Zero node padding before passing to reader sinks

5 files changed, 41 insertions, 2 deletions

diff --git a/src/n3.c b/src/n3.c
index 7042233f..a6d8d801 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -1211,7 +1211,7 @@ read_object(SerdReader* const  reader,
     o->flags = flags;
   }
 
-  if (!st && emit && simple) {
+  if (!st && emit && simple && o) {
     st = emit_statement(reader, *ctx, o);
   } else if (!st && !emit) {
     ctx->object = o;
@@ -1451,6 +1451,7 @@ read_base(SerdReader* const reader, const bool sparql, const bool token)
 
   SerdNode* uri = NULL;
   TRY(st, read_IRIREF(reader, &uri));
+  serd_node_zero_pad(uri);
   TRY(st, serd_sink_write_base(reader->sink, uri));
 
   read_ws_star(reader);
@@ -1489,6 +1490,8 @@ read_prefixID(SerdReader* const reader, const bool sparql, const bool token)
   SerdNode* uri = NULL;
   TRY(st, read_IRIREF(reader, &uri));
 
+  serd_node_zero_pad(name);
+  serd_node_zero_pad(uri);
   st = serd_sink_write_prefix(reader->sink, name, uri);
 
   if (!sparql) {
diff --git a/src/node.c b/src/node.c
index ec14e611..f7ee0457 100644
--- a/src/node.c
+++ b/src/node.c
@@ -152,6 +152,27 @@ serd_node_set(SerdNode** const dst, const SerdNode* const src)
   memcpy(*dst, src, size);
 }
 
+/**
+   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->length;
+  const size_t padded_size = serd_node_pad_size(size);
+
+  memset(buf + size, 0, padded_size - size);
+
+  if (node->flags & (SERD_HAS_DATATYPE | SERD_HAS_LANGUAGE)) {
+    serd_node_zero_pad(serd_node_meta(node));
+  }
+}
+
 SerdNode*
 serd_new_token(const SerdNodeType type, const SerdStringView str)
 {
@@ -275,7 +296,14 @@ serd_node_copy(const SerdNode* node)
   }
 
   const size_t size = serd_node_total_size(node);
-  SerdNode*    copy = (SerdNode*)serd_calloc_aligned(serd_node_align, size);
+#ifndef NDEBUG
+  const size_t unpadded_size = node->length;
+  const size_t padded_size   = serd_node_pad_size(node->length);
+  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*)serd_calloc_aligned(serd_node_align, size);
   memcpy(copy, node, size);
   return copy;
 }
diff --git a/src/node.h b/src/node.h
index a87871b4..783fe574 100644
--- a/src/node.h
+++ b/src/node.h
@@ -39,6 +39,9 @@ void
 serd_node_set(SerdNode* SERD_NONNULL* SERD_NONNULL dst,
               const SerdNode* SERD_NONNULL         src);
 
+void
+serd_node_zero_pad(SerdNode* SERD_NONNULL node);
+
 /// Create a new URI from a string, resolved against a base URI
 SerdNode* SERD_ALLOCATED
 serd_new_resolved_uri(SerdStringView string, SerdURIView base_uri);
diff --git a/src/reader.c b/src/reader.c
index ae50adab..fc39d7fa 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -114,6 +114,10 @@ emit_statement(SerdReader* const reader,
     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);
+
   const SerdStatus st = serd_sink_write(
     reader->sink, *ctx.flags, ctx.subject, ctx.predicate, o, graph);
 
diff --git a/src/stack.h b/src/stack.h
index 4592705e..6ad3385d 100644
--- a/src/stack.h
+++ b/src/stack.h
@@ -72,6 +72,7 @@ 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;
 }