JasonWoof Got questions, comments, patches, etc.? Contact Jason Woofenden
finish ins_mode_in_body and missing tok_state
authorJason Woofenden <jason@jasonwoof.com>
Wed, 23 Dec 2015 17:05:17 +0000 (12:05 -0500)
committerJason Woofenden <jason@jasonwoof.com>
Wed, 23 Dec 2015 17:05:17 +0000 (12:05 -0500)
parse-html.coffee

index 8fdba4f..a6894cd 100644 (file)
@@ -92,6 +92,7 @@ class Node
                @namespace = args.namespace ? NS_HTML
                @parent = args.parent ? null
                @token = args.token ? null
+               @flags = args.flags ? {}
                if args.id?
                        @id = "#{args.id}+"
                else
@@ -101,8 +102,11 @@ class Node
                        @token.flag 'did_self_close'
                else
                        @flag 'did_self_close', true
-       flag: ->
-               # fixfull
+       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
@@ -320,6 +324,7 @@ 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
@@ -328,6 +333,7 @@ foster_parenting_targets = {
        tr: true
 }
 
+# FIXME namespacify
 # all html I presume
 end_tag_implied = {
        dd: true
@@ -349,6 +355,84 @@ adp_els = { address: NS_HTML, div: NS_HTML, p: NS_HTML }
 el_is_special_not_adp = (el) ->
        return special_elements[el.name] is el.namespace and adp_els[el.name] isnt el.namespace
 
+svg_attribute_fixes = {
+       attributename: 'attributeName'
+       attributetype: 'attributeType'
+       basefrequency: 'baseFrequency'
+       baseprofile: 'baseProfile'
+       calcmode: 'calcMode'
+       clippathunits: 'clipPathUnits'
+       contentscripttype: 'contentScriptType'
+       contentstyletype: 'contentStyleType'
+       diffuseconstant: 'diffuseConstant'
+       edgemode: 'edgeMode'
+       externalresourcesrequired: 'externalResourcesRequired'
+       filterres: 'filterRes'
+       filterunits: 'filterUnits'
+       glyphref: 'glyphRef'
+       gradienttransform: 'gradientTransform'
+       gradientunits: 'gradientUnits'
+       kernelmatrix: 'kernelMatrix'
+       kernelunitlength: 'kernelUnitLength'
+       keypoints: 'keyPoints'
+       keysplines: 'keySplines'
+       keytimes: 'keyTimes'
+       lengthadjust: 'lengthAdjust'
+       limitingconeangle: 'limitingConeAngle'
+       markerheight: 'markerHeight'
+       markerunits: 'markerUnits'
+       markerwidth: 'markerWidth'
+       maskcontentunits: 'maskContentUnits'
+       maskunits: 'maskUnits'
+       numoctaves: 'numOctaves'
+       pathlength: 'pathLength'
+       patterncontentunits: 'patternContentUnits'
+       patterntransform: 'patternTransform'
+       patternunits: 'patternUnits'
+       pointsatx: 'pointsAtX'
+       pointsaty: 'pointsAtY'
+       pointsatz: 'pointsAtZ'
+       preservealpha: 'preserveAlpha'
+       preserveaspectratio: 'preserveAspectRatio'
+       primitiveunits: 'primitiveUnits'
+       refx: 'refX'
+       refy: 'refY'
+       repeatcount: 'repeatCount'
+       repeatdur: 'repeatDur'
+       requiredextensions: 'requiredExtensions'
+       requiredfeatures: 'requiredFeatures'
+       specularconstant: 'specularConstant'
+       specularexponent: 'specularExponent'
+       spreadmethod: 'spreadMethod'
+       startoffset: 'startOffset'
+       stddeviation: 'stdDeviation'
+       stitchtiles: 'stitchTiles'
+       surfacescale: 'surfaceScale'
+       systemlanguage: 'systemLanguage'
+       tablevalues: 'tableValues'
+       targetx: 'targetX'
+       targety: 'targetY'
+       textlength: 'textLength'
+       viewbox: 'viewBox'
+       viewtarget: 'viewTarget'
+       xchannelselector: 'xChannelSelector'
+       ychannelselector: 'yChannelSelector'
+       zoomandpan: 'zoomAndPan'
+}
+adjust_mathml_attributes = (t) ->
+       for a in t.attrs_a
+               if a[0] is 'definitionurl'
+                       a[0] = 'definitionURL'
+       return
+adjust_svg_attributes = (t) ->
+       for a in t.attrs_a
+               if svg_attribute_fixes[a[0]]?
+                       a[0] = svg_attribute_fixes[a[0]]
+       return
+adjust_foreign_attributes = (t) ->
+       # fixfull
+       return
+
 # decode_named_char_ref()
 #
 # The list of named character references is _huge_ so ask the browser to decode
@@ -1120,7 +1204,8 @@ parse_html = (txt, parse_error_cb = null) ->
                open_els.unshift el
                return el
        # http://www.w3.org/TR/html5/syntax.html#insert-an-html-element
-       insert_html_element = insert_foreign_element # (token, namespace) ->
+       insert_html_element = (token) ->
+               insert_foreign_element token, NS_HTML
 
        # http://www.w3.org/TR/html5/syntax.html#insert-a-comment
        # position should be [node, index_within_children]
@@ -1862,15 +1947,70 @@ parse_html = (txt, parse_error_cb = null) ->
                        flag_frameset_ok = false
                        parse_generic_raw_text t
                        return
-
-               # FIXME CONTINUE
-
+               if t.type is TYPE_START_TAG and t.name is 'iframe'
+                       flag_frameset_ok = false
+                       parse_generic_raw_text t
+                       return
+               if t.type is TYPE_START_TAG and (t.name is 'noembed' or (t.name is 'noscript' and flag_scripting))
+                       parse_generic_raw_text t
+                       return
+               if t.type is TYPE_START_TAG and t.name is 'select'
+                       reconstruct_afe()
+                       insert_html_element t
+                       flag_frameset_ok = false
+                       if ins_mode is ins_mode_in_table or ins_mode is ins_mode_in_caption or ins_mode is ins_mode_in_table_body or ins_mode is ins_mode_in_row or ins_mode is ins_mode_in_cell
+                               ins_mode = ins_mode_in_select_in_table
+                       else
+                               ins_mode = ins_mode_in_select
+                       return
+               if t.type is TYPE_START_TAG and (t.name is 'optgroup' or t.name is 'option')
+                       if open_els[0].name is 'option' and open_els[0].namespace is NS_HTML
+                               open_els.shift()
+                       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')
+                       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
+               if t.type is TYPE_START_TAG and t.name is 'math'
+                       reconstruct_afe()
+                       adjust_mathml_attributes t
+                       adjust_foreign_attributes t
+                       insert_foreign_element t, NS_MATHML
+                       if t.flag 'self-closing'
+                               open_els.shift()
+                               t.acknowledge_self_closing()
+                       return
+               if t.type is TYPE_START_TAG and t.name is 'svg'
+                       reconstruct_afe()
+                       adjust_svg_attributes t
+                       adjust_foreign_attributes t
+                       insert_foreign_element t, NS_SVG
+                       if t.flag 'self-closing'
+                               open_els.shift()
+                               t.acknowledge_self_closing()
+                       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 'frame' or t.name is 'head' 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
                if t.type is TYPE_START_TAG # any other start tag
                        reconstruct_afe()
                        insert_html_element t
                        return
                if t.type is TYPE_END_TAG # any other end tag
                        in_body_any_other_end_tag t.name
+                       return
                return
 
        ins_mode_in_table_else = (t) ->
@@ -3981,6 +4121,18 @@ parse_html = (txt, parse_error_cb = null) ->
                # Anything else
                return null
 
+       # 8.2.4.68 http://www.w3.org/TR/html5/syntax.html#cdata-section-state
+       tok_state_cdata_section = ->
+               tok_state = tok_state_data
+               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 + 3
+               val = val.replace "\u0000", "\ufffd" # fixfull spec doesn't say this
+               return new_character_token val # fixfull split
 
        # 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