lisp/progmodes/ebnf-dtd.el

   1 ;;; ebnf-dtd.el --- parser for DTD (Data Type Description for XML)
   2
   3 ;; Copyright (C) 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
   4 ;;   Free Software Foundation, Inc.
   5
   6 ;; Author: Vinicius Jose Latorre <viniciusjl@ig.com.br>
   7 ;; Maintainer: Vinicius Jose Latorre <viniciusjl@ig.com.br>
   8 ;; Keywords: wp, ebnf, PostScript
   9 ;; Version: 1.1
  10
  11 ;; This file is part of GNU Emacs.
  12
  13 ;; GNU Emacs is free software: you can redistribute it and/or modify
  14 ;; it under the terms of the GNU General Public License as published by
  15 ;; the Free Software Foundation, either version 3 of the License, or
  16 ;; (at your option) any later version.
  17
  18 ;; GNU Emacs is distributed in the hope that it will be useful,
  19 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  20 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21 ;; GNU General Public License for more details.
  22
  23 ;; You should have received a copy of the GNU General Public License
  24 ;; along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
  25
  26 ;;; Commentary:
  27
  28 ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  29 ;;
  30 ;;
  31 ;; This is part of ebnf2ps package.
  32 ;;
  33 ;; This package defines a parser for DTD (Data Type Description for XML).
  34 ;;
  35 ;; See ebnf2ps.el for documentation.
  36 ;;
  37 ;;
  38 ;; DTD Syntax
  39 ;; ----------
  40 ;;
  41 ;;      See the URLs:
  42 ;;      `http://www.w3.org/TR/2004/REC-xml-20040204/'
  43 ;;      (Extensible Markup Language (XML) 1.0 (Third Edition))
  44 ;;      `http://www.w3.org/TR/html40/'
  45 ;;      (HTML 4.01 Specification)
  46 ;;      `http://www.w3.org/TR/NOTE-html-970421'
  47 ;;      (HTML DTD with support for Style Sheets)
  48 ;;
  49 ;;
  50 ;; /* Document */
  51 ;;
  52 ;; document ::= prolog element Misc*
  53 ;; /* Note that *only* the prolog will be parsed */
  54 ;;
  55 ;;
  56 ;; /* Characters */
  57 ;;
  58 ;; Char ::= #x9 | #xA | #xD
  59 ;;        | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
  60 ;; /* any Unicode character, excluding the surrogate blocks, FFFE, and FFFF. */
  61 ;;
  62 ;; /* NOTE:
  63 ;;
  64 ;;    Document authors are encouraged to avoid "compatibility characters", as
  65 ;;    defined in section 6.8 of [Unicode] (see also D21 in section 3.6 of
  66 ;;    [Unicode3]). The characters defined in the following ranges are also
  67 ;;    discouraged. They are either control characters or permanently undefined
  68 ;;    Unicode characters:
  69 ;;
  70 ;;    [#x7F-#x84],      [#x86-#x9F],      [#xFDD0-#xFDDF],
  71 ;;    [#1FFFE-#x1FFFF], [#2FFFE-#x2FFFF], [#3FFFE-#x3FFFF],
  72 ;;    [#4FFFE-#x4FFFF], [#5FFFE-#x5FFFF], [#6FFFE-#x6FFFF],
  73 ;;    [#7FFFE-#x7FFFF], [#8FFFE-#x8FFFF], [#9FFFE-#x9FFFF],
  74 ;;    [#AFFFE-#xAFFFF], [#BFFFE-#xBFFFF], [#CFFFE-#xCFFFF],
  75 ;;    [#DFFFE-#xDFFFF], [#EFFFE-#xEFFFF], [#FFFFE-#xFFFFF],
  76 ;;    [#10FFFE-#x10FFFF]. */
  77 ;;
  78 ;;
  79 ;; /* White Space */
  80 ;;
  81 ;; S ::= (#x20 | #x9 | #xD | #xA)+
  82 ;;
  83 ;;
  84 ;; /* Names and Tokens */
  85 ;;
  86 ;; NameChar ::= Letter | Digit | '.' | '-' | '_' | ':'
  87 ;;            | CombiningChar | Extender
  88 ;;
  89 ;; Name ::= (Letter | '_' | ':') (NameChar)*
  90 ;;
  91 ;; Names ::= Name (#x20 Name)*
  92 ;;
  93 ;; Nmtoken ::= (NameChar)+
  94 ;;
  95 ;; Nmtokens ::= Nmtoken (#x20 Nmtoken)*
  96 ;;
  97 ;;
  98 ;; /* Literals */
  99 ;;
 100 ;; EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 101 ;;               | "'" ([^%&'] | PEReference | Reference)* "'"
 102 ;;
 103 ;; AttValue ::= '"' ([^<&"] | Reference)* '"'
 104 ;;            | "'" ([^<&'] | Reference)* "'"
 105 ;;
 106 ;; SystemLiteral ::= ('"' [^"]* '"')
 107 ;;                 | ("'" [^']* "'")
 108 ;;
 109 ;; PubidLiteral ::= '"' PubidChar* '"'
 110 ;;                | "'" (PubidChar - "'")* "'"
 111 ;;
 112 ;; PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%]
 113 ;;
 114 ;; /* NOTE:
 115 ;;
 116 ;;    Although the EntityValue production allows the definition of a general
 117 ;;    entity consisting of a single explicit < in the literal (e.g., <!ENTITY
 118 ;;    mylt "<">), it is strongly advised to avoid this practice since any
 119 ;;    reference to that entity will cause a well-formedness error. */
 120 ;;
 121 ;;
 122 ;; /* Character Data */
 123 ;;
 124 ;; CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
 125 ;;
 126 ;;
 127 ;; /* Comments */
 128 ;;
 129 ;; Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
 130 ;;
 131 ;;
 132 ;; /* Processing Instructions */
 133 ;;
 134 ;; PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
 135 ;;
 136 ;; PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
 137 ;;
 138 ;;
 139 ;; /* CDATA Sections */
 140 ;;
 141 ;; CDSect ::= CDStart CData CDEnd
 142 ;;
 143 ;; CDStart ::= '<![CDATA['
 144 ;;
 145 ;; CData ::= (Char* - (Char* ']]>' Char*))
 146 ;;
 147 ;; CDEnd ::= ']]>'
 148 ;;
 149 ;;
 150 ;; /* Prolog */
 151 ;;
 152 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
 153 ;;
 154 ;; XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
 155 ;;
 156 ;; VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"')
 157 ;;
 158 ;; Eq ::= S? '=' S?
 159 ;;
 160 ;; VersionNum ::= '1.0'
 161 ;;
 162 ;; Misc ::= Comment | PI | S
 163 ;;
 164 ;;
 165 ;; /* Document Type Definition */
 166 ;;
 167 ;; doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
 168 ;;                 ('[' intSubset ']' S?)? '>'
 169 ;;               [VC: Root Element Type]
 170 ;;               [WFC: External Subset]
 171 ;;
 172 ;; DeclSep ::= PEReference | S
 173 ;;           [WFC: PE Between Declarations]
 174 ;;
 175 ;; intSubset ::= (markupdecl | DeclSep)*
 176 ;;
 177 ;; markupdecl ::= elementdecl | AttlistDecl | EntityDecl
 178 ;;              | NotationDecl | PI | Comment
 179 ;;              [VC: Proper Declaration/PE Nesting]
 180 ;;              [WFC: PEs in Internal Subset]
 181 ;;
 182 ;;
 183 ;; /* External Subset */
 184 ;;
 185 ;; extSubset ::= TextDecl? extSubsetDecl
 186 ;;
 187 ;; extSubsetDecl ::= ( markupdecl | conditionalSect | DeclSep)*
 188 ;;
 189 ;;
 190 ;; /* Standalone Document Declaration */
 191 ;;
 192 ;; SDDecl ::= S 'standalone' Eq
 193 ;;            (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"'))
 194 ;;          [VC: Standalone Document Declaration]
 195 ;;
 196 ;;
 197 ;; /* Element */
 198 ;;
 199 ;; element ::= EmptyElemTag | STag content ETag
 200 ;;           [WFC: Element Type Match]
 201 ;;           [VC: Element Valid]
 202 ;;
 203 ;;
 204 ;; /* Start-tag */
 205 ;;
 206 ;; STag ::= '<' Name (S Attribute)* S? '>'
 207 ;;        [WFC: Unique Att Spec]
 208 ;;
 209 ;; Attribute ::= Name Eq AttValue
 210 ;;             [VC: Attribute Value Type]
 211 ;;             [WFC: No External Entity References]
 212 ;;             [WFC: No < in Attribute Values]
 213 ;;
 214 ;;
 215 ;; /* End-tag */
 216 ;;
 217 ;; ETag ::= '</' Name S? '>'
 218 ;;
 219 ;;
 220 ;; /* Content of Elements */
 221 ;;
 222 ;; content ::= CharData?
 223 ;;             ((element | Reference | CDSect | PI | Comment) CharData?)*
 224 ;;
 225 ;;
 226 ;; /* Tags for Empty Elements */
 227 ;;
 228 ;; EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
 229 ;;                [WFC: Unique Att Spec]
 230 ;;
 231 ;;
 232 ;; /* Element Type Declaration */
 233 ;;
 234 ;; elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
 235 ;;               [VC: Unique Element Type Declaration]
 236 ;;
 237 ;; contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
 238 ;;
 239 ;;
 240 ;; /* Element-content Models */
 241 ;;
 242 ;; children ::= (choice | seq) ('?' | '*' | '+')?
 243 ;;
 244 ;; cp ::= (Name | choice | seq) ('?' | '*' | '+')?
 245 ;;
 246 ;; choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
 247 ;;          [VC: Proper Group/PE Nesting]
 248 ;;
 249 ;; seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
 250 ;;       [VC: Proper Group/PE Nesting]
 251 ;;
 252 ;;
 253 ;; /* Mixed-content Declaration */
 254 ;;
 255 ;; Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'
 256 ;;         | '(' S? '#PCDATA' S? ')'
 257 ;;         [VC: Proper Group/PE Nesting]
 258 ;;         [VC: No Duplicate Types]
 259 ;;
 260 ;;
 261 ;; /* Attribute-list Declaration */
 262 ;;
 263 ;; AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
 264 ;;
 265 ;; AttDef ::= S Name S AttType S DefaultDecl
 266 ;;
 267 ;;
 268 ;; /* Attribute Types */
 269 ;;
 270 ;; AttType ::= StringType | TokenizedType | EnumeratedType
 271 ;;
 272 ;; StringType ::= 'CDATA'
 273 ;;
 274 ;; TokenizedType ::= 'ID'       [VC: ID]
 275 ;;                              [VC: One ID per Element Type]
 276 ;;                              [VC: ID Attribute Default]
 277 ;;                 | 'IDREF'    [VC: IDREF]
 278 ;;                 | 'IDREFS'   [VC: IDREF]
 279 ;;                 | 'ENTITY'   [VC: Entity Name]
 280 ;;                 | 'ENTITIES' [VC: Entity Name]
 281 ;;                 | 'NMTOKEN'  [VC: Name Token]
 282 ;;                 | 'NMTOKENS' [VC: Name Token]
 283 ;;
 284 ;;
 285 ;; /* Enumerated Attribute Types */
 286 ;;
 287 ;; EnumeratedType ::= NotationType | Enumeration
 288 ;;
 289 ;; NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 290 ;;                [VC: Notation Attributes]
 291 ;;                [VC: One Notation Per Element Type]
 292 ;;                [VC: No Notation on Empty Element]
 293 ;;                [VC: No Duplicate Tokens]
 294 ;;
 295 ;; Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
 296 ;;               [VC: Enumeration]
 297 ;;               [VC: No Duplicate Tokens]
 298 ;;
 299 ;;
 300 ;; /* Attribute Defaults */
 301 ;;
 302 ;; DefaultDecl ::= '#REQUIRED' | '#IMPLIED'
 303 ;;               | (('#FIXED' S)? AttValue)
 304 ;;               [VC: Required Attribute]
 305 ;;               [VC: Attribute Default Value Syntactically Correct]
 306 ;;               [WFC: No < in Attribute Values]
 307 ;;               [VC: Fixed Attribute Default]
 308 ;;
 309 ;;
 310 ;; /* Conditional Section */
 311 ;;
 312 ;; conditionalSect ::= includeSect | ignoreSect
 313 ;;
 314 ;; includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
 315 ;;               [VC: Proper Conditional Section/PE Nesting]
 316 ;;
 317 ;; ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
 318 ;;              [VC: Proper Conditional Section/PE Nesting]
 319 ;;
 320 ;; ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)*
 321 ;;
 322 ;; Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
 323 ;;
 324 ;;
 325 ;; /* Character Reference */
 326 ;;
 327 ;; CharRef ::= '&#' [0-9]+ ';'
 328 ;;           | '&#x' [0-9a-fA-F]+ ';'
 329 ;;           [WFC: Legal Character]
 330 ;;
 331 ;;
 332 ;; /* Entity Reference */
 333 ;;
 334 ;; Reference ::= EntityRef | CharRef
 335 ;;
 336 ;; EntityRef ::= '&' Name ';'
 337 ;;             [WFC: Entity Declared]
 338 ;;             [VC: Entity Declared]
 339 ;;             [WFC: Parsed Entity]
 340 ;;             [WFC: No Recursion]
 341 ;;
 342 ;; PEReference ::= '%' Name ';'
 343 ;;               [VC: Entity Declared]
 344 ;;               [WFC: No Recursion]
 345 ;;               [WFC: In DTD]
 346 ;;
 347 ;;
 348 ;; /* Entity Declaration */
 349 ;;
 350 ;; EntityDecl ::= GEDecl | PEDecl
 351 ;;
 352 ;; GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
 353 ;;
 354 ;; PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 355 ;;
 356 ;; EntityDef ::= EntityValue | (ExternalID NDataDecl?)
 357 ;;
 358 ;; PEDef ::= EntityValue | ExternalID
 359 ;;
 360 ;;
 361 ;; /* External Entity Declaration */
 362 ;;
 363 ;; ExternalID ::= 'SYSTEM' S SystemLiteral
 364 ;;              | 'PUBLIC' S PubidLiteral S SystemLiteral
 365 ;;
 366 ;; NDataDecl ::= S 'NDATA' S Name
 367 ;;             [VC: Notation Declared]
 368 ;;
 369 ;;
 370 ;; /* Text Declaration */
 371 ;;
 372 ;; TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>'
 373 ;;
 374 ;;
 375 ;; /* Well-Formed External Parsed Entity */
 376 ;;
 377 ;; extParsedEnt ::= TextDecl? content
 378 ;;
 379 ;;
 380 ;; /* Encoding Declaration */
 381 ;;
 382 ;; EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'" )
 383 ;;
 384 ;; EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
 385 ;; /* Encoding name contains only Latin characters */
 386 ;;
 387 ;;
 388 ;; /* Notation Declarations */
 389 ;;
 390 ;; NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>'
 391 ;;                [VC: Unique Notation Name]
 392 ;;
 393 ;; PublicID ::= 'PUBLIC' S PubidLiteral
 394 ;;
 395 ;;
 396 ;; /* Characters */
 397 ;;
 398 ;; Letter ::= BaseChar | Ideographic
 399 ;;
 400 ;; BaseChar ::= [#x0041-#x005A] | [#x0061-#x007A] | [#x00C0-#x00D6]
 401 ;;            | [#x00D8-#x00F6] | [#x00F8-#x00FF] | [#x0100-#x0131]
 402 ;;            | [#x0134-#x013E] | [#x0141-#x0148] | [#x014A-#x017E]
 403 ;;            | [#x0180-#x01C3] | [#x01CD-#x01F0] | [#x01F4-#x01F5]
 404 ;;            | [#x01FA-#x0217] | [#x0250-#x02A8] | [#x02BB-#x02C1]
 405 ;;            | #x0386          | [#x0388-#x038A] | #x038C
 406 ;;            | [#x038E-#x03A1] | [#x03A3-#x03CE] | [#x03D0-#x03D6]
 407 ;;            | #x03DA          | #x03DC          | #x03DE
 408 ;;            | #x03E0          | [#x03E2-#x03F3] | [#x0401-#x040C]
 409 ;;            | [#x040E-#x044F] | [#x0451-#x045C] | [#x045E-#x0481]
 410 ;;            | [#x0490-#x04C4] | [#x04C7-#x04C8] | [#x04CB-#x04CC]
 411 ;;            | [#x04D0-#x04EB] | [#x04EE-#x04F5] | [#x04F8-#x04F9]
 412 ;;            | [#x0531-#x0556] | #x0559          | [#x0561-#x0586]
 413 ;;            | [#x05D0-#x05EA] | [#x05F0-#x05F2] | [#x0621-#x063A]
 414 ;;            | [#x0641-#x064A] | [#x0671-#x06B7] | [#x06BA-#x06BE]
 415 ;;            | [#x06C0-#x06CE] | [#x06D0-#x06D3] | #x06D5
 416 ;;            | [#x06E5-#x06E6] | [#x0905-#x0939] | #x093D
 417 ;;            | [#x0958-#x0961] | [#x0985-#x098C] | [#x098F-#x0990]
 418 ;;            | [#x0993-#x09A8] | [#x09AA-#x09B0] | #x09B2
 419 ;;            | [#x09B6-#x09B9] | [#x09DC-#x09DD] | [#x09DF-#x09E1]
 420 ;;            | [#x09F0-#x09F1] | [#x0A05-#x0A0A] | [#x0A0F-#x0A10]
 421 ;;            | [#x0A13-#x0A28] | [#x0A2A-#x0A30] | [#x0A32-#x0A33]
 422 ;;            | [#x0A35-#x0A36] | [#x0A38-#x0A39] | [#x0A59-#x0A5C]
 423 ;;            | #x0A5E          | [#x0A72-#x0A74] | [#x0A85-#x0A8B]
 424 ;;            | #x0A8D          | [#x0A8F-#x0A91] | [#x0A93-#x0AA8]
 425 ;;            | [#x0AAA-#x0AB0] | [#x0AB2-#x0AB3] | [#x0AB5-#x0AB9]
 426 ;;            | #x0ABD          | #x0AE0          | [#x0B05-#x0B0C]
 427 ;;            | [#x0B0F-#x0B10] | [#x0B13-#x0B28] | [#x0B2A-#x0B30]
 428 ;;            | [#x0B32-#x0B33] | [#x0B36-#x0B39] | #x0B3D
 429 ;;            | [#x0B5C-#x0B5D] | [#x0B5F-#x0B61] | [#x0B85-#x0B8A]
 430 ;;            | [#x0B8E-#x0B90] | [#x0B92-#x0B95] | [#x0B99-#x0B9A]
 431 ;;            | #x0B9C          | [#x0B9E-#x0B9F] | [#x0BA3-#x0BA4]
 432 ;;            | [#x0BA8-#x0BAA] | [#x0BAE-#x0BB5] | [#x0BB7-#x0BB9]
 433 ;;            | [#x0C05-#x0C0C] | [#x0C0E-#x0C10] | [#x0C12-#x0C28]
 434 ;;            | [#x0C2A-#x0C33] | [#x0C35-#x0C39] | [#x0C60-#x0C61]
 435 ;;            | [#x0C85-#x0C8C] | [#x0C8E-#x0C90] | [#x0C92-#x0CA8]
 436 ;;            | [#x0CAA-#x0CB3] | [#x0CB5-#x0CB9] | #x0CDE
 437 ;;            | [#x0CE0-#x0CE1] | [#x0D05-#x0D0C] | [#x0D0E-#x0D10]
 438 ;;            | [#x0D12-#x0D28] | [#x0D2A-#x0D39] | [#x0D60-#x0D61]
 439 ;;            | [#x0E01-#x0E2E] | #x0E30          | [#x0E32-#x0E33]
 440 ;;            | [#x0E40-#x0E45] | [#x0E81-#x0E82] | #x0E84
 441 ;;            | [#x0E87-#x0E88] | #x0E8A          | #x0E8D
 442 ;;            | [#x0E94-#x0E97] | [#x0E99-#x0E9F] | [#x0EA1-#x0EA3]
 443 ;;            | #x0EA5          | #x0EA7          | [#x0EAA-#x0EAB]
 444 ;;            | [#x0EAD-#x0EAE] | #x0EB0          | [#x0EB2-#x0EB3]
 445 ;;            | #x0EBD          | [#x0EC0-#x0EC4] | [#x0F40-#x0F47]
 446 ;;            | [#x0F49-#x0F69] | [#x10A0-#x10C5] | [#x10D0-#x10F6]
 447 ;;            | #x1100          | [#x1102-#x1103] | [#x1105-#x1107]
 448 ;;            | #x1109          | [#x110B-#x110C] | [#x110E-#x1112]
 449 ;;            | #x113C          | #x113E          | #x1140
 450 ;;            | #x114C          | #x114E          | #x1150
 451 ;;            | [#x1154-#x1155] | #x1159          | [#x115F-#x1161]
 452 ;;            | #x1163          | #x1165          | #x1167
 453 ;;            | #x1169          | [#x116D-#x116E] | [#x1172-#x1173]
 454 ;;            | #x1175          | #x119E          | #x11A8
 455 ;;            | #x11AB          | [#x11AE-#x11AF] | [#x11B7-#x11B8]
 456 ;;            | #x11BA          | [#x11BC-#x11C2] | #x11EB
 457 ;;            | #x11F0          | #x11F9          | [#x1E00-#x1E9B]
 458 ;;            | [#x1EA0-#x1EF9] | [#x1F00-#x1F15] | [#x1F18-#x1F1D]
 459 ;;            | [#x1F20-#x1F45] | [#x1F48-#x1F4D] | [#x1F50-#x1F57]
 460 ;;            | #x1F59          | #x1F5B          | #x1F5D
 461 ;;            | [#x1F5F-#x1F7D] | [#x1F80-#x1FB4] | [#x1FB6-#x1FBC]
 462 ;;            | #x1FBE          | [#x1FC2-#x1FC4] | [#x1FC6-#x1FCC]
 463 ;;            | [#x1FD0-#x1FD3] | [#x1FD6-#x1FDB] | [#x1FE0-#x1FEC]
 464 ;;            | [#x1FF2-#x1FF4] | [#x1FF6-#x1FFC] | #x2126
 465 ;;            | [#x212A-#x212B] | #x212E          | [#x2180-#x2182]
 466 ;;            | [#x3041-#x3094] | [#x30A1-#x30FA] | [#x3105-#x312C]
 467 ;;            | [#xAC00-#xD7A3]
 468 ;;
 469 ;; Ideographic ::= [#x4E00-#x9FA5] | #x3007 | [#x3021-#x3029]
 470 ;;
 471 ;; CombiningChar ::= [#x0300-#x0345] | [#x0360-#x0361] | [#x0483-#x0486]
 472 ;;                 | [#x0591-#x05A1] | [#x05A3-#x05B9] | [#x05BB-#x05BD]
 473 ;;                 | #x05BF          | [#x05C1-#x05C2] | #x05C4
 474 ;;                 | [#x064B-#x0652] | #x0670          | [#x06D6-#x06DC]
 475 ;;                 | [#x06DD-#x06DF] | [#x06E0-#x06E4] | [#x06E7-#x06E8]
 476 ;;                 | [#x06EA-#x06ED] | [#x0901-#x0903] | #x093C
 477 ;;                 | [#x093E-#x094C] | #x094D          | [#x0951-#x0954]
 478 ;;                 | [#x0962-#x0963] | [#x0981-#x0983] | #x09BC
 479 ;;                 | #x09BE          | #x09BF          | [#x09C0-#x09C4]
 480 ;;                 | [#x09C7-#x09C8] | [#x09CB-#x09CD] | #x09D7
 481 ;;                 | [#x09E2-#x09E3] | #x0A02          | #x0A3C
 482 ;;                 | #x0A3E          | #x0A3F          | [#x0A40-#x0A42]
 483 ;;                 | [#x0A47-#x0A48] | [#x0A4B-#x0A4D] | [#x0A70-#x0A71]
 484 ;;                 | [#x0A81-#x0A83] | #x0ABC          | [#x0ABE-#x0AC5]
 485 ;;                 | [#x0AC7-#x0AC9] | [#x0ACB-#x0ACD] | [#x0B01-#x0B03]
 486 ;;                 | #x0B3C          | [#x0B3E-#x0B43] | [#x0B47-#x0B48]
 487 ;;                 | [#x0B4B-#x0B4D] | [#x0B56-#x0B57] | [#x0B82-#x0B83]
 488 ;;                 | [#x0BBE-#x0BC2] | [#x0BC6-#x0BC8] | [#x0BCA-#x0BCD]
 489 ;;                 | #x0BD7          | [#x0C01-#x0C03] | [#x0C3E-#x0C44]
 490 ;;                 | [#x0C46-#x0C48] | [#x0C4A-#x0C4D] | [#x0C55-#x0C56]
 491 ;;                 | [#x0C82-#x0C83] | [#x0CBE-#x0CC4] | [#x0CC6-#x0CC8]
 492 ;;                 | [#x0CCA-#x0CCD] | [#x0CD5-#x0CD6] | [#x0D02-#x0D03]
 493 ;;                 | [#x0D3E-#x0D43] | [#x0D46-#x0D48] | [#x0D4A-#x0D4D]
 494 ;;                 | #x0D57          | #x0E31          | [#x0E34-#x0E3A]
 495 ;;                 | [#x0E47-#x0E4E] | #x0EB1          | [#x0EB4-#x0EB9]
 496 ;;                 | [#x0EBB-#x0EBC] | [#x0EC8-#x0ECD] | [#x0F18-#x0F19]
 497 ;;                 | #x0F35          | #x0F37          | #x0F39
 498 ;;                 | #x0F3E          | #x0F3F          | [#x0F71-#x0F84]
 499 ;;                 | [#x0F86-#x0F8B] | [#x0F90-#x0F95] | #x0F97
 500 ;;                 | [#x0F99-#x0FAD] | [#x0FB1-#x0FB7] | #x0FB9
 501 ;;                 | [#x20D0-#x20DC] | #x20E1          | [#x302A-#x302F]
 502 ;;                 | #x3099          | #x309A
 503 ;;
 504 ;; Digit ::= [#x0030-#x0039] | [#x0660-#x0669] | [#x06F0-#x06F9]
 505 ;;         | [#x0966-#x096F] | [#x09E6-#x09EF] | [#x0A66-#x0A6F]
 506 ;;         | [#x0AE6-#x0AEF] | [#x0B66-#x0B6F] | [#x0BE7-#x0BEF]
 507 ;;         | [#x0C66-#x0C6F] | [#x0CE6-#x0CEF] | [#x0D66-#x0D6F]
 508 ;;         | [#x0E50-#x0E59] | [#x0ED0-#x0ED9] | [#x0F20-#x0F29]
 509 ;;
 510 ;; Extender ::= #x00B7 | #x02D0 | #x02D1 | #x0387 | #x0640 | #x0E46 | #x0EC6
 511 ;;            | #x3005 | [#x3031-#x3035] | [#x309D-#x309E] | [#x30FC-#x30FE]
 512 ;;
 513 ;;
 514 ;; NOTES
 515 ;; -----
 516 ;;
 517 ;; At moment, only the `<!ELEMENT' generates a syntactic chart.  The
 518 ;; `<!ATTLIST', `<!NOTATION' and `<!ENTITY' are syntacticly checked but they
 519 ;; don't generate a syntactic chart.
 520 ;;
 521 ;; Besides the syntax above, ebnf-dtd also accepts a `pure' dtd file.  An
 522 ;; example of a `pure' dtd file is:
 523 ;;
 524 ;;    <?xml version="1.0" encoding="UTF-8"?>
 525 ;;    <!--
 526 ;;    The main element.
 527 ;;    -->
 528 ;;    <!ELEMENT workflow (registers?, trigger-functions?, initial-actions,
 529 ;;                        steps, splits?, joins?)>
 530 ;;    <!--
 531 ;;    An action that can be executed (id must be unique among actions for
 532 ;;    the enclosing step).
 533 ;;    Used in: actions
 534 ;;    -->
 535 ;;    <!ELEMENT action (restrict-to, validators?, pre-functions?, results,
 536 ;;                      post-functions?)>
 537 ;;    <!ATTLIST action
 538 ;;      id CDATA #REQUIRED
 539 ;;      name CDATA #REQUIRED
 540 ;;    >
 541 ;;
 542 ;;
 543 ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 544
 545 ;;; Code:
 546
 547
 548 (require 'ebnf-otz)
 549
 550
 551 (defvar ebnf-dtd-lex nil
 552   "Value returned by `ebnf-dtd-lex' function.")
 553
 554 \f
 555 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 556 ;; Syntactic analyzer
 557
 558
 559 ;;; document ::= prolog element Misc*
 560 ;;; /* Note that *only* the prolog will be parsed */
 561
 562 (defun ebnf-dtd-parser (start)
 563   "DTD parser."
 564   (let ((total (+ (- ebnf-limit start) 1))
 565         (bias (1- start))
 566         (origin (point))
 567         rule-list token rule the-end)
 568     (goto-char start)
 569     (setq token (ebnf-dtd-lex))
 570     (and (eq token 'end-of-input)
 571          (error "Empty DTD file"))
 572     (setq token (ebnf-dtd-prolog token))
 573     (unless (eq (car token) 'end-prolog)
 574       (setq the-end (cdr token)
 575             token   (car token))
 576       (while (not (eq token the-end))
 577         (ebnf-message-float
 578          "Parsing...%s%%"
 579          (/ (* (- (point) bias) 100.0) total))
 580         (setq token (ebnf-dtd-intsubset token)
 581               rule  (cdr token)
 582               token (car token))
 583         (or (null rule)
 584             (ebnf-add-empty-rule-list rule)
 585             (setq rule-list (cons rule rule-list))))
 586       (or (eq the-end 'end-of-input)
 587           (eq (ebnf-dtd-lex) 'end-decl)
 588           (error "Missing end of DOCTYPE"))
 589       ;; adjust message, 'cause *only* prolog will be parsed
 590       (ebnf-message-float "Parsing...%s%%" 100.0))
 591     (goto-char origin)
 592     rule-list))
 593
 594
 595 ;;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
 596 ;;;
 597 ;;; XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
 598 ;;;
 599 ;;; VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"')
 600 ;;;
 601 ;;; Eq ::= S? '=' S?
 602 ;;;
 603 ;;; VersionNum ::= '1.0'
 604 ;;;
 605 ;;; Misc ::= Comment | PI | S
 606 ;;;
 607 ;;; EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'" )
 608 ;;;
 609 ;;; EncName ::= [A-Za-z] ([-A-Za-z0-9._])*
 610 ;;; /* Encoding name contains only Latin characters */
 611 ;;;
 612 ;;; SDDecl ::= S 'standalone' Eq
 613 ;;;            (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"'))
 614 ;;;
 615 ;;; doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
 616 ;;;                 ('[' intSubset ']' S?)? '>'
 617
 618
 619 (defun ebnf-dtd-prolog (token)
 620   (when (and (eq token 'begin-pi) (string= ebnf-dtd-lex "xml"))
 621     ;; version = "1.0"
 622     (setq token (ebnf-dtd-attribute (ebnf-dtd-lex) 'version-attr
 623                                     "^1\\.0$" "XML version"))
 624     ;; ( encoding = "encoding name" )?
 625     (setq token (ebnf-dtd-attribute-optional
 626                  token 'encoding-attr
 627                  "^[A-Za-z][-A-Za-z0-9._]*$" "XML encoding"))
 628     ;; ( standalone = ( "yes" | "no" ) )?
 629     (setq token (ebnf-dtd-attribute-optional
 630                  token 'standalone-attr
 631                  "^yes|no$" "XML standalone"))
 632     (or (eq token 'end-pi)
 633         (error "Missing end of XML processing instruction")))
 634   ;; processing instructions
 635   (setq token (ebnf-dtd-pi (ebnf-dtd-lex)))
 636   (cond
 637    ;; DOCTYPE
 638    ((eq token 'doctype-decl)
 639     (or (eq (ebnf-dtd-lex) 'name)
 640         (error "Document type name is missing"))
 641     (cons (if (eq (ebnf-dtd-externalid) 'begin-subset)
 642               (ebnf-dtd-lex)
 643             'end-prolog)
 644           'end-subset))
 645    ((memq token '(element-decl attlist-decl entity-decl notation-decl))
 646     (cons token 'end-of-input))
 647    (t
 648     '(end-prolog . end-subset))
 649    ))
 650
 651
 652 (defun ebnf-dtd-attribute (token attr match attr-name)
 653   (or (eq token attr)
 654       (error "%s attribute is missing" attr-name))
 655   (ebnf-dtd-attribute-optional token attr match attr-name))
 656
 657
 658 (defun ebnf-dtd-attribute-optional (token attr match attr-name)
 659   (when (eq token attr)
 660     (or (and (eq (ebnf-dtd-lex) 'equal)
 661              (eq (ebnf-dtd-lex) 'string)
 662              (string-match match ebnf-dtd-lex))
 663         (error "XML %s attribute is invalid" attr-name))
 664     (setq token (ebnf-dtd-lex)))
 665   token)
 666
 667
 668 ;;; ExternalID ::= 'SYSTEM' S SystemLiteral
 669 ;;;              | 'PUBLIC' S PubidLiteral S SystemLiteral
 670
 671
 672 (defun ebnf-dtd-externalid (&optional token)
 673   (let ((must-have token))
 674     (or token (setq token (ebnf-dtd-lex)))
 675     (cond ((eq token 'system)
 676            (ebnf-dtd-systemliteral))
 677           ((eq token 'public)
 678            (ebnf-dtd-pubidliteral)
 679            (ebnf-dtd-systemliteral))
 680           (must-have
 681            (error "Missing `SYSTEM' or `PUBLIC' in external id"))
 682           (t
 683            token))))
 684
 685
 686 ;;; SystemLiteral ::= ('"' [^"]* '"')
 687 ;;;                 | ("'" [^']* "'")
 688
 689
 690 (defun ebnf-dtd-systemliteral ()
 691   (or (eq (ebnf-dtd-lex) 'string)
 692       (error "System identifier is invalid"))
 693   (ebnf-dtd-lex))
 694
 695
 696 ;;; PubidLiteral ::= '"' PubidChar* '"'
 697 ;;;                | "'" (PubidChar - "'")* "'"
 698 ;;;
 699 ;;; PubidChar ::= [-'()+,./:=?;!*#@$_%\n\r a-zA-Z0-9]
 700
 701
 702 (defun ebnf-dtd-pubidliteral ()
 703   (or (and (eq (ebnf-dtd-lex) 'string)
 704            (string-match "^[-'()+,./:=?;!*#@$_%\n\r a-zA-Z0-9]*$"
 705                          ebnf-dtd-lex))
 706       (error "Public identifier is invalid")))
 707
 708
 709 ;;; PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
 710 ;;;
 711 ;;; PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
 712
 713
 714 (defun ebnf-dtd-pi (token)
 715   (while (eq token 'begin-pi)
 716     (and (string-match "^[xX][mM][lL]$" ebnf-dtd-lex)
 717          (error "Processing instruction name can not be `XML'"))
 718     (while (not (eq (ebnf-dtd-lex) 'end-pi)))
 719     (setq token (ebnf-dtd-lex)))
 720   token)
 721
 722
 723 ;;; doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
 724 ;;;                 ('[' intSubset ']' S?)? '>'
 725 ;;;
 726 ;;; intSubset ::= (markupdecl | DeclSep)*
 727 ;;;
 728 ;;; DeclSep ::= PEReference | S
 729 ;;;
 730 ;;; markupdecl ::= elementdecl | AttlistDecl | EntityDecl
 731 ;;;              | NotationDecl | PI | Comment
 732
 733
 734 (defun ebnf-dtd-intsubset (token)
 735   ;; PI - Processing Instruction
 736   (and (eq token 'begin-pi)
 737        (setq token (ebnf-dtd-pi token)))
 738   (cond
 739    ((memq token '(end-subset end-of-input))
 740     (cons token nil))
 741    ((eq token 'pe-ref)
 742     (cons (ebnf-dtd-lex) nil))          ; annotation
 743    ((eq token 'element-decl)
 744     (ebnf-dtd-elementdecl))             ; rule
 745    ((eq token 'attlist-decl)
 746     (ebnf-dtd-attlistdecl))             ; annotation
 747    ((eq token 'entity-decl)
 748     (ebnf-dtd-entitydecl))              ; annotation
 749    ((eq token 'notation-decl)
 750     (ebnf-dtd-notationdecl))            ; annotation
 751    (t
 752     (error "Invalid DOCTYPE element"))
 753    ))
 754
 755
 756 ;;; elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
 757 ;;;
 758 ;;; contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
 759 ;;;
 760 ;;; Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'
 761 ;;;         | '(' S? '#PCDATA' S? ')'
 762 ;;;
 763 ;;; children ::= (choice | seq) ('?' | '*' | '+')?
 764 ;;;
 765 ;;; choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
 766 ;;;
 767 ;;; seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
 768 ;;;
 769 ;;; cp ::= (Name | choice | seq) ('?' | '*' | '+')?
 770
 771
 772 (defun ebnf-dtd-elementdecl ()
 773   (let ((action ebnf-action)
 774         name token body)
 775     (setq ebnf-action nil)
 776     (or (eq (ebnf-dtd-lex) 'name)
 777         (error "Invalid ELEMENT name"))
 778     (setq name  ebnf-dtd-lex
 779           token (ebnf-dtd-lex)
 780           body  (cond ((memq token '(empty any))
 781                        (let ((term (ebnf-make-terminal ebnf-dtd-lex)))
 782                          (cons (ebnf-dtd-lex) term)))
 783                       ((eq token 'begin-group)
 784                        (setq token (ebnf-dtd-lex))
 785                        (if (eq token 'pcdata)
 786                            (ebnf-dtd-mixed)
 787                          (ebnf-dtd-children token)))
 788                       (t
 789                        (error "Invalid ELEMENT content"))
 790                       ))
 791     (or (eq (car body) 'end-decl)
 792         (error "Missing `>' in ELEMENT declaration"))
 793     (ebnf-eps-add-production name)
 794     (cons (ebnf-dtd-lex)
 795           (ebnf-make-production name (cdr body) action))))
 796
 797
 798 ;;; Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'
 799 ;;;         | '(' S? '#PCDATA' S? ')'
 800
 801
 802 (defun ebnf-dtd-mixed ()
 803   (let* ((alt             (cons (ebnf-make-terminal ebnf-dtd-lex) nil))
 804          (token           (ebnf-dtd-lex))
 805          (has-alternative (eq token 'alternative)))
 806     (while (eq token 'alternative)
 807       (or (eq (ebnf-dtd-lex) 'name)
 808           (error "Invalid name"))
 809       (setq alt   (cons ebnf-dtd-lex alt)
 810             token (ebnf-dtd-lex)))
 811     (or (eq token 'end-group)
 812         (error "Missing `)'"))
 813     (and has-alternative
 814          (or (eq (ebnf-dtd-lex) 'zero-or-more)
 815              (error "Missing `*'")))
 816     (ebnf-token-alternative alt (cons (ebnf-dtd-lex) nil))))
 817
 818
 819 ;;; children ::= (choice | seq) ('?' | '*' | '+')?
 820
 821
 822 (defun ebnf-dtd-children (token)
 823   (ebnf-dtd-operators (ebnf-dtd-choice-seq token)))
 824
 825
 826 ;;; choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
 827 ;;;
 828 ;;; seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
 829
 830
 831 (defun ebnf-dtd-choice-seq (token)
 832   (setq token (ebnf-dtd-cp token))
 833   (let (elist)
 834     (cond
 835      ;; choice
 836      ((eq (car token) 'alternative)
 837       (while (eq (car token) 'alternative)
 838         (setq elist (cons (cdr token) elist)
 839               token (ebnf-dtd-cp (ebnf-dtd-lex))))
 840       (setq elist (ebnf-token-alternative elist token)))
 841      ;; seq
 842      ((eq (car token) 'comma)
 843       (while (eq (car token) 'comma)
 844         (setq elist (cons (cdr token) elist)
 845               token (ebnf-dtd-cp (ebnf-dtd-lex))))
 846       (setq elist (ebnf-token-sequence (cons (cdr token) elist))))
 847      ;; only one element
 848      (t
 849       (setq elist (cdr token))))
 850     (or (eq (car token) 'end-group)
 851         (error "Missing `)' in ELEMENT content"))
 852     elist))
 853
 854
 855 ;;; cp ::= (Name | choice | seq) ('?' | '*' | '+')?
 856
 857
 858 (defun ebnf-dtd-cp (token)
 859   (ebnf-dtd-operators (cond ((eq token 'name)
 860                              (ebnf-make-terminal ebnf-dtd-lex))
 861                             ((eq token 'begin-group)
 862                              (ebnf-dtd-choice-seq (ebnf-dtd-lex)))
 863                             (t
 864                              (error "Invalid element"))
 865                             )))
 866
 867
 868 ;;; elm ('?' | '*' | '+')?
 869
 870
 871 (defun ebnf-dtd-operators (elm)
 872   (let ((token (ebnf-dtd-lex)))
 873     (cond ((eq token 'optional)         ; ? -  optional
 874            (cons (ebnf-dtd-lex) (ebnf-token-optional elm)))
 875           ((eq token 'zero-or-more)     ; * - zero or more
 876            (cons (ebnf-dtd-lex) (ebnf-make-zero-or-more elm)))
 877           ((eq token 'one-or-more)      ; + - one or more
 878            (cons (ebnf-dtd-lex) (ebnf-make-one-or-more elm)))
 879           (t                            ; only element
 880            (cons token elm))
 881           )))
 882
 883
 884 ;;; AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
 885 ;;;
 886 ;;; AttDef ::= S Name S AttType S DefaultDecl
 887 ;;;
 888 ;;; AttType ::= StringType | TokenizedType | EnumeratedType
 889 ;;;
 890 ;;; StringType ::= 'CDATA'
 891 ;;;
 892 ;;; TokenizedType ::= 'ID'
 893 ;;;                 | 'IDREF'
 894 ;;;                 | 'IDREFS'
 895 ;;;                 | 'ENTITY'
 896 ;;;                 | 'ENTITIES'
 897 ;;;                 | 'NMTOKEN'
 898 ;;;                 | 'NMTOKENS'
 899 ;;;
 900 ;;; EnumeratedType ::= NotationType | Enumeration
 901 ;;;
 902 ;;; NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 903 ;;;
 904 ;;; Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
 905 ;;;
 906 ;;; DefaultDecl ::= '#REQUIRED'
 907 ;;;               | '#IMPLIED'
 908 ;;;               | (('#FIXED' S)? AttValue)
 909 ;;;
 910 ;;;
 911 ;;; AttValue ::= '"' ([^<&"] | Reference)* '"'
 912 ;;;            | "'" ([^<&'] | Reference)* "'"
 913 ;;;
 914 ;;; Reference ::= EntityRef | CharRef
 915 ;;;
 916 ;;; EntityRef ::= '&' Name ';'
 917 ;;;
 918 ;;; CharRef ::= '&#' [0-9]+ ';'
 919 ;;;           | '&#x' [0-9a-fA-F]+ ';'
 920
 921 ;;; "^\\(&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^<&]\\)*$"
 922
 923
 924 (defun ebnf-dtd-attlistdecl ()
 925   (or (eq (ebnf-dtd-lex) 'name)
 926       (error "Invalid ATTLIST name"))
 927   (let (token)
 928     (while (eq (setq token (ebnf-dtd-lex)) 'name)
 929       ;; type
 930       (setq token (ebnf-dtd-lex))
 931       (cond
 932        ((eq token 'notation)
 933         (or (eq (ebnf-dtd-lex) 'begin-group)
 934             (error "Missing `(' in NOTATION type in ATTLIST declaration"))
 935         (ebnf-dtd-namelist "NOTATION" '(name)))
 936        ((eq token 'begin-group)
 937         (ebnf-dtd-namelist "enumeration" '(name name-char)))
 938        ((memq token
 939               '(cdata id idref idrefs entity entities nmtoken nmtokens)))
 940        (t
 941         (error "Invalid type in ATTLIST declaration")))
 942       ;; default value
 943       (setq token (ebnf-dtd-lex))
 944       (unless (memq token '(required implied))
 945         (and (eq token 'fixed)
 946              (setq token (ebnf-dtd-lex)))
 947         (or (and (eq token 'string)
 948                  (string-match
 949                   "^\\(&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^<&]\\)*$"
 950                   ebnf-dtd-lex))
 951             (error "Invalid default value in ATTLIST declaration"))))
 952     (or (eq token 'end-decl)
 953         (error "Missing `>' in end of ATTLIST"))
 954     (cons (ebnf-dtd-lex) nil)))
 955
 956
 957 (defun ebnf-dtd-namelist (type name-list)
 958   (let (token)
 959     (while (progn
 960              (or (memq (ebnf-dtd-lex) name-list)
 961                  (error "Invalid name in %s type in ATTLIST declaration" type))
 962              (eq (setq token (ebnf-dtd-lex)) 'alternative)))
 963     (or (eq token 'end-group)
 964         (error "Missing `)' in %s type in ATTLIST declaration" type))))
 965
 966
 967 ;;; EntityDecl ::= GEDecl | PEDecl
 968 ;;;
 969 ;;; GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
 970 ;;;
 971 ;;; PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 972 ;;;
 973 ;;; EntityDef ::= EntityValue | (ExternalID NDataDecl?)
 974 ;;;
 975 ;;; PEDef ::= EntityValue | ExternalID
 976 ;;;
 977 ;;; NDataDecl ::= S 'NDATA' S Name
 978 ;;;
 979 ;;;
 980 ;;; EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 981 ;;;               | "'" ([^%&'] | PEReference | Reference)* "'"
 982 ;;;
 983 ;;; PEReference ::= '%' Name ';'
 984 ;;;
 985 ;;; Reference ::= EntityRef | CharRef
 986 ;;;
 987 ;;; EntityRef ::= '&' Name ';'
 988 ;;;
 989 ;;; CharRef ::= '&#' [0-9]+ ';'
 990 ;;;           | '&#x' [0-9a-fA-F]+ ';'
 991
 992 ;;; "^\\(%[A-Za-z_:][-A-Za-z0-9._:]*;\\|&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^%&]\\)*$"
 993
 994
 995 (defun ebnf-dtd-entitydecl ()
 996   (let* ((token (ebnf-dtd-lex))
 997          (pedecl (eq token 'percent)))
 998     (and pedecl
 999          (setq token (ebnf-dtd-lex)))
1000     (or (eq token 'name)
1001         (error "Invalid name of ENTITY"))
1002     (setq token (ebnf-dtd-lex))
1003     (if (eq token 'string)
1004         (if (string-match
1005              "^\\(%[A-Za-z_:][-A-Za-z0-9._:]*;\\|&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^%&]\\)*$"
1006              ebnf-dtd-lex)
1007             (setq token (ebnf-dtd-lex))
1008           (error "Invalid ENTITY definition"))
1009       (setq token (ebnf-dtd-externalid token))
1010       (when (and (not pedecl) (eq token 'ndata))
1011         (or (eq (ebnf-dtd-lex) 'name)
1012             (error "Invalid NDATA name"))
1013         (setq token (ebnf-dtd-lex))))
1014     (or (eq token 'end-decl)
1015         (error "Missing `>' in end of ENTITY"))
1016     (cons (ebnf-dtd-lex) nil)))
1017
1018
1019 ;;; NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>'
1020 ;;;
1021 ;;; PublicID ::= 'PUBLIC' S PubidLiteral
1022
1023
1024 (defun ebnf-dtd-notationdecl ()
1025   (or (eq (ebnf-dtd-lex) 'name)
1026       (error "Invalid name NOTATION"))
1027   (or (eq (ebnf-dtd-externalid-or-publicid) 'end-decl)
1028       (error "Missing `>' in end of NOTATION"))
1029   (cons (ebnf-dtd-lex) nil))
1030
1031
1032 ;;; ExternalID ::= 'SYSTEM' S SystemLiteral
1033 ;;;              | 'PUBLIC' S PubidLiteral S SystemLiteral
1034 ;;;
1035 ;;; PublicID ::= 'PUBLIC' S PubidLiteral
1036
1037
1038 (defun ebnf-dtd-externalid-or-publicid ()
1039   (let ((token (ebnf-dtd-lex)))
1040     (cond ((eq token 'system)
1041            (ebnf-dtd-systemliteral))
1042           ((eq token 'public)
1043            (ebnf-dtd-pubidliteral)
1044            (and (eq (setq token (ebnf-dtd-lex)) 'string)
1045                 (setq token (ebnf-dtd-lex)))
1046            token)
1047           (t
1048            (error "Missing `SYSTEM' or `PUBLIC'")))))
1049
1050 \f
1051 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1052 ;; Lexical analyzer
1053
1054
1055 (defconst ebnf-dtd-token-table (make-vector 256 'error)
1056   "Vector used to map characters to a lexical token.")
1057
1058
1059 (defun ebnf-dtd-initialize ()
1060   "Initialize EBNF token table."
1061   ;; control character & control 8-bit character are set to `error'
1062   (let ((char ?\060))
1063     ;; digits: 0-9
1064     (while (< char ?\072)
1065       (aset ebnf-dtd-token-table char 'name-char)
1066       (setq char (1+ char)))
1067     ;; printable character: A-Z
1068     (setq char ?\101)
1069     (while (< char ?\133)
1070       (aset ebnf-dtd-token-table char 'name)
1071       (setq char (1+ char)))
1072     ;; printable character: a-z
1073     (setq char ?\141)
1074     (while (< char ?\173)
1075       (aset ebnf-dtd-token-table char 'name)
1076       (setq char (1+ char)))
1077     ;; European 8-bit accentuated characters:
1078     (setq char ?\240)
1079     (while (< char ?\400)
1080       (aset ebnf-dtd-token-table char 'name)
1081       (setq char (1+ char)))
1082     ;; Override name characters:
1083     (aset ebnf-dtd-token-table ?_ 'name)
1084     (aset ebnf-dtd-token-table ?: 'name)
1085     (aset ebnf-dtd-token-table ?. 'name-char)
1086     (aset ebnf-dtd-token-table ?- 'name-char)
1087     ;; Override space characters:
1088     (aset ebnf-dtd-token-table ?\n 'space) ; [NL] linefeed
1089     (aset ebnf-dtd-token-table ?\r 'space) ; [CR] carriage return
1090     (aset ebnf-dtd-token-table ?\t 'space) ; [HT] horizontal tab
1091     (aset ebnf-dtd-token-table ?\  'space) ; [SP] space
1092     ;; Override other lexical characters:
1093     (aset ebnf-dtd-token-table ?=  'equal)
1094     (aset ebnf-dtd-token-table ?,  'comma)
1095     (aset ebnf-dtd-token-table ?*  'zero-or-more)
1096     (aset ebnf-dtd-token-table ?+  'one-or-more)
1097     (aset ebnf-dtd-token-table ?|  'alternative)
1098     (aset ebnf-dtd-token-table ?%  'percent)
1099     (aset ebnf-dtd-token-table ?&  'ampersand)
1100     (aset ebnf-dtd-token-table ?#  'hash)
1101     (aset ebnf-dtd-token-table ?\? 'interrogation)
1102     (aset ebnf-dtd-token-table ?\" 'double-quote)
1103     (aset ebnf-dtd-token-table ?\' 'single-quote)
1104     (aset ebnf-dtd-token-table ?<  'less-than)
1105     (aset ebnf-dtd-token-table ?>  'end-decl)
1106     (aset ebnf-dtd-token-table ?\( 'begin-group)
1107     (aset ebnf-dtd-token-table ?\) 'end-group)
1108     (aset ebnf-dtd-token-table ?\[ 'begin-subset)
1109     (aset ebnf-dtd-token-table ?\] 'end-subset)))
1110
1111
1112 ;; replace the range "\240-\377" (see `ebnf-range-regexp').
1113 (defconst ebnf-dtd-name-chars
1114   (ebnf-range-regexp "-._:0-9A-Za-z" ?\240 ?\377))
1115
1116
1117 (defconst ebnf-dtd-decl-alist
1118   '(("ATTLIST"  . attlist-decl)
1119     ("DOCTYPE"  . doctype-decl)
1120     ("ELEMENT"  . element-decl)
1121     ("ENTITY"   . entity-decl)
1122     ("NOTATION" . notation-decl)))
1123
1124
1125 (defconst ebnf-dtd-element-alist
1126   '(("#FIXED"    . fixed)
1127     ("#IMPLIED"  . implied)
1128     ("#PCDATA"   . pcdata)
1129     ("#REQUIRED" . required)))
1130
1131
1132 (defconst ebnf-dtd-name-alist
1133   '(("ANY"        . any)
1134     ("CDATA"      . cdata)
1135     ("EMPTY"      . empty)
1136     ("ENTITIES"   . entities)
1137     ("ENTITY"     . entity)
1138     ("ID"         . id)
1139     ("IDREF"      . idref)
1140     ("IDREFS"     . idrefs)
1141     ("NDATA"      . ndata)
1142     ("NMTOKEN"    . nmtoken)
1143     ("NMTOKENS"   . nmtokens)
1144     ("NOTATION"   . notation)
1145     ("PUBLIC"     . public)
1146     ("SYSTEM"     . system)
1147     ("encoding"   . encoding-attr)
1148     ("standalone" . standalone-attr)
1149     ("version"    . version-attr)))
1150
1151
1152 (defun ebnf-dtd-lex ()
1153   "Lexical analyzer for DTD.
1154
1155 Return a lexical token.
1156
1157 See documentation for variable `ebnf-dtd-lex'."
1158   (if (>= (point) ebnf-limit)
1159       'end-of-input
1160     (let (token)
1161       ;; skip spaces and comments
1162       (while (if (> (following-char) 255)
1163                  (progn
1164                    (setq token 'error)
1165                    nil)
1166                (setq token (aref ebnf-dtd-token-table (following-char)))
1167                (cond
1168                 ((eq token 'space)
1169                  (skip-chars-forward " \n\r\t" ebnf-limit)
1170                  (< (point) ebnf-limit))
1171                 ((and (eq token 'less-than)
1172                       (looking-at "<!--"))
1173                  (ebnf-dtd-skip-comment))
1174                 (t nil)
1175                 )))
1176       (cond
1177        ;; end of input
1178        ((>= (point) ebnf-limit)
1179         'end-of-input)
1180        ;; error
1181        ((eq token 'error)
1182         (error "Invalid character"))
1183        ;; beginning of declaration:
1184        ;; <?name, <!ATTLIST, <!DOCTYPE, <!ELEMENT, <!ENTITY, <!NOTATION
1185        ((eq token 'less-than)
1186         (forward-char)
1187         (let ((char (following-char)))
1188           (cond ((= char ?\?)           ; <?
1189                  (forward-char)
1190                  (setq ebnf-dtd-lex (ebnf-buffer-substring ebnf-dtd-name-chars))
1191                  'begin-pi)
1192                 ((= char ?!)            ; <!
1193                  (forward-char)
1194                  (let ((decl (ebnf-buffer-substring ebnf-dtd-name-chars)))
1195                    (or (cdr (assoc decl ebnf-dtd-decl-alist))
1196                        (error "Invalid declaration name `%s'" decl))))
1197                 (t                      ; <x
1198                  (error "Invalid declaration `<%c'" char)))))
1199        ;; name, namechar
1200        ((memq  token '(name name-char))
1201         (setq ebnf-dtd-lex (ebnf-buffer-substring ebnf-dtd-name-chars))
1202         (or (cdr (assoc ebnf-dtd-lex ebnf-dtd-name-alist))
1203             token))
1204        ;; ?, ?>
1205        ((eq token 'interrogation)
1206         (forward-char)
1207         (if (/= (following-char) ?>)
1208             'optional
1209           (forward-char)
1210           'end-pi))
1211        ;; #FIXED, #IMPLIED, #PCDATA, #REQUIRED
1212        ((eq token 'hash)
1213         (forward-char)
1214         (setq ebnf-dtd-lex
1215               (concat "#" (ebnf-buffer-substring ebnf-dtd-name-chars)))
1216         (or (cdr (assoc ebnf-dtd-lex ebnf-dtd-element-alist))
1217             (error "Invalid element `%s'" ebnf-dtd-lex)))
1218        ;; "string"
1219        ((eq token 'double-quote)
1220         (setq ebnf-dtd-lex (ebnf-dtd-string ?\"))
1221         'string)
1222        ;; 'string'
1223        ((eq token 'single-quote)
1224         (setq ebnf-dtd-lex (ebnf-dtd-string ?\'))
1225         'string)
1226        ;; %, %name;
1227        ((eq token 'percent)
1228         (forward-char)
1229         (if (looking-at "[ \n\r\t]")
1230             'percent
1231           (setq ebnf-dtd-lex (ebnf-dtd-name-ref "%"))
1232           'pe-ref))
1233        ;; &#...;, &#x...;, &name;
1234        ((eq token 'ampersand)
1235         (forward-char)
1236         (if (/= (following-char) ?#)
1237             (progn
1238               ;; &name;
1239               (setq ebnf-dtd-lex (ebnf-dtd-name-ref "&"))
1240               'entity-ref)
1241           ;; &#...;, &#x...;
1242           (forward-char)
1243           (setq ebnf-dtd-lex (if (/= (following-char) ?x)
1244                                  (ebnf-dtd-char-ref "&#" "0-9")
1245                                (forward-char)
1246                                (ebnf-dtd-char-ref "&#x" "0-9a-fA-F")))
1247           'char-ref))
1248        ;; miscellaneous: (, ), [, ], =, |, *, +, >, `,'
1249        (t
1250         (forward-char)
1251         token)
1252        ))))
1253
1254
1255 (defun ebnf-dtd-name-ref (start)
1256   (ebnf-dtd-char-ref start ebnf-dtd-name-chars))
1257
1258
1259 (defun ebnf-dtd-char-ref (start chars)
1260   (let ((char (ebnf-buffer-substring chars)))
1261     (or (= (following-char) ?\;)
1262         (error "Invalid element `%s%s%c'" start char (following-char)))
1263     (forward-char)
1264     (format "%s%s;" start char)))
1265
1266
1267 ;; replace the range "\240-\377" (see `ebnf-range-regexp').
1268 (defconst ebnf-dtd-double-string-chars
1269   (ebnf-range-regexp "\t -!#-~" ?\240 ?\377))
1270 (defconst ebnf-dtd-single-string-chars
1271   (ebnf-range-regexp "\t -&(-~" ?\240 ?\377))
1272
1273
1274 (defun ebnf-dtd-string (delim)
1275   (buffer-substring-no-properties
1276    (progn
1277      (forward-char)
1278      (point))
1279    (progn
1280      (skip-chars-forward (if (= delim ?\")
1281                              ebnf-dtd-double-string-chars
1282                            ebnf-dtd-single-string-chars)
1283                          ebnf-limit)
1284      (or (= (following-char) delim)
1285          (error "Missing string delimiter `%c'" delim))
1286      (prog1
1287          (point)
1288        (forward-char)))))
1289
1290
1291 ;; replace the range "\177-\237" (see `ebnf-range-regexp').
1292 (defconst ebnf-dtd-comment-chars
1293   (ebnf-range-regexp "^-\000-\010\013\014\016-\037" ?\177 ?\237))
1294 (defconst ebnf-dtd-filename-chars
1295   (ebnf-range-regexp "^-\000-\037" ?\177 ?\237))
1296
1297
1298 (defun ebnf-dtd-skip-comment ()
1299   (forward-char 4)                      ; <!--
1300   (cond
1301    ;; open EPS file
1302    ((and ebnf-eps-executing (= (following-char) ?\[))
1303     (ebnf-eps-add-context (ebnf-dtd-eps-filename)))
1304    ;; close EPS file
1305    ((and ebnf-eps-executing (= (following-char) ?\]))
1306     (ebnf-eps-remove-context (ebnf-dtd-eps-filename)))
1307    ;; EPS header
1308    ((and ebnf-eps-executing (= (following-char) ?H))
1309     (ebnf-eps-header-comment (ebnf-dtd-eps-filename)))
1310    ;; EPS footer
1311    ((and ebnf-eps-executing (= (following-char) ?F))
1312     (ebnf-eps-footer-comment (ebnf-dtd-eps-filename)))
1313    ;; any other action in comment
1314    (t
1315     (setq ebnf-action (aref ebnf-comment-table (following-char))))
1316    )
1317   (while (progn
1318            (skip-chars-forward ebnf-dtd-comment-chars ebnf-limit)
1319            (and (< (point) ebnf-limit)
1320                 (not (looking-at "-->"))))
1321     (skip-chars-forward "-" ebnf-limit))
1322   ;; check for a valid end of comment
1323   (cond ((>= (point) ebnf-limit)
1324          nil)
1325         ((looking-at "-->")
1326          (forward-char 3)
1327          t)
1328         (t
1329          (error "Invalid character"))
1330         ))
1331
1332
1333 (defun ebnf-dtd-eps-filename ()
1334   (forward-char)
1335   (let (fname)
1336     (while (progn
1337              (setq fname
1338                    (concat fname
1339                            (ebnf-buffer-substring ebnf-dtd-filename-chars)))
1340              (and (< (point) ebnf-limit)
1341                   (= (following-char) ?-)       ; may be \n, \t, \r
1342                   (not (looking-at "-->"))))
1343       (setq fname (concat fname (ebnf-buffer-substring "-"))))
1344     fname))
1345
1346 \f
1347 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1348
1349
1350 (provide 'ebnf-dtd)
1351
1352 ;; arch-tag: c21bb640-135f-4afa-8712-fa11d86301c4
1353 ;;; ebnf-dtd.el ends here