@namespace = args.namespace ? NS_HTML
@parent = args.parent ? null
@token = args.token ? null
+ @flags = args.flags ? {}
if args.id?
@id = "#{args.id}+"
else
@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
u: true
}
+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
+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
+ return false
+ if el.namespace is NS_SVG
+ if el.name is 'foreignObject' or el.name is 'desc' or el.name is 'title'
+ return true
+ return false
+
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
tr: true
}
+# FIXME namespacify
# all html I presume
end_tag_implied = {
dd: true
el_is_special_not_adp = (el) ->
return special_elements[el.name] is el.namespace and adp_els[el.name] isnt el.namespace
+svg_name_fixes = {
+ altglyph: 'altGlyph'
+ altglyphdef: 'altGlyphDef'
+ altglyphitem: 'altGlyphItem'
+ animatecolor: 'animateColor'
+ animatemotion: 'animateMotion'
+ animatetransform: 'animateTransform'
+ clippath: 'clipPath'
+ feblend: 'feBlend'
+ fecolormatrix: 'feColorMatrix'
+ fecomponenttransfer: 'feComponentTransfer'
+ fecomposite: 'feComposite'
+ feconvolvematrix: 'feConvolveMatrix'
+ fediffuselighting: 'feDiffuseLighting'
+ fedisplacementmap: 'feDisplacementMap'
+ fedistantlight: 'feDistantLight'
+ fedropshadow: 'feDropShadow'
+ feflood: 'feFlood'
+ fefunca: 'feFuncA'
+ fefuncb: 'feFuncB'
+ fefuncg: 'feFuncG'
+ fefuncr: 'feFuncR'
+ fegaussianblur: 'feGaussianBlur'
+ feimage: 'feImage'
+ femerge: 'feMerge'
+ femergenode: 'feMergeNode'
+ femorphology: 'feMorphology'
+ feoffset: 'feOffset'
+ fepointlight: 'fePointLight'
+ fespecularlighting: 'feSpecularLighting'
+ fespotlight: 'feSpotLight'
+ fetile: 'feTile'
+ feturbulence: 'feTurbulence'
+ foreignobject: 'foreignObject'
+ glyphref: 'glyphRef'
+ lineargradient: 'linearGradient'
+ radialgradient: 'radialGradient'
+ textpath: 'textPath'
+}
+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
return null if decoded is txt
return g_dncr.cache[txt] = decoded
-parse_html = (txt, parse_error_cb = null) ->
- cur = 0 # index of next char in txt to be parsed
+parse_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
flag_parsing = false
parse_error = ->
- if parse_error_cb?
- parse_error_cb cur
+ if args.error_cb?
+ args.error_cb cur
else
console.log "Parse error at character #{cur} of #{txt.length}"
return
dest[0].children.splice dest[1], 0, t
+
+ # 8.2.5 http://www.w3.org/TR/html5/syntax.html#tree-construction
+ process_token = (t) ->
+ acn = adjusted_current_node()
+ unless acn?
+ ins_mode t
+ return
+ if acn.namespace is NS_HTML
+ 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')
+ ins_mode t
+ return
+ if t.type is TYPE_TEXT
+ ins_mode t
+ return
+ if acn.namespace is NS_MATHML and acn.name is 'annotation-xml' and t.type is TYPE_START_TAG and t.name is 'svg'
+ ins_mode t
+ return
+ if is_html_integration acn
+ if t.type is TYPE_START_TAG or t.type is TYPE_TEXT
+ ins_mode t
+ return
+ if t.type is TYPE_EOF
+ ins_mode t
+ return
+ in_foreign_content t
+ return
+
# 8.2.5.1
# http://www.w3.org/TR/html5/syntax.html#creating-and-inserting-nodes
# http://www.w3.org/TR/html5/syntax.html#appropriate-place-for-inserting-a-node
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
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
# 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.
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]
# Anything else
#fixfull (iframe, quirks)
ins_mode = ins_mode_before_html
- ins_mode t # reprocess the token
+ process_token t
return
# 8.2.5.4.2 http://www.w3.org/TR/html5/syntax.html#the-before-html-insertion-mode
open_els.unshift el
# ?fixfull browsing context
ins_mode = ins_mode_before_head
- ins_mode t
+ process_token t
return
# 8.2.5.4.3 http://www.w3.org/TR/html5/syntax.html#the-before-head-insertion-mode
el = insert_html_element head_tok
head_element_pointer = el
ins_mode = ins_mode_in_head
- ins_mode t # reprocess current token
+ process_token t
# 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
- ins_mode t
+ process_token t
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
parse_error()
open_els.shift()
ins_mode = ins_mode_in_head
- ins_mode t
+ process_token t
ins_mode_in_head_noscript = (t) ->
if t.type is TYPE_DOCTYPE
parse_error()
body_tok = new_open_tag 'body'
insert_html_element body_tok
ins_mode = ins_mode_in_body
- ins_mode t # reprocess token
+ process_token t
return
ins_mode_after_head = (t) ->
if is_space_tok t
parse_error()
break
ins_mode = ins_mode_after_body
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG and (t.name is 'address' or t.name is 'article' or t.name is 'aside' or t.name is 'blockquote' or t.name is 'center' or t.name is 'details' or t.name is 'dialog' or t.name is 'dir' or t.name is 'div' or t.name is 'dl' or t.name is 'fieldset' or t.name is 'figcaption' or t.name is 'figure' or t.name is 'footer' or t.name is 'header' or t.name is 'hgroup' or t.name is 'main' or t.name is 'nav' or t.name is 'ol' or t.name is 'p' or t.name is 'section' or t.name is 'summary' or t.name is 'ul')
close_p_if_in_button_scope()
if t.type is TYPE_START_TAG and t.name is 'image'
parse_error()
t.name = 'img'
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG and t.name is 'isindex'
parse_error()
input_el.attrs_a.push ['name', 'isindex']
# fixfull this next bit is in english... internationalize?
prompt ?= "This is a searchable index. Enter search keywords: "
- insert_character prompt # fixfull split
+ insert_character new_character_token prompt # fixfull split
# TODO submit typo "balue" in spec
insert_html_element input_el
open_els.shift()
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) ->
- 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
open_els[0].flag 'already started', true
open_els.shift()
ins_mode = original_ins_mode
- ins_mode t
+ process_token t
return
if t.type is TYPE_END_TAG and t.name is 'script'
open_els.shift()
# 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
+ can_in_table = { # FIXME do this inline like everywhere else
+ 'table': true
+ 'tbody': true
+ 'tfoot': true
+ 'thead': true
+ 'tr': true
+ }
ins_mode_in_table = (t) ->
switch t.type
when TYPE_TEXT
if can_in_table[t.name]
original_ins_mode = ins_mode
ins_mode = ins_mode_in_table_text
- ins_mode t
+ process_token t
else
ins_mode_in_table_else t
when TYPE_COMMENT
clear_stack_to_table_context()
insert_html_element new_open_tag 'colgroup'
ins_mode = ins_mode_in_column_group
- ins_mode t
+ process_token t
when 'tbody', 'tfoot', 'thead'
clear_stack_to_table_context()
insert_html_element t
clear_stack_to_table_context()
insert_html_element new_open_tag 'tbody'
ins_mode = ins_mode_in_table_body
- ins_mode t
+ process_token t
when 'table'
parse_error()
if is_in_table_scope 'table'
if el.name is 'table'
break
reset_ins_mode()
- ins_mode t
+ 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()
ins_mode_table_else old
pending_table_character_tokens = [] # FIXME test (spec doesn't say this)
ins_mode = original_ins_mode
- ins_mode t
+ process_token t
# 8.2.5.4.11 http://www.w3.org/TR/html5/syntax.html#parsing-main-incaption
ins_mode_in_caption = (t) ->
break
clear_afe_to_marker()
ins_mode = ins_mode_in_table
- ins_mode t
+ process_token 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')
return
open_els.shift()
ins_mode = ins_mode_in_table
- ins_mode t
+ process_token t
return
# 8.2.5.4.13 http://www.w3.org/TR/html5/syntax.html#parsing-main-intbody
clear_stack_to_table_body_context()
insert_html_element new_open_tag 'tr'
ins_mode = ins_mode_in_row
- ins_mode t
+ 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
clear_stack_to_table_body_context()
open_els.shift()
ins_mode = ins_mode_in_table
- ins_mode t
+ process_token t
return
if t.type is TYPE_END_TAG and (t.name is 'body' or t.name is 'caption' or t.name is 'col' or t.name is 'colgroup' or t.name is 'html' or t.name is 'td' or t.name is 'th' or t.name is 'tr')
parse_error()
clear_stack_to_table_row_context()
open_els.shift()
ins_mode = ins_mode_in_table_body
- ins_mode t
+ process_token t
else
parse_error()
return
clear_stack_to_table_row_context()
open_els.shift()
ins_mode = ins_mode_in_table_body
- ins_mode t
+ process_token t
else
parse_error()
return
parse_error()
return
close_the_cell()
- ins_mode t
+ process_token t
return
if t.type is TYPE_END_TAG and (t.name is 'body' or t.name is 'caption' or t.name is 'col' or t.name is 'colgroup' or t.name is 'html')
parse_error()
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
close_the_cell()
- ins_mode t
+ process_token t
else
parse_error()
return
if el.name is 'select'
break
reset_ins_mode()
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG and (t.name is 'script' or t.name is 'template')
ins_mode_in_head t
if el.name is 'select'
break
reset_ins_mode()
- ins_mode t
+ process_token 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()
if el.name is 'select'
break
reset_ins_mode()
- ins_mode t
+ process_token t
return
# Anything else
ins_mode_in_select t
template_ins_modes.shift()
template_ins_modes.unshift ins_mode_in_table
ins_mode = ins_mode_in_table
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG and t.name is 'col'
template_ins_modes.shift()
template_ins_modes.unshift ins_mode_in_column_group
ins_mode = ins_mode_in_column_group
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG and t.name is 'tr'
template_ins_modes.shift()
template_ins_modes.unshift ins_mode_in_table_body
ins_mode = ins_mode_in_table_body
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG and (t.name is 'td' or t.name is 'th')
template_ins_modes.shift()
template_ins_modes.unshift ins_mode_in_row
ins_mode = ins_mode_in_row
- ins_mode t
+ process_token t
return
if t.type is TYPE_START_TAG
template_ins_modes.shift()
template_ins_modes.unshift ins_mode_in_body
ins_mode = ins_mode_in_body
- ins_mode t
+ process_token t
return
if t.type is TYPE_END_TAG
parse_error()
clear_afe_to_marker()
template_ins_modes.shift()
reset_ins_mode()
- ins_mode t
+ process_token t
# 8.2.5.4.19 http://www.w3.org/TR/html5/syntax.html#parsing-main-afterbody
ins_mode_after_body = (t) ->
# Anything ELse
parse_error()
ins_mode = ins_mode_in_body
- ins_mode t
+ process_token t
# 8.2.5.4.20 http://www.w3.org/TR/html5/syntax.html#parsing-main-inframeset
ins_mode_in_frameset = (t) ->
parse_error()
return
-
-
+ # 8.2.5.5 http://www.w3.org/TR/html5/syntax.html#parsing-main-inforeign
+ has_color_face_or_size = (t) ->
+ for a in t.attrs_a
+ if a[0] is 'color' or a[0] is 'face' or a[0] is 'size'
+ return true
+ return false
+ in_foreign_content_end_script = ->
+ open_els.shift()
+ # fixfull
+ return
+ in_foreign_content_other_start = (t) ->
+ acn = adjusted_current_node()
+ if acn.namespace is NS_MATHML
+ adjust_mathml_attributes t
+ if acn.namespace is NS_SVG and svg_name_fixes[t.name]?
+ t.name = svg_name_fixes[t.name]
+ if acn.namespace is NS_SVG
+ adjust_svg_attributes t
+ adjust_foreign_attributes t
+ insert_foreign_element t, acn.namespace
+ if t.flag 'self-closing'
+ if t.name is 'script'
+ t.acknowledge_self_closing()
+ in_foreign_content_end_script()
+ else
+ open_els.shift()
+ t.acknowledge_self_closing()
+ return
+ in_foreign_content = (t) ->
+ if t.type is TYPE_TEXT and t.text is "\u0000"
+ parse_error()
+ insert_character new_character_token "\ufffd"
+ return
+ if is_space_tok t
+ insert_character t
+ return
+ if t.type is TYPE_TEXT
+ flag_frameset_ok = false
+ 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 'b' or t.name is 'big' or t.name is 'blockquote' or t.name is 'body' or t.name is 'br' or t.name is 'center' or t.name is 'code' or t.name is 'dd' or t.name is 'div' or t.name is 'dl' or t.name is 'dt' or t.name is 'em' or t.name is 'embed' or t.name is 'h1' or t.name is 'h2' or t.name is 'h3' or t.name is 'h4' or t.name is 'h5' or t.name is 'h6' or t.name is 'head' or t.name is 'hr' or t.name is 'i' or t.name is 'img' or t.name is 'li' or t.name is 'listing' or t.name is 'main' or t.name is 'meta' or t.name is 'nobr' or t.name is 'ol' or t.name is 'p' or t.name is 'pre' or t.name is 'ruby' or t.name is 's' or t.name is 'small' or t.name is 'span' or t.name is 'strong' or t.name is 'strike' or t.name is 'sub' or t.name is 'sup' or t.name is 'table' or t.name is 'tt' or t.name is 'u' or t.name is 'ul' or t.name is 'var' or (t.name is 'font' and has_color_face_or_size(t)))
+ parse_error()
+ if flag_fragment_parsing
+ in_foreign_content_other_start t
+ 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
+ break
+ process_token t
+ return
+ if t.type is TYPE_START_TAG
+ in_foreign_content_other_start t
+ return
+ if t.type is TYPE_END_TAG and t.name is 'script' and open_els[0].name is 'script' and open_els[0].namespace is NS_SVG
+ in_foreign_content_end_script()
+ return
+ if t.type is TYPE_END_TAG
+ if open_els[0].name.toLowerCase() isnt t.name
+ parse_error()
+ for node in open_els
+ if node is open_els[open_els.length - 1]
+ return
+ if node.name.toLowerCase() is t.name
+ loop
+ el = open_els.shift()
+ if el is node
+ return
+ if node.namespace is NS_HTML
+ break
+ ins_mode t # explicitly call HTML insertion mode
# 8.2.4.1 http://www.w3.org/TR/html5/syntax.html#data-state
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
+ # the various "raw" states, so it's hopefully ok to assume that
+ # open_els[0].name will work instead TODO: verify this after the script
+ # data states are implemented
debug_log "#{t.type}, #{t.name} open_els: #{serialize_els open_els, true, true}"
return t.type is TYPE_END_TAG and t.name is open_els[0].name
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
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()
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
# 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(new RegExp("\u0000", 'g'), "\ufffd") # fixfull spec doesn't say this
+ val = val.replace(new RegExp("\r\n", 'g'), "\n") # fixfull spec doesn't say this
+ val = val.replace(new RegExp("\r", 'g'), "\n") # 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
# tree constructor initialization
# see comments on TYPE_TAG/etc for the structure of this data
+ txt = args.html
+ cur = 0
doc = new Node TYPE_TAG, name: 'html', namespace: NS_HTML
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 <noscript> to parse correctly
+ flag_scripting = args.scripting ? true # TODO might need an extra flag to get <noscript> to parse correctly
flag_frameset_ok = true
flag_parsing = true
flag_foster_parenting = false
tok_state = tok_state_data
# proccess input
+ # http://www.w3.org/TR/html5/syntax.html#tree-construction
while flag_parsing
t = tok_state()
if t?
- ins_mode t
+ process_token t
# fixfull parse error if has self-closing flag, but it wasn't acknolwedged
return doc.children
module.exports.TYPE_TEXT = TYPE_TEXT
module.exports.TYPE_COMMENT = TYPE_COMMENT
module.exports.TYPE_DOCTYPE = TYPE_DOCTYPE
+module.exports.NS_HTML = NS_HTML
+module.exports.NS_MATHML = NS_MATHML
+module.exports.NS_SVG = NS_SVG
Got questions, comments, patches, etc.? Contact Jason Woofenden