All the remaining changes for handling characters with more than one other

case.


git-svn-id: svn://vcs.exim.org/pcre/code/trunk@1046 2f5784b3-3f2a-0410-8824-cb99058d5e15

1 files changed, 40 insertions, 7 deletions

diff --git a/pcre_printint.c b/pcre_printint.c
index 1767172..108d72f 100644
--- a/pcre_printint.c
+++ b/pcre_printint.c
@@ -120,7 +120,7 @@ int c = *ptr;
 
 (void)utf;  /* Avoid compiler warning */
 if (PRINTABLE(c)) fprintf(f, "%c", c);
-else if (c <= 0xff) fprintf(f, "\\x%02x", c);
+else if (c <= 0x80) fprintf(f, "\\x%02x", c);
 else fprintf(f, "\\x{%x}", c);
 return 0;
 
@@ -233,6 +233,40 @@ return (ptype == pvalue)? "??" : "??";
 }
 
 
+/*************************************************
+*       Print Unicode property value             *
+*************************************************/
+
+/* "Normal" properties can be printed from tables. The PT_CLIST property is a 
+pseudo-property that contains a pointer to a list of case-equivalent 
+characters. This is used only when UCP support is available and UTF mode is 
+selected. It should never occur otherwise, but just in case it does, have 
+something ready to print. */
+
+static void
+print_prop(FILE *f, pcre_uchar *code, const char *before, const char *after)
+{
+if (code[1] != PT_CLIST)
+  {
+  fprintf(f, "%s%s %s%s", before, priv_OP_names[*code], get_ucpname(code[1], 
+    code[2]), after);
+  }
+else
+  {
+  const char *not = (*code == OP_PROP)? "" : "not ";
+#ifndef SUPPORT_UCP   
+  fprintf(f, "%s%sclist %d%s", before, not, code[2], after);
+#else
+  const pcre_uint32 *p = PRIV(ucd_caseless_sets) + code[2];
+  fprintf (f, "%s%sclist", before, not);
+  while (*p < NOTACHAR) fprintf(f, " %04x", *p++); 
+  fprintf(f, "%s", after); 
+#endif  
+  }     
+}
+
+
+
 
 /*************************************************
 *         Print compiled regex                   *
@@ -427,12 +461,12 @@ for(;;)
     fprintf(f, " %s ", flag);
     if (*code >= OP_TYPESTAR)
       {
-      fprintf(f, "%s", priv_OP_names[code[1]]);
       if (code[1] == OP_PROP || code[1] == OP_NOTPROP)
         {
-        fprintf(f, " %s ", get_ucpname(code[2], code[3]));
+        print_prop(f, code + 1, "", " ");
         extra = 2;
         }
+      else fprintf(f, "%s", priv_OP_names[code[1]]);
       }
     else extra = print_char(f, code+1, utf);
     fprintf(f, "%s", priv_OP_names[*code]);
@@ -461,13 +495,12 @@ for(;;)
     case OP_TYPEUPTO:
     case OP_TYPEMINUPTO:
     case OP_TYPEPOSUPTO:
-    fprintf(f, "    %s", priv_OP_names[code[1 + IMM2_SIZE]]);
     if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
       {
-      fprintf(f, " %s ", get_ucpname(code[1 + IMM2_SIZE + 1],
-        code[1 + IMM2_SIZE + 2]));
+      print_prop(f, code + IMM2_SIZE + 1, "    ", " ");
       extra = 2;
       }
+    else fprintf(f, "    %s", priv_OP_names[code[1 + IMM2_SIZE]]);
     fprintf(f, "{");
     if (*code != OP_TYPEEXACT) fprintf(f, "0,");
     fprintf(f, "%d}", GET2(code,1));
@@ -552,7 +585,7 @@ for(;;)
 
     case OP_PROP:
     case OP_NOTPROP:
-    fprintf(f, "    %s %s", priv_OP_names[*code], get_ucpname(code[1], code[2]));
+    print_prop(f, code, "    ", "");
     break;
 
     /* OP_XCLASS can only occur in UTF or PCRE16 modes. However, there's no