fix most fragment tests, fix tree bugs

[peach-html5-editor.git] / parse-html.coffee

diff --git a/parse-html.coffee b/parse-html.coffee
index 8b2712d..255892f 100644 (file)
--- a/parse-html.coffee
+++ b/parse-html.coffee
@@ -1395,6 +1395,7 @@ parse_html = (args) ->
                        el = insert_html_element t
                        head_element_pointer = el
                        ins_mode = ins_mode_in_head
+                       return
                if t.type is TYPE_END_TAG
                        if t.name is 'head' or t.name is 'body' or t.name is 'html' or t.name is 'br'
                                # fall through to Anything else below
@@ -1440,7 +1441,7 @@ parse_html = (args) ->
                if t.type is TYPE_START_TAG and t.name is 'title'
                        parse_generic_rcdata_text t
                        return
-               if t.type is TYPE_START_TAG and ((t.name is 'noscript' and flag_scripting) or (t.name is 'noframes' or t.name is 'style'))
+               if t.type is TYPE_START_TAG and ((t.name is 'noscript' and flag_scripting) or t.name is 'noframes' or t.name is 'style')
                        parse_generic_raw_text t
                        return
                if t.type is TYPE_START_TAG and t.name is 'noscript' and flag_scripting is false
@@ -1502,14 +1503,14 @@ parse_html = (args) ->
                if t.type is TYPE_DOCTYPE
                        parse_error()
                        return
-               if t.type is TYPE_START_TAG
+               if t.type is TYPE_START_TAG and t.name is 'html'
                        ins_mode_in_body t
                        return
                if t.type is TYPE_END_TAG and t.name is 'noscript'
                        open_els.shift()
                        ins_mode = ins_mode_in_head
                        return
-               if (t.type is TYPE_TEXT and (t.text is "\t" or t.text is "\u000a" or t.text is "\u000c" or t.text is "\u000d" or t.text is ' ')) or t.type is TYPE_COMMENT or (t.type is TYPE_START_TAG and (t.name is 'basefont' or t.name is 'bgsound' or t.name is 'link' or t.name is 'meta' or t.name is 'noframes' or t.name is 'style'))
+               if is_space_tok(t) or t.type is TYPE_COMMENT or (t.type is TYPE_START_TAG and (t.name is 'basefont' or t.name is 'bgsound' or t.name is 'link' or t.name is 'meta' or t.name is 'noframes' or t.name is 'style'))
                        ins_mode_in_head t
                        return
                if t.type is TYPE_END_TAG and t.name is 'br'
@@ -3735,7 +3736,7 @@ parse_html = (args) ->
                        return
                # Otherwise
                parse_error()
-               tok_cur_tag = new_comment_token '!' # TODO test ("!" right?)
+               tok_cur_tag = new_comment_token ''
                tok_state = tok_state_bogus_comment
                return
 
@@ -3746,6 +3747,7 @@ parse_html = (args) ->
                                tok_state = tok_state_comment_start_dash
                        when "\u0000"
                                parse_error()
+                               tok_state = tok_state_comment
                                return new_character_token "\ufffd"
                        when '>'
                                parse_error()
@@ -3758,6 +3760,7 @@ parse_html = (args) ->
                                return tok_cur_tag
                        else
                                tok_cur_tag.text += c
+                               tok_state = tok_state_comment
                return null
 
        # 8.2.4.47 http://www.w3.org/TR/html5/syntax.html#comment-start-dash-state
@@ -4411,6 +4414,8 @@ parse_html = (args) ->
        # tokenizer initialization
        tok_state = tok_state_data
 
+       if args.name is "one_that_breaks #1"
+               throw "hi" # console.log "hi"
        # proccess input
        # http://www.w3.org/TR/html5/syntax.html#tree-construction
        while flag_parsing