Patch to PCRE2 to handle ill-formed UTF8 data (assuming it is safe to skip the length check)

pcre-handle-ill-formed-utf8.diff

diff -prU12 pcre2-10.30-release/src/pcre2_internal.h pcre2-10.30/src/pcre2_internal.h
--- pcre2-10.30-release/src/pcre2_internal.h	2017-07-19 12:00:20.000000000 -0400
+++ pcre2-10.30/src/pcre2_internal.h	2018-06-20 17:03:09.000000000 -0400
@@ -271,103 +271,157 @@ is not supported. */
 
 /* The following macros were originally written in the form of loops that used
 data from the tables whose names start with PRIV(utf8_table). They were
 rewritten by a user so as not to use loops, because in some environments this
 gives a significant performance advantage, and it seems never to do any harm.
 */
 
 /* Base macro to pick up the remaining bytes of a UTF-8 character, not
 advancing the pointer. */
 
 #define GETUTF8(c, eptr) \
     { \
-    if ((c & 0x20u) == 0) \
+    if ((eptr[1] & 0xc0) != 0x80) \
+      c = 0xFFFD; \
+    else if ((c & 0x20u) == 0) \
       c = ((c & 0x1fu) << 6) | (eptr[1] & 0x3fu); \
+    else if ((eptr[2] & 0xc0) != 0x80) \
+      c = 0xFFFD; \
     else if ((c & 0x10u) == 0) \
       c = ((c & 0x0fu) << 12) | ((eptr[1] & 0x3fu) << 6) | (eptr[2] & 0x3fu); \
+    else if ((eptr[3] & 0xc0) != 0x80) \
+      c = 0xFFFD; \
     else if ((c & 0x08u) == 0) \
       c = ((c & 0x07u) << 18) | ((eptr[1] & 0x3fu) << 12) | \
       ((eptr[2] & 0x3fu) << 6) | (eptr[3] & 0x3fu); \
+    else if ((eptr[4] & 0xc0) != 0x80) \
+      c = 0xFFFD; \
     else if ((c & 0x04u) == 0) \
       c = ((c & 0x03u) << 24) | ((eptr[1] & 0x3fu) << 18) | \
           ((eptr[2] & 0x3fu) << 12) | ((eptr[3] & 0x3fu) << 6) | \
           (eptr[4] & 0x3fu); \
+    else if ((eptr[5] & 0xc0) != 0x80) \
+      c = 0xFFFD; \
     else \
       c = ((c & 0x01u) << 30) | ((eptr[1] & 0x3fu) << 24) | \
           ((eptr[2] & 0x3fu) << 18) | ((eptr[3] & 0x3fu) << 12) | \
           ((eptr[4] & 0x3fu) << 6) | (eptr[5] & 0x3fu); \
     }
 
 /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing
 the pointer. */
 
 #define GETUTF8INC(c, eptr) \
     { \
-    if ((c & 0x20u) == 0) \
+    if ((*eptr & 0xc0) != 0x80) \
+      c = 0xFFFD; \
+    else if ((c & 0x20u) == 0) \
       c = ((c & 0x1fu) << 6) | (*eptr++ & 0x3fu); \
+    else if ((eptr[1] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      eptr += 1; \
+      } \
     else if ((c & 0x10u) == 0) \
       { \
       c = ((c & 0x0fu) << 12) | ((*eptr & 0x3fu) << 6) | (eptr[1] & 0x3fu); \
       eptr += 2; \
       } \
+    else if ((eptr[2] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      eptr += 2; \
+      } \
     else if ((c & 0x08u) == 0) \
       { \
       c = ((c & 0x07u) << 18) | ((*eptr & 0x3fu) << 12) | \
           ((eptr[1] & 0x3fu) << 6) | (eptr[2] & 0x3fu); \
       eptr += 3; \
       } \
+    else if ((eptr[3] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      eptr += 3; \
+      } \
     else if ((c & 0x04u) == 0) \
       { \
       c = ((c & 0x03u) << 24) | ((*eptr & 0x3fu) << 18) | \
           ((eptr[1] & 0x3fu) << 12) | ((eptr[2] & 0x3fu) << 6) | \
           (eptr[3] & 0x3fu); \
       eptr += 4; \
       } \
+    else if ((eptr[4] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      eptr += 4; \
+      } \
     else \
       { \
       c = ((c & 0x01u) << 30) | ((*eptr & 0x3fu) << 24) | \
           ((eptr[1] & 0x3fu) << 18) | ((eptr[2] & 0x3fu) << 12) | \
           ((eptr[3] & 0x3fu) << 6) | (eptr[4] & 0x3fu); \
       eptr += 5; \
       } \
     }
 
 /* Base macro to pick up the remaining bytes of a UTF-8 character, not
 advancing the pointer, incrementing the length. */
 
 #define GETUTF8LEN(c, eptr, len) \
     { \
-    if ((c & 0x20u) == 0) \
+    if ((eptr[1] & 0xc0) != 0x80) \
+      c = 0xFFFD; \
+    else if ((c & 0x20u) == 0) \
       { \
       c = ((c & 0x1fu) << 6) | (eptr[1] & 0x3fu); \
       len++; \
       } \
+    else if ((eptr[2] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      len++; \
+      } \
     else if ((c & 0x10u)  == 0) \
       { \
       c = ((c & 0x0fu) << 12) | ((eptr[1] & 0x3fu) << 6) | (eptr[2] & 0x3fu); \
       len += 2; \
       } \
+    else if ((eptr[3] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      len += 2; \
+      } \
     else if ((c & 0x08u)  == 0) \
       {\
       c = ((c & 0x07u) << 18) | ((eptr[1] & 0x3fu) << 12) | \
           ((eptr[2] & 0x3fu) << 6) | (eptr[3] & 0x3fu); \
       len += 3; \
       } \
+    else if ((eptr[4] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      len += 3; \
+      } \
     else if ((c & 0x04u)  == 0) \
       { \
       c = ((c & 0x03u) << 24) | ((eptr[1] & 0x3fu) << 18) | \
           ((eptr[2] & 0x3fu) << 12) | ((eptr[3] & 0x3fu) << 6) | \
           (eptr[4] & 0x3fu); \
       len += 4; \
       } \
+    else if ((eptr[5] & 0xc0) != 0x80) \
+      { \
+      c = 0xFFFD; \
+      len += 4; \
+      } \
     else \
       {\
       c = ((c & 0x01u) << 30) | ((eptr[1] & 0x3fu) << 24) | \
           ((eptr[2] & 0x3fu) << 18) | ((eptr[3] & 0x3fu) << 12) | \
           ((eptr[4] & 0x3fu) << 6) | (eptr[5] & 0x3fu); \
       len += 5; \
       } \
     }
 
 /* --------------- Whitespace macros ---------------- */
 
 /* Tests for Unicode horizontal and vertical whitespace characters must check a