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 => 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";
    }
  }
}

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 $result;
  unless (defined $test->{element}) {
    Whatpm::XML::Parser->parse_char_string
        ($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};
  
  ok join ("\n", @errors), join ("\n", @{$test->{errors}->[0] or []}),
    'Parse error: ' . Data::Dumper::qquote ($test->{data}->[0]) . '; ' . 
    join (', ', @errors) . ';' . join (', ', @{$test->{errors}->[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

my @FILES = grep {$_} split /\s+/, qq[
  ${test_dir_name}cdata-1.dat
  ${test_dir_name}attrs-1.dat
  ${test_dir_name}tree-1.dat
];

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

## License: Public Domain.
## $Date: 2008/10/14 09:00:57 $
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			BEGIN { plan tests => 4935 }
11
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			my $result;
78			unless (defined $test->{element}) {
79			Whatpm::XML::Parser->parse_char_string
80			($test->{data}->[0] => $doc, $onerror, $chk);
81			$result = dumptree ($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 $test->{errors};
91
92			ok join ("\n", @errors), join ("\n", @{$test->{errors}->[0] or []}),
93			'Parse error: ' . Data::Dumper::qquote ($test->{data}->[0]) . '; ' .
94			join (', ', @errors) . ';' . join (', ', @{$test->{errors}->[0] or []});
95
96			$test->{document}->[0] .= "\x0A" if length $test->{document}->[0];
97			ok $result, $test->{document}->[0],
98			'Document tree: ' . Data::Dumper::qquote ($test->{data}->[0]);
99			} # test
100
101			my @FILES = grep {$_} split /\s+/, qq[
102	wakaba	1.2	${test_dir_name}cdata-1.dat
103			${test_dir_name}attrs-1.dat
104			${test_dir_name}tree-1.dat
105			];
106	wakaba	1.1
107			require 't/testfiles.pl';
108			execute_test ($_, {
109			errors => {is_list => 1},
110			document => {is_prefixed => 1},
111			'document-fragment' => {is_prefixed => 1},
112			}, \&test) for @FILES;
113
114			## License: Public Domain.
115	wakaba	1.2	## $Date: 2008/10/14 09:00:57 $