whatpm/t/ContentChecker.t

#!/usr/bin/perl
use strict;

use Test;
BEGIN { plan tests => 873 }

my @FILES = qw[
  t/content-model-1.dat
  t/content-model-2.dat
  t/content-model-3.dat
  t/content-model-4.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: $!";

  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: 2007/06/24 14:24:22 $
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4			use Test;
5	wakaba	1.11	BEGIN { plan tests => 873 }
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	1.5	];
13	wakaba	1.1
14			require Whatpm::ContentChecker;
15
16			## ISSUE: Currently we require manakai XML parser to test arbitrary XML tree.
17			use lib qw[/home/wakaba/work/manakai/lib];
18	wakaba	1.12	require Message::DOM::DOMImplementation;
19			require Message::DOM::XMLParserTemp;
20	wakaba	1.4	require Whatpm::HTML;
21			require Whatpm::NanoDOM;
22	wakaba	1.1
23	wakaba	1.12	my $dom = Message::DOM::DOMImplementation->____new;
24	wakaba	1.1
25			for my $file_name (@FILES) {
26			open my $file, '<', $file_name or die "$0: $file_name: $!";
27
28			my $test;
29			my $mode = 'data';
30			while (<$file>) {
31			s/\x0D\x0A/\x0A/;
32			if (/^#data$/) {
33			undef $test;
34			$test->{data} = '';
35			$mode = 'data';
36	wakaba	1.4	$test->{parse_as} = 'xml';
37			} elsif (/^#data html$/) {
38			undef $test;
39			$test->{data} = '';
40			$mode = 'data';
41			$test->{parse_as} = 'html';
42	wakaba	1.1	} elsif (/^#errors$/) {
43			$test->{errors} = [];
44			$mode = 'errors';
45			$test->{data} =~ s/\x0D?\x0A\z//;
46			} elsif (defined $test->{errors} and /^$/) {
47			test ($test);
48			undef $test;
49			} else {
50			if ($mode eq 'data') {
51			$test->{$mode} .= $_;
52			} elsif ($mode eq 'errors') {
53			tr/\x0D\x0A//d;
54			push @{$test->{errors}}, $_;
55			}
56			}
57			}
58			} # @FILES
59
60			sub test ($) {
61			my $test = shift;
62
63	wakaba	1.4	my $doc;
64			if ($test->{parse_as} eq 'xml') {
65	wakaba	1.12	open my $fh, '<', \($test->{data});
66			$doc = Message::DOM::XMLParserTemp->parse_byte_stream
67			($fh => $dom, sub { }, charset => 'utf-8');
68	wakaba	1.4	## NOTE: There should be no well-formedness error; if there is,
69			## then it is an error of the test case itself.
70			} else {
71			$doc = Whatpm::NanoDOM::Document->new;
72			Whatpm::HTML->parse_string ($test->{data} => $doc);
73			}
74	wakaba	1.1
75			my @error;
76	wakaba	1.10	Whatpm::ContentChecker->check_element
77	wakaba	1.1	($doc->document_element, sub {
78			my %opt = @_;
79	wakaba	1.14	push @error, get_node_path ($opt{node}) . ';' . $opt{type} .
80			(defined $opt{level} ? ';'.$opt{level} : '');
81	wakaba	1.1	});
82
83			ok join ("\n", sort {$a cmp $b} @error),
84			join ("\n", sort {$a cmp $b} @{$test->{errors}}), $test->{data};
85			} # test
86
87			sub get_node_path ($) {
88			my $node = shift;
89			my @r;
90			while (defined $node) {
91			my $rs;
92			if ($node->node_type == 1) {
93			$rs = $node->manakai_local_name;
94	wakaba	1.4	$node = $node->parent_node;
95	wakaba	1.3	} elsif ($node->node_type == 2) {
96			$rs = '@' . $node->manakai_local_name;
97	wakaba	1.4	$node = $node->owner_element;
98	wakaba	1.1	} elsif ($node->node_type == 3) {
99			$rs = '"' . $node->data . '"';
100	wakaba	1.4	$node = $node->parent_node;
101	wakaba	1.1	} elsif ($node->node_type == 9) {
102			$rs = '';
103	wakaba	1.4	$node = $node->parent_node;
104	wakaba	1.1	} else {
105			$rs = '#' . $node->node_type;
106	wakaba	1.4	$node = $node->parent_node;
107	wakaba	1.1	}
108			unshift @r, $rs;
109			}
110			return join '/', @r;
111			} # get_node_path
112
113			## License: Public Domain.
114	wakaba	1.14	## $Date: 2007/06/24 14:24:22 $