Commit | Line | Data |
---|---|---|
86530b38 AT |
1 | package Switch; |
2 | ||
3 | use strict; | |
4 | use vars qw($VERSION); | |
5 | use Carp; | |
6 | ||
7 | $VERSION = '2.09'; | |
8 | ||
9 | ||
10 | # LOAD FILTERING MODULE... | |
11 | use Filter::Util::Call; | |
12 | ||
13 | sub __(); | |
14 | ||
15 | # CATCH ATTEMPTS TO CALL case OUTSIDE THE SCOPE OF ANY switch | |
16 | ||
17 | $::_S_W_I_T_C_H = sub { croak "case/when statement not in switch/given block" }; | |
18 | ||
19 | my $offset; | |
20 | my $fallthrough; | |
21 | my ($Perl5, $Perl6) = (0,0); | |
22 | ||
23 | sub import | |
24 | { | |
25 | $fallthrough = grep /\bfallthrough\b/, @_; | |
26 | $offset = (caller)[2]+1; | |
27 | filter_add({}) unless @_>1 && $_[1] eq 'noimport'; | |
28 | my $pkg = caller; | |
29 | no strict 'refs'; | |
30 | for ( qw( on_defined on_exists ) ) | |
31 | { | |
32 | *{"${pkg}::$_"} = \&$_; | |
33 | } | |
34 | *{"${pkg}::__"} = \&__ if grep /__/, @_; | |
35 | $Perl6 = 1 if grep(/Perl\s*6/i, @_); | |
36 | $Perl5 = 1 if grep(/Perl\s*5/i, @_) || !grep(/Perl\s*6/i, @_); | |
37 | 1; | |
38 | } | |
39 | ||
40 | sub unimport | |
41 | { | |
42 | filter_del() | |
43 | } | |
44 | ||
45 | sub filter | |
46 | { | |
47 | my($self) = @_ ; | |
48 | local $Switch::file = (caller)[1]; | |
49 | ||
50 | my $status = 1; | |
51 | $status = filter_read(10_000); | |
52 | return $status if $status<0; | |
53 | $_ = filter_blocks($_,$offset); | |
54 | $_ = "# line $offset\n" . $_ if $offset; undef $offset; | |
55 | return $status; | |
56 | } | |
57 | ||
58 | use Text::Balanced ':ALL'; | |
59 | ||
60 | sub line | |
61 | { | |
62 | my ($pretext,$offset) = @_; | |
63 | ($pretext=~tr/\n/\n/)+($offset||0); | |
64 | } | |
65 | ||
66 | sub is_block | |
67 | { | |
68 | local $SIG{__WARN__}=sub{die$@}; | |
69 | local $^W=1; | |
70 | my $ishash = defined eval 'my $hr='.$_[0]; | |
71 | undef $@; | |
72 | return !$ishash; | |
73 | } | |
74 | ||
75 | ||
76 | my $EOP = qr/\n\n|\Z/; | |
77 | my $CUT = qr/\n=cut.*$EOP/; | |
78 | my $pod_or_DATA = qr/ ^=(?:head[1-4]|item) .*? $CUT | |
79 | | ^=pod .*? $CUT | |
80 | | ^=for .*? $EOP | |
81 | | ^=begin \s* (\S+) .*? \n=end \s* \1 .*? $EOP | |
82 | | ^__(DATA|END)__\n.* | |
83 | /smx; | |
84 | ||
85 | my $casecounter = 1; | |
86 | sub filter_blocks | |
87 | { | |
88 | my ($source, $line) = @_; | |
89 | return $source unless $Perl5 && $source =~ /case|switch/ | |
90 | || $Perl6 && $source =~ /when|given/; | |
91 | pos $source = 0; | |
92 | my $text = ""; | |
93 | $DB::single = 1; | |
94 | component: while (pos $source < length $source) | |
95 | { | |
96 | if ($source =~ m/(\G\s*use\s+Switch\b)/gc) | |
97 | { | |
98 | $text .= q{use Switch 'noimport'}; | |
99 | next component; | |
100 | } | |
101 | my @pos = Text::Balanced::_match_quotelike(\$source,qr/\s*/,1,0); | |
102 | if (defined $pos[0]) | |
103 | { | |
104 | $text .= " " . substr($source,$pos[2],$pos[18]-$pos[2]); | |
105 | next component; | |
106 | } | |
107 | if ($source =~ m/\G\s*($pod_or_DATA)/gc) { | |
108 | next component; | |
109 | } | |
110 | @pos = Text::Balanced::_match_variable(\$source,qr/\s*/); | |
111 | if (defined $pos[0]) | |
112 | { | |
113 | $text .= " " . substr($source,$pos[0],$pos[4]-$pos[0]); | |
114 | next component; | |
115 | } | |
116 | ||
117 | if ($Perl5 && $source =~ m/\G(\n*)(\s*)(switch)\b(?=\s*[(])/gc | |
118 | || $Perl6 && $source =~ m/\G(\n*)(\s*)(given)\b(?=\s*[(])/gc | |
119 | || $Perl6 && $source =~ m/\G(\n*)(\s*)(given)\b(.*)(?=\{)/gc) | |
120 | { | |
121 | my $keyword = $3; | |
122 | my $arg = $4; | |
123 | # print STDERR "[$arg]\n"; | |
124 | $text .= $1.$2.'S_W_I_T_C_H: while (1) '; | |
125 | unless ($arg) { | |
126 | @pos = Text::Balanced::_match_codeblock(\$source,qr/\s*/,qr/\(/,qr/\)/,qr/[[{(<]/,qr/[]})>]/,undef) | |
127 | or do { | |
128 | die "Bad $keyword statement (problem in the parentheses?) near $Switch::file line ", line(substr($source,0,pos $source),$line), "\n"; | |
129 | }; | |
130 | $arg = filter_blocks(substr($source,$pos[0],$pos[4]-$pos[0]),line(substr($source,0,$pos[0]),$line)); | |
131 | } | |
132 | $arg =~ s {^\s*[(]\s*%} { ( \\\%} || | |
133 | $arg =~ s {^\s*[(]\s*m\b} { ( qr} || | |
134 | $arg =~ s {^\s*[(]\s*/} { ( qr/} || | |
135 | $arg =~ s {^\s*[(]\s*qw} { ( \\qw}; | |
136 | @pos = Text::Balanced::_match_codeblock(\$source,qr/\s*/,qr/\{/,qr/\}/,qr/\{/,qr/\}/,undef) | |
137 | or do { | |
138 | die "Bad $keyword statement (problem in the code block?) near $Switch::file line ", line(substr($source,0, pos $source), $line), "\n"; | |
139 | }; | |
140 | my $code = filter_blocks(substr($source,$pos[0],$pos[4]-$pos[0]),line(substr($source,0,$pos[0]),$line)); | |
141 | $code =~ s/{/{ local \$::_S_W_I_T_C_H; Switch::switch $arg;/; | |
142 | $text .= $code . 'continue {last}'; | |
143 | next component; | |
144 | } | |
145 | elsif ($Perl5 && $source =~ m/\G(\s*)(case\b)(?!\s*=>)/gc | |
146 | || $Perl6 && $source =~ m/\G(\s*)(when\b)(?!\s*=>)/gc) | |
147 | { | |
148 | my $keyword = $2; | |
149 | $text .= $1."if (Switch::case"; | |
150 | if (@pos = Text::Balanced::_match_codeblock(\$source,qr/\s*/,qr/\{/,qr/\}/,qr/\{/,qr/\}/,undef)) { | |
151 | my $code = substr($source,$pos[0],$pos[4]-$pos[0]); | |
152 | $text .= " sub" if is_block $code; | |
153 | $text .= " " . filter_blocks($code,line(substr($source,0,$pos[0]),$line)) . ")"; | |
154 | } | |
155 | elsif (@pos = Text::Balanced::_match_codeblock(\$source,qr/\s*/,qr/[[(]/,qr/[])]/,qr/[[({]/,qr/[])}]/,undef)) { | |
156 | my $code = filter_blocks(substr($source,$pos[0],$pos[4]-$pos[0]),line(substr($source,0,$pos[0]),$line)); | |
157 | $code =~ s {^\s*[(]\s*%} { ( \\\%} || | |
158 | $code =~ s {^\s*[(]\s*m\b} { ( qr} || | |
159 | $code =~ s {^\s*[(]\s*/} { ( qr/} || | |
160 | $code =~ s {^\s*[(]\s*qw} { ( \\qw}; | |
161 | $text .= " $code)"; | |
162 | } | |
163 | elsif ($Perl6 && do{@pos = Text::Balanced::_match_variable(\$source,qr/\s*/)}) { | |
164 | my $code = filter_blocks(substr($source,$pos[0],$pos[4]-$pos[0]),line(substr($source,0,$pos[0]),$line)); | |
165 | $code =~ s {^\s*%} { \%} || | |
166 | $code =~ s {^\s*@} { \@}; | |
167 | $text .= " $code)"; | |
168 | } | |
169 | elsif ( @pos = Text::Balanced::_match_quotelike(\$source,qr/\s*/,1,0)) { | |
170 | my $code = substr($source,$pos[2],$pos[18]-$pos[2]); | |
171 | $code = filter_blocks($code,line(substr($source,0,$pos[2]),$line)); | |
172 | $code =~ s {^\s*m} { qr} || | |
173 | $code =~ s {^\s*/} { qr/} || | |
174 | $code =~ s {^\s*qw} { \\qw}; | |
175 | $text .= " $code)"; | |
176 | } | |
177 | elsif ($Perl5 && $source =~ m/\G\s*(([^\$\@{])[^\$\@{]*)(?=\s*{)/gc | |
178 | || $Perl6 && $source =~ m/\G\s*([^;{]*)()/gc) { | |
179 | my $code = filter_blocks($1,line(substr($source,0,pos $source),$line)); | |
180 | $text .= ' \\' if $2 eq '%'; | |
181 | $text .= " $code)"; | |
182 | } | |
183 | else { | |
184 | die "Bad $keyword statement (invalid $keyword value?) near $Switch::file line ", line(substr($source,0,pos $source), $line), "\n"; | |
185 | } | |
186 | ||
187 | die "Missing opening brace or semi-colon after 'when' value near $Switch::file line ", line(substr($source,0,pos $source), $line), "\n" | |
188 | unless !$Perl6 || $source =~ m/\G(\s*)(?=;|\{)/gc; | |
189 | ||
190 | do{@pos = Text::Balanced::_match_codeblock(\$source,qr/\s*/,qr/\{/,qr/\}/,qr/\{/,qr/\}/,undef)} | |
191 | or do { | |
192 | if ($source =~ m/\G\s*(?=([};]|\Z))/gc) { | |
193 | $casecounter++; | |
194 | next component; | |
195 | } | |
196 | die "Bad $keyword statement (problem in the code block?) near $Switch::file line ", line(substr($source,0,pos $source),$line), "\n"; | |
197 | }; | |
198 | my $code = filter_blocks(substr($source,$pos[0],$pos[4]-$pos[0]),line(substr($source,0,$pos[0]),$line)); | |
199 | $code =~ s/}(?=\s*\Z)/;last S_W_I_T_C_H }/ | |
200 | unless $fallthrough; | |
201 | $text .= "{ while (1) $code continue { goto C_A_S_E_$casecounter } last S_W_I_T_C_H; C_A_S_E_$casecounter: }"; | |
202 | $casecounter++; | |
203 | next component; | |
204 | } | |
205 | ||
206 | $source =~ m/\G(\s*(-[sm]\s+|\w+|#.*\n|\W))/gc; | |
207 | $text .= $1; | |
208 | } | |
209 | $text; | |
210 | } | |
211 | ||
212 | ||
213 | ||
214 | sub in | |
215 | { | |
216 | my ($x,$y) = @_; | |
217 | my @numy; | |
218 | for my $nextx ( @$x ) | |
219 | { | |
220 | my $numx = ref($nextx) || defined $nextx && (~$nextx&$nextx) eq 0; | |
221 | for my $j ( 0..$#$y ) | |
222 | { | |
223 | my $nexty = $y->[$j]; | |
224 | push @numy, ref($nexty) || defined $nexty && (~$nexty&$nexty) eq 0 | |
225 | if @numy <= $j; | |
226 | return 1 if $numx && $numy[$j] && $nextx==$nexty | |
227 | || $nextx eq $nexty; | |
228 | ||
229 | } | |
230 | } | |
231 | return ""; | |
232 | } | |
233 | ||
234 | sub on_exists | |
235 | { | |
236 | my $ref = @_==1 && ref($_[0]) eq 'HASH' ? $_[0] : { @_ }; | |
237 | [ keys %$ref ] | |
238 | } | |
239 | ||
240 | sub on_defined | |
241 | { | |
242 | my $ref = @_==1 && ref($_[0]) eq 'HASH' ? $_[0] : { @_ }; | |
243 | [ grep { defined $ref->{$_} } keys %$ref ] | |
244 | } | |
245 | ||
246 | sub switch(;$) | |
247 | { | |
248 | my ($s_val) = @_ ? $_[0] : $_; | |
249 | my $s_ref = ref $s_val; | |
250 | ||
251 | if ($s_ref eq 'CODE') | |
252 | { | |
253 | $::_S_W_I_T_C_H = | |
254 | sub { my $c_val = $_[0]; | |
255 | return $s_val == $c_val if ref $c_val eq 'CODE'; | |
256 | return $s_val->(@$c_val) if ref $c_val eq 'ARRAY'; | |
257 | return $s_val->($c_val); | |
258 | }; | |
259 | } | |
260 | elsif ($s_ref eq "" && defined $s_val && (~$s_val&$s_val) eq 0) # NUMERIC SCALAR | |
261 | { | |
262 | $::_S_W_I_T_C_H = | |
263 | sub { my $c_val = $_[0]; | |
264 | my $c_ref = ref $c_val; | |
265 | return $s_val == $c_val if $c_ref eq "" | |
266 | && defined $c_val | |
267 | && (~$c_val&$c_val) eq 0; | |
268 | return $s_val eq $c_val if $c_ref eq ""; | |
269 | return in([$s_val],$c_val) if $c_ref eq 'ARRAY'; | |
270 | return $c_val->($s_val) if $c_ref eq 'CODE'; | |
271 | return $c_val->call($s_val) if $c_ref eq 'Switch'; | |
272 | return scalar $s_val=~/$c_val/ | |
273 | if $c_ref eq 'Regexp'; | |
274 | return scalar $c_val->{$s_val} | |
275 | if $c_ref eq 'HASH'; | |
276 | return; | |
277 | }; | |
278 | } | |
279 | elsif ($s_ref eq "") # STRING SCALAR | |
280 | { | |
281 | $::_S_W_I_T_C_H = | |
282 | sub { my $c_val = $_[0]; | |
283 | my $c_ref = ref $c_val; | |
284 | return $s_val eq $c_val if $c_ref eq ""; | |
285 | return in([$s_val],$c_val) if $c_ref eq 'ARRAY'; | |
286 | return $c_val->($s_val) if $c_ref eq 'CODE'; | |
287 | return $c_val->call($s_val) if $c_ref eq 'Switch'; | |
288 | return scalar $s_val=~/$c_val/ | |
289 | if $c_ref eq 'Regexp'; | |
290 | return scalar $c_val->{$s_val} | |
291 | if $c_ref eq 'HASH'; | |
292 | return; | |
293 | }; | |
294 | } | |
295 | elsif ($s_ref eq 'ARRAY') | |
296 | { | |
297 | $::_S_W_I_T_C_H = | |
298 | sub { my $c_val = $_[0]; | |
299 | my $c_ref = ref $c_val; | |
300 | return in($s_val,[$c_val]) if $c_ref eq ""; | |
301 | return in($s_val,$c_val) if $c_ref eq 'ARRAY'; | |
302 | return $c_val->(@$s_val) if $c_ref eq 'CODE'; | |
303 | return $c_val->call(@$s_val) | |
304 | if $c_ref eq 'Switch'; | |
305 | return scalar grep {$_=~/$c_val/} @$s_val | |
306 | if $c_ref eq 'Regexp'; | |
307 | return scalar grep {$c_val->{$_}} @$s_val | |
308 | if $c_ref eq 'HASH'; | |
309 | return; | |
310 | }; | |
311 | } | |
312 | elsif ($s_ref eq 'Regexp') | |
313 | { | |
314 | $::_S_W_I_T_C_H = | |
315 | sub { my $c_val = $_[0]; | |
316 | my $c_ref = ref $c_val; | |
317 | return $c_val=~/s_val/ if $c_ref eq ""; | |
318 | return scalar grep {$_=~/s_val/} @$c_val | |
319 | if $c_ref eq 'ARRAY'; | |
320 | return $c_val->($s_val) if $c_ref eq 'CODE'; | |
321 | return $c_val->call($s_val) if $c_ref eq 'Switch'; | |
322 | return $s_val eq $c_val if $c_ref eq 'Regexp'; | |
323 | return grep {$_=~/$s_val/ && $c_val->{$_}} keys %$c_val | |
324 | if $c_ref eq 'HASH'; | |
325 | return; | |
326 | }; | |
327 | } | |
328 | elsif ($s_ref eq 'HASH') | |
329 | { | |
330 | $::_S_W_I_T_C_H = | |
331 | sub { my $c_val = $_[0]; | |
332 | my $c_ref = ref $c_val; | |
333 | return $s_val->{$c_val} if $c_ref eq ""; | |
334 | return scalar grep {$s_val->{$_}} @$c_val | |
335 | if $c_ref eq 'ARRAY'; | |
336 | return $c_val->($s_val) if $c_ref eq 'CODE'; | |
337 | return $c_val->call($s_val) if $c_ref eq 'Switch'; | |
338 | return grep {$_=~/$c_val/ && $s_val->{"$_"}} keys %$s_val | |
339 | if $c_ref eq 'Regexp'; | |
340 | return $s_val==$c_val if $c_ref eq 'HASH'; | |
341 | return; | |
342 | }; | |
343 | } | |
344 | elsif ($s_ref eq 'Switch') | |
345 | { | |
346 | $::_S_W_I_T_C_H = | |
347 | sub { my $c_val = $_[0]; | |
348 | return $s_val == $c_val if ref $c_val eq 'Switch'; | |
349 | return $s_val->call(@$c_val) | |
350 | if ref $c_val eq 'ARRAY'; | |
351 | return $s_val->call($c_val); | |
352 | }; | |
353 | } | |
354 | else | |
355 | { | |
356 | croak "Cannot switch on $s_ref"; | |
357 | } | |
358 | return 1; | |
359 | } | |
360 | ||
361 | sub case($) { local $SIG{__WARN__} = \&carp; | |
362 | $::_S_W_I_T_C_H->(@_); } | |
363 | ||
364 | # IMPLEMENT __ | |
365 | ||
366 | my $placeholder = bless { arity=>1, impl=>sub{$_[1+$_[0]]} }; | |
367 | ||
368 | sub __() { $placeholder } | |
369 | ||
370 | sub __arg($) | |
371 | { | |
372 | my $index = $_[0]+1; | |
373 | bless { arity=>0, impl=>sub{$_[$index]} }; | |
374 | } | |
375 | ||
376 | sub hosub(&@) | |
377 | { | |
378 | # WRITE THIS | |
379 | } | |
380 | ||
381 | sub call | |
382 | { | |
383 | my ($self,@args) = @_; | |
384 | return $self->{impl}->(0,@args); | |
385 | } | |
386 | ||
387 | sub meta_bop(&) | |
388 | { | |
389 | my ($op) = @_; | |
390 | sub | |
391 | { | |
392 | my ($left, $right, $reversed) = @_; | |
393 | ($right,$left) = @_ if $reversed; | |
394 | ||
395 | my $rop = ref $right eq 'Switch' | |
396 | ? $right | |
397 | : bless { arity=>0, impl=>sub{$right} }; | |
398 | ||
399 | my $lop = ref $left eq 'Switch' | |
400 | ? $left | |
401 | : bless { arity=>0, impl=>sub{$left} }; | |
402 | ||
403 | my $arity = $lop->{arity} + $rop->{arity}; | |
404 | ||
405 | return bless { | |
406 | arity => $arity, | |
407 | impl => sub { my $start = shift; | |
408 | return $op->($lop->{impl}->($start,@_), | |
409 | $rop->{impl}->($start+$lop->{arity},@_)); | |
410 | } | |
411 | }; | |
412 | }; | |
413 | } | |
414 | ||
415 | sub meta_uop(&) | |
416 | { | |
417 | my ($op) = @_; | |
418 | sub | |
419 | { | |
420 | my ($left) = @_; | |
421 | ||
422 | my $lop = ref $left eq 'Switch' | |
423 | ? $left | |
424 | : bless { arity=>0, impl=>sub{$left} }; | |
425 | ||
426 | my $arity = $lop->{arity}; | |
427 | ||
428 | return bless { | |
429 | arity => $arity, | |
430 | impl => sub { $op->($lop->{impl}->(@_)) } | |
431 | }; | |
432 | }; | |
433 | } | |
434 | ||
435 | ||
436 | use overload | |
437 | "+" => meta_bop {$_[0] + $_[1]}, | |
438 | "-" => meta_bop {$_[0] - $_[1]}, | |
439 | "*" => meta_bop {$_[0] * $_[1]}, | |
440 | "/" => meta_bop {$_[0] / $_[1]}, | |
441 | "%" => meta_bop {$_[0] % $_[1]}, | |
442 | "**" => meta_bop {$_[0] ** $_[1]}, | |
443 | "<<" => meta_bop {$_[0] << $_[1]}, | |
444 | ">>" => meta_bop {$_[0] >> $_[1]}, | |
445 | "x" => meta_bop {$_[0] x $_[1]}, | |
446 | "." => meta_bop {$_[0] . $_[1]}, | |
447 | "<" => meta_bop {$_[0] < $_[1]}, | |
448 | "<=" => meta_bop {$_[0] <= $_[1]}, | |
449 | ">" => meta_bop {$_[0] > $_[1]}, | |
450 | ">=" => meta_bop {$_[0] >= $_[1]}, | |
451 | "==" => meta_bop {$_[0] == $_[1]}, | |
452 | "!=" => meta_bop {$_[0] != $_[1]}, | |
453 | "<=>" => meta_bop {$_[0] <=> $_[1]}, | |
454 | "lt" => meta_bop {$_[0] lt $_[1]}, | |
455 | "le" => meta_bop {$_[0] le $_[1]}, | |
456 | "gt" => meta_bop {$_[0] gt $_[1]}, | |
457 | "ge" => meta_bop {$_[0] ge $_[1]}, | |
458 | "eq" => meta_bop {$_[0] eq $_[1]}, | |
459 | "ne" => meta_bop {$_[0] ne $_[1]}, | |
460 | "cmp" => meta_bop {$_[0] cmp $_[1]}, | |
461 | "\&" => meta_bop {$_[0] & $_[1]}, | |
462 | "^" => meta_bop {$_[0] ^ $_[1]}, | |
463 | "|" => meta_bop {$_[0] | $_[1]}, | |
464 | "atan2" => meta_bop {atan2 $_[0], $_[1]}, | |
465 | ||
466 | "neg" => meta_uop {-$_[0]}, | |
467 | "!" => meta_uop {!$_[0]}, | |
468 | "~" => meta_uop {~$_[0]}, | |
469 | "cos" => meta_uop {cos $_[0]}, | |
470 | "sin" => meta_uop {sin $_[0]}, | |
471 | "exp" => meta_uop {exp $_[0]}, | |
472 | "abs" => meta_uop {abs $_[0]}, | |
473 | "log" => meta_uop {log $_[0]}, | |
474 | "sqrt" => meta_uop {sqrt $_[0]}, | |
475 | "bool" => sub { croak "Can't use && or || in expression containing __" }, | |
476 | ||
477 | # "&()" => sub { $_[0]->{impl} }, | |
478 | ||
479 | # "||" => meta_bop {$_[0] || $_[1]}, | |
480 | # "&&" => meta_bop {$_[0] && $_[1]}, | |
481 | # fallback => 1, | |
482 | ; | |
483 | 1; | |
484 | ||
485 | __END__ | |
486 | ||
487 | ||
488 | =head1 NAME | |
489 | ||
490 | Switch - A switch statement for Perl | |
491 | ||
492 | =head1 VERSION | |
493 | ||
494 | This document describes version 2.09 of Switch, | |
495 | released June 12, 2002. | |
496 | ||
497 | =head1 SYNOPSIS | |
498 | ||
499 | use Switch; | |
500 | ||
501 | switch ($val) { | |
502 | ||
503 | case 1 { print "number 1" } | |
504 | case "a" { print "string a" } | |
505 | case [1..10,42] { print "number in list" } | |
506 | case (@array) { print "number in list" } | |
507 | case /\w+/ { print "pattern" } | |
508 | case qr/\w+/ { print "pattern" } | |
509 | case (%hash) { print "entry in hash" } | |
510 | case (\%hash) { print "entry in hash" } | |
511 | case (\&sub) { print "arg to subroutine" } | |
512 | else { print "previous case not true" } | |
513 | } | |
514 | ||
515 | =head1 BACKGROUND | |
516 | ||
517 | [Skip ahead to L<"DESCRIPTION"> if you don't care about the whys | |
518 | and wherefores of this control structure] | |
519 | ||
520 | In seeking to devise a "Swiss Army" case mechanism suitable for Perl, | |
521 | it is useful to generalize this notion of distributed conditional | |
522 | testing as far as possible. Specifically, the concept of "matching" | |
523 | between the switch value and the various case values need not be | |
524 | restricted to numeric (or string or referential) equality, as it is in other | |
525 | languages. Indeed, as Table 1 illustrates, Perl | |
526 | offers at least eighteen different ways in which two values could | |
527 | generate a match. | |
528 | ||
529 | Table 1: Matching a switch value ($s) with a case value ($c) | |
530 | ||
531 | Switch Case Type of Match Implied Matching Code | |
532 | Value Value | |
533 | ====== ===== ===================== ============= | |
534 | ||
535 | number same numeric or referential match if $s == $c; | |
536 | or ref equality | |
537 | ||
538 | object method result of method call match if $s->$c(); | |
539 | ref name match if defined $s->$c(); | |
540 | or ref | |
541 | ||
542 | other other string equality match if $s eq $c; | |
543 | non-ref non-ref | |
544 | scalar scalar | |
545 | ||
546 | string regexp pattern match match if $s =~ /$c/; | |
547 | ||
548 | array scalar array entry existence match if 0<=$c && $c<@$s; | |
549 | ref array entry definition match if defined $s->[$c]; | |
550 | array entry truth match if $s->[$c]; | |
551 | ||
552 | array array array intersection match if intersects(@$s, @$c); | |
553 | ref ref (apply this table to | |
554 | all pairs of elements | |
555 | $s->[$i] and | |
556 | $c->[$j]) | |
557 | ||
558 | array regexp array grep match if grep /$c/, @$s; | |
559 | ref | |
560 | ||
561 | hash scalar hash entry existence match if exists $s->{$c}; | |
562 | ref hash entry definition match if defined $s->{$c}; | |
563 | hash entry truth match if $s->{$c}; | |
564 | ||
565 | hash regexp hash grep match if grep /$c/, keys %$s; | |
566 | ref | |
567 | ||
568 | sub scalar return value defn match if defined $s->($c); | |
569 | ref return value truth match if $s->($c); | |
570 | ||
571 | sub array return value defn match if defined $s->(@$c); | |
572 | ref ref return value truth match if $s->(@$c); | |
573 | ||
574 | ||
575 | In reality, Table 1 covers 31 alternatives, because only the equality and | |
576 | intersection tests are commutative; in all other cases, the roles of | |
577 | the C<$s> and C<$c> variables could be reversed to produce a | |
578 | different test. For example, instead of testing a single hash for | |
579 | the existence of a series of keys (C<match if exists $s-E<gt>{$c}>), | |
580 | one could test for the existence of a single key in a series of hashes | |
581 | (C<match if exists $c-E<gt>{$s}>). | |
582 | ||
583 | As L<perltodo> observes, a Perl case mechanism must support all these | |
584 | "ways to do it". | |
585 | ||
586 | ||
587 | =head1 DESCRIPTION | |
588 | ||
589 | The Switch.pm module implements a generalized case mechanism that covers | |
590 | the numerous possible combinations of switch and case values described above. | |
591 | ||
592 | The module augments the standard Perl syntax with two new control | |
593 | statements: C<switch> and C<case>. The C<switch> statement takes a | |
594 | single scalar argument of any type, specified in parentheses. | |
595 | C<switch> stores this value as the | |
596 | current switch value in a (localized) control variable. | |
597 | The value is followed by a block which may contain one or more | |
598 | Perl statements (including the C<case> statement described below). | |
599 | The block is unconditionally executed once the switch value has | |
600 | been cached. | |
601 | ||
602 | A C<case> statement takes a single scalar argument (in mandatory | |
603 | parentheses if it's a variable; otherwise the parens are optional) and | |
604 | selects the appropriate type of matching between that argument and the | |
605 | current switch value. The type of matching used is determined by the | |
606 | respective types of the switch value and the C<case> argument, as | |
607 | specified in Table 1. If the match is successful, the mandatory | |
608 | block associated with the C<case> statement is executed. | |
609 | ||
610 | In most other respects, the C<case> statement is semantically identical | |
611 | to an C<if> statement. For example, it can be followed by an C<else> | |
612 | clause, and can be used as a postfix statement qualifier. | |
613 | ||
614 | However, when a C<case> block has been executed control is automatically | |
615 | transferred to the statement after the immediately enclosing C<switch> | |
616 | block, rather than to the next statement within the block. In other | |
617 | words, the success of any C<case> statement prevents other cases in the | |
618 | same scope from executing. But see L<"Allowing fall-through"> below. | |
619 | ||
620 | Together these two new statements provide a fully generalized case | |
621 | mechanism: | |
622 | ||
623 | use Switch; | |
624 | ||
625 | # AND LATER... | |
626 | ||
627 | %special = ( woohoo => 1, d'oh => 1 ); | |
628 | ||
629 | while (<>) { | |
630 | switch ($_) { | |
631 | ||
632 | case (%special) { print "homer\n"; } # if $special{$_} | |
633 | case /a-z/i { print "alpha\n"; } # if $_ =~ /a-z/i | |
634 | case [1..9] { print "small num\n"; } # if $_ in [1..9] | |
635 | ||
636 | case { $_[0] >= 10 } { # if $_ >= 10 | |
637 | my $age = <>; | |
638 | switch (sub{ $_[0] < $age } ) { | |
639 | ||
640 | case 20 { print "teens\n"; } # if 20 < $age | |
641 | case 30 { print "twenties\n"; } # if 30 < $age | |
642 | else { print "history\n"; } | |
643 | } | |
644 | } | |
645 | ||
646 | print "must be punctuation\n" case /\W/; # if $_ ~= /\W/ | |
647 | } | |
648 | ||
649 | Note that C<switch>es can be nested within C<case> (or any other) blocks, | |
650 | and a series of C<case> statements can try different types of matches | |
651 | -- hash membership, pattern match, array intersection, simple equality, | |
652 | etc. -- against the same switch value. | |
653 | ||
654 | The use of intersection tests against an array reference is particularly | |
655 | useful for aggregating integral cases: | |
656 | ||
657 | sub classify_digit | |
658 | { | |
659 | switch ($_[0]) { case 0 { return 'zero' } | |
660 | case [2,4,6,8] { return 'even' } | |
661 | case [1,3,4,7,9] { return 'odd' } | |
662 | case /[A-F]/i { return 'hex' } | |
663 | } | |
664 | } | |
665 | ||
666 | ||
667 | =head2 Allowing fall-through | |
668 | ||
669 | Fall-though (trying another case after one has already succeeded) | |
670 | is usually a Bad Idea in a switch statement. However, this | |
671 | is Perl, not a police state, so there I<is> a way to do it, if you must. | |
672 | ||
673 | If a C<case> block executes an untargetted C<next>, control is | |
674 | immediately transferred to the statement I<after> the C<case> statement | |
675 | (i.e. usually another case), rather than out of the surrounding | |
676 | C<switch> block. | |
677 | ||
678 | For example: | |
679 | ||
680 | switch ($val) { | |
681 | case 1 { handle_num_1(); next } # and try next case... | |
682 | case "1" { handle_str_1(); next } # and try next case... | |
683 | case [0..9] { handle_num_any(); } # and we're done | |
684 | case /\d/ { handle_dig_any(); next } # and try next case... | |
685 | case /.*/ { handle_str_any(); next } # and try next case... | |
686 | } | |
687 | ||
688 | If $val held the number C<1>, the above C<switch> block would call the | |
689 | first three C<handle_...> subroutines, jumping to the next case test | |
690 | each time it encountered a C<next>. After the thrid C<case> block | |
691 | was executed, control would jump to the end of the enclosing | |
692 | C<switch> block. | |
693 | ||
694 | On the other hand, if $val held C<10>, then only the last two C<handle_...> | |
695 | subroutines would be called. | |
696 | ||
697 | Note that this mechanism allows the notion of I<conditional fall-through>. | |
698 | For example: | |
699 | ||
700 | switch ($val) { | |
701 | case [0..9] { handle_num_any(); next if $val < 7; } | |
702 | case /\d/ { handle_dig_any(); } | |
703 | } | |
704 | ||
705 | If an untargetted C<last> statement is executed in a case block, this | |
706 | immediately transfers control out of the enclosing C<switch> block | |
707 | (in other words, there is an implicit C<last> at the end of each | |
708 | normal C<case> block). Thus the previous example could also have been | |
709 | written: | |
710 | ||
711 | switch ($val) { | |
712 | case [0..9] { handle_num_any(); last if $val >= 7; next; } | |
713 | case /\d/ { handle_dig_any(); } | |
714 | } | |
715 | ||
716 | ||
717 | =head2 Automating fall-through | |
718 | ||
719 | In situations where case fall-through should be the norm, rather than an | |
720 | exception, an endless succession of terminal C<next>s is tedious and ugly. | |
721 | Hence, it is possible to reverse the default behaviour by specifying | |
722 | the string "fallthrough" when importing the module. For example, the | |
723 | following code is equivalent to the first example in L<"Allowing fall-through">: | |
724 | ||
725 | use Switch 'fallthrough'; | |
726 | ||
727 | switch ($val) { | |
728 | case 1 { handle_num_1(); } | |
729 | case "1" { handle_str_1(); } | |
730 | case [0..9] { handle_num_any(); last } | |
731 | case /\d/ { handle_dig_any(); } | |
732 | case /.*/ { handle_str_any(); } | |
733 | } | |
734 | ||
735 | Note the explicit use of a C<last> to preserve the non-fall-through | |
736 | behaviour of the third case. | |
737 | ||
738 | ||
739 | ||
740 | =head2 Alternative syntax | |
741 | ||
742 | Perl 6 will provide a built-in switch statement with essentially the | |
743 | same semantics as those offered by Switch.pm, but with a different | |
744 | pair of keywords. In Perl 6 C<switch> will be spelled C<given>, and | |
745 | C<case> will be pronounced C<when>. In addition, the C<when> statement | |
746 | will not require switch or case values to be parenthesized. | |
747 | ||
748 | This future syntax is also (largely) available via the Switch.pm module, by | |
749 | importing it with the argument C<"Perl6">. For example: | |
750 | ||
751 | use Switch 'Perl6'; | |
752 | ||
753 | given ($val) { | |
754 | when 1 { handle_num_1(); } | |
755 | when ($str1) { handle_str_1(); } | |
756 | when [0..9] { handle_num_any(); last } | |
757 | when /\d/ { handle_dig_any(); } | |
758 | when /.*/ { handle_str_any(); } | |
759 | } | |
760 | ||
761 | Note that scalars still need to be parenthesized, since they would be | |
762 | ambiguous in Perl 5. | |
763 | ||
764 | Note too that you can mix and match both syntaxes by importing the module | |
765 | with: | |
766 | ||
767 | use Switch 'Perl5', 'Perl6'; | |
768 | ||
769 | ||
770 | =head2 Higher-order Operations | |
771 | ||
772 | One situation in which C<switch> and C<case> do not provide a good | |
773 | substitute for a cascaded C<if>, is where a switch value needs to | |
774 | be tested against a series of conditions. For example: | |
775 | ||
776 | sub beverage { | |
777 | switch (shift) { | |
778 | ||
779 | case sub { $_[0] < 10 } { return 'milk' } | |
780 | case sub { $_[0] < 20 } { return 'coke' } | |
781 | case sub { $_[0] < 30 } { return 'beer' } | |
782 | case sub { $_[0] < 40 } { return 'wine' } | |
783 | case sub { $_[0] < 50 } { return 'malt' } | |
784 | case sub { $_[0] < 60 } { return 'Moet' } | |
785 | else { return 'milk' } | |
786 | } | |
787 | } | |
788 | ||
789 | The need to specify each condition as a subroutine block is tiresome. To | |
790 | overcome this, when importing Switch.pm, a special "placeholder" | |
791 | subroutine named C<__> [sic] may also be imported. This subroutine | |
792 | converts (almost) any expression in which it appears to a reference to a | |
793 | higher-order function. That is, the expression: | |
794 | ||
795 | use Switch '__'; | |
796 | ||
797 | __ < 2 + __ | |
798 | ||
799 | is equivalent to: | |
800 | ||
801 | sub { $_[0] < 2 + $_[1] } | |
802 | ||
803 | With C<__>, the previous ugly case statements can be rewritten: | |
804 | ||
805 | case __ < 10 { return 'milk' } | |
806 | case __ < 20 { return 'coke' } | |
807 | case __ < 30 { return 'beer' } | |
808 | case __ < 40 { return 'wine' } | |
809 | case __ < 50 { return 'malt' } | |
810 | case __ < 60 { return 'Moet' } | |
811 | else { return 'milk' } | |
812 | ||
813 | The C<__> subroutine makes extensive use of operator overloading to | |
814 | perform its magic. All operations involving __ are overloaded to | |
815 | produce an anonymous subroutine that implements a lazy version | |
816 | of the original operation. | |
817 | ||
818 | The only problem is that operator overloading does not allow the | |
819 | boolean operators C<&&> and C<||> to be overloaded. So a case statement | |
820 | like this: | |
821 | ||
822 | case 0 <= __ && __ < 10 { return 'digit' } | |
823 | ||
824 | doesn't act as expected, because when it is | |
825 | executed, it constructs two higher order subroutines | |
826 | and then treats the two resulting references as arguments to C<&&>: | |
827 | ||
828 | sub { 0 <= $_[0] } && sub { $_[0] < 10 } | |
829 | ||
830 | This boolean expression is inevitably true, since both references are | |
831 | non-false. Fortunately, the overloaded C<'bool'> operator catches this | |
832 | situation and flags it as a error. | |
833 | ||
834 | =head1 DEPENDENCIES | |
835 | ||
836 | The module is implemented using Filter::Util::Call and Text::Balanced | |
837 | and requires both these modules to be installed. | |
838 | ||
839 | =head1 AUTHOR | |
840 | ||
841 | Damian Conway (damian@conway.org) | |
842 | ||
843 | =head1 BUGS | |
844 | ||
845 | There are undoubtedly serious bugs lurking somewhere in code this funky :-) | |
846 | Bug reports and other feedback are most welcome. | |
847 | ||
848 | =head1 LIMITATION | |
849 | ||
850 | Due to the heuristic nature of Switch.pm's source parsing, the presence | |
851 | of regexes specified with raw C<?...?> delimiters may cause mysterious | |
852 | errors. The workaround is to use C<m?...?> instead. | |
853 | ||
854 | =head1 COPYRIGHT | |
855 | ||
856 | Copyright (c) 1997-2001, Damian Conway. All Rights Reserved. | |
857 | This module is free software. It may be used, redistributed | |
858 | and/or modified under the same terms as Perl itself. |