handle nulls properly

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

diff --git a/parse-html.coffee b/parse-html.coffee
index 487fe99..20bc99c 100644 (file)
--- a/parse-html.coffee
+++ b/parse-html.coffee
@@ -2157,7 +2157,8 @@ parse_html = (args) ->
                        return
                if t.type is TYPE_END_TAG and t.name is 'br'
                        parse_error()
-                       t.type = TYPE_START_TAG
+                       # W3C: t.type = TYPE_START_TAG
+                       t = new_open_tag 'br' # WHATWG
                        # fall through
                if t.type is TYPE_START_TAG and (t.name is 'area' or t.name is 'br' or t.name is 'embed' or t.name is 'img' or t.name is 'keygen' or t.name is 'wbr')
                        reconstruct_afe()
@@ -3092,7 +3093,7 @@ parse_html = (args) ->
                                tok_state = tok_state_tag_open
                        when "\u0000"
                                parse_error()
-                               return new_text_node "\ufffd"
+                               return new_text_node c
                        when '' # EOF
                                return new_eof_token()
                        else
@@ -3789,7 +3790,7 @@ parse_html = (args) ->
                        return
                if c is '>'
                        tok_state = tok_state_data
-                       return
+                       return tok_cur_tag
                if is_uc_alpha(c)
                        tok_cur_tag.attrs_a.unshift [c.toLowerCase(), '']
                        tok_state = tok_state_attribute_name
@@ -4542,6 +4543,7 @@ parse_html = (args) ->
                else
                        val = txt.substr cur, (next_gt - cur)
                        cur = next_gt + 3
+               val = val.replace(new RegExp("\u0000", 'g'), "\ufffd")
                if val.length > 0
                        return new_character_token val # fixfull split
                return null
@@ -4681,11 +4683,10 @@ parse_html = (args) ->
 
        # text pre-processing
        # FIXME http://www.w3.org/TR/html5/syntax.html#preprocessing-the-input-stream
-       txt = txt.replace(new RegExp("\u0000", 'g'), "\ufffd") # fixfull spec doesn't say this
        txt = txt.replace(new RegExp("\r\n", 'g'), "\n") # fixfull spec doesn't say this
        txt = txt.replace(new RegExp("\r", 'g'), "\n") # fixfull spec doesn't say this
 
-       if args.name is "tests23.dat #1"
+       if args.name is "webkit01.dat #12"
                console.log "hi"
        # proccess input
        # http://www.w3.org/TR/html5/syntax.html#tree-construction