622620750190521\n\nOr with rationals:\n\nuse 5.010;\nuse bigrat;\n$x = 3/22;\n$y = 4/6;\nsay \"x/y is \", $x/$y;\nsay \"x*y is \", $x*$y;\nx/y is 9/44\nx*y is 1/11\n\nSeveral modules let you calculate with unlimited or fixed precision (bound only by memory and\nCPU time). There are also some non-standard modules that provide faster implementations via\nexternal C libraries.\n\nHere is a short, but incomplete summary:\n\nMath::String treat string sequences like numbers\nMath::FixedPrecision calculate with a fixed precision\nMath::Currency for currency calculations\nBit::Vector manipulate bit vectors fast (uses C)\nMath::BigIntFast Bit::Vector wrapper for big numbers\nMath::Pari provides access to the Pari C library\nMath::Cephes uses the external Cephes C library (no\nbig numbers)\nMath::Cephes::Fraction fractions via the Cephes library\nMath::GMP another one using an external C library\nMath::GMPz an alternative interface to libgmp's big ints\nMath::GMPq an interface to libgmp's fraction numbers\nMath::GMPf an interface to libgmp's floating point numbers\n\nChoose wisely.\n\nperl v5.34.0 2025-07-25 PERLOP(1)", "subsections": [] } } } }