Misc stuff

Signed-off-by: Mattias Andrée <m@maandree.se>

1 files changed, 216 insertions, 101 deletions

diff --git a/libcharconv_latin.c b/libcharconv_latin.c
index 374a097..20e3fd9 100644
--- a/libcharconv_latin.c
+++ b/libcharconv_latin.c
@@ -220,6 +220,38 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			c -= (uint_least32_t)UINT32_C(0x1D4EA) - (uint_least32_t)'a';
 			goto conv;
 
+		} else if (UINT32_C(0x1743) <= c && c <= UINT32_C(0x1751)) {
+			/* buhid */
+			c1 = "kgGtdnpbmyrlwsh"[c - UINT32_C(0x1743)];
+			if (*n)
+				goto no_conv;
+			c2 = 'a';
+			s = &s[clen];
+			*n += clen;
+			if (!slen) {
+				ret = LIBCHARCONV_CONVERT_IF_END;
+				goto budih_conv;
+			}
+			clen = libcharconv_decode_utf8_(s, slen, &c);
+			if (clen > slen)
+				return LIBCHARCONV_INDETERMINATE;
+			if (!clen)
+				goto budih_conv;
+			switch (c) {
+			case UINT32_C(0x1752): c2 = 'i'; *n += clen; break;
+			case UINT32_C(0x1753): c2 = 'u'; *n += clen; break;
+			default:
+				break;
+			}
+		budih_conv:
+			if (c1 == 'G') {
+				c3 = c2;
+				c2 = 'g';
+				c1 = 'n';
+				goto conv3_prechecked;
+			}
+			goto conv2_prechecked;
+
 		} else if (UINT32_C(0x1F110) <= c && c <= UINT32_C(0x1F129)) {
 			/* bracketed (parenthesised capital) */
 			c -= (uint_least32_t)UINT32_C(0x1F110) - (uint_least32_t)'A';
@@ -627,127 +659,180 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			c2 = (char)c;
 			goto conv2;
 
+		} else if (UINT32_C(0x10330) <= c && c <= UINT32_C(0x10349)) {
+			/* gothic */
+			if (c == UINT32_C(0x10341))
+			    goto use_switch;
+			c = (uint_least32_t)"abgdeqzhTiklmnjup rstwfxvo"[c - UINT32_C(0x10330)];
+			goto conv;
+
+		} else if (UINT32_C(0x0840) <= c && c <= UINT32_C(0x085B)) {
+			/* mandaic */
+			c = (uint_least32_t)"abgdhuzHTiklmnsepZqrStDKo\'*:"[c - UINT32_C(0x0840)];
+			goto conv;
+
+		} else if (UINT32_C(0x10E60) <= c && c <= UINT32_C(0x10E68)) {
+			/* rumi */
+			c -= (uint_least32_t)UINT32_C(0x10E60) - (uint_least32_t)'1';
+			c1 = (char)c;
+			goto conv1;
+		} else if (UINT32_C(0x10E69) <= c && c <= UINT32_C(0x10E71)) {
+			/* rumi */
+			c -= (uint_least32_t)UINT32_C(0x10E69) - (uint_least32_t)'1';
+			c1 = (char)c;
+			c2 = '0';
+			goto conv2;
+		} else if (UINT32_C(0x10E72) <= c && c <= UINT32_C(0x10E7A)) {
+			/* rumi */
+			c -= (uint_least32_t)UINT32_C(0x10E72) - (uint_least32_t)'1';
+			c1 = (char)c;
+			c2 = '0';
+			c3 = '0';
+			goto conv3;
+
+		} else if (UINT32_C(0x1A00) <= c && c <= UINT32_C(0x1A1B)) {
+			/* buginese */
+			c1 = "kgGKpbmMtdnRcjYCyrlvsahiueoA"[c - UINT32_C(0x1A00)];
+			if (c >= UINT32_C(0x1A17))
+				goto combining;
+			if (*n)
+				goto no_conv;
+			c2 = 'a';
+			s = &s[clen];
+			*n += clen;
+			if (!slen) {
+				ret = LIBCHARCONV_CONVERT_IF_END;
+				goto buginese_conv;
+			}
+			clen = libcharconv_decode_utf8_(s, slen, &c);
+			if (clen > slen)
+				return LIBCHARCONV_INDETERMINATE;
+			if (!clen)
+				goto buginese_conv;
+			switch (c) {
+			case UINT32_C(0x1A17): c2 = 'i'; *n += clen; break;
+			case UINT32_C(0x1A18): c2 = 'u'; *n += clen; break;
+			case UINT32_C(0x1A19): c2 = 'e'; *n += clen; break;
+			case UINT32_C(0x1A1A): c2 = 'o'; *n += clen; break;
+			case UINT32_C(0x1A1B): c2 = 'A'; *n += clen; break;
+			default:
+				break;
+			}
+		buginese_conv:
+			if (c1 == 'a') {
+				c = (uint_least32_t)c2;
+				goto conv_prechecked;
+			}
+			goto conv2_prechecked;
+
+		} else if (UINT32_C(0x11EE0) <= c && c <= UINT32_C(0x11EF6)) {
+			/* makasar */
+			c1 = "kgGpbmtdncjYyrlvsa-iueo"[c - UINT32_C(0x11EE0)];
+			if (c >= UINT32_C(0x11EF3))
+				goto combining;
+			if (*n)
+				goto no_conv;
+			c2 = 'a';
+			s = &s[clen];
+			*n += clen;
+			if (!slen) {
+				ret = LIBCHARCONV_CONVERT_IF_END;
+				goto makasar_conv;
+			}
+			clen = libcharconv_decode_utf8_(s, slen, &c);
+			if (clen > slen)
+				return LIBCHARCONV_INDETERMINATE;
+			if (!clen)
+				goto makasar_conv;
+			switch (c) {
+			case UINT32_C(0x11EF3): c2 = 'i'; *n += clen; break;
+			case UINT32_C(0x11EF4): c2 = 'u'; *n += clen; break;
+			case UINT32_C(0x11EF5): c2 = 'e'; *n += clen; break;
+			case UINT32_C(0x11EF6): c2 = 'o'; *n += clen; break;
+			default:
+				break;
+			}
+		makasar_conv:
+			if (c1 == 'a') {
+				c = (uint_least32_t)c2;
+				goto conv_prechecked;
+			}
+			goto conv2_prechecked;
+
 		} else {
 		use_switch:
 			switch (c) {
 			/* shogi */
-			case UINT32_C(0x2616): c = (uint_least32_t)'w'; goto conv;
-			case UINT32_C(0x2617): c = (uint_least32_t)'b'; goto conv;
-			case UINT32_C(0x26C9): c = (uint_least32_t)'W'; goto conv;
-			case UINT32_C(0x26CA): c = (uint_least32_t)'B'; goto conv;
+			case UINT32_C(0x2616): c1 = 'w'; goto conv1;
+			case UINT32_C(0x2617): c1 = 'b'; goto conv1;
+			case UINT32_C(0x26C9): c1 = 'W'; goto conv1;
+			case UINT32_C(0x26CA): c1 = 'B'; goto conv1;
 
 			/* go (common) */
-			case UINT32_C(0x25CF): c = (uint_least32_t)'b'; goto conv;
-			case UINT32_C(0x25CB): c = (uint_least32_t)'w'; goto conv;
+			case UINT32_C(0x25CB): c1 = 'w'; goto conv1;
+			case UINT32_C(0x25CF): c1 = 'b'; goto conv1;
 
 			/* go (white) */
-			case UINT32_C(0x2686): c = (uint_least32_t)'1'; goto conv;
-			case UINT32_C(0x2687): c = (uint_least32_t)'2'; goto conv;
+			case UINT32_C(0x2686): c1 = '1'; goto conv1;
+			case UINT32_C(0x2687): c1 = '2'; goto conv1;
 
 			/* go (black) */
-			case UINT32_C(0x2688): c = (uint_least32_t)'1'; goto conv;
-			case UINT32_C(0x2689): c = (uint_least32_t)'2'; goto conv;
+			case UINT32_C(0x2688): c1 = '1'; goto conv1;
+			case UINT32_C(0x2689): c1 = '2'; goto conv1;
 
 			/* draughts */
-			case UINT32_C(0x26C0): c = (uint_least32_t)'m'; goto conv;
-			case UINT32_C(0x26C1): c = (uint_least32_t)'k'; goto conv;
-			case UINT32_C(0x26C2): c = (uint_least32_t)'M'; goto conv;
-			case UINT32_C(0x26C3): c = (uint_least32_t)'K'; goto conv;
+			case UINT32_C(0x26C0): c1 = 'm'; goto conv1;
+			case UINT32_C(0x26C1): c1 = 'k'; goto conv1;
+			case UINT32_C(0x26C2): c1 = 'M'; goto conv1;
+			case UINT32_C(0x26C3): c1 = 'K'; goto conv1;
 
 			/* gender symbols */
-			case UINT32_C(0x2640): c = (uint_least32_t)'f'; goto conv;
-			case UINT32_C(0x2642): c = (uint_least32_t)'m'; goto conv;
-			case UINT32_C(0x263F): c = (uint_least32_t)'i'; goto conv;
+			case UINT32_C(0x2640): c1 = 'f'; goto conv1;
+			case UINT32_C(0x2642): c1 = 'm'; goto conv1;
+			case UINT32_C(0x263F): c1 = 'i'; goto conv1;
 
 			/* double-struck */
-			case UINT32_C(0x2102): c = (uint_least32_t)'C'; goto conv;
-			case UINT32_C(0x210D): c = (uint_least32_t)'H'; goto conv;
-			case UINT32_C(0x2115): c = (uint_least32_t)'N'; goto conv;
-			case UINT32_C(0x2119): c = (uint_least32_t)'P'; goto conv;
-			case UINT32_C(0x211A): c = (uint_least32_t)'Q'; goto conv;
-			case UINT32_C(0x211D): c = (uint_least32_t)'R'; goto conv;
-			case UINT32_C(0x2124): c = (uint_least32_t)'Z'; goto conv;
+			case UINT32_C(0x2102): c1 = 'C'; goto conv1;
+			case UINT32_C(0x210D): c1 = 'H'; goto conv1;
+			case UINT32_C(0x2115): c1 = 'N'; goto conv1;
+			case UINT32_C(0x2119): c1 = 'P'; goto conv1;
+			case UINT32_C(0x211A): c1 = 'Q'; goto conv1;
+			case UINT32_C(0x211D): c1 = 'R'; goto conv1;
+			case UINT32_C(0x2124): c1 = 'Z'; goto conv1;
 
 			/* double-struck italic */
-			case UINT32_C(0x2145): c = (uint_least32_t)'D'; goto conv;
-			case UINT32_C(0x2146): c = (uint_least32_t)'d'; goto conv;
-			case UINT32_C(0x2147): c = (uint_least32_t)'e'; goto conv;
-			case UINT32_C(0x2148): c = (uint_least32_t)'i'; goto conv;
-			case UINT32_C(0x2149): c = (uint_least32_t)'j'; goto conv;
+			case UINT32_C(0x2145): c1 = 'D'; goto conv1;
+			case UINT32_C(0x2146): c1 = 'd'; goto conv1;
+			case UINT32_C(0x2147): c1 = 'e'; goto conv1;
+			case UINT32_C(0x2148): c1 = 'i'; goto conv1;
+			case UINT32_C(0x2149): c1 = 'j'; goto conv1;
 
 			/* fraktur */
-			case UINT32_C(0x212D): c = (uint_least32_t)'C'; goto conv;
-			case UINT32_C(0x210C): c = (uint_least32_t)'H'; goto conv;
-			case UINT32_C(0x2111): c = (uint_least32_t)'I'; goto conv;
-			case UINT32_C(0x211C): c = (uint_least32_t)'R'; goto conv;
-			case UINT32_C(0x2128): c = (uint_least32_t)'Z'; goto conv;
+			case UINT32_C(0x212D): c1 = 'C'; goto conv1;
+			case UINT32_C(0x210C): c1 = 'H'; goto conv1;
+			case UINT32_C(0x2111): c1 = 'I'; goto conv1;
+			case UINT32_C(0x211C): c1 = 'R'; goto conv1;
+			case UINT32_C(0x2128): c1 = 'Z'; goto conv1;
 
 			/* script */
-			case UINT32_C(0x212C): c = (uint_least32_t)'B'; goto conv;
-			case UINT32_C(0x2130): c = (uint_least32_t)'E'; goto conv;
-			case UINT32_C(0x2131): c = (uint_least32_t)'F'; goto conv;
-			case UINT32_C(0x210B): c = (uint_least32_t)'H'; goto conv;
-			case UINT32_C(0x2110): c = (uint_least32_t)'I'; goto conv;
-			case UINT32_C(0x2112): c = (uint_least32_t)'L'; goto conv;
-			case UINT32_C(0x2133): c = (uint_least32_t)'M'; goto conv;
-			case UINT32_C(0x211B): c = (uint_least32_t)'R'; goto conv;
-			case UINT32_C(0x212F): c = (uint_least32_t)'e'; goto conv;
-			case UINT32_C(0x210A): c = (uint_least32_t)'g'; goto conv;
-			case UINT32_C(0x2134): c = (uint_least32_t)'o'; goto conv;
+			case UINT32_C(0x212C): c1 = 'B'; goto conv1;
+			case UINT32_C(0x2130): c1 = 'E'; goto conv1;
+			case UINT32_C(0x2131): c1 = 'F'; goto conv1;
+			case UINT32_C(0x210B): c1 = 'H'; goto conv1;
+			case UINT32_C(0x2110): c1 = 'I'; goto conv1;
+			case UINT32_C(0x2112): c1 = 'L'; goto conv1;
+			case UINT32_C(0x2133): c1 = 'M'; goto conv1;
+			case UINT32_C(0x211B): c1 = 'R'; goto conv1;
+			case UINT32_C(0x212F): c1 = 'e'; goto conv1;
+			case UINT32_C(0x210A): c1 = 'g'; goto conv1;
+			case UINT32_C(0x2134): c1 = 'o'; goto conv1;
 
 			/* buhid */
-			case UINT32_C(0x1740): c = (uint_least32_t)'a'; goto conv;
-			case UINT32_C(0x1741): c = (uint_least32_t)'i'; goto conv;
-			case UINT32_C(0x1742): c = (uint_least32_t)'u'; goto conv;
-			case UINT32_C(0x1752): c2 = 'i'; goto budih_combining;
-			case UINT32_C(0x1753): c2 = 'u'; goto budih_combining;
-			budih_combining:
-				c1 = '^';
-				goto conv2;
-			case UINT32_C(0x174A): c1 = 'b'; goto budih;
-			case UINT32_C(0x1747): c1 = 'd'; goto budih;
-			case UINT32_C(0x1744): c1 = 'g'; goto budih;
-			case UINT32_C(0x1751): c1 = 'h'; goto budih;
-			case UINT32_C(0x1743): c1 = 'k'; goto budih;
-			case UINT32_C(0x174E): c1 = 'l'; goto budih;
-			case UINT32_C(0x174B): c1 = 'm'; goto budih;
-			case UINT32_C(0x1748): c1 = 'n'; goto budih;
-			case UINT32_C(0x1749): c1 = 'p'; goto budih;
-			case UINT32_C(0x174D): c1 = 'r'; goto budih;
-			case UINT32_C(0x1750): c1 = 's'; goto budih;
-			case UINT32_C(0x1746): c1 = 't'; goto budih;
-			case UINT32_C(0x174C): c1 = 'y'; goto budih;
-			case UINT32_C(0x174F): c1 = 'w'; goto budih;
-			case UINT32_C(0x1745): c1 = '-'; goto budih;
-			budih:
-				if (*n)
-					goto no_conv;
-				c2 = 'a';
-				s = &s[clen];
-				*n += clen;
-				if (!slen) {
-					ret = LIBCHARCONV_CONVERT_IF_END;
-					goto budih_conv;
-				}
-				clen = libcharconv_decode_utf8_(s, slen, &c);
-				if (clen > slen)
-					return LIBCHARCONV_INDETERMINATE;
-				if (!clen)
-					goto budih_conv;
-				switch (c) {
-				case UINT32_C(0x1752): c2 = 'i'; *n += clen; break;
-				case UINT32_C(0x1753): c2 = 'u'; *n += clen; break;
-				default:
-					break;
-				}
-			budih_conv:
-				if (c1 == '-') {
-					c3 = c2;
-					c2 = 'g';
-					c1 = 'n';
-					goto conv3_prechecked;
-				}
-				goto conv2_prechecked;
+			case UINT32_C(0x1740): c1 = 'a'; goto conv1;
+			case UINT32_C(0x1741): c1 = 'i'; goto conv1;
+			case UINT32_C(0x1742): c1 = 'u'; goto conv1;
+			case UINT32_C(0x1752): c1 = 'i'; goto combining;
+			case UINT32_C(0x1753): c1 = 'u'; goto combining;
 
 			/* replacement */
 			case UINT32_C(0xFFFC):
@@ -1038,10 +1123,10 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			case UINT32_C(0x00AD):  c1 = 'S'; c2 = 'H'; c3 = 'Y'; goto conv3;
 
 			/* invisible */
-			case UINT32_C(0x2061): c = (uint_least32_t)'('; goto conv;
-			case UINT32_C(0x2062): c = (uint_least32_t)'*'; goto conv;
-			case UINT32_C(0x2063): c = (uint_least32_t)'|'; goto conv;
-			case UINT32_C(0x2064): c = (uint_least32_t)'+'; goto conv;
+			case UINT32_C(0x2061): c1 = '('; goto conv1;
+			case UINT32_C(0x2062): c1 = '*'; goto conv1;
+			case UINT32_C(0x2063): c1 = '|'; goto conv1;
+			case UINT32_C(0x2064): c1 = '+'; goto conv1;
 
 			/* enclosed (positive) */
 			case UINT32_C(0x24EA): c1 = '('; c2 = '0'; c3 = ')'; goto conv3;
@@ -1097,6 +1182,32 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			case UINT32_C(0x2051): c1 = '*'; c2 = '*'; goto conv2;
 			case UINT32_C(0x2E49): c1 = ','; c2 = ','; goto conv2;
 
+			/* gothic */
+			case UINT32_C(0x10341): c1 = '9'; c2 = '0'; goto conv2;
+			case UINT32_C(0x1034A): c1 = '9'; c2 = '0'; c3 = '0'; goto conv3;
+
+			/* mandaic */
+			case UINT32_C(0x085E): c1 = '.'; goto conv1;
+
+			/* rumi */
+			case UINT32_C(0x10E7B): c1 = '1'; c2 = '/'; c3 = '2'; goto conv3;
+			case UINT32_C(0x10E7C): c1 = '1'; c2 = '/'; c3 = '4'; goto conv3;
+			case UINT32_C(0x10E7D): c1 = '1'; c2 = '/'; c3 = '3'; goto conv3;
+			case UINT32_C(0x10E7E): c1 = '2'; c2 = '/'; c3 = '3'; goto conv3;
+
+			/* scores */
+			case UINT32_C(0x1F4AE): c1 = '*'; goto conv1;
+			case UINT32_C(0x1F51F): c1 = '1'; c2 = '0'; goto conv2;
+			case UINT32_C(0x1F4AF): c1 = '1'; c2 = '0'; c3 = '0'; goto conv3;
+
+			/* buginese */
+			case UINT32_C(0x1A1E): c1 = '.'; goto conv1;
+			case UINT32_C(0x1A1F): c1 = '|'; goto conv1;
+
+			/* makasar */
+			case UINT32_C(0x11EF7): c1 = '.'; goto conv1;
+			case UINT32_C(0x11EF8): c1 = '|'; goto conv1;
+
 			default:
 			no_match:
 				*n += clen;
@@ -1114,11 +1225,15 @@ conv:
 	if (*n)
 		goto no_conv;
 	*n += clen;
+conv_prechecked:
 	if (*ncp)
 		*cp = c;
 	*ncp = 1u;
 	return ret;
 
+combining:
+	c2 = c1;
+	c1 = '^';
 conv2:
 	if (*n)
 		goto no_conv;