Add lisu

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

5 files changed, 182 insertions, 2 deletions

diff --git a/Makefile b/Makefile
index c1b5690..e3e3c01 100644
--- a/Makefile
+++ b/Makefile
@@ -53,7 +53,8 @@ BIN =\
 	convert-to-clock-faces\
 	convert-to-ocr\
 	convert-to-crop-marks\
-	convert-to-braille
+	convert-to-braille\
+	convert-to-lisu
 
 LIBOBJ =\
 	libcharconv_decode_utf8_.o\
@@ -95,7 +96,8 @@ LIBOBJ =\
 	libcharconv_clock_faces.o\
 	libcharconv_ocr.o\
 	libcharconv_crop_marks.o\
-	libcharconv_braille.o
+	libcharconv_braille.o\
+	libcharconv_lisu.o
 
 LOBJ = $(LIBOBJ:.o=.lo)
 
diff --git a/convert-to-lisu.c b/convert-to-lisu.c
new file mode 100644
index 0000000..3a8320a
--- /dev/null
+++ b/convert-to-lisu.c
@@ -0,0 +1,18 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+USAGE("");
+
+
+int
+main(int argc, char *argv[])
+{
+	ARGBEGIN {
+	default:
+		usage();
+	} ARGEND;
+	if (argc)
+		usage();
+
+	return convert(&libcharconv_lisu);
+}
diff --git a/libcharconv.h b/libcharconv.h
index 613ef1b..9c77b1f 100644
--- a/libcharconv.h
+++ b/libcharconv.h
@@ -1313,4 +1313,36 @@ enum libcharconv_result libcharconv_crop_marks(const char *s, size_t slen, size_
 enum libcharconv_result libcharconv_braille(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp);
 
 
+/**
+ * Preforms convertion from Latin to Lisu
+ * 
+ * @param   s     Text to convert
+ * @param   slen  The number of bytes available in `s`
+ * @param   n     Output parameter for the number of consumed bytes
+ * @param   cp    Output buffer for the codepoints
+ * @param   ncp   Input parameter for the number of codepoints that
+ *                fit in `cp`, and output parameter for the number
+ *                of output codepoints (if it exceeds the original
+ *                value of `ncp`, a larger buffer is needed)
+ * @return        LIBCHARCONV_NO_CONVERT:
+ *                  `*n` is the number of bytes from the beginning
+ *                  of `s` that cannot be converted
+ *                LIBCHARCONV_CONVERTED:
+ *                  `*n` is the number of bytes from the beginning
+ *                  of `s` that was converted to a codepoint which
+ *                  is stored in `*cp`
+ *                LIBCHARCONV_INDETERMINATE:
+ *                  If all text has been input, no more can be
+ *                  converted, otherwise more of the text most
+ *                  be made available before the function can
+ *                  determine whether the beginning of `s` can be
+ *                  converted or what it should be converted to
+ *                LIBCHARCONV_CONVERT_IF_END:
+ *                  As LIBCHARCONV_CONVERTED the entire text has
+ *                  been input, as LIBCHARCONV_INDETERMINATE
+ *                  otherwise
+ */
+enum libcharconv_result libcharconv_lisu(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp);
+
+
 #endif
diff --git a/libcharconv_latin.c b/libcharconv_latin.c
index 5124c53..9dc81cf 100644
--- a/libcharconv_latin.c
+++ b/libcharconv_latin.c
@@ -382,7 +382,15 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			*ncp = i;
 			return ret;
 
+		} else if (UINT32_C(0xA4D0) <= c && c <= UINT32_C(0xA4FF)) {
+			/* lisu */
+			c1 = "bpPdtYgkKjcCzfFmnlsrRVvhGJwxyBaAeEiouULD.,  ;:-="[c - UINT32_C(0xA4D0)];
+			if (c1 == ' ')
+				goto use_switch;
+			goto conv1;
+
 		} else {
+		use_switch:
 			switch (c) {
 			/* shogi */
 			case UINT32_C(0x2616): c = (uint_least32_t)'w'; goto conv;
@@ -529,6 +537,11 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			case UINT32_C(0x230E): c1 = '|'; c2 = '_'; goto conv2;
 			case UINT32_C(0x230F): c1 = '_'; c2 = '|'; goto conv2;
 
+			/* lisu */
+			case UINT32_C(0xA4FB): c1 = '.'; c2 = ','; goto conv2;
+			case UINT32_C(0xA4FA): c1 = '.'; c2 = '.'; goto conv2;
+			case UINT32_C(0x11FB0): c1 = 'Y'; goto conv1;
+
 			default:
 			no_match:
 				*n += clen;
diff --git a/libcharconv_lisu.c b/libcharconv_lisu.c
new file mode 100644
index 0000000..2a0293a
--- /dev/null
+++ b/libcharconv_lisu.c
@@ -0,0 +1,115 @@
+/* See LICENSE file for copyright and license details. */
+#include "lib-common.h"
+
+
+static struct {
+	unsigned char cp_low;
+	char latin;
+} lisu[] = {
+	{0xEE, 'a'},
+	{0xD0, 'b'},
+	{0xDA, 'c'},
+	{0xD3, 'd'},
+	{0xF0, 'e'},
+	{0xDD, 'f'},
+	{0xD6, 'g'},
+	{0xE7, 'h'},
+	{0xF2, 'i'},
+	{0xD9, 'j'},
+	{0xD7, 'k'},
+	{0xE1, 'l'},
+	{0xDF, 'm'},
+	{0xE0, 'n'},
+	{0xF3, 'o'},
+	{0xD1, 'p'},
+	{0xE3, 'r'},
+	{0xE2, 's'},
+	{0xD4, 't'},
+	{0xF4, 'u'},
+	{0xE6, 'v'},
+	{0xEA, 'w'},
+	{0xEB, 'x'},
+	{0xEC, 'y'},
+	{0xDC, 'z'},
+	{0xEF, 'A'},
+	{0xED, 'B'},
+	{0xDB, 'C'},
+	{0xF7, 'D'},
+	{0xF1, 'E'},
+	{0xDE, 'F'},
+	{0xE8, 'G'},
+	{0xE7, 'H'}, /* symmetry */
+	{0xF2, 'I'}, /* symmetry */
+	{0xE9, 'J'},
+	{0xD8, 'K'},
+	{0xF6, 'L'},
+	{0xEA, 'M'}, /* W */
+	{0xE0, 'N'}, /* symmetry */
+	{0xF3, 'O'}, /* symmetry */
+	{0xD2, 'P'},
+	{0xE4, 'R'},
+	{0xE2, 'S'}, /* symmetry */
+	{0xD5, 'Y'},
+	{0xF5, 'U'},
+	{0xE5, 'V'},
+	{0xDF, 'W'}, /* M */
+	{0xEB, 'X'}, /* symmetry */
+	{0xDC, 'Z'}, /* symmetry */
+	{0xFE, '-'},
+	{0xFF, '='},
+	{0xFD, ':'},
+	{0xFC, ';'},
+	{0xF9, ','}
+};
+
+
+enum libcharconv_result
+libcharconv_lisu(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
+{
+	uint_least32_t c;
+	size_t i;
+	*n = 0;
+	for (; slen--; s++) {
+		if (*s == 'Y') {
+			c = UINT32_C(0x11FB0);
+			goto conv;
+		} else if (*s == '.') {
+			if (*n)
+				goto no_conv;
+			if (*ncp)
+				*cp = UINT32_C(0xA4F8);
+			*n += 1u;
+			*ncp = 1u;
+			if (!slen--)
+				return LIBCHARCONV_CONVERT_IF_END;
+			if (s[1] == '.') {
+				if (*ncp)
+					*cp = UINT32_C(0xA4FA);
+				*n += 1u;
+			} else if (s[1] == ',') {
+				if (*ncp)
+					*cp = UINT32_C(0xA4FB);
+				*n += 1u;
+			}
+			return LIBCHARCONV_CONVERTED;
+		} else {
+			for (i = 0u; i < sizeof(lisu) / sizeof(*lisu); i++)
+				if (*s == lisu[i].latin)
+					goto conv_listed;
+			*n += 1u;
+		}
+	}
+no_conv:
+	return LIBCHARCONV_NO_CONVERT;
+
+conv_listed:
+	c = (uint_least32_t)(UINT32_C(0xA400) | lisu[i].cp_low);
+conv:
+	if (*n)
+		goto no_conv;
+	if (*ncp)
+		*cp = c;
+	*n += 1u;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERTED;
+}