gh18770: stop scanning for substrs after *COMMIT

*ACCEPT already avoids this (because it is "ENDLIKE"), but gets a related fix to stop scanning for start class.
Perl · Jun 1, 2021 · 499aa13 · 499aa13
1 parent 403d7eb
commit 499aa13
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 6 deletions.
diff --git a/regcomp.c b/regcomp.c
@@ -6350,18 +6350,32 @@ Perl_re_printf( aTHX_  "LHS=%" UVuf " RHS=%" UVuf "\n",
                 *(data->last_closep) = ARG(scan);
         }
         else if (OP(scan) == EVAL) {
-                if (data)
-                    data->flags |= SF_HAS_EVAL;
+            if (data)
+                data->flags |= SF_HAS_EVAL;
         }
         else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
             if (flags & SCF_DO_SUBSTR) {
                 scan_commit(pRExC_state, data, minlenp, is_inf);
                 flags &= ~SCF_DO_SUBSTR;
             }
-            if (data && OP(scan)==ACCEPT) {
-                data->flags |= SCF_SEEN_ACCEPT;
-                if (stopmin > min)
-                    stopmin = min;
+            if (OP(scan)==ACCEPT) {
+                /* m{(*ACCEPT)x} does not have to start with 'x' */
+                flags &= ~SCF_DO_STCLASS;
+                if (data) {
+                    data->flags |= SCF_SEEN_ACCEPT;
+                    if (stopmin > min)
+                        stopmin = min;
+                }
+            }
+        }
+        else if (OP(scan) == COMMIT) {
+            /* gh18770: m{abc(*COMMIT)xyz} must fail on "abc abcxyz", so we
+             * must not end up with "abcxyz" as a fixed substring else we'll
+             * skip straight to attempting to match at offset 4.
+             */
+            if (flags & SCF_DO_SUBSTR) {
+                scan_commit(pRExC_state, data, minlenp, is_inf);
+                flags &= ~SCF_DO_SUBSTR;
             }
         }
         else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */

diff --git a/t/re/opt.t b/t/re/opt.t
@@ -268,3 +268,8 @@ acdef|bcdeg	5	1+cde	-	Tanchored,stclass=~[ab]
 a(b){2,3}c	4	-abb	1+bbc
 a(b|bb)c	3	-ab	1-bc	Tfloating,Tfloating min offset
 a(b|bb){2}c	4	-abb	1-bbc	Tanchored,Tfloating,Tfloating min offset
+
+abc(*COMMIT)xyz	6	0+abc	-	-
+abc(*ACCEPT)xyz	3	0+abc	-	-
+# Must not have stclass=[x]
+(*ACCEPT)xyz	0	-	-	-