anolis2bikeshed.rb for URL Standard

anolis2bikeshed.rb

src = Dir.chdir File.dirname(__FILE__) do
  system 'git checkout url.src.html'
  File.read('url.src.html')
end

# replace boilerplate with metadata
src.sub! /\A.*?\n<h2/m, <<-EOF + "\n<h2"
<pre class="metadata">
Title: URL Standard
Group: WHATWG
H1: URL
Shortname: url
Status: LS
No Editor: true
Abstract: This specification provides APIs for creating, parsing, and working with universal identifiers.
Logo: https://resources.whatwg.org/logo-url.svg
!Version History: <a href="https://github.com/whatwg/url/commits">https://github.com/whatwg/url/commits</a> <a href="https://twitter.com/urlstandard">@urlstandard</a>
!Participate: <a href="https://www.w3.org/Bugs/Public/enter_bug.cgi?product=WHATWG&amp;component=URL">file a bug</a> (<a href="https://www.w3.org/Bugs/Public/buglist.cgi?product=WHATWG&amp;component=URL&amp;resolution=---">open bugs</a>)
!Participate: <a href="https://whatwg.org/mailing-list">[email protected]</a> (<a href="https://whatwg.org/mailing-list#specs">archives</a>)
!Participate: <a href="http://wiki.whatwg.org/wiki/IRC">IRC: #whatwg on Freenode</a>
</pre>
EOF

# remove TOC
src.sub! /^<h2[^\n]*>Table of Contents<\/h2>.*?<h2/m, '<h2'

# remove Conformance
src.sub! /^<h2[^\n]*>Conformance<\/h2>.*?<h2/m, '<h2'

# remove References
src.sub! /^<h2[^\n]*>References<\/h2>.*?<h2/m, '<h2'

# add ids to headers
src.gsub! /<h\d.*?>(.*?)<\/h\d>/ do |head|
  id = $1.gsub(/<.*?>/, '').gsub(' ', '-').downcase
  id = '"' + id + '"' unless id =~ /^[-\w]+$/
  head[/<h\d()/, 1] = " id=#{id}" unless head =~ /<h\d id=/
  head
end

# change document references to bikeshed format
externs = {
  "URI"      => "RFC3986",
  "IRI"      => "RFC3987",
  "ORIGIN"   => "RFC6454",
  "IPV6"     => "RFC4291",
  "IPV6TEXT" => "RFC5952"
}

src.gsub! /<span data-anolis-ref.*?>(.*?)<\/span>/ do |ref|
  name = externs[$1] || $1
  "[[#{'!' unless ref.include? 'informative'}#{name}]]"
end

# change links from <span> to <a>
src.gsub! /<span(.*?>.*?)<\/span>/ do |link|
  if link.include? 'XXX'
    link
  else
    "<a#{$1}</a>"
  end
end

# zap var title
src.gsub! /<var title>/, '<var>'

# interpret <code> links
src.gsub! /<code(.*?)>(.*?)<\/code>/ do |link|
  term = $2
  if link.include? '<code title>'
    link.sub '<code title>', '<code>'
  elsif link.include? 'title=dom-URL-'
    if $2 == 'href'
      "<a stringifier for=URLUtils><code>#{$2}</code></a>"
    else
      "<a attribute for=URLUtils><code>#{$2}</code></a>"
    end
  elsif 
    %w(
      url
      urlsearchparams
      urlutils
      urlutilsreadonly
    ).include? $2.downcase
  then
    "<a interface><code#{$1}>#{$2}</code></a>"
  elsif %w(file:/path/ / messageevent).include? $2.downcase
    "<code#{$1}>#{$2}</code>"
  else
    "<a#{$1}><code>#{$2}</code></a>"
  end
end

# eliminate some self-referential links
src.gsub! /(<dfn.*?>)<a>(.*?)<\/a>(<\/dfn>)/, '\1\2\3'

# change data-anolis-spec attributes to hrefs
externs = {
# "encoding" => "https://encoding.spec.whatwg.org/#",
# "idna"     => "http://www.unicode.org/reports/tr46/#",
# "fileapi"  => "http://dev.w3.org/2006/webapi/FileAPI/#",
# "html"     => "https://html.spec.whatwg.org/multipage/#",
# "dom"      => "https://dom.spec.whatwg.org/#",
# "webidl"   => "http://heycam.github.io/webidl/#dfn-"
}

src.gsub! /<a data-anolis-spec=(\w+)(.*?)>(.*?)<\/a>/ do |link|
  spec, remainder, text = $1, $2, $3
  title = link[/title="(.*?)"/,1] || link[/title=([-\w]+)/,1] || text
  if externs.include? spec
    "<a href=#{externs[spec]}#{title.gsub(' ', '-')}#{remainder}>#{text}</a>"
  else
    "<a#{remainder}>#{text}</a>"
  end
end

# DISABLED
# change unadorned <i> elements to use <var> instead
src.gsub /<i>(.*?)<\/i>/ do
  "<var>#{$1}</var>"
end

# add parens to function <dfn>s
src.gsub! /<dfn title=(.*?)>(<a>)?<code>\w+\(.*?\<\/dfn>/ do |fn|
  fn.sub $1, '"' + $1 + '()"'
end

# fix for html5lib "DataLossWarning: Coercing non-XML name"
src.sub! ">><", ">&gt;<"

# fix application/x-www-form-urlencoded id collision TODO
src.gsub! /<(dfn|code).*>application\/x-www-form-urlencoded<\S*\/\1>/ do |axwfu|
  if axwfu =~ /<dfn/
    axwfu[/<dfn()/, 1] = ' id=application-x-www-form-urlencoded-0'
  elsif axwfu !~ /code title/
    axwfu[/<code()/, 1] = ' title=application-x-www-form-urlencoded-0'
  end

  axwfu
end

# idl
src.gsub! /<pre class=idl>.*?<\/pre?>/m do |idl|
  "<pre class=idl>#{idl.gsub(/<.*?>/, '')}</pre>"
end

%w(URL URLSearchParams).each do |constructor|
  src.gsub! "<dfn title=\"dom-#{constructor}()\">",
    "<dfn constructor dfn-for=#{constructor}>"
end

%w(domainToASCII domainToUnicode).each do |method|
  src.gsub! "dfn title=\"dom-URL-#{method}()\"", 'dfn method dfn-for=URL'
end

URLUTILS_ATTRS = %w(
  href origin protocol username password hostname host port pathname
  searchParams search hash
)

URLUTILSREADONLY_ATTRS = %w(
  href origin protocol hostname host port pathname search hash
)

# TODO: Constructor, Stringifier
URLUTILS_ATTRS.each do |attr|
  type = (attr=='href' ? 'stringifier' : 'attribute')
  if URLUTILSREADONLY_ATTRS.include? attr
    src.gsub!(/(<dfn title=dom-URL-#{attr}.*?<\/dfn>)/, 
      '<dfn ' + type + ' dfn-for=URLUtilsReadOnly>\1</dfn>')
  end

  src.gsub!("dfn title=dom-URL-#{attr}", "dfn #{type} dfn-for=URLUtils")
  src.gsub!("dom-URL-#{attr}", "dom-URLUtils-#{attr}")
end

src.gsub! /<dfn title="dom-URLSearchParams-\w+\(\)">/,
  "<dfn method dfn-for=URLSearchParams>"

src.gsub! '<dfn title=dom-URLSearchParams-stringifier>',
  '<dfn stringifier dfn-for=URLSearchParams>'

# swap a and code
src.gsub! /(<a.*?>)(<code.*?>)(.*?)(<\/code>)(<\/a>)/, '\2\1\3\5\4'

# DISABLED
# comment out problematic line for now
# src[/().* implements .*/, 1] = '// '

src += "\n" + <<-EOF
<pre class="biblio"
{
    "IDNA": {
        "authors": [
            "Mark Davis",
            "Michel Suignard"
        ],
        "href": "http://www.unicode.org/reports/tr46/",
        "title": "Unicode IDNA Compatibility Processing",
        "publisher": "Unicode Consortium"
    },
    "DOM": {
        "authors": [
            "Anne van Kesteren",
            "Aryeh Gregor",
            "Ms2ger"
        ],
        "href": "https://dom.spec.whatwg.org/",
        "title": "DOM",
        "publisher": "WHATWG"
    },
    "ENCODING": {
        "authors": [
            "Anne van Kesteren"
        ],
        "href": "https://encoding.spec.whatwg.org/",
        "title": "Encoding",
        "publisher": "WHATWG"
    }
}
</pre>

<pre class="anchors">
[
    {
        "linkingText": "ascii whitespace",
        "url": "https://encoding.spec.whatwg.org/#ascii-whitespace",
        "shortname": "encoding", "type": "dfn", "level": 2, "status": "current"
    },
    {
        "linkingText": "blob",
        "url": "http://dev.w3.org/2006/webapi/FileAPI/#blob",
        "shortname": "fileapi", "type": "dfn", "level": 2, "status": "current"
    },
    {
        "linkingText": "blob url store",
        "url": "http://dev.w3.org/2006/webapi/FileAPI/#BlobURLStore",
        "shortname": "fileapi", "type": "dfn", "level": 2, "status": "current"
    },
    {
        "linkingText": "concept-attribute-value",
        "url": "https://dom.spec.whatwg.org/#concept-attribute-value",
        "shortname": "webidl", "type": "dfn", "level": 2, "status": "current"
    },
    {
        "linkingText": "concept-document",
        "url": "https://dom.spec.whatwg.org/#concept-document",
        "shortname": "dom", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "concept-encoding-get",
        "url": "https://encoding.spec.whatwg.org/#concept-encoding-get",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "context object",
        "url": "https://dom.spec.whatwg.org/#context-object",
        "shortname": "dom", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "decoder",
        "url": "https://encoding.spec.whatwg.org/#decoder",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "dom-node-baseuri",
        "url": "https://dom.spec.whatwg.org/#dom-node-baseuri",
        "shortname": "dom", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "encoding",
        "url": "https://encoding.spec.whatwg.org/#encoding",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "error",
        "url": "https://encoding.spec.whatwg.org/#error",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "eventsource",
        "url": "https://html.spec.whatwg.org/multipage/comms.html#eventsource",
        "shortname": "html", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "hashchangeevent",
        "url": "https://html.spec.whatwg.org/multipage/comms.html#hashchangeevent",
        "shortname": "html", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "structured clone",
        "url": "https://html.spec.whatwg.org/multipage/infrastructure.html#structured-clone",
        "shortname": "html", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "throw",
        "url": "http://heycam.github.io/webidl/#dfn-throw",
        "shortname": "webidl", "type": "dfn", "level": 2, "status": "current"
    },
    {
        "linkingText": "toascii",
        "url": "http://www.unicode.org/reports/tr46/#ToASCII",
        "shortname": "idna", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "tounicode",
        "url": "http://www.unicode.org/reports/tr46/#ToUnicode",
        "shortname": "idna", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "utf-8",
        "url": "https://encoding.spec.whatwg.org/#utf-8",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "utf-8 decode without bom",
        "url": "https://encoding.spec.whatwg.org/#utf-8-decode-without-bom",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "utf-8 encode",
        "url": "https://encoding.spec.whatwg.org/#utf-8-encode",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "utf-8 decoder",
        "url": "https://encoding.spec.whatwg.org/#utf-8-decoder",
        "shortname": "encoding", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "unicode serialization of an origin",
        "url": "https://html.spec.whatwg.org/multipage/browsers.html#unicode-serialisation-of-an-origin",
        "shortname": "html", "type": "dfn", "level": 1, "status": "current"
    },
    {
        "linkingText": "value pairs to iterate over",
        "url": "http://heycam.github.io/webidl/#dfn-value-pairs-to-iterate-over",
        "shortname": "webidl", "type": "dfn", "level": 2, "status": "current"
    }
]
</pre>
EOF
Dir.chdir File.expand_path('../../url-bikeshed', __FILE__) do
  File.open('url.bs', 'w') {|file| file.write src}
end