| 1 |
wakaba |
1.1 |
|
| 2 |
|
|
=head1 NAME
|
| 3 |
|
|
|
| 4 |
|
|
SuikaWiki::Markup::XML::Parser --- SuikaWiki: Simple XML parser
|
| 5 |
|
|
|
| 6 |
|
|
=head1 DESCRIPTION
|
| 7 |
|
|
|
| 8 |
|
|
This is a simple XML parser intended to be used with SuikaWiki::Markup::XML.
|
| 9 |
|
|
After parsing of the XML document, this module returns a SuikaWiki::Markup::XML
|
| 10 |
|
|
object so that you can handle XML document with that module (and other modules
|
| 11 |
|
|
implementing same interface).
|
| 12 |
|
|
|
| 13 |
|
|
This module is part of SuikaWiki.
|
| 14 |
|
|
|
| 15 |
|
|
=cut
|
| 16 |
|
|
|
| 17 |
|
|
package SuikaWiki::Markup::XML::Parser;
|
| 18 |
|
|
use strict;
|
| 19 |
|
|
our $VERSION = do{my @r=(q$Revision: 1.2 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
|
| 20 |
|
|
use Char::Class::XML qw!InXML_NameStartChar InXMLNameChar InXML_NCNameStartChar InXMLNCNameChar!;
|
| 21 |
|
|
use SuikaWiki::Markup::XML;
|
| 22 |
|
|
|
| 23 |
|
|
my %NS = (
|
| 24 |
|
|
SGML => 'urn:x-suika-fam-cx:markup:sgml:',
|
| 25 |
|
|
internal_ns_invalid => 'http://suika.fam.cx/~wakaba/-temp/2003/05/17/unknown-namespace#',
|
| 26 |
|
|
internal_attr_duplicate => 'http://suika.fam.cx/~wakaba/-temp/2003/05/17/invalid-attr#',
|
| 27 |
|
|
);
|
| 28 |
|
|
|
| 29 |
|
|
=head1 METHODS
|
| 30 |
|
|
|
| 31 |
|
|
WARNING: This module is under construction. Interface of this module is not yet fixed.
|
| 32 |
|
|
|
| 33 |
|
|
=cut
|
| 34 |
|
|
|
| 35 |
|
|
my %xml_re;
|
| 36 |
|
|
# [1] document = prolog element *Misc
|
| 37 |
|
|
# [2] Char = %x09 / %x0A / %x0D / U+0020-U+D7FF / U+E000-U+FFFD / U+10000-U+10FFFF ;; 1.0
|
| 38 |
|
|
# [2] Char = %x09 / %x0A / %x0D / %x20-7E / U+0085 / U+00A0-U+D7FF / U+E000-U+FFFD
|
| 39 |
|
|
# / U+10000-U+10FFFF ;; 1.1
|
| 40 |
|
|
# [3] s = 1*(%x20 / %x09 / %x0D / %x0A)
|
| 41 |
|
|
$xml_re{s} = qr/[\x09\x0D\x0A\x20]+/s;
|
| 42 |
|
|
# [4] NameChar = Letter / Digit / "." / "-" / "_" / ":" / CombiningChar / Extender ;; 1.0
|
| 43 |
|
|
# [4] NameStartChar = ":" / ALPHA / "_" / U+00C0-U+02FF / U+0370-U+037D
|
| 44 |
|
|
# / U+037F-U+1FFF / U+200C-U+200D / U+2070-U+218F
|
| 45 |
|
|
# / U+2C00-U+2FEF / U+3001-U+D7FF / U+F900-U+EFFFF ;; 1.1
|
| 46 |
|
|
# [4a] NameChar = NameStartChar / "-" / "." / DIGIT / U+00B7 / U+0300-U+036F
|
| 47 |
|
|
# / U+203F-U+2040 ;; 1.1
|
| 48 |
|
|
# $xml_re{NameChar} = qr/[A-Za-z0-9._:-]|[^\x00-\x7F]/;
|
| 49 |
|
|
$xml_re{NameChar} = qr/\p{InXMLNameChar}/;
|
| 50 |
|
|
# [5] Name = (Letter / "_" / ":") *NameChar ;; 1.0
|
| 51 |
|
|
# [5] Name = NameStartChar *NameChar ;; 1.1
|
| 52 |
|
|
# $xml_re{Name} = qr/(?:[A-Z_:]|[^\x00-\x7F])(?:$xml_re{NameChar})*/;
|
| 53 |
|
|
$xml_re{Name} = qr/\p{InXML_NameStartChar}\p{InXMLNameChar}*/;
|
| 54 |
|
|
# [6] Names = Name *(S Name) ;; 1.0 FE & 1.0 FE-errata (2000-09-27) & 1.0 SE
|
| 55 |
|
|
# [6] Names = Name *(%x20 Name) ;; 1.0 FE-errata (2000-04-09) & 1.0 SE-errata
|
| 56 |
|
|
$xml_re{Names} = qr/$xml_re{Name}(?:$xml_re{s}$xml_re{Name})*/s;
|
| 57 |
|
|
# [7] Nmtoken = 1*NameChar
|
| 58 |
|
|
#$xml_re{Nmtoken} = qr/(?:$xml_re{NameChar})+/;
|
| 59 |
|
|
$xml_re{Nmtoken} = qr/\p{InXMLNameChar}+/;
|
| 60 |
|
|
# [8] Nmtokens = Nmtoken *(S Nmtoken) ;; 1.0 FE & 1.0 FE-errata (2000-09-27) & 1.0 SE
|
| 61 |
|
|
# [8] Nmtokens = Nmtoken *(%x20 Nmtoken) ;; 1.0 FE-errata (2000-04-09) & 1.0 SE-errata
|
| 62 |
|
|
$xml_re{Nmtokens} = qr/$xml_re{Nmtoken}(?:$xml_re{s}$xml_re{Nmtoken})*/s;
|
| 63 |
|
|
# [9] EntityValue = <"> *((Char - ("%" / "&" / <">)) / PEReference / Reference) <">
|
| 64 |
|
|
# / "'" *((Char - ("%" / "&" / "'")) / PEReference / Reference) "'"
|
| 65 |
|
|
# [10] AttValue = <"> *((Char - ("<" / "&" / <">)) / Reference) <">
|
| 66 |
|
|
# / "'" *((Char - ("<" / "&" / "'")) / Reference) "'"
|
| 67 |
|
|
$xml_re{__AttValue_simple} = qr/"[^"]*"|'[^']*'/s;
|
| 68 |
|
|
# [11] SystemLiteral = <"> *(Char - <">) <"> / "'" *(Char - "'") "'"
|
| 69 |
|
|
$xml_re{SystemLiteral} = qr/"[^"]*"|'[^']*'/;
|
| 70 |
|
|
# [12] PublicLiteral = <"> *PubidChar <"> / "'" *(PubidChar - "'") "'"
|
| 71 |
|
|
# [13] PubidChar = %x20 / %x0D / %x0A / ALPHA / DIGIT / "-" / "'" / "(" / ")"
|
| 72 |
|
|
# / "+" / "," / "." / "/" / ":" / "=" / "?" / ";" / "!" / "*"
|
| 73 |
|
|
# / "#" / "@" / "$" / "_" / "%"
|
| 74 |
|
|
$xml_re{PubidChar} = qr[[\x0D\x0A\x20!\x24#%'()+*,./0-9:;=?\x40A-Z_a-z-]];
|
| 75 |
|
|
$xml_re{__non_PubidChar} = qr[[^\x0D\x0A\x20!\x24#%'()+*,./0-9:;=?\x40A-Z_a-z-]];
|
| 76 |
|
|
$xml_re{__PubidChar2} = qr[[\x0D\x0A\x20!\x24#%()+*,./0-9:;=?\x40A-Z_a-z-]];
|
| 77 |
|
|
$xml_re{PublicLiteral} = qr/"(?:$xml_re{PubidChar})*"|'(?:$xml_re{PubidChar2})*'/;
|
| 78 |
|
|
# [14] CharData = *(Char - ("<" / "&")) - (*(Char - ("<" / "&")) "]]>" *(Char - ("<" / "&")))
|
| 79 |
|
|
$xml_re{CharData} = qr/(?:(?!\]\]>)[^<&])*/s;
|
| 80 |
|
|
$xml_re{__CharDataP} = qr/(?:(?!\]\]>)[^<&])+/s;
|
| 81 |
|
|
# [15] Comment = "<!--" *((Char - "-") / ("-" (Char - "-")))) "-->"
|
| 82 |
|
|
$xml_re{Comment_M} = qr/<!--((?:(?!--).)*)-->/s;
|
| 83 |
|
|
# [16] PI = "<?" PITarget [S (*Char - (*Char "?>" *Char))] "?>"
|
| 84 |
|
|
$xml_re{PI_M} = qr/<\?($xml_re{Name})(?:$xml_re{s}((?:(?!\?>).)*))?\?>/s;
|
| 85 |
|
|
# [17] PITarget = Name - "xml"
|
| 86 |
|
|
# [18] CDSect = CDStart CData CDEnd
|
| 87 |
|
|
# [19] CDStart = '<![CDATA['
|
| 88 |
|
|
# [20] CDATA = (*Char - (*Char "]]>" *Char))
|
| 89 |
|
|
# [21] CDEnd = "]]>"
|
| 90 |
|
|
$xml_re{CDSect_M} = qr/<!\[CDATA\[((?:(?!\]\]>).)*)\]\]>/s;
|
| 91 |
|
|
# [22] prolog = [XMLDecl] *Misc [doctypedecl *Misc]
|
| 92 |
|
|
# [23] XMLDecl = '<?xml' VersionInfo [EncodingDecl] [SDDecl] [S] "?>"
|
| 93 |
|
|
# [24] VersionInfo = S 'version' Eq ("'" VersionNum "'" / <"> VersionNum <">)
|
| 94 |
|
|
# [25] Eq = [S] "=" [S]
|
| 95 |
|
|
# [26] VersionNum = 1*(ALPHA / DIGIT / "_" / "." / ":" / "-") ;; 1.0 FE & 1.0 SE
|
| 96 |
|
|
# [26] VersionNum = "1.0" ;; 1.0 SE-errata
|
| 97 |
|
|
# [26] VersionNum = "1.1" ;; 1.1
|
| 98 |
|
|
# [27] Misc = Comment / PI / S
|
| 99 |
|
|
# [28] doctypedecl = '<!DOCTYPE' S Name [S ExternalID] [S]
|
| 100 |
|
|
# ["[" *(markupdecl / PEReference / S) "]" [S]] ">" ;; 1.0 FE
|
| 101 |
|
|
# [28] doctypedecl = '<!DOCTYPE' S Name [S ExternalID] [S] ["[" *(markupdecl / DeclSep) "]" [S]] ">"
|
| 102 |
|
|
# ;; 1.0 FE-errata & 1.0 SE
|
| 103 |
|
|
# [28] doctypedecl = '<!DOCTYPE' S Name [S ExternalID] [S] ["[" intSubset "]" [S]] ">"
|
| 104 |
|
|
# ;; 1.0 SE-errata
|
| 105 |
|
|
$xml_re{__doctypedecl_start_simple_M} = qr/(<!DOCTYPE$xml_re{s})($xml_re{Name})($xml_re{s}SYSTEM$xml_re{s}$xml_re{__AttValue_simple}|$xml_re{s}PUBLIC$xml_re{s}$xml_re{__AttValue_simple}$xml_re{s}$xml_re{__AttValue_simple})?/s;
|
| 106 |
|
|
# [28a] DeclSep = PEReference / S ;; 1.0 FE-errata & 1.0 SE
|
| 107 |
|
|
# [28b] intSubset = *(markupdecl / DeclSep) ;; 1.0 SE-errata
|
| 108 |
|
|
# [29] markupdecl = elementdecl / AttlistDecl / EntityDecl / NotationDecl / PI / Comment
|
| 109 |
|
|
# [30] extSubset = [TextDecl] extSubsetDecl
|
| 110 |
|
|
# [31] extSubsetDecl = *(markupdecl / conditionalSect / PEReference / S) ;; 1.0 FE
|
| 111 |
|
|
# [31] extSubsetDecl = *(markupdecl / conditionalSect / DeclSep) ;; 1.0 FE-errata & 1.0 SE
|
| 112 |
|
|
# [32] SDDecl = S 'standalone' Eq ("'" ('yes' / 'no') "'" / <"> ('yes' / 'no') <">)
|
| 113 |
|
|
# [33] LanguageID = Langcode *("-" Subcode) ;; 1.0 FE (removed by 1.0 SE)
|
| 114 |
|
|
# [34] Langcode = ISO639Cocde / IanaCode / UserCode ;; 1.0 FE (ditto)
|
| 115 |
|
|
# [35] ISO639Code = 2ALPHA ;; 1.0 FE (ditto)
|
| 116 |
|
|
# [36] IanaCode = "i-" 1*ALPHA ;; 1.0 FE (ditto)
|
| 117 |
|
|
# [37] UserCode = "x-" 1*ALPHA ;; 1.0 FE (ditto)
|
| 118 |
|
|
# [38] Subcode = 1*ALPHA ;; 1.0 FE (ditto)
|
| 119 |
|
|
# [39] element = EmptyElemTag / STag content ETag
|
| 120 |
|
|
# [40] STag = "<" Name *(S Attribute) [S] ">"
|
| 121 |
|
|
# [41] Attribute = Name Eq AttValue
|
| 122 |
|
|
$xml_re{Attribute} = qr/$xml_re{Name}(?:$xml_re{s})?=(?:$xml_re{s})?$xml_re{__AttValue_simple}/s;
|
| 123 |
|
|
$xml_re{Attribute_M} = qr/($xml_re{Name})(?:$xml_re{s})?=(?:$xml_re{s})?($xml_re{__AttValue_simple})/s;
|
| 124 |
|
|
$xml_re{STag} = qr/<$xml_re{Name}(?:$xml_re{s}$xml_re{Attribute})*(?:$xml_re{s})?>/s;
|
| 125 |
|
|
# [42] ETag = "</" Name [S] ">"
|
| 126 |
|
|
$xml_re{ETag_M} = qr!</($xml_re{Name})(?:$xml_re{s})?>!s;
|
| 127 |
|
|
# [43] content = *(element / CharData / Reference / CDSect / PI / Comment) ;; 1.0 FE
|
| 128 |
|
|
# [43] content = [CharData] *((element / Reference / CDSect / PI / Comment) [CharData])
|
| 129 |
|
|
# ;; 1.0 FE-errata & 1.0 SE
|
| 130 |
|
|
# [44] EmptyElemTag = "<" Name *(S Attribute) [S] "/>"
|
| 131 |
|
|
$xml_re{__STag_or_EmptyElemTag} = qr!<$xml_re{Name}(?:$xml_re{s}$xml_re{Attribute})*(?:$xml_re{s})?/?>!s;
|
| 132 |
|
|
$xml_re{__STag_or_EmptyElemTag_simple} = qr!<$xml_re{ame}(?:$xml_re{s}|$xml_re{Name}|$xml_re{__AttValue_simple}|=)*/?>!s;
|
| 133 |
|
|
# [45] elementdecl = '<!ELEMENT' S Name S contentspec [S] ">"
|
| 134 |
|
|
# [46] contentspec = 'EMPTY' / 'ANY' / Mixed / children
|
| 135 |
|
|
$xml_re{__contentspec_simple} = qr/(?:$xml_re{Name}|\#PCDATA|[()|,?*+]|$xml_re{s})/s;
|
| 136 |
|
|
# [47] children = (choice / seq) ["?" / "*" / "+"]
|
| 137 |
|
|
# [48] cp = (Name / choice / seq) ["?" / "*" / "+"]
|
| 138 |
|
|
# [49] choice = "(" [S] cp *([S] "|" [S] cp) [S] ")" ;; 1.0 FE
|
| 139 |
|
|
# [49] choice = "(" [S] cp 1*([S] "|" [S] cp) [S] ")" ;; 1.0 FE-errata & 1.0 SE
|
| 140 |
|
|
# [50] seq = "(" [S] cp *([S] "," [S] cp) [S] ")"
|
| 141 |
|
|
# [51] Mixed = "(" [S] '#PCDATA' *([S] "|" [S] Name) [S] ")*"
|
| 142 |
|
|
# / "(" [S] '#PCDATA' [S] ")"
|
| 143 |
|
|
#$xml_re{seq} = qr/\($xml_re{cp}(?:(?:$xml_re{s})?,(?:$xml_re{s})?$xml_re{cp})*(?:$xml_re{s})?\)/;
|
| 144 |
|
|
#$xml_re{cp} = qr/(?:$xml_re{Name}|$xml_re{choice}|$xml_re{seq})[?*+]/;
|
| 145 |
|
|
#$xml_re{choice} = qr/\($xml_re{cp}(?:(?:$xml_re{s})?\|(?:$xml_re{s})?$xml_re{cp})+(?:$xml_re{s})?\)/;
|
| 146 |
|
|
#$xml_re{children} = qr/(?:$xml_re{choice}|$xml_re{seq})[?*+]/;
|
| 147 |
|
|
#$xml_re{Mixed} = qr/(?:\((?:$xml_re{s})?\#PCDATA(?:$xml_re{s})?(?:(?:$xml_re{s})?|(?:$xml_re{s})?$xml_re{Name})*(?:$xml_re{s})?\)|\((?:$xml_re{s})?\#PCDATA(?:$xml_re{s})?\))/;
|
| 148 |
|
|
#$xml_re{contentspec} = qr/(?:EMPTY|ANY|$xml_re{Mixed}|$xml_re{children})/;
|
| 149 |
|
|
# [52] AttlistDecl = '<!ATTLIST' S Name *AttDef [S] ">"
|
| 150 |
|
|
# [53] AttDef = S Name S AttType S DefaultDecl
|
| 151 |
|
|
# [54] AttType = StringType / TokenizedType / EnumeratedType
|
| 152 |
|
|
# [55] StringType = 'CDATA'
|
| 153 |
|
|
# [56] TokenizedType = 'ID' / 'IDREF' / 'IDREFS' / 'ENTITY' / 'ENTITIES' / 'NMTOKEN' / 'NMTOKENS'
|
| 154 |
|
|
# [57] EnumeratedType = NotationType / Enumeration
|
| 155 |
|
|
# [58] NotationType = 'NOTATION' S "(" [S] Name *([S] "|" [S] Name) [S] ")"
|
| 156 |
|
|
# [59] Enumeration = "(" [S] Nmtoken *([S] "|" [S] Nmtoken) [S} ")"
|
| 157 |
|
|
# [60] DefaultDecl = '#REQUIRED' / '#IMPLIED' / ['#FIXED' S] AttValue
|
| 158 |
|
|
# [61] conditionalSect = includeSect / ignoreSect
|
| 159 |
|
|
# [62] includeSect = "<![" [S} 'INCLUDE' [S] "[" extSubsetDecl "]]>"
|
| 160 |
|
|
# [63] ignoreSect = "<![" [S] 'IGNORE' [S] "[" *ignoreSectContents "]]>"
|
| 161 |
|
|
# [64] ignoreSectContents = Ignore *("<![" ignoreSectContents "]]>" Ignore)
|
| 162 |
|
|
# [65] Ignore = *Char - (*Char ("<![" / "]]>") *Char)
|
| 163 |
|
|
# [66] CharRef = '&#' 1*DIGIT ";" / '&#x' 1*HEXDIGIT ";"
|
| 164 |
|
|
$xml_re{CharRef} = qr/&#[0-9]+;|&#x[0-9A-Fa-f]+;/;
|
| 165 |
|
|
# [67] Reference = EntityRef / CharRef
|
| 166 |
|
|
# [68] EntityRef = "&" Name ";"
|
| 167 |
|
|
$xml_re{EntityRef} = qr/&$xml_re{Name};/;
|
| 168 |
|
|
$xml_re{EntityRef_M} = qr/&($xml_re{Name});/;
|
| 169 |
|
|
$xml_re{Reference} = qr/$xml_re{EntityRef}|$xml_re{CharRef}/;
|
| 170 |
|
|
$xml_re{AttValue} = qr/"(?:$xml_re{Reference}|[^&<"])*"|'(?:$xml_re{Reference}|[^&<'])*'/s;
|
| 171 |
|
|
# [69] PEReference = "%" Name ";"
|
| 172 |
|
|
$xml_re{PEReference} = qr/%(?:$xml_re{Name});/;
|
| 173 |
|
|
$xml_re{PEReference_M} = qr/%($xml_re{Name});/;
|
| 174 |
|
|
$xml_re{__elementdecl_simple} = qr/<!ELEMENT(?:$xml_re{s}|$xml_re{PEReference}|$xml_re{Name}|\#PCDATA|[()|,?*+])+>/s;
|
| 175 |
|
|
$xml_re{__AttlistDecl_simple} = qr/<!ATTLIST(?:$xml_re{PEReference}|$xml_re{Name}|\#$xml_re{Name}|$xml_re{s}|$xml_re{__AttValue_simple})*>/s;
|
| 176 |
|
|
# [70] EntityDecl = GEDecl / PEDecl
|
| 177 |
|
|
$xml_re{__EntityDecl_simple} = qr/<!ENTITY$xml_re{s}(?:%$xml_re{s})?(?:$xml_re{__AttValue_simple}|$xml_re{Name}|$xml_re{PEReference}|$xml_re{s})*>/s;
|
| 178 |
|
|
# [71] GEDecl = '<!ENTITY' S Name S EntityDef [S] ">"
|
| 179 |
|
|
# [72] PEDecl = '<!ENTITY' S "%" S Name S PEDef [S] ">"
|
| 180 |
|
|
# [73] EntityDef = EntityValue / ExternalID [NDataDecl]
|
| 181 |
|
|
# [74] PEDef = EntityValue / ExternalID
|
| 182 |
|
|
# [75] ExternalID = 'SYSTEM' S SystemLiteral / 'PUBLIC' S PublicLiteral S SystemLiteral
|
| 183 |
|
|
# [76] NDataDecl = S 'NDATA' S Name
|
| 184 |
|
|
# [77] TextDecl = '?xml' [VersionInfo] EncodingDecl [S] "?>"
|
| 185 |
|
|
# [78] extParsedEnt = [TextDecl] content
|
| 186 |
|
|
# [79] extPE ;; 1.0 FE (removed by errata)
|
| 187 |
|
|
# [80] EncodingDecl = S 'encoding' Eq (<"> EncName <"> / "'" EncName "'")
|
| 188 |
|
|
# [81] EncName = ALPHA *(ALPHA / DIGIT / "." / "_" / "-")
|
| 189 |
|
|
# [82] NotationDecl = '<!NOTATION' S Name S (ExternalID / PublicID) [S] ">"
|
| 190 |
|
|
$xml_re{__NotationDecl_simple} = qr/<!NOTATION$xml_re{s}(?:$xml_re{__AttValue_simple}|$xml_re{Name}|$xml_re{PEReference}|$xml_re{s})*>/s;
|
| 191 |
|
|
# [83] PublicID = 'PUBLIC' S PubidLiteral
|
| 192 |
|
|
# [84] Letter = BaseChar / Ideographic
|
| 193 |
|
|
# [85] Basechar = ...
|
| 194 |
|
|
# [86] Ideographic = ...
|
| 195 |
|
|
# [87] CombiningChar = ...
|
| 196 |
|
|
# [88] Digit = ...
|
| 197 |
|
|
# [89] Extender = U+00B7 / U+02D0 / U+02D1 / U+0387 / U+0640 / U+0E46 / U+0EC6
|
| 198 |
|
|
# / U+3005 / U+3031-U+3035 / U+309D / U+309E / U+30FC-U+30FE
|
| 199 |
|
|
## [84]-[89] removed by 1.1
|
| 200 |
|
|
|
| 201 |
|
|
# [XMLNames]
|
| 202 |
|
|
# [1] NSAttName = PrefixedAttName / DefaultAttName
|
| 203 |
|
|
# [2] PrefixedAttName = 'xmlns:' NCName
|
| 204 |
|
|
# [3] DefaultAttName = 'xmlns'
|
| 205 |
|
|
# [4] NCName = (Letter / "_") *NCNameChar ;; 1.0
|
| 206 |
|
|
# [4] NCName = NCNameStartChar *NCNameChar ;; 1.1
|
| 207 |
|
|
# [5] NCNameChar = Letter / Digit / "." / "-" / "_" / CombiningChar / Extender ;; 1.0
|
| 208 |
|
|
# [5] NCNameChar = NameChar - ":" ;; 1.1
|
| 209 |
|
|
# [5a] NCNameStartChar = NameStartChar - ":" ;; 1.1
|
| 210 |
|
|
# [6] QName = [Prefix ":"] LocalPart ;; 1.0
|
| 211 |
|
|
# [6] QName = PrefixedName / UnprefixedName ;; 1.1
|
| 212 |
|
|
# [6a] PrefixedName = Prefix ":" LocalPart ;; 1.1
|
| 213 |
|
|
# [6b] UnprefixedName = LocalPart ;; 1.1
|
| 214 |
|
|
# [7] Prefix = NCName
|
| 215 |
|
|
# [8] LocalPart = NCName
|
| 216 |
|
|
# [9] STag = "<" QName *(S Attribute) [S] ">"
|
| 217 |
|
|
$xml_re{__NCSTag} = qr/<$xml_re{QName}(?:$xml_re{s}$xml_re{Attribute})*(?:$xml_re{s})?>/s;
|
| 218 |
|
|
# [10] ETag = "</" QName [S] ">"
|
| 219 |
|
|
$xml_re{__NCETag} = qr!</$xml_re{QName}(?:$xml_re{s})?>!s;
|
| 220 |
|
|
# [11] EmptyElemTag = "<" QName *(S Attribute) [S] "/>"
|
| 221 |
|
|
# [12] Attribute = NSAttName Eq AttValue / QName Eq AttValue
|
| 222 |
|
|
# [13] doctypedecl = '<!DOCTYPE' S QName [S ExternalID] [S]
|
| 223 |
|
|
# ["[" *(markupdecl / PEReference / S) "]" [S]] ">"
|
| 224 |
|
|
# [14] elementdecl = '<!ELEMENT' S QName S contentspec [S] ">"
|
| 225 |
|
|
# [15] cp = (QName / choice / sep) ["?" / "*" / "+"]
|
| 226 |
|
|
# [16] Mixed = "(" [S] '#PCDATA' *([S] "|" [S] QName) [S] ")*" / "(" [S] '#PCDATA' [S] ")"
|
| 227 |
|
|
# [17] AttlistDecl = '<!ATTLIST' S QName *AttDef [S] ">"
|
| 228 |
|
|
# [18] AttDef = S (QName / NSAttName) S AttType S DefaultDecl
|
| 229 |
|
|
# [19] Name = NameStartChar *NameChar ;; 1.1 draft
|
| 230 |
|
|
# [20] NameChar = {XML1.1}.NameChar ;; 1.1 draft
|
| 231 |
|
|
# [21] NameStartChar = {XML1.1}.NameStartChar ;; 1.1 draft
|
| 232 |
|
|
|
| 233 |
|
|
|
| 234 |
|
|
sub new ($) {
|
| 235 |
|
|
bless {}, shift;
|
| 236 |
|
|
}
|
| 237 |
|
|
|
| 238 |
|
|
sub parse_text ($$;$) {
|
| 239 |
|
|
my $self = shift;
|
| 240 |
|
|
my $s = shift;
|
| 241 |
|
|
my $o = shift || {line => 0, pos => 0, entity_type => ''};
|
| 242 |
|
|
my $r = SuikaWiki::Markup::XML->new (type => '#document');
|
| 243 |
|
|
my $c = $r;
|
| 244 |
|
|
if ($s =~ /([^\x09\x0A\x0D\x20-\x{D7FF}\x{E000}-\x{FFFD}\x{10000}-\x{10FFFF}])/) {
|
| 245 |
|
|
my $o = {line => 0, pos => 0};
|
| 246 |
|
|
_count_lp ($`, $o);
|
| 247 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_LEGAL_CHARACTER', t => ord ($1));
|
| 248 |
|
|
## NOTE: Even if there are more than one non-XML-Char, error is raised only one time.
|
| 249 |
|
|
}
|
| 250 |
|
|
while ($s) {
|
| 251 |
|
|
my $is_dtd = ($o->{entity_type} eq 'dtd_external_subset'
|
| 252 |
|
|
|| $o->{entity_type} eq 'dtd_internal_subset');
|
| 253 |
|
|
my $can_be_dtd = ($o->{entity_type} eq 'external_parsed_entity');
|
| 254 |
|
|
my $is_ext_entity = ($o->{entity_type} eq 'dtd_external_subset'
|
| 255 |
|
|
|| $o->{entity_type} eq 'general_external_parsed_entity'
|
| 256 |
|
|
|| $o->{entity_type} eq 'external_parsed_entity');
|
| 257 |
|
|
if ($s =~ s/^($xml_re{__STag_or_EmptyElemTag_simple})//s) {
|
| 258 |
|
|
my ($stag) = ($1);
|
| 259 |
|
|
if ($is_dtd) {
|
| 260 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_INVALID_POSITION', t => $stag);
|
| 261 |
|
|
} elsif ($c->node_type eq '#document') {
|
| 262 |
|
|
if ($o->{entity_type} eq 'document_entity' && $self->_is_brother_of_root_element ($c)) {
|
| 263 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_DATA_OUT_OF_ROOT_ELEMENT', t => $stag);
|
| 264 |
|
|
} elsif (!$o->{entity_type} || $can_be_dtd) {
|
| 265 |
|
|
$o->{entity_type} = 'general_external_parsed_entity';
|
| 266 |
|
|
}
|
| 267 |
|
|
}
|
| 268 |
|
|
## Parse element type name
|
| 269 |
|
|
$stag =~ s/^<($xml_re{Name})//s;
|
| 270 |
|
|
my ($prefix, $lname) = _ns_parse_qname ($1); ## TODO: invalid qname valid name
|
| 271 |
|
|
$c = $c->append_new_node (local_name => $lname);
|
| 272 |
|
|
$c->flag (p_original_qname => $1);
|
| 273 |
|
|
$c->flag (p_o_start => $self->_make_clone_of ($o));
|
| 274 |
|
|
_count_lp ($&, $o);
|
| 275 |
|
|
## Parse attributes
|
| 276 |
|
|
$stag =~ s/>$//;
|
| 277 |
|
|
my $etag = 0;
|
| 278 |
|
|
if ($stag =~ s!/$!!) { # empty element
|
| 279 |
|
|
$c->option (use_EmptyElemTag => 1);
|
| 280 |
|
|
$etag = 1;
|
| 281 |
|
|
}
|
| 282 |
|
|
$self->_parse_attribute_spec_list ($c, $stag, $o) if $stag;
|
| 283 |
|
|
_count_lp (($etag ? '/>' : '>'), $o);
|
| 284 |
|
|
## Parse element type name (namespace)
|
| 285 |
|
|
my $uri = $c->defined_namespace_prefix ($prefix || '');
|
| 286 |
|
|
if ($uri) {
|
| 287 |
|
|
$c->namespace_uri ($uri);
|
| 288 |
|
|
} elsif (!$prefix) { ## Default NS
|
| 289 |
|
|
#$c->namespace_uri ('');
|
| 290 |
|
|
$c->define_new_namespace ('' => '');
|
| 291 |
|
|
} else {
|
| 292 |
|
|
my $o_etypename = $self->_make_clone_of ($c->flag ('p_o_start'));
|
| 293 |
|
|
$o_etypename->{pos}++; # '<'
|
| 294 |
|
|
$self->_raise_error ($o_etypename, type => 'NC_PREFIX_NOT_DEFINED', t => $prefix);
|
| 295 |
|
|
my $uri = $NS{internal_ns_invalid}.$self->_uri_escape ($prefix);
|
| 296 |
|
|
$c->namespace_uri ($uri);
|
| 297 |
|
|
$c->define_new_namespace ($prefix => $uri);
|
| 298 |
|
|
}
|
| 299 |
|
|
## Ending the element if EmptyElemTag
|
| 300 |
|
|
if ($etag) {
|
| 301 |
|
|
$c = $c->{parent};
|
| 302 |
|
|
}
|
| 303 |
|
|
} elsif ($s =~ s/^$xml_re{ETag_M}//s) {
|
| 304 |
|
|
if (!$is_dtd) {
|
| 305 |
|
|
my $ename = $1;
|
| 306 |
|
|
if ($ename eq $c->flag ('p_original_qname') || $ename eq $c->qname) {
|
| 307 |
|
|
$c = $c->{parent};
|
| 308 |
|
|
} else { ## Element type name does not match
|
| 309 |
|
|
my $o_etn = $self->_make_clone_of ($o);
|
| 310 |
|
|
$o_etn->{pos} += 2;
|
| 311 |
|
|
$self->_raise_error ($o_etn, p => $c, type => 'WFC_ELEMENT_TYPE_MATCH', t => [$ename, $c->qname]);
|
| 312 |
|
|
}
|
| 313 |
|
|
} else {
|
| 314 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_INVALID_POSITION', t => $&);
|
| 315 |
|
|
}
|
| 316 |
|
|
_count_lp ($&, $o);
|
| 317 |
|
|
} elsif (($is_dtd || $can_be_dtd) && $s =~ s/^$xml_re{PEReference_M}//s) {
|
| 318 |
|
|
if ($can_be_dtd) {
|
| 319 |
|
|
$o->{entity_type} = 'dtd_external_subset'; $can_be_dtd = 0;
|
| 320 |
|
|
}
|
| 321 |
|
|
$c->append_new_node (type => '#reference', local_name => $1, ## BUG: %in:valid;
|
| 322 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:entity:parameter');
|
| 323 |
|
|
_count_lp ($1.'%;', $o);
|
| 324 |
|
|
} elsif (($is_dtd || $can_be_dtd || $c->node_type eq '#document') && ($s =~ s/^$xml_re{s}//s)) {
|
| 325 |
|
|
$c->append_text ($&);
|
| 326 |
|
|
_count_lp ($&, $o);
|
| 327 |
|
|
} elsif (!$is_dtd && ($s =~ s/^$xml_re{__CharDataP}//s)) {
|
| 328 |
|
|
if ($c->node_type eq '#document') {
|
| 329 |
|
|
if ($o->{entity_type} eq 'document_entity') {
|
| 330 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_DATA_OUT_OF_ROOT_ELEMENT', t => $&);
|
| 331 |
|
|
} elsif (!$o->{entity_type} || $can_be_dtd) {
|
| 332 |
|
|
$o->{entity_type} = 'general_external_parsed_entity'; $can_be_dtd = 0;
|
| 333 |
|
|
}
|
| 334 |
|
|
}
|
| 335 |
|
|
$c->append_text ($&);
|
| 336 |
|
|
_count_lp ($&, $o);
|
| 337 |
|
|
} elsif ($s =~ s/^($xml_re{Reference})//s) {
|
| 338 |
|
|
if ($is_dtd) {
|
| 339 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_INVALID_POSITION', t => $1);
|
| 340 |
|
|
} elsif ($c->node_type eq '#document') {
|
| 341 |
|
|
if ($self->_is_brother_of_root_element ($c)) {
|
| 342 |
|
|
if ($o->{entity_type} eq 'document_entity') {
|
| 343 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_DATA_OUT_OF_ROOT_ELEMENT', t => $1);
|
| 344 |
|
|
} elsif (!$o->{entity_type} || $can_be_dtd) {
|
| 345 |
|
|
$o->{entity_type} = 'general_external_parsed_entity'; $can_be_dtd = 0;
|
| 346 |
|
|
}
|
| 347 |
|
|
}
|
| 348 |
|
|
}
|
| 349 |
|
|
$self->_parse_reference ($c, $1, $o);
|
| 350 |
|
|
} elsif ($s =~ s/^$xml_re{CDSect_M}//s) {
|
| 351 |
|
|
if ($is_dtd) {
|
| 352 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_INVALID_POSITION', t => '<?>//s) {
|
| 435 |
|
|
$o->{entity_type} = 'document_entity';
|
| 436 |
|
|
_count_lp ($&, $o);
|
| 437 |
|
|
$c = $c->{parent};
|
| 438 |
|
|
} elsif ($s =~ s/^($xml_re{__NotationDecl_simple})//s) {
|
| 439 |
|
|
if ($can_be_dtd) {
|
| 440 |
|
|
$o->{entity_type} = 'dtd_external_subset'; $can_be_dtd = 0;
|
| 441 |
|
|
} elsif (!$is_dtd) {
|
| 442 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_INVALID_POSITION', t => '<!NOTATION');
|
| 443 |
|
|
}
|
| 444 |
|
|
$self->_parse_entity_declaration ($1, $c, $o);
|
| 445 |
|
|
} elsif ($s =~ s/^$xml_re{PI_M}//s) {
|
| 446 |
|
|
my ($target, $data, $all) = ($1, $2, $&);
|
| 447 |
|
|
if ($target eq 'xml') {
|
| 448 |
|
|
if ($c->node_type eq '#document' && $c->count == 0) {
|
| 449 |
|
|
_count_lp ('<?xml', $o);
|
| 450 |
|
|
if (length ($data)) {
|
| 451 |
|
|
$self->_parse_xml_declaration ($c->append_new_node (type => '#pi', local_name => 'xml'), $data, $o);
|
| 452 |
|
|
} else {
|
| 453 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_XML_DECLARE_NO_ATTR');
|
| 454 |
|
|
}
|
| 455 |
|
|
_count_lp ('?>', $o);
|
| 456 |
|
|
} else {
|
| 457 |
|
|
_raise_fatal_error ($o, desc => 'INVALID_XML_DECLARE');
|
| 458 |
|
|
}
|
| 459 |
|
|
} else {
|
| 460 |
|
|
$c->append_new_node (type => '#pi', local_name => $1, value => $2);
|
| 461 |
|
|
_count_lp ($all, $o);
|
| 462 |
|
|
}
|
| 463 |
|
|
} else {
|
| 464 |
|
|
$self->_raise_error ($o, p => $c, type => 'SYNTAX_INVALID_CHAR', t => substr ($s, 0, 1));
|
| 465 |
|
|
substr ($s, 0, 1) = '';
|
| 466 |
|
|
}
|
| 467 |
|
|
}
|
| 468 |
|
|
while (1) {
|
| 469 |
|
|
if (ref $c->{parent}) {
|
| 470 |
|
|
if ($c->node_type eq '#element') {
|
| 471 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_END_TAG_NOT_FOUND', t => $c);
|
| 472 |
|
|
} else {
|
| 473 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_END_OF_MARKUP_NOT_FOUND', t => $c);
|
| 474 |
|
|
}
|
| 475 |
|
|
$c = $c->{parent};
|
| 476 |
|
|
} else {
|
| 477 |
|
|
last;
|
| 478 |
|
|
}
|
| 479 |
|
|
}
|
| 480 |
|
|
wantarray ? ($r, $o) : $r;
|
| 481 |
|
|
}
|
| 482 |
|
|
|
| 483 |
|
|
sub _uri_escape ($$) {
|
| 484 |
|
|
shift;
|
| 485 |
|
|
my $s = shift; ## TODO: support utf8 flag
|
| 486 |
|
|
$s =~ s/([^0-9A-Za-z_.-])/sprintf '%%%02X', ord $1/ge;
|
| 487 |
|
|
$s;
|
| 488 |
|
|
}
|
| 489 |
|
|
|
| 490 |
|
|
sub _make_clone_of ($$) {
|
| 491 |
|
|
my ($self, $mother) = @_;
|
| 492 |
|
|
if (ref $mother eq 'HASH') {
|
| 493 |
|
|
my $child = {};
|
| 494 |
|
|
for (keys %$mother) {
|
| 495 |
|
|
$child->{$_} = $mother->{$_}; ## BUG: recursive!
|
| 496 |
|
|
}
|
| 497 |
|
|
return $child;
|
| 498 |
|
|
} else {
|
| 499 |
|
|
## BUG: not supported
|
| 500 |
|
|
}
|
| 501 |
|
|
}
|
| 502 |
|
|
|
| 503 |
|
|
sub _count_lp ($$) {
|
| 504 |
|
|
my ($s, $o) = @_;
|
| 505 |
|
|
$s =~ s/[^\x0A\x0D]*(?:\x0D\x0A?|\x0A)/$o->{line}++;$o->{pos}=0;''/ges;
|
| 506 |
|
|
$o->{pos} += length $s;
|
| 507 |
|
|
}
|
| 508 |
|
|
sub _ns_parse_qname ($) {
|
| 509 |
|
|
my $qname = shift;
|
| 510 |
|
|
if ($qname =~ /:/) {
|
| 511 |
|
|
return split /:/, $qname, 2;
|
| 512 |
|
|
} else {
|
| 513 |
|
|
return (undef, $qname);
|
| 514 |
|
|
}
|
| 515 |
|
|
}
|
| 516 |
|
|
sub _parse_attribute_spec_list ($$$$) {
|
| 517 |
|
|
my ($self, $c, $attrs, $o) = @_;
|
| 518 |
|
|
my @attrs;
|
| 519 |
|
|
my (%defined_attr, %defined_ns_attr);
|
| 520 |
|
|
my $no_s = 0;
|
| 521 |
|
|
while ($attrs) {
|
| 522 |
|
|
if (!$no_s && $attrs =~ s/^$xml_re{Attribute_M}//s) {
|
| 523 |
|
|
my ($qname, $qvalue) = ($1, $2);
|
| 524 |
|
|
my ($prefix, $name) = _ns_parse_qname ($qname);
|
| 525 |
|
|
push @attrs, {prefix => $prefix, lname => $name, qvalue => $qvalue, qname => $qname,
|
| 526 |
|
|
o => {line => $o->{line}, pos => ($o->{pos} + length ($&) - length ($qvalue))},
|
| 527 |
|
|
o_attr_start => {line => $o->{line}, pos => $o->{pos}}};
|
| 528 |
|
|
_count_lp ($&, $o);
|
| 529 |
|
|
$no_s = 1;
|
| 530 |
|
|
} elsif ($attrs =~ s/^($xml_re{s})//s) {
|
| 531 |
|
|
_count_lp ($1, $o);
|
| 532 |
|
|
$no_s = 0;
|
| 533 |
|
|
} else {
|
| 534 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_INVALID_CHAR', t => substr ($attrs, 0, 1));
|
| 535 |
|
|
substr ($attrs, 0, 1) = '';
|
| 536 |
|
|
$no_s = 1;
|
| 537 |
|
|
}
|
| 538 |
|
|
}
|
| 539 |
|
|
for (grep {($_->{prefix} eq 'xmlns') || (!$_->{prefix} && ($_->{lname} eq 'xmlns'))} @attrs) {
|
| 540 |
|
|
my $value = SuikaWiki::Markup::XML->new (type => '#text');
|
| 541 |
|
|
$self->_parse_attribute_value ($value, $_->{qvalue}, $_->{o});
|
| 542 |
|
|
if (!$defined_attr{$_->{qname}}) {
|
| 543 |
|
|
if ($_->{prefix} eq 'xmlns') {
|
| 544 |
|
|
$c->define_new_namespace ($_->{lname} => $value); # BUG: Reference
|
| 545 |
|
|
} else {
|
| 546 |
|
|
$c->define_new_namespace ('' => $value); # BUG: Reference
|
| 547 |
|
|
}
|
| 548 |
|
|
$defined_attr{$_->{qname}} = 1;
|
| 549 |
|
|
} else { ## Already defined
|
| 550 |
|
|
$self->_raise_error ($_->{o_attr_start}, type => 'WFC_UNIQUE_ATT_SPEC', t => $_->{qname});
|
| 551 |
|
|
my $attr;
|
| 552 |
|
|
if ($_->{prefix} eq 'xmlns') {
|
| 553 |
|
|
$attr = $c->set_attribute ($_->{lname} => '', namespace_uri => $NS{internal_attr_duplicate}.'xmlns');
|
| 554 |
|
|
} else { ## BUG: xmlns="" (1.1)
|
| 555 |
|
|
$attr = $c->set_attribute (xmlns => '', namespace_uri => $NS{internal_attr_duplicate});
|
| 556 |
|
|
}
|
| 557 |
|
|
$attr->append_node ($value);
|
| 558 |
|
|
}
|
| 559 |
|
|
}
|
| 560 |
|
|
for (grep {($_->{prefix} ne 'xmlns') && !(!$_->{prefix} && ($_->{lname} eq 'xmlns'))} @attrs) {
|
| 561 |
|
|
my $attr;
|
| 562 |
|
|
my $uri; $uri = $c->defined_namespace_prefix ($_->{prefix}) if $_->{prefix};
|
| 563 |
|
|
if ($defined_attr{$_->{qname}}) { ## Already-defined-attr is found
|
| 564 |
|
|
$self->_raise_error ($_->{o_attr_start}, type => 'WFC_UNIQUE_ATT_SPEC', t => $_->{qname});
|
| 565 |
|
|
$uri = $NS{internal_attr_duplicate} . $defined_attr{$_->{qname}};
|
| 566 |
|
|
$_->{prefix} = 'dup.' . $defined_attr{$_->{qname}} . ($_->{prefix} ? '.'.$_->{prefix}:'');
|
| 567 |
|
|
$c->define_new_namespace ($_->{prefix} => $uri);
|
| 568 |
|
|
$defined_attr{$_->{qname}}++;
|
| 569 |
|
|
} elsif (defined $uri && $defined_ns_attr{$_->{lname}.':'.$uri}) {
|
| 570 |
|
|
## ns:attr="a" ns2:attr="b" xmlns:ns="ns" xmlns:ns2="ns"
|
| 571 |
|
|
$self->_raise_error ($_->{o_attr_start}, type => 'NC_unique_att_spec', t => $_->{qname});
|
| 572 |
|
|
my $i = $defined_ns_attr{$_->{lname}.':'.$uri}++;
|
| 573 |
|
|
$uri = $NS{internal_attr_duplicate} . 'ns.' . $i;
|
| 574 |
|
|
$_->{prefix} = 'dup.ns.' . $i . ($_->{prefix} ? '.'.$_->{prefix}:'');
|
| 575 |
|
|
$c->define_new_namespace ($_->{prefix} => $uri);
|
| 576 |
|
|
} else {
|
| 577 |
|
|
$defined_attr{$_->{qname}} = 1;
|
| 578 |
|
|
$defined_ns_attr{$_->{lname}.':'.$uri} = 1;
|
| 579 |
|
|
}
|
| 580 |
|
|
if ($_->{prefix}) {
|
| 581 |
|
|
if (defined $uri) {
|
| 582 |
|
|
$attr = $c->set_attribute ($_->{lname} => '', namespace_uri => $uri);
|
| 583 |
|
|
} else {
|
| 584 |
|
|
$self->_raise_error ($o, type => 'NC_PREFIX_NOT_DEFINED', t => $_->{prefix});
|
| 585 |
|
|
my $uri = $NS{internal_ns_invalid}.$self->_uri_escape ($_->{prefix});
|
| 586 |
|
|
$attr = $c->set_attribute ($_->{lname} => '', namespace_uri => $uri);
|
| 587 |
|
|
$c->define_new_namespace ($_->{prefix} => $uri);
|
| 588 |
|
|
}
|
| 589 |
|
|
} else {
|
| 590 |
|
|
$attr = $c->set_attribute ($_->{lname} => '');
|
| 591 |
|
|
}
|
| 592 |
|
|
$self->_parse_attribute_value ($attr, $_->{qvalue}, $_->{o}) if $attr;
|
| 593 |
|
|
}
|
| 594 |
|
|
}
|
| 595 |
|
|
sub _parse_attribute_value ($$$$) {
|
| 596 |
|
|
my ($self, $attr, $qvalue, $o) = @_;
|
| 597 |
|
|
$qvalue = substr ($qvalue, 1, length ($qvalue) - 2);
|
| 598 |
|
|
_count_lp ('"', $o);
|
| 599 |
|
|
while ($qvalue) {
|
| 600 |
|
|
if ($qvalue =~ s/^($xml_re{Reference})//) {
|
| 601 |
|
|
$self->_parse_reference ($attr, $1);
|
| 602 |
|
|
_count_lp ($1, $o);
|
| 603 |
|
|
} elsif ($qvalue =~ s/^([^&<]+)//) {
|
| 604 |
|
|
$attr->append_text ($1);
|
| 605 |
|
|
_count_lp ($1, $o);
|
| 606 |
|
|
} else {
|
| 607 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_INVALID_CHAR', t => substr ($qvalue, 0, 1));
|
| 608 |
|
|
substr ($qvalue, 0, 1) = '';
|
| 609 |
|
|
}
|
| 610 |
|
|
}
|
| 611 |
|
|
_count_lp ('"', $o);
|
| 612 |
|
|
}
|
| 613 |
|
|
sub _parse_entity_value ($$$$) {
|
| 614 |
|
|
my ($self, $attr, $qvalue, $o) = @_;
|
| 615 |
|
|
$qvalue = substr ($qvalue, 1, length ($qvalue) - 2);
|
| 616 |
|
|
_count_lp ('"', $o);
|
| 617 |
|
|
while ($qvalue) {
|
| 618 |
|
|
if ($qvalue =~ s/^($xml_re{Reference})//) {
|
| 619 |
|
|
$self->_parse_reference ($attr, $1, $o);
|
| 620 |
|
|
} elsif ($qvalue =~ s/^$xml_re{PEReference_M}//) {
|
| 621 |
|
|
if ($o->{entity_type} eq 'dtd_internal_subset') {
|
| 622 |
|
|
$self->_raise_error ($o, type => 'WFC_PE_IN_INTERNAL_SUBSET', t => $1);
|
| 623 |
|
|
}
|
| 624 |
|
|
$attr->append_new_node (type => '#reference', local_name => $1,
|
| 625 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:entity:parameter');
|
| 626 |
|
|
_count_lp ($1.'%;', $o);
|
| 627 |
|
|
} elsif ($qvalue =~ s/^([^%&]+)//) {
|
| 628 |
|
|
$attr->append_text ($1);
|
| 629 |
|
|
_count_lp ($1, $o);
|
| 630 |
|
|
} else {
|
| 631 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_INVALID_CHAR', t => substr ($qvalue, 0, 1));
|
| 632 |
|
|
substr ($qvalue, 0, 1) = '';
|
| 633 |
|
|
}
|
| 634 |
|
|
}
|
| 635 |
|
|
_count_lp ('"', $o);
|
| 636 |
|
|
}
|
| 637 |
|
|
sub _parse_parametered_markup_declaration ($$$$) {
|
| 638 |
|
|
my ($self, $attr, $qvalue, $o) = @_;
|
| 639 |
|
|
while ($qvalue) {
|
| 640 |
|
|
if ($qvalue =~ s/^$xml_re{PEReference_M}//) {
|
| 641 |
|
|
if ($o->{entity_type} eq 'dtd_internal_subset') {
|
| 642 |
|
|
$self->_raise_error ($o, type => 'WFC_PE_IN_INTERNAL_SUBSET', t => $1);
|
| 643 |
|
|
}
|
| 644 |
|
|
$attr->append_new_node (type => '#reference', local_name => $1,
|
| 645 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:entity:parameter');
|
| 646 |
|
|
_count_lp ($1.'%;', $o);
|
| 647 |
|
|
} elsif ($qvalue =~ s/^($xml_re{__AttValue_simple})//s) {
|
| 648 |
|
|
$attr->append_new_node (type => '#xml', value => $1); ## TODO: parse and check PEref
|
| 649 |
|
|
_count_lp ($1, $o);
|
| 650 |
|
|
} elsif ($qvalue =~ s/^(\#?$xml_re{Name})//) {
|
| 651 |
|
|
$attr->append_text ($1);
|
| 652 |
|
|
_count_lp ($1, $o);
|
| 653 |
|
|
} elsif ($qvalue =~ s/^($xml_re{s})//s) {
|
| 654 |
|
|
$attr->append_text ($1);
|
| 655 |
|
|
_count_lp ($1, $o);
|
| 656 |
|
|
} else {
|
| 657 |
|
|
$self->_raise_error ($o, type => 'SYNTAX_INVALID_CHAR', t => substr ($qvalue, 0, 1));
|
| 658 |
|
|
substr ($qvalue, 0, 1) = '';
|
| 659 |
|
|
}
|
| 660 |
|
|
}
|
| 661 |
|
|
}
|
| 662 |
|
|
sub _parse_reference ($$$$) {
|
| 663 |
|
|
my ($self, $c, $ref, $o) = @_;
|
| 664 |
|
|
if ($ref =~ /$xml_re{EntityRef_M}/) { ## BUG: QName
|
| 665 |
|
|
$c->append_new_node (type => '#reference', local_name => $1,
|
| 666 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:entity');
|
| 667 |
|
|
} elsif ($ref =~ /x([0-9A-Fa-f]+)/) {
|
| 668 |
|
|
my $ch = hex $1;
|
| 669 |
|
|
if (($ch < 0x20 && $ch != 0x09 && $ch != 0x0A && $ch != 0x0D)
|
| 670 |
|
|
|| (0xD7FF < $ch && $ch < 0xE000) || $ch == 0xFFFE || $ch == 0xFFFF
|
| 671 |
|
|
|| $ch > 0x10FFFF) {
|
| 672 |
|
|
$self->_raise_error ($o, type => 'WFC_LEGAL_CHARACTER', t => $ch);
|
| 673 |
|
|
}
|
| 674 |
|
|
$c->append_new_node (type => '#reference', value => $ch,
|
| 675 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:char:ref:hex');
|
| 676 |
|
|
} elsif ($ref =~ /([0-9]+)/) {
|
| 677 |
|
|
my $ch = $1;
|
| 678 |
|
|
if (($ch < 0x20 && $ch != 0x09 && $ch != 0x0A && $ch != 0x0D)
|
| 679 |
|
|
|| (0xD7FF < $ch && $ch < 0xE000) || $ch == 0xFFFE || $ch == 0xFF
|
| 680 |
|
|
|| $ch > 0x10FFFF) {
|
| 681 |
|
|
$self->_raise_error ($o, type => 'WFC_LEGAL_CHARACTER', t => $ch);
|
| 682 |
|
|
}
|
| 683 |
|
|
$c->append_new_node (type => '#reference', value => $ch,
|
| 684 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:char:ref');
|
| 685 |
|
|
} else {
|
| 686 |
|
|
$self->_raise_error ($o, type => 'UNKNOWN', t => $ref);
|
| 687 |
|
|
}
|
| 688 |
|
|
_count_lp ($ref, $o);
|
| 689 |
|
|
}
|
| 690 |
|
|
sub _parse_xml_declaration ($$$$) {
|
| 691 |
|
|
my ($self, $c, $attrs, $o) = @_;
|
| 692 |
|
|
my $stage = 0; # 0: <?xml, 1: version="", 2: encoding="", 3: standalone="", 4: ?>
|
| 693 |
|
|
$attrs = ' ' . $attrs;
|
| 694 |
|
|
while ($attrs) {
|
| 695 |
|
|
if ($attrs =~ s/^$xml_re{s}version(?:$xml_re{s})?=(?:$xml_re{s})?("[A-Za-z0-9_.:-]+"|'[A-Za-z0-9_.:-]+')//s) {
|
| 696 |
|
|
my $version = substr ($1, 1, length ($1) - 2);
|
| 697 |
|
|
if ($stage > 0) {
|
| 698 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE', t => 'version');
|
| 699 |
|
|
}
|
| 700 |
|
|
if ($version eq '1.0') {
|
| 701 |
|
|
$c->set_attribute (version => $version);
|
| 702 |
|
|
} else {
|
| 703 |
|
|
_raise_fatal_error ($o, desc => 'UNSUPPORTED_XML_VERSION', t => $version);
|
| 704 |
|
|
}
|
| 705 |
|
|
_count_lp ($&, $o); $stage++;
|
| 706 |
|
|
} elsif ($attrs =~ s/^$xml_re{s}encoding(?:$xml_re{s})?=(?:$xml_re{s})?("[A-Za-z0-9_.-]+"|'[A-Za-z0-9_.:-]+')//s) {
|
| 707 |
|
|
if ($stage > 2) {
|
| 708 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE', t => 'encoding');
|
| 709 |
|
|
} elsif ($stage == 0) { ## No version pseudo-attr
|
| 710 |
|
|
if ($o->{entity_type} eq 'document_entity') {
|
| 711 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE_NO_ATTR', t => 'version');
|
| 712 |
|
|
$c->set_attribute (version => '1.0');
|
| 713 |
|
|
} else {
|
| 714 |
|
|
$o->{entity_type} = 'external_parsed_entity';
|
| 715 |
|
|
}
|
| 716 |
|
|
}
|
| 717 |
|
|
$c->set_attribute (encoding => substr ($1, 1, length ($1) - 2));
|
| 718 |
|
|
_count_lp ($&, $o); $stage = 2;
|
| 719 |
|
|
} elsif ($attrs =~ s/^$xml_re{s}standalone(?:$xml_re{s})?=(?:$xml_re{s})?("(?:yes|no)"|'(?:yes|no)')//s) {
|
| 720 |
|
|
if ($stage == 0 || $stage > 3 || $o->{entity_type} eq 'external_parsed_entity'
|
| 721 |
|
|
|| $o->{entity_type} eq 'dtd_external_subset'
|
| 722 |
|
|
|| $o->{entity_type} eq 'general_external_parsed_entity') {
|
| 723 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE_NO_ATTR', t => 'standalone');
|
| 724 |
|
|
}
|
| 725 |
|
|
$c->set_attribute (standalone => (substr ($1, 1, 1) eq 'y' ? 'yes' : 'no'));
|
| 726 |
|
|
_count_lp ($&, $o); $stage = 3;
|
| 727 |
|
|
} elsif ($attrs =~ s/^($xml_re{s})//s) {
|
| 728 |
|
|
my $s = $1;
|
| 729 |
|
|
if ($stage == 0) {
|
| 730 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE_NO_ATTR');
|
| 731 |
|
|
$c->set_attribute (version => '1.0');
|
| 732 |
|
|
}
|
| 733 |
|
|
_count_lp ($s, $o); $stage = 4;
|
| 734 |
|
|
} else {
|
| 735 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE', t => $attrs);
|
| 736 |
|
|
_count_lp ($attrs, $o); undef $attrs;
|
| 737 |
|
|
}
|
| 738 |
|
|
} # while
|
| 739 |
|
|
if ($stage == 0) {
|
| 740 |
|
|
$self->_raise_error ($o, c => $attrs, type => 'SYNTAX_XML_DECLARE_NO_ATTR');
|
| 741 |
|
|
$c->set_attribute (version => '1.0');
|
| 742 |
|
|
}
|
| 743 |
|
|
}
|
| 744 |
|
|
sub _parse_entity_declaration ($$$$) {
|
| 745 |
|
|
my ($self, $all, $c, $o) = @_;
|
| 746 |
|
|
my ($e, $p) = (undef, 0);
|
| 747 |
|
|
$e = $c->append_new_node (type => '#declaration');
|
| 748 |
|
|
## Entity/notation type
|
| 749 |
|
|
if ($all =~ s/^<!ENTITY$xml_re{s}(%$xml_re{s})?//s) {
|
| 750 |
|
|
$p = $1;
|
| 751 |
|
|
if ($p) {
|
| 752 |
|
|
$e->namespace_uri ('urn:x-suika-fam-cx:markup:sgml:entity:parameter');
|
| 753 |
|
|
} else {
|
| 754 |
|
|
$e->namespace_uri ('urn:x-suika-fam-cx:markup:sgml:entity');
|
| 755 |
|
|
}
|
| 756 |
|
|
_count_lp ($&, $o);
|
| 757 |
|
|
} elsif ($all =~ s/^<!NOTATION$xml_re{s}//s) {
|
| 758 |
|
|
$e->namespace_uri ('urn:x-suika-fam-cx:markup:sgml:notation');
|
| 759 |
|
|
_count_lp ($&, $o); $p = 'n';
|
| 760 |
|
|
}
|
| 761 |
|
|
## Entity/notation name
|
| 762 |
|
|
if ($all =~ s/^($xml_re{Name})//s) {
|
| 763 |
|
|
$e->local_name ($1);
|
| 764 |
|
|
_count_lp ($1, $o);
|
| 765 |
|
|
}
|
| 766 |
|
|
## EntityValue (ENTITY only)
|
| 767 |
|
|
if ($p ne 'n' && ($all =~ s/^$xml_re{s}($xml_re{__AttValue_simple})//s)) {
|
| 768 |
|
|
_count_lp (length ($&) - length ($1), $o);
|
| 769 |
|
|
$self->_parse_entity_value ($e->set_attribute ('value'), $1, $o);
|
| 770 |
|
|
## System ID
|
| 771 |
|
|
} elsif ($all =~ s/^$xml_re{s}SYSTEM$xml_re{s}($xml_re{SystemLiteral})//s) {
|
| 772 |
|
|
$e->set_attribute (SYSTEM => substr ($1, 1, length ($1) - 2));
|
| 773 |
|
|
_count_lp ($&, $o); ## TODO: PEref
|
| 774 |
|
|
## Public ID + System ID
|
| 775 |
|
|
} elsif ($all =~ s/^$xml_re{s}PUBLIC$xml_re{s}($xml_re{PublicLiteral})$xml_re{s}($xml_re{SystemLiteral})//s) {
|
| 776 |
|
|
$e->set_attribute (PUBLIC => substr ($1, 1, length ($1) - 2));
|
| 777 |
|
|
$e->set_attribute (SYSTEM => substr ($2, 1, length ($1) - 2));
|
| 778 |
|
|
_count_lp ($&, $o);
|
| 779 |
|
|
## Public ID (NOTATION only)
|
| 780 |
|
|
} elsif ($p eq 'n' && ($all =~ s/^$xml_re{s}PUBLIC$xml_re{s}($xml_re{PublicLiteral})//s)) {
|
| 781 |
|
|
$e->set_attribute (PUBLIC => substr ($1, 1, length ($1) - 2));
|
| 782 |
|
|
_count_lp ($&, $o);
|
| 783 |
|
|
## Including parameter entity reference
|
| 784 |
|
|
} elsif ($all =~ s/^(?:$xml_re{s})?(?:$xml_re{PEReference}|$xml_re{Name}|$xml_re{__AttValue_simple})(?:$xml_re{s}|$xml_re{PEReference}|$xml_re{Name}|$xml_re{__AttValue_simple})*//s) {
|
| 785 |
|
|
$self->_parse_parametered_markup_declaration ($e, $&, $o);
|
| 786 |
|
|
} else {
|
| 787 |
|
|
_raise_fatal_error ($o, desc => 'INVALID_DECLARE_SYNTAX', t => $all);
|
| 788 |
|
|
}
|
| 789 |
|
|
## NDATA (General entity only)
|
| 790 |
|
|
if (!$p && $all =~ s/^$xml_re{s}NDATA$xml_re{s}($xml_re{Name})//s) {
|
| 791 |
|
|
$e->set_attribute (NDATA => $1);
|
| 792 |
|
|
_count_lp ($&, $o);
|
| 793 |
|
|
}
|
| 794 |
|
|
if ($all =~ s/^((?:$xml_re{s})?>)$//s) {
|
| 795 |
|
|
_count_lp ($1, $o);
|
| 796 |
|
|
} else {
|
| 797 |
|
|
_raise_fatal_error ($o, desc => 'INVALID_DECLARE_SYNTAX', t => $all);
|
| 798 |
|
|
}
|
| 799 |
|
|
}
|
| 800 |
|
|
|
| 801 |
|
|
sub _parse_element_declaration ($$$$) {
|
| 802 |
|
|
my ($self, $all, $c, $o) = (@_);
|
| 803 |
|
|
my $e = undef;
|
| 804 |
|
|
$e = $c->append_new_node (type => '#declaration',
|
| 805 |
|
|
namespace_uri => 'urn:x-suika-fam-cx:markup:sgml:element');
|
| 806 |
|
|
$all =~ s/^<!ELEMENT//s;
|
| 807 |
|
|
_count_lp ('<!ELEMENT', $o);
|
| 808 |
|
|
## Element type name
|
| 809 |
|
|
if ($all =~ s/^$xml_re{s}($xml_re{Name})//s) {
|
| 810 |
|
|
$e->set_attribute (qname => $1);
|
| 811 |
|
|
_count_lp ($&, $o);
|
| 812 |
|
|
}
|
| 813 |
|
|
## contentspec / PEReference
|
| 814 |
|
|
if ($all =~ s/^(?:$xml_re{s})?(?:$xml_re{PEReference}|$xml_re{Name}|\#PCDATA|\()(?:$xml_re{s}|$xml_re{PEReference}|$xml_re{Name}|\#PCDATA|[()|,+*?])*//s) {
|
| 815 |
|
|
$e->append_new_node (type => '#xml', value => $&); # TODO: temporary
|
| 816 |
|
|
_count_lp ($&, $o);
|
| 817 |
|
|
}
|
| 818 |
|
|
if ($all =~ s/^((?:$xml_re{s})?>)$//s) {
|
| 819 |
|
|
_count_lp ($1, $o);
|
| 820 |
|
|
} else {
|
| 821 |
|
|
_raise_fatal_error ($o, desc => 'INVALID_DECLARE_SYNTAX', t => $all);
|
| 822 |
|
|
}
|
| 823 |
|
|
}
|
| 824 |
|
|
|
| 825 |
|
|
|
| 826 |
|
|
sub _parse_attlist_declaration ($$$$) {
|
| 827 |
|
|
my ($self, $all, $c, $o) = (@_);
|
| 828 |
|
|
my $e = undef;
|
| 829 |
|
|
$e = $c->append_new_node (type => '#declaration', local_name => 'ATTLIST');
|
| 830 |
|
|
$all =~ s/^<!ATTLIST//s;
|
| 831 |
|
|
_count_lp ('<!ATTLIST', $o);
|
| 832 |
|
|
## Element type name
|
| 833 |
|
|
if ($all =~ s/^$xml_re{s}($xml_re{Name})//s) {
|
| 834 |
|
|
$e->target_name ($1);
|
| 835 |
|
|
_count_lp ($&, $o);
|
| 836 |
|
|
}
|
| 837 |
|
|
## Definition
|
| 838 |
|
|
if ($all =~ s/^(?:$xml_re{PEReference}|$xml_re{Name}|\#$xml_re{Name}|$xml_re{s}|$xml_re{__AttValue_simple})+//s) {
|
| 839 |
|
|
$e->append_new_node (type => '#xml', value => $&); # TODO: temporary
|
| 840 |
|
|
_count_lp ($&, $o);
|
| 841 |
|
|
}
|
| 842 |
|
|
if ($all =~ s/^((?:$xml_re{s})?>)$//s) {
|
| 843 |
|
|
_count_lp ($1, $o);
|
| 844 |
|
|
} else {
|
| 845 |
|
|
_raise_fatal_error ($o, desc => 'INVALID_DECLARE_SYNTAX', t => $all);
|
| 846 |
|
|
}
|
| 847 |
|
|
}
|
| 848 |
|
|
|
| 849 |
|
|
## TODO: remove this function. this function is already obsoleted.
|
| 850 |
|
|
sub _raise_fatal_error ($%) {
|
| 851 |
|
|
require Carp;
|
| 852 |
|
|
my ($o, %o) = @_;
|
| 853 |
|
|
$o{desc} = {
|
| 854 |
|
|
INVALID_CHAR => 'Invalid character (%s) at this context',
|
| 855 |
|
|
INVALID_XML_DECLARE => 'XML declaration must be at the top of the entity',
|
| 856 |
|
|
NONDECLARED_NS_PREFIX => 'Undeclared namespace prefix (%s) is used',
|
| 857 |
|
|
NONMATCH_ETAG => 'End tag (element type name = '.$o{_end_qname}.') does not match with start tag (element type name = '.$o{_start_qname}.')',
|
| 858 |
|
|
NOT_ALLOWED_HERE => 'This type of markup (%s) cannot appear here',
|
| 859 |
|
|
NOT_ALLOWED_THIS_MODE => 'This type of markup (%s) cannot be used '.({
|
| 860 |
|
|
document_entity => 'out of DTD',
|
| 861 |
|
|
dtd_internal_subset => 'in internal subset of DTD',
|
| 862 |
|
|
dtd_external_subset => 'in external parsed entity (external subset of DTD)',
|
| 863 |
|
|
}->{$o->{entity_type}||'document_entity'}||'in '.$o->{entity_type}),
|
| 864 |
|
|
UNKNOWN => 'Unknown error (%s)',
|
| 865 |
|
|
}->{$o{desc}} || $o{desc};
|
| 866 |
|
|
$o{desc} .= ' (%s)' if length $o{t} && $o{desc} !~ /%s/;
|
| 867 |
|
|
Carp::croak ("Line $o->{line}, position $o->{pos}: ".sprintf $o{desc}, $o{t});
|
| 868 |
|
|
}
|
| 869 |
|
|
|
| 870 |
|
|
sub _is_brother_of_root_element ($$) {
|
| 871 |
|
|
my ($self, $c) = @_;
|
| 872 |
|
|
for (@{$c->child_nodes}) {
|
| 873 |
|
|
if ($_->node_type eq '#element') {
|
| 874 |
|
|
return 1;
|
| 875 |
|
|
}
|
| 876 |
|
|
}
|
| 877 |
|
|
return 0;
|
| 878 |
|
|
}
|
| 879 |
|
|
|
| 880 |
|
|
my %_Error = (
|
| 881 |
|
|
## Syntax errors
|
| 882 |
|
|
SYNTAX_DATA_OUT_OF_ROOT_ELEMENT => {
|
| 883 |
|
|
description => 'Invalid data or markup out of root element',
|
| 884 |
|
|
level => 'wfc',
|
| 885 |
|
|
},
|
| 886 |
|
|
SYNTAX_END_OF_MARKUP_NOT_FOUND => {
|
| 887 |
|
|
description => sub {
|
| 888 |
|
|
my ($o, $err) = @_;
|
| 889 |
|
|
my $o_start = $err->{t}->flag ('p_o_start');
|
| 890 |
|
|
my $r = $err->{t}->qname;
|
| 891 |
|
|
$r = sprintf 'line %d, position %d%s', $o_start->{line},
|
| 892 |
|
|
$o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
|
| 893 |
|
|
$r ? $r = '; '.$r : 0;
|
| 894 |
|
|
$err->{t} = $err->{t}->node_type;
|
| 895 |
|
|
'Markup is not closed (%s'.$r.')';
|
| 896 |
|
|
},
|
| 897 |
|
|
level => 'wfc',
|
| 898 |
|
|
},
|
| 899 |
|
|
SYNTAX_END_TAG_NOT_FOUND => {
|
| 900 |
|
|
description => sub {
|
| 901 |
|
|
my ($o, $err) = @_;
|
| 902 |
|
|
my $o_start = $err->{t}->flag ('p_o_start');
|
| 903 |
|
|
my $r = '';
|
| 904 |
|
|
$r = sprintf 'line %d, position %d%s', $o_start->{line},
|
| 905 |
|
|
$o_start->{pos}, ($r ? '; '.$r : '') if ref $o_start;
|
| 906 |
|
|
$r ? $r = '; '.$r : 0;
|
| 907 |
|
|
$err->{t} = $err->{t}->qname;
|
| 908 |
|
|
'End tag of element (type = %s'.$r.') not found';
|
| 909 |
|
|
},
|
| 910 |
|
|
level => 'wfc',
|
| 911 |
|
|
},
|
| 912 |
|
|
SYNTAX_INVALID_CHAR => {
|
| 913 |
|
|
description => 'Invalid character (%s) at this context',
|
| 914 |
|
|
level => 'wfc',
|
| 915 |
|
|
},
|
| 916 |
|
|
SYNTAX_INVALID => {
|
| 917 |
|
|
description => 'This type of markup (%s) cannot appear here',
|
| 918 |
|
|
level => 'wfc',
|
| 919 |
|
|
},
|
| 920 |
|
|
SYNTAX_INVALID_DOCTYPE_POSITION => {
|
| 921 |
|
|
description => 'DOCTYPE declaration must be between xml declaration and the root element',
|
| 922 |
|
|
level => 'wfc',
|
| 923 |
|
|
},
|
| 924 |
|
|
SYNTAX_INVALID_POSITION => {
|
| 925 |
|
|
description => sub {
|
| 926 |
|
|
my ($o) = shift;
|
| 927 |
|
|
'This type of markup (%s) cannot be used '.({
|
| 928 |
|
|
document_entity => 'out of DTD',
|
| 929 |
|
|
dtd_internal_subset => 'in the internal subset of DTD',
|
| 930 |
|
|
dtd_external_subset => 'in the external subset of DTD',
|
| 931 |
|
|
external_parsed_entity => 'in the external parsed entity',
|
| 932 |
|
|
general_external_parsed_entity => 'in the general external parsed entity',
|
| 933 |
|
|
}->{$o->{entity_type}||'document_entity'}||'in '.$o->{entity_type})},
|
| 934 |
|
|
level => 'wfc',
|
| 935 |
|
|
},
|
| 936 |
|
|
SYNTAX_LEGAL_CHARACTER => {
|
| 937 |
|
|
description => sub {
|
| 938 |
|
|
my $r = sprintf 'The character U-%08X is not a legal XML Char',
|
| 939 |
|
|
$_[1]->{t};
|
| 940 |
|
|
$_[1]->{t} = undef;
|
| 941 |
|
|
$r;
|
| 942 |
|
|
},
|
| 943 |
|
|
level => 'wfc',
|
| 944 |
|
|
},
|
| 945 |
|
|
SYNTAX_XML_DECLARE => {
|
| 946 |
|
|
description => 'Syntax of XML (or text) declaration is invalid',
|
| 947 |
|
|
level => 'wfc',
|
| 948 |
|
|
},
|
| 949 |
|
|
SYNTAX_XML_DECLARE_NO_ATTR => {
|
| 950 |
|
|
description => 'XML (or text) declaration does not have pseudo attribute',
|
| 951 |
|
|
level => 'wfc',
|
| 952 |
|
|
},
|
| 953 |
|
|
SYNTAX_XML_DECLARE_POSITION => {
|
| 954 |
|
|
description => 'XML declaration must be at the top of the entity',
|
| 955 |
|
|
level => 'wfc',
|
| 956 |
|
|
},
|
| 957 |
|
|
## Well-formedness error
|
| 958 |
|
|
WFC_ELEMENT_TYPE_MATCH => {
|
| 959 |
|
|
description => 'End tag (type = %s) does not match with start tag (type = %s)',
|
| 960 |
|
|
level => 'wfc',
|
| 961 |
|
|
},
|
| 962 |
|
|
WFC_LEGAL_CHARACTER => {
|
| 963 |
|
|
description => sub {
|
| 964 |
|
|
my $r = sprintf 'The character referred (U-%08X) is not a legal XML Char',
|
| 965 |
|
|
$_[1]->{t};
|
| 966 |
|
|
$_[1]->{t} = undef;
|
| 967 |
|
|
$r;
|
| 968 |
|
|
},
|
| 969 |
|
|
level => 'wfc',
|
| 970 |
|
|
},
|
| 971 |
|
|
WFC_NO_LT_IN_ATTRIBUTE_VALUE => {
|
| 972 |
|
|
description => 'Replacement text of entity reference in an attribute value literal cannot contain LESS-THAN SIGN (<) itself',
|
| 973 |
|
|
level => 'wfc',
|
| 974 |
|
|
},
|
| 975 |
|
|
WFC_PE_IN_INTERNAL_SUBSET => {
|
| 976 |
|
|
description => 'Parameter entity (%s) cannot be referred in markup declaration in internal subset of DTD',
|
| 977 |
|
|
level => 'wfc',
|
| 978 |
|
|
},
|
| 979 |
|
|
WFC_UNIQUE_ATT_SPEC => {
|
| 980 |
|
|
description => 'Dupulicate attribute specification',
|
| 981 |
|
|
level => 'wfc',
|
| 982 |
|
|
},
|
| 983 |
|
|
## Validity error
|
| 984 |
|
|
## Namespace well-formedness error
|
| 985 |
|
|
NC_PREFIX_NOT_DEFINED => {
|
| 986 |
|
|
description => 'Undeclared namespace prefix (%s) is used',
|
| 987 |
|
|
level => 'nc',
|
| 988 |
|
|
},
|
| 989 |
|
|
## Namespace validity error
|
| 990 |
|
|
## Misc
|
| 991 |
|
|
UNKNOWN => {
|
| 992 |
|
|
description => 'Unknown error',
|
| 993 |
|
|
level => 'wfc',
|
| 994 |
|
|
},
|
| 995 |
|
|
);
|
| 996 |
|
|
## TODO: error handling should be customizable (hookable) by user of this module
|
| 997 |
|
|
sub _raise_error ($$%) {
|
| 998 |
|
|
my ($self, $o, %err) = @_;
|
| 999 |
|
|
my $error_type = $_Error{$err{type}} || $_Error{UNKNOWN};
|
| 1000 |
|
|
my $error_msg = ref $error_type->{description} ? &{$error_type->{description}} ($o, \%err)
|
| 1001 |
|
|
: $error_type->{description};
|
| 1002 |
|
|
my @err_msg;
|
| 1003 |
|
|
ref $err{t} eq 'ARRAY' ? @err_msg = @{$err{t}} : defined $err{t} ? @err_msg = $err{t} : undef;
|
| 1004 |
|
|
$error_msg .= ' (%s)' if scalar (@err_msg) && ($error_msg !~ /%s/);
|
| 1005 |
|
|
$error_msg = sprintf $error_msg, @err_msg;
|
| 1006 |
|
|
require Carp;
|
| 1007 |
|
|
Carp::carp ("Line $o->{line}, position $o->{pos}: ".$error_msg);
|
| 1008 |
|
|
#Carp::croak ("Line $o->{line}, position $o->{pos}: ".$error_msg);
|
| 1009 |
|
|
}
|
| 1010 |
|
|
|
| 1011 |
|
|
=head1 LICENSE
|
| 1012 |
|
|
|
| 1013 |
|
|
Copyright 2003 Wakaba <[email protected]>
|
| 1014 |
|
|
|
| 1015 |
|
|
This program is free software; you can redistribute it and/or
|
| 1016 |
|
|
modify it under the same terms as Perl itself.
|
| 1017 |
|
|
|
| 1018 |
|
|
=cut
|
| 1019 |
|
|
|
| 1020 |
|
|
1; # $Date: 2003/05/10 05:58:06 $
|
Parent Directory
| 
Revision Log