Add flipping and turning for yijing n-grams, and add invisible

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

7 files changed, 143 insertions, 6 deletions

diff --git a/Makefile b/Makefile
index 293dc8b..d6a2431 100644
--- a/Makefile
+++ b/Makefile
@@ -79,7 +79,8 @@ BIN =\
 	convert-to-rotated-45deg-cw\
 	convert-to-rotated-90deg-cw\
 	convert-to-rotated-45deg-ccw\
-	convert-to-rotated-90deg-ccw
+	convert-to-rotated-90deg-ccw\
+	convert-to-invisible
 
 LIBOBJ =\
 	libcharconv_decode_utf8_.o\
@@ -150,7 +151,8 @@ LIBOBJ =\
 	libcharconv_rotated_45deg_cw.o\
 	libcharconv_rotated_90deg_cw.o\
 	libcharconv_rotated_45deg_ccw.o\
-	libcharconv_rotated_90deg_ccw.o
+	libcharconv_rotated_90deg_ccw.o\
+	libcharconv_invisible.o
 
 LOBJ = $(LIBOBJ:.o=.lo)
 
diff --git a/convert-to-invisible.c b/convert-to-invisible.c
new file mode 100644
index 0000000..8f8539b
--- /dev/null
+++ b/convert-to-invisible.c
@@ -0,0 +1,4 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+SIMPLE(libcharconv_invisible)
diff --git a/libcharconv.h b/libcharconv.h
index fb8fa60..00aa7a9 100644
--- a/libcharconv.h
+++ b/libcharconv.h
@@ -450,6 +450,15 @@ LIBCHARCONV_FUNC_(libcharconv_rotated_90deg_cw);
  */
 LIBCHARCONV_FUNC_(libcharconv_rotated_90deg_ccw);
 
+/**
+ * Convert
+ * '(' to FUNCTION APPLICATION,
+ * '*' to INVISIBLE TIMES,
+ * '|' to INVISIBLE SEPARATOR, and
+ * '+' to INVISIBLE PLUS
+ */
+LIBCHARCONV_FUNC_(libcharconv_invisible);
+
 
 #undef LIBCHARCONV_FUNC_
 #endif
diff --git a/libcharconv_flipped.c b/libcharconv_flipped.c
index fd47f99..f015c99 100644
--- a/libcharconv_flipped.c
+++ b/libcharconv_flipped.c
@@ -2,6 +2,18 @@
 #include "lib-common.h"
 
 
+static unsigned char yijing_hexagrams[] = {
+	0xC0, 0xEB, 0xCC, 0xE0, 0xC9, 0xC5, 0xD8, 0xCB,
+	0xC8, 0xF8, 0xE4, 0xF4, 0xFC, 0xFA, 0xE9, 0xD3,
+	0xCD, 0xF1, 0xDD, 0xF7, 0xE5, 0xFF, 0xD4, 0xE2,
+	0xD9, 0xD1, 0xD5, 0xF3, 0xE8, 0xC3, 0xDA, 0xD6,
+	0xEA, 0xDB, 0xF0, 0xDE, 0xF9, 0xEE, 0xD0, 0xEC,
+	0xC4, 0xEF, 0xFE, 0xE6, 0xFB, 0xDC, 0xC2, 0xC7,
+	0xE1, 0xDF, 0xF6, 0xFD, 0xF5, 0xE7, 0xF2, 0xCF,
+	0xCA, 0xED, 0xE3, 0xCE, 0xD2, 0xC6, 0xD7, 0xC1
+};
+
+
 static struct {
 	uint_least32_t a;
 	uint_least32_t b;
@@ -13,14 +25,17 @@ static struct {
 	{UINT32_C(0xA4E5), (uint_least32_t)'V'},
 	{UINT32_C(0x1D5B8), UINT32_C(0x2144)},
 	{UINT32_C(0x230D), UINT32_C(0x230F)},
-	{UINT32_C(0x230C), UINT32_C(0x230E)}
+	{UINT32_C(0x230C), UINT32_C(0x230E)},
+	{UINT32_C(0x268E), UINT32_C(0x268D)},
+	{UINT32_C(0x1D301), UINT32_C(0x1D303)},
+	{UINT32_C(0x1D302), UINT32_C(0x1D304)}
 };
 
 
 enum libcharconv_result
 libcharconv_flipped(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
 {
-	uint_least32_t c;
+	uint_least32_t c, x1, x2, x3, x4;
 	size_t i, clen;
 	*n = 0;
 	while (slen) {
@@ -42,6 +57,33 @@ libcharconv_flipped(const char *s, size_t slen, size_t *n, uint_least32_t *cp, s
 			c = c % 7u * 7u + c / 7u;
 			c += UINT32_C(0x1F063);
 			goto conv;
+		} else if (UINT32_C(0x2630) <= c && c <= UINT32_C(0x2637)) {
+			c = ((c & 1u) << 2) | ((c & 4u) >> 2) | (c & ~5u);
+			goto conv;
+		} else if (UINT32_C(0x1D306) <= c && c <= UINT32_C(0x1D356)) {
+			c -= UINT32_C(0x1D306);
+			x1 = c /  1u % 3u;
+			x2 = c /  3u % 3u;
+			x3 = c /  9u % 3u;
+			x4 = c / 27u % 3u;
+			c = UINT32_C(0x1D306);
+			c += x4 *  1u;
+			c += x3 *  3u;
+			c += x2 *  9u;
+			c += x1 * 27u;
+			goto conv;
+		} else if (UINT32_C(0x4DC0) <= c && c <= UINT32_C(0x4DFF)) {
+			for (i = 0u;; i++)
+				if ((c & 0xFFu) == yijing_hexagrams[i])
+					break;
+			c =  (i & 32u) ?  1u : 0u;
+			c |= (i & 16u) ?  2u : 0u;
+			c |= (i &  8u) ?  4u : 0u;
+			c |= (i &  4u) ?  8u : 0u;
+			c |= (i &  2u) ? 16u : 0u;
+			c |= (i &  1u) ? 32u : 0u;
+			c = UINT32_C(0x4D00) | yijing_hexagrams[c];
+			goto conv;
 		} else {
 			for (i = 0u; i < sizeof(pairs) / sizeof(*pairs); i++) {
 				if (c == pairs[i].a) {
diff --git a/libcharconv_invisible.c b/libcharconv_invisible.c
new file mode 100644
index 0000000..a3bda95
--- /dev/null
+++ b/libcharconv_invisible.c
@@ -0,0 +1,32 @@
+/* See LICENSE file for copyright and license details. */
+#include "lib-common.h"
+
+
+enum libcharconv_result
+libcharconv_invisible(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
+{
+	uint_least32_t c;
+	*n = 0;
+	for (; slen--; s++) {
+		switch (*s) {
+		case '(': c = UINT32_C(0x2061); goto conv;
+		case '*': c = UINT32_C(0x2062); goto conv;
+		case '|': c = UINT32_C(0x2063); goto conv;
+		case '+': c = UINT32_C(0x2064); goto conv;
+		default:
+			*n += 1u;
+			break;
+		}
+	}
+no_conv:
+	return LIBCHARCONV_NO_CONVERT;
+
+conv:
+	if (*n)
+		goto no_conv;
+	if (*ncp)
+		*cp = c;
+	*n += 1u;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERTED;
+}
diff --git a/libcharconv_latin.c b/libcharconv_latin.c
index 2668ab1..5e90e1e 100644
--- a/libcharconv_latin.c
+++ b/libcharconv_latin.c
@@ -882,6 +882,12 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			/* control characters (conditional characters) */
 			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;
+
 			default:
 			no_match:
 				*n += clen;
diff --git a/libcharconv_turned.c b/libcharconv_turned.c
index 835ec00..4dd587f 100644
--- a/libcharconv_turned.c
+++ b/libcharconv_turned.c
@@ -2,6 +2,18 @@
 #include "lib-common.h"
 
 
+static unsigned char yijing_hexagrams[] = {
+	0xC0, 0xEB, 0xCC, 0xE0, 0xC9, 0xC5, 0xD8, 0xCB,
+	0xC8, 0xF8, 0xE4, 0xF4, 0xFC, 0xFA, 0xE9, 0xD3,
+	0xCD, 0xF1, 0xDD, 0xF7, 0xE5, 0xFF, 0xD4, 0xE2,
+	0xD9, 0xD1, 0xD5, 0xF3, 0xE8, 0xC3, 0xDA, 0xD6,
+	0xEA, 0xDB, 0xF0, 0xDE, 0xF9, 0xEE, 0xD0, 0xEC,
+	0xC4, 0xEF, 0xFE, 0xE6, 0xFB, 0xDC, 0xC2, 0xC7,
+	0xE1, 0xDF, 0xF6, 0xFD, 0xF5, 0xE7, 0xF2, 0xCF,
+	0xCA, 0xED, 0xE3, 0xCE, 0xD2, 0xC6, 0xD7, 0xC1
+};
+
+
 static struct {
 	uint_least32_t a;
 	uint_least32_t b;
@@ -35,14 +47,17 @@ static struct {
 	{UINT32_C(0x1D5B8), UINT32_C(0x2144)},
 	{UINT32_C(0x1F12F), UINT32_C(0x00A9)},
 	{UINT32_C(0x230C), UINT32_C(0x230F)},
-	{UINT32_C(0x230E), UINT32_C(0x230D)}
+	{UINT32_C(0x230E), UINT32_C(0x230D)},
+	{UINT32_C(0x268E), UINT32_C(0x268D)},
+	{UINT32_C(0x1D301), UINT32_C(0x1D303)},
+	{UINT32_C(0x1D302), UINT32_C(0x1D304)}
 };
 
 
 enum libcharconv_result
 libcharconv_turned(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
 {
-	uint_least32_t c;
+	uint_least32_t c, x1, x2, x3, x4;
 	size_t i, clen;
 	*n = 0;
 	while (slen) {
@@ -81,6 +96,33 @@ libcharconv_turned(const char *s, size_t slen, size_t *n, uint_least32_t *cp, si
 			c = c % 7u * 7u + c / 7u;
 			c += UINT32_C(0x1F063);
 			goto conv;
+		} else if (UINT32_C(0x2630) <= c && c <= UINT32_C(0x2637)) {
+			c = ((c & 1u) << 2) | ((c & 4u) >> 2) | (c & ~5u);
+			goto conv;
+		} else if (UINT32_C(0x1D306) <= c && c <= UINT32_C(0x1D356)) {
+			c -= UINT32_C(0x1D306);
+			x1 = c /  1u % 3u;
+			x2 = c /  3u % 3u;
+			x3 = c /  9u % 3u;
+			x4 = c / 27u % 3u;
+			c = UINT32_C(0x1D306);
+			c += x4 *  1u;
+			c += x3 *  3u;
+			c += x2 *  9u;
+			c += x1 * 27u;
+			goto conv;
+		} else if (UINT32_C(0x4DC0) <= c && c <= UINT32_C(0x4DFF)) {
+			for (i = 0u;; i++)
+				if ((c & 0xFFu) == yijing_hexagrams[i])
+					break;
+			c =  (i & 32u) ?  1u : 0u;
+			c |= (i & 16u) ?  2u : 0u;
+			c |= (i &  8u) ?  4u : 0u;
+			c |= (i &  4u) ?  8u : 0u;
+			c |= (i &  2u) ? 16u : 0u;
+			c |= (i &  1u) ? 32u : 0u;
+			c = UINT32_C(0x4D00) | yijing_hexagrams[c];
+			goto conv;
 		} else {
 			for (i = 0u; i < sizeof(pairs) / sizeof(*pairs); i++) {
 				if (c == pairs[i].a) {