/[suikacvs]/messaging/manakai/lib/Message/Markup/XML/Parser.pm
Suika

Contents of /messaging/manakai/lib/Message/Markup/XML/Parser.pm

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1.17 - (hide annotations) (download)
Wed Sep 17 09:17:13 2003 UTC (22 years, 10 months ago) by wakaba
Branch: MAIN
Changes since 1.16: +19 -17 lines 
Bug fix
1 wakaba 1.1
2     =head1 NAME
3    
4 wakaba 1.12 Message::Markup::XML::Parser --- manakai: Simple XML parser
5 wakaba 1.1
6     =head1 DESCRIPTION
7    
8 wakaba 1.12 This is a simple XML parser intended to be used with Message::Markup::XML.
9     After parsing of the XML document, this module returns a Message::Markup::XML
10 wakaba 1.1 object so that you can handle XML document with that module (and other modules
11     implementing same interface).
12    
13 wakaba 1.12 This module is part of manakai.
14 wakaba 1.1
15     =cut
16    
17 wakaba 1.12 package Message::Markup::XML::Parser;
18 wakaba 1.1 use strict;
19 wakaba 1.17 our $VERSION = do{my @r=(q$Revision: 1.16 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
20 wakaba 1.6 use Char::Class::XML qw!InXML_NameStartChar InXMLNameChar InXMLChar
21     InXML_deprecated_noncharacter InXML_unicode_xml_not_suitable!;
22 wakaba 1.12 require Message::Markup::XML;
23     require Message::Markup::XML::Error;
24     *_raise_error = \&Message::Markup::XML::Error::raise;
25 wakaba 1.6 our %NS;
26 wakaba 1.12 *NS = \%Message::Markup::XML::NS;
27 wakaba 1.1
28     =head1 METHODS
29    
30     WARNING: This module is under construction. Interface of this module is not yet fixed.
31    
32     =cut
33    
34 w 1.10 our %xml_re;
35 wakaba 1.1 # [1] document = prolog element *Misc
36     # [2] Char = %x09 / %x0A / %x0D / U+0020-U+D7FF / U+E000-U+FFFD / U+10000-U+10FFFF ;; 1.0
37     # [2] Char = %x09 / %x0A / %x0D / %x20-7E / U+0085 / U+00A0-U+D7FF / U+E000-U+FFFD
38     # / U+10000-U+10FFFF ;; 1.1
39     # [3] s = 1*(%x20 / %x09 / %x0D / %x0A)
40 wakaba 1.6 $xml_re{s} = qr/[\x09\x0A\x0D\x20]+/s;
41 w 1.8 $xml_re{_s__chars} = qr/\x09\x0A\x0D\x20/s;
42 wakaba 1.1 # [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     # $xml_re{NameChar} = qr/[A-Za-z0-9._:-]|[^\x00-\x7F]/;
49 wakaba 1.4 #$xml_re{NameChar} = qr/\p{InXMLNameChar}/;
50 wakaba 1.1 # [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     $xml_re{Name} = qr/\p{InXML_NameStartChar}\p{InXMLNameChar}*/;
54     # [6] Names = Name *(S Name) ;; 1.0 FE & 1.0 FE-errata (2000-09-27) & 1.0 SE
55     # [6] Names = Name *(%x20 Name) ;; 1.0 FE-errata (2000-04-09) & 1.0 SE-errata
56 wakaba 1.4 #$xml_re{Names} = qr/$xml_re{Name}(?:$xml_re{s}$xml_re{Name})*/s;
57 wakaba 1.1 # [7] Nmtoken = 1*NameChar
58     #$xml_re{Nmtoken} = qr/(?:$xml_re{NameChar})+/;
59 wakaba 1.4 #$xml_re{Nmtoken} = qr/\p{InXMLNameChar}+/;
60 wakaba 1.1 # [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 wakaba 1.4 #$xml_re{Nmtokens} = qr/$xml_re{Nmtoken}(?:$xml_re{s}$xml_re{Nmtoken})*/s;
63 wakaba 1.1 # [9] EntityValue = <"> *((Char - ("%" / "&" / <">)) / PEReference / Reference) <">
64     # / "'" *((Char - ("%" / "&" / "'")) / PEReference / Reference) "'"
65     # [10] AttValue = <"> *((Char - ("<" / "&" / <">)) / Reference) <">
66     # / "'" *((Char - ("<" / "&" / "'")) / Reference) "'"
67     $xml_re{__AttValue_simple} = qr/"[^"]*"|'[^']*'/s;
68     # [11] SystemLiteral = <"> *(Char - <">) <"> / "'" *(Char - "'") "'"
69 wakaba 1.4 #$xml_re{SystemLiteral} = qr/"[^"]*"|'[^']*'/;
70 wakaba 1.1 # [12] PublicLiteral = <"> *PubidChar <"> / "'" *(PubidChar - "'") "'"
71     # [13] PubidChar = %x20 / %x0D / %x0A / ALPHA / DIGIT / "-" / "'" / "(" / ")"
72     # / "+" / "," / "." / "/" / ":" / "=" / "?" / ";" / "!" / "*"
73     # / "#" / "@" / "$" / "_" / "%"
74 wakaba 1.6 #$xml_re{PubidChar} = qr[[\x0D\x0A\x20!\x24#%'()+*,./0-9:;=?\x40A-Z_a-z-]];
75     #$xml_re{__non_PubidChar} = qr[[^\x0D\x0A\x20!\x24#%'()+*,./0-9:;=?\x40A-Z_a-z-]];
76 wakaba 1.4 #$xml_re{__PubidChar2} = qr[[\x0D\x0A\x20!\x24#%()+*,./0-9:;=?\x40A-Z_a-z-]];
77     #$xml_re{PublicLiteral} = qr/"(?:$xml_re{PubidChar})*"|'(?:$xml_re{__PubidChar2})*'/;
78 wakaba 1.1 # [14] CharData = *(Char - ("<" / "&")) - (*(Char - ("<" / "&")) "]]>" *(Char - ("<" / "&")))
79 wakaba 1.4 #$xml_re{CharData} = qr/(?:(?!\]\]>)[^<&])*/s;
80 wakaba 1.1 $xml_re{__CharDataP} = qr/(?:(?!\]\]>)[^<&])+/s;
81     # [15] Comment = "<!--" *((Char - "-") / ("-" (Char - "-")))) "-->"
82 wakaba 1.16 #$xml_re{Comment_M} = qr/<!--((?:(?!--).)*)-->/s;
83 wakaba 1.1 # [16] PI = "<?" PITarget [S (*Char - (*Char "?>" *Char))] "?>"
84     $xml_re{PI_M} = qr/<\?($xml_re{Name})(?:$xml_re{s}((?:(?!\?>).)*))?\?>/s;
85 wakaba 1.4 $xml_re{_xml_PI_M} = qr/<\?xml(?:$xml_re{s}((?:(?!\?>).)*))?\?>/s;
86 wakaba 1.1 # [17] PITarget = Name - "xml"
87     # [18] CDSect = CDStart CData CDEnd
88     # [19] CDStart = '<![CDATA['
89     # [20] CDATA = (*Char - (*Char "]]>" *Char))
90     # [21] CDEnd = "]]>"
91     $xml_re{CDSect_M} = qr/<!\[CDATA\[((?:(?!\]\]>).)*)\]\]>/s;
92     # [22] prolog = [XMLDecl] *Misc [doctypedecl *Misc]
93     # [23] XMLDecl = '<?xml' VersionInfo [EncodingDecl] [SDDecl] [S] "?>"
94     # [24] VersionInfo = S 'version' Eq ("'" VersionNum "'" / <"> VersionNum <">)
95     # [25] Eq = [S] "=" [S]
96     # [26] VersionNum = 1*(ALPHA / DIGIT / "_" / "." / ":" / "-") ;; 1.0 FE & 1.0 SE
97     # [26] VersionNum = "1.0" ;; 1.0 SE-errata
98     # [26] VersionNum = "1.1" ;; 1.1
99     # [27] Misc = Comment / PI / S
100     # [28] doctypedecl = '<!DOCTYPE' S Name [S ExternalID] [S]
101     # ["[" *(markupdecl / PEReference / S) "]" [S]] ">" ;; 1.0 FE
102     # [28] doctypedecl = '<!DOCTYPE' S Name [S ExternalID] [S] ["[" *(markupdecl / DeclSep) "]" [S]] ">"
103     # ;; 1.0 FE-errata & 1.0 SE
104     # [28] doctypedecl = '<!DOCTYPE' S Name [S ExternalID] [S] ["[" intSubset "]" [S]] ">"
105     # ;; 1.0 SE-errata
106 wakaba 1.6 #$xml_re{__doctypedecl_start_simple_M} = qr/(<!DOCTYPE$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;
107 wakaba 1.1 # [28a] DeclSep = PEReference / S ;; 1.0 FE-errata & 1.0 SE
108     # [28b] intSubset = *(markupdecl / DeclSep) ;; 1.0 SE-errata
109     # [29] markupdecl = elementdecl / AttlistDecl / EntityDecl / NotationDecl / PI / Comment
110     # [30] extSubset = [TextDecl] extSubsetDecl
111     # [31] extSubsetDecl = *(markupdecl / conditionalSect / PEReference / S) ;; 1.0 FE
112     # [31] extSubsetDecl = *(markupdecl / conditionalSect / DeclSep) ;; 1.0 FE-errata & 1.0 SE
113     # [32] SDDecl = S 'standalone' Eq ("'" ('yes' / 'no') "'" / <"> ('yes' / 'no') <">)
114     # [33] LanguageID = Langcode *("-" Subcode) ;; 1.0 FE (removed by 1.0 SE)
115     # [34] Langcode = ISO639Cocde / IanaCode / UserCode ;; 1.0 FE (ditto)
116     # [35] ISO639Code = 2ALPHA ;; 1.0 FE (ditto)
117     # [36] IanaCode = "i-" 1*ALPHA ;; 1.0 FE (ditto)
118     # [37] UserCode = "x-" 1*ALPHA ;; 1.0 FE (ditto)
119     # [38] Subcode = 1*ALPHA ;; 1.0 FE (ditto)
120     # [39] element = EmptyElemTag / STag content ETag
121     # [40] STag = "<" Name *(S Attribute) [S] ">"
122     # [41] Attribute = Name Eq AttValue
123     $xml_re{Attribute} = qr/$xml_re{Name}(?:$xml_re{s})?=(?:$xml_re{s})?$xml_re{__AttValue_simple}/s;
124     $xml_re{Attribute_M} = qr/($xml_re{Name})(?:$xml_re{s})?=(?:$xml_re{s})?($xml_re{__AttValue_simple})/s;
125 wakaba 1.4 #$xml_re{STag} = qr/<$xml_re{Name}(?:$xml_re{s}$xml_re{Attribute})*(?:$xml_re{s})?>/s;
126 wakaba 1.1 # [42] ETag = "</" Name [S] ">"
127     $xml_re{ETag_M} = qr!</($xml_re{Name})(?:$xml_re{s})?>!s;
128     # [43] content = *(element / CharData / Reference / CDSect / PI / Comment) ;; 1.0 FE
129     # [43] content = [CharData] *((element / Reference / CDSect / PI / Comment) [CharData])
130     # ;; 1.0 FE-errata & 1.0 SE
131     # [44] EmptyElemTag = "<" Name *(S Attribute) [S] "/>"
132 wakaba 1.4 #$xml_re{__STag_or_EmptyElemTag} = qr!<$xml_re{Name}(?:$xml_re{s}$xml_re{Attribute})*(?:$xml_re{s})?/?>!s;
133 wakaba 1.1 $xml_re{__STag_or_EmptyElemTag_simple} = qr!<$xml_re{ame}(?:$xml_re{s}|$xml_re{Name}|$xml_re{__AttValue_simple}|=)*/?>!s;
134     # [45] elementdecl = '<!ELEMENT' S Name S contentspec [S] ">"
135     # [46] contentspec = 'EMPTY' / 'ANY' / Mixed / children
136 wakaba 1.4 #$xml_re{__contentspec_simple} = qr/(?:$xml_re{Name}|\#PCDATA|[()|,?*+]|$xml_re{s})/s;
137 wakaba 1.1 # [47] children = (choice / seq) ["?" / "*" / "+"]
138     # [48] cp = (Name / choice / seq) ["?" / "*" / "+"]
139     # [49] choice = "(" [S] cp *([S] "|" [S] cp) [S] ")" ;; 1.0 FE
140     # [49] choice = "(" [S] cp 1*([S] "|" [S] cp) [S] ")" ;; 1.0 FE-errata & 1.0 SE
141     # [50] seq = "(" [S] cp *([S] "," [S] cp) [S] ")"
142     # [51] Mixed = "(" [S] '#PCDATA' *([S] "|" [S] Name) [S] ")*"
143     # / "(" [S] '#PCDATA' [S] ")"
144     #$xml_re{seq} = qr/\($xml_re{cp}(?:(?:$xml_re{s})?,(?:$xml_re{s})?$xml_re{cp})*(?:$xml_re{s})?\)/;
145     #$xml_re{cp} = qr/(?:$xml_re{Name}|$xml_re{choice}|$xml_re{seq})[?*+]/;
146     #$xml_re{choice} = qr/\($xml_re{cp}(?:(?:$xml_re{s})?\|(?:$xml_re{s})?$xml_re{cp})+(?:$xml_re{s})?\)/;
147     #$xml_re{children} = qr/(?:$xml_re{choice}|$xml_re{seq})[?*+]/;
148     #$xml_re{Mixed} = qr/(?:\((?:$xml_re{s})?\#PCDATA(?:$xml_re{s})?(?:(?:$xml_re{s})?|(?:$xml_re{s})?$xml_re{Name})*(?:$xml_re{s})?\)|\((?:$xml_re{s})?\#PCDATA(?:$xml_re{s})?\))/;
149     #$xml_re{contentspec} = qr/(?:EMPTY|ANY|$xml_re{Mixed}|$xml_re{children})/;
150     # [52] AttlistDecl = '<!ATTLIST' S Name *AttDef [S] ">"
151     # [53] AttDef = S Name S AttType S DefaultDecl
152     # [54] AttType = StringType / TokenizedType / EnumeratedType
153     # [55] StringType = 'CDATA'
154     # [56] TokenizedType = 'ID' / 'IDREF' / 'IDREFS' / 'ENTITY' / 'ENTITIES' / 'NMTOKEN' / 'NMTOKENS'
155     # [57] EnumeratedType = NotationType / Enumeration
156     # [58] NotationType = 'NOTATION' S "(" [S] Name *([S] "|" [S] Name) [S] ")"
157     # [59] Enumeration = "(" [S] Nmtoken *([S] "|" [S] Nmtoken) [S} ")"
158     # [60] DefaultDecl = '#REQUIRED' / '#IMPLIED' / ['#FIXED' S] AttValue
159     # [61] conditionalSect = includeSect / ignoreSect
160     # [62] includeSect = "<![" [S} 'INCLUDE' [S] "[" extSubsetDecl "]]>"
161     # [63] ignoreSect = "<![" [S] 'IGNORE' [S] "[" *ignoreSectContents "]]>"
162     # [64] ignoreSectContents = Ignore *("<![" ignoreSectContents "]]>" Ignore)
163     # [65] Ignore = *Char - (*Char ("<![" / "]]>") *Char)
164     # [66] CharRef = '&#' 1*DIGIT ";" / '&#x' 1*HEXDIGIT ";"
165     $xml_re{CharRef} = qr/&#[0-9]+;|&#x[0-9A-Fa-f]+;/;
166     # [67] Reference = EntityRef / CharRef
167     # [68] EntityRef = "&" Name ";"
168     $xml_re{EntityRef} = qr/&$xml_re{Name};/;
169     $xml_re{EntityRef_M} = qr/&($xml_re{Name});/;
170     $xml_re{Reference} = qr/$xml_re{EntityRef}|$xml_re{CharRef}/;
171 wakaba 1.4 #$xml_re{AttValue} = qr/"(?:$xml_re{Reference}|[^&<"])*"|'(?:$xml_re{Reference}|[^&<'])*'/s;
172 wakaba 1.1 # [69] PEReference = "%" Name ";"
173     $xml_re{PEReference} = qr/%(?:$xml_re{Name});/;
174     $xml_re{PEReference_M} = qr/%($xml_re{Name});/;
175     $xml_re{__elementdecl_simple} = qr/<!ELEMENT(?:$xml_re{s}|$xml_re{PEReference}|$xml_re{Name}|\#PCDATA|[()|,?*+])+>/s;
176 wakaba 1.5 $xml_re{__AttlistDecl_simple} = qr/<!ATTLIST(?:$xml_re{PEReference}|$xml_re{Name}|[#()|]|$xml_re{s}|$xml_re{__AttValue_simple})*>/s;
177 wakaba 1.1 # [70] EntityDecl = GEDecl / PEDecl
178 wakaba 1.6 #$xml_re{__EntityDecl_simple} = qr/<!ENTITY(?:$xml_re{__AttValue_simple}|[^"'>])*>/s;
179 wakaba 1.1 # [71] GEDecl = '<!ENTITY' S Name S EntityDef [S] ">"
180     # [72] PEDecl = '<!ENTITY' S "%" S Name S PEDef [S] ">"
181     # [73] EntityDef = EntityValue / ExternalID [NDataDecl]
182     # [74] PEDef = EntityValue / ExternalID
183     # [75] ExternalID = 'SYSTEM' S SystemLiteral / 'PUBLIC' S PublicLiteral S SystemLiteral
184     # [76] NDataDecl = S 'NDATA' S Name
185     # [77] TextDecl = '?xml' [VersionInfo] EncodingDecl [S] "?>"
186     # [78] extParsedEnt = [TextDecl] content
187     # [79] extPE ;; 1.0 FE (removed by errata)
188     # [80] EncodingDecl = S 'encoding' Eq (<"> EncName <"> / "'" EncName "'")
189     # [81] EncName = ALPHA *(ALPHA / DIGIT / "." / "_" / "-")
190     # [82] NotationDecl = '<!NOTATION' S Name S (ExternalID / PublicID) [S] ">"
191 wakaba 1.6 #$xml_re{__NotationDecl_simple} = qr/<!NOTATION(?:$xml_re{__AttValue_simple}|[^"'>])*>/s;
192 wakaba 1.1 # [83] PublicID = 'PUBLIC' S PubidLiteral
193     # [84] Letter = BaseChar / Ideographic
194     # [85] Basechar = ...
195     # [86] Ideographic = ...
196     # [87] CombiningChar = ...
197     # [88] Digit = ...
198     # [89] Extender = U+00B7 / U+02D0 / U+02D1 / U+0387 / U+0640 / U+0E46 / U+0EC6
199     # / U+3005 / U+3031-U+3035 / U+309D / U+309E / U+30FC-U+30FE
200     ## [84]-[89] removed by 1.1
201    
202     # [XMLNames]
203     # [1] NSAttName = PrefixedAttName / DefaultAttName
204     # [2] PrefixedAttName = 'xmlns:' NCName
205     # [3] DefaultAttName = 'xmlns'
206     # [4] NCName = (Letter / "_") *NCNameChar ;; 1.0
207     # [4] NCName = NCNameStartChar *NCNameChar ;; 1.1
208     # [5] NCNameChar = Letter / Digit / "." / "-" / "_" / CombiningChar / Extender ;; 1.0
209     # [5] NCNameChar = NameChar - ":" ;; 1.1
210     # [5a] NCNameStartChar = NameStartChar - ":" ;; 1.1
211     # [6] QName = [Prefix ":"] LocalPart ;; 1.0
212     # [6] QName = PrefixedName / UnprefixedName ;; 1.1
213