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,
75 $flags{ALL} = -1 & ~($flags{OFFSETS}|$flags{OFFSETSDBG}|$flags{BUFFERS});
76 $flags{All} = $flags{all} = $flags{DUMP} | $flags{EXECUTE};
77 $flags{Extra} = $flags{EXECUTE} | $flags{COMPILE};
78 $flags{More} = $flags{MORE} = $flags{All} | $flags{TRIEC} | $flags{TRIEM} | $flags{STATE};
79 $flags{State} = $flags{DUMP} | $flags{EXECUTE} | $flags{STATE};
80 $flags{TRIE} = $flags{DUMP} | $flags{EXECUTE} | $flags{TRIEC};
86 if ( ! defined($installed) ) {
88 $installed = eval { XSLoader::load('re', $VERSION) } || 0;
89 $installed_error = $@;
98 die "'re' not installed!? ($installed_error)";
100 # We call install() every time, as if we didn't, we wouldn't
101 # "see" any changes to the color environment var since
102 # the last time it was called.
104 # install() returns an integer, which if casted properly
105 # in C resolves to a structure containing the regex
106 # hooks. Setting it to a random integer will guarantee
108 $^H{regcomp} = install();
120 Carp::carp("Useless use of \"re\" pragma");
122 foreach my $idx (0..$#_){
124 if ($s eq 'Debug' or $s eq 'Debugcolor') {
125 setcolor() if $s =~/color/i;
126 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
127 for my $idx ($idx+1..$#_) {
128 if ($flags{$_[$idx]}) {
130 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
132 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
136 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
137 join(", ",sort keys %flags ) );
140 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
142 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
143 setcolor() if $s =~/color/i;
146 } elsif (exists $bitmask{$s}) {
147 $bits |= $bitmask{$s};
148 } elsif ($XS_FUNCTIONS{$s}) {
152 Carp::croak("\"re\" function '$s' not available");
155 re->export_to_level(2, 're', $s);
156 } elsif ($EXPORT_OK{$s}) {
158 re->export_to_level(2, 're', $s);
161 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
162 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
176 $^H &= ~ bits(0, @_);
185 re - Perl pragma to alter regular expression behaviour
190 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
192 $pat = '(?{ $foo = 1 })';
194 /foo${pat}bar/; # won't fail (when not under -T switch)
197 no re 'taint'; # the default
198 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
200 no re 'eval'; # the default
201 /foo${pat}bar/; # disallowed (with or without -T switch)
204 use re 'debug'; # output debugging info during
205 /^(.*)$/s; # compile and run time
208 use re 'debugcolor'; # same as 'debug', but with colored output
211 use re qw(Debug All); # Finer tuned debugging options.
212 use re qw(Debug More);
213 no re qw(Debug ALL); # Turn of all re debugging in this scope
215 use re qw(is_regexp regexp_pattern); # import utility functions
216 my ($pat,$mods)=regexp_pattern(qr/foo/i);
217 if (is_regexp($obj)) {
218 print "Got regexp: ",
219 scalar regexp_pattern($obj); # just as perl would stringify it
220 } # but no hassle with blessed re's.
222 (We use $^X in these examples because it's tainted by default.)
228 When C<use re 'taint'> is in effect, and a tainted string is the target
229 of a regex, the regex memories (or values returned by the m// operator
230 in list context) are tainted. This feature is useful when regex operations
231 on tainted data aren't meant to extract safe substrings, but to perform
232 other transformations.
236 When C<use re 'eval'> is in effect, a regex is allowed to contain
237 C<(?{ ... })> zero-width assertions even if regular expression contains
238 variable interpolation. That is normally disallowed, since it is a
239 potential security risk. Note that this pragma is ignored when the regular
240 expression is obtained from tainted data, i.e. evaluation is always
241 disallowed with tainted regular expressions. See L<perlre/(?{ code })>.
243 For the purpose of this pragma, interpolation of precompiled regular
244 expressions (i.e., the result of C<qr//>) is I<not> considered variable
249 I<is> allowed if $pat is a precompiled regular expression, even
250 if $pat contains C<(?{ ... })> assertions.
254 When C<use re 'debug'> is in effect, perl emits debugging messages when
255 compiling and using regular expressions. The output is the same as that
256 obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
257 B<-Dr> switch. It may be quite voluminous depending on the complexity
258 of the match. Using C<debugcolor> instead of C<debug> enables a
259 form of output that can be used to get a colorful display on terminals
260 that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
261 comma-separated list of C<termcap> properties to use for highlighting
262 strings on/off, pre-point part on/off.
263 See L<perldebug/"Debugging regular expressions"> for additional info.
265 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
266 lexically scoped, as the other directives are. However they have both
267 compile-time and run-time effects.
269 See L<perlmodlib/Pragmatic Modules>.
273 Similarly C<use re 'Debug'> produces debugging output, the difference
274 being that it allows the fine tuning of what debugging output will be
275 emitted. Options are divided into three groups, those related to
276 compilation, those related to execution and those related to special
277 purposes. The options are as follows:
281 =item Compile related options
287 Turns on all compile related debug options.
291 Turns on debug output related to the process of parsing the pattern.
295 Enables output related to the optimisation phase of compilation.
299 Detailed info about trie compilation.
303 Dump the final program out after it is compiled and optimised.
307 =item Execute related options
313 Turns on all execute related debug options.
317 Turns on debugging of the main matching loop.
321 Extra debugging of how tries execute.
325 Enable debugging of start point optimisations.
329 =item Extra debugging options
335 Turns on all "extra" debugging options.
339 Enable debugging the capture buffer storage during match. Warning,
340 this can potentially produce extremely large output.
344 Enable enhanced TRIE debugging. Enhances both TRIEE
349 Enable debugging of states in the engine.
353 Enable debugging of the recursion stack in the engine. Enabling
354 or disabling this option automatically does the same for debugging
355 states as well. This output from this can be quite large.
359 Enable enhanced optimisation debugging and start point optimisations.
360 Probably not useful except when debugging the regex engine itself.
364 Dump offset information. This can be used to see how regops correlate
365 to the pattern. Output format is
367 NODENUM:POSITION[LENGTH]
369 Where 1 is the position of the first char in the string. Note that position
370 can be 0, or larger than the actual length of the pattern, likewise length
375 Enable debugging of offsets information. This emits copious
376 amounts of trace information and doesn't mesh well with other
379 Almost definitely only useful to people hacking
380 on the offsets part of the debug engine.
384 =item Other useful flags
386 These are useful shortcuts to save on the typing.
392 Enable all options at once except OFFSETS, OFFSETSDBG and BUFFERS
396 Enable DUMP and all execute options. Equivalent to:
404 Enable TRIEM and all execute compile and execute options.
410 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
411 lexically scoped, as the other directives are. However they have both
412 compile-time and run-time effects.
414 =head2 Exportable Functions
416 As of perl 5.9.5 're' debug contains a number of utility functions that
417 may be optionally exported into the caller's namespace. They are listed
422 =item is_regexp($ref)
424 Returns true if the argument is a compiled regular expression as returned
425 by C<qr//>, false if it is not.
427 This function will not be confused by overloading or blessing. In
428 internals terms, this extracts the regexp pointer out of the
429 PERL_MAGIC_qr structure so it it cannot be fooled.
431 =item regexp_pattern($ref)
433 If the argument is a compiled regular expression as returned by C<qr//>,
434 then this function returns the pattern.
436 In list context it returns a two element list, the first element
437 containing the pattern and the second containing the modifiers used when
438 the pattern was compiled.
440 my ($pat, $mods) = regexp_pattern($ref);
442 In scalar context it returns the same as perl would when strigifying a raw
443 C<qr//> with the same pattern inside. If the argument is not a compiled
444 reference then this routine returns false but defined in scalar context,
445 and the empty list in list context. Thus the following
447 if (regexp_pattern($ref) eq '(?i-xsm:foo)')
449 will be warning free regardless of what $ref actually is.
451 Like C<is_regexp> this function will not be confused by overloading
452 or blessing of the object.
456 If the argument is a compiled regular expression as returned by C<qr//>,
457 then this function returns what the optimiser consiers to be the longest
458 anchored fixed string and longest floating fixed string in the pattern.
460 A I<fixed string> is defined as being a substring that must appear for the
461 pattern to match. An I<anchored fixed string> is a fixed string that must
462 appear at a particular offset from the beginning of the match. A I<floating
463 fixed string> is defined as a fixed string that can appear at any point in
464 a range of positions relative to the start of the match. For example,
466 my $qr = qr/here .* there/x;
467 my ($anchored, $floating) = regmust($qr);
468 print "anchored:'$anchored'\nfloating:'$floating'\n";
475 Because the C<here> is before the C<.*> in the pattern, its position
476 can be determined exactly. That's not true, however, for the C<there>;
477 it could appear at any point after where the anchored string appeared.
478 Perl uses both for its optimisations, prefering the longer, or, if they are
481 B<NOTE:> This may not necessarily be the definitive longest anchored and
482 floating string. This will be what the optimiser of the Perl that you
483 are using thinks is the longest. If you believe that the result is wrong
484 please report it via the L<perlbug> utility.
486 =item regname($name,$all)
488 Returns the contents of a named buffer of the last successful match. If
489 $all is true, then returns an array ref containing one entry per buffer,
490 otherwise returns the first defined buffer.
494 Returns a list of all of the named buffers defined in the last successful
495 match. If $all is true, then it returns all names defined, if not it returns
496 only names which were involved in the match.
498 =item regnames_count()
500 Returns the number of distinct names defined in the pattern used
501 for the last successful match.
503 B<Note:> this result is always the actual number of distinct
504 named buffers defined, it may not actually match that which is
505 returned by C<regnames()> and related routines when those routines
506 have not been called with the $all parameter set.
512 L<perlmodlib/Pragmatic Modules>.