Fixed The adoption agency & related things

8 files changed, 107 insertions, 47 deletions

diff --git a/src/treebuilder/in_body.c b/src/treebuilder/in_body.c
index e685e0e..3e8650d 100644
--- a/src/treebuilder/in_body.c
+++ b/src/treebuilder/in_body.c
@@ -23,6 +23,7 @@ typedef struct bookmark {
 	formatting_list_entry *next;	/**< Next entry */
 } bookmark;
 
+static hubbub_error close_p_in_body(hubbub_treebuilder *treebuilder);
 static hubbub_error process_character(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
 static hubbub_error process_start_tag(hubbub_treebuilder *treebuilder,
@@ -564,8 +565,8 @@ hubbub_error process_container_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err;
 
-	if (element_in_scope(treebuilder, P, false)) {
-		err = process_0p_in_body(treebuilder);
+	if (element_in_scope(treebuilder, P, false, true)) {
+		err = close_p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
 	}
@@ -585,7 +586,7 @@ hubbub_error process_hN_in_body(hubbub_treebuilder *treebuilder,
 	hubbub_error err;
 	element_type type;
 
-	if (element_in_scope(treebuilder, P, false)) {
+	if (element_in_scope(treebuilder, P, false, false)) {
 		err = process_0p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
@@ -627,7 +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)) {
+		if (element_in_scope(treebuilder, P, false, false)) {
 			err = process_0p_in_body(treebuilder);
 			if (err != HUBBUB_OK)
 				return err;
@@ -668,7 +669,7 @@ hubbub_error process_dd_dt_li_in_body(hubbub_treebuilder *treebuilder,
 
 	treebuilder->context.frameset_ok = false;
 
-	if (element_in_scope(treebuilder, P, false)) {
+	if (element_in_scope(treebuilder, P, false, false)) {
 		err = process_0p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
@@ -731,7 +732,7 @@ hubbub_error process_plaintext_in_body(hubbub_treebuilder *treebuilder,
 	hubbub_error err;
 	hubbub_tokeniser_optparams params;
 
-	if (element_in_scope(treebuilder, P, false)) {
+	if (element_in_scope(treebuilder, P, false, false)) {
 		err = process_0p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
@@ -919,7 +920,7 @@ hubbub_error process_nobr_in_body(hubbub_treebuilder *treebuilder,
 	if (err != HUBBUB_OK)
 		return err;
 
-	if (element_in_scope(treebuilder, NOBR, false)) {
+	if (element_in_scope(treebuilder, NOBR, false, false)) {
 		/** \todo parse error */
 
 		/* Act as if </nobr> were seen */
@@ -981,7 +982,7 @@ hubbub_error process_button_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err;
 
-	if (element_in_scope(treebuilder, BUTTON, false)) {
+	if (element_in_scope(treebuilder, BUTTON, false, false)) {
 		/** \todo parse error */
 
 		/* Act as if </button> has been seen */
@@ -1070,7 +1071,7 @@ hubbub_error process_hr_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err;
 
-	if (element_in_scope(treebuilder, P, false)) {
+	if (element_in_scope(treebuilder, P, false, false)) {
 		err = process_0p_in_body(treebuilder);
 		if (err != HUBBUB_OK)
 			return err;
@@ -1332,7 +1333,7 @@ hubbub_error process_opt_in_body(hubbub_treebuilder *treebuilder,
 {
 	hubbub_error err; 
 
-	if (element_in_scope(treebuilder, OPTION, false)) {
+	if (element_in_scope(treebuilder, OPTION, false, false)) {
 		err = process_0generic_in_body(treebuilder, OPTION);
 		/* Cannot fail */
 		assert(err == HUBBUB_OK);
@@ -1373,7 +1374,7 @@ hubbub_error process_0body_in_body(hubbub_treebuilder *treebuilder)
 {
 	hubbub_error err = HUBBUB_OK;
 
-	if (!element_in_scope(treebuilder, BODY, false)) {
+	if (!element_in_scope(treebuilder, BODY, false, false)) {
 		/** \todo parse error */
 	} else {
 		element_context *stack = treebuilder->context.element_stack;
@@ -1412,7 +1413,7 @@ 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)) {
+	if (!element_in_scope(treebuilder, type, false, false)) {
 		/** \todo parse error */
 	} else {
 		uint32_t popped = 0;
@@ -1457,7 +1458,7 @@ 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);
+	idx = element_in_scope(treebuilder, FORM, false, false);
 
 	if (idx == 0 || node == NULL || 
 			treebuilder->context.element_stack[idx].node != node) {
@@ -1486,25 +1487,27 @@ hubbub_error process_0form_in_body(hubbub_treebuilder *treebuilder)
 	return HUBBUB_OK;
 }
 
-
 /**
- * Process a p end tag as if in "in body"
- *
+ * Close a p Element by repeated popping
  * \param treebuilder  The treebuilder instance
  */
-hubbub_error process_0p_in_body(hubbub_treebuilder *treebuilder)
+hubbub_error close_p_in_body(hubbub_treebuilder *treebuilder)
 {
 	hubbub_error err = HUBBUB_OK;
 	uint32_t popped = 0;
 
-	if (treebuilder->context.element_stack[
+	close_implied_end_tags(treebuilder, P);
+
+	if(treebuilder->context.element_stack[
 			treebuilder->context.current_node].type != P) {
-		/** \todo parse error */
+		/* todo parse error */
+
 	}
 
-	while (element_in_scope(treebuilder, P, false)) {
+	element_type type;
+
+	do {
 		hubbub_ns ns;
-		element_type type;
 		void *node;
 
 		err = element_stack_pop(treebuilder, &ns, &type, &node);
@@ -1514,7 +1517,7 @@ hubbub_error process_0p_in_body(hubbub_treebuilder *treebuilder)
 				treebuilder->tree_handler->ctx, node);
 
 		popped++;
-	}
+	} while (type != P);
 
 	if (popped == 0) {
 		hubbub_token dummy;
@@ -1536,11 +1539,39 @@ hubbub_error process_0p_in_body(hubbub_treebuilder *treebuilder)
 		/* Cannot fail */
 		assert(err == HUBBUB_OK);
 	}
-
 	return err;
 }
 
 /**
+ * Process a p end tag as if in "in body"
+ *
+ * \param treebuilder  The treebuilder instance
+ */
+hubbub_error process_0p_in_body(hubbub_treebuilder *treebuilder)
+{
+
+	if (!element_in_scope(treebuilder, P, false, true)) {
+		/** \todo parse error */
+
+		hubbub_tag tag;
+		hubbub_error err;
+
+		tag.ns = HUBBUB_NS_HTML;
+		tag.name.ptr = (const uint8_t *) "p";
+		tag.name.len = SLEN("p");
+		tag.n_attributes = 0;
+		tag.attributes = NULL;
+
+		err = insert_element(treebuilder, &tag, true);
+		if(err != HUBBUB_OK)
+			return err;
+
+	}
+
+	return close_p_in_body(treebuilder);
+}
+
+/**
  * Process a dd, dt, or li end tag as if in "in body"
  *
  * \param treebuilder  The treebuilder instance
@@ -1549,7 +1580,7 @@ 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)) {
+	if (!element_in_scope(treebuilder, type, false, false)) {
 		/** \todo parse error */
 	} else {
 		uint32_t popped = 0;
@@ -1591,12 +1622,12 @@ hubbub_error process_0h_in_body(hubbub_treebuilder *treebuilder,
 	UNUSED(type);
 
 	/** \todo optimise this */
-	if (element_in_scope(treebuilder, H1, false) ||
-			element_in_scope(treebuilder, H2, false) ||
-			element_in_scope(treebuilder, H3, false) ||
-			element_in_scope(treebuilder, H4, false) ||
-			element_in_scope(treebuilder, H5, false) ||
-			element_in_scope(treebuilder, H6, false)) {
+	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)) {
 		uint32_t popped = 0;
 		element_type otype;
 
@@ -1839,7 +1870,7 @@ hubbub_error aa_find_and_validate_formatting_element(
 
 	if (entry->stack_index != 0 &&
 			element_in_scope(treebuilder, entry->details.type,
-					false) != entry->stack_index) {
+					false, false) != entry->stack_index) {
 		/** \todo parse error */
 		return HUBBUB_OK;
 	}
@@ -2291,7 +2322,7 @@ 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)) {
+	if (!element_in_scope(treebuilder, type, false, false)) {
 		/** \todo parse error */
 	} else {
 		uint32_t popped = 0;
diff --git a/src/treebuilder/in_cell.c b/src/treebuilder/in_cell.c
index 416f988..6787c51 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)) {
+	if (element_in_scope(treebuilder, TD, true, false)) {
 		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)) {
+			if (element_in_scope(treebuilder, type, true, false)) {
 				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)) {
+			if (element_in_scope(treebuilder, type, true, false)) {
 				close_cell(treebuilder);
 				err = HUBBUB_REPROCESS;
 			} else {
diff --git a/src/treebuilder/in_select.c b/src/treebuilder/in_select.c
index ab40c6c..9f06b8f 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)) {
+			if (element_in_scope(treebuilder, SELECT, true, false)) {
 				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)) {
+			if (element_in_scope(treebuilder, SELECT, true, false)) {
 				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 7a299b2..69d4203 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)) ||
+							true, false)) ||
 					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 8d1a6ab..7d4fb0f 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) ||
-			element_in_scope(treebuilder, THEAD, true) ||
-			element_in_scope(treebuilder, TFOOT, true)) {
+	if (element_in_scope(treebuilder, TBODY, true, false) ||
+			element_in_scope(treebuilder, THEAD, true, false) ||
+			element_in_scope(treebuilder, TFOOT, true, false)) {
 		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)) {
+			if (!element_in_scope(treebuilder, type, true, false)) {
 				/** \todo parse error */
 				/* Ignore the token */
 			} else {
diff --git a/src/treebuilder/internal.h b/src/treebuilder/internal.h
index 1b59267..d87e89e 100644
--- a/src/treebuilder/internal.h
+++ b/src/treebuilder/internal.h
@@ -137,7 +137,7 @@ 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);
+		element_type type, bool in_table, bool in_button);
 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 2358ff9..45197b1 100644
--- a/src/treebuilder/treebuilder.c
+++ b/src/treebuilder/treebuilder.c
@@ -496,10 +496,11 @@ 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
  * \return Element stack index, or 0 if not in scope
  */
 uint32_t element_in_scope(hubbub_treebuilder *treebuilder,
-		element_type type, bool in_table)
+		element_type type, bool in_table, bool in_button)
 {
 	uint32_t node;
 
@@ -525,7 +526,8 @@ uint32_t element_in_scope(hubbub_treebuilder *treebuilder,
 		 * in the previous conditional and HTML should only occur
 		 * as the first node in the stack, which is never processed
 		 * in this loop. */
-		if (!in_table && node_type != BUTTON &&
+		if (!in_table && (node_type != BUTTON || (
+						in_button && node_type == BUTTON)) &&
 				(is_scoping_element(node_type) ||
 				(node_ns == HUBBUB_NS_SVG && (
 							      node_type == FOREIGNOBJECT ||
@@ -1304,6 +1306,33 @@ hubbub_error formatting_list_append(hubbub_treebuilder *treebuilder,
 		uint32_t stack_index)
 {
 	formatting_list_entry *entry;
+	uint32_t n_elements = 0;
+	formatting_list_entry *remove_entry;
+	for (entry = treebuilder->context.formatting_list_end;
+			entry != NULL; entry = entry->prev) {
+		/* Assumption: HTML and TABLE elements are not in the list */
+		if (is_scoping_element(entry->details.type))
+			break;
+
+		if(entry->details.type == type &&
+				entry->details.ns == ns)
+		{
+			remove_entry = entry;
+			n_elements += 1;
+		}
+	}
+	if(n_elements == 3) {
+		hubbub_ns ons;
+		element_type otype;
+		void *onode;
+		uint32_t oindex;
+
+		formatting_list_remove(treebuilder, remove_entry,
+				&ons, &otype, &onode, &oindex);
+
+		treebuilder->tree_handler->unref_node(
+				treebuilder->tree_handler->ctx, onode);
+	}
 
 	entry = malloc(sizeof(formatting_list_entry));
 	if (entry == NULL)
diff --git a/test/data/tree-construction/INDEX b/test/data/tree-construction/INDEX
index 7e7e38b..49f0d0c 100644
--- a/test/data/tree-construction/INDEX
+++ b/test/data/tree-construction/INDEX
@@ -2,8 +2,8 @@
 #
 # Test			Description
 
-#adoption01.dat		NA
-#adoption02.dat		NA
+adoption01.dat		html5lib tests for the adoption agency algorithm
+adoption02.dat		html5lib tests for the adoption agency algorithm
 #after-after-body.dat	NA
 #after-after-frameset.dat	NA
 #after-body.dat		NA