hi
") # -# the spec uses a many different words do indicate which ends of lists/stacks -# they are talking about (and relative movement within the lists/stacks). This -# section splains. I'm implementing "lists" (afe and open_els) the same way -# (both as stacks) +# Or, if you don't want /etc, do this: +# +# wheic.parse_html("hi
", {fragment: "body"}) +# +# return value is an array of Nodes, see "class Node" below. + +# This code is a work in progress, eg try search this file for "fixfull", +# "TODO" and "FIXME" + + +# Notes: stacks/lists +# +# Jason was frequently confused by the terminology used to refer to different +# parts of the stacks and lists in the spec, so he made this chart to help keep +# his head straight: # # stacks grow downward (current element is index=0) # # example: open_els = [a, b, c, d, e, f, g] # -# "grows downwards" means it's visualized like this: (index: el, names) +# "grows downwards" means it's visualized like this: (index: el "names") # # 6: g "start of the list", "topmost", "first" # 5: f @@ -46,13 +75,22 @@ # 1: b # 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? +if (typeof module) isnt 'undefined' and module.exports? + context = 'module' + exports = module.exports +else + context = 'browser' window.wheic = {} - module = exports: window.wheic + exports = window.wheic + +from_code_point = (x) -> + if String.fromCodePoint? + return String.fromCodePoint x + else + if x <= 0xffff + return String.fromCharCode x + x -= 0x10000 + return String.fromCharCode((x >> 10) + 0xd800, (x % 0x400) + 0xdc00) # Each node is an obect of the Node class. Here are the Node types: TYPE_TAG = 0 # name, {attributes}, [children] @@ -71,14 +109,23 @@ NS_HTML = 1 NS_MATHML = 2 NS_SVG = 3 +# quirks mode constants +QUIRKS_NO = 1 +QUIRKS_LIMITED = 2 +QUIRKS_YES = 3 + +# queue up debug logs, so eg they can be shown only for tests that fail g_debug_log = [] debug_log_reset = -> g_debug_log = [] + return debug_log = (str) -> g_debug_log.push str + return debug_log_each = (cb) -> for str in g_debug_log cb str + return prev_node_id = 0 class Node @@ -99,58 +146,16 @@ class Node @id = "#{++prev_node_id}" acknowledge_self_closing: -> if @token? - @token.flag 'did_self_close' + @token.flag 'did_self_close', true else @flag 'did_self_close', true + return flag: (key, value = null) -> if value? @flags[key] = value else return @flags[key] - serialize: (shallow = false, show_ids = false) -> # for unit tests - ret = '' - switch @type - when TYPE_TAG - ret += 'tag:' - ret += JSON.stringify @name - ret += ',' - if show_ids - ret += "##{@id}," - if shallow - break - attr_keys = [] - for k of @attrs - attr_keys.push k - attr_keys.sort() - ret += '{' - sep = '' - for k in attr_keys - ret += sep - sep = ',' - ret += "#{JSON.stringify k}:#{JSON.stringify @attrs[k]}" - ret += '},[' - sep = '' - for c in @children - ret += sep - sep = ',' - ret += c.serialize shallow, show_ids - ret += ']' - when TYPE_TEXT - ret += 'text:' - ret += JSON.stringify @text - when TYPE_COMMENT - ret += 'comment:' - ret += JSON.stringify @text - when TYPE_DOCTYPE - ret += "doctype:#{@name},#{JSON.stringify(@public_identifier ? '')},#{JSON.stringify(@system_identifier ? '')}" - when TYPE_AFE_MARKER - ret += 'marker' - when TYPE_AAA_BOOKMARK - ret += 'aaa_bookmark' - else - ret += 'unknown:' - console.log "unknown: #{JSON.stringify @}" # backtrace is just as well - return ret + return # helpers: (only take args that are normally known when parser creates nodes) new_open_tag = (name) -> @@ -195,8 +200,8 @@ is_space_tok = (t) -> return t.type is TYPE_TEXT && t.text.length is 1 and space_chars.indexOf(t.text) > -1 is_input_hidden_tok = (t) -> - return unless t.type is TYPE_START_TAG - for a of t.attrs_a + return false unless t.type is TYPE_START_TAG + for a in t.attrs_a if a[0] is 'type' if a[1].toLowerCase() is 'hidden' return true @@ -206,6 +211,94 @@ is_input_hidden_tok = (t) -> # 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" +unicode_fixes = {} +unicode_fixes[0x00] = "\uFFFD" +unicode_fixes[0x80] = "\u20AC" +unicode_fixes[0x82] = "\u201A" +unicode_fixes[0x83] = "\u0192" +unicode_fixes[0x84] = "\u201E" +unicode_fixes[0x85] = "\u2026" +unicode_fixes[0x86] = "\u2020" +unicode_fixes[0x87] = "\u2021" +unicode_fixes[0x88] = "\u02C6" +unicode_fixes[0x89] = "\u2030" +unicode_fixes[0x8A] = "\u0160" +unicode_fixes[0x8B] = "\u2039" +unicode_fixes[0x8C] = "\u0152" +unicode_fixes[0x8E] = "\u017D" +unicode_fixes[0x91] = "\u2018" +unicode_fixes[0x92] = "\u2019" +unicode_fixes[0x93] = "\u201C" +unicode_fixes[0x94] = "\u201D" +unicode_fixes[0x95] = "\u2022" +unicode_fixes[0x96] = "\u2013" +unicode_fixes[0x97] = "\u2014" +unicode_fixes[0x98] = "\u02DC" +unicode_fixes[0x99] = "\u2122" +unicode_fixes[0x9A] = "\u0161" +unicode_fixes[0x9B] = "\u203A" +unicode_fixes[0x9C] = "\u0153" +unicode_fixes[0x9E] = "\u017E" +unicode_fixes[0x9F] = "\u0178" + +quirks_yes_pi_prefixes = [ + "+//silmaril//dtd html pro v0r11 19970101//" + "-//as//dtd html 3.0 aswedit + extensions//" + "-//advasoft ltd//dtd html 3.0 aswedit + extensions//" + "-//ietf//dtd html 2.0 level 1//" + "-//ietf//dtd html 2.0 level 2//" + "-//ietf//dtd html 2.0 strict level 1//" + "-//ietf//dtd html 2.0 strict level 2//" + "-//ietf//dtd html 2.0 strict//" + "-//ietf//dtd html 2.0//" + "-//ietf//dtd html 2.1e//" + "-//ietf//dtd html 3.0//" + "-//ietf//dtd html 3.2 final//" + "-//ietf//dtd html 3.2//" + "-//ietf//dtd html 3//" + "-//ietf//dtd html level 0//" + "-//ietf//dtd html level 1//" + "-//ietf//dtd html level 2//" + "-//ietf//dtd html level 3//" + "-//ietf//dtd html strict level 0//" + "-//ietf//dtd html strict level 1//" + "-//ietf//dtd html strict level 2//" + "-//ietf//dtd html strict level 3//" + "-//ietf//dtd html strict//" + "-//ietf//dtd html//" + "-//metrius//dtd metrius presentational//" + "-//microsoft//dtd internet explorer 2.0 html strict//" + "-//microsoft//dtd internet explorer 2.0 html//" + "-//microsoft//dtd internet explorer 2.0 tables//" + "-//microsoft//dtd internet explorer 3.0 html strict//" + "-//microsoft//dtd internet explorer 3.0 html//" + "-//microsoft//dtd internet explorer 3.0 tables//" + "-//netscape comm. corp.//dtd html//" + "-//netscape comm. corp.//dtd strict html//" + "-//o'reilly and associates//dtd html 2.0//" + "-//o'reilly and associates//dtd html extended 1.0//" + "-//o'reilly and associates//dtd html extended relaxed 1.0//" + "-//sq//dtd html 2.0 hotmetal + extensions//" + "-//softquad software//dtd hotmetal pro 6.0::19990601::extensions to html 4.0//" + "-//softquad//dtd hotmetal pro 4.0::19971010::extensions to html 4.0//" + "-//spyglass//dtd html 2.0 extended//" + "-//sun microsystems corp.//dtd hotjava html//" + "-//sun microsystems corp.//dtd hotjava strict html//" + "-//w3c//dtd html 3 1995-03-24//" + "-//w3c//dtd html 3.2 draft//" + "-//w3c//dtd html 3.2 final//" + "-//w3c//dtd html 3.2//" + "-//w3c//dtd html 3.2s draft//" + "-//w3c//dtd html 4.0 frameset//" + "-//w3c//dtd html 4.0 transitional//" + "-//w3c//dtd html experimental 19960712//" + "-//w3c//dtd html experimental 970421//" + "-//w3c//dtd w3 html//" + "-//w3o//dtd w3 html 3.0//" + "-//webtechs//dtd mozilla html 2.0//" + "-//webtechs//dtd mozilla html//" +] + # These are the character references that don't need a terminating semicolon # min length: 2, max: 6, none are a prefix of any other. legacy_char_refs = { @@ -297,14 +390,17 @@ special_elements = { h2:NS_HTML, h3:NS_HTML, h4:NS_HTML, h5:NS_HTML, h6:NS_HTML, head:NS_HTML, header:NS_HTML, hgroup:NS_HTML, hr:NS_HTML, html:NS_HTML, iframe:NS_HTML, img:NS_HTML, input:NS_HTML, isindex:NS_HTML, li:NS_HTML, link:NS_HTML, - listing:NS_HTML, main:NS_HTML, marquee:NS_HTML, meta:NS_HTML, nav:NS_HTML, - noembed:NS_HTML, noframes:NS_HTML, noscript:NS_HTML, object:NS_HTML, - ol:NS_HTML, p:NS_HTML, param:NS_HTML, plaintext:NS_HTML, pre:NS_HTML, - script:NS_HTML, section:NS_HTML, select:NS_HTML, source:NS_HTML, - style:NS_HTML, summary:NS_HTML, table:NS_HTML, tbody:NS_HTML, td:NS_HTML, - template:NS_HTML, textarea:NS_HTML, tfoot:NS_HTML, th:NS_HTML, - thead:NS_HTML, title:NS_HTML, tr:NS_HTML, track:NS_HTML, ul:NS_HTML, - wbr:NS_HTML, xmp:NS_HTML, + listing:NS_HTML, main:NS_HTML, marquee:NS_HTML, + + menu:NS_HTML,menuitem:NS_HTML, # WHATWG adds these + + meta:NS_HTML, nav:NS_HTML, noembed:NS_HTML, noframes:NS_HTML, + noscript:NS_HTML, object:NS_HTML, ol:NS_HTML, p:NS_HTML, param:NS_HTML, + plaintext:NS_HTML, pre:NS_HTML, script:NS_HTML, section:NS_HTML, + select:NS_HTML, source:NS_HTML, style:NS_HTML, summary:NS_HTML, + table:NS_HTML, tbody:NS_HTML, td:NS_HTML, template:NS_HTML, + textarea:NS_HTML, tfoot:NS_HTML, th:NS_HTML, thead:NS_HTML, title:NS_HTML, + tr:NS_HTML, track:NS_HTML, ul:NS_HTML, wbr:NS_HTML, xmp:NS_HTML, # MathML: mi:NS_MATHML, mo:NS_MATHML, mn:NS_MATHML, ms:NS_MATHML, mtext:NS_MATHML, @@ -324,14 +420,15 @@ mathml_text_integration = { mi: NS_MATHML, mo: NS_MATHML, mn: NS_MATHML, ms: NS_MATHML, mtext: NS_MATHML } is_mathml_text_integration_point = (el) -> - return mathml_text_integration[el.name] = el.namespace + return mathml_text_integration[el.name] is el.namespace is_html_integration = (el) -> # DON'T PASS A TOKEN - if el.namespace is NS_MATHML and el.name is 'annotation-xml' - if el.attrs.encoding? - if el.attrs.encoding.toLowerCase() is 'text/html' - return true - if el.attrs.encoding.toLowerCase() is 'application/xhtml+xml' - return true + if el.namespace is NS_MATHML + if el.name is 'annotation-xml' + if el.attrs.encoding? + if el.attrs.encoding.toLowerCase() is 'text/html' + return true + if el.attrs.encoding.toLowerCase() is 'application/xhtml+xml' + return true return false if el.namespace is NS_SVG if el.name is 'foreignObject' or el.name is 'desc' or el.name is 'title' @@ -342,28 +439,25 @@ h_tags = { h1:NS_HTML, h2:NS_HTML, h3:NS_HTML, h4:NS_HTML, h5:NS_HTML, h6:NS_HTML } -# FIXME namespacify foster_parenting_targets = { - table: true - tbody: true - tfoot: true - thead: true - tr: true + table: NS_HTML + tbody: NS_HTML + tfoot: NS_HTML + thead: NS_HTML + tr: NS_HTML } -# FIXME namespacify -# all html I presume end_tag_implied = { - dd: true - dt: true - li: true - option: true - optgroup: true - p: true - rb: true - rp: true - rt: true - rtc: true + dd: NS_HTML + dt: NS_HTML + li: NS_HTML + option: NS_HTML + optgroup: NS_HTML + p: NS_HTML + rb: NS_HTML + rp: NS_HTML + rt: NS_HTML + rtc: NS_HTML } el_is_special = (e) -> @@ -424,7 +518,7 @@ svg_attribute_fixes = { diffuseconstant: 'diffuseConstant' edgemode: 'edgeMode' externalresourcesrequired: 'externalResourcesRequired' - filterres: 'filterRes' + # WHATWG removes this: filterres: 'filterRes' filterunits: 'filterUnits' glyphref: 'glyphRef' gradienttransform: 'gradientTransform' @@ -476,6 +570,20 @@ svg_attribute_fixes = { ychannelselector: 'yChannelSelector' zoomandpan: 'zoomAndPan' } +foreign_attr_fixes = { + 'xlink:actuate': 'xlink actuate' + 'xlink:arcrole': 'xlink arcrole' + 'xlink:href': 'xlink href' + 'xlink:role': 'xlink role' + 'xlink:show': 'xlink show' + 'xlink:title': 'xlink title' + 'xlink:type': 'xlink type' + 'xml:base': 'xml base' + 'xml:lang': 'xml lang' + 'xml:space': 'xml space' + 'xmlns': 'xmlns' + 'xmlns:xlink': 'xmlns xlink' +} adjust_mathml_attributes = (t) -> for a in t.attrs_a if a[0] is 'definitionurl' @@ -488,32 +596,40 @@ adjust_svg_attributes = (t) -> return adjust_foreign_attributes = (t) -> # fixfull + for a in t.attrs_a + if foreign_attr_fixes[a[0]]? + a[0] = foreign_attr_fixes[a[0]] return # decode_named_char_ref() # -# The list of named character references is _huge_ so ask the browser to decode -# for us instead of wasting bandwidth/space on including the table here. -# -# Pass without the "&" but with the ";" examples: -# for "&" pass "amp;" -# for "′" pass "x2032;" -g_dncr = { - cache: {} - textarea: document.createElement('textarea') -} -# TODO test this in IE8 +# The list of named character references is _huge_ so if we're running in a +# browser, we get the browser to decode them, rather than increasing the code +# size to include the table. +if context is 'module' + _decode_named_char_ref = require './html5-named-entities.coffee' +else + # TODO test this in IE8 + decode_named_char_ref_el = document.createElement('textarea') + _decode_named_char_ref = (txt) -> + txt = "&#{txt};" + decode_named_char_ref_el.innerHTML = txt + decoded = decode_named_char_ref_el.value + return null if decoded is txt + return decoded +# Pass the name of a named entity _that has a terminating semicolon_ +# Entities without terminating semicolons should use legacy_char_refs[] +# Do not include the "&" or ";" in your argument, eg pass "alpha" +decode_named_char_ref_cache = {} decode_named_char_ref = (txt) -> - txt = "&#{txt}" - decoded = g_dncr.cache[txt] + decoded = decode_named_char_ref_cache[txt] return decoded if decoded? - g_dncr.textarea.innerHTML = txt - decoded = g_dncr.textarea.value - return null if decoded is txt - return g_dncr.cache[txt] = decoded + decoded = _decode_named_char_ref txt + return decode_named_char_ref_cache[txt] = decoded -parse_html = (txt, parse_error_cb = null) -> - cur = 0 # index of next char in txt to be parsed +parse_html = (args_html, args = {}) -> + txt = null + cur = null # index of next char in txt to be parsed # declare doc and tokenizer variables so they're in scope below doc = null open_els = null # stack of open elements @@ -536,61 +652,75 @@ parse_html = (txt, parse_error_cb = null) -> stop_parsing = -> flag_parsing = false + return parse_error = -> - if parse_error_cb? - parse_error_cb cur - else - console.log "Parse error at character #{cur} of #{txt.length}" + if args.error_cb? + args.error_cb cur + return + # http://www.w3.org/TR/html5/syntax.html#push-onto-the-list-of-active-formatting-elements + # "Noah's Ark clause" but with three afe_push = (new_el) -> matches = 0 for el, i in afe + if el.type is TYPE_AFE_MARKER + break if el.name is new_el.name and el.namespace is new_el.namespace + attrs_match = true for k, v of el.attrs - continue unless new_el.attrs[k] is v - for k, v of new_el.attrs - continue unless el.attrs[k] is v - matches += 1 - if matches is 3 - afe.splice i, 1 - break + unless new_el.attrs[k] is v + attrs_match = false + break + if attrs_match + for k, v of new_el.attrs + unless el.attrs[k] is v + attrs_match = false + break + if attrs_match + matches += 1 + if matches is 3 + afe.splice i, 1 + break afe.unshift new_el + return + afe_push_marker = -> afe.unshift new_afe_marker() + return # the functions below impliment the Tree Contstruction algorithm # http://www.w3.org/TR/html5/syntax.html#tree-construction # But first... the helpers template_tag_is_open = -> - for t in open_els - if t.name is 'template' # maybe should also check: and t.namespace is 'html' + for el in open_els + if el.name is 'template' and el.namespace is NS_HTML return true return false is_in_scope_x = (tag_name, scope, namespace) -> - for t in open_els - if t.name is tag_name and (namespace is null or namespace is t.namespace) + for el in open_els + if el.name is tag_name and (namespace is null or namespace is el.namespace) return true - if scope[t.name] is t.namespace + if scope[el.name] is el.namespace return false return false is_in_scope_x_y = (tag_name, scope, scope2, namespace) -> - for t in open_els - if t.name is tag_name and (namespace is null or namespace is t.namespace) + for el in open_els + if el.name is tag_name and (namespace is null or namespace is el.namespace) return true - if scope[t.name] is t.namespace + if scope[el.name] is el.namespace return false - if scope2[t.name] is t.namespace + if scope2[el.name] is el.namespace return false return false standard_scopers = { applet: NS_HTML, caption: NS_HTML, html: NS_HTML, table: NS_HTML, td: NS_HTML, th: NS_HTML, marquee: NS_HTML, object: NS_HTML, - template: NS_HTML, mi: NS_MATHML, + template: NS_HTML, - mo: NS_MATHML, mn: NS_MATHML, ms: NS_MATHML, mtext: NS_MATHML, - 'annotation-xml': NS_MATHML, + mi: NS_MATHML, mo: NS_MATHML, mn: NS_MATHML, ms: NS_MATHML, + mtext: NS_MATHML, 'annotation-xml': NS_MATHML, foreignObject: NS_SVG, desc: NS_SVG, title: NS_SVG } @@ -610,15 +740,16 @@ parse_html = (txt, parse_error_cb = null) -> for t in open_els if t.name is tag_name and (namespace is null or namespace is t.namespace) return true - if t.ns isnt NS_HTML and t.name isnt 'optgroup' and t.name isnt 'option' + if t.namespace isnt NS_HTML and t.name isnt 'optgroup' and t.name isnt 'option' return false return false # this checks for a particular element, not by name - el_is_in_scope = (el) -> - for t in open_els - if t is el + # this requires a namespace match + el_is_in_scope = (needle) -> + for el in open_els + if el is needle return true - if standard_scopers[t.name] is t.namespace + if standard_scopers[el.name] is el.namespace return false return false @@ -634,15 +765,15 @@ parse_html = (txt, parse_error_cb = null) -> open_els.shift() return clear_to_table_body_stopers = { - 'tbody': true - 'tfoot': true - 'thead': true - 'template': true - 'html': true + tbody: NS_HTML + tfoot: NS_HTML + thead: NS_HTML + template: NS_HTML + html: NS_HTML } clear_stack_to_table_body_context = -> loop - if clear_to_table_body_stopers[open_els[0].name]? + if clear_to_table_body_stopers[open_els[0].name] is open_els[0].namespace break open_els.shift() return @@ -680,10 +811,10 @@ parse_html = (txt, parse_error_cb = null) -> loop if node_i is open_els.length - 1 last = true - # fixfull (fragment case) - + if flag_fragment_parsing + node = context_element # 4. If node is a select element, run these substeps: - if node.name is 'select' + if node.name is 'select' and node.namespace is NS_HTML # 1. If last is true, jump to the step below labeled done. unless last # 2. Let ancestor be node. @@ -700,11 +831,11 @@ parse_html = (txt, parse_error_cb = null) -> ancestor = open_els[ancestor_i] # 5. If ancestor is a template node, jump to the step below # labeled done. - if ancestor.name is 'template' + if ancestor.name is 'template' and ancestor.namespace is NS_HTML break # 6. If ancestor is a table node, switch the insertion mode # to "in select in table" and abort these steps. - if ancestor.name is 'table' + if ancestor.name is 'table' and ancestor.namespace is NS_HTML ins_mode = ins_mode_in_select_in_table return # 7. Jump back to the step labeled loop. @@ -714,61 +845,62 @@ parse_html = (txt, parse_error_cb = null) -> return # 5. If node is a td or th element and last is false, then switch # the insertion mode to "in cell" and abort these steps. - if (node.name is 'td' or node.name is 'th') and last is false + if (node.name is 'td' or node.name is 'th') and node.namespace is NS_HTML and last is false ins_mode = ins_mode_in_cell return # 6. If node is a tr element, then switch the insertion mode to "in # row" and abort these steps. - if node.name is 'tr' + if node.name is 'tr' and node.namespace is NS_HTML ins_mode = ins_mode_in_row return # 7. If node is a tbody, thead, or tfoot element, then switch the # insertion mode to "in table body" and abort these steps. - if node.name is 'tbody' or node.name is 'thead' or node.name is 'tfoot' + if (node.name is 'tbody' or node.name is 'thead' or node.name is 'tfoot') and node.namespace is NS_HTML ins_mode = ins_mode_in_table_body return # 8. If node is a caption element, then switch the insertion mode # to "in caption" and abort these steps. - if node.name is 'caption' + if node.name is 'caption' and node.namespace is NS_HTML ins_mode = ins_mode_in_caption return # 9. If node is a colgroup element, then switch the insertion mode # to "in column group" and abort these steps. - if node.name is 'colgroup' + if node.name is 'colgroup' and node.namespace is NS_HTML ins_mode = ins_mode_in_column_group return # 10. If node is a table element, then switch the insertion mode to # "in table" and abort these steps. - if node.name is 'table' + if node.name is 'table' and node.namespace is NS_HTML ins_mode = ins_mode_in_table return # 11. If node is a template element, then switch the insertion mode # to the current template insertion mode and abort these steps. - # fixfull (template insertion mode stack) - + if node.name is 'template' and node.namespace is NS_HTML + ins_mode = template_ins_modes[0] + return # 12. If node is a head element and last is true, then switch the # insertion mode to "in body" ("in body"! not "in head"!) and abort # these steps. (fragment case) - if node.name is 'head' and last + if node.name is 'head' and node.namespace is NS_HTML and last ins_mode = ins_mode_in_body return # 13. If node is a head element and last is false, then switch the # insertion mode to "in head" and abort these steps. - if node.name is 'head' and last is false + if node.name is 'head' and node.namespace is NS_HTML and last is false ins_mode = ins_mode_in_head return # 14. If node is a body element, then switch the insertion mode to # "in body" and abort these steps. - if node.name is 'body' + if node.name is 'body' and node.namespace is NS_HTML ins_mode = ins_mode_in_body return # 15. If node is a frameset element, then switch the insertion mode # to "in frameset" and abort these steps. (fragment case) - if node.name is 'frameset' + if node.name is 'frameset' and node.namespace is NS_HTML ins_mode = ins_mode_in_frameset return # 16. If node is an html element, run these substeps: - if node.name is 'html' + if node.name is 'html' and node.namespace is NS_HTML # 1. If the head element pointer is null, switch the insertion # mode to "before head" and abort these steps. (fragment case) if head_element_pointer is null @@ -789,6 +921,7 @@ parse_html = (txt, parse_error_cb = null) -> node_i += 1 node = open_els[node_i] # 19. Return to the step labeled loop. + return # 8.2.3.2 @@ -820,6 +953,7 @@ parse_html = (txt, parse_error_cb = null) -> afe[i] = el break if i is 0 i -= 1 # Advance + return # http://www.w3.org/TR/html5/syntax.html#adoption-agency-algorithm # adoption agency algorithm @@ -828,20 +962,43 @@ parse_html = (txt, parse_error_cb = null) -> # http://www.w3.org/TR/html5/syntax.html#misnested-tags:-b-p-/b-/p # http://www.w3.org/TR/html5/syntax.html#unclosed-formatting-elements adoption_agency = (subject) -> - debug_log "adoption_agency()" - 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 - el = open_els[0] - open_els.shift() +# this block implements tha W3C spec +# # 1. If the current node is an HTML element whose tag name is subject, +# # then run these substeps: +# # +# # 1. Let element be the current node. +# # +# # 2. Pop element off the stack of open elements. +# # +# # 3. If element is also in the list of active formatting elements, +# # remove the element from the list. +# # +# # 4. Abort the adoption agency algorithm. +# if open_els[0].name is subject and open_els[0].namespace is NS_HTML +# el = open_els.shift() +# # remove it from the list of active formatting elements (if found) +# for t, i in afe +# if t is el +# afe.splice i, 1 +# break +# return +# WHATWG: https://html.spec.whatwg.org/multipage/syntax.html#adoption-agency-algorithm + # If the current node is an HTML element whose tag name is subject, and + # the current node is not in the list of active formatting elements, + # then pop the current node off the stack of open elements, and abort + # these steps. + if open_els[0].name is subject and open_els[0].namespace is NS_HTML # remove it from the list of active formatting elements (if found) - for t, i in afe - if t is el - afe.splice i, 1 + in_afe = false + for el, i in afe + if el is open_els[0] + in_afe = true break - debug_log "aaa: starting off with subject on top of stack, exiting" - return + unless in_afe + open_els.shift() + return + # fall through +# END WHATWG outer = 0 loop if outer >= 8 @@ -861,7 +1018,6 @@ parse_html = (txt, parse_error_cb = null) -> # If there is no such element, then abort these steps and instead # act as described in the "any other end tag" entry above. if fe is null - debug_log "aaa: fe not found in afe" in_body_any_other_end_tag subject return # 6. If formatting element is not in the stack of open elements, @@ -873,7 +1029,6 @@ parse_html = (txt, parse_error_cb = null) -> in_open_els = true break unless in_open_els - debug_log "aaa: fe not found in open_els" parse_error() # "remove it from the list" must mean afe, since it's not in open_els afe.splice fe_of_afe, 1 @@ -882,7 +1037,6 @@ parse_html = (txt, parse_error_cb = null) -> # the element is not in scope, then this is a parse error; abort # these steps. unless el_is_in_scope fe - debug_log "aaa: fe not in scope" parse_error() return # 8. If formatting element is not the current node, this is a parse @@ -908,7 +1062,6 @@ parse_html = (txt, parse_error_cb = null) -> # formatting element from the list of active formatting elements, # and finally abort these steps. if fb is null - debug_log "aaa: no fb" loop t = open_els.shift() if t is fe @@ -940,21 +1093,12 @@ parse_html = (txt, parse_error_cb = null) -> node_next = open_els[i + 1] break node = node_next ? node_above - debug_log "inner loop #{inner}" - 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}" - 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 "node: #{node.serialize true, true}" # TODO make sure node_above gets re-set if/when node is removed from open_els # 4. If node is formatting element, then go to the next step in # the overall algorithm. if node is fe break - debug_log "the meat" # 5. If inner loop counter is greater than three and node is in # the list of active formatting elements, then remove node from # the list of active formatting elements. @@ -963,23 +1107,19 @@ parse_html = (txt, parse_error_cb = null) -> if t is node if inner > 3 afe.splice i, 1 - debug_log "max out inner" else node_in_afe = true - debug_log "in afe" break # 6. If node is not in the list of active formatting elements, # then remove node from the stack of open elements and then go # back to the step labeled inner loop. unless node_in_afe - debug_log "not in afe" for t, i in open_els if t is node node_above = open_els[i + 1] open_els.splice i, 1 break continue - debug_log "the bones" # 7. create an element for the token for which the element node # was created, in the HTML namespace, with common ancestor as # the intended parent; replace the entry for node in the list @@ -991,13 +1131,11 @@ parse_html = (txt, parse_error_cb = null) -> for t, i in afe if t is node afe[i] = new_node - debug_log "replaced in afe" break for t, i in open_els if t is node node_above = open_els[i + 1] open_els[i] = new_node - debug_log "replaced in open_els" break node = new_node # 8. If last node is furthest block, then move the @@ -1007,29 +1145,23 @@ parse_html = (txt, parse_error_cb = null) -> for t, i in afe if t is bookmark afe.splice i, 1 - debug_log "removed bookmark" break for t, i in afe if t is node # "after" means lower afe.splice i, 0, bookmark # "after as <- - debug_log "placed bookmark after node" - debug_log "node: #{node.id} afe: #{serialize_els afe, true, true}" break # 9. Insert last node into node, first removing it from its # previous parent node if any. if last_node.parent? - debug_log "last_node has parent" for c, i in last_node.parent.children if c is last_node - debug_log "removing last_node from parent" last_node.parent.children.splice i, 1 break node.children.push last_node last_node.parent = node # 10. Let last node be node. last_node = node - debug_log "at last" # 11. Return to the step labeled inner loop. # 14. Insert whatever last node ended up being in the previous step # at the appropriate place for inserting a node, but using common @@ -1040,36 +1172,15 @@ parse_html = (txt, parse_error_cb = null) -> # * last_node is fb # * last_node is still in the tree (not a duplicate) if last_node.parent? - debug_log "FEFIRST? last_node has parent" for c, i in last_node.parent.children if c is last_node - debug_log "removing last_node from parent" last_node.parent.children.splice i, 1 break - - debug_log "after aaa inner loop" - debug_log "ca: #{ca.name}##{ca.id} children: #{serialize_els ca.children, true, true}" - 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 doc.children, false, true}" - - debug_log "insert" - - # can't use standard insert token thing, because it's already in # open_els and must stay at it's current position in open_els dest = adjusted_insertion_location ca dest[0].children.splice dest[1], 0, last_node last_node.parent = dest[0] - - - debug_log "ca: #{ca.name}##{ca.id} children: #{serialize_els ca.children, true, true}" - 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 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 # intended parent. @@ -1107,24 +1218,22 @@ parse_html = (txt, parse_error_cb = null) -> open_els.splice i, 0, new_element 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 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" + return # http://www.w3.org/TR/html5/syntax.html#close-a-p-element close_p_element = -> generate_implied_end_tags 'p' # arg is exception - if open_els[0].name isnt 'p' + unless open_els[0].name is 'p' and open_els[0].namespace is NS_HTML parse_error() while open_els.length > 1 # just in case el = open_els.shift() - if el.name is 'p' + if el.name is 'p' and el.namespace is NS_HTML return + return close_p_if_in_button_scope = -> - if is_in_button_scope 'p' + if is_in_button_scope 'p', NS_HTML close_p_element() + return # http://www.w3.org/TR/html5/syntax.html#insert-a-character # aka insert_a_character = (t) -> @@ -1137,7 +1246,7 @@ parse_html = (txt, parse_error_cb = null) -> prev.text += t.text return dest[0].children.splice dest[1], 0, t - + return # 8.2.5 http://www.w3.org/TR/html5/syntax.html#tree-construction process_token = (t) -> @@ -1149,7 +1258,7 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode t return if is_mathml_text_integration_point(acn) - if t.type is TYPE_START_TAG and (t.name is 'mglyph' or t.name is 'malignmark') + if t.type is TYPE_START_TAG and not (t.name is 'mglyph' or t.name is 'malignmark') ins_mode t return if t.type is TYPE_TEXT @@ -1184,14 +1293,14 @@ parse_html = (txt, parse_error_cb = null) -> # If foster parenting is enabled and target is a table, tbody, tfoot, # thead, or tr element Foster parenting happens when content is # misnested in tables. - if flag_foster_parenting and foster_parenting_targets[target.name] + if flag_foster_parenting and foster_parenting_targets[target.name] is target.namespace loop # once. this is here so we can ``break`` to "abort these substeps" # 1. Let last template be the last template element in the # stack of open elements, if any. last_template = null last_template_i = null for el, i in open_els - if el.name is 'template' + if el.name is 'template' and el.namespace is NS_HTML last_template = el last_template_i = i break @@ -1200,7 +1309,7 @@ parse_html = (txt, parse_error_cb = null) -> last_table = null last_table_i for el, i in open_els - if el.name is 'table' + if el.name is 'table' and el.namespace is NS_HTML last_table = el last_table_i = i break @@ -1222,6 +1331,7 @@ parse_html = (txt, parse_error_cb = null) -> # this is odd target = open_els[open_els.length - 1] target_i = target.children.length + break # 5. If last table has a parent element, then let adjusted # insertion location be inside last table's parent element, # immediately before last table, and abort these substeps. @@ -1292,13 +1402,14 @@ parse_html = (txt, parse_error_cb = null) -> return el # http://www.w3.org/TR/html5/syntax.html#insert-an-html-element insert_html_element = (token) -> - insert_foreign_element token, NS_HTML + return insert_foreign_element token, NS_HTML # http://www.w3.org/TR/html5/syntax.html#insert-a-comment # position should be [node, index_within_children] insert_comment = (t, position = null) -> position ?= adjusted_insertion_location() position[0].children.splice position[1], 0, t + return # 8.2.5.2 # http://www.w3.org/TR/html5/syntax.html#generic-raw-text-element-parsing-algorithm @@ -1307,23 +1418,55 @@ parse_html = (txt, parse_error_cb = null) -> tok_state = tok_state_rawtext original_ins_mode = ins_mode ins_mode = ins_mode_text + return parse_generic_rcdata_text = (t) -> insert_html_element t tok_state = tok_state_rcdata original_ins_mode = ins_mode ins_mode = ins_mode_text + return # 8.2.5.3 http://www.w3.org/TR/html5/syntax.html#closing-elements-that-have-implied-end-tags # http://www.w3.org/TR/html5/syntax.html#generate-implied-end-tags generate_implied_end_tags = (except = null) -> - while end_tag_implied[open_els[0].name] and open_els[0].name isnt except + while end_tag_implied[open_els[0].name] is open_els[0].namespace and open_els[0].name isnt except open_els.shift() + return # 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 + is_quirks_yes_doctype = (t) -> + if t.flag 'force-quirks' + return true + if t.name isnt 'html' + return true + if t.public_identifier? + pi = t.public_identifier.toLowerCase() + for p in quirks_yes_pi_prefixes + if pi.substr(0, p.length) is p + return true + if pi is '-//w3o//dtd w3 html strict 3.0//en//' or pi is '-/w3c/dtd html 4.0 transitional/en' or pi is 'html' + return true + if t.system_identifier? + if t.system_identifier.toLowerCase() is 'http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd' + return true + else if t.public_identifier? + # already did this: pi = t.public_identifier.toLowerCase() + if pi.substr(0, 32) is '-//w3c//dtd html 4.01 frameset//' or pi.substr(0, 36) is '-//w3c//dtd html 4.01 transitional//' + return true + return false + is_quirks_limited_doctype = (t) -> + if t.public_identifier? + pi = t.public_identifier.toLowerCase() + if pi.substr(0, 32) is '-//w3c//dtd xhtml 1.0 frameset//' or pi.substr(0, 36) is '-//w3c//dtd xhtml 1.0 transitional//' + return true + if t.system_identifier? + if pi.substr(0, 32) is '-//w3c//dtd html 4.01 frameset//' or pi.substr(0, 36) is '-//w3c//dtd html 4.01 transitional//' + return true + return false ins_mode_initial = (t) -> if is_space_tok t return @@ -1332,13 +1475,20 @@ parse_html = (txt, parse_error_cb = null) -> doc.children.push t return if t.type is TYPE_DOCTYPE - # FIXME check identifiers, set quirks, etc - # fixfull + # fixfull syntax error from first paragraph and following bullets + # fixfull set doc.doctype + # fixfull is the "not an iframe srcdoc" thing relevant? + if is_quirks_yes_doctype t + doc.flag 'quirks mode', QUIRKS_YES + else if is_quirks_limited_doctype t + doc.flag 'quirks mode', QUIRKS_LIMITED doc.children.push t ins_mode = ins_mode_before_html return # Anything else - #fixfull (iframe, quirks) + # fixfull not iframe srcdoc? + parse_error() + doc.flag 'quirks mode', QUIRKS_YES ins_mode = ins_mode_before_html process_token t return @@ -1356,6 +1506,7 @@ parse_html = (txt, parse_error_cb = null) -> if t.type is TYPE_START_TAG and t.name is 'html' el = token_to_element t, NS_HTML, doc doc.children.push el + el.document = doc open_els.unshift(el) # fixfull (big paragraph in spec about manifest, fragment, urls, etc) ins_mode = ins_mode_before_head @@ -1367,9 +1518,9 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return # Anything else - html_tok = new_open_tag 'html' - el = token_to_element html_tok, NS_HTML, doc + el = token_to_element new_open_tag('html'), NS_HTML, doc doc.children.push el + el.document = doc open_els.unshift el # ?fixfull browsing context ins_mode = ins_mode_before_head @@ -1393,6 +1544,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -1400,17 +1552,18 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return # Anything else - head_tok = new_open_tag 'head' - el = insert_html_element head_tok + el = insert_html_element new_open_tag 'head' head_element_pointer = el ins_mode = ins_mode_in_head process_token t + return # 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 ins_mode = ins_mode_after_head process_token t + return ins_mode_in_head = (t) -> if t.type is TYPE_TEXT and (t.text is "\t" or t.text is "\n" or t.text is "\u000c" or t.text is ' ') insert_character t @@ -1438,7 +1591,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -1477,7 +1630,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() loop el = open_els.shift() - if el.name is 'template' + if el.name is 'template' and el.namespace is NS_HTML break clear_afe_to_marker() template_ins_modes.shift() @@ -1489,6 +1642,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return ins_mode_in_head_else t + return # 8.2.5.4.5 http://www.w3.org/TR/html5/syntax.html#parsing-main-inheadnoscript ins_mode_in_head_noscript_else = (t) -> @@ -1496,18 +1650,19 @@ parse_html = (txt, parse_error_cb = null) -> open_els.shift() ins_mode = ins_mode_in_head process_token t + return ins_mode_in_head_noscript = (t) -> 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' @@ -1520,8 +1675,6 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode_in_head_noscript_else t return - - # 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' @@ -1555,11 +1708,10 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() open_els.unshift head_element_pointer ins_mode_in_head t - for el, i of open_els + for el, i in 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 @@ -1572,20 +1724,27 @@ parse_html = (txt, parse_error_cb = null) -> return # Anything else ins_mode_after_head_else t + return # 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 el, i in open_els - if el.namespace is NS_HTML and el.name is name + node = open_els[0] + loop + if node.name is name and node.namespace is NS_HTML generate_implied_end_tags name # arg is exception - parse_error() unless i is 0 - while i >= 0 - open_els.shift() - i -= 1 - return - if special_elements[el.name] is el.namespace + unless node is open_els[0] + parse_error() + loop + el = open_els.shift() + if el is node + return + if special_elements[node.name] is node.namespace parse_error() return + for el, i in open_els + if node is el + node = open_els[i + 1] + break return ins_mode_in_body = (t) -> if t.type is TYPE_TEXT and t.text is "\u0000" @@ -1610,7 +1769,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return if template_tag_is_open() root_attrs = open_els[open_els.length - 1].attrs - for a of t.attrs_a + for a in t.attrs_a root_attrs[a[0]] = a[1] unless root_attrs[a[0]]? return @@ -1621,11 +1780,11 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return if open_els.length < 2 second = open_els[open_els.length - 2] - return unless second.ns is NS_HTML + return unless second.namespace is NS_HTML return unless second.name is 'body' return if template_tag_is_open() - frameset_ok_flag = false - for a of t.attrs_a + flag_frameset_ok = false + for a in t.attrs_a second.attrs[a[0]] = a[1] unless second.attrs[a[0]]? return if t.type is TYPE_START_TAG and t.name is 'frameset' @@ -1633,9 +1792,10 @@ parse_html = (txt, parse_error_cb = null) -> return if open_els.length < 2 second_i = open_els.length - 2 second = open_els[second_i] - return unless second.ns is NS_HTML + return unless second.namespace is NS_HTML return unless second.name is 'body' - flag_frameset_ok = false + if flag_frameset_ok is false + return if second.parent? for el, i in second.parent.children if el is second @@ -1664,7 +1824,7 @@ parse_html = (txt, parse_error_cb = null) -> stop_parsing() return if t.type is TYPE_END_TAG and t.name is 'body' - unless is_in_scope 'body' + unless is_in_scope 'body', NS_HTML parse_error() return ok_tags = { @@ -1681,7 +1841,7 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode = ins_mode_after_body return if t.type is TYPE_END_TAG and t.name is 'html' - unless is_in_scope 'body' + unless is_in_scope 'body', NS_HTML parse_error() return ok_tags = { @@ -1704,7 +1864,7 @@ parse_html = (txt, parse_error_cb = null) -> return if t.type is TYPE_START_TAG and h_tags[t.name]? close_p_if_in_button_scope() - if h_tags[open_els[0]] is NS_HTML + if h_tags[open_els[0].name] is open_els[0].namespace parse_error() open_els.shift() insert_html_element t @@ -1712,11 +1872,7 @@ parse_html = (txt, parse_error_cb = null) -> 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' @@ -1911,6 +2067,10 @@ parse_html = (txt, parse_error_cb = null) -> return if t.type is TYPE_START_TAG and t.name is 'nobr' reconstruct_afe() + if is_in_scope 'nobr', NS_HTML + parse_error() + adoption_agency 'nobr' + reconstruct_afe() el = insert_html_element t afe_push el return @@ -1937,14 +2097,16 @@ parse_html = (txt, parse_error_cb = null) -> clear_afe_to_marker() return if t.type is TYPE_START_TAG and t.name is 'table' - close_p_if_in_button_scope() # fixfull quirksmode thing + unless doc.flag('quirks mode') is QUIRKS_YES + close_p_if_in_button_scope() # test insert_html_element t flag_frameset_ok = false ins_mode = ins_mode_in_table return if t.type is TYPE_END_TAG and t.name is 'br' parse_error() - t.type is 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() @@ -1961,7 +2123,8 @@ parse_html = (txt, parse_error_cb = null) -> unless is_input_hidden_tok t flag_frameset_ok = false return - if t.type is TYPE_START_TAG and (t.name is 'param' or t.name is 'source' or t.name is 'track') + if t.type is TYPE_START_TAG and (t.name is 'menuitem' or t.name is 'param' or t.name is 'source' or t.name is 'track') + # WHATWG adds 'menuitem' for this block insert_html_element t open_els.shift() t.acknowledge_self_closing() @@ -2021,8 +2184,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -2056,20 +2218,37 @@ parse_html = (txt, parse_error_cb = null) -> reconstruct_afe() insert_html_element t return - if t.type is TYPE_START_TAG and (t.name is 'rb' or t.name is 'rp' or t.name is 'rtc') +# this comment block implements the W3C spec +# if t.type is TYPE_START_TAG and (t.name is 'rb' or t.name is 'rp' or t.name is 'rtc') +# if is_in_scope 'ruby', NS_HTML +# generate_implied_end_tags() +# unless open_els[0].name is 'ruby' and open_els[0].namespace is NS_HTML +# parse_error() +# insert_html_element t +# return +# if t.type is TYPE_START_TAG and t.name is 'rt' +# if is_in_scope 'ruby', NS_HTML +# generate_implied_end_tags 'rtc' # arg is exception +# unless (open_els[0].name is 'ruby' or open_els[0].name is 'rtc') and open_els[0].namespace is NS_HTML +# parse_error() +# insert_html_element t +# return +# below implements the WHATWG spec https://html.spec.whatwg.org/multipage/syntax.html#parsing-main-inbody + if t.type is TYPE_START_TAG and (t.name is 'rb' or t.name is 'rtc') if is_in_scope 'ruby', NS_HTML generate_implied_end_tags() unless open_els[0].name is 'ruby' and open_els[0].namespace is NS_HTML parse_error() insert_html_element t return - if t.type is TYPE_START_TAG and t.name is 'rt' + if t.type is TYPE_START_TAG and (t.name is 'rp' or t.name is 'rt') if is_in_scope 'ruby', NS_HTML - generate_implied_end_tags 'rtc' # arg is exception + generate_implied_end_tags 'rtc' unless (open_els[0].name is 'ruby' or open_els[0].name is 'rtc') and open_els[0].namespace is NS_HTML parse_error() insert_html_element t return +# end WHATWG chunk if t.type is TYPE_START_TAG and t.name is 'math' reconstruct_afe() adjust_mathml_attributes t @@ -2100,19 +2279,6 @@ parse_html = (txt, parse_error_cb = null) -> return return - ins_mode_in_table_else = (t) -> - parse_error() - flag_foster_parenting = true # FIXME - ins_mode_in_body t - flag_foster_parenting = false - can_in_table = { # FIXME do this inline like everywhere else - 'table': true - 'tbody': true - 'tfoot': true - 'thead': true - 'tr': true - } - # 8.2.5.4.8 http://www.w3.org/TR/html5/syntax.html#parsing-main-incdata ins_mode_text = (t) -> if t.type is TYPE_TEXT @@ -2120,7 +2286,7 @@ parse_html = (txt, parse_error_cb = null) -> return if t.type is TYPE_EOF parse_error() - if open_els[0].name is 'script' + if open_els[0].name is 'script' and open_els[0].namespace is NS_HTML open_els[0].flag 'already started', true open_els.shift() ins_mode = original_ins_mode @@ -2136,16 +2302,23 @@ parse_html = (txt, parse_error_cb = null) -> open_els.shift() ins_mode = original_ins_mode return - console.log 'warning: end of ins_mode_text reached' + return # the functions below implement the tokenizer stats described here: # http://www.w3.org/TR/html5/syntax.html#tokenization # 8.2.5.4.9 http://www.w3.org/TR/html5/syntax.html#parsing-main-intable + ins_mode_in_table_else = (t) -> + parse_error() + flag_foster_parenting = true + ins_mode_in_body t + flag_foster_parenting = false + return ins_mode_in_table = (t) -> switch t.type when TYPE_TEXT - if can_in_table[t.name] + if (open_els[0].name is 'table' or open_els[0].name is 'tbody' or open_els[0].name is 'tfoot' or open_els[0].name is 'thead' or open_els[0].name is 'tr') and open_els[0].namespace is NS_HTML + pending_table_character_tokens = [] original_ins_mode = ins_mode ins_mode = ins_mode_in_table_text process_token t @@ -2182,17 +2355,17 @@ parse_html = (txt, parse_error_cb = null) -> process_token t when 'table' parse_error() - if is_in_table_scope 'table' + if is_in_table_scope 'table', NS_HTML loop el = open_els.shift() - if el.name is 'table' + if el.name is 'table' and el.namespace is NS_HTML break reset_ins_mode() process_token t when 'style', 'script', 'template' ins_mode_in_head t when 'input' - if is_input_hidden_tok t + unless is_input_hidden_tok t ins_mode_in_table_else t else parse_error() @@ -2212,14 +2385,14 @@ parse_html = (txt, parse_error_cb = null) -> when TYPE_END_TAG switch t.name when 'table' - if is_in_table_scope 'table' + if is_in_table_scope 'table', NS_HTML loop el = open_els.shift() - if el.name is 'table' + if el.name is 'table' and el.namespace is NS_HTML break reset_ins_mode() else - parse_error + parse_error() when 'body', 'caption', 'col', 'colgroup', 'html', 'tbody', 'td', 'tfoot', 'th', 'thead', 'tr' parse_error() when 'template' @@ -2230,12 +2403,13 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode_in_body t else ins_mode_in_table_else t + return # 8.2.5.4.10 http://www.w3.org/TR/html5/syntax.html#parsing-main-intabletext ins_mode_in_table_text = (t) -> if t.type is TYPE_TEXT and t.text is "\u0000" - # huh? I thought the tokenizer didn't emit these + # from javascript? parse_error() return if t.type is TYPE_TEXT @@ -2252,21 +2426,22 @@ parse_html = (txt, parse_error_cb = null) -> 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) + ins_mode_in_table_else old + pending_table_character_tokens = [] ins_mode = original_ins_mode process_token t + return # 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' + if is_in_table_scope 'caption', NS_HTML generate_implied_end_tags() if open_els[0].name isnt 'caption' parse_error() loop el = open_els.shift() - if el.name is 'caption' + if el.name is 'caption' and el.namespace is NS_HTML break clear_afe_to_marker() ins_mode = ins_mode_in_table @@ -2276,10 +2451,10 @@ parse_html = (txt, parse_error_cb = null) -> 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' + if is_in_table_scope 'caption', NS_HTML loop el = open_els.shift() - if el.name is 'caption' + if el.name is 'caption' and el.namespace is NS_HTML break clear_afe_to_marker() ins_mode = ins_mode_in_table @@ -2291,6 +2466,7 @@ parse_html = (txt, parse_error_cb = null) -> return # Anything else ins_mode_in_body t + return # 8.2.5.4.12 http://www.w3.org/TR/html5/syntax.html#parsing-main-incolgroup ins_mode_in_column_group = (t) -> @@ -2312,7 +2488,7 @@ parse_html = (txt, parse_error_cb = null) -> t.acknowledge_self_closing() return if t.type is TYPE_END_TAG and t.name is 'colgroup' - if open_els[0].name is 'colgroup' + if open_els[0].name is 'colgroup' and open_els.namespace is NS_HTML open_els.shift() ins_mode = ins_mode_in_table else @@ -2351,7 +2527,7 @@ parse_html = (txt, parse_error_cb = null) -> process_token t return if t.type is TYPE_END_TAG and (t.name is 'tbody' or t.name is 'tfoot' or t.name is 'thead') - unless is_in_table_scope t.name # fixfull check namespace + unless is_in_table_scope t.name, NS_HTML parse_error() return clear_stack_to_table_body_context() @@ -2361,10 +2537,10 @@ parse_html = (txt, parse_error_cb = null) -> 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 'tfoot' or t.name is 'thead')) or (t.type is TYPE_END_TAG and t.name is 'table') has = false for el in open_els - if el.name is 'tbody' or el.name is 'tfoot' or el.name is 'thead' + if el.namespace is NS_HTML and (el.name is 'tbody' or el.name is 'tfoot' or el.name is 'thead') has = true break - if table_scopers[el.name] + if table_scopers[el.name] is el.namespace break if !has parse_error() @@ -2379,6 +2555,7 @@ parse_html = (txt, parse_error_cb = null) -> return # Anything else ins_mode_in_table t + return # 8.2.5.4.14 http://www.w3.org/TR/html5/syntax.html#parsing-main-intr ins_mode_in_row = (t) -> @@ -2389,7 +2566,7 @@ parse_html = (txt, parse_error_cb = null) -> afe_push_marker() return if t.type is TYPE_END_TAG and t.name is 'tr' - if is_in_table_scope 'tr' + if is_in_table_scope 'tr', NS_HTML clear_stack_to_table_row_context() open_els.shift() ins_mode = ins_mode_in_table_body @@ -2397,7 +2574,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() 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 'tfoot' or t.name is 'thead' or t.name is 'tr')) or t.type is TYPE_END_TAG and t.name is 'table' - if is_in_table_scope 'tr' + if is_in_table_scope 'tr', NS_HTML clear_stack_to_table_row_context() open_els.shift() ins_mode = ins_mode_in_table_body @@ -2406,8 +2583,8 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return if t.type is TYPE_END_TAG and (t.name is 'tbody' or t.name is 'tfoot' or t.name is 'thead') - if is_in_table_scope t.name # fixfull namespace - if is_in_table_scope 'tr' + if is_in_table_scope t.name, NS_HTML + if is_in_table_scope 'tr', NS_HTML clear_stack_to_table_row_context() open_els.shift() ins_mode = ins_mode_in_table_body @@ -2420,29 +2597,31 @@ parse_html = (txt, parse_error_cb = null) -> return # Anything else ins_mode_in_table t + return # http://www.w3.org/TR/html5/syntax.html#close-the-cell close_the_cell = -> generate_implied_end_tags() - unless open_els[0].name is 'td' or open_els[0] is 'th' + unless (open_els[0].name is 'td' or open_els[0] is 'th') and open_els[0].namespace is NS_HTML parse_error() loop el = open_els.shift() - if el.name is 'td' or el.name is 'th' + if el.namespace is NS_HTML and (el.name is 'td' or el.name is 'th') break clear_afe_to_marker() ins_mode = ins_mode_in_row + return # 8.2.5.4.15 http://www.w3.org/TR/html5/syntax.html#parsing-main-intd ins_mode_in_cell = (t) -> if t.type is TYPE_END_TAG and (t.name is 'td' or t.name is 'th') - if is_in_table_scope t.name + if is_in_table_scope t.name, NS_HTML generate_implied_end_tags() - if open_els[0].name isnt t.name - parse_error + unless (open_els[0].name is t.name) and open_els[0].namespace is NS_HTML + parse_error() loop el = open_els.shift() - if el.name is t.name + if el.name is t.name and el.namespace is NS_HTML break clear_afe_to_marker() ins_mode = ins_mode_in_row @@ -2452,10 +2631,10 @@ parse_html = (txt, parse_error_cb = null) -> 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') has = false for el in open_els - if el.name is 'td' or el.name is 'th' + if el.namespace is NS_HTML and (el.name is 'td' or el.name is 'th') has = true break - if table_scopers[el.name] + if table_scopers[el.name] is el.namespace break if !has parse_error() @@ -2467,7 +2646,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() return if t.type is TYPE_END_TAG and (t.name is 'table' or t.name is 'tbody' or t.name is 'tfoot' or t.name is 'thead' or t.name is 'tr') - if is_in_table_scope t.name # fixfull namespace + if is_in_table_scope t.name, NS_HTML close_the_cell() process_token t else @@ -2475,6 +2654,7 @@ parse_html = (txt, parse_error_cb = null) -> return # Anything Else ins_mode_in_body t + return # 8.2.5.4.16 http://www.w3.org/TR/html5/syntax.html#parsing-main-inselect ins_mode_in_select = (t) -> @@ -2494,36 +2674,37 @@ parse_html = (txt, parse_error_cb = null) -> 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' + if open_els[0].name is 'option' and open_els[0].namespace is NS_HTML 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' + if open_els[0].name is 'option' and open_els[0].namespace is NS_HTML open_els.shift() - if open_els[0].name is 'optgroup' + if open_els[0].name is 'optgroup' and open_els[0].namespace is NS_HTML 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' + if open_els[0].name is 'option' and open_els[0].namespace is NS_HTML + if open_els[1].name is 'optgroup' and open_els[0].namespace is NS_HTML + open_els.shift() + if open_els[0].name is 'optgroup' and open_els[0].namespace is NS_HTML 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' + if open_els[0].name is 'option' and open_els[0].namespace is NS_HTML 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' + if is_in_select_scope 'select', NS_HTML loop el = open_els.shift() - if el.name is 'select' + if el.name is 'select' and el.namespace is NS_HTML break reset_ins_mode() else @@ -2533,7 +2714,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() loop el = open_els.shift() - if el.name is 'select' + if el.name is 'select' and el.namespace is NS_HTML break reset_ins_mode() # spec says that this is the same as but it doesn't say @@ -2541,11 +2722,11 @@ parse_html = (txt, parse_error_cb = null) -> 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' + unless is_in_select_scope 'select', NS_HTML return loop el = open_els.shift() - if el.name is 'select' + if el.name is 'select' and el.namespace is NS_HTML break reset_ins_mode() process_token t @@ -2566,7 +2747,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() loop el = open_els.shift() - if el.name is 'select' + if el.name is 'select' and el.namespace is NS_HTML break reset_ins_mode() process_token t @@ -2577,7 +2758,7 @@ parse_html = (txt, parse_error_cb = null) -> return loop el = open_els.shift() - if el.name is 'select' + if el.name is 'select' and el.namespace is NS_HTML break reset_ins_mode() process_token t @@ -2634,12 +2815,13 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() loop el = open_els.shift() - if el.name is 'template' # fixfull check namespace + if el.name is 'template' and el.namespace is NS_HTML break clear_afe_to_marker() template_ins_modes.shift() reset_ins_mode() process_token t + return # 8.2.5.4.19 http://www.w3.org/TR/html5/syntax.html#parsing-main-afterbody ins_mode_after_body = (t) -> @@ -2647,7 +2829,8 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode_in_body t return if t.type is TYPE_COMMENT - insert_comment t, [open_els[0], open_els[0].children.length] + first = open_els[open_els.length - 1] + insert_comment t, [first, first.children.length] return if t.type is TYPE_DOCTYPE parse_error() @@ -2656,7 +2839,9 @@ parse_html = (txt, parse_error_cb = null) -> ins_mode_in_body t return if t.type is TYPE_END_TAG and t.name is 'html' - # fixfull fragment case + if flag_fragment_parsing + parse_error() + return ins_mode = ins_mode_after_after_body return if t.type is TYPE_EOF @@ -2666,6 +2851,7 @@ parse_html = (txt, parse_error_cb = null) -> parse_error() ins_mode = ins_mode_in_body process_token t + return # 8.2.5.4.20 http://www.w3.org/TR/html5/syntax.html#parsing-main-inframeset ins_mode_in_frameset = (t) -> @@ -2685,7 +2871,6 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -2702,7 +2887,6 @@ parse_html = (txt, parse_error_cb = null) -> 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() @@ -2726,7 +2910,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 + ins_mode = ins_mode_after_after_frameset return if t.type is TYPE_START_TAG and t.name is 'noframes' ins_mode_in_head t @@ -2752,6 +2936,7 @@ parse_html = (txt, parse_error_cb = null) -> # Anything else parse_error() ins_mode = ins_mode_in_body + process_token t return # 8.2.5.4.23 http://www.w3.org/TR/html5/syntax.html#the-after-after-frameset-insertion-mode @@ -2796,6 +2981,7 @@ parse_html = (txt, parse_error_cb = null) -> if t.name is 'script' t.acknowledge_self_closing() in_foreign_content_end_script() + # fixfull else open_els.shift() t.acknowledge_self_closing() @@ -2825,8 +3011,7 @@ parse_html = (txt, parse_error_cb = null) -> return loop # is this safe? open_els.shift() - cn = open_els[0] - if is_mathml_text_integration_point(cn) or is_html_integration(cn) or cn.namespace is NS_HTML + if is_mathml_text_integration_point(open_els[0]) or is_html_integration(open_els[0]) or open_els[0].namespace is NS_HTML break process_token t return @@ -2837,9 +3022,11 @@ parse_html = (txt, parse_error_cb = null) -> in_foreign_content_end_script() return if t.type is TYPE_END_TAG - if open_els[0].name.toLowerCase() isnt t.name + i = 0 + node = open_els[i] + if node.name.toLowerCase() isnt t.name parse_error() - for node in open_els + loop if node is open_els[open_els.length - 1] return if node.name.toLowerCase() is t.name @@ -2847,9 +3034,12 @@ parse_html = (txt, parse_error_cb = null) -> el = open_els.shift() if el is node return + i += 1 + node = open_els[i] if node.namespace is NS_HTML break ins_mode t # explicitly call HTML insertion mode + return # 8.2.4.1 http://www.w3.org/TR/html5/syntax.html#data-state @@ -2935,50 +3125,55 @@ parse_html = (txt, parse_error_cb = null) -> # 8.2.4.8 http://www.w3.org/TR/html5/syntax.html#tag-open-state tok_state_tag_open = -> - switch c = txt.charAt(cur++) - when '!' - tok_state = tok_state_markup_declaration_open - when '/' - tok_state = tok_state_end_tag_open - when '?' - parse_error() - tok_cur_tag = new_comment_token '?' - tok_state = tok_state_bogus_comment - else - if is_lc_alpha(c) - tok_cur_tag = new_open_tag c - tok_state = tok_state_tag_name - else if is_uc_alpha(c) - tok_cur_tag = new_open_tag c.toLowerCase() - tok_state = tok_state_tag_name - else - parse_error() - tok_state = tok_state_data - cur -= 1 # we didn't parse/handle the char after < - return new_text_node '<' - return null + c = txt.charAt(cur++) + if c is '!' + tok_state = tok_state_markup_declaration_open + return + if c is '/' + tok_state = tok_state_end_tag_open + return + if is_uc_alpha(c) + tok_cur_tag = new_open_tag c.toLowerCase() + tok_state = tok_state_tag_name + return + if is_lc_alpha(c) + tok_cur_tag = new_open_tag c + tok_state = tok_state_tag_name + return + if c is '?' + parse_error() + tok_cur_tag = new_comment_token '?' # FIXME right? + tok_state = tok_state_bogus_comment + return + # Anything else + parse_error() + tok_state = tok_state_data + cur -= 1 # we didn't parse/handle the char after < + return new_text_node '<' # 8.2.4.9 http://www.w3.org/TR/html5/syntax.html#end-tag-open-state tok_state_end_tag_open = -> - switch c = txt.charAt(cur++) - when '>' - parse_error() - tok_state = tok_state_data - when '' # EOF - parse_error() - tok_state = tok_state_data - return new_text_node '' + parse_error() + tok_state = tok_state_data + return + if c is '' # EOF + parse_error() + tok_state = tok_state_data + return new_text_node ' # http://www.w3.org/TR/html5/syntax.html#appropriate-end-tag-token is_appropriate_end_tag = (t) -> - # spec says to check against "the tag name of the last start tag to - # have been emitted from this tokenizer", but this is only called from - # the various "raw" states, which I'm pretty sure all push the start - # token onto open_els. TODO: verify this after the script data states - # are implemented - debug_log "#{t.type}, #{t.name} open_els: #{serialize_els open_els, true, true}" + # fixfull: this assumes that open_els[0].name is "the tag name of the last + # start tag to have been emitted from this tokenizer" return t.type is TYPE_END_TAG and t.name is open_els[0].name # 8.2.4.13 http://www.w3.org/TR/html5/syntax.html#rcdata-end-tag-name-state @@ -3184,6 +3375,11 @@ parse_html = (txt, parse_error_cb = null) -> tok_state = tok_state_self_closing_start_tag return # fall through + if c is '>' + if is_appropriate_end_tag tok_cur_tag + tok_state = tok_state_data + return tok_cur_tag + # fall through if is_uc_alpha(c) tok_cur_tag.name += c.toLowerCase() temporary_buffer += c @@ -3303,7 +3499,7 @@ parse_html = (txt, parse_error_cb = null) -> # Anything else tok_state = tok_state_script_data_escaped cur -= 1 # Reconsume - return new_character_token c + return new_character_token '<' # 8.2.4.26 http://www.w3.org/TR/html5/syntax.html#script-data-escaped-end-tag-open-state tok_state_script_data_escaped_end_tag_open = -> @@ -3336,6 +3532,11 @@ parse_html = (txt, parse_error_cb = null) -> tok_state = tok_state_self_closing_start_tag return # fall through + if c is '>' + if is_appropriate_end_tag tok_cur_tag + tok_state = tok_state_data + return tok_cur_tag + # fall through if is_uc_alpha(c) tok_cur_tag.name += c.toLowerCase() temporary_buffer += c.toLowerCase() @@ -3516,16 +3717,16 @@ parse_html = (txt, parse_error_cb = null) -> return tmp when "\u0000" parse_error() - tok_cur_tag.attrs_a[0][0] = "\ufffd" + tok_cur_tag.attrs_a[0][0] += "\ufffd" when '"', "'", '<' parse_error() - tok_cur_tag.attrs_a[0][0] = c + tok_cur_tag.attrs_a[0][0] += c when '' # EOF parse_error() tok_state = tok_state_data else if is_uc_alpha(c) - tok_cur_tag.attrs_a[0][0] = c.toLowerCase() + tok_cur_tag.attrs_a[0][0] += c.toLowerCase() else tok_cur_tag.attrs_a[0][0] += c return null @@ -3543,7 +3744,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -3564,6 +3765,7 @@ parse_html = (txt, parse_error_cb = null) -> # Anything else tok_cur_tag.attrs_a.unshift [c, ''] tok_state = tok_state_attribute_name + return # 8.2.4.37 http://www.w3.org/TR/html5/syntax.html#before-attribute-value-state tok_state_before_attribute_value = -> @@ -3676,7 +3878,7 @@ parse_html = (txt, parse_error_cb = null) -> tok_state_self_closing_start_tag = -> c = txt.charAt(cur++) if c is '>' - tok_cur_tag.flag 'self-closing' + tok_cur_tag.flag 'self-closing', true tok_state = tok_state_data return tok_cur_tag if c is '' @@ -3700,7 +3902,7 @@ parse_html = (txt, parse_error_cb = null) -> else val = txt.substr cur, (next_gt - cur) cur = next_gt + 1 - val = val.replace "\u0000", "\ufffd" + val = val.replace(new RegExp("\u0000", 'g'), "\ufffd") tok_cur_tag.text += val tok_state = tok_state_data return tok_cur_tag @@ -3723,7 +3925,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -3734,6 +3936,7 @@ parse_html = (txt, parse_error_cb = null) -> tok_state = tok_state_comment_start_dash when "\u0000" parse_error() + tok_state = tok_state_comment return new_character_token "\ufffd" when '>' parse_error() @@ -3746,6 +3949,7 @@ parse_html = (txt, parse_error_cb = null) -> 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 @@ -3969,12 +4173,12 @@ parse_html = (txt, parse_error_cb = null) -> return if c is '"' parse_error() - tok_cur_tag.public_identifier = '' # FIXME should this go in @attrs or @text? + tok_cur_tag.public_identifier = '' tok_state = tok_state_doctype_public_identifier_double_quoted return if c is "'" parse_error() - tok_cur_tag.public_identifier = '' # FIXME should this go in @attrs or @text? + tok_cur_tag.public_identifier = '' tok_state = tok_state_doctype_public_identifier_single_quoted return if c is '>' @@ -4001,12 +4205,12 @@ parse_html = (txt, parse_error_cb = null) -> return if c is '"' parse_error() - tok_cur_tag.public_identifier = '' # FIXME should this go in @attrs or @text? + tok_cur_tag.public_identifier = '' tok_state = tok_state_doctype_public_identifier_double_quoted return if c is "'" parse_error() - tok_cur_tag.public_identifier = '' # FIXME should this go in @attrs or @text? + tok_cur_tag.public_identifier = '' tok_state = tok_state_doctype_public_identifier_single_quoted return if c is '>' @@ -4294,8 +4498,10 @@ parse_html = (txt, parse_error_cb = null) -> else val = txt.substr cur, (next_gt - cur) cur = next_gt + 3 - val = val.replace "\u0000", "\ufffd" # fixfull spec doesn't say this - return new_character_token val # fixfull split + val = val.replace(new RegExp("\u0000", 'g'), "\ufffd") + if val.length > 0 + return new_character_token val # fixfull split + 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 @@ -4314,26 +4520,39 @@ parse_html = (txt, parse_error_cb = null) -> if cur + 1 >= txt.length return '&' if txt.charAt(cur + 1).toLowerCase() is 'x' - prefix = '#x' + base = 16 charset = hex_chars start = cur + 2 else charset = digits start = cur + 1 - prefix = '#' + base = 10 i = 0 while start + i < txt.length and charset.indexOf(txt.charAt(start + i)) > -1 i += 1 if i is 0 return '&' + cur = start + i if txt.charAt(start + i) is ';' - i += 1 - # FIXME This is supposed to generate parse errors for some chars - decoded = decode_named_char_ref(prefix + txt.substr(start, i).toLowerCase()) - if decoded? - cur = start + i - return decoded - return '&' + cur += 1 + else + parse_error() + code_point = txt.substr(start, i) + while code_point.charAt(0) is '0' and code_point.length > 1 + code_point = code_point.substr 1 + code_point = parseInt(code_point, base) + if unicode_fixes[code_point]? + parse_error() + return unicode_fixes[code_point] + else + if (code_point >= 0xd800 and code_point <= 0xdfff) or code_point > 0x10ffff + parse_error() + return "\ufffd" + else + if (code_point >= 0x0001 and code_point <= 0x0008) or (code_point >= 0x000D and code_point <= 0x001F) or (code_point >= 0x007F and code_point <= 0x009F) or (code_point >= 0xFDD0 and code_point <= 0xFDEF) or code_point is 0x000B or code_point is 0xFFFE or code_point is 0xFFFF or code_point is 0x1FFFE or code_point is 0x1FFFF or code_point is 0x2FFFE or code_point is 0x2FFFF or code_point is 0x3FFFE or code_point is 0x3FFFF or code_point is 0x4FFFE or code_point is 0x4FFFF or code_point is 0x5FFFE or code_point is 0x5FFFF or code_point is 0x6FFFE or code_point is 0x6FFFF or code_point is 0x7FFFE or code_point is 0x7FFFF or code_point is 0x8FFFE or code_point is 0x8FFFF or code_point is 0x9FFFE or code_point is 0x9FFFF or code_point is 0xAFFFE or code_point is 0xAFFFF or code_point is 0xBFFFE or code_point is 0xBFFFF or code_point is 0xCFFFE or code_point is 0xCFFFF or code_point is 0xDFFFE or code_point is 0xDFFFF or code_point is 0xEFFFE or code_point is 0xEFFFF or code_point is 0xFFFFE or code_point is 0xFFFFF or code_point is 0x10FFFE or code_point is 0x10FFFF + parse_error() + return from_code_point code_point + return else for i in [0...31] if alnum.indexOf(txt.charAt(cur + i)) is -1 @@ -4342,46 +4561,66 @@ parse_html = (txt, parse_error_cb = null) -> # exit early, because parse_error() below needs at least one alnum return '&' if txt.charAt(cur + i) is ';' - i += 1 # include ';' terminator in value decoded = decode_named_char_ref txt.substr(cur, i) + i += 1 # scan past the ';' (after, so we dno't pass it to decode) if decoded? cur += i return decoded - parse_error() - return '&' - else - # no ';' terminator (only legacy char refs) - max = i - for i in [2..max] # no prefix matches, so ok to check shortest first - c = legacy_char_refs[txt.substr(cur, i)] - if c? - if in_attr - if txt.charAt(cur + i) is '=' - # "because some legacy user agents will - # misinterpret the markup in those cases" - parse_error() - return '&' - if alnum.indexOf(txt.charAt(cur + i)) > -1 - # this makes attributes forgiving about url args - return '&' - # ok, and besides the weird exceptions for attributes... - # return the matching char - cur += i # consume entity chars - parse_error() # because no terminating ";" - return c - parse_error() - return '&' + # else FALL THROUGH (check for match without last char(s) or ";") + # no ';' terminator (only legacy char refs) + max = i + for i in [2..max] # no prefix matches, so ok to check shortest first + c = legacy_char_refs[txt.substr(cur, i)] + if c? + if in_attr + if txt.charAt(cur + i) is '=' + # "because some legacy user agents will + # misinterpret the markup in those cases" + parse_error() + return '&' + if alnum.indexOf(txt.charAt(cur + i)) > -1 + # this makes attributes forgiving about url args + return '&' + # ok, and besides the weird exceptions for attributes... + # return the matching char + cur += i # consume entity chars + parse_error() # because no terminating ";" + return c + parse_error() + 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 - doc = new Node TYPE_TAG, name: 'html', namespace: NS_HTML + txt = args_html + cur = 0 + doc = new Node TYPE_TAG, name: 'document', namespace: NS_HTML + doc.flag 'quirks mode', QUIRKS_NO # TODO bugreport spec for not specifying this + fragment_root = null # fragment parsing algorithm returns children of this open_els = [] afe = [] # active formatting elements template_ins_modes = [] ins_mode = ins_mode_initial original_ins_mode = ins_mode # TODO check spec - flag_scripting = true # TODO might need an extra flag to get