| 2 |
use strict; |
use strict; |
| 3 |
|
|
| 4 |
my $dir_name; |
my $dir_name; |
| 5 |
|
my $test_dir_name; |
| 6 |
BEGIN { |
BEGIN { |
| 7 |
my $test_dir_name = 't/'; |
$test_dir_name = 't/'; |
| 8 |
$dir_name = 't/tokenizer/'; |
$dir_name = 't/tokenizer/'; |
| 9 |
my $skip = "You don't have JSON module"; |
my $skip = "You don't have JSON module"; |
| 10 |
eval q{ |
eval q{ |
| 24 |
} |
} |
| 25 |
|
|
| 26 |
use Test; |
use Test; |
| 27 |
BEGIN { plan tests => 67 } |
BEGIN { plan tests => 94 } |
| 28 |
|
|
| 29 |
use Data::Dumper; |
use Data::Dumper; |
| 30 |
$Data::Dumper::Useqq = 1; |
$Data::Dumper::Useqq = 1; |
| 34 |
return q<qq'> . $s . q<'>; |
return q<qq'> . $s . q<'>; |
| 35 |
} # Data::Dumper::qquote |
} # Data::Dumper::qquote |
| 36 |
|
|
| 37 |
use What::HTML; |
use Whatpm::HTML; |
| 38 |
|
|
| 39 |
for my $file_name (qw[ |
for my $file_name (grep {$_} split /\s+/, qq[ |
| 40 |
test1.test |
${dir_name}test1.test |
| 41 |
test2.test |
${dir_name}test2.test |
| 42 |
contentModelFlags.test |
${dir_name}contentModelFlags.test |
| 43 |
|
${dir_name}escapeFlag.test |
| 44 |
|
${test_dir_name}tokenizer-test-1.test |
| 45 |
]) { |
]) { |
| 46 |
open my $file, '<', $dir_name.$file_name |
open my $file, '<', $file_name |
| 47 |
or die "$0: $dir_name$file_name: $!"; |
or die "$0: $file_name: $!"; |
| 48 |
local $/ = undef; |
local $/ = undef; |
| 49 |
my $js = <$file>; |
my $js = <$file>; |
| 50 |
close $file; |
close $file; |
| 51 |
|
|
| 52 |
|
print "# $file_name\n"; |
| 53 |
my $tests = jsonToObj ($js)->{tests}; |
my $tests = jsonToObj ($js)->{tests}; |
| 54 |
TEST: for my $test (@$tests) { |
TEST: for my $test (@$tests) { |
| 55 |
my $s = $test->{input}; |
my $s = $test->{input}; |
| 70 |
my @cm = @{$test->{contentModelFlags} || ['PCDATA']}; |
my @cm = @{$test->{contentModelFlags} || ['PCDATA']}; |
| 71 |
my $last_start_tag = $test->{lastStartTag}; |
my $last_start_tag = $test->{lastStartTag}; |
| 72 |
for my $cm (@cm) { |
for my $cm (@cm) { |
| 73 |
my $p = What::HTML->new; |
my $p = Whatpm::HTML->new; |
| 74 |
my $i = 0; |
my $i = 0; |
| 75 |
|
my @token; |
| 76 |
$p->{set_next_input_character} = sub { |
$p->{set_next_input_character} = sub { |
| 77 |
my $self = shift; |
my $self = shift; |
| 78 |
$self->{next_input_character} = -1 and return if $i >= length $s; |
$self->{next_input_character} = -1 and return if $i >= length $s; |
| 92 |
$self->{next_input_character} = 0x000A; # LF # MUST |
$self->{next_input_character} = 0x000A; # LF # MUST |
| 93 |
} elsif ($self->{next_input_character} > 0x10FFFF) { |
} elsif ($self->{next_input_character} > 0x10FFFF) { |
| 94 |
$self->{next_input_character} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
$self->{next_input_character} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
| 95 |
|
push @token, 'ParseError'; |
| 96 |
} elsif ($self->{next_input_character} == 0x0000) { # NULL |
} elsif ($self->{next_input_character} == 0x0000) { # NULL |
| 97 |
$self->{next_input_character} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
$self->{next_input_character} = 0xFFFD; # REPLACEMENT CHARACTER # MUST |
| 98 |
|
push @token, 'ParseError'; |
| 99 |
} |
} |
| 100 |
}; |
}; |
| 101 |
|
|
|
my @token; |
|
| 102 |
$p->{parse_error} = sub { |
$p->{parse_error} = sub { |
| 103 |
push @token, 'ParseError'; |
push @token, 'ParseError'; |
| 104 |
}; |
}; |