Whatpm/ContentChecker/HTML.pm

package Whatpm::ContentChecker;
use strict;
require Whatpm::ContentChecker;

use Char::Class::XML qw/InXML_NCNameStartChar10 InXMLNCNameChar10/;

my $HTML_NS = q<http://www.w3.org/1999/xhtml>;

## --- Feature Status ---

sub FEATURE_HTML5_REC () {
  ## NOTE: Part of HTML5, the implemented status.
  Whatpm::ContentChecker::FEATURE_STATUS_REC |
  Whatpm::ContentChecker::FEATURE_ALLOWED

      ## Strictly speaking, HTML5's "implemented and widely deployed"
      ## status does not necessarily satisfy the condition for
      ## FEATURE_STATUS_REC, since there is no test cases for most of
      ## features marked as "implemented" in HTML5.  Nevertheless, we
      ## special-case HTML5's this status as if that had passed the CR
      ## phase, considering HTML's history.
}

sub FEATURE_HTML5_CR () {
  ## NOTE: Part of HTML5, the awaiting implementation feedback status.
  Whatpm::ContentChecker::FEATURE_STATUS_CR |
  Whatpm::ContentChecker::FEATURE_ALLOWED
}
sub FEATURE_HTML5_LC () {
  ## NOTE: Part of HTML5, the last call of comments status.
  Whatpm::ContentChecker::FEATURE_STATUS_LC |
  Whatpm::ContentChecker::FEATURE_ALLOWED
}
sub FEATURE_HTML5_AT_RISK () {
  ## NOTE: Part of HTML5, but in the being considered for removal
  ## status.
  Whatpm::ContentChecker::FEATURE_STATUS_WD |
  Whatpm::ContentChecker::FEATURE_ALLOWED
}
sub FEATURE_HTML5_WD () {
  ## NOTE: Part of HTML5, the working draft status.
  Whatpm::ContentChecker::FEATURE_STATUS_WD |
  Whatpm::ContentChecker::FEATURE_ALLOWED
}
sub FEATURE_HTML5_FD () {
  ## NOTE: Part of HTML5, the first draft status.
  Whatpm::ContentChecker::FEATURE_STATUS_WD |
  Whatpm::ContentChecker::FEATURE_ALLOWED
}
sub FEATURE_HTML5_DEFAULT () {
  ## NOTE: Part of HTML5, but not annotated.
  Whatpm::ContentChecker::FEATURE_STATUS_WD |
  Whatpm::ContentChecker::FEATURE_ALLOWED
}
sub FEATURE_HTML5_DROPPED () {
  ## NOTE: Was part of HTML5, in a status before the last call of
  ## comments, but then dropped.
  Whatpm::ContentChecker::FEATURE_STATUS_WD
}
sub FEATURE_HTML5_LC_DROPPED () {
  ## NOTE: Was part of HTML5, in the last call of comments status, but
  ## then dropped.
  Whatpm::ContentChecker::FEATURE_STATUS_LC
}

sub FEATURE_WF2X () {
  ## NOTE: Defined in WF2 (whether deprecated or not) and then
  ## incorporated into the HTML5 spec.
  Whatpm::ContentChecker::FEATURE_STATUS_LC
}
sub FEATURE_WF2 () {
  ## NOTE: Features introduced or modified in WF2, which were not
  ## merged into HTML5.
  Whatpm::ContentChecker::FEATURE_STATUS_LC
}
sub FEATURE_WF2_INFORMATIVE () {
  ## NOTE: Features mentioned in WF2's informative appendix A, which
  ## were not merged into HTML5.
  Whatpm::ContentChecker::FEATURE_STATUS_LC
}

sub FEATURE_RDFA_REC () {
  Whatpm::ContentChecker::FEATURE_STATUS_REC
}
sub FEATURE_RDFA_LC_DROPPED () {
  ## NOTE: The feature that was defined in a RDFa last call working
  ## draft, but then dropped.
  Whatpm::ContentChecker::FEATURE_STATUS_LC
}

## NOTE: XHTML Role LCWD has almost no information on how the |role|
## attribute can be used- the only requirements for that matter is:
## "the attribute MUST be referenced using its namespace-qualified form" (and
## this is a host language conformance!).
sub FEATURE_ROLE_LC () {
  Whatpm::ContentChecker::FEATURE_STATUS_LC
}

sub FEATURE_XHTML2_ED () {
  ## NOTE: XHTML 2.0 Editor's Draft, in which the namespace URI is
  ## "http://www.w3.org/1999/xhtml".
  Whatpm::ContentChecker::FEATURE_STATUS_WD
}

sub FEATURE_XHTMLBASIC11_CR () {
  ## NOTE: XHTML Basic 1.1 Recommendation, new features (not in XHTML
  ## M12N).
  Whatpm::ContentChecker::FEATURE_STATUS_REC
}
sub FEATURE_XHTMLBASIC11_CR_DEPRECATED () {
  ## NOTE: XHTML Basic 1.1 Recommendation, new but deprecated
  ## features.
  Whatpm::ContentChecker::FEATURE_STATUS_REC |
  Whatpm::ContentChecker::FEATURE_DEPRECATED_INFO
}

sub FEATURE_RUBY_REC () {
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

sub FEATURE_M12N11_LC () {
  ## NOTE: XHTML M12N 1.1 Recommendation, new features (not in 1.0).
  Whatpm::ContentChecker::FEATURE_STATUS_REC;
}

## NOTE: M12N10 status is based on its abstract module definition.
## It contains a number of problems.  (However, again, it's a REC!)
sub FEATURE_M12N10_REC () {
  ## NOTE: Oh, XHTML m12n 1.0 passed the CR phase!  W3C Process sucks!
  Whatpm::ContentChecker::FEATURE_STATUS_REC
}
sub FEATURE_M12N10_REC_DEPRECATED () {
  Whatpm::ContentChecker::FEATURE_STATUS_REC |
  Whatpm::ContentChecker::FEATURE_DEPRECATED_INFO
}

## NOTE: XHTML10 status is based on its transitional and frameset DTDs
## (second edition).  Only missing attributes from M12N10 abstract
## definition are added.
sub FEATURE_XHTML10_REC () {
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

## NOTE: Diff from HTML4.
sub FEATURE_ISOHTML_PREPARATION () { ## Informative documentation
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

## NOTE: HTML4 status is based on its transitional and frameset DTDs (HTML
## 4.01).  Only missing attributes from XHTML10 are added.
sub FEATURE_HTML4_REC_RESERVED () {
  Whatpm::ContentChecker::FEATURE_STATUS_WD
}

## TODO: According to HTML4 definition, authors SHOULD use style sheets
## rather than presentational attributes (deprecated or not deprecated).

## NOTE: Diff from HTML4.
sub FEATURE_HTML32_REC_OBSOLETE () {
  Whatpm::ContentChecker::FEATURE_STATUS_CR |
  Whatpm::ContentChecker::FEATURE_DEPRECATED_SHOULD
      ## NOTE: Lowercase normative "should".
}

sub FEATURE_RFC2659 () { ## Experimental RFC
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

## NOTE: HTML 2.x - diff from HTML 2.0 and not in newer versions.
sub FEATURE_HTML2X_RFC () { ## Proposed Standard, obsolete
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

## NOTE: Diff from HTML 2.0.
sub FEATURE_RFC1942 () { ## Experimental RFC, obsolete
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

## NOTE: Diff from HTML 3.2.
sub FEATURE_HTML20_RFC () { ## Proposed Standard, obsolete
  Whatpm::ContentChecker::FEATURE_STATUS_CR
}

## --- Content Model ---

## December 2007 HTML5 Classification

my $HTMLMetadataContent = {
  $HTML_NS => {
    title => 1, base => 1, link => 1, style => 1, script => 1, noscript => 1,
    'event-source' => 1, eventsource => 1,
    command => 1, datatemplate => 1,
    ## NOTE: A |meta| with no |name| element is not allowed as
    ## a metadata content other than |head| element.
    meta => 1,
  },
  ## NOTE: RDF is mentioned in the HTML5 spec.
  ## TODO: Other RDF elements?
  q<http://www.w3.org/1999/02/22-rdf-syntax-ns#> => {RDF => 1},
};

my $HTMLFlowContent = {
  $HTML_NS => {
    section => 1, nav => 1, article => 1, blockquote => 1, aside => 1,
    h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, header => 1,
    footer => 1, address => 1, p => 1, hr => 1, dialog => 1, pre => 1,
    ol => 1, ul => 1, dl => 1, figure => 1, map => 1, table => 1,
    form => 1, fieldset => 1,
    details => 1, ## ISSUE: "Flow element" in spec.
    datagrid => 1, ## ISSUE: "Flow element" in spec.
    data
1	wakaba	1.1	package Whatpm::ContentChecker;
2			use strict;
3			require Whatpm::ContentChecker;
4
5	wakaba	1.117	use Char::Class::XML qw/InXML_NCNameStartChar10 InXMLNCNameChar10/;
6
7	wakaba	1.1	my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
8
9	wakaba	1.174	## --- Feature Status ---
10
11	wakaba	1.187	sub FEATURE_HTML5_REC () {
12	wakaba	1.154	## NOTE: Part of HTML5, the implemented status.
13	wakaba	1.89	Whatpm::ContentChecker::FEATURE_STATUS_REC \|
14			Whatpm::ContentChecker::FEATURE_ALLOWED
15	wakaba	1.187
16			## Strictly speaking, HTML5's "implemented and widely deployed"
17			## status does not necessarily satisfy the condition for
18			## FEATURE_STATUS_REC, since there is no test cases for most of
19			## features marked as "implemented" in HTML5. Nevertheless, we
20			## special-case HTML5's this status as if that had passed the CR
21			## phase, considering HTML's history.
22	wakaba	1.89	}
23	wakaba	1.187
24	wakaba	1.154	sub FEATURE_HTML5_CR () {
25			## NOTE: Part of HTML5, the awaiting implementation feedback status.
26			Whatpm::ContentChecker::FEATURE_STATUS_CR \|
27			Whatpm::ContentChecker::FEATURE_ALLOWED
28			}
29	wakaba	1.54	sub FEATURE_HTML5_LC () {
30	wakaba	1.154	## NOTE: Part of HTML5, the last call of comments status.
31	wakaba	1.54	Whatpm::ContentChecker::FEATURE_STATUS_LC \|
32			Whatpm::ContentChecker::FEATURE_ALLOWED
33			}
34			sub FEATURE_HTML5_AT_RISK () {
35	wakaba	1.154	## NOTE: Part of HTML5, but in the being considered for removal
36			## status.
37	wakaba	1.54	Whatpm::ContentChecker::FEATURE_STATUS_WD \|
38			Whatpm::ContentChecker::FEATURE_ALLOWED
39			}
40			sub FEATURE_HTML5_WD () {
41	wakaba	1.154	## NOTE: Part of HTML5, the working draft status.
42	wakaba	1.54	Whatpm::ContentChecker::FEATURE_STATUS_WD \|
43			Whatpm::ContentChecker::FEATURE_ALLOWED
44			}
45			sub FEATURE_HTML5_FD () {
46	wakaba	1.154	## NOTE: Part of HTML5, the first draft status.
47	wakaba	1.54	Whatpm::ContentChecker::FEATURE_STATUS_WD \|
48			Whatpm::ContentChecker::FEATURE_ALLOWED
49			}
50			sub FEATURE_HTML5_DEFAULT () {
51	wakaba	1.154	## NOTE: Part of HTML5, but not annotated.
52	wakaba	1.54	Whatpm::ContentChecker::FEATURE_STATUS_WD \|
53			Whatpm::ContentChecker::FEATURE_ALLOWED
54	wakaba	1.49	}
55	wakaba	1.54	sub FEATURE_HTML5_DROPPED () {
56	wakaba	1.154	## NOTE: Was part of HTML5, in a status before the last call of
57			## comments, but then dropped.
58	wakaba	1.49	Whatpm::ContentChecker::FEATURE_STATUS_WD
59			}
60	wakaba	1.118	sub FEATURE_HTML5_LC_DROPPED () {
61	wakaba	1.154	## NOTE: Was part of HTML5, in the last call of comments status, but
62			## then dropped.
63	wakaba	1.118	Whatpm::ContentChecker::FEATURE_STATUS_LC
64			}
65	wakaba	1.154
66	wakaba	1.119	sub FEATURE_WF2X () {
67	wakaba	1.154	## NOTE: Defined in WF2 (whether deprecated or not) and then
68			## incorporated into the HTML5 spec.
69	wakaba	1.119	Whatpm::ContentChecker::FEATURE_STATUS_LC
70			}
71	wakaba	1.54	sub FEATURE_WF2 () {
72	wakaba	1.154	## NOTE: Features introduced or modified in WF2, which were not
73			## merged into HTML5.
74	wakaba	1.119	Whatpm::ContentChecker::FEATURE_STATUS_LC
75	wakaba	1.54	}
76	wakaba	1.126	sub FEATURE_WF2_INFORMATIVE () {
77	wakaba	1.154	## NOTE: Features mentioned in WF2's informative appendix A, which
78			## were not merged into HTML5.
79	wakaba	1.56	Whatpm::ContentChecker::FEATURE_STATUS_LC
80			}
81	wakaba	1.49
82	wakaba	1.154	sub FEATURE_RDFA_REC () {
83			Whatpm::ContentChecker::FEATURE_STATUS_REC
84	wakaba	1.121	}
85	wakaba	1.154	sub FEATURE_RDFA_LC_DROPPED () {
86			## NOTE: The feature that was defined in a RDFa last call working
87			## draft, but then dropped.
88	wakaba	1.61	Whatpm::ContentChecker::FEATURE_STATUS_LC
89			}
90	wakaba	1.58
91			## NOTE: XHTML Role LCWD has almost no information on how the \|role\|
92			## attribute can be used- the only requirements for that matter is:
93			## "the attribute MUST be referenced using its namespace-qualified form" (and
94			## this is a host language conformance!).
95	wakaba	1.82	sub FEATURE_ROLE_LC () {
96			Whatpm::ContentChecker::FEATURE_STATUS_LC
97			}
98
99			sub FEATURE_XHTML2_ED () {
100	wakaba	1.154	## NOTE: XHTML 2.0 Editor's Draft, in which the namespace URI is
101			## "http://www.w3.org/1999/xhtml".
102	wakaba	1.82	Whatpm::ContentChecker::FEATURE_STATUS_WD
103			}
104	wakaba	1.58
105	wakaba	1.55	sub FEATURE_XHTMLBASIC11_CR () {
106	wakaba	1.154	## NOTE: XHTML Basic 1.1 Recommendation, new features (not in XHTML
107			## M12N).
108			Whatpm::ContentChecker::FEATURE_STATUS_REC
109	wakaba	1.55	}
110			sub FEATURE_XHTMLBASIC11_CR_DEPRECATED () {
111	wakaba	1.154	## NOTE: XHTML Basic 1.1 Recommendation, new but deprecated
112			## features.
113			Whatpm::ContentChecker::FEATURE_STATUS_REC \|
114	wakaba	1.55	Whatpm::ContentChecker::FEATURE_DEPRECATED_INFO
115			}
116
117	wakaba	1.154	sub FEATURE_RUBY_REC () {
118			Whatpm::ContentChecker::FEATURE_STATUS_CR
119	wakaba	1.82	}
120
121	wakaba	1.154	sub FEATURE_M12N11_LC () {
122			## NOTE: XHTML M12N 1.1 Recommendation, new features (not in 1.0).
123			Whatpm::ContentChecker::FEATURE_STATUS_REC;
124	wakaba	1.99	}
125
126	wakaba	1.49	## NOTE: M12N10 status is based on its abstract module definition.
127			## It contains a number of problems. (However, again, it's a REC!)
128	wakaba	1.54	sub FEATURE_M12N10_REC () {
129	wakaba	1.154	## NOTE: Oh, XHTML m12n 1.0 passed the CR phase! W3C Process sucks!
130	wakaba	1.54	Whatpm::ContentChecker::FEATURE_STATUS_REC
131			}
132			sub FEATURE_M12N10_REC_DEPRECATED () {
133			Whatpm::ContentChecker::FEATURE_STATUS_REC \|
134			Whatpm::ContentChecker::FEATURE_DEPRECATED_INFO
135			}
136	wakaba	1.49
137			## NOTE: XHTML10 status is based on its transitional and frameset DTDs
138			## (second edition). Only missing attributes from M12N10 abstract
139			## definition are added.
140	wakaba	1.54	sub FEATURE_XHTML10_REC () {
141			Whatpm::ContentChecker::FEATURE_STATUS_CR
142			}
143
144	wakaba	1.61	## NOTE: Diff from HTML4.
145			sub FEATURE_ISOHTML_PREPARATION () { ## Informative documentation
146			Whatpm::ContentChecker::FEATURE_STATUS_CR
147			}
148	wakaba	1.58
149	wakaba	1.49	## NOTE: HTML4 status is based on its transitional and frameset DTDs (HTML
150			## 4.01). Only missing attributes from XHTML10 are added.
151	wakaba	1.54	sub FEATURE_HTML4_REC_RESERVED () {
152			Whatpm::ContentChecker::FEATURE_STATUS_WD
153			}
154
155			## TODO: According to HTML4 definition, authors SHOULD use style sheets
156			## rather than presentational attributes (deprecated or not deprecated).
157	wakaba	1.48
158	wakaba	1.61	## NOTE: Diff from HTML4.
159			sub FEATURE_HTML32_REC_OBSOLETE () {
160			Whatpm::ContentChecker::FEATURE_STATUS_CR \|
161			Whatpm::ContentChecker::FEATURE_DEPRECATED_SHOULD
162			## NOTE: Lowercase normative "should".
163			}
164
165			sub FEATURE_RFC2659 () { ## Experimental RFC
166			Whatpm::ContentChecker::FEATURE_STATUS_CR
167			}
168
169			## NOTE: HTML 2.x - diff from HTML 2.0 and not in newer versions.
170			sub FEATURE_HTML2X_RFC () { ## Proposed Standard, obsolete
171			Whatpm::ContentChecker::FEATURE_STATUS_CR
172			}
173
174			## NOTE: Diff from HTML 2.0.
175			sub FEATURE_RFC1942 () { ## Experimental RFC, obsolete
176			Whatpm::ContentChecker::FEATURE_STATUS_CR
177			}
178
179			## NOTE: Diff from HTML 3.2.
180			sub FEATURE_HTML20_RFC () { ## Proposed Standard, obsolete
181			Whatpm::ContentChecker::FEATURE_STATUS_CR
182			}
183	wakaba	1.58
184	wakaba	1.174	## --- Content Model ---
185
186	wakaba	1.29	## December 2007 HTML5 Classification
187
188			my $HTMLMetadataContent = {
189			$HTML_NS => {
190			title => 1, base => 1, link => 1, style => 1, script => 1, noscript => 1,
191	wakaba	1.118	'event-source' => 1, eventsource => 1,
192			command => 1, datatemplate => 1,
193	wakaba	1.29	## NOTE: A \|meta\| with no \|name\| element is not allowed as
194			## a metadata content other than \|head\| element.
195			meta => 1,
196			},
197			## NOTE: RDF is mentioned in the HTML5 spec.
198			## TODO: Other RDF elements?
199			q<http://www.w3.org/1999/02/22-rdf-syntax-ns#> => {RDF => 1},
200			};
201
202	wakaba	1.72	my $HTMLFlowContent = {
203	wakaba	1.29	$HTML_NS => {
204			section => 1, nav => 1, article => 1, blockquote => 1, aside => 1,
205			h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1, header => 1,
206			footer => 1, address => 1, p => 1, hr => 1, dialog => 1, pre => 1,
207			ol => 1, ul => 1, dl => 1, figure => 1, map => 1, table => 1,
208	wakaba	1.119	form => 1, fieldset => 1,
209	wakaba	1.72	details => 1, ## ISSUE: "Flow element" in spec.
210			datagrid => 1, ## ISSUE: "Flow element" in spec.
211	wakaba	1.29	data