Add replacement

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

5 files changed, 112 insertions, 2 deletions

diff --git a/Makefile b/Makefile
index 61f9469..76d3333 100644
--- a/Makefile
+++ b/Makefile
@@ -41,7 +41,8 @@ BIN =\
 	convert-to-bold-fraktur\
 	convert-to-script\
 	convert-to-bold-script\
-	convert-to-buhid
+	convert-to-buhid\
+	convert-to-replacement
 
 LIBOBJ =\
 	libcharconv_shogi.o\
@@ -69,7 +70,8 @@ LIBOBJ =\
 	libcharconv_bold_fraktur.o\
 	libcharconv_script.o\
 	libcharconv_bold_script.o\
-	libcharconv_buhid.o
+	libcharconv_buhid.o\
+	libcharconv_replacement.o
 
 HDR =\
 	libcharconv.h
diff --git a/convert-to-replacement.c b/convert-to-replacement.c
new file mode 100644
index 0000000..6407b52
--- /dev/null
+++ b/convert-to-replacement.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_replacement);
+}
diff --git a/libcharconv.h b/libcharconv.h
index aa86a16..8e3c4bc 100644
--- a/libcharconv.h
+++ b/libcharconv.h
@@ -867,4 +867,38 @@ enum libcharconv_result libcharconv_bold_script(const char *s, size_t slen, size
 enum libcharconv_result libcharconv_buhid(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp);
 
 
+/**
+ * Convert
+ * "obj" to OBJECT REPLACEMENT CHARACTER, and
+ * "?" to REPLACEMENT CHARACTER
+ * 
+ * @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_replacement(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 d682b17..be0b53c 100644
--- a/libcharconv_latin.c
+++ b/libcharconv_latin.c
@@ -367,6 +367,16 @@ libcharconv_latin(const char *s, size_t slen, size_t *n, uint_least32_t *cp, siz
 				}
 				goto conv2_prechecked;
 
+			/* replacement */
+			case UINT32_C(0xFFFC):
+				c1 = 'o';
+				c2 = 'b';
+				c3 = 'j';
+				goto conv3;
+			case UINT32_C(0xFFFD):
+				c = (uint_least32_t)'?';
+				goto conv;
+
 			default:
 			no_match:
 				*n += clen;
@@ -398,6 +408,10 @@ conv2_prechecked:
 	*ncp = 2u;
 	return ret;
 
+conv3:
+	if (*n)
+		goto no_conv;
+	*n += clen;
 conv3_prechecked:
 	if (*ncp >= 1u)
 		cp[0] = (uint_least32_t)c1;
diff --git a/libcharconv_replacement.c b/libcharconv_replacement.c
new file mode 100644
index 0000000..a9bd3c5
--- /dev/null
+++ b/libcharconv_replacement.c
@@ -0,0 +1,42 @@
+/* See LICENSE file for copyright and license details. */
+#include "libcharconv.h"
+#include <strings.h>
+
+
+enum libcharconv_result
+libcharconv_replacement(const char *s, size_t slen, size_t *n, uint_least32_t *cp, size_t *ncp)
+{
+	uint_least32_t c;
+	*n = 0;
+	while (slen) {
+		if (slen < 3u && !strncasecmp(s, "obj", slen)) {
+			if (*n)
+				goto no_conv;
+			return LIBCHARCONV_INDETERMINATE;
+		} else if (slen >= 3u && !strncasecmp(s, "obj", 3u)) {
+			if (*n)
+				goto no_conv;
+			c = UINT32_C(0xFFFC);
+			*n = 3u;
+			goto conv;
+		} else if (*s == '?') {
+			if (*n)
+				goto no_conv;
+			c = UINT32_C(0xFFFD);
+			*n = 1u;
+			goto conv;
+		} else {
+			*n += 1u;
+			s++;
+			slen--;
+		}
+	}
+no_conv:
+	return LIBCHARCONV_NO_CONVERT;
+
+conv:
+	if (*ncp)
+		*cp = c;
+	*ncp = 1u;
+	return LIBCHARCONV_CONVERTED;
+}