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: 1.4 $=~/\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_ATTR_LITERAL_NOT_FOUND   => {
                description     => 'Attribute value literal of the attribute (name = %s) is expected',
                level   => 'wfc',
        },
        SYNTAX_ATTR_NAME_OMITTED        => {
                description     => 'Attribute name corresponding to the value (%s) must be specified in XML',
                level   => 'wfc',
        },
        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_TAG_NOT_CLOSED   => {
                description     => 'Tag must be closed in XML',
                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
        NS_SYNTAX_LNAME_IS_NCNAME       => {
                description     => 'Character just after the colon (:) in QName (%s) must be one of NameStartChar in namespaced XML document',
                level   => 'nswfc',
        },
        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 here in namespaced XML document',
                level   => 'nswfc',
        },
        NS_SYNTAX_NAME_IS_QNAME => {
                description     => 'Name with colon (%s) must match with QName in namespaced XML document',
                level   => 'nswfc',
        },
        NC_UNIQUE_ATT_SPEC      => {
                description     => 'Dupulicate attribute specification (%s == <%s>:%s)',
                level   => 'wfc',
        },
        ## 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',
        },
        ## XMLName recommendation
        WARN_XML_NS_URI_IS_RELATIVE     => {
                description     => 'URI of XML Namespace name is a relative URI',
                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_NS_NAME        => {
                description     => 'URI of XML Namespace name contains at least one non-URI character',
                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: 2003/06/29 08:34:37 $
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	1.5	our $VERSION = do{my @r=(q$Revision: 1.4 $=~/\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	wakaba	1.5	SYNTAX_ATTR_LITERAL_NOT_FOUND => {
47			description => 'Attribute value literal of the attribute (name = %s) is expected',
48			level => 'wfc',
49			},
50			SYNTAX_ATTR_NAME_OMITTED => {
51			description => 'Attribute name corresponding to the value (%s) must be specified in XML',
52			level => 'wfc',
53			},
54	wakaba	1.1	SYNTAX_DATA_OUT_OF_ROOT_ELEMENT => {
55			description => 'Invalid data or markup out of root element',
56			level => 'wfc',
57			},
58	wakaba	1.4	SYNTAX_DOCTYPE_NAME_NOT_FOUND => {
59			description => 'Document type name must be specified in the document type declaration',
60			level => 'wfc',
61			},
62			SYNTAX_DOCTYPE_PID_LITERAL_NOT_FOUND => {
63			description => 'Minimum literal for the public identifier must follow the keyword PUBLIC',
64			level => 'wfc',
65			},
66			SYNTAX_DOCTYPE_SYSID_LITERAL_NOT_FOUND => {
67			description => 'Literal of the system identifier must follow the keyword SYSTEM or the minimum literal of the public identifier in XML',
68			level => 'wfc',
69			},
70	wakaba	1.1	SYNTAX_END_OF_MARKUP_NOT_FOUND => {
71			description => sub {
72			my ($o, $err) = @_;
73			my $o_start = $err->{t}->flag ('p_o_start');
74			my $r = $err->{t}->qname;
75			$r = sprintf 'line %d, position %d%s', $o_start->{line},
76			$o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
77			$r ? $r = '; '.$r : 0;
78			$err->{t} = $err->{t}->node_type;
79			'Markup is not closed (%s'.$r.')';
80			},
81			level => 'wfc',
82			},
83			SYNTAX_END_TAG_NOT_FOUND => {
84			description => sub {
85			my ($o, $err) = @_;
86			my $o_start = $err->{t}->flag ('p_o_start');
87			my $r = '';
88			$r = sprintf 'line %d, position %d%s', $o_start->{line},
89			$o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
90			$r ? $r = '; '.$r : 0;
91			$err->{t} = $err->{t}->qname;
92			'End tag of element (type = %s'.$r.') not found';
93			},
94			level => 'wfc',
95			},
96			SYNTAX_INVALID => {
97			description => 'This type of markup (%s) cannot appear here',
98			level => 'wfc',
99			},
100			SYNTAX_INVALID_CHAR => {
101			description => 'Invalid character (%s) at this context',
102			level => 'wfc',
103			},
104			SYNTAX_INVALID_DOCTYPE_POSITION => {
105			description => 'DOCTYPE declaration must be between xml declaration and the root element',
106			level => 'wfc',
107			},
108			SYNTAX_INVALID_KEYWORD => {
109			description => 'Invalid keyword (%s) at this context',
110			level => 'wfc',
111			},
112			SYNTAX_INVALID_LITERAL => {
113			description => 'Literal cannot be here ("%s")',
114			level => 'wfc',
115			},
116			SYNTAX_INVALID_MD => {
117			description => 'Invalid syntax of markup declaration',
118			level => 'wfc',
119			},
120			SYNTAX_INVALID_POSITION => {
121			description => sub {
122			my ($o) = shift;
123			'This type of markup (%s) cannot be used '.({
124			document_entity => 'out of DTD',
125			dtd_internal_subset => 'in the internal subset of DTD',
126			dtd_external_subset => 'in the external subset of DTD',
127			external_parsed_entity => 'in the external parsed entity',
128			general_external_parsed_entity => 'in the general external parsed entity',
129			}->{$o->{entity_type}\|\|'document_entity'}\|\|'in '.$o->{entity_type})},
130			level => 'wfc',
131			},
132	wakaba	1.3	SYNTAX_INVALID_PUBID => {
133			description => 'This public identifier contains at least one invalid character (%s)',
134			level => 'wfc',
135			},
136	wakaba	1.1	SYNTAX_LEGAL_CHARACTER => {
137			description => sub {
138			my $r = sprintf 'The character U-%08X is not a legal XML Char',
139			$_[1]->{t};
140			$_[1]->{t} = undef;
141			$r;
142			},
143			level => 'wfc',
144			},
145			SYNTAX_MD_NAME_NOT_FOUND => {
146			description => 'Name is required by this type of declaration',
147			level => 'wfc',
148			},
149			SYNTAX_MD_SYSID_NOT_FOUND => {
150			description => 'System identifier is required by this type of declaration',
151			level => 'wfc',
152			},
153	wakaba	1.4	SYNTAX_MS_IN_INTERNAL_SUBSET => {
154			description => 'Marked section cannot be used in the internal subset of the DTD in XML',
155			level => 'wfc',
156			},
157			SYNTAX_MS_INVALID_STATUS_STRING => {
158			description => 'Invalid string in the status keyword list',
159			level => 'wfc',
160			},
161			SYNTAX_MS_MULTIPLE_STATUS => {
162			description => 'Multiple status keyword (%s) cannot be used',
163			level => 'wfc',
164			},
165			SYNTAX_MS_NO_STATUS_KEYWORD => {
166			description => 'No status keyword found',
167			level => 'wfc',
168			},
169			SYNTAX_MS_NON_XML_STATUS => {
170			description => 'This status keyword (%s) cannot be used in XML',
171			level => 'wfc',
172			},
173			SYNTAX_MS_UNKNOWN_STATUS => {
174			description => 'Unknown status keyword (%s) is used',
175			level => 'wfc',
176			},
177	wakaba	1.1	SYNTAX_PE_NDATA => {
178			description => 'Parameter entity must be a parsed entity',
179			level => 'wfc',
180			},
181	wakaba	1.4	SYNTAX_ROOT_ELEMENT_NOT_FOUND => {
182			description => 'There is no root element (type = %s) in this document entity',
183			level => 'wfc',
184			},
185	wakaba	1.5	SYNTAX_TAG_NOT_CLOSED => {
186			description => 'Tag must be closed in XML',
187			level => 'wfc',
188			},
189	wakaba	1.1	SYNTAX_XML_DECLARE => {
190			description => 'Syntax of XML (or text) declaration is invalid',
191			level => 'wfc',
192			},
193			SYNTAX_XML_DECLARE_NO_ATTR => {
194	wakaba	1.3	description => 'XML (or text) declaration has no (valid) pseudo attribute',
195			level => 'wfc',
196			},
197			SYNTAX_XML_DECLARE_NO_ENCODING_ATTR => {
198			description => q(Text declaration must have 'encoding' pseudo attribute),
199			level => 'wfc',
200			},
201			SYNTAX_XML_DECLARE_NO_VERSION_ATTR => {
202			description => q(XML declaration must have 'version' pseudo attribute),
203	wakaba	1.1	level => 'wfc',
204			},
205			SYNTAX_XML_DECLARE_POSITION => {
206			description => 'XML declaration must be at the top of the entity',
207			level => 'wfc',
208			},
209	wakaba	1.3	SYNTAX_XML_DECLARE_STANDALONE_ATTR => {
210			description => q(Text declaration cannot have 'standalone' pseudo attribute),
211			level => 'wfc',
212			},
213	wakaba	1.1	## Well-formedness error
214			WFC_ELEMENT_TYPE_MATCH => {
215			description => 'End tag (type = %s) does not match with start tag (type = %s)',
216			level => 'wfc',
217			},
218			WFC_ENTITY_DECLARED => {
219			description => 'Entity (%s) must be declared before it is referred',
220			level => 'wfc',
221			},
222			WFC_LEGAL_CHARACTER => {
223			description => sub {
224			my $r = sprintf 'The character referred (U-%08X) is not a legal XML Char',
225			$_[1]->{t};
226			$_[1]->{t} = undef;
227			$r;
228			},
229			level => 'wfc',
230			},
231			WFC_NO_EXTERNAL_ENTITY_REFERENCE => {
232			description => 'Attribute value must not contain reference to an external entity',
233			level => 'wfc',
234			},
235			WFC_NO_LE_IN_ATTRIBUTE_VALUE => {
236			description => 'Attribute value must not contain a less-than sign (<)',
237			level => 'wfc',
238			},
239			WFC_NO_RECURSION => {
240			description => 'Parsed entity (%s) must not contain a recursive reference to itself',
241			level => 'wfc',
242			},
243			WFC_NO_LT_IN_ATTRIBUTE_VALUE => {
244			description => 'Replacement text of entity reference in an attribute value literal cannot contain LESS-THAN SIGN (<) itself',
245			level => 'wfc',
246			},
247	wakaba	1.3	WFC_PARSED_ENTITY => {
248			description => 'Entity reference (%s) must not refer non-parsed entity',
249			level => 'wfc',
250			},
251	wakaba	1.1	WFC_PE_IN_INTERNAL_SUBSET => {
252			description => 'Parameter entity (%s) cannot be referred in markup declaration in internal subset of DTD',
253			level => 'wfc',
254			},
255			WFC_UNIQUE_ATT_SPEC => {
256			description => 'Dupulicate attribute specification',
257			level => 'wfc',
258			},
259	wakaba	1.3	FATAL_ERR_DECODE_IMPL_ERR => {
260			description => 'Decode error (%s)',
261			level => 'fatal',
262			},
263			FATAL_ERR_PREDEFINED_ENTITY => {
264	wakaba	1.4	description => 'Predefined entity (%s := %s) must be declared as of the XML specification defined',
265	wakaba	1.3	level => 'fatal',
266	wakaba	1.2	},
267	wakaba	1.1	## Validity error
268			VC_ENTITY_DECLARED => {
269			description => 'Entity %s should (or must to be valid) be declared before it is referred',
270			level => 'vc',
271			},
272			VC_NOTATION_DECLARED => {
273			description => 'Notation %s should (or must to be valid) be declared',
274			level => 'vc',
275			},
276	wakaba	1.4	VC_ROOT_ELEMENT_TYPE => {
277			description => 'Document type name (%s) and element type name of the root element (%s) should (or must to be valid) match',
278			level => 'vc',
279			},
280	wakaba	1.1	VC_UNIQUE_NOTATION_NAME => {
281			description => 'Notation %s is already declared',
282			level => 'warn',
283			},
284			## Namespace well-formedness error
285	wakaba	1.5	NS_SYNTAX_LNAME_IS_NCNAME => {
286			description => 'Character just after the colon (:) in QName (%s) must be one of NameStartChar in namespaced XML document',
287			level => 'nswfc',
288			},
289	wakaba	1.1	NC_PREFIX_NOT_DEFINED => {
290			description => 'Undeclared namespace prefix (%s) is used',
291			level => 'nswfc',
292			},
293			NS_SYNTAX_NAME_IS_NCNAME => {
294	wakaba	1.5	description => 'Name with colon (%s) cannot be used here in namespaced XML document',
295	wakaba	1.1	level => 'nswfc',
296			},
297	wakaba	1.5	NS_SYNTAX_NAME_IS_QNAME => {
298			description => 'Name with colon (%s) must match with QName in namespaced XML document',
299			level => 'nswfc',
300			},
301			NC_UNIQUE_ATT_SPEC => {
302			description => 'Dupulicate attribute specification (%s == <%s>:%s)',
303			level => 'wfc',
304			},
305	wakaba	1.1	## Namespace validity error
306	wakaba	1.4	## XML (non-fatal) error
307	wakaba	1.3	ERR_EXT_ENTITY_NOT_FOUND => {
308			description => 'External entity (%s == <%s>) cannot be retrived (%s)',
309			level => 'vc',
310			},
311	wakaba	1.4	ERR_XML_NDATA_REF_IN_ENTITY_VALUE => {
312			description => 'Unparsed entity (%s) cannot be referred in EntityValue',
313			level => 'warn', ## Was fatal error but refined by SE Errata
314			},
315			ERR_XML_SYSID_HAS_FRAGMENT => {
316	wakaba	1.3	description => 'URI Reference converted from system identifier should not have the fragment identifier (%s)',
317			level => 'warn',
318			},
319	wakaba	1.4	## XML warning
320	wakaba	1.1	WARN_PREDEFINED_ENTITY_NOT_DECLARED => {
321			description => 'Predefined general entity (%s) should be declared before it is referred for interoperability',
322			level => 'warn',
323			},
324			WARN_UNICODE_COMPAT_CHARACTER => {
325			description => sub {
326			my $r = sprintf 'Use of the character U+%04X is deprecated, since it is classified as compatible character in the Unicode Standard',
327			$_[1]->{t};
328			$_[1]->{t} = undef;
329			$r;
330			},
331			level => 'warn',
332			},
333			WARN_UNICODE_NONCHARACTER => {
334			description => sub {
335			my $r = sprintf 'Use of the code point U+%04X is deprecated, since it is noncharacter in the Unicode Standard',
336			$_[1]->{t};
337			$_[1]->{t} = undef;
338			$r;
339			},
340			level => 'warn',
341			},
342			WARN_UNICODE_XML_NOT_SUITABLE_CHARACTER => {
343			description => sub {
344			my $r = sprintf 'Use of the character U+%04X is deprecated by W3C Note unicode-xml',
345			$_[1]->{t};
346			$_[1]->{t} = undef;
347			$r;
348			},
349			level => 'warn',
350			},
351			WARN_UNIQUE_ENTITY_NAME => {
352			description => 'General entity %s is already declared',
353			level => 'warn',
354			},
355			WARN_UNIQUE_PARAMETER_ENTITY_NAME => {
356			description => 'Parameter entity %s is already declared',
357			level => 'warn',
358			},
359	wakaba	1.5	## XMLName recommendation
360			WARN_XML_NS_URI_IS_RELATIVE => {
361			description => 'URI of XML Namespace name is a relative URI',
362			level => 'warn',
363			},
364	wakaba	1.3	## RFC 3023 'SHOULD'
365			WARN_MT_DTD_EXTERNAL_SUBSET => {
366			description => q(Media type "application/xml-dtd" SHOULD be used for the external subset of the DTD or the external parameter entity),
367			level => 'warn',
368			},
369			WARN_MT_EXTERNAL_GENERAL_PARSED_ENTITY => {
370			description => q(Media type "application/xml-external-parsed-entity" SHOULD be used for the external general parsed entity),
371			level => 'warn',
372			},
373			WARN_MT_XML_FOR_EXT_GENERAL_ENTITY => {
374			description => 'Using media type %s for external general parsed entity is now forbidden unless the entity is also well-formed as a document entity',
375			level => 'warn',
376			},
377	wakaba	1.1	## Implementation's warning
378	wakaba	1.4	WARN_DOCTYPE_NOT_FOUND => {
379			description => 'No document type definition provided for this document',
380			level => 'warn',
381			},
382			WARN_ENTITY_DECLARATION_NOT_PROCESSED => {
383			description => 'This entity declaration is not processed because unread parameter entity is referred before this declaration',
384			level => 'warn',
385			},
386			WARN_EXTERNALLY_DEFINED_ENTITY_REFERRED => {
387			description => 'The entity referred (%s) is declared in the external entity, so different groove can be constructed when external entity is not read',
388			level => 'warn',
389			},
390	wakaba	1.3	WARN_GUESS_ENCODING_IMPL_ERR => {
391			description => 'Guessing encoding procedure cause some error (%s)',
392			level => 'warn',
393			},
394	wakaba	1.5	WARN_INVALID_URI_CHAR_IN_NS_NAME => {
395			description => 'URI of XML Namespace name contains at least one non-URI character',
396			level => 'warn',
397			},
398	wakaba	1.3	WARN_INVALID_URI_CHAR_IN_SYSID => {
399			description => 'System identifier has at least one character that is invalid as part of URI Reference (%s)',
400			level => 'warn',
401			},
402			WARN_MT_TEXT_XML => {
403			description => q(In many case, labeling with the media type "text/xml" is inappropriate. Use "application/xml" or markup language specified type instead),
404			level => 'warn',
405			},
406			WARN_MT_TEXT_XML_EXTERNAL_PARSED_ENTITY => {
407			description => q(In many case, labeling with the media type "text/xml-external-parsed-entity" is inappropriate. Use "application/xml-external-parsed-entity" instead),
408			level => 'warn',
409			},
410			WARN_NO_CHARSET_PARAM => { ## charset parameter is optional
411			description => 'Charset parameter should be specified (%s)',
412			level => 'warn',
413			},
414			WARN_NO_CHARSET_PARAM_FOR_TEXT => { ## charset parameter have default value of ascii
415			description => q(Charset parameter is not specified, so default value 'us-ascii' is assumed (%s)),
416			level => 'warn',
417			},
418			WARN_NO_EXPLICIT_ENCODING_INFO => { ## BOM and '<?' guessing is failed (so general encoding guess is to be called)
419			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)),
420			level => 'warn',
421			},
422	wakaba	1.4	WARN_PI_TARGET_NOTATION => {
423			description => 'Target name of the process instruction (%s) should be declared as a notation name to ensure interoperability',
424			level => 'warn',
425			},
426			WARN_PID_EMPTY => {
427			description => 'Public identifier is empty',
428			level => 'warn',
429			},
430	wakaba	1.3	WARN_PID_IS_INVALID_URN => {
431			description => 'Public identifier (%s) seems an invalid URN',
432			level => 'warn',
433			},
434			WARN_PID_IS_NOT_FPI_NOR_URN => {
435			description => 'Public identifier (%s) should be a formal public identifier or a uniform resource name to ensure interoperability',
436			level => 'warn',
437			},
438			WARN_PID_IS_TOO_LONG => {
439			description => 'Public identifier (%s) should be shorter for interoperability',
440			level => 'warn',
441			},
442			WARN_PID_IS_URN_WITH_RESERVED_CHAR => {
443			description => 'Public identifier (%s) seems a URN and it contains one or more reserved character ("/" and/or "?") which should not be used',
444			level => 'warn',
445			},
446	wakaba	1.4	WARN_SYSID_EMPTY => {
447			description => 'System identifier is empty, in most case it is wrong',
448			level => 'warn',
449			},
450	wakaba	1.3	WARN_XML_DECLARE_NO_VERSION_ATTR => {
451			description => 'Text declaration does not have the version pseudo attribute',
452			level => 'warn',
453			},
454	wakaba	1.1	## Misc
455			UNKNOWN => {
456			description => 'Unknown error',
457	wakaba	1.3	level => 'fatal',
458	wakaba	1.1	},
459			);
460			## TODO: error handling should be customizable (hookable) by user of this module
461			sub raise ($$%) {
462			my (undef, $o, %err) = @_;
463			my $error_type = $_Error{$err{type}} \|\| $_Error{UNKNOWN};
464			my $error_msg = ref $error_type->{description} ? &{$error_type->{description}} ($o, \%err)
465			: $error_type->{description};
466			my @err_msg;
467			ref $err{t} eq 'ARRAY' ? @err_msg = @{$err{t}} : defined $err{t} ? @err_msg = $err{t} : undef;
468			$error_msg .= ' (%s)' if scalar (@err_msg) && ($error_msg !~ /%s/);
469			$error_msg = sprintf $error_msg, @err_msg;
470			$error_msg = "Line $o->{line}, position $o->{pos}: " . $error_msg;
471			$error_msg = 'Entity '.($err{entity}\|\|$o->{entity}) . ": " . $error_msg if ($err{entity}\|\|$o->{entity});
472	wakaba	1.4	$error_msg = '<'.($o->{uri}) . ">: " . $error_msg ;#if defined $o->{uri};
473	wakaba	1.1	require Carp;
474			Carp::carp ($error_msg);
475			#Carp::croak ("Line $o->{line}, position $o->{pos}: ".$error_msg);
476			}
477
478			=head1 LICENSE
479
480			Copyright 2003 Wakaba <[email protected]>
481
482			This program is free software; you can redistribute it and/or
483			modify it under the same terms as Perl itself.
484
485			=cut
486
487	wakaba	1.5	1; # $Date: 2003/06/29 08:34:37 $