- Fix process_character_expect_whitespace()'s idea of whitespace to match current spec.

 - Add a few more asserts (they can't hurt).
 - Fix infinite loop "in frameset"
 - Make the "after body" mode behave properly wrt character data

svn path=/trunk/hubbub/; revision=4543

3 files changed, 50 insertions, 8 deletions

diff --git a/src/treebuilder/after_body.c b/src/treebuilder/after_body.c
index 66bcd06..81c654a 100644
--- a/src/treebuilder/after_body.c
+++ b/src/treebuilder/after_body.c
@@ -28,8 +28,42 @@ bool handle_after_body(hubbub_treebuilder *treebuilder,
 
 	switch (token->type) {
 	case HUBBUB_TOKEN_CHARACTER:
-		reprocess = process_characters_expect_whitespace(treebuilder,
-				token, true);
+	{
+		/* mostly cribbed from process_characters_expect_whitespace */
+
+		const uint8_t *data = treebuilder->input_buffer +
+				token->data.character.data.off;
+
+		size_t len = token->data.character.len;
+		size_t c;
+
+		/** \todo utf-16 */
+
+		/* Scan for whitespace */
+		for (c = 0; c < len; c++) {
+			if (data[c] != 0x09 && data[c] != 0x0A &&
+					data[c] != 0x0C && data[c] != 0x20)
+				break;
+		}
+
+		/* Non-whitespace characters in token, so handle as in body */
+		if (c > 0) {
+			hubbub_token temp = *token;
+			temp.data.character.len = c;
+
+			handle_in_body(treebuilder, &temp);
+		}
+
+		/* Anything else, switch to in body */
+		if (c != len) {
+			/* Update token data to strip leading whitespace */
+			((hubbub_token *) token)->data.character.data.off += c;
+			((hubbub_token *) token)->data.character.len -= c;
+
+			treebuilder->context.mode = IN_BODY;
+			reprocess = true;
+		}
+	}
 		break;
 	case HUBBUB_TOKEN_COMMENT:
 		process_comment_append(treebuilder, token,
diff --git a/src/treebuilder/in_frameset.c b/src/treebuilder/in_frameset.c
index f46431c..465f7e6 100644
--- a/src/treebuilder/in_frameset.c
+++ b/src/treebuilder/in_frameset.c
@@ -28,8 +28,10 @@ bool handle_in_frameset(hubbub_treebuilder *treebuilder,
 
 	switch (token->type) {
 	case HUBBUB_TOKEN_CHARACTER:
-		reprocess = process_characters_expect_whitespace(treebuilder,
-				token, true);
+		if (process_characters_expect_whitespace(treebuilder,
+				token, true)) {
+			/** \todo parser error */
+		}
 		break;
 	case HUBBUB_TOKEN_COMMENT:
 		process_comment_append(treebuilder, token,
diff --git a/src/treebuilder/treebuilder.c b/src/treebuilder/treebuilder.c
index 670c1db..20649e1 100644
--- a/src/treebuilder/treebuilder.c
+++ b/src/treebuilder/treebuilder.c
@@ -407,7 +407,7 @@ void hubbub_treebuilder_token_handler(const hubbub_token *token,
 bool process_characters_expect_whitespace(hubbub_treebuilder *treebuilder,
 		const hubbub_token *token, bool insert_into_current_node)
 {
-	const uint8_t *data = treebuilder->input_buffer + 
+	const uint8_t *data = treebuilder->input_buffer +
 			token->data.character.data.off;
 	size_t len = token->data.character.len;
 	size_t c;
@@ -415,9 +415,8 @@ bool process_characters_expect_whitespace(hubbub_treebuilder *treebuilder,
 	/** \todo UTF-16 */
 
 	for (c = 0; c < len; c++) {
-		if (data[c] != 0x09 && data[c] != 0x0A && 
-				data[c] != 0x0B && data[c] != 0x0C &&
-				data[c] != 0x20)
+		if (data[c] != 0x09 && data[c] != 0x0A &&
+				data[c] != 0x0C && data[c] != 0x20)
 			break;
 	}
 	/* Non-whitespace characters in token, so reprocess */
@@ -713,6 +712,8 @@ void insert_element(hubbub_treebuilder *treebuilder, const hubbub_tag *tag)
 	int success;
 	void *node, *appended;
 
+	/** \todo handle treebuilder->context.in_table_foster */
+
 	success = treebuilder->tree_handler->create_element(
 			treebuilder->tree_handler->ctx, tag, &node);
 	if (success != 0) {
@@ -751,6 +752,8 @@ void insert_element_no_push(hubbub_treebuilder *treebuilder,
 	int success;
 	void *node, *appended;
 
+	/** \todo handle treebuilder->context.in_table_foster */
+
 	success = treebuilder->tree_handler->create_element(
 			treebuilder->tree_handler->ctx, tag, &node);
 	if (success != 0) {
@@ -1081,6 +1084,7 @@ bool element_stack_pop(hubbub_treebuilder *treebuilder,
 	/** \todo reduce allocated stack size once there's enough free */
 
 	treebuilder->context.current_node = slot - 1;
+	assert((signed) treebuilder->context.current_node >= 0);
 
 	return true;
 }
@@ -1105,6 +1109,8 @@ bool element_stack_pop_until(hubbub_treebuilder *treebuilder,
 
 		treebuilder->tree_handler->unref_node(
 				treebuilder->tree_handler->ctx, node);
+
+		assert((signed) treebuilder->context.current_node >= 0);
 	}
 
 	return true;