[Bug 3464428] string is graph \u0120 is wrong

1 files changed, 90 insertions, 85 deletions

diff --git a/generic/regc_locale.c b/generic/regc_locale.c
index 77cfd8e..ef8ecdc 100644
--- a/generic/regc_locale.c
+++ b/generic/regc_locale.c
@@ -12,7 +12,7 @@
 
 /* ASCII character-name table */
 
-static struct cname {
+static CONST struct cname {
     CONST char *name;
     CONST char code;
 } cnames[] = {
@@ -224,6 +224,23 @@ static CONST chr alphaCharTable[] = {
 #define NUM_ALPHA_CHAR (sizeof(alphaCharTable)/sizeof(chr))
 
 /*
+ * Unicode: control characters.
+ */
+
+static CONST crange controlRangeTable[] = {
+    {0x007f, 0x009f}, {0x0600, 0x0603}, {0x200b, 0x200f}, {0x202a, 0x202e},
+    {0x2060, 0x2064}, {0x206a, 0x206f}, {0xe000, 0xf8ff}, {0xfff9, 0xfffb}
+};
+
+#define NUM_CONTROL_RANGE (sizeof(controlRangeTable)/sizeof(crange))
+
+static CONST chr controlCharTable[] = {
+    0x00ad, 0x06dd, 0x070f, 0x17b4, 0x17b5, 0xfeff
+};
+
+#define NUM_CONTROL_CHAR (sizeof(controlCharTable)/sizeof(chr))
+
+/*
  * Unicode: decimal digit characters.
  */
 
@@ -478,7 +495,7 @@ static CONST chr upperCharTable[] = {
  */
 
 static CONST crange graphRangeTable[] = {
-    {0x0021, 0x007e}, {0x00a0, 0x00ac}, {0x00ae, 0x0377}, {0x037a, 0x037e},
+    {0x0021, 0x007e}, {0x00a1, 0x00ac}, {0x00ae, 0x0377}, {0x037a, 0x037e},
     {0x0384, 0x038a}, {0x038e, 0x03a1}, {0x03a3, 0x0527}, {0x0531, 0x0556},
     {0x0559, 0x055f}, {0x0561, 0x0587}, {0x0591, 0x05c7}, {0x05d0, 0x05ea},
     {0x05f0, 0x05f4}, {0x0606, 0x061b}, {0x061e, 0x06dc}, {0x06de, 0x070d},
@@ -513,21 +530,21 @@ static CONST crange graphRangeTable[] = {
     {0x1250, 0x1256}, {0x125a, 0x125d}, {0x1260, 0x1288}, {0x128a, 0x128d},
     {0x1290, 0x12b0}, {0x12b2, 0x12b5}, {0x12b8, 0x12be}, {0x12c2, 0x12c5},
     {0x12c8, 0x12d6}, {0x12d8, 0x1310}, {0x1312, 0x1315}, {0x1318, 0x135a},
-    {0x135d, 0x137c}, {0x1380, 0x1399}, {0x13a0, 0x13f4}, {0x1400, 0x169c},
-    {0x16a0, 0x16f0}, {0x1700, 0x170c}, {0x170e, 0x1714}, {0x1720, 0x1736},
-    {0x1740, 0x1753}, {0x1760, 0x176c}, {0x176e, 0x1770}, {0x1780, 0x17b3},
-    {0x17b6, 0x17dd}, {0x17e0, 0x17e9}, {0x17f0, 0x17f9}, {0x1800, 0x180e},
-    {0x1810, 0x1819}, {0x1820, 0x1877}, {0x1880, 0x18aa}, {0x18b0, 0x18f5},
-    {0x1900, 0x191c}, {0x1920, 0x192b}, {0x1930, 0x193b}, {0x1944, 0x196d},
-    {0x1970, 0x1974}, {0x1980, 0x19ab}, {0x19b0, 0x19c9}, {0x19d0, 0x19da},
-    {0x19de, 0x1a1b}, {0x1a1e, 0x1a5e}, {0x1a60, 0x1a7c}, {0x1a7f, 0x1a89},
-    {0x1a90, 0x1a99}, {0x1aa0, 0x1aad}, {0x1b00, 0x1b4b}, {0x1b50, 0x1b7c},
-    {0x1b80, 0x1baa}, {0x1bae, 0x1bb9}, {0x1bc0, 0x1bf3}, {0x1bfc, 0x1c37},
-    {0x1c3b, 0x1c49}, {0x1c4d, 0x1c7f}, {0x1cd0, 0x1cf2}, {0x1d00, 0x1de6},
-    {0x1dfc, 0x1f15}, {0x1f18, 0x1f1d}, {0x1f20, 0x1f45}, {0x1f48, 0x1f4d},
-    {0x1f50, 0x1f57}, {0x1f5f, 0x1f7d}, {0x1f80, 0x1fb4}, {0x1fb6, 0x1fc4},
-    {0x1fc6, 0x1fd3}, {0x1fd6, 0x1fdb}, {0x1fdd, 0x1fef}, {0x1ff2, 0x1ff4},
-    {0x1ff6, 0x1ffe}, {0x2000, 0x200a}, {0x2010, 0x2029}, {0x202f, 0x205f},
+    {0x135d, 0x137c}, {0x1380, 0x1399}, {0x13a0, 0x13f4}, {0x1400, 0x167f},
+    {0x1681, 0x169c}, {0x16a0, 0x16f0}, {0x1700, 0x170c}, {0x170e, 0x1714},
+    {0x1720, 0x1736}, {0x1740, 0x1753}, {0x1760, 0x176c}, {0x176e, 0x1770},
+    {0x1780, 0x17b3}, {0x17b6, 0x17dd}, {0x17e0, 0x17e9}, {0x17f0, 0x17f9},
+    {0x1800, 0x180d}, {0x1810, 0x1819}, {0x1820, 0x1877}, {0x1880, 0x18aa},
+    {0x18b0, 0x18f5}, {0x1900, 0x191c}, {0x1920, 0x192b}, {0x1930, 0x193b},
+    {0x1944, 0x196d}, {0x1970, 0x1974}, {0x1980, 0x19ab}, {0x19b0, 0x19c9},
+    {0x19d0, 0x19da}, {0x19de, 0x1a1b}, {0x1a1e, 0x1a5e}, {0x1a60, 0x1a7c},
+    {0x1a7f, 0x1a89}, {0x1a90, 0x1a99}, {0x1aa0, 0x1aad}, {0x1b00, 0x1b4b},
+    {0x1b50, 0x1b7c}, {0x1b80, 0x1baa}, {0x1bae, 0x1bb9}, {0x1bc0, 0x1bf3},
+    {0x1bfc, 0x1c37}, {0x1c3b, 0x1c49}, {0x1c4d, 0x1c7f}, {0x1cd0, 0x1cf2},
+    {0x1d00, 0x1de6}, {0x1dfc, 0x1f15}, {0x1f18, 0x1f1d}, {0x1f20, 0x1f45},
+    {0x1f48, 0x1f4d}, {0x1f50, 0x1f57}, {0x1f5f, 0x1f7d}, {0x1f80, 0x1fb4},
+    {0x1fb6, 0x1fc4}, {0x1fc6, 0x1fd3}, {0x1fd6, 0x1fdb}, {0x1fdd, 0x1fef},
+    {0x1ff2, 0x1ff4}, {0x1ff6, 0x1ffe}, {0x2010, 0x2027}, {0x2030, 0x205e},
     {0x2074, 0x208e}, {0x2090, 0x209c}, {0x20a0, 0x20b9}, {0x20d0, 0x20f0},
     {0x2100, 0x2189}, {0x2190, 0x23f3}, {0x2400, 0x2426}, {0x2440, 0x244a},
     {0x2460, 0x26ff}, {0x2701, 0x27ca}, {0x27ce, 0x2b4c}, {0x2b50, 0x2b59},
@@ -535,7 +552,7 @@ static CONST crange graphRangeTable[] = {
     {0x2d30, 0x2d65}, {0x2d7f, 0x2d96}, {0x2da0, 0x2da6}, {0x2da8, 0x2dae},
     {0x2db0, 0x2db6}, {0x2db8, 0x2dbe}, {0x2dc0, 0x2dc6}, {0x2dc8, 0x2dce},
     {0x2dd0, 0x2dd6}, {0x2dd8, 0x2dde}, {0x2de0, 0x2e31}, {0x2e80, 0x2e99},
-    {0x2e9b, 0x2ef3}, {0x2f00, 0x2fd5}, {0x2ff0, 0x2ffb}, {0x3000, 0x303f},
+    {0x2e9b, 0x2ef3}, {0x2f00, 0x2fd5}, {0x2ff0, 0x2ffb}, {0x3001, 0x303f},
     {0x3041, 0x3096}, {0x3099, 0x30ff}, {0x3105, 0x312d}, {0x3131, 0x318e},
     {0x3190, 0x31ba}, {0x31c0, 0x31e3}, {0x31f0, 0x321e}, {0x3220, 0x32fe},
     {0x3300, 0x4db5}, {0x4dc0, 0x9fcb}, {0xa000, 0xa48c}, {0xa490, 0xa4c6},
@@ -582,43 +599,6 @@ static CONST chr graphCharTable[] = {
 #define	CH	NOCELT
 
 /*
- - nmcces - how many distinct MCCEs are there?
- ^ static int nmcces(struct vars *);
- */
-static int
-nmcces(v)
-    struct vars *v;			/* context */
-{
-    /*
-     * No multi-character collating elements defined at the moment.
-     */
-    return 0;
-}
-
-/*
- - nleaders - how many chrs can be first chrs of MCCEs?
- ^ static int nleaders(struct vars *);
- */
-static int
-nleaders(v)
-    struct vars *v;			/* context */
-{
-    return 0;
-}
-
-/*
- - allmcces - return a cvec with all the MCCEs of the locale
- ^ static struct cvec *allmcces(struct vars *, struct cvec *);
- */
-static struct cvec *
-allmcces(v, cv)
-    struct vars *v;			/* context */
-    struct cvec *cv;			/* this is supposed to have enough room */
-{
-    return clearcvec(cv);
-}
-
-/*
  - element - map collating-element name to celt
  ^ static celt element(struct vars *, CONST chr *, CONST chr *);
  */
@@ -690,7 +670,7 @@ range(v, a, b, cases)
     }
 
     if (!cases) {		/* easy version */
-	cv = getcvec(v, 0, 1, 0);
+	cv = getcvec(v, 0, 1);
 	NOERRN();
 	addrange(cv, a, b);
 	return cv;
@@ -704,7 +684,7 @@ range(v, a, b, cases)
 
     nchrs = (b - a + 1)*2 + 4;
 
-    cv = getcvec(v, nchrs, 0, 0);
+    cv = getcvec(v, nchrs, 0);
     NOERRN();
 
     for (c=a; c<=b; c++) {
@@ -759,7 +739,7 @@ eclass(v, c, cases)
      */
 
     if ((v->cflags&REG_FAKE) && c == 'x') {
-	cv = getcvec(v, 4, 0, 0);
+	cv = getcvec(v, 4, 0);
 	addchr(cv, (chr)'x');
 	addchr(cv, (chr)'y');
 	if (cases) {
@@ -776,7 +756,7 @@ eclass(v, c, cases)
     if (cases) {
 	return allcases(v, c);
     }
-    cv = getcvec(v, 1, 0, 0);
+    cv = getcvec(v, 1, 0);
     assert(cv != NULL);
     addchr(cv, (chr)c);
     return cv;
@@ -825,15 +805,6 @@ cclass(v, startp, endp, cases)
     np = Tcl_UniCharToUtfDString(startp, (int)len, &ds);
 
     /*
-     * Remap lower and upper to alpha if the match is case insensitive.
-     */
-
-    if (cases && len == 5 && (strncmp("lower", np, 5) == 0
-	    || strncmp("upper", np, 5) == 0)) {
-	np = "alpha";
-    }
-
-    /*
      * Map the name to the corresponding enumerated value.
      */
 
@@ -851,13 +822,20 @@ cclass(v, startp, endp, cases)
     }
 
     /*
+     * Remap lower and upper to alpha if the match is case insensitive.
+     */
+
+    if (cases && ((index == CC_LOWER) || (index == CC_UPPER))) {
+	index = CC_ALNUM;
+    }
+
+    /*
      * Now compute the character class contents.
      */
 
     switch((enum classes) index) {
-    case CC_PRINT:
     case CC_ALNUM:
-	cv = getcvec(v, NUM_ALPHA_CHAR, NUM_DIGIT_RANGE + NUM_ALPHA_RANGE, 0);
+	cv = getcvec(v, NUM_ALPHA_CHAR, NUM_DIGIT_RANGE + NUM_ALPHA_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_ALPHA_CHAR ; i++) {
 		addchr(cv, alphaCharTable[i]);
@@ -873,7 +851,7 @@ cclass(v, startp, endp, cases)
 	}
 	break;
     case CC_ALPHA:
-	cv = getcvec(v, NUM_ALPHA_CHAR, NUM_ALPHA_RANGE, 0);
+	cv = getcvec(v, NUM_ALPHA_CHAR, NUM_ALPHA_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_ALPHA_RANGE ; i++) {
 		addrange(cv, alphaRangeTable[i].start,
@@ -885,23 +863,30 @@ cclass(v, startp, endp, cases)
 	}
 	break;
     case CC_ASCII:
-	cv = getcvec(v, 0, 1, 0);
+	cv = getcvec(v, 0, 1);
 	if (cv) {
 	    addrange(cv, 0, 0x7f);
 	}
 	break;
     case CC_BLANK:
-	cv = getcvec(v, 2, 0, 0);
+	cv = getcvec(v, 2, 0);
 	addchr(cv, '\t');
 	addchr(cv, ' ');
 	break;
     case CC_CNTRL:
-	cv = getcvec(v, 0, 2, 0);
-	addrange(cv, 0x0, 0x1f);
-	addrange(cv, 0x7f, 0x9f);
+	cv = getcvec(v, NUM_CONTROL_CHAR, NUM_CONTROL_RANGE);
+	if (cv) {
+	    for (i=0 ; (size_t)i<NUM_CONTROL_RANGE ; i++) {
+		addrange(cv, controlRangeTable[i].start,
+			controlRangeTable[i].end);
+	    }
+	    for (i=0 ; (size_t)i<NUM_CONTROL_CHAR ; i++) {
+		addchr(cv, controlCharTable[i]);
+	    }
+	}
 	break;
     case CC_DIGIT:
-	cv = getcvec(v, 0, NUM_DIGIT_RANGE, 0);
+	cv = getcvec(v, 0, NUM_DIGIT_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_DIGIT_RANGE ; i++) {
 		addrange(cv, digitRangeTable[i].start,
@@ -910,7 +895,7 @@ cclass(v, startp, endp, cases)
 	}
 	break;
     case CC_PUNCT:
-	cv = getcvec(v, NUM_PUNCT_CHAR, NUM_PUNCT_RANGE, 0);
+	cv = getcvec(v, NUM_PUNCT_CHAR, NUM_PUNCT_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_PUNCT_RANGE ; i++) {
 		addrange(cv, punctRangeTable[i].start,
@@ -930,7 +915,8 @@ cclass(v, startp, endp, cases)
 	 * Whatever is actually the case, the number of ranges is fixed (until
 	 * someone comes up with a better arrangement!)
 	 */
-	cv = getcvec(v, 0, 3, 0);
+
+	cv = getcvec(v, 0, 3);
 	if (cv) {
 	    addrange(cv, '0', '9');
 	    addrange(cv, 'a', 'f');
@@ -938,7 +924,7 @@ cclass(v, startp, endp, cases)
 	}
 	break;
     case CC_SPACE:
-	cv = getcvec(v, NUM_SPACE_CHAR, NUM_SPACE_RANGE, 0);
+	cv = getcvec(v, NUM_SPACE_CHAR, NUM_SPACE_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_SPACE_RANGE ; i++) {
 		addrange(cv, spaceRangeTable[i].start,
@@ -950,7 +936,7 @@ cclass(v, startp, endp, cases)
 	}
 	break;
     case CC_LOWER:
-	cv  = getcvec(v, NUM_LOWER_CHAR, NUM_LOWER_RANGE, 0);
+	cv  = getcvec(v, NUM_LOWER_CHAR, NUM_LOWER_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_LOWER_RANGE ; i++) {
 		addrange(cv, lowerRangeTable[i].start,
@@ -962,7 +948,7 @@ cclass(v, startp, endp, cases)
 	}
 	break;
     case CC_UPPER:
-	cv  = getcvec(v, NUM_UPPER_CHAR, NUM_UPPER_RANGE, 0);
+	cv  = getcvec(v, NUM_UPPER_CHAR, NUM_UPPER_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_UPPER_RANGE ; i++) {
 		addrange(cv, upperRangeTable[i].start,
@@ -973,8 +959,27 @@ cclass(v, startp, endp, cases)
 	    }
 	}
 	break;
+    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++) {
+    		addrange(cv, spaceRangeTable[i].start,
+    				spaceRangeTable[i].end);
+    	    }
+    	    for (i=0 ; (size_t)i<NUM_SPACE_CHAR ; i++) {
+    		addchr(cv, spaceCharTable[i]);
+    	    }
+    	    for (i=0 ; (size_t)i<NUM_GRAPH_RANGE ; i++) {
+    		addrange(cv, graphRangeTable[i].start,
+    				graphRangeTable[i].end);
+    	    }
+    	    for (i=0 ; (size_t)i<NUM_GRAPH_CHAR ; i++) {
+    		addchr(cv, graphCharTable[i]);
+    	    }
+    	}
+    	break;
     case CC_GRAPH:
-	cv  = getcvec(v, NUM_GRAPH_CHAR, NUM_GRAPH_RANGE, 0);
+	cv  = getcvec(v, NUM_GRAPH_CHAR, NUM_GRAPH_RANGE);
 	if (cv) {
 	    for (i=0 ; (size_t)i<NUM_GRAPH_RANGE ; i++) {
 		addrange(cv, graphRangeTable[i].start,
@@ -1012,10 +1017,10 @@ allcases(v, pc)
     tc = Tcl_UniCharToTitle((chr)c);
 
     if (tc != uc) {
-	cv = getcvec(v, 3, 0, 0);
+	cv = getcvec(v, 3, 0);
 	addchr(cv, tc);
     } else {
-	cv = getcvec(v, 2, 0, 0);
+	cv = getcvec(v, 2, 0);
     }
     addchr(cv, lc);
     if (lc != uc) {