whatpm/t/XML-Parser.t

#!/usr/bin/perl
use strict;

my $DEBUG = $ENV{DEBUG};

use lib qw[/home/wakaba/work/manakai2/lib];
my $test_dir_name = 't/xml/';

use Test;
BEGIN { plan tests => 1573 }

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";
    }
  }
}

use Whatpm::XML::Parser;
use Whatpm::NanoDOM;
use Whatpm::Charset::UnicodeChecker;
use Whatpm::HTML::Dumper qw/dumptree/;

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;
  
  $SIG{INT} = sub {
    print scalar dumptree ($doc);
    exit;
  };

  my $onerror = sub {
    my %opt = @_;
    push @errors, join ';',
        $opt{token}->{line} || $opt{line},
        $opt{token}->{column} || $opt{column},
        $opt{type},
        defined $opt{text} ? $opt{text} : '',
        defined $opt{value} ? $opt{value} : '',
        $opt{level};
  };

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

  my $p = Whatpm::XML::Parser->new;
  my $result;
  unless (defined $test->{element}) {
    $p->parse_char_string ($test->{data}->[0] => $doc, $onerror, $chk);
    $result = dumptree ($doc);
  } else {
    ## TODO: ...
    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);
    $result = dumptree ($el);
  }
  
  warn "No #errors section ($test->{data}->[0])" unless $test->{errors};

  @errors = sort {$a cmp $b} @errors;
  @{$test->{errors}->[0]} = sort {$a cmp $b} @{$test->{errors}->[0] ||= []};
  
  ok join ("\n", @errors), join ("\n", @{$test->{errors}->[0] or []}),
    'Parse error: ' . Data::Dumper::qquote ($test->{data}->[0]);

  if ($test->{'xml-version'}) {
    ok $doc->xml_version, $test->{'xml-version'}->[0],
        'XML version: ' . Data::Dumper::qquote ($test->{data}->[0]);
  }

  if ($test->{'xml-encoding'}) {
    if ($test->{'xml-encoding'}->[1]->[0] eq 'null') {
      ok $doc->xml_encoding, undef, 
        'XML encoding: ' . Data::Dumper::qquote ($test->{data}->[0]);
    } else {
      ok $doc->xml_encoding, $test->{'xml-encoding'}->[0],
          'XML encoding: ' . Data::Dumper::qquote ($test->{data}->[0]);
    }
  }

  if ($test->{'xml-standalone'}) {
    ok $doc->xml_standalone ? 1 : 0,
        $test->{'xml-standalone'}->[1]->[0] eq 'true' ? 1 : 0,
        'XML standalone: ' . Data::Dumper::qquote ($test->{data}->[0]);
  }

  if ($test->{entities}) {
    my @e;
    for (keys %{$p->{ge}}) {
      my $ent = $p->{ge}->{$_};
      my $v = '<!ENTITY ' . $ent->{name} . ' "'; 
      $v .= $ent->{value} if defined $ent->{value};
      $v .= '" "';
      $v .= $ent->{pubid} if defined $ent->{pubid};
      $v .= '" "';
      $v .= $ent->{sysid} if defined $ent->{sysid};
      $v .= '" ';
      $v .= $ent->{notation} if defined $ent->{notation};
      $v .= '>';
      push @e, $v;
    }
    for (keys %{$p->{pe}}) {
      my $ent = $p->{pe}->{$_};
      my $v = '<!ENTITY % ' . $ent->{name} . ' "'; 
      $v .= $ent->{value} if defined $ent->{value};
      $v .= '" "';
      $v .= $ent->{pubid} if defined $ent->{pubid};
      $v .= '" "';
      $v .= $ent->{sysid} if defined $ent->{sysid};
      $v .= '" ';
      $v .= $ent->{notation} if defined $ent->{notation};
      $v .= '>';
      push @e, $v;
    }
    ok join ("\x0A", @e), $test->{entities}->[0],
        'Entities: ' . Data::Dumper::qquote ($test->{data}->[0]);
  }
  
  $test->{document}->[0] .= "\x0A" if length $test->{document}->[0];
  ok $result, $test->{document}->[0],
      'Document tree: ' . Data::Dumper::qquote ($test->{data}->[0]);
} # test

my @FILES = grep {$_} split /\s+/, qq[
  ${test_dir_name}elements-1.dat
  ${test_dir_name}attrs-1.dat
  ${test_dir_name}attrs-2.dat
  ${test_dir_name}texts-1.dat
  ${test_dir_name}cdata-1.dat
  ${test_dir_name}charref-1.dat
  ${test_dir_name}comments-2.dat
  ${test_dir_name}pis-1.dat
  ${test_dir_name}pis-2.dat
  ${test_dir_name}xmldecls-1.dat
  ${test_dir_name}tree-1.dat
  ${test_dir_name}ns-elements-1.dat
  ${test_dir_name}ns-attrs-1.dat
  ${test_dir_name}doctypes-1.dat
  ${test_dir_name}doctypes-2.dat
  ${test_dir_name}eldecls-1.dat
  ${test_dir_name}attlists-1.dat
  ${test_dir_name}entities-1.dat
  ${test_dir_name}entities-2.dat
  ${test_dir_name}notations-1.dat
  ${test_dir_name}entrefs-1.dat
  ${test_dir_name}entrefs-2.dat
];

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

## License: Public Domain.
## $Date: 2008/10/19 10:12:54 $
1	wakaba	1.1	#!/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/xml/';
8
9			use Test;
10	wakaba	1.18	BEGIN { plan tests => 1573 }
11	wakaba	1.1
12			use Data::Dumper;
13			$Data::Dumper::Useqq = 1;
14			sub Data::Dumper::qquote {
15			my $s = shift;
16			$s =~ s/([^\x20\x21-\x26\x28-\x5B\x5D-\x7E])/sprintf '\x{%02X}', ord $1/ge;
17			return q<qq'> . $s . q<'>;
18			} # Data::Dumper::qquote
19
20			if ($DEBUG) {
21			my $not_found = {%{$Whatpm::HTML::Debug::cp or {}}};
22			$Whatpm::HTML::Debug::cp_pass = sub {
23			my $id = shift;
24			delete $not_found->{$id};
25			};
26
27			END {
28			for my $id (sort {$a <=> $b \|\| $a cmp $b} keys %$not_found) {
29			print "# checkpoint $id is not reached\n";
30			}
31			}
32			}
33
34			use Whatpm::XML::Parser;
35			use Whatpm::NanoDOM;
36			use Whatpm::Charset::UnicodeChecker;
37			use Whatpm::HTML::Dumper qw/dumptree/;
38
39			sub test ($) {
40			my $test = shift;
41
42			if ($test->{'document-fragment'}) {
43			if (@{$test->{'document-fragment'}->[1]}) {
44			## NOTE: Old format.
45			$test->{element} = $test->{'document-fragment'}->[1]->[0];
46			$test->{document} \|\|= $test->{'document-fragment'};
47			} else {
48			## NOTE: New format.
49			$test->{element} = $test->{'document-fragment'}->[0];
50			}
51			}
52
53			my $doc = Whatpm::NanoDOM::Document->new;
54			my @errors;
55
56			$SIG{INT} = sub {
57			print scalar dumptree ($doc);
58			exit;
59			};
60
61			my $onerror = sub {
62			my %opt = @_;
63			push @errors, join ';',
64			$opt{token}->{line} \|\| $opt{line},
65			$opt{token}->{column} \|\| $opt{column},
66			$opt{type},
67			defined $opt{text} ? $opt{text} : '',
68			defined $opt{value} ? $opt{value} : '',
69			$opt{level};
70			};
71
72			my $chk = sub {
73			return $_[0];
74			#return Whatpm::Charset::UnicodeChecker->new_handle ($_[0], 'html5');
75			}; # $chk
76
77	wakaba	1.13	my $p = Whatpm::XML::Parser->new;
78	wakaba	1.1	my $result;
79			unless (defined $test->{element}) {
80	wakaba	1.13	$p->parse_char_string ($test->{data}->[0] => $doc, $onerror, $chk);
81	wakaba	1.1	$result = dumptree ($doc);
82			} else {
83			## TODO: ...
84			my $el = $doc->create_element_ns
85			('http://www.w3.org/1999/xhtml', [undef, $test->{element}]);
86			Whatpm::HTML->set_inner_html ($el, $test->{data}->[0], $onerror, $chk);
87			$result = dumptree ($el);
88			}
89
90			warn "No #errors section ($test->{data}->[0])" unless $test->{errors};
91	wakaba	1.5
92			@errors = sort {$a cmp $b} @errors;
93			@{$test->{errors}->[0]} = sort {$a cmp $b} @{$test->{errors}->[0] \|\|= []};
94	wakaba	1.1
95			ok join ("\n", @errors), join ("\n", @{$test->{errors}->[0] or []}),
96	wakaba	1.7	'Parse error: ' . Data::Dumper::qquote ($test->{data}->[0]);
97
98			if ($test->{'xml-version'}) {
99			ok $doc->xml_version, $test->{'xml-version'}->[0],
100			'XML version: ' . Data::Dumper::qquote ($test->{data}->[0]);
101			}
102
103			if ($test->{'xml-encoding'}) {
104			if ($test->{'xml-encoding'}->[1]->[0] eq 'null') {
105			ok $doc->xml_encoding, undef,
106			'XML encoding: ' . Data::Dumper::qquote ($test->{data}->[0]);
107			} else {
108			ok $doc->xml_encoding, $test->{'xml-encoding'}->[0],
109			'XML encoding: ' . Data::Dumper::qquote ($test->{data}->[0]);
110			}
111			}
112
113			if ($test->{'xml-standalone'}) {
114			ok $doc->xml_standalone ? 1 : 0,
115			$test->{'xml-standalone'}->[1]->[0] eq 'true' ? 1 : 0,
116			'XML standalone: ' . Data::Dumper::qquote ($test->{data}->[0]);
117			}
118	wakaba	1.13
119			if ($test->{entities}) {
120			my @e;
121			for (keys %{$p->{ge}}) {
122			my $ent = $p->{ge}->{$_};
123	wakaba	1.16	my $v = '<!ENTITY ' . $ent->{name} . ' "';
124	wakaba	1.14	$v .= $ent->{value} if defined $ent->{value};
125	wakaba	1.13	$v .= '" "';
126			$v .= $ent->{pubid} if defined $ent->{pubid};
127			$v .= '" "';
128			$v .= $ent->{sysid} if defined $ent->{sysid};
129			$v .= '" ';
130			$v .= $ent->{notation} if defined $ent->{notation};
131			$v .= '>';
132			push @e, $v;
133			}
134			for (keys %{$p->{pe}}) {
135			my $ent = $p->{pe}->{$_};
136	wakaba	1.14	my $v = '<!ENTITY % ' . $ent->{name} . ' "';
137			$v .= $ent->{value} if defined $ent->{value};
138	wakaba	1.13	$v .= '" "';
139			$v .= $ent->{pubid} if defined $ent->{pubid};
140			$v .= '" "';
141			$v .= $ent->{sysid} if defined $ent->{sysid};
142			$v .= '" ';
143			$v .= $ent->{notation} if defined $ent->{notation};
144			$v .= '>';
145			push @e, $v;
146			}
147			ok join ("\x0A", @e), $test->{entities}->[0],
148			'Entities: ' . Data::Dumper::qquote ($test->{data}->[0]);
149			}
150	wakaba	1.1
151			$test->{document}->[0] .= "\x0A" if length $test->{document}->[0];
152			ok $result, $test->{document}->[0],
153			'Document tree: ' . Data::Dumper::qquote ($test->{data}->[0]);
154			} # test
155
156			my @FILES = grep {$_} split /\s+/, qq[
157	wakaba	1.4	${test_dir_name}elements-1.dat
158	wakaba	1.5	${test_dir_name}attrs-1.dat
159	wakaba	1.18	${test_dir_name}attrs-2.dat
160	wakaba	1.5	${test_dir_name}texts-1.dat
161	wakaba	1.2	${test_dir_name}cdata-1.dat
162	wakaba	1.6	${test_dir_name}charref-1.dat
163	wakaba	1.10	${test_dir_name}comments-2.dat
164	wakaba	1.7	${test_dir_name}pis-1.dat
165	wakaba	1.10	${test_dir_name}pis-2.dat
166	wakaba	1.7	${test_dir_name}xmldecls-1.dat
167	wakaba	1.2	${test_dir_name}tree-1.dat
168	wakaba	1.8	${test_dir_name}ns-elements-1.dat
169	wakaba	1.3	${test_dir_name}ns-attrs-1.dat
170	wakaba	1.4	${test_dir_name}doctypes-1.dat
171	wakaba	1.9	${test_dir_name}doctypes-2.dat
172	wakaba	1.15	${test_dir_name}eldecls-1.dat
173	wakaba	1.11	${test_dir_name}attlists-1.dat
174	wakaba	1.13	${test_dir_name}entities-1.dat
175			${test_dir_name}entities-2.dat
176	wakaba	1.15	${test_dir_name}notations-1.dat
177	wakaba	1.16	${test_dir_name}entrefs-1.dat
178	wakaba	1.17	${test_dir_name}entrefs-2.dat
179	wakaba	1.2	];
180	wakaba	1.1
181			require 't/testfiles.pl';
182			execute_test ($_, {
183			errors => {is_list => 1},
184			document => {is_prefixed => 1},
185			'document-fragment' => {is_prefixed => 1},
186	wakaba	1.13	entities => {is_prefixed => 1},
187	wakaba	1.1	}, \&test) for @FILES;
188
189			## License: Public Domain.
190	wakaba	1.18	## $Date: 2008/10/19 10:12:54 $