whatpm/t/HTML-tree.t

#!/usr/bin/perl
use strict;

my $DEBUG = $ENV{DEBUG};

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 => 1794 }

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";
    }
  }
}

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
                      ${dir_name}tests7.dat
                      ${test_dir_name}tree-test-1.dat
                      ${test_dir_name}tree-test-2.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;
      $test->{data} =~ s/\\U([0-9A-Fa-f]{8})/chr hex $1/ge if $escaped;
      undef $escaped;
    } elsif (/^#document$/) {
      $test->{document} = '';
      $mode = 'document';
      undef $escaped;
    } elsif (/^#document escaped$/) {
      $test->{document} = '';
      $mode = 'document';
      $escaped = 1;
    } elsif (/^#document-fragment$/) {
      $test->{element} = '';
      $mode = 'element';
      undef $escaped;
    } elsif (/^#document-fragment (\S+)$/) {
      $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->{document} =~ s/\\U([0-9A-Fa-f]{8})/chr hex $1/ge if $escaped;
      test ($test);
      undef $test;
    } else {
      if ($mode eq 'data' or $mode eq 'document') {
        $test->{$mode} .= $_;
      } elsif ($mode eq 'element') {
        tr/\x0D\x0A//d;
        $test->{$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 @{$test->{errors}},
    'Parse error: ' . Data::Dumper::qquote ($test->{data}) . '; ' . 
    join (', ', @errors) . ';' . join (', ', @{$test->{errors}});

  ok $result, $test->{document},
      'Document tree: ' . Data::Dumper::qquote ($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;
      my $pubid = $child->[0]->public_id;
      $r .= ' PUBLIC "' . $pubid . '"' if length $pubid;
      my $sysid = $child->[0]->system_id;
      $r .= ' SYSTEM' if not length $pubid and length $sysid;
      $r .= ' "' . $sysid . '"' if length $sysid;
      $r .= ">\x0A";
    } else {
      $r .= '| ' . $child->[1] . $child->[0]->node_type . "\x0A"; # error
    }
  }
  
  return $r;
} # serialize

## License: Public Domain.
## $Date: 2008/03/21 02:39:37 $
1	#!/usr/bin/perl
2	use strict;
3
4	my $DEBUG = $ENV{DEBUG};
5
6	my $dir_name;
7	my $test_dir_name;
8	BEGIN {
9	$test_dir_name = 't/';
10	$dir_name = 't/tree-construction/';
11	my $skip = "You don't have make command";
12	eval q{
13	system ("cd $test_dir_name; make tree-construction-files") == 0 or die
14	unless -f $dir_name.'tests1.dat';
15	$skip = '';
16	};
17	if ($skip) {
18	print "1..1\n";
19	print "ok 1 # $skip\n";
20	exit;
21	}
22	}
23
24	use Test;
25	BEGIN { plan tests => 1794 }
26
27	use Data::Dumper;
28	$Data::Dumper::Useqq = 1;
29	sub Data::Dumper::qquote {
30	my $s = shift;
31	$s =~ s/([^\x20\x21-\x26\x28-\x5B\x5D-\x7E])/sprintf '\x{%02X}', ord $1/ge;
32	return q<qq'> . $s . q<'>;
33	} # Data::Dumper::qquote
34
35
36	if ($DEBUG) {
37	my $not_found = {%{$Whatpm::HTML::Debug::cp or {}}};
38	$Whatpm::HTML::Debug::cp_pass = sub {
39	my $id = shift;
40	delete $not_found->{$id};
41	};
42
43	END {
44	for my $id (sort {$a <=> $b \|\| $a cmp $b} keys %$not_found) {
45	print "# checkpoint $id is not reached\n";
46	}
47	}
48	}
49
50	for my $file_name (grep {$_} split /\s+/, qq[
51	${test_dir_name}tokenizer-test-2.dat
52	${dir_name}tests1.dat
53	${dir_name}tests2.dat
54	${dir_name}tests3.dat
55	${dir_name}tests4.dat
56	${dir_name}tests5.dat
57	${dir_name}tests6.dat
58	${dir_name}tests7.dat
59	${test_dir_name}tree-test-1.dat
60	${test_dir_name}tree-test-2.dat
61	]) {
62	open my $file, '<', $file_name
63	or die "$0: $file_name: $!";
64	print "# $file_name\n";
65
66	my $test;
67	my $mode = 'data';
68	my $escaped;
69	while (<$file>) {
70	s/\x0D\x0A/\x0A/;
71	if (/^#data$/) {
72	undef $test;
73	$test->{data} = '';
74	$mode = 'data';
75	undef $escaped;
76	} elsif (/^#data escaped$/) {
77	undef $test;
78	$test->{data} = '';
79	$mode = 'data';
80	$escaped = 1;
81	} elsif (/^#errors$/) {
82	$test->{errors} = [];
83	$mode = 'errors';
84	$test->{data} =~ s/\x0D?\x0A\z//;
85	$test->{data} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
86	$test->{data} =~ s/\\U([0-9A-Fa-f]{8})/chr hex $1/ge if $escaped;
87	undef $escaped;
88	} elsif (/^#document$/) {
89	$test->{document} = '';
90	$mode = 'document';
91	undef $escaped;
92	} elsif (/^#document escaped$/) {
93	$test->{document} = '';
94	$mode = 'document';
95	$escaped = 1;
96	} elsif (/^#document-fragment$/) {
97	$test->{element} = '';
98	$mode = 'element';
99	undef $escaped;
100	} elsif (/^#document-fragment (\S+)$/) {
101	$test->{document} = '';
102	$mode = 'document';
103	$test->{element} = $1;
104	undef $escaped;
105	} elsif (/^#document-fragment (\S+) escaped$/) {
106	$test->{document} = '';
107	$mode = 'document';
108	$test->{element} = $1;
109	$escaped = 1;
110	} elsif (defined $test->{document} and /^$/) {
111	$test->{document} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
112	$test->{document} =~ s/\\U([0-9A-Fa-f]{8})/chr hex $1/ge if $escaped;
113	test ($test);
114	undef $test;
115	} else {
116	if ($mode eq 'data' or $mode eq 'document') {
117	$test->{$mode} .= $_;
118	} elsif ($mode eq 'element') {
119	tr/\x0D\x0A//d;
120	$test->{$mode} .= $_;
121	} elsif ($mode eq 'errors') {
122	tr/\x0D\x0A//d;
123	push @{$test->{errors}}, $_;
124	}
125	}
126	}
127	test ($test) if $test->{errors};
128	}
129
130	use Whatpm::HTML;
131	use Whatpm::NanoDOM;
132
133	sub test ($) {
134	my $test = shift;
135
136	my $doc = Whatpm::NanoDOM::Document->new;
137	my @errors;
138
139	$SIG{INT} = sub {
140	print scalar serialize ($doc);
141	exit;
142	};
143
144	my $onerror = sub {
145	my %opt = @_;
146	push @errors, join ':', $opt{line}, $opt{column}, $opt{type};
147	};
148	my $result;
149	unless (defined $test->{element}) {
150	Whatpm::HTML->parse_string ($test->{data} => $doc, $onerror);
151	$result = serialize ($doc);
152	} else {
153	my $el = $doc->create_element_ns
154	('http://www.w3.org/1999/xhtml', [undef, $test->{element}]);
155	Whatpm::HTML->set_inner_html ($el, $test->{data}, $onerror);
156	$result = serialize ($el);
157	}
158
159	ok scalar @errors, scalar @{$test->{errors}},
160	'Parse error: ' . Data::Dumper::qquote ($test->{data}) . '; ' .
161	join (', ', @errors) . ';' . join (', ', @{$test->{errors}});
162
163	ok $result, $test->{document},
164	'Document tree: ' . Data::Dumper::qquote ($test->{data});
165	} # test
166
167	sub serialize ($) {
168	my $node = shift;
169	my $r = '';
170
171	my @node = map { [$_, ''] } @{$node->child_nodes};
172	while (@node) {
173	my $child = shift @node;
174	my $nt = $child->[0]->node_type;
175	if ($nt == $child->[0]->ELEMENT_NODE) {
176	$r .= '\| ' . $child->[1] . '<' . $child->[0]->tag_name . ">\x0A"; ## ISSUE: case?
177
178	for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value] }
179	@{$child->[0]->attributes}) {
180	$r .= '\| ' . $child->[1] . ' ' . $attr->[0] . '="'; ## ISSUE: case?
181	$r .= $attr->[1] . '"' . "\x0A";
182	}
183
184	unshift @node,
185	map { [$_, $child->[1] . ' '] } @{$child->[0]->child_nodes};
186	} elsif ($nt == $child->[0]->TEXT_NODE) {
187	$r .= '\| ' . $child->[1] . '"' . $child->[0]->data . '"' . "\x0A";
188	} elsif ($nt == $child->[0]->COMMENT_NODE) {
189	$r .= '\| ' . $child->[1] . '<!-- ' . $child->[0]->data . " -->\x0A";
190	} elsif ($nt == $child->[0]->DOCUMENT_TYPE_NODE) {
191	$r .= '\| ' . $child->[1] . '<!DOCTYPE ' . $child->[0]->name;
192	my $pubid = $child->[0]->public_id;
193	$r .= ' PUBLIC "' . $pubid . '"' if length $pubid;
194	my $sysid = $child->[0]->system_id;
195	$r .= ' SYSTEM' if not length $pubid and length $sysid;
196	$r .= ' "' . $sysid . '"' if length $sysid;
197	$r .= ">\x0A";
198	} else {
199	$r .= '\| ' . $child->[1] . $child->[0]->node_type . "\x0A"; # error
200	}
201	}
202
203	return $r;
204	} # serialize
205
206	## License: Public Domain.
207	## $Date: 2008/03/21 02:39:37 $