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 $flags{ALL} = -1 & ~($flags{OFFSETS}|$flags{OFFSETSDBG}|$flags{BUFFERS});
73 $flags{All} = $flags{all} = $flags{DUMP} | $flags{EXECUTE};
74 $flags{Extra} = $flags{EXECUTE} | $flags{COMPILE};
75 $flags{More} = $flags{MORE} = $flags{All} | $flags{TRIEC} | $flags{TRIEM} | $flags{STATE};
76 $flags{State} = $flags{DUMP} | $flags{EXECUTE} | $flags{STATE};
77 $flags{TRIE} = $flags{DUMP} | $flags{EXECUTE} | $flags{TRIEC};
83 if ( ! defined($installed) ) {
85 $installed = eval { XSLoader::load('re', $VERSION) } || 0;
86 $installed_error = $@;
95 die "'re' not installed!? ($installed_error)";
97 # We call install() every time, as if we didn't, we wouldn't
98 # "see" any changes to the color environment var since
99 # the last time it was called.
101 # install() returns an integer, which if casted properly
102 # in C resolves to a structure containing the regex
103 # hooks. Setting it to a random integer will guarantee
105 $^H{regcomp} = install();
117 Carp::carp("Useless use of \"re\" pragma");
119 foreach my $idx (0..$#_){
121 if ($s eq 'Debug' or $s eq 'Debugcolor') {
122 setcolor() if $s =~/color/i;
123 ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS};
124 for my $idx ($idx+1..$#_) {
125 if ($flags{$_[$idx]}) {
127 ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]};
129 ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]};
133 Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ",
134 join(", ",sort keys %flags ) );
137 _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS});
139 } elsif ($s eq 'debug' or $s eq 'debugcolor') {
140 setcolor() if $s =~/color/i;
143 } elsif (exists $bitmask{$s}) {
144 $bits |= $bitmask{$s};
145 } elsif ($EXPORT_OK{$s}) {
148 re->export_to_level(2, 're', $s);
151 Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ",
152 join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask),
166 $^H &= ~ bits(0, @_);
175 re - Perl pragma to alter regular expression behaviour
180 ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here
182 $pat = '(?{ $foo = 1 })';
184 /foo${pat}bar/; # won't fail (when not under -T switch)
187 no re 'taint'; # the default
188 ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here
190 no re 'eval'; # the default
191 /foo${pat}bar/; # disallowed (with or without -T switch)
194 use re 'debug'; # output debugging info during
195 /^(.*)$/s; # compile and run time
198 use re 'debugcolor'; # same as 'debug', but with colored output
201 use re qw(Debug All); # Finer tuned debugging options.
202 use re qw(Debug More);
203 no re qw(Debug ALL); # Turn of all re debugging in this scope
205 use re qw(is_regexp regexp_pattern); # import utility functions
206 my ($pat,$mods)=regexp_pattern(qr/foo/i);
207 if (is_regexp($obj)) {
208 print "Got regexp: ",
209 scalar regexp_pattern($obj); # just as perl would stringify it
210 } # but no hassle with blessed re's.
212 (We use $^X in these examples because it's tainted by default.)
218 When C<use re 'taint'> is in effect, and a tainted string is the target
219 of a regex, the regex memories (or values returned by the m// operator
220 in list context) are tainted. This feature is useful when regex operations
221 on tainted data aren't meant to extract safe substrings, but to perform
222 other transformations.
226 When C<use re 'eval'> is in effect, a regex is allowed to contain
227 C<(?{ ... })> zero-width assertions even if regular expression contains
228 variable interpolation. That is normally disallowed, since it is a
229 potential security risk. Note that this pragma is ignored when the regular
230 expression is obtained from tainted data, i.e. evaluation is always
231 disallowed with tainted regular expressions. See L<perlre/(?{ code })>.
233 For the purpose of this pragma, interpolation of precompiled regular
234 expressions (i.e., the result of C<qr//>) is I<not> considered variable
239 I<is> allowed if $pat is a precompiled regular expression, even
240 if $pat contains C<(?{ ... })> assertions.
244 When C<use re 'debug'> is in effect, perl emits debugging messages when
245 compiling and using regular expressions. The output is the same as that
246 obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the
247 B<-Dr> switch. It may be quite voluminous depending on the complexity
248 of the match. Using C<debugcolor> instead of C<debug> enables a
249 form of output that can be used to get a colorful display on terminals
250 that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a
251 comma-separated list of C<termcap> properties to use for highlighting
252 strings on/off, pre-point part on/off.
253 See L<perldebug/"Debugging regular expressions"> for additional info.
255 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
256 lexically scoped, as the other directives are. However they have both
257 compile-time and run-time effects.
259 See L<perlmodlib/Pragmatic Modules>.
263 Similarly C<use re 'Debug'> produces debugging output, the difference
264 being that it allows the fine tuning of what debugging output will be
265 emitted. Options are divided into three groups, those related to
266 compilation, those related to execution and those related to special
267 purposes. The options are as follows:
271 =item Compile related options
277 Turns on all compile related debug options.
281 Turns on debug output related to the process of parsing the pattern.
285 Enables output related to the optimisation phase of compilation.
289 Detailed info about trie compilation.
293 Dump the final program out after it is compiled and optimised.
297 =item Execute related options
303 Turns on all execute related debug options.
307 Turns on debugging of the main matching loop.
311 Extra debugging of how tries execute.
315 Enable debugging of start point optimisations.
319 =item Extra debugging options
325 Turns on all "extra" debugging options.
329 Enable debugging the capture buffer storage during match. Warning,
330 this can potentially produce extremely large output.
334 Enable enhanced TRIE debugging. Enhances both TRIEE
339 Enable debugging of states in the engine.
343 Enable debugging of the recursion stack in the engine. Enabling
344 or disabling this option automatically does the same for debugging
345 states as well. This output from this can be quite large.
349 Enable enhanced optimisation debugging and start point optimisations.
350 Probably not useful except when debugging the regex engine itself.
354 Dump offset information. This can be used to see how regops correlate
355 to the pattern. Output format is
357 NODENUM:POSITION[LENGTH]
359 Where 1 is the position of the first char in the string. Note that position
360 can be 0, or larger than the actual length of the pattern, likewise length
365 Enable debugging of offsets information. This emits copious
366 amounts of trace information and doesn't mesh well with other
369 Almost definitely only useful to people hacking
370 on the offsets part of the debug engine.
374 =item Other useful flags
376 These are useful shortcuts to save on the typing.
382 Enable all options at once except OFFSETS, OFFSETSDBG and BUFFERS
386 Enable DUMP and all execute options. Equivalent to:
394 Enable TRIEM and all execute compile and execute options.
400 As of 5.9.5 the directive C<use re 'debug'> and its equivalents are
401 lexically scoped, as the other directives are. However they have both
402 compile-time and run-time effects.
404 =head2 Exportable Functions
406 As of perl 5.9.5 're' debug contains a number of utility functions that
407 may be optionally exported into the caller's namespace. They are listed
412 =item is_regexp($ref)
414 Returns true if the argument is a compiled regular expression as returned
415 by C<qr//>, false if it is not.
417 This function will not be confused by overloading or blessing. In
418 internals terms, this extracts the regexp pointer out of the
419 PERL_MAGIC_qr structure so it it cannot be fooled.
421 =item regexp_pattern($ref)
423 If the argument is a compiled regular expression as returned by C<qr//>,
424 then this function returns the pattern.
426 In list context it returns a two element list, the first element
427 containing the pattern and the second containing the modifiers used when
428 the pattern was compiled.
430 my ($pat, $mods) = regexp_pattern($ref);
432 In scalar context it returns the same as perl would when strigifying a raw
433 C<qr//> with the same pattern inside. If the argument is not a compiled
434 reference then this routine returns false but defined in scalar context,
435 and the empty list in list context. Thus the following
437 if (regexp_pattern($ref) eq '(?i-xsm:foo)')
439 will be warning free regardless of what $ref actually is.
441 Like C<is_regexp> this function will not be confused by overloading
442 or blessing of the object.
446 If the argument is a compiled regular expression as returned by C<qr//>,
447 then this function returns what the optimiser consiers to be the longest
448 anchored fixed string and longest floating fixed string in the pattern.
450 A I<fixed string> is defined as being a substring that must appear for the
451 pattern to match. An I<anchored fixed string> is a fixed string that must
452 appear at a particular offset from the beginning of the match. A I<floating
453 fixed string> is defined as a fixed string that can appear at any point in
454 a range of positions relative to the start of the match. For example,
456 my $qr = qr/here .* there/x;
457 my ($anchored, $floating) = regmust($qr);
458 print "anchored:'$anchored'\nfloating:'$floating'\n";
465 Because the C<here> is before the C<.*> in the pattern, its position
466 can be determined exactly. That's not true, however, for the C<there>;
467 it could appear at any point after where the anchored string appeared.
468 Perl uses both for its optimisations, prefering the longer, or, if they are
471 B<NOTE:> This may not necessarily be the definitive longest anchored and
472 floating string. This will be what the optimiser of the Perl that you
473 are using thinks is the longest. If you believe that the result is wrong
474 please report it via the L<perlbug> utility.
476 =item regname($name,$all)
478 Returns the contents of a named buffer of the last successful match. If
479 $all is true, then returns an array ref containing one entry per buffer,
480 otherwise returns the first defined buffer.
484 Returns a list of all of the named buffers defined in the last successful
485 match. If $all is true, then it returns all names defined, if not it returns
486 only names which were involved in the match.
488 =item regnames_iterinit()
490 Initializes the internal hash iterator associated to the last successful
491 matches named capture buffers.
493 =item regnames_iternext($all)
495 Gets the next key from the named capture buffer hash associated with the
496 last successful match. If $all is true returns the keys of all of the
497 distinct named buffers in the pattern, if not returns only those names
498 used in the last successful match.
500 =item regnames_count()
502 Returns the number of distinct names defined in the pattern used
503 for the last successful match.
505 B<Note:> 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>.