whatpm/t/ContentChecker.t

#!/usr/bin/perl
use strict;

use Test;
BEGIN { plan tests => 1435 }

my @FILES = qw[
  t/content-model-1.dat
  t/content-model-2.dat
  t/content-model-3.dat
  t/content-model-4.dat
  t/table-1.dat
  t/content-model-atom-1.dat
  t/content-model-atom-2.dat
];

require Whatpm::ContentChecker;

## ISSUE: Currently we require manakai XML parser to test arbitrary XML tree.
use lib qw[/home/wakaba/work/manakai/lib];
require Message::DOM::DOMImplementation;
require Message::DOM::XMLParserTemp;
require Whatpm::HTML;
require Whatpm::NanoDOM;

my $dom = Message::DOM::DOMImplementation->new;

for my $file_name (@FILES) {
  open my $file, '<', $file_name or die "$0: $file_name: $!";
  print "# $file_name\n";

  my $test;
  my $mode = 'data';
  while (<$file>) {
    s/\x0D\x0A/\x0A/;
    if (/^#data$/) {
      undef $test;
      $test->{data} = '';
      $mode = 'data';
      $test->{parse_as} = 'xml';
    } elsif (/^#data html$/) {
      undef $test;
      $test->{data} = '';
      $mode = 'data';
      $test->{parse_as} = 'html';
    } elsif (/^#errors$/) {
      $test->{errors} = [];
      $mode = 'errors';
      $test->{data} =~ s/\x0D?\x0A\z//;       
    } elsif (defined $test->{errors} and /^$/) {
      test ($test);
      undef $test;
    } else {
      if ($mode eq 'data') {
        $test->{$mode} .= $_;
      } elsif ($mode eq 'errors') {
        tr/\x0D\x0A//d;
        push @{$test->{errors}}, $_;
      }
    }
  }
} # @FILES

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

  my $doc;
  if ($test->{parse_as} eq 'xml') {
    open my $fh, '<', \($test->{data});
    $doc = Message::DOM::XMLParserTemp->parse_byte_stream
      ($fh => $dom, sub { }, charset => 'utf-8');
    ## NOTE: There should be no well-formedness error; if there is,
    ## then it is an error of the test case itself.
  } else {
    $doc = Whatpm::NanoDOM::Document->new;
    Whatpm::HTML->parse_string ($test->{data} => $doc);
  }

  my @error;
  Whatpm::ContentChecker->check_element
    ($doc->document_element, sub {
       my %opt = @_;
       push @error, get_node_path ($opt{node}) . ';' . $opt{type} .
           (defined $opt{level} ? ';'.$opt{level} : '');
     });
  
  ok join ("\n", sort {$a cmp $b} @error),
    join ("\n", sort {$a cmp $b} @{$test->{errors}}), $test->{data};
} # test

sub get_node_path ($) {
  my $node = shift;
  my @r;
  while (defined $node) {
    my $rs;
    if ($node->node_type == 1) {
      $rs = $node->manakai_local_name;
      $node = $node->parent_node;
    } elsif ($node->node_type == 2) {
      $rs = '@' . $node->manakai_local_name;
      $node = $node->owner_element;
    } elsif ($node->node_type == 3) {
      $rs = '"' . $node->data . '"';
      $node = $node->parent_node;
    } elsif ($node->node_type == 9) {
      $rs = '';
      $node = $node->parent_node;
    } else {
      $rs = '#' . $node->node_type;
      $node = $node->parent_node;
    }
    unshift @r, $rs;
  }
  return join '/', @r;
} # get_node_path

## License: Public Domain.
## $Date: 2007/09/29 04:45:10 $
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4			use Test;
5	wakaba	1.21	BEGIN { plan tests => 1435 }
6	wakaba	1.1
7	wakaba	1.5	my @FILES = qw[
8			t/content-model-1.dat
9			t/content-model-2.dat
10			t/content-model-3.dat
11	wakaba	1.13	t/content-model-4.dat
12	wakaba	1.15	t/table-1.dat
13	wakaba	1.19	t/content-model-atom-1.dat
14			t/content-model-atom-2.dat
15	wakaba	1.5	];
16	wakaba	1.1
17			require Whatpm::ContentChecker;
18
19			## ISSUE: Currently we require manakai XML parser to test arbitrary XML tree.
20			use lib qw[/home/wakaba/work/manakai/lib];
21	wakaba	1.12	require Message::DOM::DOMImplementation;
22			require Message::DOM::XMLParserTemp;
23	wakaba	1.4	require Whatpm::HTML;
24			require Whatpm::NanoDOM;
25	wakaba	1.1
26	wakaba	1.20	my $dom = Message::DOM::DOMImplementation->new;
27	wakaba	1.1
28			for my $file_name (@FILES) {
29			open my $file, '<', $file_name or die "$0: $file_name: $!";
30	wakaba	1.19	print "# $file_name\n";
31	wakaba	1.1
32			my $test;
33			my $mode = 'data';
34			while (<$file>) {
35			s/\x0D\x0A/\x0A/;
36			if (/^#data$/) {
37			undef $test;
38			$test->{data} = '';
39			$mode = 'data';
40	wakaba	1.4	$test->{parse_as} = 'xml';
41			} elsif (/^#data html$/) {
42			undef $test;
43			$test->{data} = '';
44			$mode = 'data';
45			$test->{parse_as} = 'html';
46	wakaba	1.1	} elsif (/^#errors$/) {
47			$test->{errors} = [];
48			$mode = 'errors';
49			$test->{data} =~ s/\x0D?\x0A\z//;
50			} elsif (defined $test->{errors} and /^$/) {
51			test ($test);
52			undef $test;
53			} else {
54			if ($mode eq 'data') {
55			$test->{$mode} .= $_;
56			} elsif ($mode eq 'errors') {
57			tr/\x0D\x0A//d;
58			push @{$test->{errors}}, $_;
59			}
60			}
61			}
62			} # @FILES
63
64			sub test ($) {
65			my $test = shift;
66
67	wakaba	1.4	my $doc;
68			if ($test->{parse_as} eq 'xml') {
69	wakaba	1.12	open my $fh, '<', \($test->{data});
70			$doc = Message::DOM::XMLParserTemp->parse_byte_stream
71			($fh => $dom, sub { }, charset => 'utf-8');
72	wakaba	1.4	## NOTE: There should be no well-formedness error; if there is,
73			## then it is an error of the test case itself.
74			} else {
75			$doc = Whatpm::NanoDOM::Document->new;
76			Whatpm::HTML->parse_string ($test->{data} => $doc);
77			}
78	wakaba	1.1
79			my @error;
80	wakaba	1.10	Whatpm::ContentChecker->check_element
81	wakaba	1.1	($doc->document_element, sub {
82			my %opt = @_;
83	wakaba	1.14	push @error, get_node_path ($opt{node}) . ';' . $opt{type} .
84			(defined $opt{level} ? ';'.$opt{level} : '');
85	wakaba	1.1	});
86
87			ok join ("\n", sort {$a cmp $b} @error),
88			join ("\n", sort {$a cmp $b} @{$test->{errors}}), $test->{data};
89			} # test
90
91			sub get_node_path ($) {
92			my $node = shift;
93			my @r;
94			while (defined $node) {
95			my $rs;
96			if ($node->node_type == 1) {
97			$rs = $node->manakai_local_name;
98	wakaba	1.4	$node = $node->parent_node;
99	wakaba	1.3	} elsif ($node->node_type == 2) {
100			$rs = '@' . $node->manakai_local_name;
101	wakaba	1.4	$node = $node->owner_element;
102	wakaba	1.1	} elsif ($node->node_type == 3) {
103			$rs = '"' . $node->data . '"';
104	wakaba	1.4	$node = $node->parent_node;
105	wakaba	1.1	} elsif ($node->node_type == 9) {
106			$rs = '';
107	wakaba	1.4	$node = $node->parent_node;
108	wakaba	1.1	} else {
109			$rs = '#' . $node->node_type;
110	wakaba	1.4	$node = $node->parent_node;
111	wakaba	1.1	}
112			unshift @r, $rs;
113			}
114			return join '/', @r;
115			} # get_node_path
116
117			## License: Public Domain.
118	wakaba	1.21	## $Date: 2007/09/29 04:45:10 $