Add serd_node_from_substring()

This allows creating nodes in-place from substrings of other strings to allow
zero-copy serialization from existing delimited buffers.

3 files changed, 61 insertions, 10 deletions

diff --git a/src/node.c b/src/node.c
index 42992917..10df10d9 100644
--- a/src/node.c
+++ b/src/node.c
@@ -48,6 +48,22 @@ serd_node_from_string(SerdType type, const uint8_t* str)
 
 SERD_API
 SerdNode
+serd_node_from_substring(SerdType type, const uint8_t* str, const size_t len)
+{
+	if (!str) {
+		return SERD_NODE_NULL;
+	}
+
+	uint32_t     flags       = 0;
+	size_t       buf_n_bytes = 0;
+	const size_t buf_n_chars = serd_substrlen(str, len, &buf_n_bytes, &flags);
+	assert(buf_n_bytes <= len);
+	SerdNode ret = { str, buf_n_bytes, buf_n_chars, flags, type };
+	return ret;
+}
+
+SERD_API
+SerdNode
 serd_node_copy(const SerdNode* node)
 {
 	if (!node || !node->buf) {
@@ -187,7 +203,7 @@ serd_node_new_file_uri(const uint8_t* path,
 		serd_uri_parse(chunk.buf, out);
 	}
 
-	return serd_node_from_string(SERD_URI, chunk.buf);
+	return serd_node_from_substring(SERD_URI, chunk.buf, chunk.len);
 }
 
 SERD_API
diff --git a/src/serd_internal.h b/src/serd_internal.h
index afbea5fb..55f6a6b6 100644
--- a/src/serd_internal.h
+++ b/src/serd_internal.h
@@ -325,6 +325,12 @@ is_windows_path(const uint8_t* path)
 		&& (path[2] == '/' || path[2] == '\\');
 }
 
+size_t
+serd_substrlen(const uint8_t* str,
+               const size_t   len,
+               size_t*        n_bytes,
+               SerdNodeFlags* flags);
+
 /* URI utilities */
 
 static inline bool
diff --git a/src/string.c b/src/string.c
index 9381015d..dedd0713 100644
--- a/src/string.c
+++ b/src/string.c
@@ -36,6 +36,42 @@ serd_strerror(SerdStatus status)
 	return (const uint8_t*)"Unknown error";  // never reached
 }
 
+static inline void
+serd_update_flags(const uint8_t c, SerdNodeFlags* const flags)
+{
+	switch (c) {
+	case '\r': case '\n':
+		*flags |= SERD_HAS_NEWLINE;
+		break;
+	case '"':
+		*flags |= SERD_HAS_QUOTE;
+	}
+}
+
+size_t
+serd_substrlen(const uint8_t* const str,
+               const size_t         len,
+               size_t* const        n_bytes,
+               SerdNodeFlags* const flags)
+{
+	size_t        n_chars = 0;
+	size_t        i       = 0;
+	SerdNodeFlags f       = 0;
+	for (; i < len && str[i]; ++i) {
+		if ((str[i] & 0xC0) != 0x80) {  // Start of new character
+			++n_chars;
+			serd_update_flags(str[i], &f);
+		}
+	}
+	if (n_bytes) {
+		*n_bytes = i;
+	}
+	if (flags) {
+		*flags = f;
+	}
+	return n_chars;
+}
+
 SERD_API
 size_t
 serd_strlen(const uint8_t* str, size_t* n_bytes, SerdNodeFlags* flags)
@@ -44,16 +80,9 @@ serd_strlen(const uint8_t* str, size_t* n_bytes, SerdNodeFlags* flags)
 	size_t        i       = 0;
 	SerdNodeFlags f       = 0;
 	for (; str[i]; ++i) {
-		if ((str[i] & 0xC0) != 0x80) {
-			// Does not start with `10', start of a new character
+		if ((str[i] & 0xC0) != 0x80) {  // Start of new character
 			++n_chars;
-			switch (str[i]) {
-			case '\r': case '\n':
-				f |= SERD_HAS_NEWLINE;
-				break;
-			case '"':
-				f |= SERD_HAS_QUOTE;
-			}
+			serd_update_flags(str[i], &f);
 		}
 	}
 	if (n_bytes) {