fix #131649 - extended charclass can trigger assert

The extended charclass parser makes some assumptions during the
first pass which are only true on well structured input, and it
does not properly catch various errors. later on the code assumes
that things the first pass will let through are valid, when in
fact they should trigger errors.

(cherry picked from commit 19a498a461d7c81ae3507c450953d1148efecf4f)

6 files changed, 68 insertions, 32 deletions

diff --git a/pod/perldiag.pod b/pod/perldiag.pod
index 106fe41121..c29925a2a4 100644
--- a/pod/perldiag.pod
+++ b/pod/perldiag.pod
@@ -5904,7 +5904,7 @@ yourself.
 a perl4 interpreter, especially if the next 2 tokens are "use strict"
 or "my $var" or "our $var".
 
-=item Syntax error in (?[...]) in regex m/%s/
+=item Syntax error in (?[...]) in regex; marked by <-- HERE in m/%s/
 
 (F) Perl could not figure out what you meant inside this construct; this
 notifies you that it is giving up trying.
@@ -6402,6 +6402,31 @@ to find out why that isn't happening.
 (F) The unexec() routine failed for some reason.  See your local FSF
 representative, who probably put it there in the first place.
 
+=item Unexpected ']' with no following ')' in (?[... in regex; marked by <-- HERE in m/%s/
+
+(F) While parsing an extended character class a ']' character was encountered
+at a point in the definition where the only legal use of ']' is to close the
+character class definition as part of a '])', you may have forgotten the close
+paren, or otherwise confused the parser.
+
+=item Expecting close paren for nested extended charclass in regex; marked by <-- HERE in m/%s/
+
+(F) While parsing a nested extended character class like:
+
+    (?[ ... (?flags:(?[ ... ])) ... ])
+                             ^
+
+we expected to see a close paren ')' (marked by ^) but did not.
+
+=item Expecting close paren for wrapper for nested extended charclass in regex; marked by <-- HERE in m/%s/
+
+(F) While parsing a nested extended character class like:
+
+    (?[ ... (?flags:(?[ ... ])) ... ])
+                              ^
+
+we expected to see a close paren ')' (marked by ^) but did not.
+
 =item Unexpected binary operator '%c' with no preceding operand in regex;
 marked by S<<-- HERE> in m/%s/
 
diff --git a/pod/perlrecharclass.pod b/pod/perlrecharclass.pod
index 79480e4131..8c008507d1 100644
--- a/pod/perlrecharclass.pod
+++ b/pod/perlrecharclass.pod
@@ -1128,8 +1128,8 @@ hence both of the following work:
 Any contained POSIX character classes, including things like C<\w> and C<\D>
 respect the C<E<sol>a> (and C<E<sol>aa>) modifiers.
 
-C<< (?[ ]) >> is a regex-compile-time construct.  Any attempt to use
-something which isn't knowable at the time the containing regular
+Note that C<< (?[ ]) >> is a regex-compile-time construct.  Any attempt
+to use something which isn't knowable at the time the containing regular
 expression is compiled is a fatal error.  In practice, this means
 just three limitations:
 
diff --git a/regcomp.c b/regcomp.c
index 4ee48ede42..ddac290d2b 100644
--- a/regcomp.c
+++ b/regcomp.c
@@ -14840,8 +14840,9 @@ S_handle_regex_sets(pTHX_ RExC_state_t *pRExC_state, SV** return_invlist,
                                     TRUE /* Force /x */ );
 
             switch (*RExC_parse) {
-                case '?':
-                    if (RExC_parse[1] == '[') depth++, RExC_parse++;
+                case '(':
+                    if (RExC_parse[1] == '?' && RExC_parse[2] == '[')
+                        depth++, RExC_parse+=2;
                     /* FALLTHROUGH */
                 default:
                     break;
@@ -14898,9 +14899,9 @@ S_handle_regex_sets(pTHX_ RExC_state_t *pRExC_state, SV** return_invlist,
                 }
 
                 case ']':
-                    if (depth--) break;
-                    RExC_parse++;
-                    if (*RExC_parse == ')') {
+                    if (RExC_parse[1] == ')') {
+                        RExC_parse++;
+                        if (depth--) break;
                         node = reganode(pRExC_state, ANYOF, 0);
                         RExC_size += ANYOF_SKIP;
                         nextchar(pRExC_state);
@@ -14912,20 +14913,25 @@ S_handle_regex_sets(pTHX_ RExC_state_t *pRExC_state, SV** return_invlist,
 
                         return node;
                     }
-                    goto no_close;
+                    /* We output the messages even if warnings are off, because we'll fail
+                     * the very next thing, and these give a likely diagnosis for that */
+                    if (posix_warnings && av_tindex_skip_len_mg(posix_warnings) >= 0) {
+                        output_or_return_posix_warnings(pRExC_state, posix_warnings, NULL);
+                    }
+                    RExC_parse++;
+                    vFAIL("Unexpected ']' with no following ')' in (?[...");
             }
 
             RExC_parse += UTF ? UTF8SKIP(RExC_parse) : 1;
         }
 
-      no_close:
         /* We output the messages even if warnings are off, because we'll fail
          * the very next thing, and these give a likely diagnosis for that */
         if (posix_warnings && av_tindex_skip_len_mg(posix_warnings) >= 0) {
             output_or_return_posix_warnings(pRExC_state, posix_warnings, NULL);
         }
 
-        FAIL("Syntax error in (?[...])");
+        vFAIL("Syntax error in (?[...])");
     }
 
     /* Pass 2 only after this. */
@@ -15105,12 +15111,14 @@ redo_curchar:
                      * inversion list, and RExC_parse points to the trailing
                      * ']'; the next character should be the ')' */
                     RExC_parse++;
-                    assert(UCHARAT(RExC_parse) == ')');
+                    if (UCHARAT(RExC_parse) != ')')
+                        vFAIL("Expecting close paren for nested extended charclass");
 
                     /* Then the ')' matching the original '(' handled by this
                      * case: statement */
                     RExC_parse++;
-                    assert(UCHARAT(RExC_parse) == ')');
+                    if (UCHARAT(RExC_parse) != ')')
+                        vFAIL("Expecting close paren for wrapper for nested extended charclass");
 
                     RExC_flags = save_flags;
                     goto handle_operand;
diff --git a/t/lib/warnings/regcomp b/t/lib/warnings/regcomp
index 2b084c59b0..51ad57ccbe 100644
--- a/t/lib/warnings/regcomp
+++ b/t/lib/warnings/regcomp
@@ -59,21 +59,21 @@ Unmatched [ in regex; marked by <-- HERE in m/abc[ <-- HERE fi[.00./ at - line
 qr/(?[[[:word]]])/;
 EXPECT
 Assuming NOT a POSIX class since there is no terminating ':' in regex; marked by <-- HERE in m/(?[[[:word <-- HERE ]]])/ at - line 2.
-syntax error in (?[...]) in regex m/(?[[[:word]]])/ at - line 2.
+Unexpected ']' with no following ')' in (?[... in regex; marked by <-- HERE in m/(?[[[:word]] <-- HERE ])/ at - line 2.
 ########
 # NAME qr/(?[ [[:digit: ])/
 # OPTION fatal
 qr/(?[[[:digit: ])/;
 EXPECT
 Assuming NOT a POSIX class since no blanks are allowed in one in regex; marked by <-- HERE in m/(?[[[:digit: ] <-- HERE )/ at - line 2.
-syntax error in (?[...]) in regex m/(?[[[:digit: ])/ at - line 2.
+syntax error in (?[...]) in regex; marked by <-- HERE in m/(?[[[:digit: ]) <-- HERE / at - line 2.
 ########
 # NAME qr/(?[ [:digit: ])/
 # OPTION fatal
 qr/(?[[:digit: ])/
 EXPECT
 Assuming NOT a POSIX class since no blanks are allowed in one in regex; marked by <-- HERE in m/(?[[:digit: ] <-- HERE )/ at - line 2.
-syntax error in (?[...]) in regex m/(?[[:digit: ])/ at - line 2.
+syntax error in (?[...]) in regex; marked by <-- HERE in m/(?[[:digit: ]) <-- HERE / at - line 2.
 ########
 # NAME [perl #126141]
 # OPTION fatal
diff --git a/t/re/reg_mesg.t b/t/re/reg_mesg.t
index d26a7caf37..5194d93751 100644
--- a/t/re/reg_mesg.t
+++ b/t/re/reg_mesg.t
@@ -215,8 +215,9 @@ my @death =
  '/\b{gc}/' => "'gc' is an unknown bound type {#} m/\\b{gc{#}}/",
  '/\B{gc}/' => "'gc' is an unknown bound type {#} m/\\B{gc{#}}/",
 
- '/(?[[[::]]])/' => "Syntax error in (?[...]) in regex m/(?[[[::]]])/",
- '/(?[[[:w:]]])/' => "Syntax error in (?[...]) in regex m/(?[[[:w:]]])/",
+
+ '/(?[[[::]]])/' => "Unexpected ']' with no following ')' in (?[... {#} m/(?[[[::]]{#}])/",
+ '/(?[[[:w:]]])/' => "Unexpected ']' with no following ')' in (?[... {#} m/(?[[[:w:]]{#}])/",
  '/(?[[:w:]])/' => "",
  '/([.].*)[.]/'   => "",    # [perl #127582]
  '/[.].*[.]/'     => "",    # [perl #127604]
@@ -239,11 +240,12 @@ my @death =
  '/(?[ \p{foo} ])/' => 'Can\'t find Unicode property definition "foo" {#} m/(?[ \p{foo}{#} ])/',
  '/(?[ \p{ foo = bar } ])/' => 'Can\'t find Unicode property definition "foo = bar" {#} m/(?[ \p{ foo = bar }{#} ])/',
  '/(?[ \8 ])/' => 'Unrecognized escape \8 in character class {#} m/(?[ \8{#} ])/',
- '/(?[ \t ]/' => 'Syntax error in (?[...]) in regex m/(?[ \t ]/',
- '/(?[ [ \t ]/' => 'Syntax error in (?[...]) in regex m/(?[ [ \t ]/',
- '/(?[ \t ] ]/' => 'Syntax error in (?[...]) in regex m/(?[ \t ] ]/',
- '/(?[ [ ] ]/' => 'Syntax error in (?[...]) in regex m/(?[ [ ] ]/',
- '/(?[ \t + \e # This was supposed to be a comment ])/' => 'Syntax error in (?[...]) in regex m/(?[ \t + \e # This was supposed to be a comment ])/',
+ '/(?[ \t ]/' => "Unexpected ']' with no following ')' in (?[... {#} m/(?[ \\t ]{#}/",
+ '/(?[ [ \t ]/' => "Syntax error in (?[...]) {#} m/(?[ [ \\t ]{#}/",
+ '/(?[ \t ] ]/' => "Unexpected ']' with no following ')' in (?[... {#} m/(?[ \\t ]{#} ]/",
+ '/(?[ [ ] ]/' => "Syntax error in (?[...]) {#} m/(?[ [ ] ]{#}/",
+ '/(?[ \t + \e # This was supposed to be a comment ])/' =>
+    "Syntax error in (?[...]) {#} m/(?[ \\t + \\e # This was supposed to be a comment ]){#}/",
  '/(?[ ])/' => 'Incomplete expression within \'(?[ ])\' {#} m/(?[ {#}])/',
  'm/(?[[a-\d]])/' => 'False [] range "a-\d" {#} m/(?[[a-\d{#}]])/',
  'm/(?[[\w-x]])/' => 'False [] range "\w-" {#} m/(?[[\w-{#}x]])/',
@@ -431,10 +433,10 @@ my @death_utf8 = mark_as_utf8(
 
  '/ネ\p{}ネ/' => 'Empty \p{} {#} m/ネ\p{{#}}ネ/',
 
- '/ネ(?[[[:ネ]]])ネ/' => "Syntax error in (?[...]) in regex m/ネ(?[[[:ネ]]])ネ/",
- '/ネ(?[[[:ネ: ])ネ/' => "Syntax error in (?[...]) in regex m/ネ(?[[[:ネ: ])ネ/",
- '/ネ(?[[[::]]])ネ/' => "Syntax error in (?[...]) in regex m/ネ(?[[[::]]])ネ/",
- '/ネ(?[[[:ネ:]]])ネ/' => "Syntax error in (?[...]) in regex m/ネ(?[[[:ネ:]]])ネ/",
+ '/ネ(?[[[:ネ]]])ネ/' => "Unexpected ']' with no following ')' in (?[... {#} m/ネ(?[[[:ネ]]{#}])ネ/",
+ '/ネ(?[[[:ネ: ])ネ/' => "Syntax error in (?[...]) {#} m/ネ(?[[[:ネ: ])ネ{#}/",
+ '/ネ(?[[[::]]])ネ/' => "Unexpected ']' with no following ')' in (?[... {#} m/ネ(?[[[::]]{#}])ネ/",
+ '/ネ(?[[[:ネ:]]])ネ/' => "Unexpected ']' with no following ')' in (?[... {#} m/ネ(?[[[:ネ:]]{#}])ネ/",
  '/ネ(?[[:ネ:]])ネ/' => "",
  '/ネ(?[ネ])ネ/' =>  'Unexpected character {#} m/ネ(?[ネ{#}])ネ/',
  '/ネ(?[ + [ネ] ])/' => 'Unexpected binary operator \'+\' with no preceding operand {#} m/ネ(?[ +{#} [ネ] ])/',
@@ -447,8 +449,9 @@ my @death_utf8 = mark_as_utf8(
  '/(?[ \x{ネ} ])ネ/' => 'Non-hex character {#} m/(?[ \x{ネ{#}} ])ネ/',
  '/(?[ \p{ネ} ])/' => 'Can\'t find Unicode property definition "ネ" {#} m/(?[ \p{ネ}{#} ])/',
  '/(?[ \p{ ネ = bar } ])/' => 'Can\'t find Unicode property definition "ネ = bar" {#} m/(?[ \p{ ネ = bar }{#} ])/',
- '/ネ(?[ \t ]/' => 'Syntax error in (?[...]) in regex m/ネ(?[ \t ]/',
- '/(?[ \t + \e # ネ This was supposed to be a comment ])/' => 'Syntax error in (?[...]) in regex m/(?[ \t + \e # ネ This was supposed to be a comment ])/',
+ '/ネ(?[ \t ]/' => "Unexpected ']' with no following ')' in (?[... {#} m/ネ(?[ \\t ]{#}/",
+ '/(?[ \t + \e # ネ This was supposed to be a comment ])/' =>
+    "Syntax error in (?[...]) {#} m/(?[ \\t + \\e # ネ This was supposed to be a comment ]){#}/",
  'm/(*ネ)ネ/' => q<Unknown verb pattern 'ネ' {#} m/(*ネ){#}ネ/>,
  '/\cネ/' => "Character following \"\\c\" must be printable ASCII",
  '/\b{ネ}/' => "'ネ' is an unknown bound type {#} m/\\b{ネ{#}}/",
diff --git a/t/re/regex_sets.t b/t/re/regex_sets.t
index 6a79f9d692..e9644bd4e6 100644
--- a/t/re/regex_sets.t
+++ b/t/re/regex_sets.t
@@ -158,13 +158,13 @@ for my $char ("٠", "٥", "٩") {
 	eval { $_ = '/(?[(\c]) /'; qr/$_/ };
 	like($@, qr/^Syntax error/, '/(?[(\c]) / should not panic');
 	eval { $_ = '(?[\c#]' . "\n])"; qr/$_/ };
-	like($@, qr/^Syntax error/, '/(?[(\c]) / should not panic');
+	like($@, qr/^Unexpected/, '/(?[(\c]) / should not panic');
 	eval { $_ = '(?[(\c])'; qr/$_/ };
 	like($@, qr/^Syntax error/, '/(?[(\c])/ should be a syntax error');
 	eval { $_ = '(?[(\c]) ]\b'; qr/$_/ };
-	like($@, qr/^Syntax error/, '/(?[(\c]) ]\b/ should be a syntax error');
+	like($@, qr/^Unexpected/, '/(?[(\c]) ]\b/ should be a syntax error');
 	eval { $_ = '(?[\c[]](])'; qr/$_/ };
-	like($@, qr/^Syntax error/, '/(?[\c[]](])/ should be a syntax error');
+	like($@, qr/^Unexpected/, '/(?[\c[]](])/ should be a syntax error');
 	like("\c#", qr/(?[\c#])/, '\c# should match itself');
 	like("\c[", qr/(?[\c[])/, '\c[ should match itself');
 	like("\c\ ", qr/(?[\c\])/, '\c\ should match itself');