Add crop marks

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

5 files changed, 119 insertions, 2 deletions

diff --git a/Makefile b/Makefile
index a33bec8..f2e0bb9 100644
--- a/Makefile
+++ b/Makefile
@@ -51,7 +51,8 @@ BIN =\
 	convert-to-lycian\
 	convert-to-domino-tiles\
 	convert-to-clock-faces\
-	convert-to-ocr
+	convert-to-ocr\
+	convert-to-crop-marks
 
 LIBOBJ =\
 	libcharconv_decode_utf8_.o\
@@ -91,7 +92,8 @@ LIBOBJ =\
 	libcharconv_domino_tiles_horizontal.o\
 	libcharconv_domino_tiles_vertical.o\
 	libcharconv_clock_faces.o\
-	libcharconv_ocr.o
+	libcharconv_ocr.o\
+	libcharconv_crop_marks.o
 
 LOBJ = $(LIBOBJ:.o=.lo)
 
diff --git a/convert-to-crop-marks.c b/convert-to-crop-marks.c
new file mode 100644
index 0000000..cc33009
--- /dev/null
+++ b/convert-to-crop-marks.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_crop_marks);
+}
diff --git a/libcharconv.h b/libcharconv.h
index f8d15c1..5b165b6 100644
--- a/libcharconv.h
+++ b/libcharconv.h
@@ -1246,4 +1246,36 @@ enum libcharconv_result libcharconv_clock_faces(const char *s, size_t slen, size
 enum libcharconv_result libcharconv_ocr(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp);
 
 
+/**
+ * Convert "-|", "|-", "_|", and "|_" to CROP marks
+ * 
+ * @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_crop_marks(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp);
+
+
 #endif
diff --git a/libcharconv_crop_marks.c b/libcharconv_crop_marks.c
new file mode 100644
index 0000000..81d3aa1
--- /dev/null
+++ b/libcharconv_crop_marks.c
@@ -0,0 +1,59 @@
+/* See LICENSE file for copyright and license details. */
+#include "lib-common.h"
+
+
+enum libcharconv_result
+libcharconv_crop_marks(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++) {
+		if (s[0] == '|') {
+			if (!slen)
+				goto indeterminate;
+			if (s[1] == '-')
+				c = UINT32_C(0x230C);
+			else if (s[1] == '_')
+				c = UINT32_C(0x230E);
+			else
+				goto no_conv;
+			goto conv;
+		} else if (s[0] == '-') {
+			if (!slen)
+				goto indeterminate;
+			if (s[1] != '|')
+				goto no_conv;
+			c = UINT32_C(0x230D);
+			goto conv;
+		} else if (s[0] == '_') {
+			if (!slen)
+				goto indeterminate;
+			if (s[1] != '|')
+				goto no_conv;
+			c = UINT32_C(0x230F);
+			goto conv;
+		} else {
+			*n += 1u;
+		}
+	}
+	return LIBCHARCONV_NO_CONVERT;
+
+no_conv:
+	if (!*n)
+		*n = 1u;
+	return LIBCHARCONV_NO_CONVERT;
+
+indeterminate:
+	if (*n)
+		goto no_conv;
+	return LIBCHARCONV_INDETERMINATE;
+
+conv:
+	if (*n)
+		goto no_conv;
+	if (*ncp)
+		*cp = c;
+	*n += 2u;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERTED;
+}
diff --git a/libcharconv_latin.c b/libcharconv_latin.c
index 68bdd7d..de5b618 100644
--- a/libcharconv_latin.c
+++ b/libcharconv_latin.c
@@ -496,6 +496,12 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 			/* ocr (telegraphing) */
 			case UINT32_C(0x2BFF): c1 = 'E'; goto conv1;
 
+			/* crop marks */
+			case UINT32_C(0x230C): c1 = '|'; c2 = '-'; goto conv2;
+			case UINT32_C(0x230D): c1 = '-'; c2 = '|'; goto conv2;
+			case UINT32_C(0x230E): c1 = '|'; c2 = '_'; goto conv2;
+			case UINT32_C(0x230F): c1 = '_'; c2 = '|'; goto conv2;
+
 			default:
 			no_match:
 				*n += clen;