Commit | Line | Data |
b3eb6a9b |
1 | package re; |
2 | |
de8c5301 |
3 | # pragma for controlling the regex engine |
4 | use strict; |
5 | use warnings; |
6 | |
256ddcd0 |
7 | our $VERSION = "0.07"; |
de8c5301 |
8 | our @ISA = qw(Exporter); |
256ddcd0 |
9 | our @EXPORT_OK = qw(is_regexp regexp_pattern regmust); |
de8c5301 |
10 | our %EXPORT_OK = map { $_ => 1 } @EXPORT_OK; |
11 | |
12 | # *** WARNING *** WARNING *** WARNING *** WARNING *** WARNING *** |
13 | # |
14 | # If you modify these values see comment below! |
15 | |
16 | my %bitmask = ( |
17 | taint => 0x00100000, # HINT_RE_TAINT |
18 | eval => 0x00200000, # HINT_RE_EVAL |
19 | ); |
20 | |
21 | # - File::Basename contains a literal for 'taint' as a fallback. If |
22 | # taint is changed here, File::Basename must be updated as well. |
23 | # |
24 | # - ExtUtils::ParseXS uses a hardcoded |
25 | # BEGIN { $^H |= 0x00200000 } |
26 | # in it to allow re.xs to be built. So if 'eval' is changed here then |
27 | # ExtUtils::ParseXS must be changed as well. |
28 | # |
29 | # *** WARNING *** WARNING *** WARNING *** WARNING *** WARNING *** |
30 | |
31 | sub setcolor { |
32 | eval { # Ignore errors |
33 | require Term::Cap; |
34 | |
35 | my $terminal = Tgetent Term::Cap ({OSPEED => 9600}); # Avoid warning. |
36 | my $props = $ENV{PERL_RE_TC} || 'md,me,so,se,us,ue'; |
37 | my @props = split /,/, $props; |
38 | my $colors = join "\t", map {$terminal->Tputs($_,1)} @props; |
39 | |
40 | $colors =~ s/\0//g; |
41 | $ENV{PERL_RE_COLORS} = $colors; |
42 | }; |
43 | if ($@) { |
44 | $ENV{PERL_RE_COLORS} ||= qq'\t\t> <\t> <\t\t'; |
45 | } |
46 | |
47 | } |
48 | |
49 | my %flags = ( |
50 | COMPILE => 0x0000FF, |
51 | PARSE => 0x000001, |
52 | OPTIMISE => 0x000002, |
53 | TRIEC => 0x000004, |
54 | DUMP => 0x000008, |
55 | |
56 | EXECUTE => 0x00FF00, |
57 | INTUIT => 0x000100, |
58 | MATCH => 0x000200, |
59 | TRIEE => 0x000400, |
60 | |
61 | EXTRA => 0xFF0000, |
62 | TRIEM => 0x010000, |
63 | OFFSETS => 0x020000, |
64 | OFFSETSDBG => 0x040000, |
65 | STATE => 0x080000, |
66 | OPTIMISEM => 0x100000, |
67 | STACK => 0x280000, |
68 | ); |
69 | $flags{ALL} = -1; |
70 | $flags{All} = $flags{all} = $flags{DUMP} | $flags{EXECUTE}; |
71 | $flags{Extra} = $flags{EXECUTE} | $flags{COMPILE}; |
72 | $flags{More} = $flags{MORE} = $flags{All} | $flags{TRIEC} | $flags{TRIEM} | $flags{STATE}; |
73 | $flags{State} = $flags{DUMP} | $flags{EXECUTE} | $flags{STATE}; |
74 | $flags{TRIE} = $flags{DUMP} | $flags{EXECUTE} | $flags{TRIEC}; |
75 | |
76 | my $installed; |
77 | my $installed_error; |
78 | |
79 | sub _do_install { |
80 | if ( ! defined($installed) ) { |
81 | require XSLoader; |
82 | $installed = eval { XSLoader::load('re', $VERSION) } || 0; |
83 | $installed_error = $@; |
84 | } |
85 | } |
86 | |
87 | sub _load_unload { |
88 | my ($on)= @_; |
89 | if ($on) { |
90 | _do_install(); |
91 | if ( ! $installed ) { |
92 | die "'re' not installed!? ($installed_error)"; |
93 | } else { |
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. |
97 | |
98 | # install() returns an integer, which if casted properly |
99 | # in C resolves to a structure containing the regex |
100 | # hooks. Setting it to a random integer will guarantee |
101 | # segfaults. |
102 | $^H{regcomp} = install(); |
103 | } |
104 | } else { |
105 | delete $^H{regcomp}; |
106 | } |
107 | } |
108 | |
109 | sub bits { |
110 | my $on = shift; |
111 | my $bits = 0; |
112 | unless (@_) { |
113 | require Carp; |
114 | Carp::carp("Useless use of \"re\" pragma"); |
115 | } |
116 | foreach my $idx (0..$#_){ |
117 | my $s=$_[$idx]; |
118 | if ($s eq 'Debug' or $s eq 'Debugcolor') { |
119 | setcolor() if $s =~/color/i; |
120 | ${^RE_DEBUG_FLAGS} = 0 unless defined ${^RE_DEBUG_FLAGS}; |
121 | for my $idx ($idx+1..$#_) { |
122 | if ($flags{$_[$idx]}) { |
123 | if ($on) { |
124 | ${^RE_DEBUG_FLAGS} |= $flags{$_[$idx]}; |
125 | } else { |
126 | ${^RE_DEBUG_FLAGS} &= ~ $flags{$_[$idx]}; |
127 | } |
128 | } else { |
129 | require Carp; |
130 | Carp::carp("Unknown \"re\" Debug flag '$_[$idx]', possible flags: ", |
131 | join(", ",sort keys %flags ) ); |
132 | } |
133 | } |
134 | _load_unload($on ? 1 : ${^RE_DEBUG_FLAGS}); |
135 | last; |
136 | } elsif ($s eq 'debug' or $s eq 'debugcolor') { |
137 | setcolor() if $s =~/color/i; |
138 | _load_unload($on); |
139 | } elsif (exists $bitmask{$s}) { |
140 | $bits |= $bitmask{$s}; |
141 | } elsif ($EXPORT_OK{$s}) { |
142 | _do_install(); |
143 | require Exporter; |
144 | re->export_to_level(2, 're', $s); |
145 | } else { |
146 | require Carp; |
147 | Carp::carp("Unknown \"re\" subpragma '$s' (known ones are: ", |
148 | join(', ', map {qq('$_')} 'debug', 'debugcolor', sort keys %bitmask), |
149 | ")"); |
150 | } |
151 | } |
152 | $bits; |
153 | } |
154 | |
155 | sub import { |
156 | shift; |
157 | $^H |= bits(1, @_); |
158 | } |
159 | |
160 | sub unimport { |
161 | shift; |
162 | $^H &= ~ bits(0, @_); |
163 | } |
164 | |
165 | 1; |
166 | |
167 | __END__ |
56953603 |
168 | |
b3eb6a9b |
169 | =head1 NAME |
170 | |
171 | re - Perl pragma to alter regular expression behaviour |
172 | |
173 | =head1 SYNOPSIS |
174 | |
e4d48cc9 |
175 | use re 'taint'; |
176 | ($x) = ($^X =~ /^(.*)$/s); # $x is tainted here |
b3eb6a9b |
177 | |
2cd61cdb |
178 | $pat = '(?{ $foo = 1 })'; |
e4d48cc9 |
179 | use re 'eval'; |
2cd61cdb |
180 | /foo${pat}bar/; # won't fail (when not under -T switch) |
e4d48cc9 |
181 | |
182 | { |
183 | no re 'taint'; # the default |
184 | ($x) = ($^X =~ /^(.*)$/s); # $x is not tainted here |
185 | |
186 | no re 'eval'; # the default |
2cd61cdb |
187 | /foo${pat}bar/; # disallowed (with or without -T switch) |
e4d48cc9 |
188 | } |
b3eb6a9b |
189 | |
1e2e3d02 |
190 | use re 'debug'; # output debugging info during |
191 | /^(.*)$/s; # compile and run time |
192 | |
2cd61cdb |
193 | |
02ea72ae |
194 | use re 'debugcolor'; # same as 'debug', but with colored output |
195 | ... |
196 | |
a3621e74 |
197 | use re qw(Debug All); # Finer tuned debugging options. |
4ee9a43f |
198 | use re qw(Debug More); |
fe759410 |
199 | no re qw(Debug ALL); # Turn of all re debugging in this scope |
4ee9a43f |
200 | |
de8c5301 |
201 | use re qw(is_regexp regexp_pattern); # import utility functions |
202 | my ($pat,$mods)=regexp_pattern(qr/foo/i); |
203 | if (is_regexp($obj)) { |
204 | print "Got regexp: ", |
205 | scalar regexp_pattern($obj); # just as perl would stringify it |
206 | } # but no hassle with blessed re's. |
a3621e74 |
207 | |
3ffabb8c |
208 | (We use $^X in these examples because it's tainted by default.) |
209 | |
b3eb6a9b |
210 | =head1 DESCRIPTION |
211 | |
de8c5301 |
212 | =head2 'taint' mode |
213 | |
b3eb6a9b |
214 | When C<use re 'taint'> is in effect, and a tainted string is the target |
215 | of a regex, the regex memories (or values returned by the m// operator |
e4d48cc9 |
216 | in list context) are tainted. This feature is useful when regex operations |
217 | on tainted data aren't meant to extract safe substrings, but to perform |
218 | other transformations. |
b3eb6a9b |
219 | |
de8c5301 |
220 | =head2 'eval' mode |
221 | |
e4d48cc9 |
222 | When C<use re 'eval'> is in effect, a regex is allowed to contain |
2cd61cdb |
223 | C<(?{ ... })> zero-width assertions even if regular expression contains |
ffbc6a93 |
224 | variable interpolation. That is normally disallowed, since it is a |
2cd61cdb |
225 | potential security risk. Note that this pragma is ignored when the regular |
226 | expression is obtained from tainted data, i.e. evaluation is always |
3c4b39be |
227 | disallowed with tainted regular expressions. See L<perlre/(?{ code })>. |
2cd61cdb |
228 | |
ffbc6a93 |
229 | For the purpose of this pragma, interpolation of precompiled regular |
0a92e3a8 |
230 | expressions (i.e., the result of C<qr//>) is I<not> considered variable |
231 | interpolation. Thus: |
2cd61cdb |
232 | |
233 | /foo${pat}bar/ |
234 | |
ffbc6a93 |
235 | I<is> allowed if $pat is a precompiled regular expression, even |
2cd61cdb |
236 | if $pat contains C<(?{ ... })> assertions. |
237 | |
de8c5301 |
238 | =head2 'debug' mode |
239 | |
ffbc6a93 |
240 | When C<use re 'debug'> is in effect, perl emits debugging messages when |
2cd61cdb |
241 | compiling and using regular expressions. The output is the same as that |
242 | obtained by running a C<-DDEBUGGING>-enabled perl interpreter with the |
243 | B<-Dr> switch. It may be quite voluminous depending on the complexity |
02ea72ae |
244 | of the match. Using C<debugcolor> instead of C<debug> enables a |
245 | form of output that can be used to get a colorful display on terminals |
246 | that understand termcap color sequences. Set C<$ENV{PERL_RE_TC}> to a |
247 | comma-separated list of C<termcap> properties to use for highlighting |
ffbc6a93 |
248 | strings on/off, pre-point part on/off. |
2cd61cdb |
249 | See L<perldebug/"Debugging regular expressions"> for additional info. |
250 | |
de8c5301 |
251 | As of 5.9.5 the directive C<use re 'debug'> and its equivalents are |
252 | lexically scoped, as the other directives are. However they have both |
253 | compile-time and run-time effects. |
254 | |
255 | See L<perlmodlib/Pragmatic Modules>. |
256 | |
257 | =head2 'Debug' mode |
258 | |
a3621e74 |
259 | Similarly C<use re 'Debug'> produces debugging output, the difference |
260 | being that it allows the fine tuning of what debugging output will be |
be8e71aa |
261 | emitted. Options are divided into three groups, those related to |
262 | compilation, those related to execution and those related to special |
263 | purposes. The options are as follows: |
264 | |
265 | =over 4 |
266 | |
267 | =item Compile related options |
268 | |
269 | =over 4 |
270 | |
271 | =item COMPILE |
272 | |
273 | Turns on all compile related debug options. |
274 | |
275 | =item PARSE |
276 | |
277 | Turns on debug output related to the process of parsing the pattern. |
278 | |
279 | =item OPTIMISE |
280 | |
281 | Enables output related to the optimisation phase of compilation. |
282 | |
24b23f37 |
283 | =item TRIEC |
be8e71aa |
284 | |
285 | Detailed info about trie compilation. |
286 | |
287 | =item DUMP |
288 | |
289 | Dump the final program out after it is compiled and optimised. |
290 | |
be8e71aa |
291 | =back |
292 | |
293 | =item Execute related options |
294 | |
295 | =over 4 |
296 | |
297 | =item EXECUTE |
298 | |
299 | Turns on all execute related debug options. |
300 | |
301 | =item MATCH |
302 | |
303 | Turns on debugging of the main matching loop. |
304 | |
24b23f37 |
305 | =item TRIEE |
be8e71aa |
306 | |
307 | Extra debugging of how tries execute. |
308 | |
309 | =item INTUIT |
310 | |
311 | Enable debugging of start point optimisations. |
312 | |
313 | =back |
314 | |
315 | =item Extra debugging options |
316 | |
317 | =over 4 |
318 | |
319 | =item EXTRA |
320 | |
321 | Turns on all "extra" debugging options. |
322 | |
24b23f37 |
323 | =item TRIEM |
324 | |
325 | Enable enhanced TRIE debugging. Enhances both TRIEE |
326 | and TRIEC. |
327 | |
328 | =item STATE |
329 | |
4ee9a43f |
330 | Enable debugging of states in the engine. |
24b23f37 |
331 | |
332 | =item STACK |
be8e71aa |
333 | |
24b23f37 |
334 | Enable debugging of the recursion stack in the engine. Enabling |
335 | or disabling this option automatically does the same for debugging |
336 | states as well. This output from this can be quite large. |
337 | |
338 | =item OPTIMISEM |
339 | |
340 | Enable enhanced optimisation debugging and start point optimisations. |
341 | Probably not useful except when debugging the regex engine itself. |
342 | |
343 | =item OFFSETS |
344 | |
345 | Dump offset information. This can be used to see how regops correlate |
346 | to the pattern. Output format is |
347 | |
348 | NODENUM:POSITION[LENGTH] |
349 | |
350 | Where 1 is the position of the first char in the string. Note that position |
351 | can be 0, or larger than the actual length of the pattern, likewise length |
352 | can be zero. |
be8e71aa |
353 | |
24b23f37 |
354 | =item OFFSETSDBG |
be8e71aa |
355 | |
356 | Enable debugging of offsets information. This emits copious |
fe759410 |
357 | amounts of trace information and doesn't mesh well with other |
be8e71aa |
358 | debug options. |
359 | |
fe759410 |
360 | Almost definitely only useful to people hacking |
be8e71aa |
361 | on the offsets part of the debug engine. |
362 | |
363 | =back |
364 | |
365 | =item Other useful flags |
366 | |
367 | These are useful shortcuts to save on the typing. |
368 | |
369 | =over 4 |
370 | |
371 | =item ALL |
372 | |
373 | Enable all compile and execute options at once. |
374 | |
375 | =item All |
376 | |
fe759410 |
377 | Enable DUMP and all execute options. Equivalent to: |
be8e71aa |
378 | |
379 | use re 'debug'; |
380 | |
381 | =item MORE |
382 | |
383 | =item More |
384 | |
24b23f37 |
385 | Enable TRIEM and all execute compile and execute options. |
be8e71aa |
386 | |
dba3f186 |
387 | =back |
be8e71aa |
388 | |
dba3f186 |
389 | =back |
a3621e74 |
390 | |
1e2e3d02 |
391 | As of 5.9.5 the directive C<use re 'debug'> and its equivalents are |
4ee9a43f |
392 | lexically scoped, as the other directives are. However they have both |
1e2e3d02 |
393 | compile-time and run-time effects. |
b3eb6a9b |
394 | |
de8c5301 |
395 | =head2 Exportable Functions |
b3eb6a9b |
396 | |
de8c5301 |
397 | As of perl 5.9.5 're' debug contains a number of utility functions that |
4ee9a43f |
398 | may be optionally exported into the caller's namespace. They are listed |
de8c5301 |
399 | below. |
b3eb6a9b |
400 | |
de8c5301 |
401 | =over 4 |
b3eb6a9b |
402 | |
de8c5301 |
403 | =item is_regexp($ref) |
02ea72ae |
404 | |
de8c5301 |
405 | Returns true if the argument is a compiled regular expression as returned |
4ee9a43f |
406 | by C<qr//>, false if it is not. |
02ea72ae |
407 | |
4ee9a43f |
408 | This function will not be confused by overloading or blessing. In |
409 | internals terms, this extracts the regexp pointer out of the |
de8c5301 |
410 | PERL_MAGIC_qr structure so it it cannot be fooled. |
894be9b7 |
411 | |
de8c5301 |
412 | =item regexp_pattern($ref) |
02ea72ae |
413 | |
4ee9a43f |
414 | If the argument is a compiled regular expression as returned by C<qr//>, |
415 | then this function returns the pattern. |
be8e71aa |
416 | |
4ee9a43f |
417 | In list context it returns a two element list, the first element |
418 | containing the pattern and the second containing the modifiers used when |
419 | the pattern was compiled. |
be8e71aa |
420 | |
4ee9a43f |
421 | my ($pat, $mods) = regexp_pattern($ref); |
a3621e74 |
422 | |
4ee9a43f |
423 | In scalar context it returns the same as perl would when strigifying a raw |
424 | C<qr//> with the same pattern inside. If the argument is not a compiled |
425 | reference then this routine returns false but defined in scalar context, |
426 | and the empty list in list context. Thus the following |
f9f4320a |
427 | |
de8c5301 |
428 | if (regexp_pattern($ref) eq '(?i-xsm:foo)') |
dba3f186 |
429 | |
de8c5301 |
430 | will be warning free regardless of what $ref actually is. |
380e0b81 |
431 | |
4ee9a43f |
432 | Like C<is_regexp> this function will not be confused by overloading |
433 | or blessing of the object. |
b3eb6a9b |
434 | |
256ddcd0 |
435 | =item regmust($ref) |
436 | |
437 | If the argument is a compiled regular expression as returned by C<qr//> |
438 | then this function returns what the optimiser consiers to be the longest |
439 | anchored fixed string and longest floating fixed string in the pattern. |
440 | |
441 | A fixed string is defined as being a string that must appear in the string |
442 | for the pattern to match. An anchored fixed string is a fixed string that |
443 | must appear at a particular offset from the beginning of the match. A |
444 | floating fixed string is defined as a fixed string that can appear at |
445 | any point in a range of positions relative to the start of the match. |
446 | |
447 | my $qr=qr/here .* there/x; |
448 | my ($anchored,$floating)=regmust($qr); |
449 | print "anchored:'$anchored'\nfloating:'$floating'\n"; |
450 | |
451 | results in |
452 | |
453 | anchored:'here' |
454 | floating:'there' |
455 | |
456 | Because the C<here> is before the C<.*> in the pattern its position |
457 | can be determined exactly. The C<there> however is the opposite. |
458 | It could appear at any point after where the anchored string could appear. |
459 | Perl uses both for its optimisations, prefering the longer, or, if they are |
460 | equal, the floating. |
461 | |
462 | B<NOTE:> This may not necessarily be the definitive longest anchored and |
463 | floating string. This will be what the optimiser of the Perl that you |
464 | are using thinks is the longest. If you believe that the result is wrong |
465 | please report it via the L<perlbug> utility. |
466 | |
de8c5301 |
467 | =back |
b3eb6a9b |
468 | |
de8c5301 |
469 | =head1 SEE ALSO |
b3eb6a9b |
470 | |
de8c5301 |
471 | L<perlmodlib/Pragmatic Modules>. |
472 | |
473 | =cut |