1Switch(3pm) Perl Programmers Reference Guide Switch(3pm)
2
3
4
6 Switch - A switch statement for Perl
7
9 This document describes version 2.14 of Switch, released Dec 29, 2008.
10
12 use Switch;
13
14 switch ($val) {
15 case 1 { print "number 1" }
16 case "a" { print "string a" }
17 case [1..10,42] { print "number in list" }
18 case (\@array) { print "number in list" }
19 case /\w+/ { print "pattern" }
20 case qr/\w+/ { print "pattern" }
21 case (\%hash) { print "entry in hash" }
22 case (\&sub) { print "arg to subroutine" }
23 else { print "previous case not true" }
24 }
25
27 [Skip ahead to "DESCRIPTION" if you don't care about the whys and
28 wherefores of this control structure]
29
30 In seeking to devise a "Swiss Army" case mechanism suitable for Perl,
31 it is useful to generalize this notion of distributed conditional
32 testing as far as possible. Specifically, the concept of "matching"
33 between the switch value and the various case values need not be
34 restricted to numeric (or string or referential) equality, as it is in
35 other languages. Indeed, as Table 1 illustrates, Perl offers at least
36 eighteen different ways in which two values could generate a match.
37
38 Table 1: Matching a switch value ($s) with a case value ($c)
39
40 Switch Case Type of Match Implied Matching Code
41 Value Value
42 ====== ===== ===================== =============
43
44 number same numeric or referential match if $s == $c;
45 or ref equality
46
47 object method result of method call match if $s->$c();
48 ref name match if defined $s->$c();
49 or ref
50
51 other other string equality match if $s eq $c;
52 non-ref non-ref
53 scalar scalar
54
55 string regexp pattern match match if $s =~ /$c/;
56
57 array scalar array entry existence match if 0<=$c && $c<@$s;
58 ref array entry definition match if defined $s->[$c];
59 array entry truth match if $s->[$c];
60
61 array array array intersection match if intersects(@$s, @$c);
62 ref ref (apply this table to
63 all pairs of elements
64 $s->[$i] and
65 $c->[$j])
66
67 array regexp array grep match if grep /$c/, @$s;
68 ref
69
70 hash scalar hash entry existence match if exists $s->{$c};
71 ref hash entry definition match if defined $s->{$c};
72 hash entry truth match if $s->{$c};
73
74 hash regexp hash grep match if grep /$c/, keys %$s;
75 ref
76
77 sub scalar return value defn match if defined $s->($c);
78 ref return value truth match if $s->($c);
79
80 sub array return value defn match if defined $s->(@$c);
81 ref ref return value truth match if $s->(@$c);
82
83 In reality, Table 1 covers 31 alternatives, because only the equality
84 and intersection tests are commutative; in all other cases, the roles
85 of the $s and $c variables could be reversed to produce a different
86 test. For example, instead of testing a single hash for the existence
87 of a series of keys ("match if exists $s->{$c}"), one could test for
88 the existence of a single key in a series of hashes ("match if exists
89 $c->{$s}").
90
92 The Switch.pm module implements a generalized case mechanism that
93 covers most (but not all) of the numerous possible combinations of
94 switch and case values described above.
95
96 The module augments the standard Perl syntax with two new control
97 statements: "switch" and "case". The "switch" statement takes a single
98 scalar argument of any type, specified in parentheses. "switch" stores
99 this value as the current switch value in a (localized) control
100 variable. The value is followed by a block which may contain one or
101 more Perl statements (including the "case" statement described below).
102 The block is unconditionally executed once the switch value has been
103 cached.
104
105 A "case" statement takes a single scalar argument (in mandatory
106 parentheses if it's a variable; otherwise the parens are optional) and
107 selects the appropriate type of matching between that argument and the
108 current switch value. The type of matching used is determined by the
109 respective types of the switch value and the "case" argument, as
110 specified in Table 1. If the match is successful, the mandatory block
111 associated with the "case" statement is executed.
112
113 In most other respects, the "case" statement is semantically identical
114 to an "if" statement. For example, it can be followed by an "else"
115 clause, and can be used as a postfix statement qualifier.
116
117 However, when a "case" block has been executed control is automatically
118 transferred to the statement after the immediately enclosing "switch"
119 block, rather than to the next statement within the block. In other
120 words, the success of any "case" statement prevents other cases in the
121 same scope from executing. But see "Allowing fall-through" below.
122
123 Together these two new statements provide a fully generalized case
124 mechanism:
125
126 use Switch;
127
128 # AND LATER...
129
130 %special = ( woohoo => 1, d'oh => 1 );
131
132 while (<>) {
133 chomp;
134 switch ($_) {
135 case (%special) { print "homer\n"; } # if $special{$_}
136 case /[a-z]/i { print "alpha\n"; } # if $_ =~ /a-z/i
137 case [1..9] { print "small num\n"; } # if $_ in [1..9]
138 case { $_[0] >= 10 } { print "big num\n"; } # if $_ >= 10
139 print "must be punctuation\n" case /\W/; # if $_ ~= /\W/
140 }
141 }
142
143 Note that "switch"es can be nested within "case" (or any other) blocks,
144 and a series of "case" statements can try different types of matches --
145 hash membership, pattern match, array intersection, simple equality,
146 etc. -- against the same switch value.
147
148 The use of intersection tests against an array reference is
149 particularly useful for aggregating integral cases:
150
151 sub classify_digit
152 {
153 switch ($_[0]) { case 0 { return 'zero' }
154 case [2,4,6,8] { return 'even' }
155 case [1,3,5,7,9] { return 'odd' }
156 case /[A-F]/i { return 'hex' }
157 }
158 }
159
160 Allowing fall-through
161 Fall-though (trying another case after one has already succeeded) is
162 usually a Bad Idea in a switch statement. However, this is Perl, not a
163 police state, so there is a way to do it, if you must.
164
165 If a "case" block executes an untargeted "next", control is immediately
166 transferred to the statement after the "case" statement (i.e. usually
167 another case), rather than out of the surrounding "switch" block.
168
169 For example:
170
171 switch ($val) {
172 case 1 { handle_num_1(); next } # and try next case...
173 case "1" { handle_str_1(); next } # and try next case...
174 case [0..9] { handle_num_any(); } # and we're done
175 case /\d/ { handle_dig_any(); next } # and try next case...
176 case /.*/ { handle_str_any(); next } # and try next case...
177 }
178
179 If $val held the number 1, the above "switch" block would call the
180 first three "handle_..." subroutines, jumping to the next case test
181 each time it encountered a "next". After the third "case" block was
182 executed, control would jump to the end of the enclosing "switch"
183 block.
184
185 On the other hand, if $val held 10, then only the last two "handle_..."
186 subroutines would be called.
187
188 Note that this mechanism allows the notion of conditional fall-through.
189 For example:
190
191 switch ($val) {
192 case [0..9] { handle_num_any(); next if $val < 7; }
193 case /\d/ { handle_dig_any(); }
194 }
195
196 If an untargeted "last" statement is executed in a case block, this
197 immediately transfers control out of the enclosing "switch" block (in
198 other words, there is an implicit "last" at the end of each normal
199 "case" block). Thus the previous example could also have been written:
200
201 switch ($val) {
202 case [0..9] { handle_num_any(); last if $val >= 7; next; }
203 case /\d/ { handle_dig_any(); }
204 }
205
206 Automating fall-through
207 In situations where case fall-through should be the norm, rather than
208 an exception, an endless succession of terminal "next"s is tedious and
209 ugly. Hence, it is possible to reverse the default behaviour by
210 specifying the string "fallthrough" when importing the module. For
211 example, the following code is equivalent to the first example in
212 "Allowing fall-through":
213
214 use Switch 'fallthrough';
215
216 switch ($val) {
217 case 1 { handle_num_1(); }
218 case "1" { handle_str_1(); }
219 case [0..9] { handle_num_any(); last }
220 case /\d/ { handle_dig_any(); }
221 case /.*/ { handle_str_any(); }
222 }
223
224 Note the explicit use of a "last" to preserve the non-fall-through
225 behaviour of the third case.
226
227 Alternative syntax
228 Perl 6 will provide a built-in switch statement with essentially the
229 same semantics as those offered by Switch.pm, but with a different pair
230 of keywords. In Perl 6 "switch" will be spelled "given", and "case"
231 will be pronounced "when". In addition, the "when" statement will not
232 require switch or case values to be parenthesized.
233
234 This future syntax is also (largely) available via the Switch.pm
235 module, by importing it with the argument "Perl6". For example:
236
237 use Switch 'Perl6';
238
239 given ($val) {
240 when 1 { handle_num_1(); }
241 when ($str1) { handle_str_1(); }
242 when [0..9] { handle_num_any(); last }
243 when /\d/ { handle_dig_any(); }
244 when /.*/ { handle_str_any(); }
245 default { handle anything else; }
246 }
247
248 Note that scalars still need to be parenthesized, since they would be
249 ambiguous in Perl 5.
250
251 Note too that you can mix and match both syntaxes by importing the
252 module with:
253
254 use Switch 'Perl5', 'Perl6';
255
256 Higher-order Operations
257 One situation in which "switch" and "case" do not provide a good
258 substitute for a cascaded "if", is where a switch value needs to be
259 tested against a series of conditions. For example:
260
261 sub beverage {
262 switch (shift) {
263 case { $_[0] < 10 } { return 'milk' }
264 case { $_[0] < 20 } { return 'coke' }
265 case { $_[0] < 30 } { return 'beer' }
266 case { $_[0] < 40 } { return 'wine' }
267 case { $_[0] < 50 } { return 'malt' }
268 case { $_[0] < 60 } { return 'Moet' }
269 else { return 'milk' }
270 }
271 }
272
273 (This is equivalent to writing "case (sub { $_[0] < 10 })", etc.; $_[0]
274 is the argument to the anonymous subroutine.)
275
276 The need to specify each condition as a subroutine block is tiresome.
277 To overcome this, when importing Switch.pm, a special "placeholder"
278 subroutine named "__" [sic] may also be imported. This subroutine
279 converts (almost) any expression in which it appears to a reference to
280 a higher-order function. That is, the expression:
281
282 use Switch '__';
283
284 __ < 2
285
286 is equivalent to:
287
288 sub { $_[0] < 2 }
289
290 With "__", the previous ugly case statements can be rewritten:
291
292 case __ < 10 { return 'milk' }
293 case __ < 20 { return 'coke' }
294 case __ < 30 { return 'beer' }
295 case __ < 40 { return 'wine' }
296 case __ < 50 { return 'malt' }
297 case __ < 60 { return 'Moet' }
298 else { return 'milk' }
299
300 The "__" subroutine makes extensive use of operator overloading to
301 perform its magic. All operations involving __ are overloaded to
302 produce an anonymous subroutine that implements a lazy version of the
303 original operation.
304
305 The only problem is that operator overloading does not allow the
306 boolean operators "&&" and "||" to be overloaded. So a case statement
307 like this:
308
309 case 0 <= __ && __ < 10 { return 'digit' }
310
311 doesn't act as expected, because when it is executed, it constructs two
312 higher order subroutines and then treats the two resulting references
313 as arguments to "&&":
314
315 sub { 0 <= $_[0] } && sub { $_[0] < 10 }
316
317 This boolean expression is inevitably true, since both references are
318 non-false. Fortunately, the overloaded 'bool' operator catches this
319 situation and flags it as an error.
320
322 The module is implemented using Filter::Util::Call and Text::Balanced
323 and requires both these modules to be installed.
324
326 Damian Conway (damian@conway.org). This module is now maintained by
327 Rafael Garcia-Suarez (rgarciasuarez@gmail.com) and more generally by
328 the Perl 5 Porters (perl5-porters@perl.org), as part of the Perl core.
329
331 There are undoubtedly serious bugs lurking somewhere in code this funky
332 :-) Bug reports and other feedback are most welcome.
333
335 Due to the heuristic nature of Switch.pm's source parsing, the presence
336 of regexes with embedded newlines that are specified with raw "/.../"
337 delimiters and don't have a modifier "//x" are indistinguishable from
338 code chunks beginning with the division operator "/". As a workaround
339 you must use "m/.../" or "m?...?" for such patterns. Also, the presence
340 of regexes specified with raw "?...?" delimiters may cause mysterious
341 errors. The workaround is to use "m?...?" instead.
342
343 Due to the way source filters work in Perl, you can't use Switch inside
344 an string "eval".
345
346 If your source file is longer then 1 million characters and you have a
347 switch statement that crosses the 1 million (or 2 million, etc.)
348 character boundary you will get mysterious errors. The workaround is to
349 use smaller source files.
350
352 Copyright (c) 1997-2008, Damian Conway. All Rights Reserved.
353 This module is free software. It may be used, redistributed
354 and/or modified under the same terms as Perl itself.
355
356
357
358perl v5.10.1 2009-03-30 Switch(3pm)