1 files changed, 123 insertions, 0 deletions
diff --git a/libcmap_get_codepoint_data.c b/libcmap_get_codepoint_data.c
new file mode 100644
index 0000000..553636d
--- /dev/null
+++ b/libcmap_get_codepoint_data.c
@@ -0,0 +1,123 @@
+/* See LICENSE file for copyright and license details. */
+#include "common.h"
+
+
+static int
+construct_codepoint_label(const char *metalabel, uint32_t codepoint, void *buffer, size_t *bufsizep)
+{
+	/* See Unicode Core Spec 4.8.2 Code Point Labels */
+
+	size_t req = sizeof("-NNNN") + strlen(metalabel) - 2u * (size_t)(*metalabel == '<');
+	char *p = buffer;
+
+	if (codepoint >> 16)  req += 1u;
+	if (codepoint >> 20)  req += 1u;
+
+	if (*bufsizep < req) {
+		*bufsizep = req;
+		errno = ERANGE;
+		return -1;
+	}
+
+	p = stpcpy(buffer, &metalabel[*metalabel == '<']);
+	p -= *metalabel == '<';
+	*p++ = '-';
+	if (codepoint >> 20)  *p++ = "0123456789ABCDEF"[(codepoint >> 20) & 15u];
+	if (codepoint >> 16)  *p++ = "0123456789ABCDEF"[(codepoint >> 16) & 15u];
+	*p++ = "0123456789ABCDEF"[(codepoint >> 12) & 15u];
+	*p++ = "0123456789ABCDEF"[(codepoint >> 8) & 15u];
+	*p++ = "0123456789ABCDEF"[(codepoint >> 4) & 15u];
+	*p++ = "0123456789ABCDEF"[(codepoint >> 0) & 15u];
+	*p = '\0';
+
+	return 0;
+}
+
+
+int
+libcmap_get_codepoint_data(uint32_t codepoint, struct libcmap_codepoint_data *data_out, void *buffer, size_t *bufsizep)
+{
+	const struct libcmap_codepoint_data *data;
+	size_t i, namelen;
+	const char *metalabel;
+
+	memset(data_out, 0, sizeof(*data_out));
+	data_out->name = NULL;
+	data_out->unicode_1_name = NULL;
+	data_out->numeric_value = NULL;
+	data_out->decomposition_mapping = NULL;
+
+	if (codepoint > LIBCMAP_ULTIMATE_CODEPOINT)
+		return 0;
+
+	if ((0xFDD0u <= codepoint && codepoint <= 0xFDEFu) || (codepoint & 0xFFFEu) == 0xFFFEu) {
+		if (data_out) {
+			if (construct_codepoint_label("noncharacter", codepoint, buffer, bufsizep))
+				return -1;
+			data_out->name = buffer;
+			data_out->general_category = LIBCMAP_GENERAL_CATEGORY_OTHER_NOT_ASSIGNED;
+			data_out->noncharacter = 1u;
+		}
+		return 1;
+	}
+
+	for (i = 0u; i < libcmap_no_data_range_list_size; i++) {
+		if (codepoint <= libcmap_no_data_range_list[i].last) {
+			if (libcmap_no_data_range_list[i].first > codepoint)
+				break;
+			if (!data_out)
+				return 1;
+			data = libcmap_find_codepoint_data(libcmap_no_data_range_list[i].first);
+			switch ((int)data->general_category) {
+			case LIBCMAP_GENERAL_CATEGORY_OTHER_CONTROL:      metalabel = "control";      goto have_metalabel;
+			case LIBCMAP_GENERAL_CATEGORY_OTHER_SURROGATE:    metalabel = "surrogate";    goto have_metalabel;
+			case LIBCMAP_GENERAL_CATEGORY_OTHER_PRIVATE_USE:  metalabel = "private-use";  goto have_metalabel;
+			default:
+				namelen = strlen(data->name);
+				namelen -= sizeof(", First") - 1u;
+				if (namelen + 1u > *bufsizep) {
+					*bufsizep = namelen + 1u;
+					errno = ERANGE;
+					return -1;
+				}
+				memcpy(buffer, data->name, namelen - 1u);
+				((char *)buffer)[namelen - 1u] = '>';
+				((char *)buffer)[namelen] = '\0';
+				break;
+
+			have_metalabel:
+				if (construct_codepoint_label(metalabel, codepoint, buffer, bufsizep))
+					return -1;
+				break;
+			}
+			*data_out = *data;
+			data_out->name = buffer;
+			return 1;
+		}
+	}
+
+	data = libcmap_find_codepoint_data(codepoint);
+	if (!data) {
+		if (data_out) {
+			if (construct_codepoint_label("reserved", codepoint, buffer, bufsizep))
+				return -1;
+			data_out->name = buffer;
+			data_out->general_category = LIBCMAP_GENERAL_CATEGORY_OTHER_NOT_ASSIGNED;
+			data_out->noncharacter = 0u;
+		}
+		return 1;
+	}
+
+	if (data_out) {
+		if (data->name[0u] == '<') {
+			if (construct_codepoint_label(data->name, codepoint, buffer, bufsizep))
+				return -1;
+			*data_out = *data;
+			data_out->name = buffer;
+		} else {
+			*data_out = *data;
+		}
+	}
+
+	return 1;
+}