1 files changed, 97 insertions, 0 deletions
diff --git a/characters.c b/characters.c
new file mode 100644
index 0000000..dae201f
--- /dev/null
+++ b/characters.c
@@ -0,0 +1,97 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+
+int
+is_noncharacter(uint32_t cp)
+{
+	return (0xFDD0u <= cp && cp <= 0xFDEFu) || (cp & 0xFFFEu) == 0xFFFEu;
+}
+
+
+int
+is_surrogate(uint32_t cp)
+{
+	return 0xD800u <= cp && cp <= 0xDFFFu;
+}
+
+
+int
+is_valid_codepoint(uint32_t cp)
+{
+	return cp <= LIBCMAP_ULTIMATE_CODEPOINT && !is_noncharacter(cp) && !is_surrogate(cp);
+}
+
+
+struct libcmap_codepoint_data *
+get_codepoint_data(uint32_t cp)
+{
+	static char *databuf = NULL;
+	static size_t databufsize = 0u;
+	static struct libcmap_codepoint_data data;
+
+	int r;
+
+again:
+	r = libcmap_get_codepoint_data(cp, &data, databuf, &databufsize);
+	if (r <= 0) {
+		if (!r || errno != ERANGE || !databufsize)
+			return NULL;
+		databuf = erealloc(databuf, databufsize);
+		goto again;
+	}
+
+	return &data;
+}
+
+
+enum codepoint_class
+classify_from_codepoint_data(const struct libcmap_codepoint_data *data)
+{
+	if (!data)
+		return CP_OUT_OF_RANGE;
+
+	switch ((int)(data->general_category & LIBCMAP_GENERAL_SUPERCATEGORY_MASK)) {
+	case LIBCMAP_GENERAL_SUPERCATEGORY_MARK:
+		return CP_COMBINING_MARK;
+	default:
+		break;
+	}
+
+	switch ((int)data->general_category) {
+	case LIBCMAP_GENERAL_CATEGORY_OTHER_CONTROL:
+		return CP_CONTROL;
+	case LIBCMAP_GENERAL_CATEGORY_OTHER_SURROGATE:
+		return CP_SURROGATE;
+	case LIBCMAP_GENERAL_CATEGORY_OTHER_NOT_ASSIGNED:
+		return data->noncharacter ? CP_NONCHARACTER : CP_RESERVED;
+	default:
+		break;
+	}
+
+	return CP_REGULAR;
+}
+
+
+enum codepoint_class
+classify_codepoint(uint32_t cp)
+{
+	return classify_from_codepoint_data(get_codepoint_data(cp));
+}
+
+
+int
+is_in_script(uint32_t cp, const struct libcmap_script *script)
+{
+	size_t i;
+	if (!script->nranges)
+		return 0;
+	if (script->ranges[0].first > cp)
+		return 0;
+	if (cp > script->ranges[script->nranges - 1u].last)
+		return 0;
+	for (i = 0u; i < script->nranges; i++)
+		if (script->ranges[i].first <= cp && cp <= script->ranges[i].last)
+			return 1;
+	return 0;
+}