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 => 2036 }

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
                      ${test_dir_name}tree-test-3.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: 2008/04/13 10:36:41 $
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4	wakaba	1.23	my $DEBUG = $ENV{DEBUG};
5
6	wakaba	1.1	my $dir_name;
7	wakaba	1.2	my $test_dir_name;
8	wakaba	1.1	BEGIN {
9	wakaba	1.2	$test_dir_name = 't/';
10	wakaba	1.1	$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	wakaba	1.29	BEGIN { plan tests => 2036 }
26	wakaba	1.1
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'> . $s . q<'>;
33			} # Data::Dumper::qquote
34
35	wakaba	1.23
36			if ($DEBUG) {
37	wakaba	1.25	my $not_found = {%{$Whatpm::HTML::Debug::cp or {}}};
38	wakaba	1.23	$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	wakaba	1.2	for my $file_name (grep {$_} split /\s+/, qq[
51	wakaba	1.12	${test_dir_name}tokenizer-test-2.dat
52	wakaba	1.2	${dir_name}tests1.dat
53			${dir_name}tests2.dat
54			${dir_name}tests3.dat
55			${dir_name}tests4.dat
56	wakaba	1.11	${dir_name}tests5.dat
57			${dir_name}tests6.dat
58	wakaba	1.27	${dir_name}tests7.dat
59	wakaba	1.2	${test_dir_name}tree-test-1.dat
60	wakaba	1.14	${test_dir_name}tree-test-2.dat
61	wakaba	1.29	${test_dir_name}tree-test-3.dat
62	wakaba	1.1	]) {
63	wakaba	1.2	open my $file, '<', $file_name
64			or die "$0: $file_name: $!";
65	wakaba	1.13	print "# $file_name\n";
66	wakaba	1.1
67			my $test;
68			my $mode = 'data';
69	wakaba	1.12	my $escaped;
70	wakaba	1.1	while (<$file>) {
71			s/\x0D\x0A/\x0A/;
72			if (/^#data$/) {
73			undef $test;
74			$test->{data} = '';
75			$mode = 'data';
76	wakaba	1.12	undef $escaped;
77			} elsif (/^#data escaped$/) {
78			undef $test;
79			$test->{data} = '';
80			$mode = 'data';
81			$escaped = 1;
82	wakaba	1.1	} elsif (/^#errors$/) {
83			$test->{errors} = [];
84			$mode = 'errors';
85	wakaba	1.13	$test->{data} =~ s/\x0D?\x0A\z//;
86			$test->{data} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
87	wakaba	1.24	$test->{data} =~ s/\\U([0-9A-Fa-f]{8})/chr hex $1/ge if $escaped;
88	wakaba	1.12	undef $escaped;
89	wakaba	1.1	} elsif (/^#document$/) {
90			$test->{document} = '';
91			$mode = 'document';
92	wakaba	1.12	undef $escaped;
93			} elsif (/^#document escaped$/) {
94			$test->{document} = '';
95			$mode = 'document';
96			$escaped = 1;
97	wakaba	1.19	} elsif (/^#document-fragment$/) {
98			$test->{element} = '';
99			$mode = 'element';
100			undef $escaped;
101	wakaba	1.5	} elsif (/^#document-fragment (\S+)$/) {
102			$test->{document} = '';
103			$mode = 'document';
104			$test->{element} = $1;
105	wakaba	1.12	undef $escaped;
106			} elsif (/^#document-fragment (\S+) escaped$/) {
107			$test->{document} = '';
108			$mode = 'document';
109			$test->{element} = $1;
110			$escaped = 1;
111	wakaba	1.2	} elsif (defined $test->{document} and /^$/) {
112	wakaba	1.13	$test->{document} =~ s/\\u([0-9A-Fa-f]{4})/chr hex $1/ge if $escaped;
113	wakaba	1.24	$test->{document} =~ s/\\U([0-9A-Fa-f]{8})/chr hex $1/ge if $escaped;
114	wakaba	1.2	test ($test);
115	wakaba	1.1	undef $test;
116			} else {
117			if ($mode eq 'data' or $mode eq 'document') {
118	wakaba	1.13	$test->{$mode} .= $_;
119	wakaba	1.19	} elsif ($mode eq 'element') {
120			tr/\x0D\x0A//d;
121			$test->{$mode} .= $_;
122	wakaba	1.1	} elsif ($mode eq 'errors') {
123			tr/\x0D\x0A//d;
124			push @{$test->{errors}}, $_;
125			}
126			}
127			}
128			test ($test) if $test->{errors};
129			}
130
131	wakaba	1.4	use Whatpm::HTML;
132			use Whatpm::NanoDOM;
133	wakaba	1.1
134			sub test ($) {
135			my $test = shift;
136
137	wakaba	1.4	my $doc = Whatpm::NanoDOM::Document->new;
138	wakaba	1.1	my @errors;
139
140			$SIG{INT} = sub {
141	wakaba	1.3	print scalar serialize ($doc);
142	wakaba	1.1	exit;
143			};
144	wakaba	1.3
145	wakaba	1.5	my $onerror = sub {
146			my %opt = @_;
147			push @errors, join ':', $opt{line}, $opt{column}, $opt{type};
148			};
149			my $result;
150			unless (defined $test->{element}) {
151			Whatpm::HTML->parse_string ($test->{data} => $doc, $onerror);
152			$result = serialize ($doc);
153			} else {
154			my $el = $doc->create_element_ns
155			('http://www.w3.org/1999/xhtml', [undef, $test->{element}]);
156			Whatpm::HTML->set_inner_html ($el, $test->{data}, $onerror);
157			$result = serialize ($el);
158			}
159
160	wakaba	1.1	ok scalar @errors, scalar @{$test->{errors}},
161	wakaba	1.21	'Parse error: ' . Data::Dumper::qquote ($test->{data}) . '; ' .
162	wakaba	1.1	join (', ', @errors) . ';' . join (', ', @{$test->{errors}});
163
164	wakaba	1.21	ok $result, $test->{document},
165			'Document tree: ' . Data::Dumper::qquote ($test->{data});
166	wakaba	1.1	} # test
167
168			sub serialize ($) {
169			my $node = shift;
170			my $r = '';
171
172			my @node = map { [$_, ''] } @{$node->child_nodes};
173			while (@node) {
174			my $child = shift @node;
175			my $nt = $child->[0]->node_type;
176			if ($nt == $child->[0]->ELEMENT_NODE) {
177			$r .= '\| ' . $child->[1] . '<' . $child->[0]->tag_name . ">\x0A"; ## ISSUE: case?
178
179	wakaba	1.2	for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value] }
180	wakaba	1.1	@{$child->[0]->attributes}) {
181			$r .= '\| ' . $child->[1] . ' ' . $attr->[0] . '="'; ## ISSUE: case?
182			$r .= $attr->[1] . '"' . "\x0A";
183			}
184
185			unshift @node,
186			map { [$_, $child->[1] . ' '] } @{$child->[0]->child_nodes};
187			} elsif ($nt == $child->[0]->TEXT_NODE) {
188			$r .= '\| ' . $child->[1] . '"' . $child->[0]->data . '"' . "\x0A";
189			} elsif ($nt == $child->[0]->COMMENT_NODE) {
190			$r .= '\| ' . $child->[1] . '<!-- ' . $child->[0]->data . " -->\x0A";
191			} elsif ($nt == $child->[0]->DOCUMENT_TYPE_NODE) {
192	wakaba	1.24	$r .= '\| ' . $child->[1] . '<!DOCTYPE ' . $child->[0]->name;
193			my $pubid = $child->[0]->public_id;
194			$r .= ' PUBLIC "' . $pubid . '"' if length $pubid;
195			my $sysid = $child->[0]->system_id;
196			$r .= ' SYSTEM' if not length $pubid and length $sysid;
197			$r .= ' "' . $sysid . '"' if length $sysid;
198			$r .= ">\x0A";
199	wakaba	1.1	} else {
200			$r .= '\| ' . $child->[1] . $child->[0]->node_type . "\x0A"; # error
201			}
202			}
203
204			return $r;
205			} # serialize
206
207			## License: Public Domain.
208	wakaba	1.29	## $Date: 2008/04/13 10:36:41 $