fixed dd,dt&li handler. Also fixed the scoping lists. Also added a missing summary tag to the start tag processor conditional.

9 files changed, 70 insertions, 54 deletions

diff --git a/src/treebuilder/after_head.c b/src/treebuilder/after_head.c
index bd6e004..40955a8 100644
--- a/src/treebuilder/after_head.c
+++ b/src/treebuilder/after_head.c
@@ -26,7 +26,6 @@ hubbub_error handle_after_head(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err = HUBBUB_OK;
 	bool handled = false;
-
 	switch (token->type) {
 	case HUBBUB_TOKEN_CHARACTER:
 		err = process_characters_expect_whitespace(treebuilder,
diff --git a/src/treebuilder/in_body.c b/src/treebuilder/in_body.c
index 3e8650d..33ab30a 100644
--- a/src/treebuilder/in_body.c
+++ b/src/treebuilder/in_body.c
@@ -272,7 +272,7 @@ hubbub_error process_start_tag(hubbub_treebuilder *treebuilder,
 			type == FIGURE || type == FOOTER ||
 			type == HEADER || type == MENU || type == NAV ||
 			type == OL || type == P || type == SECTION ||
-			type == UL) {
+			type == SUMMARY || type == UL) {
 		err = process_container_in_body(treebuilder, token);
 	} else if (type == H1 || type == H2 || type == H3 ||
 			type == H4 || type == H5 || type == H6) {
@@ -565,7 +565,7 @@ hubbub_error process_container_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err;
 
-	if (element_in_scope(treebuilder, P, false, true)) {
+	if (element_in_scope(treebuilder, P, BUTTON_SCOPE)) {
 		err = close_p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
@@ -586,8 +586,8 @@ hubbub_error process_hN_in_body(hubbub_treebuilder *treebuilder,
 	hubbub_error err;
 	element_type type;
 
-	if (element_in_scope(treebuilder, P, false, false)) {
-		err = process_0p_in_body(treebuilder);
+	if (element_in_scope(treebuilder, P, BUTTON_SCOPE)) {
+		err = close_p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
 	}
@@ -628,7 +628,7 @@ hubbub_error process_form_in_body(hubbub_treebuilder *treebuilder,
 	if (treebuilder->context.form_element != NULL) {
 		/** \todo parse error */
 	} else {
-		if (element_in_scope(treebuilder, P, false, false)) {
+		if (element_in_scope(treebuilder, P, NONE)) {
 			err = process_0p_in_body(treebuilder);
 			if (err != HUBBUB_OK)
 				return err;
@@ -669,12 +669,6 @@ hubbub_error process_dd_dt_li_in_body(hubbub_treebuilder *treebuilder,
 
 	treebuilder->context.frameset_ok = false;
 
-	if (element_in_scope(treebuilder, P, false, false)) {
-		err = process_0p_in_body(treebuilder);
-		if (err != HUBBUB_OK)
-			return err;
-	}
-
 	/* Find last LI/(DD,DT) on stack, if any */
 	for (node = treebuilder->context.current_node; node > 0; node--) {
 		element_type ntype = stack[node].type;
@@ -689,13 +683,16 @@ hubbub_error process_dd_dt_li_in_body(hubbub_treebuilder *treebuilder,
 		if (!is_formatting_element(ntype) &&
 				!is_phrasing_element(ntype) &&
 				ntype != ADDRESS && 
-				ntype != DIV)
+				ntype != DIV &&
+				ntype != P)
 			break;
 	}
 
 	/* If we found one, then pop all nodes up to and including it */
 	if (stack[node].type == LI || stack[node].type == DD ||
 			stack[node].type == DT) {
+
+		close_implied_end_tags(treebuilder, type);
 		/* Check that we're only popping one node 
 		 * and emit a parse error if not */
 		if (treebuilder->context.current_node > node) {
@@ -717,6 +714,11 @@ hubbub_error process_dd_dt_li_in_body(hubbub_treebuilder *treebuilder,
 		} while (treebuilder->context.current_node >= node);
 	}
 
+	if(element_in_scope(treebuilder, P, BUTTON_SCOPE)) {
+		err = close_p_in_body(treebuilder);
+		if(err != HUBBUB_OK)
+			return err;
+	}
 	return insert_element(treebuilder, &token->data.tag, true);
 }
 
@@ -732,8 +734,8 @@ hubbub_error process_plaintext_in_body(hubbub_treebuilder *treebuilder,
 	hubbub_error err;
 	hubbub_tokeniser_optparams params;
 
-	if (element_in_scope(treebuilder, P, false, false)) {
-		err = process_0p_in_body(treebuilder);
+	if (element_in_scope(treebuilder, P, BUTTON_SCOPE)) {
+		err = close_p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
 	}
@@ -920,7 +922,7 @@ hubbub_error process_nobr_in_body(hubbub_treebuilder *treebuilder,
 	if (err != HUBBUB_OK)
 		return err;
 
-	if (element_in_scope(treebuilder, NOBR, false, false)) {
+	if (element_in_scope(treebuilder, NOBR, NONE)) {
 		/** \todo parse error */
 
 		/* Act as if </nobr> were seen */
@@ -982,7 +984,7 @@ hubbub_error process_button_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err;
 
-	if (element_in_scope(treebuilder, BUTTON, false, false)) {
+	if (element_in_scope(treebuilder, BUTTON, NONE)) {
 		/** \todo parse error */
 
 		/* Act as if </button> has been seen */
@@ -1071,7 +1073,7 @@ hubbub_error process_hr_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err;
 
-	if (element_in_scope(treebuilder, P, false, false)) {
+	if (element_in_scope(treebuilder, P, BUTTON_SCOPE)) {
 		err = process_0p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
@@ -1333,7 +1335,7 @@ hubbub_error process_opt_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err; 
 
-	if (element_in_scope(treebuilder, OPTION, false, false)) {
+	if (element_in_scope(treebuilder, OPTION, NONE)) {
 		err = process_0generic_in_body(treebuilder, OPTION);
 		/* Cannot fail */
 		assert(err == HUBBUB_OK);
@@ -1374,7 +1376,7 @@ hubbub_error process_0body_in_body(hubbub_treebuilder *treebuilder)
 {
 	hubbub_error err = HUBBUB_OK;
 
-	if (!element_in_scope(treebuilder, BODY, false, false)) {
+	if (!element_in_scope(treebuilder, BODY, NONE)) {
 		/** \todo parse error */
 	} else {
 		element_context *stack = treebuilder->context.element_stack;
@@ -1413,7 +1415,7 @@ hubbub_error process_0body_in_body(hubbub_treebuilder *treebuilder)
 hubbub_error process_0container_in_body(hubbub_treebuilder *treebuilder,
 		element_type type)
 {
-	if (!element_in_scope(treebuilder, type, false, false)) {
+	if (!element_in_scope(treebuilder, type, NONE)) {
 		/** \todo parse error */
 	} else {
 		uint32_t popped = 0;
@@ -1458,7 +1460,7 @@ hubbub_error process_0form_in_body(hubbub_treebuilder *treebuilder)
 				treebuilder->context.form_element);
 	treebuilder->context.form_element = NULL;
 
-	idx = element_in_scope(treebuilder, FORM, false, false);
+	idx = element_in_scope(treebuilder, FORM, NONE);
 
 	if (idx == 0 || node == NULL || 
 			treebuilder->context.element_stack[idx].node != node) {
@@ -1550,7 +1552,7 @@ hubbub_error close_p_in_body(hubbub_treebuilder *treebuilder)
 hubbub_error process_0p_in_body(hubbub_treebuilder *treebuilder)
 {
 
-	if (!element_in_scope(treebuilder, P, false, true)) {
+	if (!element_in_scope(treebuilder, P, BUTTON_SCOPE)) {
 		/** \todo parse error */
 
 		hubbub_tag tag;
@@ -1580,7 +1582,7 @@ hubbub_error process_0p_in_body(hubbub_treebuilder *treebuilder)
 hubbub_error process_0dd_dt_li_in_body(hubbub_treebuilder *treebuilder,
 		element_type type)
 {
-	if (!element_in_scope(treebuilder, type, false, false)) {
+	if (!element_in_scope(treebuilder, type, LIST_ITEM_SCOPE)) {
 		/** \todo parse error */
 	} else {
 		uint32_t popped = 0;
@@ -1622,12 +1624,12 @@ hubbub_error process_0h_in_body(hubbub_treebuilder *treebuilder,
 	UNUSED(type);
 
 	/** \todo optimise this */
-	if (element_in_scope(treebuilder, H1, false, false) ||
-			element_in_scope(treebuilder, H2, false, false) ||
-			element_in_scope(treebuilder, H3, false, false) ||
-			element_in_scope(treebuilder, H4, false, false) ||
-			element_in_scope(treebuilder, H5, false, false) ||
-			element_in_scope(treebuilder, H6, false, false)) {
+	if (element_in_scope(treebuilder, H1, NONE) ||
+			element_in_scope(treebuilder, H2, NONE) ||
+			element_in_scope(treebuilder, H3, NONE) ||
+			element_in_scope(treebuilder, H4, NONE) ||
+			element_in_scope(treebuilder, H5, NONE) ||
+			element_in_scope(treebuilder, H6, NONE)) {
 		uint32_t popped = 0;
 		element_type otype;
 
@@ -1870,7 +1872,7 @@ hubbub_error aa_find_and_validate_formatting_element(
 
 	if (entry->stack_index != 0 &&
 			element_in_scope(treebuilder, entry->details.type,
-					false, false) != entry->stack_index) {
+					NONE) != entry->stack_index) {
 		/** \todo parse error */
 		return HUBBUB_OK;
 	}
@@ -2322,7 +2324,7 @@ hubbub_error aa_insert_into_foster_parent(hubbub_treebuilder *treebuilder,
 hubbub_error process_0applet_button_marquee_object_in_body(
 		hubbub_treebuilder *treebuilder, element_type type)
 {
-	if (!element_in_scope(treebuilder, type, false, false)) {
+	if (!element_in_scope(treebuilder, type, NONE)) {
 		/** \todo parse error */
 	} else {
 		uint32_t popped = 0;
diff --git a/src/treebuilder/in_cell.c b/src/treebuilder/in_cell.c
index 6787c51..90bdf53 100644
--- a/src/treebuilder/in_cell.c
+++ b/src/treebuilder/in_cell.c
@@ -27,7 +27,7 @@ static inline void close_cell(hubbub_treebuilder *treebuilder)
 
 	element_type type;
 
-	if (element_in_scope(treebuilder, TD, true, false)) {
+	if (element_in_scope(treebuilder, TD, TABLE_SCOPE)) {
 		type = TD;
 	} else {
 		type = TH;
@@ -89,7 +89,7 @@ hubbub_error handle_in_cell(hubbub_treebuilder *treebuilder,
 				&token->data.tag.name);
 
 		if (type == TH || type == TD) {
-			if (element_in_scope(treebuilder, type, true, false)) {
+			if (element_in_scope(treebuilder, type, TABLE_SCOPE)) {
 				hubbub_ns ns;
 				element_type otype = UNKNOWN;
 				void *node;
@@ -118,7 +118,7 @@ hubbub_error handle_in_cell(hubbub_treebuilder *treebuilder,
 			/** \todo parse error */
 		} else if (type == TABLE  || type == TBODY || type == TFOOT ||
 				type == THEAD || type == TR) {
-			if (element_in_scope(treebuilder, type, true, false)) {
+			if (element_in_scope(treebuilder, type, TABLE_SCOPE)) {
 				close_cell(treebuilder);
 				err = HUBBUB_REPROCESS;
 			} else {
diff --git a/src/treebuilder/in_select.c b/src/treebuilder/in_select.c
index 9f06b8f..a49b568 100644
--- a/src/treebuilder/in_select.c
+++ b/src/treebuilder/in_select.c
@@ -86,7 +86,7 @@ hubbub_error handle_in_select(hubbub_treebuilder *treebuilder,
 		} else if (type == SELECT || type == INPUT ||
 				type == TEXTAREA) {
 
-			if (element_in_scope(treebuilder, SELECT, true, false)) {
+			if (element_in_scope(treebuilder, SELECT, TABLE_SCOPE)) {
 				element_stack_pop_until(treebuilder, 
 						SELECT);
 				reset_insertion_mode(treebuilder);
@@ -142,7 +142,7 @@ hubbub_error handle_in_select(hubbub_treebuilder *treebuilder,
 				/** \todo parse error */
 			}
 		} else if (type == SELECT) {
-			if (element_in_scope(treebuilder, SELECT, true, false)) {
+			if (element_in_scope(treebuilder, SELECT, TABLE_SCOPE)) {
 				element_stack_pop_until(treebuilder, 
 						SELECT);
 				reset_insertion_mode(treebuilder);
diff --git a/src/treebuilder/in_select_in_table.c b/src/treebuilder/in_select_in_table.c
index 69d4203..d6e1b8c 100644
--- a/src/treebuilder/in_select_in_table.c
+++ b/src/treebuilder/in_select_in_table.c
@@ -41,7 +41,7 @@ hubbub_error handle_in_select_in_table(hubbub_treebuilder *treebuilder,
 
 			if ((token->type == HUBBUB_TOKEN_END_TAG &&
 					element_in_scope(treebuilder, type,
-							true, false)) ||
+							TABLE_SCOPE)) ||
 					token->type == HUBBUB_TOKEN_START_TAG) {
 				/** \todo fragment case */
 
diff --git a/src/treebuilder/in_table_body.c b/src/treebuilder/in_table_body.c
index 7d4fb0f..c8c26cf 100644
--- a/src/treebuilder/in_table_body.c
+++ b/src/treebuilder/in_table_body.c
@@ -50,9 +50,9 @@ static void table_clear_stack(hubbub_treebuilder *treebuilder)
  */
 static hubbub_error table_sub_start_or_table_end(hubbub_treebuilder *treebuilder)
 {
-	if (element_in_scope(treebuilder, TBODY, true, false) ||
-			element_in_scope(treebuilder, THEAD, true, false) ||
-			element_in_scope(treebuilder, TFOOT, true, false)) {
+	if (element_in_scope(treebuilder, TBODY, TABLE_SCOPE) ||
+			element_in_scope(treebuilder, THEAD, TABLE_SCOPE) ||
+			element_in_scope(treebuilder, TFOOT, TABLE_SCOPE)) {
 		hubbub_ns ns;
 		element_type otype;
 		void *node;
@@ -140,7 +140,7 @@ hubbub_error handle_in_table_body(hubbub_treebuilder *treebuilder,
 				&token->data.tag.name);
 
 		if (type == TBODY || type == TFOOT || type == THEAD) {
-			if (!element_in_scope(treebuilder, type, true, false)) {
+			if (!element_in_scope(treebuilder, type, TABLE_SCOPE)) {
 				/** \todo parse error */
 				/* Ignore the token */
 			} else {
diff --git a/src/treebuilder/internal.h b/src/treebuilder/internal.h
index d87e89e..ccec4ed 100644
--- a/src/treebuilder/internal.h
+++ b/src/treebuilder/internal.h
@@ -36,6 +36,11 @@ typedef enum
 	UNKNOWN
 } element_type;
 
+typedef enum
+{
+	NONE, LIST_ITEM_SCOPE, BUTTON_SCOPE, TABLE_SCOPE, SELECT_SCOPE
+} element_scope;
+
 /**
  * Item on the element stack
  */
@@ -137,7 +142,7 @@ hubbub_error parse_generic_rcdata(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token, bool rcdata);
 
 uint32_t element_in_scope(hubbub_treebuilder *treebuilder,
-		element_type type, bool in_table, bool in_button);
+		element_type type, element_scope scope);
 hubbub_error reconstruct_active_formatting_list(
 		hubbub_treebuilder *treebuilder);
 void clear_active_formatting_list_to_marker(
diff --git a/src/treebuilder/treebuilder.c b/src/treebuilder/treebuilder.c
index 45197b1..9e5734b 100644
--- a/src/treebuilder/treebuilder.c
+++ b/src/treebuilder/treebuilder.c
@@ -495,12 +495,11 @@ hubbub_error parse_generic_rcdata(hubbub_treebuilder *treebuilder,
  *
  * \param treebuilder  Treebuilder to look in
  * \param type         Element type to find
- * \param in_table     Whether we're looking in table scope
- * \param in_button	Whether we're looking in button scope
+ * \param scope		The scope we're looking in
  * \return Element stack index, or 0 if not in scope
  */
 uint32_t element_in_scope(hubbub_treebuilder *treebuilder,
-		element_type type, bool in_table, bool in_button)
+		element_type type, element_scope scope)
 {
 	uint32_t node;
 
@@ -518,17 +517,28 @@ uint32_t element_in_scope(hubbub_treebuilder *treebuilder,
 		if (node_type == type)
 			return node;
 
-		if (node_type == TABLE)
+		if (scope == TABLE_SCOPE && (
+				node_type == TABLE ||
+				node_type == HTML ||
+				node_type == TEMPLATE))
 			break;
 
-		/* The list of element types given in the spec here are the
-		 * scoping elements excluding TABLE and HTML and BUTTON. TABLE is handled
-		 * in the previous conditional and HTML should only occur
+		if (scope == SELECT_SCOPE &&
+					node_type != OPTGROUP &&
+					node_type != OPTION)
+			break;
+
+		/* The list of element types of the normal scope given in the spec here are the
+		 * scoping elements excluding HTML and BUTTON. HTML should only occur
 		 * as the first node in the stack, which is never processed
 		 * in this loop. */
-		if (!in_table && (node_type != BUTTON || (
-						in_button && node_type == BUTTON)) &&
-				(is_scoping_element(node_type) ||
+		if (scope != TABLE_SCOPE && scope != SELECT_SCOPE &&
+				((scope == BUTTON_SCOPE &&
+						node_type == BUTTON) ||
+				(scope == LIST_ITEM_SCOPE &&
+				 (node_type == OL ||
+				  node_type == UL)) ||
+				(node_type != BUTTON && is_scoping_element(node_type)) ||
 				(node_ns == HUBBUB_NS_SVG && (
 							      node_type == FOREIGNOBJECT ||
 							      node_type == DESC ||
diff --git a/test/data/tree-construction/INDEX b/test/data/tree-construction/INDEX
index 49f0d0c..7504628 100644
--- a/test/data/tree-construction/INDEX
+++ b/test/data/tree-construction/INDEX
@@ -36,7 +36,7 @@ adoption02.dat		html5lib tests for the adoption agency algorithm
 #tests17.dat		NA
 #tests18.dat		NA
 #tests19.dat		NA
-#tests1.dat		NA
+tests1.dat		html5lib treebuilder tests
 #tests20.dat		NA
 #tests21.dat		NA
 #tests22.dat		NA