whatpm/t/HTML-tree.t

#!/usr/bin/perl
use strict;

my $dir_name;
my $test_dir_name;
BEGIN {
  $test_dir_name = 't/';
  $dir_name = 't/tree-construction/';
  my $skip = "You don't have make command";
  eval q{
         system ("cd $test_dir_name; make tree-construction-files") == 0 or die
           unless -f $dir_name.'tests1.dat';
         $skip = '';
        };
  if ($skip) {
    print "1..1\n";
    print "ok 1 # $skip\n";
    exit;
  }
}

use Test;
BEGIN { plan tests => 472 }

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

for my $file_name (grep {$_} split /\s+/, qq[
                      ${test_dir_name}tokenizer-test-2.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
                      ${test_dir_name}tree-test-1.dat
                     ]) {
  open my $file, '<', $file_name
    or die "$0: $file_name: $!";
  print "# $file_name\n";

  my $test;
  my $mode = 'data';
  my $escaped;
  while (<$file>) {
    s/\x0D\x0A/\x0A/;
    if (/^#data$/) {
      undef $test;
      $test->{data} = '';
      $mode = 'data';
      undef $escaped;
    } elsif (/^#data escaped$/) {
      undef $test;
      $test->{data} = '';
      $mode = 'data';
      $escaped = 1;
    } elsif (/^#errors$/) {
      $test->{errors} = [];
      $mode = 'errors';
      $test->{data} =~ s/\x0D?\x0A\z//;       
      $test->{data} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
      undef $escaped;
    } elsif (/^#document$/) {
      $test->{document} = '';
      $mode = 'document';
      undef $escaped;
    } elsif (/^#document escaped$/) {
      $test->{document} = '';
      $mode = 'document';
      $escaped = 1;
    } elsif (/^#document-fragment (\S+)$/) {
      $test->{document} = '';
      $mode = 'document';
      $test->{element} = $1;
      undef $escaped;
    } elsif (/^#document-fragment (\S+) escaped$/) {
      $test->{document} = '';
      $mode = 'document';
      $test->{element} = $1;
      $escaped = 1;
    } elsif (defined $test->{document} and /^$/) {
      $test->{document} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
      test ($test);
      undef $test;
    } else {
      if ($mode eq 'data' or $mode eq 'document') {
        $test->{$mode} .= $_;
      } elsif ($mode eq 'errors') {
        tr/\x0D\x0A//d;
        push @{$test->{errors}}, $_;
      }
    }
  }
  test ($test) if $test->{errors};
}

use Whatpm::HTML;
use Whatpm::NanoDOM;

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

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

  my $onerror = sub {
    my %opt = @_;
    push @errors, join ':', $opt{line}, $opt{column}, $opt{type};
  };
  my $result;
  unless (defined $test->{element}) {
    Whatpm::HTML->parse_string ($test->{data} => $doc, $onerror);
    $result = serialize ($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}, $onerror);
    $result = serialize ($el);
  }
    
  ok scalar @errors, scalar @{$test->{errors}},
    'Parse error: ' . $test->{data} . '; ' . 
    join (', ', @errors) . ';' . join (', ', @{$test->{errors}});

  ok $result, $test->{document}, 'Document tree: ' . $test->{data};
} # test

sub serialize ($) {
  my $node = shift;
  my $r = '';

  my @node = map { [$_, ''] } @{$node->child_nodes};
  while (@node) {
    my $child = shift @node;
    my $nt = $child->[0]->node_type;
    if ($nt == $child->[0]->ELEMENT_NODE) {
      $r .= '| ' . $child->[1] . '<' . $child->[0]->tag_name . ">\x0A"; ## ISSUE: case?

      for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->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 . ">\x0A";
    } else {
      $r .= '| ' . $child->[1] . $child->[0]->node_type . "\x0A"; # error
    }
  }
  
  return $r;
} # serialize

## License: Public Domain.
## $Date: 2007/06/23 03:53:35 $
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4			my $dir_name;
5	wakaba	1.2	my $test_dir_name;
6	wakaba	1.1	BEGIN {
7	wakaba	1.2	$test_dir_name = 't/';
8	wakaba	1.1	$dir_name = 't/tree-construction/';
9			my $skip = "You don't have make command";
10			eval q{
11			system ("cd $test_dir_name; make tree-construction-files") == 0 or die
12			unless -f $dir_name.'tests1.dat';
13			$skip = '';
14			};
15			if ($skip) {
16			print "1..1\n";
17			print "ok 1 # $skip\n";
18			exit;
19			}
20			}
21
22			use Test;
23	wakaba	1.10	BEGIN { plan tests => 472 }
24	wakaba	1.1
25			use Data::Dumper;
26			$Data::Dumper::Useqq = 1;
27			sub Data::Dumper::qquote {
28			my $s = shift;
29			$s =~ s/([^\x20\x21-\x26\x28-\x5B\x5D-\x7E])/sprintf '\x{%02X}', ord $1/ge;
30			return q<qq'> . $s . q<'>;
31			} # Data::Dumper::qquote
32
33	wakaba	1.2	for my $file_name (grep {$_} split /\s+/, qq[
34	wakaba	1.12	${test_dir_name}tokenizer-test-2.dat
35	wakaba	1.2	${dir_name}tests1.dat
36			${dir_name}tests2.dat
37			${dir_name}tests3.dat
38			${dir_name}tests4.dat
39	wakaba	1.11	${dir_name}tests5.dat
40			${dir_name}tests6.dat
41	wakaba	1.2	${test_dir_name}tree-test-1.dat
42	wakaba	1.1	]) {
43	wakaba	1.2	open my $file, '<', $file_name
44			or die "$0: $file_name: $!";
45	wakaba	1.13	print "# $file_name\n";
46	wakaba	1.1
47			my $test;
48			my $mode = 'data';
49	wakaba	1.12	my $escaped;
50	wakaba	1.1	while (<$file>) {
51			s/\x0D\x0A/\x0A/;
52			if (/^#data$/) {
53			undef $test;
54			$test->{data} = '';
55			$mode = 'data';
56	wakaba	1.12	undef $escaped;
57			} elsif (/^#data escaped$/) {
58			undef $test;
59			$test->{data} = '';
60			$mode = 'data';
61			$escaped = 1;
62	wakaba	1.1	} elsif (/^#errors$/) {
63			$test->{errors} = [];
64			$mode = 'errors';
65	wakaba	1.13	$test->{data} =~ s/\x0D?\x0A\z//;
66			$test->{data} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
67	wakaba	1.12	undef $escaped;
68	wakaba	1.1	} elsif (/^#document$/) {
69			$test->{document} = '';
70			$mode = 'document';
71	wakaba	1.12	undef $escaped;
72			} elsif (/^#document escaped$/) {
73			$test->{document} = '';
74			$mode = 'document';
75			$escaped = 1;
76	wakaba	1.5	} elsif (/^#document-fragment (\S+)$/) {
77			$test->{document} = '';
78			$mode = 'document';
79			$test->{element} = $1;
80	wakaba	1.12	undef $escaped;
81			} elsif (/^#document-fragment (\S+) escaped$/) {
82			$test->{document} = '';
83			$mode = 'document';
84			$test->{element} = $1;
85			$escaped = 1;
86	wakaba	1.2	} elsif (defined $test->{document} and /^$/) {
87	wakaba	1.13	$test->{document} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
88	wakaba	1.2	test ($test);
89	wakaba	1.1	undef $test;
90			} else {
91			if ($mode eq 'data' or $mode eq 'document') {
92	wakaba	1.13	$test->{$mode} .= $_;
93	wakaba	1.1	} elsif ($mode eq 'errors') {
94			tr/\x0D\x0A//d;
95			push @{$test->{errors}}, $_;
96			}
97			}
98			}
99			test ($test) if $test->{errors};
100			}
101
102	wakaba	1.4	use Whatpm::HTML;
103			use Whatpm::NanoDOM;
104	wakaba	1.1
105			sub test ($) {
106			my $test = shift;
107
108	wakaba	1.4	my $doc = Whatpm::NanoDOM::Document->new;
109	wakaba	1.1	my @errors;
110
111			$SIG{INT} = sub {
112	wakaba	1.3	print scalar serialize ($doc);
113	wakaba	1.1	exit;
114			};
115	wakaba	1.3
116	wakaba	1.5	my $onerror = sub {
117			my %opt = @_;
118			push @errors, join ':', $opt{line}, $opt{column}, $opt{type};
119			};
120			my $result;
121			unless (defined $test->{element}) {
122			Whatpm::HTML->parse_string ($test->{data} => $doc, $onerror);
123			$result = serialize ($doc);
124			} else {
125			my $el = $doc->create_element_ns
126			('http://www.w3.org/1999/xhtml', [undef, $test->{element}]);
127			Whatpm::HTML->set_inner_html ($el, $test->{data}, $onerror);
128			$result = serialize ($el);
129			}
130
131	wakaba	1.1	ok scalar @errors, scalar @{$test->{errors}},
132			'Parse error: ' . $test->{data} . '; ' .
133			join (', ', @errors) . ';' . join (', ', @{$test->{errors}});
134
135	wakaba	1.5	ok $result, $test->{document}, 'Document tree: ' . $test->{data};
136	wakaba	1.1	} # test
137
138			sub serialize ($) {
139			my $node = shift;
140			my $r = '';
141
142			my @node = map { [$_, ''] } @{$node->child_nodes};
143			while (@node) {
144			my $child = shift @node;
145			my $nt = $child->[0]->node_type;
146			if ($nt == $child->[0]->ELEMENT_NODE) {
147			$r .= '\| ' . $child->[1] . '<' . $child->[0]->tag_name . ">\x0A"; ## ISSUE: case?
148
149	wakaba	1.2	for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value] }
150	wakaba	1.1	@{$child->[0]->attributes}) {
151			$r .= '\| ' . $child->[1] . ' ' . $attr->[0] . '="'; ## ISSUE: case?
152			$r .= $attr->[1] . '"' . "\x0A";
153			}
154
155			unshift @node,
156			map { [$_, $child->[1] . ' '] } @{$child->[0]->child_nodes};
157			} elsif ($nt == $child->[0]->TEXT_NODE) {
158			$r .= '\| ' . $child->[1] . '"' . $child->[0]->data . '"' . "\x0A";
159			} elsif ($nt == $child->[0]->COMMENT_NODE) {
160			$r .= '\| ' . $child->[1] . '<!-- ' . $child->[0]->data . " -->\x0A";
161			} elsif ($nt == $child->[0]->DOCUMENT_TYPE_NODE) {
162			$r .= '\| ' . $child->[1] . '<!DOCTYPE ' . $child->[0]->name . ">\x0A";
163			} else {
164			$r .= '\| ' . $child->[1] . $child->[0]->node_type . "\x0A"; # error
165			}
166			}
167
168			return $r;
169			} # serialize
170
171			## License: Public Domain.
172	wakaba	1.13	## $Date: 2007/06/23 03:53:35 $