3 # pragma for controlling the regex engine
8 our @ISA = qw(Exporter);
9 our @EXPORT_OK = qw(is_regexp regexp_pattern regmust
11 regnames_count regnames_iterinit regnames_iternext);
12 our %EXPORT_OK = map { $_ => 1 } @EXPORT_OK;
14 # *** WARNING *** WARNING *** WARNING *** WARNING *** WARNING ***
16 # If you modify these values see comment below!
19 taint => 0x00100000, # HINT_RE_TAINT
20 eval => 0x00200000, # HINT_RE_EVAL
23 # - File::Basename contains a literal for 'taint' as a fallback. If
24 # taint is changed here, File::Basename must be updated as well.
26 # - ExtUtils::ParseXS uses a hardcoded
27 # BEGIN { $^H |= 0x00200000 }
28 # in it to allow re.xs to be built. So if 'eval' is changed here then
29 # ExtUtils::ParseXS must be changed as well.
31 # *** WARNING *** WARNING *** WARNING *** WARNING *** WARNING ***
34 eval { # Ignore errors
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;
43 $ENV{PERL_RE_COLORS} = $colors;
46 $ENV{PERL_RE_COLORS} ||= qq'\t\t> <\t> <\t\t';
66 OFFSETSDBG => 0x040000,
68 OPTIMISEM => 0x100000,
72 $flags{All} = $flags{all} = $flags{DUMP} | $flags{EXECUTE};
73 $flags{Extra} = $flags{EXECUTE} | $flags{COMPILE};
74 $flags{More} = $flags{MORE} = $flags{All} | $flags{TRIEC} | $flags{TRIEM} | $flags{STATE};
75 $flags{State} = $flags{DUMP} | $flags{EXECUTE} | $flags{STATE};
76 $flags{TRIE} = $flags{DUMP} | $flags{EXECUTE} | $flags{TRIEC};
82 if ( ! defined($installed) ) {
84 $installed = eval { XSLoader::load('re', $VERSION) } || 0;
85 $installed_error = $@;
94 die "'re' not installed!? ($installed_error)";
96 # We call install() every time, as if we didn't, we wouldn't
97 # "see" any changes to the color environment var since
98 # the last time it was called.
100 # install() returns an integer, which if casted properly
101 # in C resolves to a structure containing the regex
102 # hooks. Setting it to a random integer will guarantee
104 $^H{regcomp} = install();
116 Carp::carp("Useless use of \"re\" pragma");
118 foreach my $idx (0..$#_){
120 if ($s eq 'Debug' or $s eq 'Debugcolor') {
121 setcolor() if $s =~/color/i;
122 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
123 for my $idx ($idx+1..$#_) {
124 if ($flags{$_[$idx]}) {
126 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
128 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
132 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
133 join(", ",sort keys %flags ) );
136 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
138 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
139 setcolor() if $s =~/color/i;
142 } elsif (exists $bitmask{$s}) {
143 $bits |= $bitmask{$s};
144 } elsif ($EXPORT_OK{$s}) {
147 re->export_to_level(2, 're', $s);
150 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
151 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
165 $^H &= ~ bits(0, @_);
174 re - Perl pragma to alter regular expression behaviour
179 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
181 $pat = '(?{ $foo = 1 })';
183 /foo${pat}bar/; # won't fail (when not under -T switch)
186 no re 'taint'; # the default
187 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
189 no re 'eval'; # the default
190 /foo${pat}bar/; # disallowed (with or without -T switch)
193 use re 'debug'; # output debugging info during
194 /^(.*)$/s; # compile and run time
197 use re 'debugcolor'; # same as 'debug', but with colored output
200 use re qw(Debug All); # Finer tuned debugging options.
201 use re qw(Debug More);
202 no re qw(Debug ALL); # Turn of all re debugging in this scope
204 use re qw(is_regexp regexp_pattern); # import utility functions
205 my ($pat,$mods)=regexp_pattern(qr/foo/i);
206 if (is_regexp($obj)) {
207 print "Got regexp: ",
208 scalar regexp_pattern($obj); # just as perl would stringify it
209 } # but no hassle with blessed re's.
211 (We use $^X in these examples because it's tainted by default.)
217 When C<use re 'taint'> is in effect, and a tainted string is the target
218 of a regex, the regex memories (or values returned by the m// operator
219 in list context) are tainted. This feature is useful when regex operations
220 on tainted data aren't meant to extract safe substrings, but to perform
221 other transformations.
225 When C<use re 'eval'> is in effect, a regex is allowed to contain
226 C<(?{ ... })> zero-width assertions even if regular expression contains
227 variable interpolation. That is normally disallowed, since it is a
228 potential security risk. Note that this pragma is ignored when the regular
229 expression is obtained from tainted data, i.e. evaluation is always
230 disallowed with tainted regular expressions. See L<perlre/(?{ code })>.
232 For the purpose of this pragma, interpolation of precompiled regular
233 expressions (i.e., the result of C<qr//>) is I<not> considered variable
238 I<is> allowed if $pat is a precompiled regular expression, even
239 if $pat contains C<(?{ ... })> assertions.
243 When C<use re 'debug'> is in effect, perl emits debugging messages when
244 compiling and using regular expressions. The output is the same as that
245 obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
246 B<-Dr> switch. It may be quite voluminous depending on the complexity
247 of the match. Using C<debugcolor> instead of C<debug> enables a
248 form of output that can be used to get a colorful display on terminals
249 that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
250 comma-separated list of C<termcap> properties to use for highlighting
251 strings on/off, pre-point part on/off.
252 See L<perldebug/"Debugging regular expressions"> for additional info.
254 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
255 lexically scoped, as the other directives are. However they have both
256 compile-time and run-time effects.
258 See L<perlmodlib/Pragmatic Modules>.
262 Similarly C<use re 'Debug'> produces debugging output, the difference
263 being that it allows the fine tuning of what debugging output will be
264 emitted. Options are divided into three groups, those related to
265 compilation, those related to execution and those related to special
266 purposes. The options are as follows:
270 =item Compile related options
276 Turns on all compile related debug options.
280 Turns on debug output related to the process of parsing the pattern.
284 Enables output related to the optimisation phase of compilation.
288 Detailed info about trie compilation.
292 Dump the final program out after it is compiled and optimised.
296 =item Execute related options
302 Turns on all execute related debug options.
306 Turns on debugging of the main matching loop.
310 Extra debugging of how tries execute.
314 Enable debugging of start point optimisations.
318 =item Extra debugging options
324 Turns on all "extra" debugging options.
328 Enable enhanced TRIE debugging. Enhances both TRIEE
333 Enable debugging of states in the engine.
337 Enable debugging of the recursion stack in the engine. Enabling
338 or disabling this option automatically does the same for debugging
339 states as well. This output from this can be quite large.
343 Enable enhanced optimisation debugging and start point optimisations.
344 Probably not useful except when debugging the regex engine itself.
348 Dump offset information. This can be used to see how regops correlate
349 to the pattern. Output format is
351 NODENUM:POSITION[LENGTH]
353 Where 1 is the position of the first char in the string. Note that position
354 can be 0, or larger than the actual length of the pattern, likewise length
359 Enable debugging of offsets information. This emits copious
360 amounts of trace information and doesn't mesh well with other
363 Almost definitely only useful to people hacking
364 on the offsets part of the debug engine.
368 =item Other useful flags
370 These are useful shortcuts to save on the typing.
376 Enable all compile and execute options at once.
380 Enable DUMP and all execute options. Equivalent to:
388 Enable TRIEM and all execute compile and execute options.
394 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
395 lexically scoped, as the other directives are. However they have both
396 compile-time and run-time effects.
398 =head2 Exportable Functions
400 As of perl 5.9.5 're' debug contains a number of utility functions that
401 may be optionally exported into the caller's namespace. They are listed
406 =item is_regexp($ref)
408 Returns true if the argument is a compiled regular expression as returned
409 by C<qr//>, false if it is not.
411 This function will not be confused by overloading or blessing. In
412 internals terms, this extracts the regexp pointer out of the
413 PERL_MAGIC_qr structure so it it cannot be fooled.
415 =item regexp_pattern($ref)
417 If the argument is a compiled regular expression as returned by C<qr//>,
418 then this function returns the pattern.
420 In list context it returns a two element list, the first element
421 containing the pattern and the second containing the modifiers used when
422 the pattern was compiled.
424 my ($pat, $mods) = regexp_pattern($ref);
426 In scalar context it returns the same as perl would when strigifying a raw
427 C<qr//> with the same pattern inside. If the argument is not a compiled
428 reference then this routine returns false but defined in scalar context,
429 and the empty list in list context. Thus the following
431 if (regexp_pattern($ref) eq '(?i-xsm:foo)')
433 will be warning free regardless of what $ref actually is.
435 Like C<is_regexp> this function will not be confused by overloading
436 or blessing of the object.
440 If the argument is a compiled regular expression as returned by C<qr//>,
441 then this function returns what the optimiser consiers to be the longest
442 anchored fixed string and longest floating fixed string in the pattern.
444 A I<fixed string> is defined as being a substring that must appear for the
445 pattern to match. An I<anchored fixed string> is a fixed string that must
446 appear at a particular offset from the beginning of the match. A I<floating
447 fixed string> is defined as a fixed string that can appear at any point in
448 a range of positions relative to the start of the match. For example,
450 my $qr = qr/here .* there/x;
451 my ($anchored, $floating) = regmust($qr);
452 print "anchored:'$anchored'\nfloating:'$floating'\n";
459 Because the C<here> is before the C<.*> in the pattern, its position
460 can be determined exactly. That's not true, however, for the C<there>;
461 it could appear at any point after where the anchored string appeared.
462 Perl uses both for its optimisations, prefering the longer, or, if they are
465 B<NOTE:> This may not necessarily be the definitive longest anchored and
466 floating string. This will be what the optimiser of the Perl that you
467 are using thinks is the longest. If you believe that the result is wrong
468 please report it via the L<perlbug> utility.
470 =item regname($name,$qr,$all)
472 Returns the contents of a named buffer. If $qr is missing, or is not the
473 result of a qr// then returns the result of the last successful match. If
474 $all is true then returns an array ref containing one entry per buffer,
475 otherwise returns the first defined buffer.
477 =item regnames($qr,$all)
479 Returns a list of all of the named buffers defined in a pattern. If
480 $all is true then it returns all names defined, if not returns only
481 names which were involved in the last successful match. If $qr is omitted
482 or is not the result of a qr// then returns the details for the last
485 =item regnames_iterinit($qr)
487 Initializes the internal hash iterator associated to a regexps named capture
488 buffers. If $qr is omitted resets the iterator associated with the regexp used
489 in the last successful match.
491 =item regnames_iternext($qr,$all)
493 Gets the next key from the hash associated with a regexp. If $qr
494 is omitted resets the iterator associated with the regexp used in the
495 last successful match. If $all is true returns the keys of all of the
496 distinct named buffers in the pattern, if not returns only those names
497 used in the last successful match.
499 =item regnames_count($qr)
501 Returns the number of distinct names defined in the regexp $qr. If
502 $qr is omitted or not a regexp returns the count of names in the
503 last successful match.
505 B<Note:> that this result is always the actual number of distinct
506 named buffers defined, it may not actually match that which is
507 returned by C<regnames()> and related routines when those routines
508 have not been called with the $all parameter set..
514 L<perlmodlib/Pragmatic Modules>.