This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexec.c: Clean up initialized vars in switch
[perl5.git] / ext / re / re.pm
CommitLineData
b3eb6a9b
GS
1package re;
2
99cc5cc6 3# pragma for controlling the regexp engine
de8c5301
YO
4use strict;
5use warnings;
6
94a75ecf 7our $VERSION = "0.15";
de8c5301 8our @ISA = qw(Exporter);
ec781434 9our @EXPORT_OK = ('regmust',
192c1e27
JH
10 qw(is_regexp regexp_pattern
11 regname regnames regnames_count));
de8c5301
YO
12our %EXPORT_OK = map { $_ => 1 } @EXPORT_OK;
13
de8c5301
YO
14my %bitmask = (
15 taint => 0x00100000, # HINT_RE_TAINT
16 eval => 0x00200000, # HINT_RE_EVAL
17);
18
1e215989
FC
19my $flags_hint = 0x02000000; # HINT_RE_FLAGS
20my $PMMOD_SHIFT = 0;
21my %reflags = (
22 m => 1 << ($PMMOD_SHIFT + 0),
23 s => 1 << ($PMMOD_SHIFT + 1),
24 i => 1 << ($PMMOD_SHIFT + 2),
25 x => 1 << ($PMMOD_SHIFT + 3),
26 p => 1 << ($PMMOD_SHIFT + 4),
27# special cases:
28 l => 1 << ($PMMOD_SHIFT + 5),
29 u => 1 << ($PMMOD_SHIFT + 6),
30 d => 0,
31);
32
de8c5301
YO
33sub setcolor {
34 eval { # Ignore errors
35 require Term::Cap;
36
37 my $terminal = Tgetent Term::Cap ({OSPEED => 9600}); # Avoid warning.
38 my $props = $ENV{PERL_RE_TC} || 'md,me,so,se,us,ue';
39 my @props = split /,/, $props;
40 my $colors = join "\t", map {$terminal->Tputs($_,1)} @props;
41
42 $colors =~ s/\0//g;
43 $ENV{PERL_RE_COLORS} = $colors;
44 };
45 if ($@) {
46 $ENV{PERL_RE_COLORS} ||= qq'\t\t> <\t> <\t\t';
47 }
48
49}
50
51my %flags = (
52 COMPILE => 0x0000FF,
53 PARSE => 0x000001,
54 OPTIMISE => 0x000002,
55 TRIEC => 0x000004,
56 DUMP => 0x000008,
f7819f85 57 FLAGS => 0x000010,
de8c5301
YO
58
59 EXECUTE => 0x00FF00,
60 INTUIT => 0x000100,
61 MATCH => 0x000200,
62 TRIEE => 0x000400,
63
64 EXTRA => 0xFF0000,
65 TRIEM => 0x010000,
66 OFFSETS => 0x020000,
67 OFFSETSDBG => 0x040000,
68 STATE => 0x080000,
69 OPTIMISEM => 0x100000,
70 STACK => 0x280000,
e7707071 71 BUFFERS => 0x400000,
2c296965 72 GPOS => 0x800000,
de8c5301 73);
e7707071 74$flags{ALL} = -1 & ~($flags{OFFSETS}|$flags{OFFSETSDBG}|$flags{BUFFERS});
de8c5301 75$flags{All} = $flags{all} = $flags{DUMP} | $flags{EXECUTE};
2c296965 76$flags{Extra} = $flags{EXECUTE} | $flags{COMPILE} | $flags{GPOS};
de8c5301
YO
77$flags{More} = $flags{MORE} = $flags{All} | $flags{TRIEC} | $flags{TRIEM} | $flags{STATE};
78$flags{State} = $flags{DUMP} | $flags{EXECUTE} | $flags{STATE};
79$flags{TRIE} = $flags{DUMP} | $flags{EXECUTE} | $flags{TRIEC};
80
ec781434
NC
81if (defined &DynaLoader::boot_DynaLoader) {
82 require XSLoader;
da4061d3 83 XSLoader::load();
de8c5301 84}
ec781434
NC
85# else we're miniperl
86# We need to work for miniperl, because the XS toolchain uses Text::Wrap, which
87# uses re 'taint'.
de8c5301
YO
88
89sub _load_unload {
90 my ($on)= @_;
91 if ($on) {
ec781434
NC
92 # We call install() every time, as if we didn't, we wouldn't
93 # "see" any changes to the color environment var since
94 # the last time it was called.
95
96 # install() returns an integer, which if casted properly
99cc5cc6 97 # in C resolves to a structure containing the regexp
ec781434
NC
98 # hooks. Setting it to a random integer will guarantee
99 # segfaults.
100 $^H{regcomp} = install();
de8c5301
YO
101 } else {
102 delete $^H{regcomp};
103 }
104}
105
106sub bits {
107 my $on = shift;
108 my $bits = 0;
109 unless (@_) {
110 require Carp;
111 Carp::carp("Useless use of \"re\" pragma");
112 }
1e215989 113 ARG:
de8c5301
YO
114 foreach my $idx (0..$#_){
115 my $s=$_[$idx];
116 if ($s eq 'Debug' or $s eq 'Debugcolor') {
117 setcolor() if $s =~/color/i;
118 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
119 for my $idx ($idx+1..$#_) {
120 if ($flags{$_[$idx]}) {
121 if ($on) {
122 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
123 } else {
124 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
125 }
126 } else {
127 require Carp;
128 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
129 join(", ",sort keys %flags ) );
130 }
131 }
132 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
133 last;
134 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
135 setcolor() if $s =~/color/i;
136 _load_unload($on);
66e6b4c5 137 last;
de8c5301
YO
138 } elsif (exists $bitmask{$s}) {
139 $bits |= $bitmask{$s};
140 } elsif ($EXPORT_OK{$s}) {
de8c5301
YO
141 require Exporter;
142 re->export_to_level(2, 're', $s);
1e215989
FC
143 } elsif ($s =~ s/^\///) {
144 my $reflags = $^H{reflags} || 0;
145 for(split//, $s) {
146 if (/[dul]/) {
147 if ($on) {
148 $^H{reflags_dul} = $reflags{$_};
149 }
150 else {
151 delete $^H{reflags_dul}
152 if defined $^H{reflags_dul}
153 && $^H{reflags_dul} == $reflags{$_};
154 }
155 } elsif (exists $reflags{$_}) {
156 $on
157 ? $reflags |= $reflags{$_}
158 : ($reflags &= ~$reflags{$_});
159 } else {
160 require Carp;
161 Carp::carp(
162 qq'Unknown regular expression flag "$_"'
163 );
164 next ARG;
165 }
166 }
167 ($^H{reflags} = $reflags or defined $^H{reflags_dul})
168 ? $^H |= $flags_hint
169 : ($^H &= ~$flags_hint);
de8c5301
YO
170 } else {
171 require Carp;
172 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
173 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
174 ")");
175 }
176 }
177 $bits;
178}
179
180sub import {
181 shift;
182 $^H |= bits(1, @_);
183}
184
185sub unimport {
186 shift;
187 $^H &= ~ bits(0, @_);
188}
189
1901;
191
192__END__
56953603 193
b3eb6a9b
GS
194=head1 NAME
195
196re - Perl pragma to alter regular expression behaviour
197
198=head1 SYNOPSIS
199
e4d48cc9
GS
200 use re 'taint';
201 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
b3eb6a9b 202
2cd61cdb 203 $pat = '(?{ $foo = 1 })';
e4d48cc9 204 use re 'eval';
2cd61cdb 205 /foo${pat}bar/; # won't fail (when not under -T switch)
e4d48cc9
GS
206
207 {
208 no re 'taint'; # the default
209 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
210
211 no re 'eval'; # the default
2cd61cdb 212 /foo${pat}bar/; # disallowed (with or without -T switch)
e4d48cc9 213 }
b3eb6a9b 214
1e215989
FC
215 use re '/ix';
216 "FOO" =~ / foo /; # /ix implied
217 no re '/x';
218 "FOO" =~ /foo/; # just /i implied
219
1e2e3d02
YO
220 use re 'debug'; # output debugging info during
221 /^(.*)$/s; # compile and run time
222
2cd61cdb 223
02ea72ae
IZ
224 use re 'debugcolor'; # same as 'debug', but with colored output
225 ...
226
a3621e74 227 use re qw(Debug All); # Finer tuned debugging options.
4ee9a43f 228 use re qw(Debug More);
fe759410 229 no re qw(Debug ALL); # Turn of all re debugging in this scope
4ee9a43f 230
de8c5301
YO
231 use re qw(is_regexp regexp_pattern); # import utility functions
232 my ($pat,$mods)=regexp_pattern(qr/foo/i);
233 if (is_regexp($obj)) {
234 print "Got regexp: ",
235 scalar regexp_pattern($obj); # just as perl would stringify it
236 } # but no hassle with blessed re's.
a3621e74 237
3ffabb8c
GS
238(We use $^X in these examples because it's tainted by default.)
239
b3eb6a9b
GS
240=head1 DESCRIPTION
241
de8c5301
YO
242=head2 'taint' mode
243
b3eb6a9b 244When C<use re 'taint'> is in effect, and a tainted string is the target
99cc5cc6
A
245of a regexp, the regexp memories (or values returned by the m// operator
246in list context) are tainted. This feature is useful when regexp operations
e4d48cc9
GS
247on tainted data aren't meant to extract safe substrings, but to perform
248other transformations.
b3eb6a9b 249
de8c5301
YO
250=head2 'eval' mode
251
99cc5cc6 252When C<use re 'eval'> is in effect, a regexp is allowed to contain
0b370c0a
A
253C<(?{ ... })> zero-width assertions and C<(??{ ... })> postponed
254subexpressions, even if the regular expression contains
ffbc6a93 255variable interpolation. That is normally disallowed, since it is a
2cd61cdb
IZ
256potential security risk. Note that this pragma is ignored when the regular
257expression is obtained from tainted data, i.e. evaluation is always
0b370c0a 258disallowed with tainted regular expressions. See L<perlre/(?{ code })>
bb1773de 259and L<perlre/(??{ code })>.
2cd61cdb 260
ffbc6a93 261For the purpose of this pragma, interpolation of precompiled regular
0a92e3a8
GS
262expressions (i.e., the result of C<qr//>) is I<not> considered variable
263interpolation. Thus:
2cd61cdb
IZ
264
265 /foo${pat}bar/
266
ffbc6a93 267I<is> allowed if $pat is a precompiled regular expression, even
0b370c0a 268if $pat contains C<(?{ ... })> assertions or C<(??{ ... })> subexpressions.
2cd61cdb 269
1e215989
FC
270=head2 '/flags' mode
271
272When C<use re '/flags'> is specified, the given flags are automatically
273added to every regular expression till the end of the lexical scope.
274
275C<no re '/flags'> will turn off the effect of C<use re '/flags'> for the
276given flags.
277
278For example, if you want all your regular expressions to have /msx on by
279default, simply put
280
281 use re '/msx';
282
283at the top of your code.
284
285The /dul flags cancel each other out. So, in this example,
286
287 use re "/u";
288 "ss" =~ /\xdf/;
289 use re "/d";
290 "ss" =~ /\xdf/;
291
292The second C<use re> does an implicit C<no re '/u'>.
293
294Turning on the /l and /u flags with C<use re> takes precedence over the
295C<locale> pragma and the 'unicode_strings' C<feature>, for regular
296expressions. Turning off one of these flags when it is active reverts to
297the behaviour specified by whatever other pragmata are in scope. For
298example:
299
300 use feature "unicode_strings";
301 no re "/u"; # does nothing
302 use re "/l";
303 no re "/l"; # reverts to unicode_strings behaviour
304
de8c5301
YO
305=head2 'debug' mode
306
ffbc6a93 307When C<use re 'debug'> is in effect, perl emits debugging messages when
2cd61cdb
IZ
308compiling and using regular expressions. The output is the same as that
309obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
310B<-Dr> switch. It may be quite voluminous depending on the complexity
02ea72ae
IZ
311of the match. Using C<debugcolor> instead of C<debug> enables a
312form of output that can be used to get a colorful display on terminals
313that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
314comma-separated list of C<termcap> properties to use for highlighting
ffbc6a93 315strings on/off, pre-point part on/off.
2cd61cdb
IZ
316See L<perldebug/"Debugging regular expressions"> for additional info.
317
de8c5301
YO
318As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
319lexically scoped, as the other directives are. However they have both
320compile-time and run-time effects.
321
322See L<perlmodlib/Pragmatic Modules>.
323
324=head2 'Debug' mode
325
a3621e74
YO
326Similarly C<use re 'Debug'> produces debugging output, the difference
327being that it allows the fine tuning of what debugging output will be
be8e71aa
YO
328emitted. Options are divided into three groups, those related to
329compilation, those related to execution and those related to special
330purposes. The options are as follows:
331
332=over 4
333
334=item Compile related options
335
336=over 4
337
338=item COMPILE
339
340Turns on all compile related debug options.
341
342=item PARSE
343
344Turns on debug output related to the process of parsing the pattern.
345
346=item OPTIMISE
347
348Enables output related to the optimisation phase of compilation.
349
24b23f37 350=item TRIEC
be8e71aa
YO
351
352Detailed info about trie compilation.
353
354=item DUMP
355
356Dump the final program out after it is compiled and optimised.
357
be8e71aa
YO
358=back
359
360=item Execute related options
361
362=over 4
363
364=item EXECUTE
365
366Turns on all execute related debug options.
367
368=item MATCH
369
370Turns on debugging of the main matching loop.
371
24b23f37 372=item TRIEE
be8e71aa
YO
373
374Extra debugging of how tries execute.
375
376=item INTUIT
377
378Enable debugging of start point optimisations.
379
380=back
381
382=item Extra debugging options
383
384=over 4
385
386=item EXTRA
387
388Turns on all "extra" debugging options.
389
e7707071
YO
390=item BUFFERS
391
c27a5cfe 392Enable debugging the capture group storage during match. Warning,
e7707071
YO
393this can potentially produce extremely large output.
394
24b23f37
YO
395=item TRIEM
396
397Enable enhanced TRIE debugging. Enhances both TRIEE
398and TRIEC.
399
400=item STATE
401
4ee9a43f 402Enable debugging of states in the engine.
24b23f37
YO
403
404=item STACK
be8e71aa 405
24b23f37
YO
406Enable debugging of the recursion stack in the engine. Enabling
407or disabling this option automatically does the same for debugging
408states as well. This output from this can be quite large.
409
410=item OPTIMISEM
411
412Enable enhanced optimisation debugging and start point optimisations.
99cc5cc6 413Probably not useful except when debugging the regexp engine itself.
24b23f37
YO
414
415=item OFFSETS
416
417Dump offset information. This can be used to see how regops correlate
418to the pattern. Output format is
419
420 NODENUM:POSITION[LENGTH]
421
422Where 1 is the position of the first char in the string. Note that position
423can be 0, or larger than the actual length of the pattern, likewise length
424can be zero.
be8e71aa 425
24b23f37 426=item OFFSETSDBG
be8e71aa
YO
427
428Enable debugging of offsets information. This emits copious
fe759410 429amounts of trace information and doesn't mesh well with other
be8e71aa
YO
430debug options.
431
fe759410 432Almost definitely only useful to people hacking
be8e71aa
YO
433on the offsets part of the debug engine.
434
435=back
436
437=item Other useful flags
438
439These are useful shortcuts to save on the typing.
440
441=over 4
442
443=item ALL
444
e7707071 445Enable all options at once except OFFSETS, OFFSETSDBG and BUFFERS
be8e71aa
YO
446
447=item All
448
fe759410 449Enable DUMP and all execute options. Equivalent to:
be8e71aa
YO
450
451 use re 'debug';
452
453=item MORE
454
455=item More
456
24b23f37 457Enable TRIEM and all execute compile and execute options.
be8e71aa 458
dba3f186 459=back
be8e71aa 460
dba3f186 461=back
a3621e74 462
1e2e3d02 463As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
4ee9a43f 464lexically scoped, as the other directives are. However they have both
1e2e3d02 465compile-time and run-time effects.
b3eb6a9b 466
de8c5301 467=head2 Exportable Functions
b3eb6a9b 468
de8c5301 469As of perl 5.9.5 're' debug contains a number of utility functions that
4ee9a43f 470may be optionally exported into the caller's namespace. They are listed
de8c5301 471below.
b3eb6a9b 472
de8c5301 473=over 4
b3eb6a9b 474
de8c5301 475=item is_regexp($ref)
02ea72ae 476
de8c5301 477Returns true if the argument is a compiled regular expression as returned
4ee9a43f 478by C<qr//>, false if it is not.
02ea72ae 479
4ee9a43f
RGS
480This function will not be confused by overloading or blessing. In
481internals terms, this extracts the regexp pointer out of the
3a5e0888 482PERL_MAGIC_qr structure so it cannot be fooled.
894be9b7 483
de8c5301 484=item regexp_pattern($ref)
02ea72ae 485
4ee9a43f
RGS
486If the argument is a compiled regular expression as returned by C<qr//>,
487then this function returns the pattern.
be8e71aa 488
4ee9a43f
RGS
489In list context it returns a two element list, the first element
490containing the pattern and the second containing the modifiers used when
491the pattern was compiled.
be8e71aa 492
4ee9a43f 493 my ($pat, $mods) = regexp_pattern($ref);
a3621e74 494
99cc5cc6 495In scalar context it returns the same as perl would when stringifying a raw
4ee9a43f
RGS
496C<qr//> with the same pattern inside. If the argument is not a compiled
497reference then this routine returns false but defined in scalar context,
498and the empty list in list context. Thus the following
f9f4320a 499
dff5e0c4 500 if (regexp_pattern($ref) eq '(?^i:foo)')
dba3f186 501
de8c5301 502will be warning free regardless of what $ref actually is.
380e0b81 503
4ee9a43f
RGS
504Like C<is_regexp> this function will not be confused by overloading
505or blessing of the object.
b3eb6a9b 506
256ddcd0
YO
507=item regmust($ref)
508
432acd5f 509If the argument is a compiled regular expression as returned by C<qr//>,
99cc5cc6 510then this function returns what the optimiser considers to be the longest
432acd5f
RGS
511anchored fixed string and longest floating fixed string in the pattern.
512
513A I<fixed string> is defined as being a substring that must appear for the
514pattern to match. An I<anchored fixed string> is a fixed string that must
515appear at a particular offset from the beginning of the match. A I<floating
516fixed string> is defined as a fixed string that can appear at any point in
517a range of positions relative to the start of the match. For example,
518
519 my $qr = qr/here .* there/x;
520 my ($anchored, $floating) = regmust($qr);
256ddcd0 521 print "anchored:'$anchored'\nfloating:'$floating'\n";
432acd5f 522
256ddcd0
YO
523results in
524
525 anchored:'here'
526 floating:'there'
527
432acd5f
RGS
528Because the C<here> is before the C<.*> in the pattern, its position
529can be determined exactly. That's not true, however, for the C<there>;
530it could appear at any point after where the anchored string appeared.
256ddcd0
YO
531Perl uses both for its optimisations, prefering the longer, or, if they are
532equal, the floating.
533
534B<NOTE:> This may not necessarily be the definitive longest anchored and
432acd5f 535floating string. This will be what the optimiser of the Perl that you
256ddcd0
YO
536are using thinks is the longest. If you believe that the result is wrong
537please report it via the L<perlbug> utility.
538
28d8d7f4 539=item regname($name,$all)
44a2ac75 540
28d8d7f4
YO
541Returns the contents of a named buffer of the last successful match. If
542$all is true, then returns an array ref containing one entry per buffer,
44a2ac75
YO
543otherwise returns the first defined buffer.
544
28d8d7f4 545=item regnames($all)
44a2ac75 546
28d8d7f4
YO
547Returns a list of all of the named buffers defined in the last successful
548match. If $all is true, then it returns all names defined, if not it returns
549only names which were involved in the match.
44a2ac75 550
28d8d7f4 551=item regnames_count()
44a2ac75 552
28d8d7f4
YO
553Returns the number of distinct names defined in the pattern used
554for the last successful match.
44a2ac75 555
28d8d7f4
YO
556B<Note:> this result is always the actual number of distinct
557named buffers defined, it may not actually match that which is
558returned by C<regnames()> and related routines when those routines
559have not been called with the $all parameter set.
44a2ac75 560
de8c5301 561=back
b3eb6a9b 562
de8c5301 563=head1 SEE ALSO
b3eb6a9b 564
de8c5301
YO
565L<perlmodlib/Pragmatic Modules>.
566
567=cut