Merge with dom-stack-keep-nodes

src/document/dom/node.c

@@ -18,6 +18,8 @@
 #include "util/string.h"
 
 
+static void done_dom_node_data(struct dom_node *node);
+
 /* Node lists */
 
 #define DOM_NODE_LIST_GRANULARITY 0x7
@@ -84,6 +86,27 @@ add_to_dom_node_list(struct dom_node_list **list_ptr,
 	return list;
 }
 
+static void
+del_from_dom_node_list(struct dom_node_list *list, struct dom_node *node)
+{
+	struct dom_node *entry;
+	size_t i;
+
+	if (!list) return;
+
+	foreach_dom_node(i, entry, list) {
+		size_t successors;
+
+		if (entry != node) continue;
+
+		successors = list->size - (i + 1);
+		if (successors)
+			memmove(&list->entries[i], &list->entries[i+1],
+				sizeof(*list->entries) * successors);
+		list->size--;
+	}
+}
+
 void
 done_dom_node_list(struct dom_node_list *list)
 {
@@ -93,7 +116,8 @@ done_dom_node_list(struct dom_node_list *list)
 	assert(list);
 
 	foreach_dom_node (i, node, list) {
-		done_dom_node(node);
+		/* Avoid that the node start messing with the node list. */
+		done_dom_node_data(node);
 	}
 
 	mem_free(list);
@@ -213,6 +237,7 @@ init_dom_node_(unsigned char *file, int line,
 	node->type   = type;
 	node->string = string;
 	node->length = length;
+	node->parent = parent;
 
 	if (parent) {
 		struct dom_node_list **list = get_dom_node_list(parent, node);
@@ -236,7 +261,7 @@ init_dom_node_(unsigned char *file, int line,
 }
 
 void
-done_dom_node(struct dom_node *node)
+done_dom_node_data(struct dom_node *node)
 {
 	union dom_node_data *data;
 
@@ -259,7 +284,6 @@ done_dom_node(struct dom_node *node)
 
 		if (data->document.children)
 			done_dom_node_list(data->document.children);
-
 		break;
 
 	case DOM_NODE_ELEMENT:
@@ -268,7 +292,6 @@ done_dom_node(struct dom_node *node)
 
 		if (data->element.map)
 			done_dom_node_list(data->element.map);
-
 		break;
 
 	case DOM_NODE_TEXT:
@@ -279,8 +302,8 @@ done_dom_node(struct dom_node *node)
 	case DOM_NODE_PROCESSING_INSTRUCTION:
 		if (data->proc_instruction.map)
 			done_dom_node_list(data->proc_instruction.map);
-
 		break;
+
 	default:
 		break;
 	}
@@ -288,6 +311,38 @@ done_dom_node(struct dom_node *node)
 	mem_free(node);
 }
 
+void
+done_dom_node(struct dom_node *node)
+{
+	assert(node);
+
+	if (node->parent) {
+		struct dom_node *parent = node->parent;
+		union dom_node_data *data = &parent->data;
+
+		switch (parent->type) {
+		case DOM_NODE_DOCUMENT:
+			del_from_dom_node_list(data->document.meta_nodes, node);
+			del_from_dom_node_list(data->document.children, node);
+			break;
+
+		case DOM_NODE_ELEMENT:
+			del_from_dom_node_list(data->element.children, node);
+			del_from_dom_node_list(data->element.map, node);
+			break;
+
+		case DOM_NODE_PROCESSING_INSTRUCTION:
+			del_from_dom_node_list(data->proc_instruction.map, node);
+			break;
+
+ 		default:
+ 			break;
+		}
+ 	}
+ 
+	done_dom_node_data(node);
+}
+
 #define set_node_name(name, namelen, str)	\
 	do { (name) = (str); (namelen) = sizeof(str) - 1; } while (0)
 

src/document/dom/node.h

@@ -190,6 +190,8 @@ struct dom_node {
 	uint16_t length;
 	unsigned char *string;
 
+	struct dom_node *parent;
+
 	/* Various info depending on the type of the node. */
 	union dom_node_data {
 		struct dom_document_node	 document;
@@ -218,13 +220,13 @@ struct dom_node_list {
 	struct dom_node *entries[1];
 };
 
-#define foreach_dom_node(iterator, node, list)		\
-	for (i = 0; i < (list)->size; i++)		\
-		if (((node) = (list)->entries[i]))
+#define foreach_dom_node(i, node, list)			\
+	for ((i) = 0; (i) < (list)->size; (i)++)	\
+		if (((node) = (list)->entries[(i)]))
 
-#define foreachback_dom_node(iterator, node, list)	\
-	for (i = (list)->size - 1; i > 0; i--)		\
-		if (((node) = (list)->entries[i]))
+#define foreachback_dom_node(i, node, list)		\
+	for ((i) = (list)->size - 1; (i) > 0; (i)--)	\
+		if (((node) = (list)->entries[(i)]))
 
 #define is_dom_node_list_member(list, member)		\
 	((list) && 0 <= (member) && (member) < (list)->size)

src/document/dom/stack.c

@@ -48,7 +48,7 @@ void
 init_dom_stack(struct dom_stack *stack, void *parser, void *renderer,
 	       dom_stack_callback_T push_callbacks[DOM_NODES],
 	       dom_stack_callback_T pop_callbacks[DOM_NODES],
-	       size_t object_size)
+	       size_t object_size, int keep_nodes)
 {
 	assert(stack);
 
@@ -57,6 +57,7 @@ init_dom_stack(struct dom_stack *stack, void *parser, void *renderer,
 	stack->parser      = parser;
 	stack->renderer    = renderer;
 	stack->object_size = object_size;
+	stack->keep_nodes  = !!keep_nodes;
 
 	if (push_callbacks)
 		memcpy(stack->push_callbacks, push_callbacks, DOM_STACK_CALLBACKS_SIZE);
@@ -148,6 +149,9 @@ do_pop_dom_node(struct dom_stack *stack, struct dom_stack_state *parent)
 		callback(stack, state->node, state_data);
 	}
 
+	if (!stack->keep_nodes)
+		done_dom_node(state->node);
+
 	stack->depth--;
 	assert(stack->depth >= 0);
 

src/document/dom/stack.h

@@ -37,6 +37,9 @@ struct dom_stack {
 	struct dom_stack_state *states;
 	size_t depth;
 
+	/* Keep nodes when popping them or call done_dom_node() on them. */
+	unsigned int keep_nodes:1;
+
 	/* This is one big array of parser specific objects. */
 	/* The objects hold parser specific data. For the SGML parser this
 	 * holds DTD-oriented info about the node (recorded in struct
@@ -110,7 +113,7 @@ search_dom_stack(struct dom_stack *stack, enum dom_node_type type,
 void init_dom_stack(struct dom_stack *stack, void *parser, void *renderer,
 		    dom_stack_callback_T push_callbacks[DOM_NODES],
 		    dom_stack_callback_T pop_callbacks[DOM_NODES],
-		    size_t object_size);
+		    size_t object_size, int keep_nodes);
 void done_dom_stack(struct dom_stack *stack);
 
 /* Decends down to the given node making it the current parent */

src/document/sgml/parser.c

@@ -318,7 +318,8 @@ init_sgml_parser(enum sgml_parser_type type, void *renderer, struct uri *uri,
 	parser->info = &sgml_html_info;
 
 	init_dom_stack(&parser->stack, parser, renderer,
-		       push_callbacks, pop_callbacks, obj_size);
+		       push_callbacks, pop_callbacks, obj_size,
+		       type != SGML_PARSER_STREAM);
 
 	parser->root = add_sgml_document(&parser->stack, parser->uri);
 	if (!parser->root) {