From 1143cd05200461754ddcbf7845a690710e87c1b5 Mon Sep 17 00:00:00 2001
From: "jan.nijtmans" <nijtmans@users.sourceforge.net>
Date: Wed, 31 Jan 2024 12:42:32 +0000
Subject: Remove private characters from regexp control table, but add them
 back in [:cntrl:] class (so no change in regexp handling). Eliminated
 (size_t) type-casts. This makes implementing more character-classes easier.

---
 generic/regc_locale.c | 80 +++++++++++++++++++++++++--------------------------
 tools/uniClass.tcl    | 10 +++++--
 2 files changed, 47 insertions(+), 43 deletions(-)

diff --git a/generic/regc_locale.c b/generic/regc_locale.c
index 9a984f5..e8f9381 100644
--- a/generic/regc_locale.c
+++ b/generic/regc_locale.c
@@ -254,7 +254,7 @@ static const crange alphaRangeTable[] = {
 #endif
 };
 
-#define NUM_ALPHA_RANGE (sizeof(alphaRangeTable)/sizeof(crange))
+#define NUM_ALPHA_RANGE ((int)(sizeof(alphaRangeTable)/sizeof(crange)))
 
 static const chr alphaCharTable[] = {
     0xAA, 0xB5, 0xBA, 0x2EC, 0x2EE, 0x376, 0x377, 0x37F, 0x386,
@@ -291,7 +291,7 @@ static const chr alphaCharTable[] = {
 #endif
 };
 
-#define NUM_ALPHA_CHAR (sizeof(alphaCharTable)/sizeof(chr))
+#define NUM_ALPHA_CHAR ((int)(sizeof(alphaCharTable)/sizeof(chr)))
 
 /*
  * Unicode: control characters.
@@ -299,14 +299,13 @@ static const chr alphaCharTable[] = {
 
 static const crange controlRangeTable[] = {
     {0x0, 0x1F}, {0x7F, 0x9F}, {0x600, 0x605}, {0x200B, 0x200F},
-    {0x202A, 0x202E}, {0x2060, 0x2064}, {0x2066, 0x206F}, {0xE000, 0xF8FF},
-    {0xFFF9, 0xFFFB}
+    {0x202A, 0x202E}, {0x2060, 0x2064}, {0x2066, 0x206F}, {0xFFF9, 0xFFFB}
 #if CHRBITS > 16
     ,{0x13430, 0x1343F}, {0x1BCA0, 0x1BCA3}, {0x1D173, 0x1D17A}, {0xE0020, 0xE007F}
 #endif
 };
 
-#define NUM_CONTROL_RANGE (sizeof(controlRangeTable)/sizeof(crange))
+#define NUM_CONTROL_RANGE ((int)(sizeof(controlRangeTable)/sizeof(crange)))
 
 static const chr controlCharTable[] = {
     0xAD, 0x61C, 0x6DD, 0x70F, 0x890, 0x891, 0x8E2, 0x180E, 0xFEFF
@@ -315,7 +314,7 @@ static const chr controlCharTable[] = {
 #endif
 };
 
-#define NUM_CONTROL_CHAR (sizeof(controlCharTable)/sizeof(chr))
+#define NUM_CONTROL_CHAR ((int)(sizeof(controlCharTable)/sizeof(chr)))
 
 /*
  * Unicode: decimal digit characters.
@@ -343,7 +342,7 @@ static const crange digitRangeTable[] = {
 #endif
 };
 
-#define NUM_DIGIT_RANGE (sizeof(digitRangeTable)/sizeof(crange))
+#define NUM_DIGIT_RANGE ((int)(sizeof(digitRangeTable)/sizeof(crange)))
 
 /*
  * no singletons of digit characters.
@@ -380,7 +379,7 @@ static const crange punctRangeTable[] = {
 #endif
 };
 
-#define NUM_PUNCT_RANGE (sizeof(punctRangeTable)/sizeof(crange))
+#define NUM_PUNCT_RANGE ((int)(sizeof(punctRangeTable)/sizeof(crange)))
 
 static const chr punctCharTable[] = {
     0x3A, 0x3B, 0x3F, 0x40, 0x5F, 0x7B, 0x7D, 0xA1, 0xA7,
@@ -405,7 +404,7 @@ static const chr punctCharTable[] = {
 #endif
 };
 
-#define NUM_PUNCT_CHAR (sizeof(punctCharTable)/sizeof(chr))
+#define NUM_PUNCT_CHAR ((int)(sizeof(punctCharTable)/sizeof(chr)))
 
 /*
  * Unicode: white space characters.
@@ -415,14 +414,14 @@ static const crange spaceRangeTable[] = {
     {0x9, 0xD}, {0x2000, 0x200B}
 };
 
-#define NUM_SPACE_RANGE (sizeof(spaceRangeTable)/sizeof(crange))
+#define NUM_SPACE_RANGE ((int)(sizeof(spaceRangeTable)/sizeof(crange)))
 
 static const chr spaceCharTable[] = {
     0x20, 0x85, 0xA0, 0x1680, 0x180E, 0x2028, 0x2029, 0x202F, 0x205F,
     0x2060, 0x3000, 0xFEFF
 };
 
-#define NUM_SPACE_CHAR (sizeof(spaceCharTable)/sizeof(chr))
+#define NUM_SPACE_CHAR ((int)(sizeof(spaceCharTable)/sizeof(chr)))
 
 /*
  * Unicode: lowercase characters.
@@ -456,7 +455,7 @@ static const crange lowerRangeTable[] = {
 #endif
 };
 
-#define NUM_LOWER_RANGE (sizeof(lowerRangeTable)/sizeof(crange))
+#define NUM_LOWER_RANGE ((int)(sizeof(lowerRangeTable)/sizeof(crange)))
 
 static const chr lowerCharTable[] = {
     0xB5, 0x101, 0x103, 0x105, 0x107, 0x109, 0x10B, 0x10D, 0x10F,
@@ -529,7 +528,7 @@ static const chr lowerCharTable[] = {
 #endif
 };
 
-#define NUM_LOWER_CHAR (sizeof(lowerCharTable)/sizeof(chr))
+#define NUM_LOWER_CHAR ((int)(sizeof(lowerCharTable)/sizeof(chr)))
 
 /*
  * Unicode: uppercase characters.
@@ -559,7 +558,7 @@ static const crange upperRangeTable[] = {
 #endif
 };
 
-#define NUM_UPPER_RANGE (sizeof(upperRangeTable)/sizeof(crange))
+#define NUM_UPPER_RANGE ((int)(sizeof(upperRangeTable)/sizeof(crange)))
 
 static const chr upperCharTable[] = {
     0x100, 0x102, 0x104, 0x106, 0x108, 0x10A, 0x10C, 0x10E, 0x110,
@@ -633,7 +632,7 @@ static const chr upperCharTable[] = {
 #endif
 };
 
-#define NUM_UPPER_CHAR (sizeof(upperCharTable)/sizeof(chr))
+#define NUM_UPPER_CHAR ((int)(sizeof(upperCharTable)/sizeof(chr)))
 
 /*
  * Unicode: unicode print characters excluding space.
@@ -792,7 +791,7 @@ static const crange graphRangeTable[] = {
 #endif
 };
 
-#define NUM_GRAPH_RANGE (sizeof(graphRangeTable)/sizeof(crange))
+#define NUM_GRAPH_RANGE ((int)(sizeof(graphRangeTable)/sizeof(crange)))
 
 static const chr graphCharTable[] = {
     0x38C, 0x85E, 0x98F, 0x990, 0x9B2, 0x9C7, 0x9C8, 0x9D7, 0x9DC,
@@ -820,7 +819,7 @@ static const chr graphCharTable[] = {
 #endif
 };
 
-#define NUM_GRAPH_CHAR (sizeof(graphCharTable)/sizeof(chr))
+#define NUM_GRAPH_CHAR ((int)(sizeof(graphCharTable)/sizeof(chr)))
 
 /*
  *	End of auto-generated Unicode character ranges declarations.
@@ -1067,14 +1066,14 @@ cclass(
     case CC_ALNUM:
 	cv = getcvec(v, NUM_ALPHA_CHAR, NUM_DIGIT_RANGE + NUM_ALPHA_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_ALPHA_CHAR ; i++) {
+	    for (i=0 ; i<NUM_ALPHA_CHAR ; i++) {
 		addchr(cv, alphaCharTable[i]);
 	    }
-	    for (i=0 ; (size_t)i<NUM_ALPHA_RANGE ; i++) {
+	    for (i=0 ; i<NUM_ALPHA_RANGE ; i++) {
 		addrange(cv, alphaRangeTable[i].start,
 			alphaRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_DIGIT_RANGE ; i++) {
+	    for (i=0 ; i<NUM_DIGIT_RANGE ; i++) {
 		addrange(cv, digitRangeTable[i].start,
 			digitRangeTable[i].end);
 	    }
@@ -1083,11 +1082,11 @@ cclass(
     case CC_ALPHA:
 	cv = getcvec(v, NUM_ALPHA_CHAR, NUM_ALPHA_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_ALPHA_RANGE ; i++) {
+	    for (i=0 ; i<NUM_ALPHA_RANGE ; i++) {
 		addrange(cv, alphaRangeTable[i].start,
 			alphaRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_ALPHA_CHAR ; i++) {
+	    for (i=0 ; i<NUM_ALPHA_CHAR ; i++) {
 		addchr(cv, alphaCharTable[i]);
 	    }
 	}
@@ -1104,13 +1103,14 @@ cclass(
 	addchr(cv, ' ');
 	break;
     case CC_CNTRL:
-	cv = getcvec(v, NUM_CONTROL_CHAR, NUM_CONTROL_RANGE);
+	cv = getcvec(v, NUM_CONTROL_CHAR, NUM_CONTROL_RANGE+1);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_CONTROL_RANGE ; i++) {
+	    addrange(cv, 0xE000, 0xF8FF); /* private */
+	    for (i=0 ; i<NUM_CONTROL_RANGE ; i++) {
 		addrange(cv, controlRangeTable[i].start,
 			controlRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_CONTROL_CHAR ; i++) {
+	    for (i=0 ; i<NUM_CONTROL_CHAR ; i++) {
 		addchr(cv, controlCharTable[i]);
 	    }
 	}
@@ -1118,7 +1118,7 @@ cclass(
     case CC_DIGIT:
 	cv = getcvec(v, 0, NUM_DIGIT_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_DIGIT_RANGE ; i++) {
+	    for (i=0 ; i<NUM_DIGIT_RANGE ; i++) {
 		addrange(cv, digitRangeTable[i].start,
 			digitRangeTable[i].end);
 	    }
@@ -1127,11 +1127,11 @@ cclass(
     case CC_PUNCT:
 	cv = getcvec(v, NUM_PUNCT_CHAR, NUM_PUNCT_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_PUNCT_RANGE ; i++) {
+	    for (i=0 ; i<NUM_PUNCT_RANGE ; i++) {
 		addrange(cv, punctRangeTable[i].start,
 			punctRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_PUNCT_CHAR ; i++) {
+	    for (i=0 ; i<NUM_PUNCT_CHAR ; i++) {
 		addchr(cv, punctCharTable[i]);
 	    }
 	}
@@ -1156,11 +1156,11 @@ cclass(
     case CC_SPACE:
 	cv = getcvec(v, NUM_SPACE_CHAR, NUM_SPACE_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_SPACE_RANGE ; i++) {
+	    for (i=0 ; i<NUM_SPACE_RANGE ; i++) {
 		addrange(cv, spaceRangeTable[i].start,
 			spaceRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_SPACE_CHAR ; i++) {
+	    for (i=0 ; i<NUM_SPACE_CHAR ; i++) {
 		addchr(cv, spaceCharTable[i]);
 	    }
 	}
@@ -1168,11 +1168,11 @@ cclass(
     case CC_LOWER:
 	cv  = getcvec(v, NUM_LOWER_CHAR, NUM_LOWER_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_LOWER_RANGE ; i++) {
+	    for (i=0 ; i<NUM_LOWER_RANGE ; i++) {
 		addrange(cv, lowerRangeTable[i].start,
 			lowerRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_LOWER_CHAR ; i++) {
+	    for (i=0 ; i<NUM_LOWER_CHAR ; i++) {
 		addchr(cv, lowerCharTable[i]);
 	    }
 	}
@@ -1180,11 +1180,11 @@ cclass(
     case CC_UPPER:
 	cv  = getcvec(v, NUM_UPPER_CHAR, NUM_UPPER_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_UPPER_RANGE ; i++) {
+	    for (i=0 ; i<NUM_UPPER_RANGE ; i++) {
 		addrange(cv, upperRangeTable[i].start,
 			upperRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_UPPER_CHAR ; i++) {
+	    for (i=0 ; i<NUM_UPPER_CHAR ; i++) {
 		addchr(cv, upperCharTable[i]);
 	    }
 	}
@@ -1192,18 +1192,18 @@ cclass(
     case CC_PRINT:
     	cv  = getcvec(v, NUM_SPACE_CHAR + NUM_GRAPH_CHAR, NUM_SPACE_RANGE + NUM_GRAPH_RANGE  - 1);
     	if (cv) {
-    	    for (i=1 ; (size_t)i<NUM_SPACE_RANGE ; i++) {
+    	    for (i=1 ; i<NUM_SPACE_RANGE ; i++) {
     		addrange(cv, spaceRangeTable[i].start,
     				spaceRangeTable[i].end);
     	    }
-    	    for (i=0 ; (size_t)i<NUM_SPACE_CHAR ; i++) {
+    	    for (i=0 ; i<NUM_SPACE_CHAR ; i++) {
     		addchr(cv, spaceCharTable[i]);
     	    }
-    	    for (i=0 ; (size_t)i<NUM_GRAPH_RANGE ; i++) {
+    	    for (i=0 ; i<NUM_GRAPH_RANGE ; i++) {
     		addrange(cv, graphRangeTable[i].start,
     				graphRangeTable[i].end);
     	    }
-    	    for (i=0 ; (size_t)i<NUM_GRAPH_CHAR ; i++) {
+    	    for (i=0 ; i<NUM_GRAPH_CHAR ; i++) {
     		addchr(cv, graphCharTable[i]);
     	    }
     	}
@@ -1211,11 +1211,11 @@ cclass(
     case CC_GRAPH:
 	cv  = getcvec(v, NUM_GRAPH_CHAR, NUM_GRAPH_RANGE);
 	if (cv) {
-	    for (i=0 ; (size_t)i<NUM_GRAPH_RANGE ; i++) {
+	    for (i=0 ; i<NUM_GRAPH_RANGE ; i++) {
 		addrange(cv, graphRangeTable[i].start,
 			graphRangeTable[i].end);
 	    }
-	    for (i=0 ; (size_t)i<NUM_GRAPH_CHAR ; i++) {
+	    for (i=0 ; i<NUM_GRAPH_CHAR ; i++) {
 		addchr(cv, graphCharTable[i]);
 	    }
 	}
diff --git a/tools/uniClass.tcl b/tools/uniClass.tcl
index 39fa28d..a4a58c2 100644
--- a/tools/uniClass.tcl
+++ b/tools/uniClass.tcl
@@ -63,11 +63,15 @@ proc genTable {type} {
     set extchars 0
     set extranges 0
 
-    for {set i 0} {$i <= 0x10FFFF} {incr i} {
+    for {set i 0} {$i <= 0xEFFFF} {incr i} {
 	if {$i == 0xD800} {
 	    # Skip surrogates
 	    set i 0xE000
 	}
+	if {$i == 0xE000} {
+	    # Skip private
+	    set i 0xF900
+	}
 	if {[string is $type [format %c $i]]} {
 	    if {$i == ($last + 1)} {
 		set last $i
@@ -92,13 +96,13 @@ proc genTable {type} {
     }
     if {$ranges ne ""} {
 	puts "static const crange ${type}RangeTable\[\] = {\n$ranges\n};\n"
-	puts "#define NUM_[string toupper $type]_RANGE (sizeof(${type}RangeTable)/sizeof(crange))\n"
+	puts "#define NUM_[string toupper $type]_RANGE ((int)(sizeof(${type}RangeTable)/sizeof(crange)))\n"
     } else {
 	puts "/* no contiguous ranges of $type characters */\n"
     }
     if {$chars ne ""} {
 	puts "static const chr ${type}CharTable\[\] = {\n$chars\n};\n"
-	puts "#define NUM_[string toupper $type]_CHAR (sizeof(${type}CharTable)/sizeof(chr))\n"
+	puts "#define NUM_[string toupper $type]_CHAR ((int)(sizeof(${type}CharTable)/sizeof(chr)))\n"
     } else {
 	puts "/*\n * no singletons of $type characters.\n */\n"
     }
-- 
cgit v0.12