Add "in foreign content" handling. Not convinced this is the best way.

svn path=/trunk/hubbub/; revision=4475

5 files changed, 152 insertions, 5 deletions

diff --git a/src/treebuilder/Makefile b/src/treebuilder/Makefile
index f2e8e3a..3b736b2 100644
--- a/src/treebuilder/Makefile
+++ b/src/treebuilder/Makefile
@@ -37,6 +37,7 @@ SRCS_$(d) := treebuilder.c \
 		in_head_noscript.c after_head.c in_body.c \
 		in_caption.c in_column_group.c in_table_body.c in_row.c \
 		in_cell.c in_select.c in_select_in_table.c \
+		in_foreign_content.c \
 		generic_rcdata.c script_collect.c
 
 # Append to sources for component
diff --git a/src/treebuilder/in_foreign_content.c b/src/treebuilder/in_foreign_content.c
new file mode 100644
index 0000000..d1338e7
--- /dev/null
+++ b/src/treebuilder/in_foreign_content.c
@@ -0,0 +1,139 @@
+/*
+ * This file is part of Hubbub.
+ * Licensed under the MIT License,
+ *                http://www.opensource.org/licenses/mit-license.php
+ * Copyright 2008 Andrew Sidwell
+ */
+
+#include <assert.h>
+#include <string.h>
+
+#include "treebuilder/modes.h"
+#include "treebuilder/internal.h"
+#include "treebuilder/treebuilder.h"
+#include "utils/utils.h"
+
+
+
+static bool element_in_scope_in_non_html_ns(hubbub_treebuilder *treebuilder)
+{
+	uint32_t node;
+
+	if (treebuilder->context.element_stack == NULL)
+		return false;
+
+	for (node = treebuilder->context.current_node; node > 0; node--) {
+		element_type node_ns =
+				treebuilder->context.element_stack[node].ns;
+
+		if (node_ns != HTML)
+			return true;
+	}
+
+	return false;
+}
+
+
+
+/**
+ * Handle tokens in "in foreign content" insertion mode
+ *
+ * \param treebuilder  The treebuilder instance
+ * \param token        The token to process
+ * \return True to reprocess the token, false otherwise
+ */
+bool handle_in_foreign_content(hubbub_treebuilder *treebuilder,
+		const hubbub_token *token)
+{
+	bool reprocess = false;
+
+	switch (token->type) {
+	case HUBBUB_TOKEN_CHARACTER:
+		append_text(treebuilder, &token->data.character);
+		break;
+	case HUBBUB_TOKEN_COMMENT:
+		process_comment_append(treebuilder, token,
+				treebuilder->context.element_stack[
+				treebuilder->context.current_node].node);
+		break;
+	case HUBBUB_TOKEN_DOCTYPE:
+		/** \todo parse error */
+		break;
+	case HUBBUB_TOKEN_START_TAG:
+	{
+		element_type type = element_type_from_name(treebuilder,
+				&token->data.tag.name);
+
+		element_type cur_node = current_node(treebuilder);
+		hubbub_ns cur_node_ns = current_node_ns(treebuilder);
+
+		if (cur_node_ns == HUBBUB_NS_HTML ||
+				(cur_node_ns == HUBBUB_NS_MATHML &&
+				(type != MGLYPH && type != MALIGNMARK) &&
+				(cur_node == MI || cur_node == MO ||
+				cur_node == MN || cur_node == MS ||
+				cur_node == MTEXT))) {
+			treebuilder->context.mode =
+					treebuilder->context.second_mode;
+			hubbub_treebuilder_token_handler(token, treebuilder);
+
+			if (treebuilder->context.mode == IN_FOREIGN_CONTENT &&
+					!element_in_scope_in_non_html_ns(treebuilder)) {
+				treebuilder->context.mode =
+						treebuilder->context.second_mode;
+			}
+		} else if (type == B || type ==  BIG || type == BLOCKQUOTE ||
+				type == BODY || type == BR || type == CENTER ||
+				type == CODE || type == DD || type == DIV ||
+				type == DL || type == DT || type == EM ||
+				type == EMBED || type == FONT || type == H1 ||
+				type == H2 || type == H3 || type == H4 ||
+				type == H5 || type == H6 || type == HEAD ||
+				type == HR || type == I || type == IMG ||
+				type == LI || type == LISTING ||
+				type == MENU || type == META || type == NOBR ||
+				type == OL || type == P || type == PRE ||
+				type == RUBY || type == S || type == SMALL ||
+				type == SPAN || type == STRONG ||
+				type == STRIKE || type == SUB || type == SUP ||
+				type == TABLE || type == TT || type == U ||
+				type == UL || type == VAR) {
+			/** \todo parse error */
+
+			while (cur_node_ns != HUBBUB_NS_HTML) {
+				void *node;
+				element_stack_pop(treebuilder, &cur_node_ns,
+						&cur_node, &node);
+				cur_node_ns = current_node_ns(treebuilder);
+			}
+
+			treebuilder->context.mode =
+					treebuilder->context.second_mode;
+		} else {
+			hubbub_tag tag = token->data.tag;
+
+			adjust_foreign_attributes(treebuilder, &tag);
+
+			/* Set to the right namespace and insert */
+			tag.ns = cur_node_ns;
+
+			if (token->data.tag.self_closing) {
+				insert_element_no_push(treebuilder, &tag);
+				/** \todo ack sc flag */
+			} else {
+				insert_element(treebuilder, &tag);
+			}
+		}
+	}
+		break;
+	case HUBBUB_TOKEN_END_TAG:
+		/** \parse error */
+		break;
+	case HUBBUB_TOKEN_EOF:
+		reprocess = true;
+		break;
+	}
+
+	return reprocess;
+}
+
diff --git a/src/treebuilder/internal.h b/src/treebuilder/internal.h
index 6f7278c..f9fd09e 100644
--- a/src/treebuilder/internal.h
+++ b/src/treebuilder/internal.h
@@ -27,7 +27,9 @@ typedef enum
 	A, B, BIG, EM, FONT, I, NOBR, S, SMALL, STRIKE, STRONG, TT, U,
 /* Phrasing */
 	/**< \todo Enumerate phrasing elements */
-	LABEL, MATH, RP, RT, XMP,
+	CODE, LABEL, RP, RT, RUBY, SPAN, SUB, SUP, VAR, XMP,
+/* MathML */
+	MATH, MGLYPH, MALIGNMARK, MI, MO, MN, MS, MTEXT,
 	UNKNOWN,
 } element_type;
 
@@ -104,6 +106,8 @@ struct hubbub_treebuilder
 	void *alloc_pw;			/**< Client private data */
 };
 
+void hubbub_treebuilder_token_handler(const hubbub_token *token, void *pw);
+
 bool process_characters_expect_whitespace(
 		hubbub_treebuilder *treebuilder, const hubbub_token *token,
 		bool insert_into_current_node);
@@ -142,6 +146,7 @@ bool element_stack_pop(hubbub_treebuilder *treebuilder,
 bool element_stack_pop_until(hubbub_treebuilder *treebuilder,
 		element_type type);
 element_type current_node(hubbub_treebuilder *treebuilder);
+hubbub_ns current_node_ns(hubbub_treebuilder *treebuilder);
 element_type prev_node(hubbub_treebuilder *treebuilder);
 
 bool formatting_list_append(hubbub_treebuilder *treebuilder,
diff --git a/src/treebuilder/modes.h b/src/treebuilder/modes.h
index c97e9de..8172ddd 100644
--- a/src/treebuilder/modes.h
+++ b/src/treebuilder/modes.h
@@ -59,6 +59,8 @@ bool handle_in_caption(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
 bool handle_in_column_group(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
+bool handle_in_table_body(hubbub_treebuilder *treebuilder,
+		const hubbub_token *token);
 bool handle_in_row(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
 bool handle_in_cell(hubbub_treebuilder *treebuilder,
@@ -71,6 +73,8 @@ bool handle_generic_rcdata(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
 bool handle_script_collect_characters(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
+bool handle_in_foreign_content(hubbub_treebuilder *treebuilder,
+		const hubbub_token *token);
 
 bool process_in_head(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token);
diff --git a/src/treebuilder/treebuilder.c b/src/treebuilder/treebuilder.c
index cef73d3..3c25540 100644
--- a/src/treebuilder/treebuilder.c
+++ b/src/treebuilder/treebuilder.c
@@ -66,12 +66,10 @@ static const struct {
 
 static void hubbub_treebuilder_buffer_handler(const uint8_t *data,
 		size_t len, void *pw);
-static void hubbub_treebuilder_token_handler(const hubbub_token *token, 
-		void *pw);
 
 
 /**
- * Create a hubbub treebuilder 
+ * Create a hubbub treebuilder
  *
  * \param tokeniser  Underlying tokeniser instance
  * \param alloc      Memory (de)allocation function
@@ -109,7 +107,7 @@ hubbub_treebuilder *hubbub_treebuilder_create(hubbub_tokeniser *tokeniser,
 		return NULL;
 	}
 	tb->context.stack_alloc = ELEMENT_STACK_CHUNK;
-	/* We rely on HTML not being equal to zero to determine 
+	/* We rely on HTML not being equal to zero to determine
 	 * if the first item in the stack is in use. Assert this here. */
 	assert(HTML != 0);
 	tb->context.element_stack[0].type = 0;