Add overlaid

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

5 files changed, 95 insertions, 3 deletions

diff --git a/Makefile b/Makefile
index 55369ae..f09b1a3 100644
--- a/Makefile
+++ b/Makefile
@@ -61,7 +61,8 @@ BIN =\
 	convert-to-yijing-tetragrams\
 	convert-to-yijing-hexagrams\
 	convert-to-vulgar-fractions\
-	convert-to-flipped
+	convert-to-flipped\
+	convert-to-overlaid
 
 LIBOBJ =\
 	libcharconv_decode_utf8_.o\
@@ -111,7 +112,8 @@ LIBOBJ =\
 	libcharconv_yijing_tetragrams.o\
 	libcharconv_yijing_hexagrams.o\
 	libcharconv_vulgar_fractions.o\
-	libcharconv_flipped.o
+	libcharconv_flipped.o\
+	libcharconv_overlaid.o
 
 LOBJ = $(LIBOBJ:.o=.lo)
 
diff --git a/convert-to-overlaid.c b/convert-to-overlaid.c
new file mode 100644
index 0000000..803cdd6
--- /dev/null
+++ b/convert-to-overlaid.c
@@ -0,0 +1,4 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+SIMPLE(libcharconv_overlaid)
diff --git a/libcharconv.h b/libcharconv.h
index 4aa9d27..289f9e8 100644
--- a/libcharconv.h
+++ b/libcharconv.h
@@ -333,6 +333,11 @@ LIBCHARCONV_FUNC_(libcharconv_vulgar_fractions);
  */
 LIBCHARCONV_FUNC_(libcharconv_flipped);
 
+/**
+ * Overlay characters
+ */
+LIBCHARCONV_FUNC_(libcharconv_overlaid);
+
 
 #undef LIBCHARCONV_FUNC_
 #endif
diff --git a/libcharconv_latin.c b/libcharconv_latin.c
index c52044c..ee975e1 100644
--- a/libcharconv_latin.c
+++ b/libcharconv_latin.c
@@ -18,7 +18,7 @@ enum libcharconv_result
 libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
 {
 	enum libcharconv_result ret = LIBCHARCONV_CONVERTED;
-	uint_least32_t c;
+	uint_least32_t c, cp1, cp2;
 	char c1, c2, c3, c4, c5, c6;
 	size_t i, clen;
 	unsigned num;
@@ -626,6 +626,10 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			case UINT32_C(0x215E): c1 = '7'; c2 = '/'; c3 = '8'; goto conv3;
 			case UINT32_C(0x2152): c1 = '1'; c2 = '/'; c3 = '1'; c4 = '0'; goto conv4;
 
+			/* overlaid */
+			case UINT32_C(0x203D): c1 = '!'; c2 = '?'; goto conv2;
+			case UINT32_C(0x2E18): cp1 = UINT32_C(0xBF); cp2 = UINT32_C(0xA1); goto conv2cp;
+
 			default:
 			no_match:
 				*n += clen;
@@ -648,6 +652,17 @@ conv:
 	*ncp = 1u;
 	return ret;
 
+conv2cp:
+	if (*n)
+		goto no_conv;
+	*n += clen;
+	if (*ncp >= 1u)
+		cp[0] = cp1;
+	if (*ncp >= 2u)
+		cp[1] = cp2;
+	*ncp = 2u;
+	return ret;
+
 conv2:
 	if (*n)
 		goto no_conv;
diff --git a/libcharconv_overlaid.c b/libcharconv_overlaid.c
new file mode 100644
index 0000000..7f3cab1
--- /dev/null
+++ b/libcharconv_overlaid.c
@@ -0,0 +1,66 @@
+/* See LICENSE file for copyright and license details. */
+#include "lib-common.h"
+
+
+static struct {
+	uint_least32_t a;
+	uint_least32_t b;
+	uint_least32_t to;
+} pairs[] = {
+	{0x00BF, 0x00A1, 0x2E18},
+	{0x0021, 0x003F, 0x203D}
+};
+
+
+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;
+	size_t i, alen, blen;
+	*n = 0;
+	while (slen) {
+		alen = libcharconv_decode_utf8_(s, slen, &a);
+		if (alen > slen) {
+			if (*n)
+				goto no_conv;
+			return LIBCHARCONV_INDETERMINATE;
+		}
+		if (!alen) {
+			*n += 1u;
+			slen -= 1u;
+			s = &s[1];
+			continue;
+		}
+
+		for (i = 0u; i < sizeof(pairs) / sizeof(*pairs); i++) {
+			if (a != pairs[i].a && a != pairs[i].b)
+				continue;
+			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 ((a ^ b) == (pairs[i].a ^ pairs[i].b))
+				goto conv;
+		}
+
+		*n += alen;
+		s = &s[alen];
+		slen -= alen;
+	}
+no_conv:
+	return LIBCHARCONV_NO_CONVERT;
+
+conv:
+	if (*n)
+		goto no_conv;
+	if (*ncp)
+		*cp = pairs[i].to;
+	*n += alen + blen;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERTED;
+}