whatpm/t/HTML-tree.t

#!/usr/bin/perl
use strict;

my $DEBUG = $ENV{DEBUG};

use lib qw[/home/wakaba/work/manakai2/lib];
my $test_dir_name = 't/';
my $dir_name = 't/tree-construction/';

use Test;
BEGIN { plan tests => 4935 }

use Data::Dumper;
$Data::Dumper::Useqq = 1;
sub Data::Dumper::qquote {
  my $s = shift;
  $s =~ s/([^\x20\x21-\x26\x28-\x5B\x5D-\x7E])/sprintf '\x{%02X}', ord $1/ge;
  return q<qq'> . $s . q<'>;
} # Data::Dumper::qquote

if ($DEBUG) {
  my $not_found = {%{$Whatpm::HTML::Debug::cp or {}}};
  $Whatpm::HTML::Debug::cp_pass = sub {
    my $id = shift;
    delete $not_found->{$id};
  };

  END {
    for my $id (sort {$a <=> $b || $a cmp $b} keys %$not_found) {
      print "# checkpoint $id is not reached\n";
    }
  }
}

my @FILES = grep {$_} split /\s+/, qq[
                      ${test_dir_name}tokenizer-test-2.dat
                      ${test_dir_name}tokenizer-test-3.dat
                      ${dir_name}tests1.dat
                      ${dir_name}tests2.dat
                      ${dir_name}tests3.dat
                      ${dir_name}tests4.dat
                      ${dir_name}tests5.dat
                      ${dir_name}tests6.dat
                      ${dir_name}tests7.dat
                      ${dir_name}tests8.dat
                      ${dir_name}tests9.dat
                      ${dir_name}tests10.dat
                      ${dir_name}tests11.dat
                      ${dir_name}tests12.dat
                      ${test_dir_name}tree-test-1.dat
                      ${test_dir_name}tree-test-2.dat
                      ${test_dir_name}tree-test-3.dat
                      ${test_dir_name}tree-test-void.dat
                      ${test_dir_name}tree-test-flow.dat
                      ${test_dir_name}tree-test-phrasing.dat
                      ${test_dir_name}tree-test-form.dat
                      ${test_dir_name}tree-test-foreign.dat
                     ];

require 't/testfiles.pl';
execute_test ($_, {
  errors => {is_list => 1},
  shoulds => {is_list => 1},
  document => {is_prefixed => 1},
  'document-fragment' => {is_prefixed => 1},
}, \&test) for @FILES;

use Whatpm::HTML;
use Whatpm::NanoDOM;
use Whatpm::Charset::UnicodeChecker;

sub test ($) {
  my $test = shift;

  if ($test->{'document-fragment'}) {
    if (@{$test->{'document-fragment'}->[1]}) {
      ## NOTE: Old format.
      $test->{element} = $test->{'document-fragment'}->[1]->[0];
      $test->{document} ||= $test->{'document-fragment'};
    } else {
      ## NOTE: New format.
      $test->{element} = $test->{'document-fragment'}->[0];
    }
  }

  my $doc = Whatpm::NanoDOM::Document->new;
  my @errors;
  my @shoulds;
  
  $SIG{INT} = sub {
    print scalar serialize ($doc);
    exit;
  };

  my $onerror = sub {
    my %opt = @_;
    if ($opt{level} eq 's') {
      push @shoulds, join ':', $opt{line}, $opt{column}, $opt{type};
    } else {
      push @errors, join ':', $opt{line}, $opt{column}, $opt{type};
    }
  };

  my $chk = sub {
    return Whatpm::Charset::UnicodeChecker->new_handle ($_[0], 'html5');
  }; # $chk

  my $result;
  unless (defined $test->{element}) {
    Whatpm::HTML->parse_char_string
        ($test->{data}->[0] => $doc, $onerror, $chk);
    $result = serialize ($doc);
  } else {
    my $el = $doc->create_element_ns
      ('http://www.w3.org/1999/xhtml', [undef, $test->{element}]);
    Whatpm::HTML->set_inner_html ($el, $test->{data}->[0], $onerror, $chk);
    $result = serialize ($el);
  }
  
  warn "No #errors section ($test->{data}->[0])" unless $test->{errors};
    
  ok scalar @errors, scalar @{$test->{errors}->[0] or []},
    'Parse error: ' . Data::Dumper::qquote ($test->{data}->[0]) . '; ' . 
    join (', ', @errors) . ';' . join (', ', @{$test->{errors}->[0] or []});
  ok scalar @shoulds, scalar @{$test->{shoulds}->[0] or []},
    'SHOULD-level error: ' . Data::Dumper::qquote ($test->{data}->[0]) . '; ' .
    join (', ', @shoulds) . ';' . join (', ', @{$test->{shoulds}->[0] or []});

  $test->{document}->[0] .= "\x0A" if length $test->{document}->[0];
  ok $result, $test->{document}->[0],
      'Document tree: ' . Data::Dumper::qquote ($test->{data}->[0]);
} # test

## NOTE: Spec: <http://wiki.whatwg.org/wiki/Parser_tests>.
sub serialize ($) {
  my $node = shift;
  my $r = '';

  my $ns_id = {
    q<http://www.w3.org/1999/xhtml> => 'html',
    q<http://www.w3.org/2000/svg> => 'svg',
    q<http://www.w3.org/1998/Math/MathML> => 'math',
    q<http://www.w3.org/1999/xlink> => 'xlink',
    q<http://www.w3.org/XML/1998/namespace> => 'xml',
    q<http://www.w3.org/2002/xmlns/> => 'xmlns',
  };

  my @node = map { [$_, ''] } @{$node->child_nodes};
  while (@node) {
    my $child = shift @node;
    my $nt = $child->[0]->node_type;
    if ($nt == $child->[0]->ELEMENT_NODE) {
      my $ns = $child->[0]->namespace_uri;
      unless (defined $ns) {
        $ns = '{} ';
      } elsif ($ns eq q<http://www.w3.org/1999/xhtml>) {
        $ns = '';
      } elsif ($ns_id->{$ns}) {
        $ns = $ns_id->{$ns} . ' ';
      } else {
        $ns = '{' . $ns . '} ';
      }
      $r .= $child->[1] . '<' . $ns . $child->[0]->manakai_local_name . ">\x0A";

      for my $attr (sort {$a->[0] cmp $b->[0]} map { [do {
                      my $ns = $_->namespace_uri;
                      unless (defined $ns) {
                        $ns = '';
                      } elsif ($ns_id->{$ns}) {
                        $ns = $ns_id->{$ns} . ' ';
                      } else {
                        $ns = '{' . $ns . '} ';
                      }
                      $ns . $_->manakai_local_name;
                    }, $_->value] }
                    @{$child->[0]->attributes}) {
        $r .= $child->[1] . '  ' . $attr->[0] . '="'; ## ISSUE: case?
        $r .= $attr->[1] . '"' . "\x0A";
      }
      
      unshift @node,
        map { [$_, $child->[1] . '  '] } @{$child->[0]->child_nodes};
    } elsif ($nt == $child->[0]->TEXT_NODE) {
      $r .= $child->[1] . '"' . $child->[0]->data . '"' . "\x0A";
    } elsif ($nt == $child->[0]->COMMENT_NODE) {
      $r .= $child->[1] . '<!-- ' . $child->[0]->data . " -->\x0A";
    } elsif ($nt == $child->[0]->DOCUMENT_TYPE_NODE) {
      $r .= $child->[1] . '<!DOCTYPE ' . $child->[0]->name;
      my $pubid = $child->[0]->public_id;
      my $sysid = $child->[0]->system_id;
      if (length $pubid or length $sysid) {
        $r .= ' "' . $pubid . '"';
        $r .= ' "' . $sysid . '"';
      }
      $r .= ">\x0A";
    } else {
      $r .= $child->[1] . $child->[0]->node_type . "\x0A"; # error
    }
  }
  
  return $r;
} # serialize

## License: Public Domain.
## $Date: 2008/10/14 06:48:05 $
1	#!/usr/bin/perl
2	use strict;
3
4	my $DEBUG = $ENV{DEBUG};
5
6	use lib qw[/home/wakaba/work/manakai2/lib];
7	my $test_dir_name = 't/';
8	my $dir_name = 't/tree-construction/';
9
10	use Test;
11	BEGIN { plan tests => 4935 }
12
13	use Data::Dumper;
14	$Data::Dumper::Useqq = 1;
15	sub Data::Dumper::qquote {
16	my $s = shift;
17	$s =~ s/([^\x20\x21-\x26\x28-\x5B\x5D-\x7E])/sprintf '\x{%02X}', ord $1/ge;
18	return q<qq'> . $s . q<'>;
19	} # Data::Dumper::qquote
20
21	if ($DEBUG) {
22	my $not_found = {%{$Whatpm::HTML::Debug::cp or {}}};
23	$Whatpm::HTML::Debug::cp_pass = sub {
24	my $id = shift;
25	delete $not_found->{$id};
26	};
27
28	END {
29	for my $id (sort {$a <=> $b \|\| $a cmp $b} keys %$not_found) {
30	print "# checkpoint $id is not reached\n";
31	}
32	}
33	}
34
35	my @FILES = grep {$_} split /\s+/, qq[
36	${test_dir_name}tokenizer-test-2.dat
37	${test_dir_name}tokenizer-test-3.dat
38	${dir_name}tests1.dat
39	${dir_name}tests2.dat
40	${dir_name}tests3.dat
41	${dir_name}tests4.dat
42	${dir_name}tests5.dat
43	${dir_name}tests6.dat
44	${dir_name}tests7.dat
45	${dir_name}tests8.dat
46	${dir_name}tests9.dat
47	${dir_name}tests10.dat
48	${dir_name}tests11.dat
49	${dir_name}tests12.dat
50	${test_dir_name}tree-test-1.dat
51	${test_dir_name}tree-test-2.dat
52	${test_dir_name}tree-test-3.dat
53	${test_dir_name}tree-test-void.dat
54	${test_dir_name}tree-test-flow.dat
55	${test_dir_name}tree-test-phrasing.dat
56	${test_dir_name}tree-test-form.dat
57	${test_dir_name}tree-test-foreign.dat
58	];
59
60	require 't/testfiles.pl';
61	execute_test ($_, {
62	errors => {is_list => 1},
63	shoulds => {is_list => 1},
64	document => {is_prefixed => 1},
65	'document-fragment' => {is_prefixed => 1},
66	}, \&test) for @FILES;
67
68	use Whatpm::HTML;
69	use Whatpm::NanoDOM;
70	use Whatpm::Charset::UnicodeChecker;
71
72	sub test ($) {
73	my $test = shift;
74
75	if ($test->{'document-fragment'}) {
76	if (@{$test->{'document-fragment'}->[1]}) {
77	## NOTE: Old format.
78	$test->{element} = $test->{'document-fragment'}->[1]->[0];
79	$test->{document} \|\|= $test->{'document-fragment'};
80	} else {
81	## NOTE: New format.
82	$test->{element} = $test->{'document-fragment'}->[0];
83	}
84	}
85
86	my $doc = Whatpm::NanoDOM::Document->new;
87	my @errors;
88	my @shoulds;
89
90	$SIG{INT} = sub {
91	print scalar serialize ($doc);
92	exit;
93	};
94
95	my $onerror = sub {
96	my %opt = @_;
97	if ($opt{level} eq 's') {
98	push @shoulds, join ':', $opt{line}, $opt{column}, $opt{type};
99	} else {
100	push @errors, join ':', $opt{line}, $opt{column}, $opt{type};
101	}
102	};
103
104	my $chk = sub {
105	return Whatpm::Charset::UnicodeChecker->new_handle ($_[0], 'html5');
106	}; # $chk
107
108	my $result;
109	unless (defined $test->{element}) {
110	Whatpm::HTML->parse_char_string
111	($test->{data}->[0] => $doc, $onerror, $chk);
112	$result = serialize ($doc);
113	} else {
114	my $el = $doc->create_element_ns
115	('http://www.w3.org/1999/xhtml', [undef, $test->{element}]);
116	Whatpm::HTML->set_inner_html ($el, $test->{data}->[0], $onerror, $chk);
117	$result = serialize ($el);
118	}
119
120	warn "No #errors section ($test->{data}->[0])" unless $test->{errors};
121
122	ok scalar @errors, scalar @{$test->{errors}->[0] or []},
123	'Parse error: ' . Data::Dumper::qquote ($test->{data}->[0]) . '; ' .
124	join (', ', @errors) . ';' . join (', ', @{$test->{errors}->[0] or []});
125	ok scalar @shoulds, scalar @{$test->{shoulds}->[0] or []},
126	'SHOULD-level error: ' . Data::Dumper::qquote ($test->{data}->[0]) . '; ' .
127	join (', ', @shoulds) . ';' . join (', ', @{$test->{shoulds}->[0] or []});
128
129	$test->{document}->[0] .= "\x0A" if length $test->{document}->[0];
130	ok $result, $test->{document}->[0],
131	'Document tree: ' . Data::Dumper::qquote ($test->{data}->[0]);
132	} # test
133
134	## NOTE: Spec: <http://wiki.whatwg.org/wiki/Parser_tests>.
135	sub serialize ($) {
136	my $node = shift;
137	my $r = '';
138
139	my $ns_id = {
140	q<http://www.w3.org/1999/xhtml> => 'html',
141	q<http://www.w3.org/2000/svg> => 'svg',
142	q<http://www.w3.org/1998/Math/MathML> => 'math',
143	q<http://www.w3.org/1999/xlink> => 'xlink',
144	q<http://www.w3.org/XML/1998/namespace> => 'xml',
145	q<http://www.w3.org/2002/xmlns/> => 'xmlns',
146	};
147
148	my @node = map { [$_, ''] } @{$node->child_nodes};
149	while (@node) {
150	my $child = shift @node;
151	my $nt = $child->[0]->node_type;
152	if ($nt == $child->[0]->ELEMENT_NODE) {
153	my $ns = $child->[0]->namespace_uri;
154	unless (defined $ns) {
155	$ns = '{} ';
156	} elsif ($ns eq q<http://www.w3.org/1999/xhtml>) {
157	$ns = '';
158	} elsif ($ns_id->{$ns}) {
159	$ns = $ns_id->{$ns} . ' ';
160	} else {
161	$ns = '{' . $ns . '} ';
162	}
163	$r .= $child->[1] . '<' . $ns . $child->[0]->manakai_local_name . ">\x0A";
164
165	for my $attr (sort {$a->[0] cmp $b->[0]} map { [do {
166	my $ns = $_->namespace_uri;
167	unless (defined $ns) {
168	$ns = '';
169	} elsif ($ns_id->{$ns}) {
170	$ns = $ns_id->{$ns} . ' ';
171	} else {
172	$ns = '{' . $ns . '} ';
173	}
174	$ns . $_->manakai_local_name;
175	}, $_->value] }
176	@{$child->[0]->attributes}) {
177	$r .= $child->[1] . ' ' . $attr->[0] . '="'; ## ISSUE: case?
178	$r .= $attr->[1] . '"' . "\x0A";
179	}
180
181	unshift @node,
182	map { [$_, $child->[1] . ' '] } @{$child->[0]->child_nodes};
183	} elsif ($nt == $child->[0]->TEXT_NODE) {
184	$r .= $child->[1] . '"' . $child->[0]->data . '"' . "\x0A";
185	} elsif ($nt == $child->[0]->COMMENT_NODE) {
186	$r .= $child->[1] . '<!-- ' . $child->[0]->data . " -->\x0A";
187	} elsif ($nt == $child->[0]->DOCUMENT_TYPE_NODE) {
188	$r .= $child->[1] . '<!DOCTYPE ' . $child->[0]->name;
189	my $pubid = $child->[0]->public_id;
190	my $sysid = $child->[0]->system_id;
191	if (length $pubid or length $sysid) {
192	$r .= ' "' . $pubid . '"';
193	$r .= ' "' . $sysid . '"';
194	}
195	$r .= ">\x0A";
196	} else {
197	$r .= $child->[1] . $child->[0]->node_type . "\x0A"; # error
198	}
199	}
200
201	return $r;
202	} # serialize
203
204	## License: Public Domain.
205	## $Date: 2008/10/14 06:48:05 $