Markup/XML/Error.pm


=head1 NAME

SuikaWiki::Markup::XML::Error --- SuikaWiki XML: Error handling module for SuikaWiki::Markup::XML::*

=head1 DESCRIPTION

This module provides the common error and/or warning handling interface
for the SuikaWiki::Markup::XML::* modules.  With this module, proper error
recovering and proper message outputing (eg. output as HTML element,
localized message, etc.) is easily implementable.

This module is part of SuikaWiki XML support.

=cut

package SuikaWiki::Markup::XML::Error;
use strict;
our $VERSION = do{my @r=(q$Revision: 1.3 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};

## Prefixes:
## - 'SYNTAX_': don't match with XML 1.0 EBNF rules
## - 'WFC_':    violation of well-formedness constraint (fatal error)
## - 'VC_':     violation of validity constraint (error)
## - 'NC_':     violation of namespace constraint (error)
## - 'NS_SYNTAX_':      don't match with XML Namespace 1.0 EBNF rules (error)
## - 'FATAL_ERR_':      fatal error specified by XML 1.0 spec
## - 'ERR_':    error specified by XML 1.0 spec
## - 'WARN_':   don't fullfil XML spec's or implementor's recommendation

## Error levels:
## - wfc:       well-formedness (including syntax error)
## - vc:        validity
## - nswfc:     namespace well-formedness
## - nsvc:      namespace validity
## - fatal:     fatal error
## - warn:      warning

my %_Error = (
        ## Forward compatibility error
        SYNTAX_UNSUPPORTED_XML_VERSION  => {
                description     => 'Unsupported XML version (%s)',
                level   => 'wfc',
        },
        ## Syntax errors
        SYNTAX_DATA_OUT_OF_ROOT_ELEMENT => {
                description     => 'Invalid data or markup out of root element',
                level   => 'wfc',
        },
        SYNTAX_DOCTYPE_NAME_NOT_FOUND   => {
                description     => 'Document type name must be specified in the document type declaration',
                level   => 'wfc',
        },
        SYNTAX_DOCTYPE_PID_LITERAL_NOT_FOUND    => {
                description     => 'Minimum literal for the public identifier must follow the keyword PUBLIC',
                level   => 'wfc',
        },
        SYNTAX_DOCTYPE_SYSID_LITERAL_NOT_FOUND  => {
                description     => 'Literal of the system identifier must follow the keyword SYSTEM or the minimum literal of the public identifier in XML',
                level   => 'wfc',
        },
        SYNTAX_END_OF_MARKUP_NOT_FOUND  => {
                description     => sub {
                        my ($o, $err) = @_;
                        my $o_start = $err->{t}->flag ('p_o_start');
                        my $r = $err->{t}->qname;
                        $r = sprintf 'line %d, position %d%s', $o_start->{line},
                                     $o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
                        $r ? $r = '; '.$r : 0;
                        $err->{t} = $err->{t}->node_type;
                        'Markup is not closed (%s'.$r.')';
                },
                level   => 'wfc',
        },
        SYNTAX_END_TAG_NOT_FOUND        => {
                description     => sub {
                        my ($o, $err) = @_;
                        my $o_start = $err->{t}->flag ('p_o_start');
                        my $r = '';
                        $r = sprintf 'line %d, position %d%s', $o_start->{line},
                                     $o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
                        $r ? $r = '; '.$r : 0;
                        $err->{t} = $err->{t}->qname;
                        'End tag of element (type = %s'.$r.') not found';
                },
                level   => 'wfc',
        },
        SYNTAX_INVALID  => {
                description     => 'This type of markup (%s) cannot appear here',
                level   => 'wfc',
        },
        SYNTAX_INVALID_CHAR     => {
                description     => 'Invalid character (%s) at this context',
                level   => 'wfc',
        },
        SYNTAX_INVALID_DOCTYPE_POSITION => {
                description     => 'DOCTYPE declaration must be between xml declaration and the root element',
                level   => 'wfc',
        },
        SYNTAX_INVALID_KEYWORD  => {
                description     => 'Invalid keyword (%s) at this context',
                level   => 'wfc',
        },
        SYNTAX_INVALID_LITERAL  => {
                description     => 'Literal cannot be here ("%s")',
                level   => 'wfc',
        },
        SYNTAX_INVALID_MD       => {
                description     => 'Invalid syntax of markup declaration',
                level   => 'wfc',
        },
        SYNTAX_INVALID_POSITION => {
                description     => sub {
                        my ($o) = shift;
                                'This type of markup (%s) cannot be used '.({
                                document_entity         => 'out of DTD',
                                dtd_internal_subset     => 'in the internal subset of DTD',
                                dtd_external_subset     => 'in the external subset of DTD',
                                external_parsed_entity  => 'in the external parsed entity',
                                general_external_parsed_entity  => 'in the general external parsed entity',
                        }->{$o->{entity_type}||'document_entity'}||'in '.$o->{entity_type})},
                level   => 'wfc',
        },
        SYNTAX_INVALID_PUBID    => {
                description     => 'This public identifier contains at least one invalid character (%s)',
                level   => 'wfc',
        },
        SYNTAX_LEGAL_CHARACTER  => {
                description     => sub {
                        my $r = sprintf 'The character U-%08X is not a legal XML Char',
                                $_[1]->{t};
                        $_[1]->{t} = undef;
                        $r;
                },
                level   => 'wfc',
        },
        SYNTAX_MD_NAME_NOT_FOUND        => {
                description     => 'Name is required by this type of declaration',
                level   => 'wfc',
        },
        SYNTAX_MD_SYSID_NOT_FOUND       => {
                description     => 'System identifier is required by this type of declaration',
                level   => 'wfc',
        },
        SYNTAX_MS_IN_INTERNAL_SUBSET    => {
                description     => 'Marked section cannot be used in the internal subset of the DTD in XML',
                level   => 'wfc',
        },
        SYNTAX_MS_INVALID_STATUS_STRING => {
                description     => 'Invalid string in the status keyword list',
                level   => 'wfc',
        },
        SYNTAX_MS_MULTIPLE_STATUS       => {
                description     => 'Multiple status keyword (%s) cannot be used',
                level   => 'wfc',
        },
        SYNTAX_MS_NO_STATUS_KEYWORD     => {
                description     => 'No status keyword found',
                level   => 'wfc',
        },
        SYNTAX_MS_NON_XML_STATUS        => {
                description     => 'This status keyword (%s) cannot be used in XML',
                level   => 'wfc',
        },
        SYNTAX_MS_UNKNOWN_STATUS        => {
                description     => 'Unknown status keyword (%s) is used',
                level   => 'wfc',
        },
        SYNTAX_PE_NDATA => {
                description     => 'Parameter entity must be a parsed entity',
                level   => 'wfc',
        },
        SYNTAX_ROOT_ELEMENT_NOT_FOUND   => {
                description     => 'There is no root element (type = %s) in this document entity',
                level   => 'wfc',
        },
        SYNTAX_XML_DECLARE      => {
                description     => 'Syntax of XML (or text) declaration is invalid',
                level   => 'wfc',
        },
        SYNTAX_XML_DECLARE_NO_ATTR      => {
                description     => 'XML (or text) declaration has no (valid) pseudo attribute',
                level   => 'wfc',
        },
        SYNTAX_XML_DECLARE_NO_ENCODING_ATTR     => {
                description     => q(Text declaration must have 'encoding' pseudo attribute),
                level   => 'wfc',
        },
        SYNTAX_XML_DECLARE_NO_VERSION_ATTR      => {
                description     => q(XML declaration must have 'version' pseudo attribute),
                level   => 'wfc',
        },
        SYNTAX_XML_DECLARE_POSITION     => {
                description     => 'XML declaration must be at the top of the entity',
                level   => 'wfc',
        },
        SYNTAX_XML_DECLARE_STANDALONE_ATTR      => {
                description     => q(Text declaration cannot have 'standalone' pseudo attribute),
                level   => 'wfc',
        },
        ## Well-formedness error
        WFC_ELEMENT_TYPE_MATCH  => {
                description     => 'End tag (type = %s) does not match with start tag (type = %s)',
                level   => 'wfc',
        },
        WFC_ENTITY_DECLARED     => {
                description     => 'Entity (%s) must be declared before it is referred',
                level   => 'wfc',
        },
        WFC_LEGAL_CHARACTER     => {
                description     => sub {
                        my $r = sprintf 'The character referred (U-%08X) is not a legal XML Char',
                                $_[1]->{t};
                        $_[1]->{t} = undef;
                        $r;
                },
                level   => 'wfc',
        },
        WFC_NO_EXTERNAL_ENTITY_REFERENCE        => {
                description     => 'Attribute value must not contain reference to an external entity',
                level   => 'wfc',
        },
        WFC_NO_LE_IN_ATTRIBUTE_VALUE    => {
                description     => 'Attribute value must not contain a less-than sign (<)',
                level   => 'wfc',
        },
        WFC_NO_RECURSION        => {
                description     => 'Parsed entity (%s) must not contain a recursive reference to itself',
                level   => 'wfc',
        },
        WFC_NO_LT_IN_ATTRIBUTE_VALUE    => {
                description     => 'Replacement text of entity reference in an attribute value literal cannot contain LESS-THAN SIGN (<) itself',
                level   => 'wfc',
        },
        WFC_PARSED_ENTITY       => {
                description     => 'Entity reference (%s) must not refer non-parsed entity',
                level   => 'wfc',
        },
        WFC_PE_IN_INTERNAL_SUBSET       => {
                description     => 'Parameter entity (%s) cannot be referred in markup declaration in internal subset of DTD',
                level   => 'wfc',
        },
        WFC_UNIQUE_ATT_SPEC     => {
                description     => 'Dupulicate attribute specification',
                level   => 'wfc',
        },
        FATAL_ERR_DECODE_IMPL_ERR       => {
                description     => 'Decode error (%s)',
                level   => 'fatal',
        },
        FATAL_ERR_PREDEFINED_ENTITY     => {
                description     => 'Predefined entity (%s := %s) must be declared as of the XML specification defined',
                level   => 'fatal',
        },
        ## Validity error
        VC_ENTITY_DECLARED      => {
                description     => 'Entity %s should (or must to be valid) be declared before it is referred',
                level   => 'vc',
        },
        VC_NOTATION_DECLARED    => {
                description     => 'Notation %s should (or must to be valid) be declared',
                level   => 'vc',
        },
        VC_ROOT_ELEMENT_TYPE    => {
                description     => 'Document type name (%s) and element type name of the root element (%s) should (or must to be valid) match',
                level   => 'vc',
        },
        VC_UNIQUE_NOTATION_NAME => {
                description     => 'Notation %s is already declared',
                level   => 'warn',
        },
        ## Namespace well-formedness error
        NC_PREFIX_NOT_DEFINED   => {
                description     => 'Undeclared namespace prefix (%s) is used',
                level   => 'nswfc',
        },
        NS_SYNTAX_NAME_IS_NCNAME        => {
                description     => 'Name with colon (%s) cannot be used within namespaced XML document',
                level   => 'nswfc',
        },
        ## Namespace validity error
        ## XML (non-fatal) error
        ERR_EXT_ENTITY_NOT_FOUND        => {
                description     => 'External entity (%s == <%s>) cannot be retrived (%s)',
                level   => 'vc',
        },
        ERR_XML_NDATA_REF_IN_ENTITY_VALUE       => {
                description     => 'Unparsed entity (%s) cannot be referred in EntityValue',
                level   => 'warn',      ## Was fatal error but refined by SE Errata
        },
        ERR_XML_SYSID_HAS_FRAGMENT      => {
                description     => 'URI Reference converted from system identifier should not have the fragment identifier (%s)',
                level   => 'warn',
        },
        ## XML warning
        WARN_PREDEFINED_ENTITY_NOT_DECLARED     => {
                description     => 'Predefined general entity (%s) should be declared before it is referred for interoperability',
                level   => 'warn',
        },
        WARN_UNICODE_COMPAT_CHARACTER   => {
                description     => sub {
                        my $r = sprintf 'Use of the character U+%04X is deprecated, since it is classified as compatible character in the Unicode Standard',
                                $_[1]->{t};
                        $_[1]->{t} = undef;
                        $r;
                },
                level   => 'warn',
        },
        WARN_UNICODE_NONCHARACTER       => {
                description     => sub {
                        my $r = sprintf 'Use of the code point U+%04X is deprecated, since it is noncharacter in the Unicode Standard',
                                $_[1]->{t};
                        $_[1]->{t} = undef;
                        $r;
                },
                level   => 'warn',
        },
        WARN_UNICODE_XML_NOT_SUITABLE_CHARACTER => {
                description     => sub {
                        my $r = sprintf 'Use of the character U+%04X is deprecated by W3C Note unicode-xml',
                                $_[1]->{t};
                        $_[1]->{t} = undef;
                        $r;
                },
                level   => 'warn',
        },
        WARN_UNIQUE_ENTITY_NAME => {
                description     => 'General entity %s is already declared',
                level   => 'warn',
        },
        WARN_UNIQUE_PARAMETER_ENTITY_NAME       => {
                description     => 'Parameter entity %s is already declared',
                level   => 'warn',
        },
        ## RFC 3023 'SHOULD'
        WARN_MT_DTD_EXTERNAL_SUBSET     => {
                description     => q(Media type "application/xml-dtd" SHOULD be used for the external subset of the DTD or the external parameter entity),
                level   => 'warn',
        },
        WARN_MT_EXTERNAL_GENERAL_PARSED_ENTITY  => {
                description     => q(Media type "application/xml-external-parsed-entity" SHOULD be used for the external general parsed entity),
                level   => 'warn',
        },
        WARN_MT_XML_FOR_EXT_GENERAL_ENTITY      => {
                description     => 'Using media type %s for external general parsed entity is now forbidden unless the entity is also well-formed as a document entity',
                level   => 'warn',
        },
        ## Implementation's warning
        WARN_DOCTYPE_NOT_FOUND  => {
                description     => 'No document type definition provided for this document',
                level   => 'warn',
        },
        WARN_ENTITY_DECLARATION_NOT_PROCESSED   => {
                description     => 'This entity declaration is not processed because unread parameter entity is referred before this declaration',
                level   => 'warn',
        },
        WARN_EXTERNALLY_DEFINED_ENTITY_REFERRED => {
                description     => 'The entity referred (%s) is declared in the external entity, so different groove can be constructed when external entity is not read',
                level   => 'warn',
        },
        WARN_GUESS_ENCODING_IMPL_ERR    => {
                description     => 'Guessing encoding procedure cause some error (%s)',
                level   => 'warn',
        },
        WARN_INVALID_URI_CHAR_IN_SYSID  => {
                description     => 'System identifier has at least one character that is invalid as part of URI Reference (%s)',
                level   => 'warn',
        },
        WARN_MT_TEXT_XML        => {
                description     => q(In many case, labeling with the media type "text/xml" is inappropriate.  Use "application/xml" or markup language specified type instead),
                level   => 'warn',
        },
        WARN_MT_TEXT_XML_EXTERNAL_PARSED_ENTITY => {
                description     => q(In many case, labeling with the media type "text/xml-external-parsed-entity" is inappropriate.  Use "application/xml-external-parsed-entity" instead),
                level   => 'warn',
        },
        WARN_NO_CHARSET_PARAM   => {    ## charset parameter is optional
                description     => 'Charset parameter should be specified (%s)',
                level   => 'warn',
        },
        WARN_NO_CHARSET_PARAM_FOR_TEXT  => {    ## charset parameter have default value of ascii
                description     => q(Charset parameter is not specified, so default value 'us-ascii' is assumed (%s)),
                level   => 'warn',
        },
        WARN_NO_EXPLICIT_ENCODING_INFO  => {    ## BOM and '<?' guessing is failed (so general encoding guess is to be called)
                description     => q(Neither upper level protocol nor XML's encoding declaration provide encoding scheme information (or cannot read the encoding declaration because of lack of guessability)),
                level   => 'warn',
        },
        WARN_PI_TARGET_NOTATION => {
                description     => 'Target name of the process instruction (%s) should be declared as a notation name to ensure interoperability',
                level   => 'warn',
        },
        WARN_PID_EMPTY  => {
                description     => 'Public identifier is empty',
                level   => 'warn',
        },
        WARN_PID_IS_INVALID_URN => {
                description     => 'Public identifier (%s) seems an invalid URN',
                level   => 'warn',
        },
        WARN_PID_IS_NOT_FPI_NOR_URN     => {
                description     => 'Public identifier (%s) should be a formal public identifier or a uniform resource name to ensure interoperability',
                level   => 'warn',
        },
        WARN_PID_IS_TOO_LONG    => {
                description     => 'Public identifier (%s) should be shorter for interoperability',
                level   => 'warn',
        },
        WARN_PID_IS_URN_WITH_RESERVED_CHAR      => {
                description     => 'Public identifier (%s) seems a URN and it contains one or more reserved character ("/" and/or "?") which should not be used',
                level   => 'warn',
        },
        WARN_SYSID_EMPTY        => {
                description     => 'System identifier is empty, in most case it is wrong',
                level   => 'warn',
        },
        WARN_XML_DECLARE_NO_VERSION_ATTR        => {
                description     => 'Text declaration does not have the version pseudo attribute',
                level   => 'warn',
        },
        ## Misc
        UNKNOWN => {
                description     => 'Unknown error',
                level   => 'fatal',
        },
);
## TODO: error handling should be customizable (hookable) by user of this module
sub raise ($$%) {
  my (undef, $o, %err) = @_;
  my $error_type = $_Error{$err{type}} || $_Error{UNKNOWN};
  my $error_msg = ref $error_type->{description} ? &{$error_type->{description}} ($o, \%err)
                                                 : $error_type->{description};
  my @err_msg;
  ref $err{t} eq 'ARRAY' ? @err_msg = @{$err{t}} : defined $err{t} ? @err_msg = $err{t} : undef;
  $error_msg .= ' (%s)' if scalar (@err_msg) && ($error_msg !~ /%s/);
  $error_msg = sprintf $error_msg, @err_msg;
  $error_msg = "Line $o->{line}, position $o->{pos}: " . $error_msg;
  $error_msg = 'Entity '.($err{entity}||$o->{entity}) . ": " . $error_msg if ($err{entity}||$o->{entity});
  $error_msg = '<'.($o->{uri}) . ">: " . $error_msg ;#if defined $o->{uri};
  require Carp;
  Carp::carp ($error_msg);
  #Carp::croak ("Line $o->{line}, position $o->{pos}: ".$error_msg);
}

=head1 LICENSE

Copyright 2003 Wakaba <[email protected]>

This program is free software; you can redistribute it and/or
modify it under the same terms as Perl itself.

=cut

1; # $Date: 2003/06/27 13:05:57 $
1	wakaba	1.1
2			=head1 NAME
3
4			SuikaWiki::Markup::XML::Error --- SuikaWiki XML: Error handling module for SuikaWiki::Markup::XML::*
5
6			=head1 DESCRIPTION
7
8	wakaba	1.3	This module provides the common error and/or warning handling interface
9			for the SuikaWiki::Markup::XML::* modules. With this module, proper error
10			recovering and proper message outputing (eg. output as HTML element,
11			localized message, etc.) is easily implementable.
12
13	wakaba	1.1	This module is part of SuikaWiki XML support.
14
15			=cut
16
17			package SuikaWiki::Markup::XML::Error;
18			use strict;
19	wakaba	1.4	our $VERSION = do{my @r=(q$Revision: 1.3 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
20	wakaba	1.3
21			## Prefixes:
22			## - 'SYNTAX_': don't match with XML 1.0 EBNF rules
23			## - 'WFC_': violation of well-formedness constraint (fatal error)
24			## - 'VC_': violation of validity constraint (error)
25			## - 'NC_': violation of namespace constraint (error)
26			## - 'NS_SYNTAX_': don't match with XML Namespace 1.0 EBNF rules (error)
27			## - 'FATAL_ERR_': fatal error specified by XML 1.0 spec
28			## - 'ERR_': error specified by XML 1.0 spec
29			## - 'WARN_': don't fullfil XML spec's or implementor's recommendation
30
31			## Error levels:
32			## - wfc: well-formedness (including syntax error)
33			## - vc: validity
34			## - nswfc: namespace well-formedness
35			## - nsvc: namespace validity
36			## - fatal: fatal error
37			## - warn: warning
38	wakaba	1.1
39			my %_Error = (
40	wakaba	1.3	## Forward compatibility error
41			SYNTAX_UNSUPPORTED_XML_VERSION => {
42			description => 'Unsupported XML version (%s)',
43			level => 'wfc',
44			},
45	wakaba	1.1	## Syntax errors
46			SYNTAX_DATA_OUT_OF_ROOT_ELEMENT => {
47			description => 'Invalid data or markup out of root element',
48			level => 'wfc',
49			},
50	wakaba	1.4	SYNTAX_DOCTYPE_NAME_NOT_FOUND => {
51			description => 'Document type name must be specified in the document type declaration',
52			level => 'wfc',
53			},
54			SYNTAX_DOCTYPE_PID_LITERAL_NOT_FOUND => {
55			description => 'Minimum literal for the public identifier must follow the keyword PUBLIC',
56			level => 'wfc',
57			},
58			SYNTAX_DOCTYPE_SYSID_LITERAL_NOT_FOUND => {
59			description => 'Literal of the system identifier must follow the keyword SYSTEM or the minimum literal of the public identifier in XML',
60			level => 'wfc',
61			},
62	wakaba	1.1	SYNTAX_END_OF_MARKUP_NOT_FOUND => {
63			description => sub {
64			my ($o, $err) = @_;
65			my $o_start = $err->{t}->flag ('p_o_start');
66			my $r = $err->{t}->qname;
67			$r = sprintf 'line %d, position %d%s', $o_start->{line},
68			$o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
69			$r ? $r = '; '.$r : 0;
70			$err->{t} = $err->{t}->node_type;
71			'Markup is not closed (%s'.$r.')';
72			},
73			level => 'wfc',
74			},
75			SYNTAX_END_TAG_NOT_FOUND => {
76			description => sub {
77			my ($o, $err) = @_;
78			my $o_start = $err->{t}->flag ('p_o_start');
79			my $r = '';
80			$r = sprintf 'line %d, position %d%s', $o_start->{line},
81			$o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
82			$r ? $r = '; '.$r : 0;
83			$err->{t} = $err->{t}->qname;
84			'End tag of element (type = %s'.$r.') not found';
85			},
86			level => 'wfc',
87			},
88			SYNTAX_INVALID => {
89			description => 'This type of markup (%s) cannot appear here',
90			level => 'wfc',
91			},
92			SYNTAX_INVALID_CHAR => {
93			description => 'Invalid character (%s) at this context',
94			level => 'wfc',
95			},
96			SYNTAX_INVALID_DOCTYPE_POSITION => {
97			description => 'DOCTYPE declaration must be between xml declaration and the root element',
98			level => 'wfc',
99			},
100			SYNTAX_INVALID_KEYWORD => {
101			description => 'Invalid keyword (%s) at this context',
102			level => 'wfc',
103			},
104			SYNTAX_INVALID_LITERAL => {
105			description => 'Literal cannot be here ("%s")',
106			level => 'wfc',
107			},
108			SYNTAX_INVALID_MD => {
109			description => 'Invalid syntax of markup declaration',
110			level => 'wfc',
111			},
112			SYNTAX_INVALID_POSITION => {
113			description => sub {
114			my ($o) = shift;
115			'This type of markup (%s) cannot be used '.({
116			document_entity => 'out of DTD',
117			dtd_internal_subset => 'in the internal subset of DTD',
118			dtd_external_subset => 'in the external subset of DTD',
119			external_parsed_entity => 'in the external parsed entity',
120			general_external_parsed_entity => 'in the general external parsed entity',
121			}->{$o->{entity_type}\|\|'document_entity'}\|\|'in '.$o->{entity_type})},
122			level => 'wfc',
123			},
124	wakaba	1.3	SYNTAX_INVALID_PUBID => {
125			description => 'This public identifier contains at least one invalid character (%s)',
126			level => 'wfc',
127			},
128	wakaba	1.1	SYNTAX_LEGAL_CHARACTER => {
129			description => sub {
130			my $r = sprintf 'The character U-%08X is not a legal XML Char',
131			$_[1]->{t};
132			$_[1]->{t} = undef;
133			$r;
134			},
135			level => 'wfc',
136			},
137			SYNTAX_MD_NAME_NOT_FOUND => {
138			description => 'Name is required by this type of declaration',
139			level => 'wfc',
140			},
141			SYNTAX_MD_SYSID_NOT_FOUND => {
142			description => 'System identifier is required by this type of declaration',
143			level => 'wfc',
144			},
145	wakaba	1.4	SYNTAX_MS_IN_INTERNAL_SUBSET => {
146			description => 'Marked section cannot be used in the internal subset of the DTD in XML',
147			level => 'wfc',
148			},
149			SYNTAX_MS_INVALID_STATUS_STRING => {
150			description => 'Invalid string in the status keyword list',
151			level => 'wfc',
152			},
153			SYNTAX_MS_MULTIPLE_STATUS => {
154			description => 'Multiple status keyword (%s) cannot be used',
155			level => 'wfc',
156			},
157			SYNTAX_MS_NO_STATUS_KEYWORD => {
158			description => 'No status keyword found',
159			level => 'wfc',
160			},
161			SYNTAX_MS_NON_XML_STATUS => {
162			description => 'This status keyword (%s) cannot be used in XML',
163			level => 'wfc',
164			},
165			SYNTAX_MS_UNKNOWN_STATUS => {
166			description => 'Unknown status keyword (%s) is used',
167			level => 'wfc',
168			},
169	wakaba	1.1	SYNTAX_PE_NDATA => {
170			description => 'Parameter entity must be a parsed entity',
171			level => 'wfc',
172			},
173	wakaba	1.4	SYNTAX_ROOT_ELEMENT_NOT_FOUND => {
174			description => 'There is no root element (type = %s) in this document entity',
175			level => 'wfc',
176			},
177	wakaba	1.1	SYNTAX_XML_DECLARE => {
178			description => 'Syntax of XML (or text) declaration is invalid',
179			level => 'wfc',
180			},
181			SYNTAX_XML_DECLARE_NO_ATTR => {
182	wakaba	1.3	description => 'XML (or text) declaration has no (valid) pseudo attribute',
183			level => 'wfc',
184			},
185			SYNTAX_XML_DECLARE_NO_ENCODING_ATTR => {
186			description => q(Text declaration must have 'encoding' pseudo attribute),
187			level => 'wfc',
188			},
189			SYNTAX_XML_DECLARE_NO_VERSION_ATTR => {
190			description => q(XML declaration must have 'version' pseudo attribute),
191	wakaba	1.1	level => 'wfc',
192			},
193			SYNTAX_XML_DECLARE_POSITION => {
194			description => 'XML declaration must be at the top of the entity',
195			level => 'wfc',
196			},
197	wakaba	1.3	SYNTAX_XML_DECLARE_STANDALONE_ATTR => {
198			description => q(Text declaration cannot have 'standalone' pseudo attribute),
199			level => 'wfc',
200			},
201	wakaba	1.1	## Well-formedness error
202			WFC_ELEMENT_TYPE_MATCH => {
203			description => 'End tag (type = %s) does not match with start tag (type = %s)',
204			level => 'wfc',
205			},
206			WFC_ENTITY_DECLARED => {
207			description => 'Entity (%s) must be declared before it is referred',
208			level => 'wfc',
209			},
210			WFC_LEGAL_CHARACTER => {
211			description => sub {
212			my $r = sprintf 'The character referred (U-%08X) is not a legal XML Char',
213			$_[1]->{t};
214			$_[1]->{t} = undef;
215			$r;
216			},
217			level => 'wfc',
218			},
219			WFC_NO_EXTERNAL_ENTITY_REFERENCE => {
220			description => 'Attribute value must not contain reference to an external entity',
221			level => 'wfc',
222			},
223			WFC_NO_LE_IN_ATTRIBUTE_VALUE => {
224			description => 'Attribute value must not contain a less-than sign (<)',
225			level => 'wfc',
226			},
227			WFC_NO_RECURSION => {
228			description => 'Parsed entity (%s) must not contain a recursive reference to itself',
229			level => 'wfc',
230			},
231			WFC_NO_LT_IN_ATTRIBUTE_VALUE => {
232			description => 'Replacement text of entity reference in an attribute value literal cannot contain LESS-THAN SIGN (<) itself',
233			level => 'wfc',
234			},
235	wakaba	1.3	WFC_PARSED_ENTITY => {
236			description => 'Entity reference (%s) must not refer non-parsed entity',
237			level => 'wfc',
238			},
239	wakaba	1.1	WFC_PE_IN_INTERNAL_SUBSET => {
240			description => 'Parameter entity (%s) cannot be referred in markup declaration in internal subset of DTD',
241			level => 'wfc',
242			},
243			WFC_UNIQUE_ATT_SPEC => {
244			description => 'Dupulicate attribute specification',
245			level => 'wfc',
246			},
247	wakaba	1.3	FATAL_ERR_DECODE_IMPL_ERR => {
248			description => 'Decode error (%s)',
249			level => 'fatal',
250			},
251			FATAL_ERR_PREDEFINED_ENTITY => {
252	wakaba	1.4	description => 'Predefined entity (%s := %s) must be declared as of the XML specification defined',
253	wakaba	1.3	level => 'fatal',
254	wakaba	1.2	},
255	wakaba	1.1	## Validity error
256			VC_ENTITY_DECLARED => {
257			description => 'Entity %s should (or must to be valid) be declared before it is referred',
258			level => 'vc',
259			},
260			VC_NOTATION_DECLARED => {
261			description => 'Notation %s should (or must to be valid) be declared',
262			level => 'vc',
263			},
264	wakaba	1.4	VC_ROOT_ELEMENT_TYPE => {
265			description => 'Document type name (%s) and element type name of the root element (%s) should (or must to be valid) match',
266			level => 'vc',
267			},
268	wakaba	1.1	VC_UNIQUE_NOTATION_NAME => {
269			description => 'Notation %s is already declared',
270			level => 'warn',
271			},
272			## Namespace well-formedness error
273			NC_PREFIX_NOT_DEFINED => {
274			description => 'Undeclared namespace prefix (%s) is used',
275			level => 'nswfc',
276			},
277			NS_SYNTAX_NAME_IS_NCNAME => {
278			description => 'Name with colon (%s) cannot be used within namespaced XML document',
279			level => 'nswfc',
280			},
281			## Namespace validity error
282	wakaba	1.4	## XML (non-fatal) error
283	wakaba	1.3	ERR_EXT_ENTITY_NOT_FOUND => {
284			description => 'External entity (%s == <%s>) cannot be retrived (%s)',
285			level => 'vc',
286			},
287	wakaba	1.4	ERR_XML_NDATA_REF_IN_ENTITY_VALUE => {
288			description => 'Unparsed entity (%s) cannot be referred in EntityValue',
289			level => 'warn', ## Was fatal error but refined by SE Errata
290			},
291			ERR_XML_SYSID_HAS_FRAGMENT => {
292	wakaba	1.3	description => 'URI Reference converted from system identifier should not have the fragment identifier (%s)',
293			level => 'warn',
294			},
295	wakaba	1.4	## XML warning
296	wakaba	1.1	WARN_PREDEFINED_ENTITY_NOT_DECLARED => {
297			description => 'Predefined general entity (%s) should be declared before it is referred for interoperability',
298			level => 'warn',
299			},
300			WARN_UNICODE_COMPAT_CHARACTER => {
301			description => sub {
302			my $r = sprintf 'Use of the character U+%04X is deprecated, since it is classified as compatible character in the Unicode Standard',
303			$_[1]->{t};
304			$_[1]->{t} = undef;
305			$r;
306			},
307			level => 'warn',
308			},
309			WARN_UNICODE_NONCHARACTER => {
310			description => sub {
311			my $r = sprintf 'Use of the code point U+%04X is deprecated, since it is noncharacter in the Unicode Standard',
312			$_[1]->{t};
313			$_[1]->{t} = undef;
314			$r;
315			},
316			level => 'warn',
317			},
318			WARN_UNICODE_XML_NOT_SUITABLE_CHARACTER => {
319			description => sub {
320			my $r = sprintf 'Use of the character U+%04X is deprecated by W3C Note unicode-xml',
321			$_[1]->{t};
322			$_[1]->{t} = undef;
323			$r;
324			},
325			level => 'warn',
326			},
327			WARN_UNIQUE_ENTITY_NAME => {
328			description => 'General entity %s is already declared',
329			level => 'warn',
330			},
331			WARN_UNIQUE_PARAMETER_ENTITY_NAME => {
332			description => 'Parameter entity %s is already declared',
333			level => 'warn',
334			},
335	wakaba	1.3	## RFC 3023 'SHOULD'
336			WARN_MT_DTD_EXTERNAL_SUBSET => {
337			description => q(Media type "application/xml-dtd" SHOULD be used for the external subset of the DTD or the external parameter entity),
338			level => 'warn',
339			},
340			WARN_MT_EXTERNAL_GENERAL_PARSED_ENTITY => {
341			description => q(Media type "application/xml-external-parsed-entity" SHOULD be used for the external general parsed entity),
342			level => 'warn',
343			},
344			WARN_MT_XML_FOR_EXT_GENERAL_ENTITY => {
345			description => 'Using media type %s for external general parsed entity is now forbidden unless the entity is also well-formed as a document entity',
346			level => 'warn',
347			},
348	wakaba	1.1	## Implementation's warning
349	wakaba	1.4	WARN_DOCTYPE_NOT_FOUND => {
350			description => 'No document type definition provided for this document',
351			level => 'warn',
352			},
353			WARN_ENTITY_DECLARATION_NOT_PROCESSED => {
354			description => 'This entity declaration is not processed because unread parameter entity is referred before this declaration',
355			level => 'warn',
356			},
357			WARN_EXTERNALLY_DEFINED_ENTITY_REFERRED => {
358			description => 'The entity referred (%s) is declared in the external entity, so different groove can be constructed when external entity is not read',
359			level => 'warn',
360			},
361	wakaba	1.3	WARN_GUESS_ENCODING_IMPL_ERR => {
362			description => 'Guessing encoding procedure cause some error (%s)',
363			level => 'warn',
364			},
365			WARN_INVALID_URI_CHAR_IN_SYSID => {
366			description => 'System identifier has at least one character that is invalid as part of URI Reference (%s)',
367			level => 'warn',
368			},
369			WARN_MT_TEXT_XML => {
370			description => q(In many case, labeling with the media type "text/xml" is inappropriate. Use "application/xml" or markup language specified type instead),
371			level => 'warn',
372			},
373			WARN_MT_TEXT_XML_EXTERNAL_PARSED_ENTITY => {
374			description => q(In many case, labeling with the media type "text/xml-external-parsed-entity" is inappropriate. Use "application/xml-external-parsed-entity" instead),
375			level => 'warn',
376			},
377			WARN_NO_CHARSET_PARAM => { ## charset parameter is optional
378			description => 'Charset parameter should be specified (%s)',
379			level => 'warn',
380			},
381			WARN_NO_CHARSET_PARAM_FOR_TEXT => { ## charset parameter have default value of ascii
382			description => q(Charset parameter is not specified, so default value 'us-ascii' is assumed (%s)),
383			level => 'warn',
384			},
385			WARN_NO_EXPLICIT_ENCODING_INFO => { ## BOM and '<?' guessing is failed (so general encoding guess is to be called)
386			description => q(Neither upper level protocol nor XML's encoding declaration provide encoding scheme information (or cannot read the encoding declaration because of lack of guessability)),
387			level => 'warn',
388			},
389	wakaba	1.4	WARN_PI_TARGET_NOTATION => {
390			description => 'Target name of the process instruction (%s) should be declared as a notation name to ensure interoperability',
391			level => 'warn',
392			},
393			WARN_PID_EMPTY => {
394			description => 'Public identifier is empty',
395			level => 'warn',
396			},
397	wakaba	1.3	WARN_PID_IS_INVALID_URN => {
398			description => 'Public identifier (%s) seems an invalid URN',
399			level => 'warn',
400			},
401			WARN_PID_IS_NOT_FPI_NOR_URN => {
402			description => 'Public identifier (%s) should be a formal public identifier or a uniform resource name to ensure interoperability',
403			level => 'warn',
404			},
405			WARN_PID_IS_TOO_LONG => {
406			description => 'Public identifier (%s) should be shorter for interoperability',
407			level => 'warn',
408			},
409			WARN_PID_IS_URN_WITH_RESERVED_CHAR => {
410			description => 'Public identifier (%s) seems a URN and it contains one or more reserved character ("/" and/or "?") which should not be used',
411			level => 'warn',
412			},
413	wakaba	1.4	WARN_SYSID_EMPTY => {
414			description => 'System identifier is empty, in most case it is wrong',
415			level => 'warn',
416			},
417	wakaba	1.3	WARN_XML_DECLARE_NO_VERSION_ATTR => {
418			description => 'Text declaration does not have the version pseudo attribute',
419			level => 'warn',
420			},
421	wakaba	1.1	## Misc
422			UNKNOWN => {
423			description => 'Unknown error',
424	wakaba	1.3	level => 'fatal',
425	wakaba	1.1	},
426			);
427			## TODO: error handling should be customizable (hookable) by user of this module
428			sub raise ($$%) {
429			my (undef, $o, %err) = @_;
430			my $error_type = $_Error{$err{type}} \|\| $_Error{UNKNOWN};
431			my $error_msg = ref $error_type->{description} ? &{$error_type->{description}} ($o, \%err)
432			: $error_type->{description};
433			my @err_msg;
434			ref $err{t} eq 'ARRAY' ? @err_msg = @{$err{t}} : defined $err{t} ? @err_msg = $err{t} : undef;
435			$error_msg .= ' (%s)' if scalar (@err_msg) && ($error_msg !~ /%s/);
436			$error_msg = sprintf $error_msg, @err_msg;
437			$error_msg = "Line $o->{line}, position $o->{pos}: " . $error_msg;
438			$error_msg = 'Entity '.($err{entity}\|\|$o->{entity}) . ": " . $error_msg if ($err{entity}\|\|$o->{entity});
439	wakaba	1.4	$error_msg = '<'.($o->{uri}) . ">: " . $error_msg ;#if defined $o->{uri};
440	wakaba	1.1	require Carp;
441			Carp::carp ($error_msg);
442			#Carp::croak ("Line $o->{line}, position $o->{pos}: ".$error_msg);
443			}
444
445			=head1 LICENSE
446
447			Copyright 2003 Wakaba <[email protected]>
448
449			This program is free software; you can redistribute it and/or
450			modify it under the same terms as Perl itself.
451
452			=cut
453
454	wakaba	1.4	1; # $Date: 2003/06/27 13:05:57 $