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();
112 foreach my $idx (0..$#_){
114 if ($s eq 'Debug' or $s eq 'Debugcolor') {
115 setcolor() if $s =~/color/i;
116 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
117 for my $idx ($idx+1..$#_) {
118 if ($flags{$_[$idx]}) {
120 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
122 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
126 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
127 join(", ",sort keys %flags ) );
130 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
132 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
133 setcolor() if $s =~/color/i;
136 } elsif (exists $bitmask{$s}) {
137 $bits |= $bitmask{$s};
138 } elsif ($EXPORT_OK{$s}) {
140 re->export_to_level(2, 're', $s);
141 } elsif ($s =~ s/^\///) {
142 my $reflags = $^H{reflags} || 0;
144 while ($s =~ m/( . )/gx) {
147 # The 'a' may be repeated; hide this from the rest of the
148 # code by counting and getting rid of all of them, then
149 # changing to 'aa' if there is a repeat.
151 my $sav_pos = pos $s;
152 my $a_count = $s =~ s/a//g;
153 pos $s = $sav_pos - 1; # -1 because got rid of the 'a'
157 qq 'The "a" flag may only appear a maximum of twice'
160 elsif ($a_count == 2) {
167 if ($seen_charset ne $_) {
169 qq 'The "$seen_charset" and "$_" flags '
175 qq 'The "$seen_charset" flag may not appear '
180 $^H{reflags_charset} = $reflags{$_};
184 delete $^H{reflags_charset}
185 if defined $^H{reflags_charset}
186 && $^H{reflags_charset} == $reflags{$_};
188 } elsif (exists $reflags{$_}) {
190 ? $reflags |= $reflags{$_}
191 : ($reflags &= ~$reflags{$_});
195 qq'Unknown regular expression flag "$_"'
200 ($^H{reflags} = $reflags or defined $^H{reflags_charset})
202 : ($^H &= ~$flags_hint);
205 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
206 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
220 $^H &= ~ bits(0, @_);
229 re - Perl pragma to alter regular expression behaviour
234 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
236 $pat = '(?{ $foo = 1 })';
238 /foo${pat}bar/; # won't fail (when not under -T
242 no re 'taint'; # the default
243 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
245 no re 'eval'; # the default
246 /foo${pat}bar/; # disallowed (with or without -T
251 "FOO" =~ / foo /; # /ix implied
253 "FOO" =~ /foo/; # just /i implied
255 use re 'debug'; # output debugging info during
256 /^(.*)$/s; # compile and run time
259 use re 'debugcolor'; # same as 'debug', but with colored
263 use re qw(Debug All); # Same as "use re 'debug'", but you
264 # can use "Debug" with things other
266 use re qw(Debug More); # 'All' plus output more details
267 no re qw(Debug ALL); # Turn on (almost) all re debugging
270 use re qw(is_regexp regexp_pattern); # import utility functions
271 my ($pat,$mods)=regexp_pattern(qr/foo/i);
272 if (is_regexp($obj)) {
273 print "Got regexp: ",
274 scalar regexp_pattern($obj); # just as perl would stringify
275 } # it but no hassle with blessed
278 (We use $^X in these examples because it's tainted by default.)
284 When C<use re 'taint'> is in effect, and a tainted string is the target
285 of a regexp, the regexp memories (or values returned by the m// operator
286 in list context) are tainted. This feature is useful when regexp operations
287 on tainted data aren't meant to extract safe substrings, but to perform
288 other transformations.
292 When C<use re 'eval'> is in effect, a regexp is allowed to contain
293 C<(?{ ... })> zero-width assertions and C<(??{ ... })> postponed
294 subexpressions that are derived from variable interpolation, rather than
295 appearing literally within the regexp. That is normally disallowed, since
297 potential security risk. Note that this pragma is ignored when the regular
298 expression is obtained from tainted data, i.e. evaluation is always
299 disallowed with tainted regular expressions. See L<perlre/(?{ code })>
300 and L<perlre/(??{ code })>.
302 For the purpose of this pragma, interpolation of precompiled regular
303 expressions (i.e., the result of C<qr//>) is I<not> considered variable
308 I<is> allowed if $pat is a precompiled regular expression, even
309 if $pat contains C<(?{ ... })> assertions or C<(??{ ... })> subexpressions.
313 When C<use re '/flags'> is specified, the given flags are automatically
314 added to every regular expression till the end of the lexical scope.
316 C<no re '/flags'> will turn off the effect of C<use re '/flags'> for the
319 For example, if you want all your regular expressions to have /msx on by
324 at the top of your code.
326 The character set /adul flags cancel each other out. So, in this example,
333 the second C<use re> does an implicit C<no re '/u'>.
335 Turning on one of the character set flags with C<use re> takes precedence over the
336 C<locale> pragma and the 'unicode_strings' C<feature>, for regular
337 expressions. Turning off one of these flags when it is active reverts to
338 the behaviour specified by whatever other pragmata are in scope. For
341 use feature "unicode_strings";
342 no re "/u"; # does nothing
344 no re "/l"; # reverts to unicode_strings behaviour
348 When C<use re 'debug'> is in effect, perl emits debugging messages when
349 compiling and using regular expressions. The output is the same as that
350 obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
351 B<-Dr> switch. It may be quite voluminous depending on the complexity
352 of the match. Using C<debugcolor> instead of C<debug> enables a
353 form of output that can be used to get a colorful display on terminals
354 that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
355 comma-separated list of C<termcap> properties to use for highlighting
356 strings on/off, pre-point part on/off.
357 See L<perldebug/"Debugging Regular Expressions"> for additional info.
359 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
360 lexically scoped, as the other directives are. However they have both
361 compile-time and run-time effects.
363 See L<perlmodlib/Pragmatic Modules>.
367 Similarly C<use re 'Debug'> produces debugging output, the difference
368 being that it allows the fine tuning of what debugging output will be
369 emitted. Options are divided into three groups, those related to
370 compilation, those related to execution and those related to special
371 purposes. The options are as follows:
375 =item Compile related options
381 Turns on all compile related debug options.
385 Turns on debug output related to the process of parsing the pattern.
389 Enables output related to the optimisation phase of compilation.
393 Detailed info about trie compilation.
397 Dump the final program out after it is compiled and optimised.
401 =item Execute related options
407 Turns on all execute related debug options.
411 Turns on debugging of the main matching loop.
415 Extra debugging of how tries execute.
419 Enable debugging of start-point optimisations.
423 =item Extra debugging options
429 Turns on all "extra" debugging options.
433 Enable debugging the capture group storage during match. Warning,
434 this can potentially produce extremely large output.
438 Enable enhanced TRIE debugging. Enhances both TRIEE
443 Enable debugging of states in the engine.
447 Enable debugging of the recursion stack in the engine. Enabling
448 or disabling this option automatically does the same for debugging
449 states as well. This output from this can be quite large.
453 Enable enhanced optimisation debugging and start-point optimisations.
454 Probably not useful except when debugging the regexp engine itself.
458 Dump offset information. This can be used to see how regops correlate
459 to the pattern. Output format is
461 NODENUM:POSITION[LENGTH]
463 Where 1 is the position of the first char in the string. Note that position
464 can be 0, or larger than the actual length of the pattern, likewise length
469 Enable debugging of offsets information. This emits copious
470 amounts of trace information and doesn't mesh well with other
473 Almost definitely only useful to people hacking
474 on the offsets part of the debug engine.
478 =item Other useful flags
480 These are useful shortcuts to save on the typing.
486 Enable all options at once except OFFSETS, OFFSETSDBG and BUFFERS.
487 (To get every single option without exception, use both ALL and EXTRA.)
491 Enable DUMP and all execute options. Equivalent to:
499 Enable the options enabled by "All", plus STATE, TRIEC, and TRIEM.
505 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
506 lexically scoped, as are the other directives. However they have both
507 compile-time and run-time effects.
509 =head2 Exportable Functions
511 As of perl 5.9.5 're' debug contains a number of utility functions that
512 may be optionally exported into the caller's namespace. They are listed
517 =item is_regexp($ref)
519 Returns true if the argument is a compiled regular expression as returned
520 by C<qr//>, false if it is not.
522 This function will not be confused by overloading or blessing. In
523 internals terms, this extracts the regexp pointer out of the
524 PERL_MAGIC_qr structure so it cannot be fooled.
526 =item regexp_pattern($ref)
528 If the argument is a compiled regular expression as returned by C<qr//>,
529 then this function returns the pattern.
531 In list context it returns a two element list, the first element
532 containing the pattern and the second containing the modifiers used when
533 the pattern was compiled.
535 my ($pat, $mods) = regexp_pattern($ref);
537 In scalar context it returns the same as perl would when stringifying a raw
538 C<qr//> with the same pattern inside. If the argument is not a compiled
539 reference then this routine returns false but defined in scalar context,
540 and the empty list in list context. Thus the following
542 if (regexp_pattern($ref) eq '(?^i:foo)')
544 will be warning free regardless of what $ref actually is.
546 Like C<is_regexp> this function will not be confused by overloading
547 or blessing of the object.
551 If the argument is a compiled regular expression as returned by C<qr//>,
552 then this function returns what the optimiser considers to be the longest
553 anchored fixed string and longest floating fixed string in the pattern.
555 A I<fixed string> is defined as being a substring that must appear for the
556 pattern to match. An I<anchored fixed string> is a fixed string that must
557 appear at a particular offset from the beginning of the match. A I<floating
558 fixed string> is defined as a fixed string that can appear at any point in
559 a range of positions relative to the start of the match. For example,
561 my $qr = qr/here .* there/x;
562 my ($anchored, $floating) = regmust($qr);
563 print "anchored:'$anchored'\nfloating:'$floating'\n";
570 Because the C<here> is before the C<.*> in the pattern, its position
571 can be determined exactly. That's not true, however, for the C<there>;
572 it could appear at any point after where the anchored string appeared.
573 Perl uses both for its optimisations, preferring the longer, or, if they are
576 B<NOTE:> This may not necessarily be the definitive longest anchored and
577 floating string. This will be what the optimiser of the Perl that you
578 are using thinks is the longest. If you believe that the result is wrong
579 please report it via the L<perlbug> utility.
581 =item regname($name,$all)
583 Returns the contents of a named buffer of the last successful match. If
584 $all is true, then returns an array ref containing one entry per buffer,
585 otherwise returns the first defined buffer.
589 Returns a list of all of the named buffers defined in the last successful
590 match. If $all is true, then it returns all names defined, if not it returns
591 only names which were involved in the match.
593 =item regnames_count()
595 Returns the number of distinct names defined in the pattern used
596 for the last successful match.
598 B<Note:> this result is always the actual number of distinct
599 named buffers defined, it may not actually match that which is
600 returned by C<regnames()> and related routines when those routines
601 have not been called with the $all parameter set.
607 L<perlmodlib/Pragmatic Modules>.