run "pending spec changes" tests too

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

diff --git a/parse-html.coffee b/parse-html.coffee
index adb9bab..20bc99c 100644 (file)
--- a/parse-html.coffee
+++ b/parse-html.coffee
@@ -1924,11 +1924,7 @@ parse_html = (args) ->
                if t.type is TYPE_START_TAG and (t.name is 'pre' or t.name is 'listing')
                        close_p_if_in_button_scope()
                        insert_html_element t
-                       # spec: If the next token is a "LF" (U+000A) character token, then
-                       # ignore that token and move on to the next one. (Newlines at the
-                       # start of pre blocks are ignored as an authoring convenience.)
-                       if txt.charAt(cur) is "\u000a" # FIXME check for crlf?
-                               cur += 1
+                       eat_next_token_if_newline()
                        flag_frameset_ok = false
                        return
                if t.type is TYPE_START_TAG and t.name is 'form'
@@ -2161,7 +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()
@@ -2239,8 +2236,7 @@ parse_html = (args) ->
                        return
                if t.type is TYPE_START_TAG and t.name is 'textarea'
                        insert_html_element t
-                       if txt.charAt(cur) is "\u000a" # FIXME check for crlf?
-                               cur += 1
+                       eat_next_token_if_newline()
                        tok_state = tok_state_rcdata
                        original_ins_mode = ins_mode
                        flag_frameset_ok = false
@@ -2771,7 +2767,7 @@ parse_html = (args) ->
                        return
                if t.type is TYPE_START_TAG and (t.name is 'input' or t.name is 'keygen' or t.name is 'textarea')
                        parse_error()
-                       if is_in_select_scope 'select', NS_HTML
+                       unless is_in_select_scope 'select', NS_HTML
                                return
                        loop
                                el = open_els.shift()
@@ -3097,7 +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
@@ -3794,7 +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
@@ -4547,6 +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
@@ -4640,6 +4637,24 @@ parse_html = (args) ->
                                        return '&'
                return # never reached
 
+       eat_next_token_if_newline = ->
+               old_cur = cur
+               t = null
+               until t?
+                       t = tok_state()
+               if t.type is TYPE_TEXT
+                       # definition of a newline depends on whether it was a character ref or not
+                       if cur - old_cur is 1
+                               # not a character reference
+                               if t.text is "\u000d" or t.text is "\u000a"
+                                       return
+                       else
+                               if t.text is "\u000a"
+                                       return
+               # not a "newline"
+               cur = old_cur
+               return
+
        # tree constructor initialization
        # see comments on TYPE_TAG/etc for the structure of this data
        txt = args.html
@@ -4668,19 +4683,20 @@ 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
-       while flag_parsing
-               t = tok_state()
-               if t?
-                       process_token t
-                       # fixfull parse error if has self-closing flag, but it wasn't acknolwedged
+       parse_main_loop = ->
+               while flag_parsing
+                       t = tok_state()
+                       if t?
+                               process_token t
+                               # fixfull parse error if has self-closing flag, but it wasn't acknolwedged
+       parse_main_loop()
        return doc.children
 
 serialize_els = (els, shallow, show_ids) ->