1 |
package WebHACC::Language::XML; |
2 |
use strict; |
3 |
require WebHACC::Language::DOM; |
4 |
push our @ISA, 'WebHACC::Language::DOM'; |
5 |
|
6 |
sub new ($) { |
7 |
return bless {}, shift; |
8 |
} # new |
9 |
|
10 |
sub generate_syntax_error_section ($) { |
11 |
my $self = shift; |
12 |
|
13 |
require Message::DOM::DOMImplementation; |
14 |
require Message::DOM::XMLParserTemp; |
15 |
|
16 |
my $out = $self->output; |
17 |
$out->start_section (role => 'parse-errors'); |
18 |
$out->start_error_list (role => 'parse-errors'); |
19 |
|
20 |
my $input = $self->input; |
21 |
my $result = $self->result; |
22 |
|
23 |
my $onerror = sub { |
24 |
my $err = shift; |
25 |
$result->add_error (line => $err->location->line_number, |
26 |
column => $err->location->column_number, |
27 |
type => 'xml parse error', |
28 |
value => $err->text, |
29 |
level => [ |
30 |
$err->SEVERITY_FATAL_ERROR => 'm', |
31 |
$err->SEVERITY_ERROR => 'm', |
32 |
$err->SEVERITY_WARNING => 's', |
33 |
]->[$err->severity], |
34 |
layer => 'syntax'); |
35 |
return 1; |
36 |
}; |
37 |
|
38 |
my $t = \($input->{s}); |
39 |
if ($input->{is_char_string}) { |
40 |
require Encode; |
41 |
$t = \(Encode::encode ('utf8', $$t)); |
42 |
$input->{charset} = 'utf-8'; |
43 |
} |
44 |
|
45 |
open my $fh, '<', $t; |
46 |
my $dom = Message::DOM::DOMImplementation->new; |
47 |
$self->{structure} = Message::DOM::XMLParserTemp->parse_byte_stream |
48 |
($fh => $dom, $onerror, charset => $input->{charset}); |
49 |
$self->{structure}->manakai_charset ($input->{official_charset}) |
50 |
if defined $input->{official_charset}; |
51 |
|
52 |
$self->{structure}->document_uri ($input->{uri}); |
53 |
$self->{structure}->manakai_entity_base_uri ($input->{base_uri}); |
54 |
|
55 |
$out->end_error_list; |
56 |
$out->end_section; |
57 |
} # generate_syntax_error_section |
58 |
|
59 |
sub source_charset ($) { |
60 |
my $self = shift; |
61 |
return $self->input->{charset} || ($self->{structure}->owner_document || $self->{structure})->input_encoding; |
62 |
## TODO: Can we always use input_encoding? |
63 |
} # source_charset |
64 |
|
65 |
1; |