3 # pragma for controlling the regexp engine
8 our @ISA = qw(Exporter);
9 our @EXPORT_OK = ('regmust',
10 qw(is_regexp regexp_pattern
11 regname regnames regnames_count));
12 our %EXPORT_OK = map { $_ => 1 } @EXPORT_OK;
15 taint => 0x00100000, # HINT_RE_TAINT
16 eval => 0x00200000, # HINT_RE_EVAL
19 my $flags_hint = 0x02000000; # HINT_RE_FLAGS
22 m => 1 << ($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),
36 eval { # Ignore errors
39 my $terminal = Tgetent Term::Cap ({OSPEED => 9600}); # Avoid warning.
40 my $props = $ENV{PERL_RE_TC} || 'md,me,so,se,us,ue';
41 my @props = split /,/, $props;
42 my $colors = join "\t", map {$terminal->Tputs($_,1)} @props;
45 $ENV{PERL_RE_COLORS} = $colors;
48 $ENV{PERL_RE_COLORS} ||= qq'\t\t> <\t> <\t\t';
69 OFFSETSDBG => 0x040000,
71 OPTIMISEM => 0x100000,
76 $flags{ALL} = -1 & ~($flags{OFFSETS}|$flags{OFFSETSDBG}|$flags{BUFFERS});
77 $flags{All} = $flags{all} = $flags{DUMP} | $flags{EXECUTE};
78 $flags{Extra} = $flags{EXECUTE} | $flags{COMPILE} | $flags{GPOS};
79 $flags{More} = $flags{MORE} = $flags{All} | $flags{TRIEC} | $flags{TRIEM} | $flags{STATE};
80 $flags{State} = $flags{DUMP} | $flags{EXECUTE} | $flags{STATE};
81 $flags{TRIE} = $flags{DUMP} | $flags{EXECUTE} | $flags{TRIEC};
83 if (defined &DynaLoader::boot_DynaLoader) {
88 # We need to work for miniperl, because the XS toolchain uses Text::Wrap, which
94 # We call install() every time, as if we didn't, we wouldn't
95 # "see" any changes to the color environment var since
96 # the last time it was called.
98 # install() returns an integer, which if casted properly
99 # in C resolves to a structure containing the regexp
100 # hooks. Setting it to a random integer will guarantee
102 $^H{regcomp} = install();
113 Carp::carp("Useless use of \"re\" pragma");
116 foreach my $idx (0..$#_){
118 if ($s eq 'Debug' or $s eq 'Debugcolor') {
119 setcolor() if $s =~/color/i;
120 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
121 for my $idx ($idx+1..$#_) {
122 if ($flags{$_[$idx]}) {
124 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
126 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
130 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
131 join(", ",sort keys %flags ) );
134 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
136 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
137 setcolor() if $s =~/color/i;
140 } elsif (exists $bitmask{$s}) {
141 $bits |= $bitmask{$s};
142 } elsif ($EXPORT_OK{$s}) {
144 re->export_to_level(2, 're', $s);
145 } elsif ($s =~ s/^\///) {
146 my $reflags = $^H{reflags} || 0;
148 while ($s =~ m/( . )/gx) {
151 # The 'a' may be repeated; hide this from the rest of the
152 # code by counting and getting rid of all of them, then
153 # changing to 'aa' if there is a repeat.
155 my $sav_pos = pos $s;
156 my $a_count = $s =~ s/a//g;
157 pos $s = $sav_pos - 1; # -1 because got rid of the 'a'
161 qq 'The "a" flag may only appear a maximum of twice'
164 elsif ($a_count == 2) {
171 if ($seen_charset ne $_) {
173 qq 'The "$seen_charset" and "$_" flags '
179 qq 'The "$seen_charset" flag may not appear '
184 $^H{reflags_charset} = $reflags{$_};
188 delete $^H{reflags_charset}
189 if defined $^H{reflags_charset}
190 && $^H{reflags_charset} == $reflags{$_};
192 } elsif (exists $reflags{$_}) {
194 ? $reflags |= $reflags{$_}
195 : ($reflags &= ~$reflags{$_});
199 qq'Unknown regular expression flag "$_"'
204 ($^H{reflags} = $reflags or defined $^H{reflags_charset})
206 : ($^H &= ~$flags_hint);
209 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
210 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
224 $^H &= ~ bits(0, @_);
233 re - Perl pragma to alter regular expression behaviour
238 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
240 $pat = '(?{ $foo = 1 })';
242 /foo${pat}bar/; # won't fail (when not under -T switch)
245 no re 'taint'; # the default
246 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
248 no re 'eval'; # the default
249 /foo${pat}bar/; # disallowed (with or without -T switch)
253 "FOO" =~ / foo /; # /ix implied
255 "FOO" =~ /foo/; # just /i implied
257 use re 'debug'; # output debugging info during
258 /^(.*)$/s; # compile and run time
261 use re 'debugcolor'; # same as 'debug', but with colored output
264 use re qw(Debug All); # Finer tuned debugging options.
265 use re qw(Debug More);
266 no re qw(Debug ALL); # Turn of all re debugging in this scope
268 use re qw(is_regexp regexp_pattern); # import utility functions
269 my ($pat,$mods)=regexp_pattern(qr/foo/i);
270 if (is_regexp($obj)) {
271 print "Got regexp: ",
272 scalar regexp_pattern($obj); # just as perl would stringify it
273 } # but no hassle with blessed re's.
275 (We use $^X in these examples because it's tainted by default.)
281 When C<use re 'taint'> is in effect, and a tainted string is the target
282 of a regexp, the regexp memories (or values returned by the m// operator
283 in list context) are tainted. This feature is useful when regexp operations
284 on tainted data aren't meant to extract safe substrings, but to perform
285 other transformations.
289 When C<use re 'eval'> is in effect, a regexp is allowed to contain
290 C<(?{ ... })> zero-width assertions and C<(??{ ... })> postponed
291 subexpressions, even if the regular expression contains
292 variable interpolation. That is normally disallowed, since it is a
293 potential security risk. Note that this pragma is ignored when the regular
294 expression is obtained from tainted data, i.e. evaluation is always
295 disallowed with tainted regular expressions. See L<perlre/(?{ code })>
296 and L<perlre/(??{ code })>.
298 For the purpose of this pragma, interpolation of precompiled regular
299 expressions (i.e., the result of C<qr//>) is I<not> considered variable
304 I<is> allowed if $pat is a precompiled regular expression, even
305 if $pat contains C<(?{ ... })> assertions or C<(??{ ... })> subexpressions.
309 When C<use re '/flags'> is specified, the given flags are automatically
310 added to every regular expression till the end of the lexical scope.
312 C<no re '/flags'> will turn off the effect of C<use re '/flags'> for the
315 For example, if you want all your regular expressions to have /msx on by
320 at the top of your code.
322 The character set /adul flags cancel each other out. So, in this example,
329 the second C<use re> does an implicit C<no re '/u'>.
331 Turning on one of the character set flags with C<use re> takes precedence over the
332 C<locale> pragma and the 'unicode_strings' C<feature>, for regular
333 expressions. Turning off one of these flags when it is active reverts to
334 the behaviour specified by whatever other pragmata are in scope. For
337 use feature "unicode_strings";
338 no re "/u"; # does nothing
340 no re "/l"; # reverts to unicode_strings behaviour
344 When C<use re 'debug'> is in effect, perl emits debugging messages when
345 compiling and using regular expressions. The output is the same as that
346 obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
347 B<-Dr> switch. It may be quite voluminous depending on the complexity
348 of the match. Using C<debugcolor> instead of C<debug> enables a
349 form of output that can be used to get a colorful display on terminals
350 that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
351 comma-separated list of C<termcap> properties to use for highlighting
352 strings on/off, pre-point part on/off.
353 See L<perldebug/"Debugging Regular Expressions"> for additional info.
355 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
356 lexically scoped, as the other directives are. However they have both
357 compile-time and run-time effects.
359 See L<perlmodlib/Pragmatic Modules>.
363 Similarly C<use re 'Debug'> produces debugging output, the difference
364 being that it allows the fine tuning of what debugging output will be
365 emitted. Options are divided into three groups, those related to
366 compilation, those related to execution and those related to special
367 purposes. The options are as follows:
371 =item Compile related options
377 Turns on all compile related debug options.
381 Turns on debug output related to the process of parsing the pattern.
385 Enables output related to the optimisation phase of compilation.
389 Detailed info about trie compilation.
393 Dump the final program out after it is compiled and optimised.
397 =item Execute related options
403 Turns on all execute related debug options.
407 Turns on debugging of the main matching loop.
411 Extra debugging of how tries execute.
415 Enable debugging of start point optimisations.
419 =item Extra debugging options
425 Turns on all "extra" debugging options.
429 Enable debugging the capture group storage during match. Warning,
430 this can potentially produce extremely large output.
434 Enable enhanced TRIE debugging. Enhances both TRIEE
439 Enable debugging of states in the engine.
443 Enable debugging of the recursion stack in the engine. Enabling
444 or disabling this option automatically does the same for debugging
445 states as well. This output from this can be quite large.
449 Enable enhanced optimisation debugging and start point optimisations.
450 Probably not useful except when debugging the regexp engine itself.
454 Dump offset information. This can be used to see how regops correlate
455 to the pattern. Output format is
457 NODENUM:POSITION[LENGTH]
459 Where 1 is the position of the first char in the string. Note that position
460 can be 0, or larger than the actual length of the pattern, likewise length
465 Enable debugging of offsets information. This emits copious
466 amounts of trace information and doesn't mesh well with other
469 Almost definitely only useful to people hacking
470 on the offsets part of the debug engine.
474 =item Other useful flags
476 These are useful shortcuts to save on the typing.
482 Enable all options at once except OFFSETS, OFFSETSDBG and BUFFERS
486 Enable DUMP and all execute options. Equivalent to:
494 Enable TRIEM and all execute compile and execute options.
500 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
501 lexically scoped, as the other directives are. However they have both
502 compile-time and run-time effects.
504 =head2 Exportable Functions
506 As of perl 5.9.5 're' debug contains a number of utility functions that
507 may be optionally exported into the caller's namespace. They are listed
512 =item is_regexp($ref)
514 Returns true if the argument is a compiled regular expression as returned
515 by C<qr//>, false if it is not.
517 This function will not be confused by overloading or blessing. In
518 internals terms, this extracts the regexp pointer out of the
519 PERL_MAGIC_qr structure so it cannot be fooled.
521 =item regexp_pattern($ref)
523 If the argument is a compiled regular expression as returned by C<qr//>,
524 then this function returns the pattern.
526 In list context it returns a two element list, the first element
527 containing the pattern and the second containing the modifiers used when
528 the pattern was compiled.
530 my ($pat, $mods) = regexp_pattern($ref);
532 In scalar context it returns the same as perl would when stringifying a raw
533 C<qr//> with the same pattern inside. If the argument is not a compiled
534 reference then this routine returns false but defined in scalar context,
535 and the empty list in list context. Thus the following
537 if (regexp_pattern($ref) eq '(?^i:foo)')
539 will be warning free regardless of what $ref actually is.
541 Like C<is_regexp> this function will not be confused by overloading
542 or blessing of the object.
546 If the argument is a compiled regular expression as returned by C<qr//>,
547 then this function returns what the optimiser considers to be the longest
548 anchored fixed string and longest floating fixed string in the pattern.
550 A I<fixed string> is defined as being a substring that must appear for the
551 pattern to match. An I<anchored fixed string> is a fixed string that must
552 appear at a particular offset from the beginning of the match. A I<floating
553 fixed string> is defined as a fixed string that can appear at any point in
554 a range of positions relative to the start of the match. For example,
556 my $qr = qr/here .* there/x;
557 my ($anchored, $floating) = regmust($qr);
558 print "anchored:'$anchored'\nfloating:'$floating'\n";
565 Because the C<here> is before the C<.*> in the pattern, its position
566 can be determined exactly. That's not true, however, for the C<there>;
567 it could appear at any point after where the anchored string appeared.
568 Perl uses both for its optimisations, prefering the longer, or, if they are
571 B<NOTE:> This may not necessarily be the definitive longest anchored and
572 floating string. This will be what the optimiser of the Perl that you
573 are using thinks is the longest. If you believe that the result is wrong
574 please report it via the L<perlbug> utility.
576 =item regname($name,$all)
578 Returns the contents of a named buffer of the last successful match. If
579 $all is true, then returns an array ref containing one entry per buffer,
580 otherwise returns the first defined buffer.
584 Returns a list of all of the named buffers defined in the last successful
585 match. If $all is true, then it returns all names defined, if not it returns
586 only names which were involved in the match.
588 =item regnames_count()
590 Returns the number of distinct names defined in the pattern used
591 for the last successful match.
593 B<Note:> this result is always the actual number of distinct
594 named buffers defined, it may not actually match that which is
595 returned by C<regnames()> and related routines when those routines
596 have not been called with the $all parameter set.
602 L<perlmodlib/Pragmatic Modules>.