3 # pragma for controlling the regex engine
8 our @ISA = qw(Exporter);
9 my @XS_FUNCTIONS = qw(regmust);
10 my %XS_FUNCTIONS = map { $_ => 1 } @XS_FUNCTIONS;
11 our @EXPORT_OK = (@XS_FUNCTIONS,
12 qw(is_regexp regexp_pattern
13 regname regnames regnames_count));
14 our %EXPORT_OK = map { $_ => 1 } @EXPORT_OK;
16 # *** WARNING *** WARNING *** WARNING *** WARNING *** WARNING ***
18 # If you modify these values see comment below!
21 taint => 0x00100000, # HINT_RE_TAINT
22 eval => 0x00200000, # HINT_RE_EVAL
25 # - File::Basename contains a literal for 'taint' as a fallback. If
26 # taint is changed here, File::Basename must be updated as well.
28 # - ExtUtils::ParseXS uses a hardcoded
29 # BEGIN { $^H |= 0x00200000 }
30 # in it to allow re.xs to be built. So if 'eval' is changed here then
31 # ExtUtils::ParseXS must be changed as well.
33 # *** WARNING *** WARNING *** WARNING *** WARNING *** WARNING ***
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} | $flags{TRIEC};
87 if ( ! defined($installed) ) {
89 $installed = eval { XSLoader::load('re', $VERSION) } || 0;
90 $installed_error = $@;
99 die "'re' not installed!? ($installed_error)";
101 # We call install() every time, as if we didn't, we wouldn't
102 # "see" any changes to the color environment var since
103 # the last time it was called.
105 # install() returns an integer, which if casted properly
106 # in C resolves to a structure containing the regex
107 # hooks. Setting it to a random integer will guarantee
109 $^H{regcomp} = install();
121 Carp::carp("Useless use of \"re\" pragma");
123 foreach my $idx (0..$#_){
125 if ($s eq 'Debug' or $s eq 'Debugcolor') {
126 setcolor() if $s =~/color/i;
127 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
128 for my $idx ($idx+1..$#_) {
129 if ($flags{$_[$idx]}) {
131 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
133 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
137 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
138 join(", ",sort keys %flags ) );
141 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
143 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
144 setcolor() if $s =~/color/i;
147 } elsif (exists $bitmask{$s}) {
148 $bits |= $bitmask{$s};
149 } elsif ($XS_FUNCTIONS{$s}) {
153 Carp::croak("\"re\" function '$s' not available");
156 re->export_to_level(2, 're', $s);
157 } elsif ($EXPORT_OK{$s}) {
159 re->export_to_level(2, 're', $s);
162 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
163 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
177 $^H &= ~ bits(0, @_);
186 re - Perl pragma to alter regular expression behaviour
191 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
193 $pat = '(?{ $foo = 1 })';
195 /foo${pat}bar/; # won't fail (when not under -T switch)
198 no re 'taint'; # the default
199 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
201 no re 'eval'; # the default
202 /foo${pat}bar/; # disallowed (with or without -T switch)
205 use re 'debug'; # output debugging info during
206 /^(.*)$/s; # compile and run time
209 use re 'debugcolor'; # same as 'debug', but with colored output
212 use re qw(Debug All); # Finer tuned debugging options.
213 use re qw(Debug More);
214 no re qw(Debug ALL); # Turn of all re debugging in this scope
216 use re qw(is_regexp regexp_pattern); # import utility functions
217 my ($pat,$mods)=regexp_pattern(qr/foo/i);
218 if (is_regexp($obj)) {
219 print "Got regexp: ",
220 scalar regexp_pattern($obj); # just as perl would stringify it
221 } # but no hassle with blessed re's.
223 (We use $^X in these examples because it's tainted by default.)
229 When C<use re 'taint'> is in effect, and a tainted string is the target
230 of a regex, the regex memories (or values returned by the m// operator
231 in list context) are tainted. This feature is useful when regex operations
232 on tainted data aren't meant to extract safe substrings, but to perform
233 other transformations.
237 When C<use re 'eval'> is in effect, a regex is allowed to contain
238 C<(?{ ... })> zero-width assertions even if regular expression contains
239 variable interpolation. That is normally disallowed, since it is a
240 potential security risk. Note that this pragma is ignored when the regular
241 expression is obtained from tainted data, i.e. evaluation is always
242 disallowed with tainted regular expressions. See L<perlre/(?{ code })>.
244 For the purpose of this pragma, interpolation of precompiled regular
245 expressions (i.e., the result of C<qr//>) is I<not> considered variable
250 I<is> allowed if $pat is a precompiled regular expression, even
251 if $pat contains C<(?{ ... })> assertions.
255 When C<use re 'debug'> is in effect, perl emits debugging messages when
256 compiling and using regular expressions. The output is the same as that
257 obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
258 B<-Dr> switch. It may be quite voluminous depending on the complexity
259 of the match. Using C<debugcolor> instead of C<debug> enables a
260 form of output that can be used to get a colorful display on terminals
261 that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
262 comma-separated list of C<termcap> properties to use for highlighting
263 strings on/off, pre-point part on/off.
264 See L<perldebug/"Debugging regular expressions"> for additional info.
266 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
267 lexically scoped, as the other directives are. However they have both
268 compile-time and run-time effects.
270 See L<perlmodlib/Pragmatic Modules>.
274 Similarly C<use re 'Debug'> produces debugging output, the difference
275 being that it allows the fine tuning of what debugging output will be
276 emitted. Options are divided into three groups, those related to
277 compilation, those related to execution and those related to special
278 purposes. The options are as follows:
282 =item Compile related options
288 Turns on all compile related debug options.
292 Turns on debug output related to the process of parsing the pattern.
296 Enables output related to the optimisation phase of compilation.
300 Detailed info about trie compilation.
304 Dump the final program out after it is compiled and optimised.
308 =item Execute related options
314 Turns on all execute related debug options.
318 Turns on debugging of the main matching loop.
322 Extra debugging of how tries execute.
326 Enable debugging of start point optimisations.
330 =item Extra debugging options
336 Turns on all "extra" debugging options.
340 Enable debugging the capture buffer storage during match. Warning,
341 this can potentially produce extremely large output.
345 Enable enhanced TRIE debugging. Enhances both TRIEE
350 Enable debugging of states in the engine.
354 Enable debugging of the recursion stack in the engine. Enabling
355 or disabling this option automatically does the same for debugging
356 states as well. This output from this can be quite large.
360 Enable enhanced optimisation debugging and start point optimisations.
361 Probably not useful except when debugging the regex engine itself.
365 Dump offset information. This can be used to see how regops correlate
366 to the pattern. Output format is
368 NODENUM:POSITION[LENGTH]
370 Where 1 is the position of the first char in the string. Note that position
371 can be 0, or larger than the actual length of the pattern, likewise length
376 Enable debugging of offsets information. This emits copious
377 amounts of trace information and doesn't mesh well with other
380 Almost definitely only useful to people hacking
381 on the offsets part of the debug engine.
385 =item Other useful flags
387 These are useful shortcuts to save on the typing.
393 Enable all options at once except OFFSETS, OFFSETSDBG and BUFFERS
397 Enable DUMP and all execute options. Equivalent to:
405 Enable TRIEM and all execute compile and execute options.
411 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
412 lexically scoped, as the other directives are. However they have both
413 compile-time and run-time effects.
415 =head2 Exportable Functions
417 As of perl 5.9.5 're' debug contains a number of utility functions that
418 may be optionally exported into the caller's namespace. They are listed
423 =item is_regexp($ref)
425 Returns true if the argument is a compiled regular expression as returned
426 by C<qr//>, false if it is not.
428 This function will not be confused by overloading or blessing. In
429 internals terms, this extracts the regexp pointer out of the
430 PERL_MAGIC_qr structure so it it cannot be fooled.
432 =item regexp_pattern($ref)
434 If the argument is a compiled regular expression as returned by C<qr//>,
435 then this function returns the pattern.
437 In list context it returns a two element list, the first element
438 containing the pattern and the second containing the modifiers used when
439 the pattern was compiled.
441 my ($pat, $mods) = regexp_pattern($ref);
443 In scalar context it returns the same as perl would when strigifying a raw
444 C<qr//> with the same pattern inside. If the argument is not a compiled
445 reference then this routine returns false but defined in scalar context,
446 and the empty list in list context. Thus the following
448 if (regexp_pattern($ref) eq '(?i-xsm:foo)')
450 will be warning free regardless of what $ref actually is.
452 Like C<is_regexp> this function will not be confused by overloading
453 or blessing of the object.
457 If the argument is a compiled regular expression as returned by C<qr//>,
458 then this function returns what the optimiser consiers to be the longest
459 anchored fixed string and longest floating fixed string in the pattern.
461 A I<fixed string> is defined as being a substring that must appear for the
462 pattern to match. An I<anchored fixed string> is a fixed string that must
463 appear at a particular offset from the beginning of the match. A I<floating
464 fixed string> is defined as a fixed string that can appear at any point in
465 a range of positions relative to the start of the match. For example,
467 my $qr = qr/here .* there/x;
468 my ($anchored, $floating) = regmust($qr);
469 print "anchored:'$anchored'\nfloating:'$floating'\n";
476 Because the C<here> is before the C<.*> in the pattern, its position
477 can be determined exactly. That's not true, however, for the C<there>;
478 it could appear at any point after where the anchored string appeared.
479 Perl uses both for its optimisations, prefering the longer, or, if they are
482 B<NOTE:> This may not necessarily be the definitive longest anchored and
483 floating string. This will be what the optimiser of the Perl that you
484 are using thinks is the longest. If you believe that the result is wrong
485 please report it via the L<perlbug> utility.
487 =item regname($name,$all)
489 Returns the contents of a named buffer of the last successful match. If
490 $all is true, then returns an array ref containing one entry per buffer,
491 otherwise returns the first defined buffer.
495 Returns a list of all of the named buffers defined in the last successful
496 match. If $all is true, then it returns all names defined, if not it returns
497 only names which were involved in the match.
499 =item regnames_count()
501 Returns the number of distinct names defined in the pattern used
502 for the last successful match.
504 B<Note:> this result is always the actual number of distinct
505 named buffers defined, it may not actually match that which is
506 returned by C<regnames()> and related routines when those routines
507 have not been called with the $all parameter set.
513 L<perlmodlib/Pragmatic Modules>.