Restrict features in wildcards

The algorithm for dealing with Unicode property wildcards is to wrap the
user-supplied pattern with /miaa.  We don't want the user to be able to
override the /m and /aa parts.  Modifiers that are only specifiable as a
modifier in a qr or similar op (like /gc) can't be included in things
like (?gc).  These normally incur a warning that they are ignored, but
the texts of those warnings are misleading when using wildcards, so I
chose to just make them illegal.  Of course that could be changed to
having custom useful warning texts, but I didn't think it was worth it.

I also chose to forbid recursion of using nested \p{}, just from fear
that it might lead to issues down the road, and it really isn't useful
for this limited universe of strings to match against.  Because
wildcards currently can't handle '}' inside them, only the single letter
\p,\P are valid anyway.

Similarly, I forbid the '*' quantifier to make it harder for the
constructed subpattern to take forever to make any progress and decide
to halt.  Again, using it would be overkill on the universe of possible
match strings.

diff --git a/charclass_invlists.h b/charclass_invlists.h
index bf27eee..3c9becf 100644 (file)
--- a/charclass_invlists.h
+++ b/charclass_invlists.h
@@ -419846,7 +419846,7 @@ static const U8 WB_table[23][23] = {
  * baba9dfc133e3cb770a89aaf0973b1341fa61c2da6c176baf6428898b3b568d8 lib/unicore/extracted/DLineBreak.txt
  * 6d4a8c945dd7db83ed617cbb7d937de7f4ecf016ff22970d846e996a7c9a2a5d lib/unicore/extracted/DNumType.txt
  * 5b7c14380d5cceeaffcfbc18db1ed936391d2af2d51f5a41f1a17b692c77e59b lib/unicore/extracted/DNumValues.txt
- * 93f508a690aa8949f213d50b573710f0b4a4e843c17283938035ecf19e0220e2 lib/unicore/mktables
+ * a3f3caba903e4d39b6c7aaa7ea4d3a739e745b010ad51cf0e05f34ffa0ac2c04 lib/unicore/mktables
  * 50b85a67451145545a65cea370dab8d3444fbfe07e9c34cef560c5b7da9d3eef lib/unicore/version
  * 2680b9254eb236c5c090f11b149605043e8c8433661b96efc4a42fb4709342a5 regen/charset_translations.pl
  * 6bbad21de0848e0236b02f34f5fa0edd3cdae9ba8173cc9469a5513936b9e728 regen/mk_PL_charclass.pl

diff --git a/lib/unicore/mktables b/lib/unicore/mktables
index 1820ad3..498a94d 100644 (file)
--- a/lib/unicore/mktables
+++ b/lib/unicore/mktables
@@ -19251,6 +19251,8 @@ Error('\p{InKana}');    # 'Kana' is not a block so InKana shouldn't compile
 # Test_WB()
 Test_WB("$breakable 0020 $breakable 0020 $breakable 0308 $breakable");
 Test_LB("$nobreak 200B $nobreak 0020 $nobreak 0020 $breakable 2060 $breakable");
+Expect(1, ord(" "), '\p{gc=:(?aa)s:}', "");     # /aa is valid
+Expect(1, ord(" "), '\p{gc=:(?-s)s:}', "");     # /-s is valid
 EOF_CODE
 
     # Sort these so get results in same order on different runs of this

diff --git a/lib/unicore/uni_keywords.pl b/lib/unicore/uni_keywords.pl
index e76cf63..f8ee5e2 100644 (file)
--- a/lib/unicore/uni_keywords.pl
+++ b/lib/unicore/uni_keywords.pl
@@ -1295,7 +1295,7 @@
 # baba9dfc133e3cb770a89aaf0973b1341fa61c2da6c176baf6428898b3b568d8 lib/unicore/extracted/DLineBreak.txt
 # 6d4a8c945dd7db83ed617cbb7d937de7f4ecf016ff22970d846e996a7c9a2a5d lib/unicore/extracted/DNumType.txt
 # 5b7c14380d5cceeaffcfbc18db1ed936391d2af2d51f5a41f1a17b692c77e59b lib/unicore/extracted/DNumValues.txt
-# 93f508a690aa8949f213d50b573710f0b4a4e843c17283938035ecf19e0220e2 lib/unicore/mktables
+# a3f3caba903e4d39b6c7aaa7ea4d3a739e745b010ad51cf0e05f34ffa0ac2c04 lib/unicore/mktables
 # 50b85a67451145545a65cea370dab8d3444fbfe07e9c34cef560c5b7da9d3eef lib/unicore/version
 # 2680b9254eb236c5c090f11b149605043e8c8433661b96efc4a42fb4709342a5 regen/charset_translations.pl
 # 6bbad21de0848e0236b02f34f5fa0edd3cdae9ba8173cc9469a5513936b9e728 regen/mk_PL_charclass.pl

diff --git a/op.h b/op.h
index 4a544d9..cbb19d5 100644 (file)
--- a/op.h
+++ b/op.h
@@ -372,6 +372,10 @@ struct pmop {
 #define PMf_IS_QR      (1U<<(PMf_BASE_SHIFT+15))
 #define PMf_USE_RE_EVAL        (1U<<(PMf_BASE_SHIFT+16)) /* use re'eval' in scope */
 
+/* Means that this is a subpattern being compiled while processing a \p{}
+ * wildcard.  This isn't called from op.c, but it is passed as a pm flag. */
+#define PMf_WILDCARD    (1U<<(PMf_BASE_SHIFT+17))
+
 /* See comments at the beginning of these defines about adding bits.  The
  * highest bit position should be used, so that if PMf_BASE_SHIFT gets
  * increased, the #error below will be triggered so that you will be reminded

diff --git a/pod/perldelta.pod b/pod/perldelta.pod
index 06cab07..f76af50 100644 (file)
--- a/pod/perldelta.pod
+++ b/pod/perldelta.pod
@@ -72,6 +72,13 @@ XXX For a release on a stable branch, this section aspires to be:
 
 [ List each incompatible change as a =head2 entry ]
 
+=head2 Certain pattern matching features are now prohibited in compiling
+Unicode property value wildcard subpatterns
+
+These few features are either inappropriate or interfere with the
+algorithm used to accomplish this task.  The complete list is in
+L<perlunicode/Wildcards in Property Values>.
+
 =head2 Stop pretending C<POSIX::mbstowcs> and C<POSIX::wcstombs> are
 supported
 

diff --git a/pod/perldiag.pod b/pod/perldiag.pod
index fa51f1d..fdb2ba6 100644 (file)
--- a/pod/perldiag.pod
+++ b/pod/perldiag.pod
@@ -7331,6 +7331,15 @@ This was deprecated in Perl 5.004, and was made fatal in Perl 5.28.
 (F) You attempted to use a feature of printf that is accessible from
 only C.  This usually means there's a better way to do it in Perl.
 
+=item Use of %s is not allowed in Unicode property wildcard
+subpatterns in regex; marked by S<<-- HERE> in m/%s/
+
+(F) You were using a wildcard subpattern a Unicode property value, and
+the subpattern contained something that is illegal.  Not all regular
+expression capabilities are legal in such subpatterns, and this is one.
+Rewrite your subppattern to not use the offending construct.
+See L<perlunicode/Wildcards in Property Values>.
+
 =item Use of -l on filehandle%s
 
 (W io) A filehandle represents an opened file, and when you opened the file

diff --git a/pod/perlunicode.pod b/pod/perlunicode.pod
index 761b0db..212f3e9 100644 (file)
--- a/pod/perlunicode.pod
+++ b/pod/perlunicode.pod
@@ -1027,14 +1027,27 @@ C<$/>.
 No modifiers may follow the final delimiter.  Instead, use
 L<perlre/(?adlupimnsx-imnsx)> and/or
 L<perlre/(?adluimnsx-imnsx:pattern)> to specify modifiers.
+However, certain modifiers are illegal in your wildcard subpattern.
+The only character set modifier specifiable is C</aa>;
+any other character set, and C<-m>, and C<p>, and C<s> are all illegal.
+Specifying modifiers like C<qr/.../gc> that aren't legal in the
+C<(?...)> notation normally raise a warning, but with wildcard
+subpatterns, their use is an error.  The C<m> modifier is ineffective;
+everything that matches will be a single line.
+
+By default, your pattern is matched case-insensitively, as if C</i> had
+been specified.  You can change this by saying C<(?-i)> in your pattern.
+
+There are also certain operations that are illegal.  You can't nest
+C<\p{...}> and C<\P{...}> calls within a wildcard subpattern, and C<\G>
+doesn't make sense, so is also prohibited.
+
+And the C<*> quantifier (or its equivalent C<(0,}>) is illegal.
 
 This feature is not available when the left-hand side is prefixed by
 C<Is_>, nor for any form that is marked as "Discouraged" in
 L<perluniprops/Discouraged>.
 
-By default, your pattern is matched case-insensitively, as if C</i> had
-been specified.  You can change this by saying C<(?-i)> in your pattern.
-
 This experimental feature has been added to begin to implement
 L<https://www.unicode.org/reports/tr18/#Wildcard_Properties>.  Using it
 will raise a (default-on) warning in the

diff --git a/regcharclass.h b/regcharclass.h
index 69b85c7..35c08b8 100644 (file)
--- a/regcharclass.h
+++ b/regcharclass.h
@@ -2247,7 +2247,7 @@
  * baba9dfc133e3cb770a89aaf0973b1341fa61c2da6c176baf6428898b3b568d8 lib/unicore/extracted/DLineBreak.txt
  * 6d4a8c945dd7db83ed617cbb7d937de7f4ecf016ff22970d846e996a7c9a2a5d lib/unicore/extracted/DNumType.txt
  * 5b7c14380d5cceeaffcfbc18db1ed936391d2af2d51f5a41f1a17b692c77e59b lib/unicore/extracted/DNumValues.txt
- * 93f508a690aa8949f213d50b573710f0b4a4e843c17283938035ecf19e0220e2 lib/unicore/mktables
+ * a3f3caba903e4d39b6c7aaa7ea4d3a739e745b010ad51cf0e05f34ffa0ac2c04 lib/unicore/mktables
  * 50b85a67451145545a65cea370dab8d3444fbfe07e9c34cef560c5b7da9d3eef lib/unicore/version
  * 2680b9254eb236c5c090f11b149605043e8c8433661b96efc4a42fb4709342a5 regen/charset_translations.pl
  * f9a393e7add8c7c2728356473ce5b52246d51295b2da0c48fb6f0aa21799e2bb regen/regcharclass.pl

diff --git a/regcomp.c b/regcomp.c
index 59b346b..0a78e32 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -7399,7 +7399,7 @@ S_set_regex_pv(pTHX_ RExC_state_t *pRExC_state, REGEXP *Rx)
  *
  * pm_flags contains the PMf_* flags, typically based on those from the
  * pm_flags field of the related PMOP. Currently we're only interested in
- * PMf_HAS_CV, PMf_IS_QR, PMf_USE_RE_EVAL.
+ * PMf_HAS_CV, PMf_IS_QR, PMf_USE_RE_EVAL, PMf_WILDCARD.
  *
  * For many years this code had an initial sizing pass that calculated
  * (sometimes incorrectly, leading to security holes) the size needed for the
@@ -10765,6 +10765,24 @@ S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
         /* && memCHRs("iogcmsx", *RExC_parse) */
         /* (?g), (?gc) and (?o) are useless here
            and must be globally applied -- japhy */
+        if ((RExC_pm_flags & PMf_WILDCARD)) {
+            if (flagsp == & negflags) {
+                if (*RExC_parse == 'm') {
+                    RExC_parse++;
+                    /* diag_listed_as: Use of %s is not allowed in Unicode
+                       property wildcard subpatterns in regex; marked by <--
+                       HERE in m/%s/ */
+                    vFAIL("Use of modifier '-m' is not allowed in Unicode"
+                          " property wildcard subpatterns");
+                }
+            }
+            else {
+                if (*RExC_parse == 's') {
+                    goto modifier_illegal_in_wildcard;
+                }
+            }
+        }
+
         switch (*RExC_parse) {
 
             /* Code for the imsxn flags */
@@ -10845,6 +10863,10 @@ S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
                                     *(RExC_parse - 1));
                 NOT_REACHED; /*NOTREACHED*/
             case GLOBAL_PAT_MOD: /* 'g' */
+                if (RExC_pm_flags & PMf_WILDCARD) {
+                    goto modifier_illegal_in_wildcard;
+                }
+                /*FALLTHROUGH*/
             case ONCE_PAT_MOD: /* 'o' */
                 if (ckWARN(WARN_REGEXP)) {
                     const I32 wflagbit = *RExC_parse == 'o'
@@ -10866,6 +10888,9 @@ S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
                 break;
 
             case CONTINUE_PAT_MOD: /* 'c' */
+                if (RExC_pm_flags & PMf_WILDCARD) {
+                    goto modifier_illegal_in_wildcard;
+                }
                 if (ckWARN(WARN_REGEXP)) {
                     if (! (wastedflags & WASTED_C) ) {
                         wastedflags |= WASTED_GC;
@@ -10880,6 +10905,9 @@ S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
                 }
                 break;
             case KEEPCOPY_PAT_MOD: /* 'p' */
+                if (RExC_pm_flags & PMf_WILDCARD) {
+                    goto modifier_illegal_in_wildcard;
+                }
                 if (flagsp == &negflags) {
                     ckWARNreg(RExC_parse + 1,"Useless use of (?-p)");
                 } else {
@@ -10900,6 +10928,18 @@ S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
             case ':':
             case ')':
 
+                if (  (RExC_pm_flags & PMf_WILDCARD)
+                    && cs != REGEX_ASCII_MORE_RESTRICTED_CHARSET)
+                {
+                    RExC_parse++;
+                    /* diag_listed_as: Use of %s is not allowed in Unicode
+                       property wildcard subpatterns in regex; marked by <--
+                       HERE in m/%s/ */
+                    vFAIL2("Use of modifier '%c' is not allowed in Unicode"
+                           " property wildcard subpatterns",
+                           has_charset_modifier);
+                }
+
                 if ((posflags & (RXf_PMf_EXTENDED|RXf_PMf_EXTENDED_MORE)) == RXf_PMf_EXTENDED) {
                     negflags |= RXf_PMf_EXTENDED_MORE;
                 }
@@ -10925,6 +10965,13 @@ S_parse_lparen_question_flags(pTHX_ RExC_state_t *pRExC_state)
     }
 
     vFAIL("Sequence (?... not terminated");
+
+  modifier_illegal_in_wildcard:
+    RExC_parse++;
+    /* diag_listed_as: Use of %s is not allowed in Unicode property wildcard
+       subpatterns in regex; marked by <-- HERE in m/%s/ */
+    vFAIL2("Use of modifier '%c' is not allowed in Unicode property wildcard"
+           " subpatterns", *(RExC_parse - 1));
 }
 
 /*
@@ -12533,6 +12580,23 @@ S_regpiece(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
          do_curly:
            if ((flags&SIMPLE)) {
                 if (min == 0 && max == REG_INFTY) {
+
+                    /* Going from 0..inf is currently forbidden in wildcard
+                     * subpatterns.  The only reason is to make it harder to
+                     * write patterns that take a long long time to halt, and
+                     * because the use of this construct isn't necessary in
+                     * matching Unicode property values */
+                    if (RExC_pm_flags & PMf_WILDCARD) {
+                        RExC_parse++;
+                        /* diag_listed_as: Use of %s is not allowed in Unicode
+                           property wildcard subpatterns in regex; marked by
+                           <-- HERE in m/%s/ */
+                        vFAIL("Use of quantifier '*' is not allowed in"
+                              " Unicode property wildcard subpatterns");
+                        /* Note, don't need to worry about {0,}, as a '}' isn't
+                         * legal at all in wildcards, so wouldn't get this far
+                         * */
+                    }
                     reginsert(pRExC_state, STAR, ret, depth+1);
                     MARK_NAUGHTY(4);
                     RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
@@ -13404,13 +13468,26 @@ S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
        /* Special Escapes */
        case 'A':
            RExC_seen_zerolen++;
-           ret = reg_node(pRExC_state, SBOL);
-            /* SBOL is shared with /^/ so we set the flags so we can tell
-             * /\A/ from /^/ in split. */
-            FLAGS(REGNODE_p(ret)) = 1;
+            if (RExC_pm_flags & PMf_WILDCARD) {
+                ret = reg_node(pRExC_state, MBOL);
+            }
+            else {
+                ret = reg_node(pRExC_state, SBOL);
+                /* SBOL is shared with /^/ so we set the flags so we can tell
+                 * /\A/ from /^/ in split. */
+                FLAGS(REGNODE_p(ret)) = 1;
+            }
            *flagp |= SIMPLE;
            goto finish_meta_pat;
        case 'G':
+            if (RExC_pm_flags & PMf_WILDCARD) {
+                RExC_parse++;
+                /* diag_listed_as: Use of %s is not allowed in Unicode property
+                   wildcard subpatterns in regex; marked by <-- HERE in m/%s/
+                 */
+                vFAIL("Use of '\\G' is not allowed in Unicode property"
+                      " wildcard subpatterns");
+            }
            ret = reg_node(pRExC_state, GPOS);
             RExC_seen |= REG_GPOS_SEEN;
            *flagp |= SIMPLE;
@@ -13432,12 +13509,22 @@ S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
                 vFAIL("\\K not permitted in lookahead/lookbehind");
             }
        case 'Z':
-           ret = reg_node(pRExC_state, SEOL);
+            if (RExC_pm_flags & PMf_WILDCARD) {
+                ret = reg_node(pRExC_state, MEOL);
+            }
+            else {
+                ret = reg_node(pRExC_state, SEOL);
+            }
            *flagp |= SIMPLE;
            RExC_seen_zerolen++;                /* Do not optimize RE away */
            goto finish_meta_pat;
        case 'z':
-           ret = reg_node(pRExC_state, EOS);
+            if (RExC_pm_flags & PMf_WILDCARD) {
+                ret = reg_node(pRExC_state, MEOL);
+            }
+            else {
+                ret = reg_node(pRExC_state, EOS);
+            }
            *flagp |= SIMPLE;
            RExC_seen_zerolen++;                /* Do not optimize RE away */
            goto finish_meta_pat;
@@ -17457,6 +17544,15 @@ S_regclass(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth,
                {
                char *e;
 
+                if (RExC_pm_flags & PMf_WILDCARD) {
+                    RExC_parse++;
+                    /* diag_listed_as: Use of %s is not allowed in Unicode
+                       property wildcard subpatterns in regex; marked by <--
+                       HERE in m/%s/ */
+                    vFAIL3("Use of '\\%c%c' is not allowed in Unicode property"
+                           " wildcard subpatterns", value, *(RExC_parse - 1));
+                }
+
                /* \p means they want Unicode semantics */
                REQUIRE_UNI_RULES(flagp, 0);
 
@@ -22818,7 +22914,7 @@ STATIC REGEXP *
 S_compile_wildcard(pTHX_ const char * name, const STRLEN len,
                          const bool ignore_case)
 {
-    U32 flags = PMf_MULTILINE;
+    U32 flags = PMf_MULTILINE|PMf_WILDCARD;
     REGEXP * subpattern_re;
 
     PERL_ARGS_ASSERT_COMPILE_WILDCARD;

diff --git a/t/re/reg_mesg.t b/t/re/reg_mesg.t
index 9e8dc6d..6c26268 100644 (file)
--- a/t/re/reg_mesg.t
+++ b/t/re/reg_mesg.t
@@ -319,6 +319,18 @@ my @death =
  '/(?[\N{KEYCAP DIGIT NINE}/' => '\N{} here is restricted to one character {#} m/(?[\\N{U+39.FE0F.20E3{#}}/', # [perl #133988]
  '/0000000000000000[\N{U+0.00}0000/' => 'Unmatched [ {#} m/0000000000000000[{#}\N{U+0.00}0000/', # [perl #134059]
  '/\p{nv=\b5\b}/' => 'Can\'t find Unicode property definition "nv=\\b5\\b" {#} m/\\p{nv=\\b5\\b}{#}/',
+ '/\p{nv=:(?g)10:}/' => 'Use of modifier \'g\' is not allowed in Unicode property wildcard subpatterns {#} m/(?g{#})10/',
+ '/\p{gc=:L*:}/' => 'Use of quantifier \'*\' is not allowed in Unicode property wildcard subpatterns {#} m/L*{#}/',
+ '/\p{gc=:L\G:}/' => 'Use of \'\G\' is not allowed in Unicode property wildcard subpatterns {#} m/L\G{#}/',
+ '/\p{gc=:(?a)L:}/' => 'Use of modifier \'a\' is not allowed in Unicode property wildcard subpatterns {#} m/(?a){#}L/',
+ '/\p{gc=:(?u)L:}/' => 'Use of modifier \'u\' is not allowed in Unicode property wildcard subpatterns {#} m/(?u){#}L/',
+ '/\p{gc=:(?d)L:}/' => 'Use of modifier \'d\' is not allowed in Unicode property wildcard subpatterns {#} m/(?d){#}L/',
+ '/\p{gc=:(?l)L:}/' => 'Use of modifier \'l\' is not allowed in Unicode property wildcard subpatterns {#} m/(?l){#}L/',
+ '/\p{gc=:(?-m)L:}/' => 'Use of modifier \'-m\' is not allowed in Unicode property wildcard subpatterns {#} m/(?-m{#})L/',
+ '/\p{gc=:\pS:}/' => 'Use of \'\\pS\' is not allowed in Unicode property wildcard subpatterns {#} m/\\pS{#}/',
+ '/\p{gc=:\PS:}/' => 'Use of \'\\PS\' is not allowed in Unicode property wildcard subpatterns {#} m/\\PS{#}/',
+ '/\p{gc=:[\pS]:}/' => 'Use of \'\\pS\' is not allowed in Unicode property wildcard subpatterns {#} m/[\\pS{#}]/',
+ '/\p{gc=:[\PS]:}/' => 'Use of \'\\PS\' is not allowed in Unicode property wildcard subpatterns {#} m/[\\PS{#}]/',
 );
 
 # These are messages that are death under 'use re "strict"', and may or may
@@ -731,6 +743,7 @@ for my $strict ("", "use re 'strict';") {
         else {
             no warnings 'experimental::regex_sets';
             no warnings 'experimental::re_strict';
+            no warnings 'experimental::uniprop_wildcards';
 
             warning_is(sub {
                     my $meaning_of_life;

diff --git a/uni_keywords.h b/uni_keywords.h
index 9359e10..8dbf316 100644 (file)
--- a/uni_keywords.h
+++ b/uni_keywords.h
@@ -7540,7 +7540,7 @@ MPH_VALt match_uniprop( const unsigned char * const key, const U16 key_len ) {
  * baba9dfc133e3cb770a89aaf0973b1341fa61c2da6c176baf6428898b3b568d8 lib/unicore/extracted/DLineBreak.txt
  * 6d4a8c945dd7db83ed617cbb7d937de7f4ecf016ff22970d846e996a7c9a2a5d lib/unicore/extracted/DNumType.txt
  * 5b7c14380d5cceeaffcfbc18db1ed936391d2af2d51f5a41f1a17b692c77e59b lib/unicore/extracted/DNumValues.txt
- * 93f508a690aa8949f213d50b573710f0b4a4e843c17283938035ecf19e0220e2 lib/unicore/mktables
+ * a3f3caba903e4d39b6c7aaa7ea4d3a739e745b010ad51cf0e05f34ffa0ac2c04 lib/unicore/mktables
  * 50b85a67451145545a65cea370dab8d3444fbfe07e9c34cef560c5b7da9d3eef lib/unicore/version
  * 2680b9254eb236c5c090f11b149605043e8c8433661b96efc4a42fb4709342a5 regen/charset_translations.pl
  * 6bbad21de0848e0236b02f34f5fa0edd3cdae9ba8173cc9469a5513936b9e728 regen/mk_PL_charclass.pl