Misc

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

1 files changed, 74 insertions, 2 deletions

diff --git a/libcharconv_overlaid.c b/libcharconv_overlaid.c
index 05cf0d3..5b738b9 100644
--- a/libcharconv_overlaid.c
+++ b/libcharconv_overlaid.c
@@ -55,14 +55,26 @@ static struct {
 	{UINT32_C(0x23CA), UINT32_C(0x23C6), UINT32_C(0x23C8)},
 	{(uint_least32_t)'-', UINT32_C(0x238F), UINT32_C(0x2390)},
 	{(uint_least32_t)'-', UINT32_C(0x2391), UINT32_C(0x2392)},
-	{(uint_least32_t)'-', (uint_least32_t)'~', UINT32_C(0x23E6)}
+	{(uint_least32_t)'-', (uint_least32_t)'~', UINT32_C(0x23E6)},
+	{UINT32_C(0x1681), UINT32_C(0x1686), UINT32_C(0x168B)},
+	{UINT32_C(0x1682), UINT32_C(0x1687), UINT32_C(0x168C)},
+	{UINT32_C(0x1683), UINT32_C(0x1688), UINT32_C(0x168D)},
+	{UINT32_C(0x1684), UINT32_C(0x1689), UINT32_C(0x168E)},
+	{UINT32_C(0x1685), UINT32_C(0x168A), UINT32_C(0x168F)},
+	{UINT32_C(0x2571), UINT32_C(0x2572), UINT32_C(0x2573)},
+	{(uint_least32_t)'S', (uint_least32_t)'S', UINT32_C(0x00A7)},
+	{(uint_least32_t)'s', (uint_least32_t)'s', UINT32_C(0x00A7)},
+	{UINT32_C(0x2E2C), (uint_least32_t)'+', UINT32_C(0x205C)},
+	{UINT32_C(0x2058), (uint_least32_t)'x', UINT32_C(0x203B)},
+	{UINT32_C(0x2058), (uint_least32_t)'X', UINT32_C(0x203B)},
+	{(uint_least32_t)'>', (uint_least32_t)':', UINT32_C(0x2E16)}
 };
 
 
 enum libcharconv_result
 libcharconv_overlaid(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
 {
-	uint_least32_t a, b;
+	uint_least32_t a, b, c;
 	size_t i, alen, blen;
 	*n = 0;
 	while (slen) {
@@ -79,6 +91,44 @@ libcharconv_overlaid(const char *s, size_t slen, size_t *n, uint_least32_t *cp,
 			continue;
 		}
 
+		if (UINT32_C(0x1FBA0) <= a && a <= UINT32_C(0x1FBAE)) {
+			if (*n)
+				goto no_conv;
+			if (slen == alen)
+				return LIBCHARCONV_INDETERMINATE;
+			blen = libcharconv_decode_utf8_(&s[alen], slen - alen, &b);
+			if (blen > slen)
+				return LIBCHARCONV_INDETERMINATE;
+			if (!blen)
+				goto no_conv;
+			if (UINT32_C(0x1FBA0) > b || b > UINT32_C(0x1FBAE))
+				goto no_conv_consume;
+			a = (uint_least32_t)"\x01\x02\x04\x08\x05\x0A\x0C\x03\x09\x06\x0E\x0D\x0B\x07\x0F"[a & 0xF];
+			b = (uint_least32_t)"\x01\x02\x04\x08\x05\x0A\x0C\x03\x09\x06\x0E\x0D\x0B\x07\x0F"[b & 0xF];
+			c = (uint_least32_t)" \x00\x01\x07\x02\x04\x09\x0D\x03\x08\x05\x0C\x06\x0B\x0A\x0E"[a | b];
+			a = c |= UINT32_C(0x1FBA0);
+			alen += blen;
+			for (;;) {
+				if (slen == alen)
+					goto conv_if_end_calc;
+				blen = libcharconv_decode_utf8_(&s[alen], slen - alen, &b);
+				if (blen > slen)
+					return LIBCHARCONV_INDETERMINATE;
+				if (!blen)
+					goto conv_calc;
+				if (UINT32_C(0x1FBA0) > b || b > UINT32_C(0x1FBAE))
+					goto conv_calc;
+				a &= 0xF;
+				b &= 0xF;
+				a = (uint_least32_t)"\x01\x02\x04\x08\x05\x0A\x0C\x03\x09\x06\x0E\x0D\x0B\x07\x0F"[a & 0xF];
+				b = (uint_least32_t)"\x01\x02\x04\x08\x05\x0A\x0C\x03\x09\x06\x0E\x0D\x0B\x07\x0F"[b & 0xF];
+				c = (uint_least32_t)" \x00\x01\x07\x02\x04\x09\x0D\x03\x08\x05\x0C\x06\x0B\x0A\x0E"[a | b];
+				a = c |= UINT32_C(0x1FBA0);
+				alen += blen;
+			}
+			goto conv_calc;
+		}
+
 		for (i = 0u; i < sizeof(pairs) / sizeof(*pairs); i++) {
 			if (a != pairs[i].a && a != pairs[i].b)
 				continue;
@@ -102,6 +152,10 @@ libcharconv_overlaid(const char *s, size_t slen, size_t *n, uint_least32_t *cp,
 no_conv:
 	return LIBCHARCONV_NO_CONVERT;
 
+no_conv_consume:
+	*n += alen;
+	return LIBCHARCONV_NO_CONVERT;
+
 conv:
 	if (*n)
 		goto no_conv;
@@ -110,4 +164,22 @@ conv:
 	*n += alen + blen;
 	*ncp = 1u;
 	return LIBCHARCONV_CONVERTED;
+
+conv_calc:
+	if (*n)
+		goto no_conv;
+	if (*ncp)
+		*cp = c;
+	*n += alen;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERTED;
+
+conv_if_end_calc:
+	if (*n)
+		goto no_conv;
+	if (*ncp)
+		*cp = c;
+	*n += alen;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERT_IF_END;
 }