Misc

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

1 files changed, 66 insertions, 2 deletions

diff --git a/libcharconv_joined.c b/libcharconv_joined.c
index e8e0a59..e8ab45f 100644
--- a/libcharconv_joined.c
+++ b/libcharconv_joined.c
@@ -11,6 +11,8 @@ static struct {
 	{(uint_least32_t)'?', (uint_least32_t)'?', UINT32_C(0x2047)},
 	{(uint_least32_t)'!', (uint_least32_t)'!', UINT32_C(0x203C)},
 	{(uint_least32_t)'!', (uint_least32_t)'?', UINT32_C(0x2049)},
+	{(uint_least32_t)':', (uint_least32_t)':', UINT32_C(0x2E2C)},
+	{(uint_least32_t)'|', (uint_least32_t)'|', UINT32_C(0x2016)},
 	{UINT32_C(0x23CB), UINT32_C(0x23BE), UINT32_C(0x23C9)},
 	{UINT32_C(0x23CC), UINT32_C(0x23BF), UINT32_C(0x23CA)}
 };
@@ -19,7 +21,7 @@ static struct {
 enum libcharconv_result
 libcharconv_joined(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) {
@@ -36,6 +38,46 @@ libcharconv_joined(const char *s, size_t slen, size_t *n, uint_least32_t *cp, si
 			continue;
 		}
 
+		if (UINT32_C(0x1681) <= a && a <= UINT32_C(0x1694)) {
+			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(0x1681) > b || b > UINT32_C(0x1694))
+				goto no_conv_consume;
+			if ((a - UINT32_C(0x1681)) / 5u != (b - UINT32_C(0x1681)) / 5u)
+				goto no_conv_consume;
+			c = (a - UINT32_C(0x1681)) % 5u + (b - UINT32_C(0x1681)) % 5u + 1u;
+			if (c >= 5u)
+				goto no_conv_consume;
+			a = c += (a - UINT32_C(0x1681)) / 5u * 5u + UINT32_C(0x1681);
+			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(0x1681) > b || b > UINT32_C(0x1694))
+					goto conv_calc;
+				if ((a - UINT32_C(0x1681)) / 5u != (b - UINT32_C(0x1681)) / 5u)
+					goto conv_calc;
+				b = (a - UINT32_C(0x1681)) % 5u + (b - UINT32_C(0x1681)) % 5u + 1u;
+				if (b >= 5u)
+					goto conv_calc;
+				a = c = b + (a - UINT32_C(0x1681)) / 5u * 5u + UINT32_C(0x1681);
+				alen += blen;
+			}
+			goto conv_calc;
+		}
+
 		for (i = 0u; i < sizeof(pairs) / sizeof(*pairs); i++) {
 			if (a != pairs[i].a)
 				continue;
@@ -59,12 +101,34 @@ libcharconv_joined(const char *s, size_t slen, size_t *n, uint_least32_t *cp, si
 no_conv:
 	return LIBCHARCONV_NO_CONVERT;
 
+no_conv_consume:
+	*n += alen;
+	return LIBCHARCONV_NO_CONVERT;
+
 conv:
 	if (*n)
 		goto no_conv;
 	if (*ncp)
-		*cp = pairs[i].to;
+		*cp = c = pairs[i].to;
 	*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;
 }