X-Git-Url: https://jasonwoof.com/gitweb/?a=blobdiff_plain;f=parse-html.coffee;h=4c1d3ccbe573c28c30aee3264d0315f29a1c09ed;hb=f63dacda515ec88d2860549c14365ddab33ec1db;hp=d2717066b6c4763f821b869c9a7b6115c1553c86;hpb=7d7b5713e2ae68559b1cfb1fcb86cfc7b83f7967;p=peach-html5-editor.git diff --git a/parse-html.coffee b/parse-html.coffee index d271706..4c1d3cc 100644 --- a/parse-html.coffee +++ b/parse-html.coffee @@ -47,6 +47,12 @@ # 0: a "end of the list", "current node", "bottommost", "last" +# browser +# note: to get this to run outside a browser, you'll have to write a native +# implementation of decode_named_char_ref() +unless module?.exports? + window.wheic = {} + module = exports: window.wheic # Each node is an obect of the Node class. Here are the Node types: TYPE_TAG = 0 # name, {attributes}, [children] @@ -85,6 +91,7 @@ class Node @children = args.children ? [] @namespace = args.namespace ? NS_HTML @parent = args.parent ? null + @token = args.token ? null if args.id? @id = "#{args.id}+" else @@ -93,8 +100,13 @@ class Node # WARNING this doesn't work right on open tags that are still being parsed attrs = {} attrs[k] = v for k, v of @attrs - return new Node @type, name: @name, text: @text, attrs: attrs, namespace: @namespace, id: @id + return new Node @type, name: @name, text: @text, attrs: attrs, namespace: @namespace, id: @id, token: @token acknowledge_self_closing: -> + if @token? + @token.flag 'did_self_close' + else + @flag 'did_self_close', true + flag: -> # fixfull serialize: (shallow = false, show_ids = false) -> # for unit tests ret = '' @@ -152,7 +164,7 @@ new_element = (name) -> new_text_node = (txt) -> return new Node TYPE_TEXT, text: txt new_character_token = new_text_node -new_comment_node = (txt) -> +new_comment_token = (txt) -> return new Node TYPE_COMMENT, text: txt new_eof_token = -> return new Node TYPE_EOF @@ -172,6 +184,10 @@ tag_name_chars = alnum + "-" # http://www.w3.org/TR/html5/infrastructure.html#space-character space_chars = "\u0009\u000a\u000c\u000d\u0020" +is_space = (txt) -> + return txt.length is 1 and space_chars.indexOf(txt) > -1 +is_space_tok = (t) -> + return t.type is TYPE_TEXT && t.text.length is 1 and space_chars.indexOf(t.text) > -1 # https://en.wikipedia.org/wiki/Whitespace_character#Unicode whitespace_chars = "\u0009\u000a\u000b\u000c\u000d\u0020\u0085\u00a0\u1680\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200a\u2028\u2029\u202f\u205f\u3000" @@ -313,9 +329,7 @@ end_tag_implied = { } el_is_special = (e) -> - return special_elements[e.name]? - # FIXME it should really be: - #return special_elements[e.name] is e.namespace + return special_elements[e.name] is e.namespace # decode_named_char_ref() # @@ -341,8 +355,8 @@ decode_named_char_ref = (txt) -> parse_html = (txt, parse_error_cb = null) -> cur = 0 # index of next char in txt to be parsed - # declare tree and tokenizer variables so they're in scope below - tree = null + # declare doc and tokenizer variables so they're in scope below + doc = null open_els = null # stack of open elements afe = null # active formatting elements template_insertion_modes = null @@ -356,6 +370,13 @@ parse_html = (txt, parse_error_cb = null) -> flag_foster_parenting = null form_element_pointer = null temporary_buffer = null + pending_table_character_tokens = null + head_element_pointer = null + flag_fragment_parsing = null + context_element = null + + stop_parsing = -> + flag_parsing = false parse_error = -> if parse_error_cb? @@ -439,6 +460,47 @@ parse_html = (txt, parse_error_cb = null) -> return false return false + clear_to_table_stopers = { + 'table': true + 'template': true + 'html': true + } + clear_stack_to_table_context = -> + loop + if clear_to_table_stopers[open_els[0].name]? + break + open_els.shift() + return + clear_to_table_body_stopers = { + 'tbody': true + 'tfoot': true + 'thead': true + 'template': true + 'html': true + } + clear_stack_to_table_body_context = -> + loop + if clear_to_table_body_stopers[open_els[0].name]? + break + open_els.shift() + return + clear_to_table_row_stopers = { + 'tr': true + 'template': true + 'html': true + } + clear_stack_to_table_row_context = -> + loop + if clear_to_table_row_stopers[open_els[0].name]? + break + open_els.shift() + return + clear_afe_to_marker = -> + loop + el = afe.shift() + if el.type is TYPE_AFE_MARKER + return + # 8.2.3.1 ... # http://www.w3.org/TR/html5/syntax.html#reset-the-insertion-mode-appropriately reset_insertion_mode = -> @@ -562,6 +624,14 @@ parse_html = (txt, parse_error_cb = null) -> node = open_els[node_i] # 19. Return to the step labeled loop. + # 8.2.3.2 + + # http://www.w3.org/TR/html5/syntax.html#adjusted-current-node + adjusted_current_node = -> + if open_els.length is 1 and flag_fragment_parsing + return context_element + return open_els[0] + # http://www.w3.org/TR/html5/syntax.html#reconstruct-the-active-formatting-elements # this implementation is structured (mostly) as described at the link above. # capitalized comments are the "labels" described at the link above. @@ -594,7 +664,7 @@ parse_html = (txt, parse_error_cb = null) -> # http://www.w3.org/TR/html5/syntax.html#unclosed-formatting-elements adoption_agency = (subject) -> debug_log "adoption_agency()" - debug_log "tree: #{serialize_els tree.children, false, true}" + debug_log "tree: #{serialize_els doc.children, false, true}" debug_log "open_els: #{serialize_els open_els, true, true}" debug_log "afe: #{serialize_els afe, true, true}" if open_els[0].name is subject @@ -706,7 +776,7 @@ parse_html = (txt, parse_error_cb = null) -> break node = node_next ? node_above debug_log "inner loop #{inner}" - debug_log "tree: #{serialize_els tree.children, false, true}" + debug_log "tree: #{serialize_els doc.children, false, true}" debug_log "open_els: #{serialize_els open_els, true, true}" debug_log "afe: #{serialize_els afe, true, true}" debug_log "ca: #{ca.name}##{ca.id} children: #{serialize_els ca.children, true, true}" @@ -800,7 +870,7 @@ parse_html = (txt, parse_error_cb = null) -> # at the appropriate place for inserting a node, but using common # ancestor as the override target. - # JASON: In the case where fe is immediately followed by fb: + # In the case where fe is immediately followed by fb: # * inner loop exits out early (node==fe) # * last_node is fb # * last_node is still in the tree (not a duplicate) @@ -817,7 +887,7 @@ parse_html = (txt, parse_error_cb = null) -> debug_log "fe: #{fe.name}##{fe.id} children: #{serialize_els fe.children, true, true}" debug_log "fb: #{fb.name}##{fb.id} children: #{serialize_els fb.children, true, true}" debug_log "last_node: #{last_node.name}##{last_node.id} children: #{serialize_els last_node.children, true, true}" - debug_log "tree: #{serialize_els tree.children, false, true}" + debug_log "tree: #{serialize_els doc.children, false, true}" debug_log "insert" @@ -833,7 +903,7 @@ parse_html = (txt, parse_error_cb = null) -> debug_log "fe: #{fe.name}##{fe.id} children: #{serialize_els fe.children, true, true}" debug_log "fb: #{fb.name}##{fb.id} children: #{serialize_els fb.children, true, true}" debug_log "last_node: #{last_node.name}##{last_node.id} children: #{serialize_els last_node.children, true, true}" - debug_log "tree: #{serialize_els tree.children, false, true}" + debug_log "tree: #{serialize_els doc.children, false, true}" # 15. Create an element for the token for which formatting element # was created, in the HTML namespace, with furthest block as the @@ -873,7 +943,7 @@ parse_html = (txt, parse_error_cb = null) -> break # 20. Jump back to the step labeled outer loop. debug_log "done wrapping fb's children. new_element: #{new_element.name}##{new_element.id}" - debug_log "tree: #{serialize_els tree.children, false, true}" + debug_log "tree: #{serialize_els doc.children, false, true}" debug_log "open_els: #{serialize_els open_els, true, true}" debug_log "afe: #{serialize_els afe, true, true}" debug_log "AAA DONE" @@ -1004,7 +1074,7 @@ parse_html = (txt, parse_error_cb = null) -> while t.attrs_a.length a = t.attrs_a.pop() attrs[a[0]] = a[1] # TODO check what to do with dupilcate attrs - el = new Node TYPE_TAG, name: t.name, namespace: namespace, attrs: attrs + el = new Node TYPE_TAG, name: t.name, namespace: namespace, attrs: attrs, token: t # TODO 2. If the newly created element has an xmlns attribute in the # XMLNS namespace whose value is not exactly the same as the element's @@ -1053,7 +1123,7 @@ parse_html = (txt, parse_error_cb = null) -> # http://www.w3.org/TR/html5/syntax.html#insert-a-comment # position should be [node, index_within_children] - tree_insert_comment = (t, position = null) -> + insert_comment = (t, position = null) -> position ?= adjusted_insertion_location() position[0].children.splice position[1], 0, t @@ -1076,6 +1146,92 @@ parse_html = (txt, parse_error_cb = null) -> while end_tag_implied[open_els[0].name] and open_els[0].name isnt except open_els.shift() + # 8.2.5.4 The rules for parsing tokens in HTML content + # http://www.w3.org/TR/html5/syntax.html#parsing-main-inhtml + + # 8.2.5.4.1 The "initial" insertion mode + # http://www.w3.org/TR/html5/syntax.html#the-initial-insertion-mode + ins_mode_initial = (t) -> + if is_space_tok t + return + if t.type is TYPE_COMMENT + # fixfull this is supposed to be "the last child of the document object" + doc.children.push t + return + if t.type is TYPE_DOCTYPE + # fixfull + t.name = 'html' + doc.children.push t + insertion_mode = ins_mode_before_html + return + # Anything else + #fixfull (iframe, quirks) + insertion_mode = ins_mode_before_html + insertion_mode t # reprocess the token + return + + # 8.2.5.4.2 http://www.w3.org/TR/html5/syntax.html#the-before-html-insertion-mode + ins_mode_before_html = (t) -> + if t.type is TYPE_DOCTYPE + parse_error() + return + if t.type is TYPE_COMMENT + doc.children.push t + return + if is_space_tok t + return + if t.type is TYPE_START_TAG and t.name is 'html' + el = token_to_element t, NS_HTML, doc + open_els.unshift(el) + # fixfull (big paragraph in spec about manifest, fragment, urls, etc) + insertion_mode = ins_mode_before_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" + else + parse_error() + return + # Anything else + html_tok = new_open_tag 'html' + el = token_to_element html_tok, NS_HTML, doc + doc.children.push el + open_els.unshift el + # ?fixfull browsing context + insertion_mode = ins_mode_before_head + insertion_mode t + return + + # 8.2.5.4.3 http://www.w3.org/TR/html5/syntax.html#the-before-head-insertion-mode + ins_mode_before_head = (t) -> + if is_space_tok t + return + if t.type is TYPE_COMMENT + insert_comment t + return + if t.type is TYPE_DOCTYPE + parse_error() + return + if t.type is TYPE_START_TAG and t.name is 'html' + ins_mode_in_body t + return + if t.type is TYPE_START_TAG and t.name is 'head' + el = insert_html_element t + head_element_pointer = el + insertion_mode = ins_mode_in_head + 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 + else + parse_error() + return + # Anything else + head_tok = new_open_tag 'head' + el = insert_html_element head_tok + head_element_pointer = el + insertion_mode = ins_mode_in_head + insertion_mode t # reprocess current token + # 8.2.5.4.4 http://www.w3.org/TR/html5/syntax.html#parsing-main-inhead ins_mode_in_head_else = (t) -> # factored out for same-as-spec flow control open_els.shift() # spec says this will be a 'head' node @@ -1086,7 +1242,7 @@ parse_html = (txt, parse_error_cb = null) -> insert_character t return if t.type is TYPE_COMMENT - tree_insert_comment t + insert_comment t return if t.type is TYPE_DOCTYPE parse_error() @@ -1097,12 +1253,12 @@ parse_html = (txt, parse_error_cb = null) -> if t.type is TYPE_START_TAG and (t.name is 'base' or t.name is 'basefont' or t.name is 'bgsound' or t.name is 'link') el = insert_html_element t open_els.shift() - el.acknowledge_self_closing() + t.acknowledge_self_closing() return if t.type is TYPE_START_TAG and t.name is 'meta' el = insert_html_element t open_els.shift() - el.acknowledge_self_closing() + t.acknowledge_self_closing() # fixfull encoding stuff return if t.type is TYPE_START_TAG and t.name is 'title' @@ -1155,11 +1311,67 @@ parse_html = (txt, parse_error_cb = null) -> else parse_error() return - if (t.type is TYPE_OPEN_TAG and t.name is 'head') or t.type is TYPE_END_TAG + if (t.type is TYPE_START_TAG and t.name is 'head') or t.type is TYPE_END_TAG parse_error() return ins_mode_in_head_else t + # 8.2.5.4.5 http://www.w3.org/TR/html5/syntax.html#parsing-main-inheadnoscript + ins_mode_in_head_noscript = (t) -> + # FIXME ?fixfull + console.log "ins_mode_in_head_noscript unimplemented" + + # 8.2.5.4.6 http://www.w3.org/TR/html5/syntax.html#the-after-head-insertion-mode + ins_mode_after_head_else = (t) -> + body_tok = new_open_tag 'body' + insert_html_element body_tok + insertion_mode = ins_mode_in_body + insertion_mode t # reprocess token + return + ins_mode_after_head = (t) -> + if is_space_tok t + insert_character t + return + if t.type is TYPE_COMMENT + insert_comment t + return + if t.type is TYPE_DOCTYPE + parse_error() + return + if t.type is TYPE_START_TAG and t.name is 'html' + ins_mode_in_body t + return + if t.type is TYPE_START_TAG and t.name is 'body' + insert_html_element t + flag_frameset_ok = false + insertion_mode = ins_mode_in_body + return + if t.type is TYPE_START_TAG and t.name is 'frameset' + insert_html_element t + insertion_mode = ins_mode_in_frameset + return + if t.type is TYPE_START_TAG and (t.name is 'base' or 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 'script' or t.name is 'style' or t.name is 'template' or t.name is 'title') + parse_error() + open_els.unshift head_element_pointer + ins_mode_in_head t + for el, i of open_els + if el is head_element_pointer + open_els.splice i, 1 + return + console.log "warning: 23904 couldn't find head element in open_els" + return + if t.type is TYPE_END_TAG and t.name is 'template' + ins_mode_in_head t + return + if t.type is TYPE_END_TAG and (t.name is 'body' or t.name is 'html' or t.name is 'br') + ins_mode_after_head_else t + return + if (t.type is TYPE_START_TAG and t.name is 'head') or t.type is TYPE_END_TAG + parse_error() + return + # Anything else + ins_mode_after_head_else t + # 8.2.5.4.7 http://www.w3.org/TR/html5/syntax.html#parsing-main-inbody in_body_any_other_end_tag = (name) -> # factored out because adoption agency calls it for node, i in open_els @@ -1187,7 +1399,7 @@ parse_html = (txt, parse_error_cb = null) -> insert_character t flag_frameset_ok = false when TYPE_COMMENT - tree_insert_comment t + insert_comment t when TYPE_DOCTYPE parse_error() when TYPE_START_TAG @@ -1269,7 +1481,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() break # TODO stack of template insertion modes thing - flag_parsing = false # stop parsing + stop_parsing() when TYPE_END_TAG switch t.name when 'body' @@ -1312,53 +1524,13 @@ parse_html = (txt, parse_error_cb = null) -> flag_foster_parenting = true # FIXME ins_mode_in_body t flag_foster_parenting = false - can_in_table = { + can_in_table = { # FIXME do this inline like everywhere else 'table': true 'tbody': true 'tfoot': true 'thead': true 'tr': true } - clear_to_table_stopers = { - 'table': true - 'template': true - 'html': true - } - clear_stack_to_table_context = -> - loop - if clear_to_table_stopers[open_els[0].name]? - break - open_els.shift() - return - clear_to_table_body_stopers = { - 'tbody': true - 'tfoot': true - 'thead': true - 'template': true - 'html': true - } - clear_stack_to_table_body_context = -> - loop - if clear_to_table_body_stopers[open_els[0].name]? - break - open_els.shift() - return - clear_to_table_row_stopers = { - 'tr': true - 'template': true - 'html': true - } - clear_stack_to_table_row_context = -> - loop - if clear_to_table_row_stopers[open_els[0].name]? - break - open_els.shift() - return - clear_afe_to_marker = -> - loop - el = afe.shift() - if el.type is TYPE_AFE_MARKER - return # 8.2.5.4.8 http://www.w3.org/TR/html5/syntax.html#parsing-main-incdata ins_mode_text = (t) -> @@ -1399,7 +1571,7 @@ parse_html = (txt, parse_error_cb = null) -> else ins_mode_in_table_else t when TYPE_COMMENT - tree_insert_comment t + insert_comment t when TYPE_DOCTYPE parse_error() when TYPE_START_TAG @@ -1445,7 +1617,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() el = insert_html_element t open_els.shift() - el.acknowledge_self_closing() + t.acknowledge_self_closing() when 'form' parse_error() if form_element_pointer? @@ -1479,22 +1651,116 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode_in_table_else t - # 8.2.5.4.11 http://www.w3.org/TR/html5/syntax.html#parsing-main-intabletext + # 8.2.5.4.10 http://www.w3.org/TR/html5/syntax.html#parsing-main-intabletext ins_mode_in_table_text = (t) -> - switch t.type - when TYPE_TEXT - switch t.text - when "\u0000" - parse_error() - return - console.log "unimplemented ins_mode_in_table_text" - # FIXME CONTINUE + if t.type is TYPE_TEXT and t.text is "\u0000" + # huh? I thought the tokenizer didn't emit these + parse_error() + return + if t.type is TYPE_TEXT + pending_table_character_tokens.push t + return + # Anything else + all_space = true + for old in pending_table_character_tokens + unless is_space_tok old + all_space = false + break + if all_space + for old in pending_table_character_tokens + insert_character old + else + for old in pending_table_character_tokens + ins_mode_table_else old + pending_table_character_tokens = [] # FIXME test (spec doesn't say this) + insertion_mode = original_insertion_mode + insertion_mode t + + # 8.2.5.4.11 http://www.w3.org/TR/html5/syntax.html#parsing-main-incaption + ins_mode_in_caption = (t) -> + if t.type is TYPE_END_TAG and t.name is 'caption' + if is_in_table_scope 'caption' + generate_implied_end_tags() + if open_els[0].name isnt 'caption' + parse_error() + loop + el = open_els.shift() + if el.name is 'caption' + break + clear_afe_to_marker() + insertion_mode = in_table + else + parse_error() + # fragment case + return + if (t.type is TYPE_START_TAG and (t.name is 'caption' or t.name is 'col' or t.name is 'colgroup' or t.name is 'tbody' or t.name is 'td' or t.name is 'tfoot' or t.name is 'th' or t.name is 'thead' or t.name is 'tr')) or t.type is TYPE_END_TAG and t.name is 'table' + parse_error() + if is_in_table_scope 'caption' + loop + el = open_els.shift() + if el.name is 'caption' + break + clear_afe_to_marker() + insertion_mode = in_table + insertion_mode t + # else fragment case + return + if t.type is TYPE_END_TAG and (t.name is 'body' or t.name is 'col' or t.name is 'colgroup' or t.name is 'html' or t.name is 'tbody' or t.name is 'td' or t.name is 'tfoot' or t.name is 'th' or t.name is 'thead' or t.name is 'tr') + parse_error() + return + # Anything else + ins_mode_in_body t + + # 8.2.5.4.12 http://www.w3.org/TR/html5/syntax.html#parsing-main-incolgroup + ins_mode_in_column_group = (t) -> + if is_space_tok t + insert_character t + return + if t.type is TYPE_COMMENT + insert_comment t + return + if t.type is TYPE_DOCTYPE + parse_error() + return + if t.type is TYPE_START_TAG and t.name is 'html' + ins_mode_in_body t + return + if t.type is TYPE_START_TAG and t.name is 'col' + el = insert_html_element t + open_els.shift() + t.acknowledge_self_closing() + return + if t.type is TYPE_END_TAG and t.name is 'colgroup' + if open_els[0].name is 'colgroup' + open_els[0].shift() + insertion_mode = ins_mode_in_table + else + parse_error() + return + if t.type is TYPE_END_TAG and t.name is 'col' + parse_error() + return + if (t.type is TYPE_START_TAG or t.type is TYPE_END_TAG) and t.name is 'template' + ins_mode_in_head t + return + if t.type is TYPE_EOF + ins_mode_in_body t + return + # Anything else + if open_els[0].name isnt 'colgroup' + parse_error() + return + open_els.shift() + insertion_mode = ins_mode_in_table + insertion_mode t + return # 8.2.5.4.13 http://www.w3.org/TR/html5/syntax.html#parsing-main-intbody ins_mode_in_table_body = (t) -> if t.type is TYPE_START_TAG and t.name is 'tr' clear_stack_to_table_body_context() insert_html_element t + insertion_mode = ins_mode_in_row return if t.type is TYPE_START_TAG and (t.name is 'th' or t.name is 'td') parse_error() @@ -1629,6 +1895,306 @@ parse_html = (txt, parse_error_cb = null) -> # Anything Else ins_mode_in_body t + # 8.2.5.4.16 http://www.w3.org/TR/html5/syntax.html#parsing-main-inselect + ins_mode_in_select = (t) -> + if t.type is TYPE_TEXT and t.text is "\u0000" + parse_error() + return + if t.type is TYPE_TEXT + insert_character t + return + if t.type is TYPE_COMMENT + insert_comment t + return + if t.type is TYPE_DOCTYPE + parse_error() + return + if t.type is TYPE_START_TAG and t.name is 'html' + ins_mode_in_body t + return + if t.type is TYPE_START_TAG and t.name is 'option' + if open_els[0].name is 'option' + open_els.shift() + insert_html_element t + return + if t.type is TYPE_START_TAG and t.name is 'optgroup' + if open_els[0].name is 'option' + open_els.shift() + if open_els[0].name is 'optgroup' + open_els.shift() + insert_html_element t + return + if t.type is TYPE_END_TAG and t.name is 'optgroup' + if open_els[0].name is 'option' and open_els[1].name is 'optgroup' + open_els.shift() + if open_els[0].name is 'optgroup' + open_els.shift() + else + parse_error() + return + if t.type is TYPE_END_TAG and t.name is 'option' + if open_els[0].name is 'option' + open_els.shift() + else + parse_error() + return + if t.type is TYPE_END_TAG and t.name is 'select' + if is_in_select_scope 'select' + loop + el = open_els.shift() + if el.name is 'select' + break + reset_insertion_mode() + else + parse_error() + return + if t.type is TYPE_START_TAG and t.name is 'select' + parse_error() + loop + el = open_els.shift() + if el.name is 'select' + break + reset_insertion_mode() + # spec says that this is the same as but it doesn't say + # to check scope first + 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' + return + loop + el = open_els.shift() + if el.name is 'select' + break + reset_insertion_mode() + insertion_mode t + return + if t.type is TYPE_START_TAG and (t.name is 'script' or t.name is 'template') + ins_mode_in_head t + return + if t.type is TYPE_EOF + ins_mode_in_body t + return + # Anything else + parse_error() + return + + # 8.2.5.4.17 http://www.w3.org/TR/html5/syntax.html#parsing-main-inselectintable + ins_mode_in_select_in_table = (t) -> + if t.type is TYPE_START_TAG and (t.name is 'caption' or t.name is 'table' or t.name is 'tbody' or t.name is 'tfoot' or t.name is 'thead' or t.name is 'tr' or t.name is 'td' or t.name is 'th') + parse_error() + loop + el = open_els.shift() + if el.name is 'select' + break + reset_insertion_mode() + insertion_mode t + return + if t.type is TYPE_END_TAG and (t.name is 'caption' or t.name is 'table' or t.name is 'tbody' or t.name is 'tfoot' or t.name is 'thead' or t.name is 'tr' or t.name is 'td' or t.name is 'th') + parse_error() + unless is_in_table_scope t.name, NS_HTML + return + loop + el = open_els.shift() + if el.name is 'select' + break + reset_insertion_mode() + insertion_mode t + return + # Anything else + ins_mode_in_select t + return + + # 8.2.5.4.18 http://www.w3.org/TR/html5/syntax.html#parsing-main-intemplate + ins_mode_in_template = (t) -> + if t.type is TYPE_TEXT or t.type is TYPE_COMMENT or t.type is TYPE_DOCTYPE + ins_mode_in_body t + return + if (t.type is TYPE_START_TAG and (t.name is 'base' or 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 'script' or t.name is 'style' or t.name is 'template' or t.name is 'title')) or (t.type is TYPE_END_TAG and t.name is 'template') + ins_mode_in_head t + return + if t.type is TYPE_START_TAG and (t.name is 'caption' or t.name is 'colgroup' or t.name is 'tbody' or t.name is 'tfoot' or t.name is 'thead') + template_insertion_modes.shift() + template_insertion_modes.unshift ins_mode_in_table + insertion_mode = ins_mode_in_table + insertion_mode t + return + if t.type is TYPE_START_TAG and t.name is 'col' + template_insertion_modes.shift() + template_insertion_modes.unshift ins_mode_in_column_group + insertion_mode = ins_mode_in_column_group + insertion_mode t + return + if t.type is TYPE_START_TAG and t.name is 'tr' + template_insertion_modes.shift() + template_insertion_modes.unshift ins_mode_in_table_body + insertion_mode = ins_mode_in_table_body + insertion_mode t + return + if t.type is TYPE_START_TAG and (t.name is 'td' or t.name is 'th') + template_insertion_modes.shift() + template_insertion_modes.unshift ins_mode_in_row + insertion_mode = ins_mode_in_row + insertion_mode t + return + if t.type is TYPE_START_TAG + template_insertion_modes.shift() + template_insertion_modes.unshift ins_mode_in_body + insertion_mode = ins_mode_in_body + insertion_mode t + return + if t.type is TYPE_END_TAG + parse_error() + return + if t.type is EOF + unless template_tag_is_open() + stop_parsing() + return + parse_error() + loop + el = open_els.shift() + if el.name is 'template' # fixfull check namespace + break + clear_afe_to_marker() + template_insertion_modes.shift() + reset_insertion_mode() + insertion_mode t + + # 8.2.5.4.19 http://www.w3.org/TR/html5/syntax.html#parsing-main-afterbody + ins_mode_after_body = (t) -> + if is_space_tok t + ins_mode_in_body t + return + if t.type is TYPE_COMMENT + insert_comment t, [open_els[0], open_els[0].children.length] + return + if t.type is TYPE_DOCTYPE + parse_error() + return + 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 'html' + # fixfull fragment case + insertion_mode = ins_mode_after_after_body + return + if t.type is TYPE_EOF + stop_parsing() + return + # Anything ELse + parse_error() + insertion_mode = ins_mode_in_body + insertion_mode t + + # 8.2.5.4.20 http://www.w3.org/TR/html5/syntax.html#parsing-main-inframeset + ins_mode_in_frameset = (t) -> + if is_space_tok t + insert_character t + return + if t.type is TYPE_COMMENT + insert_comment t + return + if t.type is TYPE_DOCTYPE + parse_error() + return + if t.type is TYPE_START_TAG and t.name is 'html' + ins_mode_in_body t + return + if t.type is TYPE_START_TAG and t.name is 'frameset' + insert_html_element t + return + if t.type is TYPE_END_TAG and t.name is 'frameset' + # TODO ?correct for: "if the current node is the root html element" + if open_els.length is 1 + parse_error() + return # fragment case + open_els.shift() + if flag_fragment_parsing is false and open_els[0].name isnt 'frameset' + insertion_mode = ins_mode_after_frameset + return + if t.type is TYPE_START_TAG and t.name is 'frame' + insert_html_element t + open_els.shift() + t.acknowledge_self_closing() + return + if t.type is TYPE_START TAG and t.name is 'noframes' + ins_mode_in_head t + return + if t.type is TYPE_EOF + # TODO ?correct for: "if the current node is not the root html element" + if open_els.length isnt 1 + parse_error() + stop_parsing() + return + # Anything else + parse_error() + return + + # 8.2.5.4.21 http://www.w3.org/TR/html5/syntax.html#parsing-main-afterframeset + ins_mode_after_frameset = (t) -> + if is_space_tok t + insert_character t + return + if t.type is TYPE_COMMENT + insert_comment t + return + if t.type is TYPE_DOCTYPE + parse_error() + return + 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 'html' + insert_mode = ins_mode_after_after_frameset + return + if t.type is TYPE_START_TAG and t.name is 'noframes' + ins_mode_in_head t + return + if t.type is TYPE_EOF + stop_parsing() + return + # Anything else + parse_error() + return + + # 8.2.5.4.22 http://www.w3.org/TR/html5/syntax.html#the-after-after-body-insertion-mode + ins_mode_after_after_body = (t) -> + if t.type is TYPE_COMMENT + insert_comment t, [doc, doc.children.length] + return + if t.type is TYPE_DOCTYPE or is_space_tok(t) or (t.type is TYPE_START_TAG and t.name is 'html') + ins_mode_in_body t + return + if t.type is TYPE_EOF + stop_parsing() + return + # Anything else + parse_error() + insertion_mode = ins_mode_in_body + return + + # 8.2.5.4.23 http://www.w3.org/TR/html5/syntax.html#the-after-after-frameset-insertion-mode + ins_mode_after_after_frameset = (t) -> + if t.type is TYPE_COMMENT + insert_comment t, [doc, doc.children.length] + return + if t.type is TYPE_DOCTYPE or is_space_tok(t) or (t.type is TYPE_START_TAG and t.name is 'html') + ins_mode_in_body t + return + if t.type is TYPE_EOF + stop_parsing() + return + if t.type is TYPE_START_TAG and t.name is 'noframes' + ins_mode_in_head t + return + # Anything else + parse_error() + return + + + + + # 8.2.4.1 http://www.w3.org/TR/html5/syntax.html#data-state tok_state_data = -> switch c = txt.charAt(cur++) @@ -1719,6 +2285,7 @@ parse_html = (txt, parse_error_cb = null) -> tok_state = tok_state_end_tag_open when '?' parse_error() + tok_cur_tag = new_comment_token '?' tok_state = tok_state_bogus_comment else if lc_alpha.indexOf(c) > -1 @@ -1753,6 +2320,7 @@ parse_html = (txt, parse_error_cb = null) -> tok_state = tok_state_tag_name else parse_error() + tok_cur_tag = new_comment_token '/' tok_state = tok_state_bogus_comment return null @@ -2120,6 +2688,174 @@ parse_html = (txt, parse_error_cb = null) -> cur -= 1 # we didn't handle that char return null + # 8.2.4.44 http://www.w3.org/TR/html5/syntax.html#bogus-comment-state + # WARNING: put a comment token in tok_cur_tag before setting this state + tok_state_bogus_comment = -> + next_gt = txt.indexOf '>', cur + if next_gt is -1 + val = txt.substr cur + cur = txt.length + else + val = txt.substr cur, (next_gt - cur) + cur = next_gt + 1 + val = val.replace "\u0000", "\ufffd" + tok_cur_tag.text += val + tok_state = tok_state_data + return tok_cur_tag + + # 8.2.4.45 http://www.w3.org/TR/html5/syntax.html#markup-declaration-open-state + tok_state_markup_declaration_open = -> + if txt.substr(cur, 2) is '--' + cur += 2 + tok_cur_tag = new_comment_token '' + tok_state = tok_state_comment_start + return + if txt.substr(cur, 7).toLowerCase() is 'doctype' + cur += 7 + tok_state = tok_state_doctype + return + acn = adjusted_current_node() + if acn and acn.namespace isnt NS_HTML and text.substr(cur, 7) is '[CDATA[' + cur += 7 + tok_state = tok_state_cdata_section + return + # Otherwise + parse_errer() + tok_cur_tag = new_comment_token '!' # TODO test ("!" right?) + tok_state = tok_state_bogus_comment + return + + # 8.2.4.46 http://www.w3.org/TR/html5/syntax.html#comment-start-state + tok_state_comment_start = -> + switch c = txt.charAt(cur++) + when '-' + tok_state = tok_state_comment_start_dash + when "\u0000" + parse_error() + return new_character_token "\ufffd" + when '>' + parse_error() + tok_state = tok_state_data + return tok_cur_tag + when '' # EOF + parse_error() + tok_state = tok_state_data + cur -= 1 # Reconsume + return tok_cur_tag + else + tok_cur_tag.text += c + return null + + # 8.2.4.47 http://www.w3.org/TR/html5/syntax.html#comment-start-dash-state + tok_state_comment_start_dash = -> + switch c = txt.charAt(cur++) + when '-' + tok_state = tok_state_comment_end + when "\u0000" + parse_error() + tok_cur_tag.text += "-\ufffd" + tok_state = tok_state_comment + when '>' + parse_error() + tok_state = tok_state_data + return tok_cur_tag + when '' # EOF + parse_error() + tok_state = tok_state_data + cur -= 1 # Reconsume + return tok_cur_tag + else + tok_cur_tag.text += "-#{c}" + tok_state = tok_state_comment + return null + + # 8.2.4.48 http://www.w3.org/TR/html5/syntax.html#comment-state + tok_state_comment = -> + switch c = txt.charAt(cur++) + when '-' + tok_state = tok_state_comment_end_dash + when "\u0000" + parse_error() + tok_cur_tag.text += "\ufffd" + when '' # EOF + parse_error() + tok_state = tok_state_data + cur -= 1 # Reconsume + return tok_cur_tag + else + tok_cur_tag.text += c + return null + + # 8.2.4.49 http://www.w3.org/TR/html5/syntax.html#comment-end-dash-state + tok_state_comment_end_dash = -> + switch c = txt.charAt(cur++) + when '-' + tok_state = tok_state_comment_end + when "\u0000" + parse_error() + tok_cur_tag.text += "-\ufffd" + tok_state = tok_state_comment + when '' # EOF + parse_error() + tok_state = tok_state_data + cur -= 1 # Reconsume + return tok_cur_tag + else + tok_cur_tag.text += "-#{c}" + tok_state = tok_state_comment + return null + + # 8.2.4.50 http://www.w3.org/TR/html5/syntax.html#comment-end-state + tok_state_comment_end = -> + switch c = txt.charAt(cur++) + when '>' + tok_state = tok_state_data + return tok_cur_tag + when "\u0000" + parse_error() + tok_cur_tag.text += "--\ufffd" + tok_state = tok_state_comment + when '!' + parse_error() + tok_state = tok_state_comment_end_bang + when '-' + parse_error() + tok_cur_tag.text += '-' + when '' # EOF + parse_error() + tok_state = tok_state_data + cur -= 1 # Reconsume + return tok_cur_tag + else + parse_error() + tok_cur_tag.text += "--#{c}" + tok_state = tok_state_comment + return null + + # 8.2.4.51 http://www.w3.org/TR/html5/syntax.html#comment-end-bang-state + tok_state_comment_end_bang = -> + switch c = txt.charAt(cur++) + when '-' + tok_cur_tag.text += "--!#{c}" + tok_state = tok_state_comment_end_dash + when '>' + tok_state = tok_state_data + return tok_cur_tag + when "\u0000" + parse_error() + tok_cur_tag.text += "--!\ufffd" + tok_state = tok_state_comment + when '' # EOF + parse_error() + tok_state = tok_state_data + cur -= 1 # Reconsume + return tok_cur_tag + else + tok_cur_tag.text += "--!#{c}" + tok_state = tok_state_comment + return null + + # 8.2.4.69 http://www.w3.org/TR/html5/syntax.html#consume-a-character-reference # Don't set this as a state, just call it # returns a string (NOT a text node) @@ -2198,11 +2934,11 @@ parse_html = (txt, parse_error_cb = null) -> # tree constructor initialization # see comments on TYPE_TAG/etc for the structure of this data - tree = new Node TYPE_TAG, name: 'html', namespace: NS_HTML - open_els = [tree] + doc = new Node TYPE_TAG, name: 'html', namespace: NS_HTML + open_els = [doc] afe = [] # active formatting elements template_insertion_modes = [] - insertion_mode = ins_mode_in_body + insertion_mode = ins_mode_initial original_insertion_mode = insertion_mode # TODO check spec flag_scripting = true # TODO might need an extra flag to get