Make sord_node_get_counted return byte count

Also clarify formerly ambiguous documentation, and add
sord_node_get_string_measured to get both byte and character counts.

Technically, this breaks behaviour, but most (if not all) existing code
that uses this function was actually expecting a byte count.  This is
unsurprising since a UTF-8 character count is almost always useless.
This approach causes less problems in reality, since it fixes broken
behaviour in the wild without requiring any changes in user code.

1 files changed, 12 insertions, 2 deletions

diff --git a/src/sord.c b/src/sord.c
index c6b2b38..fbc33f7 100644
--- a/src/sord.c
+++ b/src/sord.c
@@ -919,9 +919,19 @@ sord_node_get_string(const SordNode* node)
 }
 
 const uint8_t*
-sord_node_get_string_counted(const SordNode* node, size_t* len)
+sord_node_get_string_counted(const SordNode* node, size_t* bytes)
 {
-	*len = node->node.n_chars;
+	*bytes = node->node.n_bytes;
+	return node->node.buf;
+}
+
+const uint8_t*
+sord_node_get_string_measured(const SordNode* node,
+                              size_t*         bytes,
+                              size_t*         chars)
+{
+	*bytes = node->node.n_bytes;
+	*chars = node->node.n_chars;
 	return node->node.buf;
 }