Use [^...] rather than '[...] (neg)' to print NCLASS (#580)

Author: NWilson

src/pcre2_printint.c

@@ -461,38 +461,35 @@ static void
 print_class(FILE *f, PCRE2_SPTR code, const uint8_t *char_lists_end, BOOL utf,
   const char *before, const char *after)
 {
-BOOL printmap, invertmap;
+BOOL printmap, negated;
 PCRE2_SPTR ccode;
 int i;
 
-fprintf(f, "%s[", before);
-
-/* Negative XCLASS has an inverted map whereas the original opcodes have
-already done the inversion. */
-invertmap = FALSE;
+/* Negative XCLASS and NCLASS both have a bitmap indicating which characters
+are accepted. For clarity we print this inverted and prefixed by "^". */
 if (*code == OP_XCLASS)
   {
   ccode = code + LINK_SIZE + 1;
   printmap = (*ccode & XCL_MAP) != 0;
-  if ((*ccode & XCL_NOT) != 0)
-    {
-    invertmap = TRUE;
-    fprintf(f, "^");
-    }
+  negated = (*ccode & XCL_NOT) != 0;
   ccode++;
   }
 else  /* CLASS or NCLASS */
   {
   printmap = TRUE;
+  negated = *code == OP_NCLASS;
   ccode = code + 1;
   }
 
+fprintf(f, "%s[%s", before, negated? "^" : "");
+
 /* Print a bit map */
 if (printmap)
   {
+  BOOL first = TRUE;
   uint8_t inverted_map[32];
   const uint8_t *map = (const uint8_t *)ccode;
-  if (invertmap)
+  if (negated)
     {
     /* Using 255 ^ instead of ~ avoids clang sanitize warning. */
     for (i = 0; i < 32; i++) inverted_map[i] = 255 ^ map[i];
@@ -505,13 +502,15 @@ if (printmap)
       int j;
       for (j = i+1; j < 256; j++)
         if ((map[j/8] & (1u << (j&7))) == 0) break;
-      if (i == '-' || i == ']') fprintf(f, "\\");
+      if (i == '-' || i == '\\' || i == ']' || (first && i == '^'))
+        fprintf(f, "\\");
       if (PRINTABLE(i)) fprintf(f, "%c", i);
         else fprintf(f, "\\x%02x", i);
+      first = FALSE;
       if (--j > i)
         {
         if (j != i + 1) fprintf(f, "-");
-        if (j == '-' || j == ']') fprintf(f, "\\");
+        if (j == '-' || j == '\\' || j == ']') fprintf(f, "\\");
         if (PRINTABLE(j)) fprintf(f, "%c", j);
           else fprintf(f, "\\x%02x", j);
         }
@@ -584,7 +583,7 @@ if (*code == OP_XCLASS)
   }
 
 /* Indicate a non-UTF class which was created by negation */
-fprintf(f, "]%s%s", (*code == OP_NCLASS)? " (neg)" : "", after);
+fprintf(f, "]%s", after);
 }
 
 
@@ -846,7 +845,7 @@ for(;;)
     case OP_NOT:
     fprintf(f, " %s [^", flag);
     extra = print_char(f, code + 1, utf);
-    fprintf(f, "]");
+    fprintf(f, "] (not)");
     break;
 
     case OP_NOTSTARI:
@@ -872,7 +871,7 @@ for(;;)
     case OP_NOTPOSQUERY:
     fprintf(f, " %s [^", flag);
     extra = print_char(f, code + 1, utf);
-    fprintf(f, "]%s", OP_names[*code]);
+    fprintf(f, "]%s (not)", OP_names[*code]);
     break;
 
     case OP_NOTEXACTI:
@@ -894,6 +893,7 @@ for(;;)
     if (*code == OP_NOTMINUPTO || *code == OP_NOTMINUPTOI) fprintf(f, "?");
       else
     if (*code == OP_NOTPOSUPTO || *code == OP_NOTPOSUPTOI) fprintf(f, "+");
+    fprintf(f, " (not)");
     break;
 
     case OP_RECURSE:

src/pcre2test.c

@@ -242,7 +242,7 @@ to hold them as 32-bit code units. */
 enum { PR_OK, PR_SKIP, PR_ABEND };
 
 /* The macro PRINTABLE determines whether to print an output character as-is or
-as a hex value when showing compiled patterns. is We use it in cases when the
+as a hex value when showing compiled patterns. We use it in cases when the
 locale has not been explicitly changed, so as to get consistent output from
 systems that differ in their output from isprint() even in the "C" locale. */
 

testdata/testinput2

@@ -7228,4 +7228,18 @@ a)"xI
 
 /[\E/
 
+/[\^z]/B
+
+/[ \^]/B
+
+/[\\z]/B
+
+/[0-z]/B
+
+/[0\-z]/B
+
+/[]z]/B
+
+/[ \]]/B
+
 # End of testinput2

testdata/testoutput10

@@ -461,7 +461,7 @@ Subject length lower bound = 1
 /[^ab\xC0-\xF0]/IB,utf
 ------------------------------------------------------------------
         Bra
-        [\x00-`c-\xbf\xf1-\xff] (neg)
+        [^ab\xc0-\xf0]
         Ket
         End
 ------------------------------------------------------------------
@@ -604,7 +604,7 @@ Subject length lower bound = 3
 /[^\x{c4}]/IB
 ------------------------------------------------------------------
         Bra
-        [^\x{c4}]
+        [^\x{c4}] (not)
         Ket
         End
 ------------------------------------------------------------------
@@ -648,7 +648,7 @@ Subject length lower bound = 1
 /[^\xff]/IB,utf
 ------------------------------------------------------------------
         Bra
-        [^\x{ff}]
+        [^\x{ff}] (not)
         Ket
         End
 ------------------------------------------------------------------
@@ -898,7 +898,7 @@ Subject length lower bound = 2
 /[^\x{c4}]/IB,utf
 ------------------------------------------------------------------
         Bra
-        [^\x{c4}]
+        [^\x{c4}] (not)
         Ket
         End
 ------------------------------------------------------------------
@@ -1159,7 +1159,7 @@ Subject length lower bound = 17
 /[^ⱥ]/Bi,utf
 ------------------------------------------------------------------
         Bra
-     /i [^\x{2c65}]
+     /i [^\x{2c65}] (not)
         Ket
         End
 ------------------------------------------------------------------
@@ -1559,7 +1559,7 @@ Subject length lower bound = 1
 ------------------------------------------------------------------
         Bra
         ^
-        [\x00-`c-\xff] (neg)
+        [^ab]
         Ket
         End
 ------------------------------------------------------------------

testdata/testoutput11-16

@@ -9,7 +9,7 @@
 /[^\x{c4}]/IB
 ------------------------------------------------------------------
         Bra
-        [^\x{c4}]
+        [^\x{c4}] (not)
         Ket
         End
 ------------------------------------------------------------------
@@ -364,55 +364,55 @@ Subject length lower bound = 6
 /[^\x{80}][^\x{ff}][^\x{100}][^\x{1000}][^\x{ffff}]/B
 ------------------------------------------------------------------
         Bra
-        [^\x{80}]
-        [^\x{ff}]
-        [^\x{100}]
-        [^\x{1000}]
-        [^\x{ffff}]
+        [^\x{80}] (not)
+        [^\x{ff}] (not)
+        [^\x{100}] (not)
+        [^\x{1000}] (not)
+        [^\x{ffff}] (not)
         Ket
         End
 ------------------------------------------------------------------
 
 /[^\x{80}][^\x{ff}][^\x{100}][^\x{1000}][^\x{ffff}]/Bi
 ------------------------------------------------------------------
         Bra
-     /i [^\x{80}]
-     /i [^\x{ff}]
-     /i [^\x{100}]
-     /i [^\x{1000}]
-     /i [^\x{ffff}]
+     /i [^\x{80}] (not)
+     /i [^\x{ff}] (not)
+     /i [^\x{100}] (not)
+     /i [^\x{1000}] (not)
+     /i [^\x{ffff}] (not)
         Ket
         End
 ------------------------------------------------------------------
 
 /[^\x{100}]*[^\x{1000}]+[^\x{ffff}]??[^\x{8000}]{4,}[^\x{7fff}]{2,9}?[^\x{100}]{5,6}+/B
 ------------------------------------------------------------------
         Bra
-        [^\x{100}]*
-        [^\x{1000}]+
-        [^\x{ffff}]??
-        [^\x{8000}]{4}
-        [^\x{8000}]*
-        [^\x{7fff}]{2}
-        [^\x{7fff}]{0,7}?
-        [^\x{100}]{5}
-        [^\x{100}]?+
+        [^\x{100}]* (not)
+        [^\x{1000}]+ (not)
+        [^\x{ffff}]?? (not)
+        [^\x{8000}]{4} (not)
+        [^\x{8000}]* (not)
+        [^\x{7fff}]{2} (not)
+        [^\x{7fff}]{0,7}? (not)
+        [^\x{100}]{5} (not)
+        [^\x{100}]?+ (not)
         Ket
         End
 ------------------------------------------------------------------
 
 /[^\x{100}]*[^\x{1000}]+[^\x{ffff}]??[^\x{8000}]{4,}[^\x{7fff}]{2,9}?[^\x{100}]{5,6}+/Bi
 ------------------------------------------------------------------
         Bra
-     /i [^\x{100}]*
-     /i [^\x{1000}]+
-     /i [^\x{ffff}]??
-     /i [^\x{8000}]{4}
-     /i [^\x{8000}]*
-     /i [^\x{7fff}]{2}
-     /i [^\x{7fff}]{0,7}?
-     /i [^\x{100}]{5}
-     /i [^\x{100}]?+
+     /i [^\x{100}]* (not)
+     /i [^\x{1000}]+ (not)
+     /i [^\x{ffff}]?? (not)
+     /i [^\x{8000}]{4} (not)
+     /i [^\x{8000}]* (not)
+     /i [^\x{7fff}]{2} (not)
+     /i [^\x{7fff}]{0,7}? (not)
+     /i [^\x{100}]{5} (not)
+     /i [^\x{100}]?+ (not)
         Ket
         End
 ------------------------------------------------------------------
@@ -474,25 +474,25 @@ MK: 0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF0123456789AB
 /[^\x00-a]{12,}[^b-\xff]*/B
 ------------------------------------------------------------------
         Bra
-        [b-\xff] (neg){12,}
-        [\x00-a] (neg)*+
+        [^\x00-a]{12,}
+        [^b-\xff]*+
         Ket
         End
 ------------------------------------------------------------------
 
 /[^\s]*\s* [^\W]+\W+ [^\d]*?\d0 [^\d\w]{4,6}?\w*A/B
 ------------------------------------------------------------------
         Bra
-        [\x00-\x08\x0e-\x1f!-\xff] (neg)*
+        [^\x09-\x0d ]*
         \s*
 
         [0-9A-Z_a-z]++
         \W+
 
-        [\x00-/:-\xff] (neg)*?
+        [^0-9]*?
         \d
         0 
-        [\x00-/:-@[-^`{-\xff] (neg){4,6}?
+        [^0-9A-Z_a-z]{4,6}?
         \w*
         A
         Ket