Fix \z behaviour when matching within invalid UTF

Author: PhilipHazel

ChangeLog

@@ -162,6 +162,8 @@ locks out any accidental sign-extension.
 literal character with a value greater than or equal to 0x80000000u caused 
 undefined behaviour.
 
+43. \z was misbehaving when matching fragments inside invalid UTF strings.
+
 
 Version 10.42 11-December-2022
 ------------------------------

src/pcre2_match.c

@@ -6076,10 +6076,12 @@ fprintf(stderr, "++ %2ld op=%3d %s\n", Fecode - mb->start_code, *Fecode,
     if ((mb->poptions & PCRE2_DOLLAR_ENDONLY) == 0) goto ASSERT_NL_OR_EOS;
 
     /* Fall through */
-    /* Unconditional end of subject assertion (\z) */
+    /* Unconditional end of subject assertion (\z). We must check NOTEOL
+    because it gets set for invalid UTF fragments. */
 
     case OP_EOD:
-    if (Feptr < mb->end_subject) RRETURN(MATCH_NOMATCH);
+    if (Feptr < mb->end_subject || (mb->moptions & PCRE2_NOTEOL) != 0)
+      RRETURN(MATCH_NOMATCH);
     if (mb->partial != 0)
       {
       mb->hitend = TRUE;

testdata/testinput10

@@ -642,4 +642,7 @@
     line1\nbreak
     line0\nline1\nbreak
 
+/A\z/utf,match_invalid_utf
+    A\x80\x42\n
+
 # End of testinput10

testdata/testinput12

@@ -464,6 +464,9 @@
 
 /aa/utf,ucp,match_invalid_utf,global
     \x{d800}aa
+    
+/A\z/utf,match_invalid_utf
+    A\x{df00}\n
 
 # ---------------------------------------------------- 
 

testdata/testoutput10

@@ -1921,4 +1921,8 @@ Partial match:
     line0\nline1\nbreak
 No match
 
+/A\z/utf,match_invalid_utf
+    A\x80\x42\n
+No match
+
 # End of testinput10

testdata/testoutput12-16

@@ -1607,6 +1607,10 @@ No match
 /aa/utf,ucp,match_invalid_utf,global
     \x{d800}aa
  0: aa
+    
+/A\z/utf,match_invalid_utf
+    A\x{df00}\n
+No match
 
 # ---------------------------------------------------- 
 

testdata/testoutput12-32

@@ -1605,6 +1605,10 @@ No match
 /aa/utf,ucp,match_invalid_utf,global
     \x{d800}aa
  0: aa
+    
+/A\z/utf,match_invalid_utf
+    A\x{df00}\n
+No match
 
 # ----------------------------------------------------