Zero node padding before passing to reader sinks

5 files changed, 41 insertions, 3 deletions

diff --git a/src/n3.c b/src/n3.c
index f4f296c5..6d778b1d 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -1066,10 +1066,10 @@ read_object(SerdReader* reader, ReadContext* ctx, bool emit, bool* ate_dot)
 		o->flags = flags;
 	}
 
-	if (ret && emit && simple) {
+	if (ret && emit && simple && o) {
 		ret = emit_statement(reader, *ctx, o);
 	} else if (ret && !emit) {
-		ctx->object   = o;
+		ctx->object = o;
 		return true;
 	}
 
@@ -1277,6 +1277,7 @@ read_base(SerdReader* reader, bool sparql, bool token)
 	if (!uri) {
 		return false;
 	} else if (reader->sink->base) {
+		serd_node_zero_pad(uri);
 		reader->sink->base(reader->sink->handle, uri);
 	}
 
@@ -1311,12 +1312,14 @@ read_prefixID(SerdReader* reader, bool sparql, bool token)
 	}
 
 	read_ws_star(reader);
-	const SerdNode* uri = read_IRIREF(reader);
+	SerdNode* uri = read_IRIREF(reader);
 	if (!uri) {
 		return false;
 	}
 
 	if (reader->sink->prefix) {
+		serd_node_zero_pad(name);
+		serd_node_zero_pad(uri);
 		ret = !reader->sink->prefix(reader->sink->handle, name, uri);
 	}
 	if (!sparql) {
diff --git a/src/node.c b/src/node.c
index 83242092..0211996a 100644
--- a/src/node.c
+++ b/src/node.c
@@ -218,6 +218,27 @@ serd_node_new_uri(const char* str)
 	return str ? serd_node_new_simple_node(SERD_URI, str, strlen(str)) : NULL;
 }
 
+/**
+   Zero node padding.
+
+   This is used for nodes which live in re-used stack memory during reading,
+   which must be normalized before being passed to a sink so comparison will
+   work correctly.
+*/
+void
+serd_node_zero_pad(SerdNode* node)
+{
+	char*        buf         = serd_node_buffer(node);
+	const size_t size        = node->n_bytes;
+	const size_t padded_size = serd_node_pad_size(node->n_bytes);
+
+	memset(buf + size, 0, padded_size - size);
+
+	if (node->flags & (SERD_HAS_DATATYPE|SERD_HAS_LANGUAGE)) {
+		serd_node_zero_pad(serd_node_get_meta(node));
+	}
+}
+
 SerdNode*
 serd_node_copy(const SerdNode* node)
 {
@@ -226,6 +247,13 @@ serd_node_copy(const SerdNode* node)
 	}
 
 	const size_t size = serd_node_total_size(node);
+#ifndef NDEBUG
+	const size_t unpadded_size = node->n_bytes;
+	const size_t padded_size   = serd_node_pad_size(node->n_bytes);
+	for (size_t i = 0; i < padded_size - unpadded_size; ++i) {
+		assert(serd_node_buffer_c(node)[unpadded_size + i] == '\0');
+	}
+#endif
 	SerdNode*    copy = (SerdNode*)calloc(1, size + 3);
 	memcpy(copy, node, size);
 	return copy;
diff --git a/src/node.h b/src/node.h
index 1dbb10d0..47d73a32 100644
--- a/src/node.h
+++ b/src/node.h
@@ -41,6 +41,7 @@ serd_node_buffer_c(const SerdNode* node)
 
 SerdNode* serd_node_malloc(size_t n_bytes, SerdNodeFlags flags, SerdType type);
 void      serd_node_set(SerdNode** dst, const SerdNode* src);
+void      serd_node_zero_pad(SerdNode* node);
 SerdNode* serd_node_new_resolved_uri_i(const char* str, const SerdURI* base);
 
 #endif  // SERD_NODE_H
diff --git a/src/reader.c b/src/reader.c
index 3eccacc7..9783502c 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -120,6 +120,11 @@ emit_statement(SerdReader* reader, ReadContext ctx, SerdNode* o)
 	if (!graph && reader->default_graph) {
 		graph = reader->default_graph;
 	}
+
+	/* Zero the pad of the object node on the top of the stack.  Lower nodes
+	   (subject and predicate) were already zeroed by subsequent pushes. */
+	serd_node_zero_pad(o);
+
 	bool ret = !reader->sink->statement ||
 		!reader->sink->statement(
 			reader->sink->handle, *ctx.flags, graph,
diff --git a/src/stack.h b/src/stack.h
index 5bd86f7a..31ed6c09 100644
--- a/src/stack.h
+++ b/src/stack.h
@@ -82,6 +82,7 @@ static inline void
 serd_stack_pop_to(SerdStack* stack, size_t n_bytes)
 {
 	assert(stack->size >= n_bytes);
+	memset(stack->buf + n_bytes, 0, stack->size - n_bytes);
 	stack->size = n_bytes;
 }