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
   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, or (at your option)
  16 ;; 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; see the file COPYING.  If not, write to the
  25 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  26 ;; Boston, MA 02110-1301, USA.
  27
  28 ;;; Commentary:
  29
  30 ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  31 ;;
  32 ;;
  33 ;; This is part of ebnf2ps package.
  34 ;;
  35 ;; This package defines a parser for DTD (Data Type Description for XML).
  36 ;;
  37 ;; See ebnf2ps.el for documentation.
  38 ;;
  39 ;;
  40 ;; DTD Syntax
  41 ;; ----------
  42 ;;
  43 ;;      See the URLs:
  44 ;;      `http://www.w3.org/TR/2004/REC-xml-20040204/'
  45 ;;      (Extensible Markup Language (XML) 1.0 (Third Edition))
  46 ;;      `http://www.w3.org/TR/html40/'
  47 ;;      (HTML 4.01 Specification)
  48 ;;      `http://www.w3.org/TR/NOTE-html-970421'
  49 ;;      (HTML DTD with support for Style Sheets)
  50 ;;
  51 ;;
  52 ;; /* Document */
  53 ;;
  54 ;; document ::= prolog element Misc*
  55 ;; /* Note that *only* the prolog will be parsed */
  56 ;;
  57 ;;
  58 ;; /* Characters */
  59 ;;
  60 ;; Char ::= #x9 | #xA | #xD
  61 ;;        | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
  62 ;; /* any Unicode character, excluding the surrogate blocks, FFFE, and FFFF. */
  63 ;;
  64 ;; /* NOTE:
  65 ;;
  66 ;;    Document authors are encouraged to avoid "compatibility characters", as
  67 ;;    defined in section 6.8 of [Unicode] (see also D21 in section 3.6 of
  68 ;;    [Unicode3]). The characters defined in the following ranges are also
  69 ;;    discouraged. They are either control characters or permanently undefined
  70 ;;    Unicode characters:
  71 ;;
  72 ;;    [#x7F-#x84],      [#x86-#x9F],      [#xFDD0-#xFDDF],
  73 ;;    [#1FFFE-#x1FFFF], [#2FFFE-#x2FFFF], [#3FFFE-#x3FFFF],
  74 ;;    [#4FFFE-#x4FFFF], [#5FFFE-#x5FFFF], [#6FFFE-#x6FFFF],
  75 ;;    [#7FFFE-#x7FFFF], [#8FFFE-#x8FFFF], [#9FFFE-#x9FFFF],
  76 ;;    [#AFFFE-#xAFFFF], [#BFFFE-#xBFFFF], [#CFFFE-#xCFFFF],
  77 ;;    [#DFFFE-#xDFFFF], [#EFFFE-#xEFFFF], [#FFFFE-#xFFFFF],
  78 ;;    [#10FFFE-#x10FFFF]. */
  79 ;;
  80 ;;
  81 ;; /* White Space */
  82 ;;
  83 ;; S ::= (#x20 | #x9 | #xD | #xA)+
  84 ;;
  85 ;;
  86 ;; /* Names and Tokens */
  87 ;;
  88 ;; NameChar ::= Letter | Digit | '.' | '-' | '_' | ':'
  89 ;;            | CombiningChar | Extender
  90 ;;
  91 ;; Name ::= (Letter | '_' | ':') (NameChar)*
  92 ;;
  93 ;; Names ::= Name (#x20 Name)*
  94 ;;
  95 ;; Nmtoken ::= (NameChar)+
  96 ;;
  97 ;; Nmtokens ::= Nmtoken (#x20 Nmtoken)*
  98 ;;
  99 ;;
 100 ;; /* Literals */
 101 ;;
 102 ;; EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 103 ;;               | "'" ([^%&'] | PEReference | Reference)* "'"
 104 ;;
 105 ;; AttValue ::= '"' ([^<&"] | Reference)* '"'
 106 ;;            | "'" ([^<&'] | Reference)* "'"
 107 ;;
 108 ;; SystemLiteral ::= ('"' [^"]* '"')
 109 ;;                 | ("'" [^']* "'")
 110 ;;
 111 ;; PubidLiteral ::= '"' PubidChar* '"'
 112 ;;                | "'" (PubidChar - "'")* "'"
 113 ;;
 114 ;; PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%]
 115 ;;
 116 ;; /* NOTE:
 117 ;;
 118 ;;    Although the EntityValue production allows the definition of a general
 119 ;;    entity consisting of a single explicit < in the literal (e.g., <!ENTITY
 120 ;;    mylt "<">), it is strongly advised to avoid this practice since any
 121 ;;    reference to that entity will cause a well-formedness error. */
 122 ;;
 123 ;;
 124 ;; /* Character Data */
 125 ;;
 126 ;; CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
 127 ;;
 128 ;;
 129 ;; /* Comments */
 130 ;;
 131 ;; Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->'
 132 ;;
 133 ;;
 134 ;; /* Processing Instructions */
 135 ;;
 136 ;; PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
 137 ;;
 138 ;; PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
 139 ;;
 140 ;;
 141 ;; /* CDATA Sections */
 142 ;;
 143 ;; CDSect ::= CDStart CData CDEnd
 144 ;;
 145 ;; CDStart ::= '<![CDATA['
 146 ;;
 147 ;; CData ::= (Char* - (Char* ']]>' Char*))
 148 ;;
 149 ;; CDEnd ::= ']]>'
 150 ;;
 151 ;;
 152 ;; /* Prolog */
 153 ;;
 154 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
 155 ;;
 156 ;; XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
 157 ;;
 158 ;; VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"')
 159 ;;
 160 ;; Eq ::= S? '=' S?
 161 ;;
 162 ;; VersionNum ::= '1.0'
 163 ;;
 164 ;; Misc ::= Comment | PI | S
 165 ;;
 166 ;;
 167 ;; /* Document Type Definition */
 168 ;;
 169 ;; doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
 170 ;;                 ('[' intSubset ']' S?)? '>'
 171 ;;               [VC: Root Element Type]
 172 ;;               [WFC: External Subset]
 173 ;;
 174 ;; DeclSep ::= PEReference | S
 175 ;;           [WFC: PE Between Declarations]
 176 ;;
 177 ;; intSubset ::= (markupdecl | DeclSep)*
 178 ;;
 179 ;; markupdecl ::= elementdecl | AttlistDecl | EntityDecl
 180 ;;              | NotationDecl | PI | Comment
 181 ;;              [VC: Proper Declaration/PE Nesting]
 182 ;;              [WFC: PEs in Internal Subset]
 183 ;;
 184 ;;
 185 ;; /* External Subset */
 186 ;;
 187 ;; extSubset ::= TextDecl? extSubsetDecl
 188 ;;
 189 ;; extSubsetDecl ::= ( markupdecl | conditionalSect | DeclSep)*
 190 ;;
 191 ;;
 192 ;; /* Standalone Document Declaration */
 193 ;;
 194 ;; SDDecl ::= S 'standalone' Eq
 195 ;;            (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"'))
 196 ;;          [VC: Standalone Document Declaration]
 197 ;;
 198 ;;
 199 ;; /* Element */
 200 ;;
 201 ;; element ::= EmptyElemTag | STag content ETag
 202 ;;           [WFC: Element Type Match]
 203 ;;           [VC: Element Valid]
 204 ;;
 205 ;;
 206 ;; /* Start-tag */
 207 ;;
 208 ;; STag ::= '<' Name (S Attribute)* S? '>'
 209 ;;        [WFC: Unique Att Spec]
 210 ;;
 211 ;; Attribute ::= Name Eq AttValue
 212 ;;             [VC: Attribute Value Type]
 213 ;;             [WFC: No External Entity References]
 214 ;;             [WFC: No < in Attribute Values]
 215 ;;
 216 ;;
 217 ;; /* End-tag */
 218 ;;
 219 ;; ETag ::= '</' Name S? '>'
 220 ;;
 221 ;;
 222 ;; /* Content of Elements */
 223 ;;
 224 ;; content ::= CharData?
 225 ;;             ((element | Reference | CDSect | PI | Comment) CharData?)*
 226 ;;
 227 ;;
 228 ;; /* Tags for Empty Elements */
 229 ;;
 230 ;; EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
 231 ;;                [WFC: Unique Att Spec]
 232 ;;
 233 ;;
 234 ;; /* Element Type Declaration */
 235 ;;
 236 ;; elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
 237 ;;               [VC: Unique Element Type Declaration]
 238 ;;
 239 ;; contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
 240 ;;
 241 ;;
 242 ;; /* Element-content Models */
 243 ;;
 244 ;; children ::= (choice | seq) ('?' | '*' | '+')?
 245 ;;
 246 ;; cp ::= (Name | choice | seq) ('?' | '*' | '+')?
 247 ;;
 248 ;; choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
 249 ;;          [VC: Proper Group/PE Nesting]
 250 ;;
 251 ;; seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
 252 ;;       [VC: Proper Group/PE Nesting]
 253 ;;
 254 ;;
 255 ;; /* Mixed-content Declaration */
 256 ;;
 257 ;; Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'
 258 ;;         | '(' S? '#PCDATA' S? ')'
 259 ;;         [VC: Proper Group/PE Nesting]
 260 ;;         [VC: No Duplicate Types]
 261 ;;
 262 ;;
 263 ;; /* Attribute-list Declaration */
 264 ;;
 265 ;; AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
 266 ;;
 267 ;; AttDef ::= S Name S AttType S DefaultDecl
 268 ;;
 269 ;;
 270 ;; /* Attribute Types */
 271 ;;
 272 ;; AttType ::= StringType | TokenizedType | EnumeratedType
 273 ;;
 274 ;; StringType ::= 'CDATA'
 275 ;;
 276 ;; TokenizedType ::= 'ID'       [VC: ID]
 277 ;;                              [VC: One ID per Element Type]
 278 ;;                              [VC: ID Attribute Default]
 279 ;;                 | 'IDREF'    [VC: IDREF]
 280 ;;                 | 'IDREFS'   [VC: IDREF]
 281 ;;                 | 'ENTITY'   [VC: Entity Name]
 282 ;;                 | 'ENTITIES' [VC: Entity Name]
 283 ;;                 | 'NMTOKEN'  [VC: Name Token]
 284 ;;                 | 'NMTOKENS' [VC: Name Token]
 285 ;;
 286 ;;
 287 ;; /* Enumerated Attribute Types */
 288 ;;
 289 ;; EnumeratedType ::= NotationType | Enumeration
 290 ;;
 291 ;; NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 292 ;;                [VC: Notation Attributes]
 293 ;;                [VC: One Notation Per Element Type]
 294 ;;                [VC: No Notation on Empty Element]
 295 ;;                [VC: No Duplicate Tokens]
 296 ;;
 297 ;; Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
 298 ;;               [VC: Enumeration]
 299 ;;               [VC: No Duplicate Tokens]
 300 ;;
 301 ;;
 302 ;; /* Attribute Defaults */
 303 ;;
 304 ;; DefaultDecl ::= '#REQUIRED' | '#IMPLIED'
 305 ;;               | (('#FIXED' S)? AttValue)
 306 ;;               [VC: Required Attribute]
 307 ;;               [VC: Attribute Default Value Syntactically Correct]
 308 ;;               [WFC: No < in Attribute Values]
 309 ;;               [VC: Fixed Attribute Default]
 310 ;;
 311 ;;
 312 ;; /* Conditional Section */
 313 ;;
 314 ;; conditionalSect ::= includeSect | ignoreSect
 315 ;;
 316 ;; includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
 317 ;;               [VC: Proper Conditional Section/PE Nesting]
 318 ;;
 319 ;; ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
 320 ;;              [VC: Proper Conditional Section/PE Nesting]
 321 ;;
 322 ;; ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)*
 323 ;;
 324 ;; Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
 325 ;;
 326 ;;
 327 ;; /* Character Reference */
 328 ;;
 329 ;; CharRef ::= '&#' [0-9]+ ';'
 330 ;;           | '&#x' [0-9a-fA-F]+ ';'
 331 ;;           [WFC: Legal Character]
 332 ;;
 333 ;;
 334 ;; /* Entity Reference */
 335 ;;
 336 ;; Reference ::= EntityRef | CharRef
 337 ;;
 338 ;; EntityRef ::= '&' Name ';'
 339 ;;             [WFC: Entity Declared]
 340 ;;             [VC: Entity Declared]
 341 ;;             [WFC: Parsed Entity]
 342 ;;             [WFC: No Recursion]
 343 ;;
 344 ;; PEReference ::= '%' Name ';'
 345 ;;               [VC: Entity Declared]
 346 ;;               [WFC: No Recursion]
 347 ;;               [WFC: In DTD]
 348 ;;
 349 ;;
 350 ;; /* Entity Declaration */
 351 ;;
 352 ;; EntityDecl ::= GEDecl | PEDecl
 353 ;;
 354 ;; GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
 355 ;;
 356 ;; PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 357 ;;
 358 ;; EntityDef ::= EntityValue | (ExternalID NDataDecl?)
 359 ;;
 360 ;; PEDef ::= EntityValue | ExternalID
 361 ;;
 362 ;;
 363 ;; /* External Entity Declaration */
 364 ;;
 365 ;; ExternalID ::= 'SYSTEM' S SystemLiteral
 366 ;;              | 'PUBLIC' S PubidLiteral S SystemLiteral
 367 ;;
 368 ;; NDataDecl ::= S 'NDATA' S Name
 369 ;;             [VC: Notation Declared]
 370 ;;
 371 ;;
 372 ;; /* Text Declaration */
 373 ;;
 374 ;; TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>'
 375 ;;
 376 ;;
 377 ;; /* Well-Formed External Parsed Entity */
 378 ;;
 379 ;; extParsedEnt ::= TextDecl? content
 380 ;;
 381 ;;
 382 ;; /* Encoding Declaration */
 383 ;;
 384 ;; EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'" )
 385 ;;
 386 ;; EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
 387 ;; /* Encoding name contains only Latin characters */
 388 ;;
 389 ;;
 390 ;; /* Notation Declarations */
 391 ;;
 392 ;; NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>'
 393 ;;                [VC: Unique Notation Name]
 394 ;;
 395 ;; PublicID ::= 'PUBLIC' S PubidLiteral
 396 ;;
 397 ;;
 398 ;; /* Characters */
 399 ;;
 400 ;; Letter ::= BaseChar | Ideographic
 401 ;;
 402 ;; BaseChar ::= [#x0041-#x005A] | [#x0061-#x007A] | [#x00C0-#x00D6]
 403 ;;            | [#x00D8-#x00F6] | [#x00F8-#x00FF] | [#x0100-#x0131]
 404 ;;            | [#x0134-#x013E] | [#x0141-#x0148] | [#x014A-#x017E]
 405 ;;            | [#x0180-#x01C3] | [#x01CD-#x01F0] | [#x01F4-#x01F5]
 406 ;;            | [#x01FA-#x0217] | [#x0250-#x02A8] | [#x02BB-#x02C1]
 407 ;;            | #x0386          | [#x0388-#x038A] | #x038C
 408 ;;            | [#x038E-#x03A1] | [#x03A3-#x03CE] | [#x03D0-#x03D6]
 409 ;;            | #x03DA          | #x03DC          | #x03DE
 410 ;;            | #x03E0          | [#x03E2-#x03F3] | [#x0401-#x040C]
 411 ;;            | [#x040E-#x044F] | [#x0451-#x045C] | [#x045E-#x0481]
 412 ;;            | [#x0490-#x04C4] | [#x04C7-#x04C8] | [#x04CB-#x04CC]
 413 ;;            | [#x04D0-#x04EB] | [#x04EE-#x04F5] | [#x04F8-#x04F9]
 414 ;;            | [#x0531-#x0556] | #x0559          | [#x0561-#x0586]
 415 ;;            | [#x05D0-#x05EA] | [#x05F0-#x05F2] | [#x0621-#x063A]
 416 ;;            | [#x0641-#x064A] | [#x0671-#x06B7] | [#x06BA-#x06BE]
 417 ;;            | [#x06C0-#x06CE] | [#x06D0-#x06D3] | #x06D5
 418 ;;            | [#x06E5-#x06E6] | [#x0905-#x0939] | #x093D
 419 ;;            | [#x0958-#x0961] | [#x0985-#x098C] | [#x098F-#x0990]
 420 ;;            | [#x0993-#x09A8] | [#x09AA-#x09B0] | #x09B2
 421 ;;            | [#x09B6-#x09B9] | [#x09DC-#x09DD] | [#x09DF-#x09E1]
 422 ;;            | [#x09F0-#x09F1] | [#x0A05-#x0A0A] | [#x0A0F-#x0A10]
 423 ;;            | [#x0A13-#x0A28] | [#x0A2A-#x0A30] | [#x0A32-#x0A33]
 424 ;;            | [#x0A35-#x0A36] | [#x0A38-#x0A39] | [#x0A59-#x0A5C]
 425 ;;            | #x0A5E          | [#x0A72-#x0A74] | [#x0A85-#x0A8B]
 426 ;;            | #x0A8D          | [#x0A8F-#x0A91] | [#x0A93-#x0AA8]
 427 ;;            | [#x0AAA-#x0AB0] | [#x0AB2-#x0AB3] | [#x0AB5-#x0AB9]
 428 ;;            | #x0ABD          | #x0AE0          | [#x0B05-#x0B0C]
 429 ;;            | [#x0B0F-#x0B10] | [#x0B13-#x0B28] | [#x0B2A-#x0B30]
 430 ;;            | [#x0B32-#x0B33] | [#x0B36-#x0B39] | #x0B3D
 431 ;;            | [#x0B5C-#x0B5D] | [#x0B5F-#x0B61] | [#x0B85-#x0B8A]
 432 ;;            | [#x0B8E-#x0B90] | [#x0B92-#x0B95] | [#x0B99-#x0B9A]
 433 ;;            | #x0B9C          | [#x0B9E-#x0B9F] | [#x0BA3-#x0BA4]
 434 ;;            | [#x0BA8-#x0BAA] | [#x0BAE-#x0BB5] | [#x0BB7-#x0BB9]
 435 ;;            | [#x0C05-#x0C0C] | [#x0C0E-#x0C10] | [#x0C12-#x0C28]
 436 ;;            | [#x0C2A-#x0C33] | [#x0C35-#x0C39] | [#x0C60-#x0C61]
 437 ;;            | [#x0C85-#x0C8C] | [#x0C8E-#x0C90] | [#x0C92-#x0CA8]
 438 ;;            | [#x0CAA-#x0CB3] | [#x0CB5-#x0CB9] | #x0CDE
 439 ;;            | [#x0CE0-#x0CE1] | [#x0D05-#x0D0C] | [#x0D0E-#x0D10]
 440 ;;            | [#x0D12-#x0D28] | [#x0D2A-#x0D39] | [#x0D60-#x0D61]
 441 ;;            | [#x0E01-#x0E2E] | #x0E30          | [#x0E32-#x0E33]
 442 ;;            | [#x0E40-#x0E45] | [#x0E81-#x0E82] | #x0E84
 443 ;;            | [#x0E87-#x0E88] | #x0E8A          | #x0E8D
 444 ;;            | [#x0E94-#x0E97] | [#x0E99-#x0E9F] | [#x0EA1-#x0EA3]
 445 ;;            | #x0EA5          | #x0EA7          | [#x0EAA-#x0EAB]
 446 ;;            | [#x0EAD-#x0EAE] | #x0EB0          | [#x0EB2-#x0EB3]
 447 ;;            | #x0EBD          | [#x0EC0-#x0EC4] | [#x0F40-#x0F47]
 448 ;;            | [#x0F49-#x0F69] | [#x10A0-#x10C5] | [#x10D0-#x10F6]
 449 ;;            | #x1100          | [#x1102-#x1103] | [#x1105-#x1107]
 450 ;;            | #x1109          | [#x110B-#x110C] | [#x110E-#x1112]
 451 ;;            | #x113C          | #x113E          | #x1140
 452 ;;            | #x114C          | #x114E          | #x1150
 453 ;;            | [#x1154-#x1155] | #x1159          | [#x115F-#x1161]
 454 ;;            | #x1163          | #x1165          | #x1167
 455 ;;            | #x1169          | [#x116D-#x116E] | [#x1172-#x1173]
 456 ;;            | #x1175          | #x119E          | #x11A8
 457 ;;            | #x11AB          | [#x11AE-#x11AF] | [#x11B7-#x11B8]
 458 ;;            | #x11BA          | [#x11BC-#x11C2] | #x11EB
 459 ;;            | #x11F0          | #x11F9          | [#x1E00-#x1E9B]
 460 ;;            | [#x1EA0-#x1EF9] | [#x1F00-#x1F15] | [#x1F18-#x1F1D]
 461 ;;            | [#x1F20-#x1F45] | [#x1F48-#x1F4D] | [#x1F50-#x1F57]
 462 ;;            | #x1F59          | #x1F5B          | #x1F5D
 463 ;;            | [#x1F5F-#x1F7D] | [#x1F80-#x1FB4] | [#x1FB6-#x1FBC]
 464 ;;            | #x1FBE          | [#x1FC2-#x1FC4] | [#x1FC6-#x1FCC]
 465 ;;            | [#x1FD0-#x1FD3] | [#x1FD6-#x1FDB] | [#x1FE0-#x1FEC]
 466 ;;            | [#x1FF2-#x1FF4] | [#x1FF6-#x1FFC] | #x2126
 467 ;;            | [#x212A-#x212B] | #x212E          | [#x2180-#x2182]
 468 ;;            | [#x3041-#x3094] | [#x30A1-#x30FA] | [#x3105-#x312C]
 469 ;;            | [#xAC00-#xD7A3]
 470 ;;
 471 ;; Ideographic ::= [#x4E00-#x9FA5] | #x3007 | [#x3021-#x3029]
 472 ;;
 473 ;; CombiningChar ::= [#x0300-#x0345] | [#x0360-#x0361] | [#x0483-#x0486]
 474 ;;                 | [#x0591-#x05A1] | [#x05A3-#x05B9] | [#x05BB-#x05BD]
 475 ;;                 | #x05BF          | [#x05C1-#x05C2] | #x05C4
 476 ;;                 | [#x064B-#x0652] | #x0670          | [#x06D6-#x06DC]
 477 ;;                 | [#x06DD-#x06DF] | [#x06E0-#x06E4] | [#x06E7-#x06E8]
 478 ;;                 | [#x06EA-#x06ED] | [#x0901-#x0903] | #x093C
 479 ;;                 | [#x093E-#x094C] | #x094D          | [#x0951-#x0954]
 480 ;;                 | [#x0962-#x0963] | [#x0981-#x0983] | #x09BC
 481 ;;                 | #x09BE          | #x09BF          | [#x09C0-#x09C4]
 482 ;;                 | [#x09C7-#x09C8] | [#x09CB-#x09CD] | #x09D7
 483 ;;                 | [#x09E2-#x09E3] | #x0A02          | #x0A3C
 484 ;;                 | #x0A3E          | #x0A3F          | [#x0A40-#x0A42]
 485 ;;                 | [#x0A47-#x0A48] | [#x0A4B-#x0A4D] | [#x0A70-#x0A71]
 486 ;;                 | [#x0A81-#x0A83] | #x0ABC          | [#x0ABE-#x0AC5]
 487 ;;                 | [#x0AC7-#x0AC9] | [#x0ACB-#x0ACD] | [#x0B01-#x0B03]
 488 ;;                 | #x0B3C          | [#x0B3E-#x0B43] | [#x0B47-#x0B48]
 489 ;;                 | [#x0B4B-#x0B4D] | [#x0B56-#x0B57] | [#x0B82-#x0B83]
 490 ;;                 | [#x0BBE-#x0BC2] | [#x0BC6-#x0BC8] | [#x0BCA-#x0BCD]
 491 ;;                 | #x0BD7          | [#x0C01-#x0C03] | [#x0C3E-#x0C44]
 492 ;;                 | [#x0C46-#x0C48] | [#x0C4A-#x0C4D] | [#x0C55-#x0C56]
 493 ;;                 | [#x0C82-#x0C83] | [#x0CBE-#x0CC4] | [#x0CC6-#x0CC8]
 494 ;;                 | [#x0CCA-#x0CCD] | [#x0CD5-#x0CD6] | [#x0D02-#x0D03]
 495 ;;                 | [#x0D3E-#x0D43] | [#x0D46-#x0D48] | [#x0D4A-#x0D4D]
 496 ;;                 | #x0D57          | #x0E31          | [#x0E34-#x0E3A]
 497 ;;                 | [#x0E47-#x0E4E] | #x0EB1          | [#x0EB4-#x0EB9]
 498 ;;                 | [#x0EBB-#x0EBC] | [#x0EC8-#x0ECD] | [#x0F18-#x0F19]
 499 ;;                 | #x0F35          | #x0F37          | #x0F39
 500 ;;                 | #x0F3E          | #x0F3F          | [#x0F71-#x0F84]
 501 ;;                 | [#x0F86-#x0F8B] | [#x0F90-#x0F95] | #x0F97
 502 ;;                 | [#x0F99-#x0FAD] | [#x0FB1-#x0FB7] | #x0FB9
 503 ;;                 | [#x20D0-#x20DC] | #x20E1          | [#x302A-#x302F]
 504 ;;                 | #x3099          | #x309A
 505 ;;
 506 ;; Digit ::= [#x0030-#x0039] | [#x0660-#x0669] | [#x06F0-#x06F9]
 507 ;;         | [#x0966-#x096F] | [#x09E6-#x09EF] | [#x0A66-#x0A6F]
 508 ;;         | [#x0AE6-#x0AEF] | [#x0B66-#x0B6F] | [#x0BE7-#x0BEF]
 509 ;;         | [#x0C66-#x0C6F] | [#x0CE6-#x0CEF] | [#x0D66-#x0D6F]
 510 ;;         | [#x0E50-#x0E59] | [#x0ED0-#x0ED9] | [#x0F20-#x0F29]
 511 ;;
 512 ;; Extender ::= #x00B7 | #x02D0 | #x02D1 | #x0387 | #x0640 | #x0E46 | #x0EC6
 513 ;;            | #x3005 | [#x3031-#x3035] | [#x309D-#x309E] | [#x30FC-#x30FE]
 514 ;;
 515 ;;
 516 ;; NOTES
 517 ;; -----
 518 ;;
 519 ;; At moment, only the `<!ELEMENT' generates a syntactic chart.  The
 520 ;; `<!ATTLIST', `<!NOTATION' and `<!ENTITY' are syntacticly checked but they
 521 ;; don't generate a syntactic chart.
 522 ;;
 523 ;; Besides the syntax above, ebnf-dtd also accepts a `pure' dtd file.  An
 524 ;; example of a `pure' dtd file is:
 525 ;;
 526 ;;    <?xml version="1.0" encoding="UTF-8"?>
 527 ;;    <!--
 528 ;;    The main element.
 529 ;;    -->
 530 ;;    <!ELEMENT workflow (registers?, trigger-functions?, initial-actions,
 531 ;;                        steps, splits?, joins?)>
 532 ;;    <!--
 533 ;;    An action that can be executed (id must be unique among actions for
 534 ;;    the enclosing step).
 535 ;;    Used in: actions
 536 ;;    -->
 537 ;;    <!ELEMENT action (restrict-to, validators?, pre-functions?, results,
 538 ;;                      post-functions?)>
 539 ;;    <!ATTLIST action
 540 ;;      id CDATA #REQUIRED
 541 ;;      name CDATA #REQUIRED
 542 ;;    >
 543 ;;
 544 ;;
 545 ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 546
 547 ;;; Code:
 548
 549
 550 (require 'ebnf-otz)
 551
 552
 553 (defvar ebnf-dtd-lex nil
 554   "Value returned by `ebnf-dtd-lex' function.")
 555
 556 \f
 557 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 558 ;; Syntactic analyzer
 559
 560
 561 ;;; document ::= prolog element Misc*
 562 ;;; /* Note that *only* the prolog will be parsed */
 563
 564 (defun ebnf-dtd-parser (start)
 565   "DTD parser."
 566   (let ((total (+ (- ebnf-limit start) 1))
 567         (bias (1- start))
 568         (origin (point))
 569         rule-list token rule the-end)
 570     (goto-char start)
 571     (setq token (ebnf-dtd-lex))
 572     (and (eq token 'end-of-input)
 573          (error "Empty DTD file"))
 574     (setq token (ebnf-dtd-prolog token))
 575     (unless (eq (car token) 'end-prolog)
 576       (setq the-end (cdr token)
 577             token   (car token))
 578       (while (not (eq token the-end))
 579         (ebnf-message-float
 580          "Parsing...%s%%"
 581          (/ (* (- (point) bias) 100.0) total))
 582         (setq token (ebnf-dtd-intsubset token)
 583               rule  (cdr token)
 584               token (car token))
 585         (or (null rule)
 586             (ebnf-add-empty-rule-list rule)
 587             (setq rule-list (cons rule rule-list))))
 588       (or (eq the-end 'end-of-input)
 589           (eq (ebnf-dtd-lex) 'end-decl)
 590           (error "Missing end of DOCTYPE"))
 591       ;; adjust message, 'cause *only* prolog will be parsed
 592       (ebnf-message-float "Parsing...%s%%" 100.0))
 593     (goto-char origin)
 594     rule-list))
 595
 596
 597 ;;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
 598 ;;;
 599 ;;; XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
 600 ;;;
 601 ;;; VersionInfo ::= S 'version' Eq ("'" VersionNum "'" | '"' VersionNum '"')
 602 ;;;
 603 ;;; Eq ::= S? '=' S?
 604 ;;;
 605 ;;; VersionNum ::= '1.0'
 606 ;;;
 607 ;;; Misc ::= Comment | PI | S
 608 ;;;
 609 ;;; EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'" )
 610 ;;;
 611 ;;; EncName ::= [A-Za-z] ([-A-Za-z0-9._])*
 612 ;;; /* Encoding name contains only Latin characters */
 613 ;;;
 614 ;;; SDDecl ::= S 'standalone' Eq
 615 ;;;            (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no') '"'))
 616 ;;;
 617 ;;; doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
 618 ;;;                 ('[' intSubset ']' S?)? '>'
 619
 620
 621 (defun ebnf-dtd-prolog (token)
 622   (when (and (eq token 'begin-pi) (string= ebnf-dtd-lex "xml"))
 623     ;; version = "1.0"
 624     (setq token (ebnf-dtd-attribute (ebnf-dtd-lex) 'version-attr
 625                                     "^1\\.0$" "XML version"))
 626     ;; ( encoding = "encoding name" )?
 627     (setq token (ebnf-dtd-attribute-optional
 628                  token 'encoding-attr
 629                  "^[A-Za-z][-A-Za-z0-9._]*$" "XML encoding"))
 630     ;; ( standalone = ( "yes" | "no" ) )?
 631     (setq token (ebnf-dtd-attribute-optional
 632                  token 'standalone-attr
 633                  "^yes|no$" "XML standalone"))
 634     (or (eq token 'end-pi)
 635         (error "Missing end of XML processing instruction")))
 636   ;; processing instructions
 637   (setq token (ebnf-dtd-pi (ebnf-dtd-lex)))
 638   (cond
 639    ;; DOCTYPE
 640    ((eq token 'doctype-decl)
 641     (or (eq (ebnf-dtd-lex) 'name)
 642         (error "Document type name is missing"))
 643     (cons (if (eq (ebnf-dtd-externalid) 'begin-subset)
 644               (ebnf-dtd-lex)
 645             'end-prolog)
 646           'end-subset))
 647    ((memq token '(element-decl attlist-decl entity-decl notation-decl))
 648     (cons token 'end-of-input))
 649    (t
 650     '(end-prolog . end-subset))
 651    ))
 652
 653
 654 (defun ebnf-dtd-attribute (token attr match attr-name)
 655   (or (eq token attr)
 656       (error "%s attribute is missing" attr-name))
 657   (ebnf-dtd-attribute-optional token attr match attr-name))
 658
 659
 660 (defun ebnf-dtd-attribute-optional (token attr match attr-name)
 661   (when (eq token attr)
 662     (or (and (eq (ebnf-dtd-lex) 'equal)
 663              (eq (ebnf-dtd-lex) 'string)
 664              (string-match match ebnf-dtd-lex))
 665         (error "XML %s attribute is invalid" attr-name))
 666     (setq token (ebnf-dtd-lex)))
 667   token)
 668
 669
 670 ;;; ExternalID ::= 'SYSTEM' S SystemLiteral
 671 ;;;              | 'PUBLIC' S PubidLiteral S SystemLiteral
 672
 673
 674 (defun ebnf-dtd-externalid (&optional token)
 675   (let ((must-have token))
 676     (or token (setq token (ebnf-dtd-lex)))
 677     (cond ((eq token 'system)
 678            (ebnf-dtd-systemliteral))
 679           ((eq token 'public)
 680            (ebnf-dtd-pubidliteral)
 681            (ebnf-dtd-systemliteral))
 682           (must-have
 683            (error "Missing `SYSTEM' or `PUBLIC' in external id"))
 684           (t
 685            token))))
 686
 687
 688 ;;; SystemLiteral ::= ('"' [^"]* '"')
 689 ;;;                 | ("'" [^']* "'")
 690
 691
 692 (defun ebnf-dtd-systemliteral ()
 693   (or (eq (ebnf-dtd-lex) 'string)
 694       (error "System identifier is invalid"))
 695   (ebnf-dtd-lex))
 696
 697
 698 ;;; PubidLiteral ::= '"' PubidChar* '"'
 699 ;;;                | "'" (PubidChar - "'")* "'"
 700 ;;;
 701 ;;; PubidChar ::= [-'()+,./:=?;!*#@$_%\n\r a-zA-Z0-9]
 702
 703
 704 (defun ebnf-dtd-pubidliteral ()
 705   (or (and (eq (ebnf-dtd-lex) 'string)
 706            (string-match "^[-'()+,./:=?;!*#@$_%\n\r a-zA-Z0-9]*$"
 707                          ebnf-dtd-lex))
 708       (error "Public identifier is invalid")))
 709
 710
 711 ;;; PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>'
 712 ;;;
 713 ;;; PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l'))
 714
 715
 716 (defun ebnf-dtd-pi (token)
 717   (while (eq token 'begin-pi)
 718     (and (string-match "^[xX][mM][lL]$" ebnf-dtd-lex)
 719          (error "Processing instruction name can not be `XML'"))
 720     (while (not (eq (ebnf-dtd-lex) 'end-pi)))
 721     (setq token (ebnf-dtd-lex)))
 722   token)
 723
 724
 725 ;;; doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
 726 ;;;                 ('[' intSubset ']' S?)? '>'
 727 ;;;
 728 ;;; intSubset ::= (markupdecl | DeclSep)*
 729 ;;;
 730 ;;; DeclSep ::= PEReference | S
 731 ;;;
 732 ;;; markupdecl ::= elementdecl | AttlistDecl | EntityDecl
 733 ;;;              | NotationDecl | PI | Comment
 734
 735
 736 (defun ebnf-dtd-intsubset (token)
 737   ;; PI - Processing Instruction
 738   (and (eq token 'begin-pi)
 739        (setq token (ebnf-dtd-pi token)))
 740   (cond
 741    ((memq token '(end-subset end-of-input))
 742     (cons token nil))
 743    ((eq token 'pe-ref)
 744     (cons (ebnf-dtd-lex) nil))          ; annotation
 745    ((eq token 'element-decl)
 746     (ebnf-dtd-elementdecl))             ; rule
 747    ((eq token 'attlist-decl)
 748     (ebnf-dtd-attlistdecl))             ; annotation
 749    ((eq token 'entity-decl)
 750     (ebnf-dtd-entitydecl))              ; annotation
 751    ((eq token 'notation-decl)
 752     (ebnf-dtd-notationdecl))            ; annotation
 753    (t
 754     (error "Invalid DOCTYPE element"))
 755    ))
 756
 757
 758 ;;; elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
 759 ;;;
 760 ;;; contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
 761 ;;;
 762 ;;; Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'
 763 ;;;         | '(' S? '#PCDATA' S? ')'
 764 ;;;
 765 ;;; children ::= (choice | seq) ('?' | '*' | '+')?
 766 ;;;
 767 ;;; choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
 768 ;;;
 769 ;;; seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
 770 ;;;
 771 ;;; cp ::= (Name | choice | seq) ('?' | '*' | '+')?
 772
 773
 774 (defun ebnf-dtd-elementdecl ()
 775   (let ((action ebnf-action)
 776         name token body)
 777     (setq ebnf-action nil)
 778     (or (eq (ebnf-dtd-lex) 'name)
 779         (error "Invalid ELEMENT name"))
 780     (setq name  ebnf-dtd-lex
 781           token (ebnf-dtd-lex)
 782           body  (cond ((memq token '(empty any))
 783                        (let ((term (ebnf-make-terminal ebnf-dtd-lex)))
 784                          (cons (ebnf-dtd-lex) term)))
 785                       ((eq token 'begin-group)
 786                        (setq token (ebnf-dtd-lex))
 787                        (if (eq token 'pcdata)
 788                            (ebnf-dtd-mixed)
 789                          (ebnf-dtd-children token)))
 790                       (t
 791                        (error "Invalid ELEMENT content"))
 792                       ))
 793     (or (eq (car body) 'end-decl)
 794         (error "Missing `>' in ELEMENT declaration"))
 795     (ebnf-eps-add-production name)
 796     (cons (ebnf-dtd-lex)
 797           (ebnf-make-production name (cdr body) action))))
 798
 799
 800 ;;; Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*'
 801 ;;;         | '(' S? '#PCDATA' S? ')'
 802
 803
 804 (defun ebnf-dtd-mixed ()
 805   (let* ((alt             (cons (ebnf-make-terminal ebnf-dtd-lex) nil))
 806          (token           (ebnf-dtd-lex))
 807          (has-alternative (eq token 'alternative)))
 808     (while (eq token 'alternative)
 809       (or (eq (ebnf-dtd-lex) 'name)
 810           (error "Invalid name"))
 811       (setq alt   (cons ebnf-dtd-lex alt)
 812             token (ebnf-dtd-lex)))
 813     (or (eq token 'end-group)
 814         (error "Missing `)'"))
 815     (and has-alternative
 816          (or (eq (ebnf-dtd-lex) 'zero-or-more)
 817              (error "Missing `*'")))
 818     (ebnf-token-alternative alt (cons (ebnf-dtd-lex) nil))))
 819
 820
 821 ;;; children ::= (choice | seq) ('?' | '*' | '+')?
 822
 823
 824 (defun ebnf-dtd-children (token)
 825   (ebnf-dtd-operators (ebnf-dtd-choice-seq token)))
 826
 827
 828 ;;; choice ::= '(' S? cp ( S? '|' S? cp )+ S? ')'
 829 ;;;
 830 ;;; seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
 831
 832
 833 (defun ebnf-dtd-choice-seq (token)
 834   (setq token (ebnf-dtd-cp token))
 835   (let (elist)
 836     (cond
 837      ;; choice
 838      ((eq (car token) 'alternative)
 839       (while (eq (car token) 'alternative)
 840         (setq elist (cons (cdr token) elist)
 841               token (ebnf-dtd-cp (ebnf-dtd-lex))))
 842       (setq elist (ebnf-token-alternative elist token)))
 843      ;; seq
 844      ((eq (car token) 'comma)
 845       (while (eq (car token) 'comma)
 846         (setq elist (cons (cdr token) elist)
 847               token (ebnf-dtd-cp (ebnf-dtd-lex))))
 848       (setq elist (ebnf-token-sequence (cons (cdr token) elist))))
 849      ;; only one element
 850      (t
 851       (setq elist (cdr token))))
 852     (or (eq (car token) 'end-group)
 853         (error "Missing `)' in ELEMENT content"))
 854     elist))
 855
 856
 857 ;;; cp ::= (Name | choice | seq) ('?' | '*' | '+')?
 858
 859
 860 (defun ebnf-dtd-cp (token)
 861   (ebnf-dtd-operators (cond ((eq token 'name)
 862                              (ebnf-make-terminal ebnf-dtd-lex))
 863                             ((eq token 'begin-group)
 864                              (ebnf-dtd-choice-seq (ebnf-dtd-lex)))
 865                             (t
 866                              (error "Invalid element"))
 867                             )))
 868
 869
 870 ;;; elm ('?' | '*' | '+')?
 871
 872
 873 (defun ebnf-dtd-operators (elm)
 874   (let ((token (ebnf-dtd-lex)))
 875     (cond ((eq token 'optional)         ; ? -  optional
 876            (cons (ebnf-dtd-lex) (ebnf-token-optional elm)))
 877           ((eq token 'zero-or-more)     ; * - zero or more
 878            (cons (ebnf-dtd-lex) (ebnf-make-zero-or-more elm)))
 879           ((eq token 'one-or-more)      ; + - one or more
 880            (cons (ebnf-dtd-lex) (ebnf-make-one-or-more elm)))
 881           (t                            ; only element
 882            (cons token elm))
 883           )))
 884
 885
 886 ;;; AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
 887 ;;;
 888 ;;; AttDef ::= S Name S AttType S DefaultDecl
 889 ;;;
 890 ;;; AttType ::= StringType | TokenizedType | EnumeratedType
 891 ;;;
 892 ;;; StringType ::= 'CDATA'
 893 ;;;
 894 ;;; TokenizedType ::= 'ID'
 895 ;;;                 | 'IDREF'
 896 ;;;                 | 'IDREFS'
 897 ;;;                 | 'ENTITY'
 898 ;;;                 | 'ENTITIES'
 899 ;;;                 | 'NMTOKEN'
 900 ;;;                 | 'NMTOKENS'
 901 ;;;
 902 ;;; EnumeratedType ::= NotationType | Enumeration
 903 ;;;
 904 ;;; NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 905 ;;;
 906 ;;; Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'
 907 ;;;
 908 ;;; DefaultDecl ::= '#REQUIRED'
 909 ;;;               | '#IMPLIED'
 910 ;;;               | (('#FIXED' S)? AttValue)
 911 ;;;
 912 ;;;
 913 ;;; AttValue ::= '"' ([^<&"] | Reference)* '"'
 914 ;;;            | "'" ([^<&'] | Reference)* "'"
 915 ;;;
 916 ;;; Reference ::= EntityRef | CharRef
 917 ;;;
 918 ;;; EntityRef ::= '&' Name ';'
 919 ;;;
 920 ;;; CharRef ::= '&#' [0-9]+ ';'
 921 ;;;           | '&#x' [0-9a-fA-F]+ ';'
 922
 923 ;;; "^\\(&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^<&]\\)*$"
 924
 925
 926 (defun ebnf-dtd-attlistdecl ()
 927   (or (eq (ebnf-dtd-lex) 'name)
 928       (error "Invalid ATTLIST name"))
 929   (let (token)
 930     (while (eq (setq token (ebnf-dtd-lex)) 'name)
 931       ;; type
 932       (setq token (ebnf-dtd-lex))
 933       (cond
 934        ((eq token 'notation)
 935         (or (eq (ebnf-dtd-lex) 'begin-group)
 936             (error "Missing `(' in NOTATION type in ATTLIST declaration"))
 937         (ebnf-dtd-namelist "NOTATION" '(name)))
 938        ((eq token 'begin-group)
 939         (ebnf-dtd-namelist "enumeration" '(name name-char)))
 940        ((memq token
 941               '(cdata id idref idrefs entity entities nmtoken nmtokens)))
 942        (t
 943         (error "Invalid type in ATTLIST declaration")))
 944       ;; default value
 945       (setq token (ebnf-dtd-lex))
 946       (unless (memq token '(required implied))
 947         (and (eq token 'fixed)
 948              (setq token (ebnf-dtd-lex)))
 949         (or (and (eq token 'string)
 950                  (string-match
 951                   "^\\(&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^<&]\\)*$"
 952                   ebnf-dtd-lex))
 953             (error "Invalid default value in ATTLIST declaration"))))
 954     (or (eq token 'end-decl)
 955         (error "Missing `>' in end of ATTLIST"))
 956     (cons (ebnf-dtd-lex) nil)))
 957
 958
 959 (defun ebnf-dtd-namelist (type name-list)
 960   (let (token)
 961     (while (progn
 962              (or (memq (ebnf-dtd-lex) name-list)
 963                  (error "Invalid name in %s type in ATTLIST declaration" type))
 964              (eq (setq token (ebnf-dtd-lex)) 'alternative)))
 965     (or (eq token 'end-group)
 966         (error "Missing `)' in %s type in ATTLIST declaration" type))))
 967
 968
 969 ;;; EntityDecl ::= GEDecl | PEDecl
 970 ;;;
 971 ;;; GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
 972 ;;;
 973 ;;; PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 974 ;;;
 975 ;;; EntityDef ::= EntityValue | (ExternalID NDataDecl?)
 976 ;;;
 977 ;;; PEDef ::= EntityValue | ExternalID
 978 ;;;
 979 ;;; NDataDecl ::= S 'NDATA' S Name
 980 ;;;
 981 ;;;
 982 ;;; EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 983 ;;;               | "'" ([^%&'] | PEReference | Reference)* "'"
 984 ;;;
 985 ;;; PEReference ::= '%' Name ';'
 986 ;;;
 987 ;;; Reference ::= EntityRef | CharRef
 988 ;;;
 989 ;;; EntityRef ::= '&' Name ';'
 990 ;;;
 991 ;;; CharRef ::= '&#' [0-9]+ ';'
 992 ;;;           | '&#x' [0-9a-fA-F]+ ';'
 993
 994 ;;; "^\\(%[A-Za-z_:][-A-Za-z0-9._:]*;\\|&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^%&]\\)*$"
 995
 996
 997 (defun ebnf-dtd-entitydecl ()
 998   (let* ((token (ebnf-dtd-lex))
 999          (pedecl (eq token 'percent)))
1000     (and pedecl
1001          (setq token (ebnf-dtd-lex)))
1002     (or (eq token 'name)
1003         (error "Invalid name of ENTITY"))
1004     (setq token (ebnf-dtd-lex))
1005     (if (eq token 'string)
1006         (if (string-match
1007              "^\\(%[A-Za-z_:][-A-Za-z0-9._:]*;\\|&\\([A-Za-z_:][-A-Za-z0-9._:]*\\|#\\(x[0-9a-fA-F]+\\|[0-9]+\\)\\);\\|[^%&]\\)*$"
1008              ebnf-dtd-lex)
1009             (setq token (ebnf-dtd-lex))
1010           (error "Invalid ENTITY definition"))
1011       (setq token (ebnf-dtd-externalid token))
1012       (when (and (not pedecl) (eq token 'ndata))
1013         (or (eq (ebnf-dtd-lex) 'name)
1014             (error "Invalid NDATA name"))
1015         (setq token (ebnf-dtd-lex))))
1016     (or (eq token 'end-decl)
1017         (error "Missing `>' in end of ENTITY"))
1018     (cons (ebnf-dtd-lex) nil)))
1019
1020
1021 ;;; NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>'
1022 ;;;
1023 ;;; PublicID ::= 'PUBLIC' S PubidLiteral
1024
1025
1026 (defun ebnf-dtd-notationdecl ()
1027   (or (eq (ebnf-dtd-lex) 'name)
1028       (error "Invalid name NOTATION"))
1029   (or (eq (ebnf-dtd-externalid-or-publicid) 'end-decl)
1030       (error "Missing `>' in end of NOTATION"))
1031   (cons (ebnf-dtd-lex) nil))
1032
1033
1034 ;;; ExternalID ::= 'SYSTEM' S SystemLiteral
1035 ;;;              | 'PUBLIC' S PubidLiteral S SystemLiteral
1036 ;;;
1037 ;;; PublicID ::= 'PUBLIC' S PubidLiteral
1038
1039
1040 (defun ebnf-dtd-externalid-or-publicid ()
1041   (let ((token (ebnf-dtd-lex)))
1042     (cond ((eq token 'system)
1043            (ebnf-dtd-systemliteral))
1044           ((eq token 'public)
1045            (ebnf-dtd-pubidliteral)
1046            (and (eq (setq token (ebnf-dtd-lex)) 'string)
1047                 (setq token (ebnf-dtd-lex)))
1048            token)
1049           (t
1050            (error "Missing `SYSTEM' or `PUBLIC'")))))
1051
1052 \f
1053 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1054 ;; Lexical analyzer
1055
1056
1057 (defconst ebnf-dtd-token-table (make-vector 256 'error)
1058   "Vector used to map characters to a lexical token.")
1059
1060
1061 (defun ebnf-dtd-initialize ()
1062   "Initialize EBNF token table."
1063   ;; control character & control 8-bit character are set to `error'
1064   (let ((char ?\060))
1065     ;; digits: 0-9
1066     (while (< char ?\072)
1067       (aset ebnf-dtd-token-table char 'name-char)
1068       (setq char (1+ char)))
1069     ;; printable character: A-Z
1070     (setq char ?\101)
1071     (while (< char ?\133)
1072       (aset ebnf-dtd-token-table char 'name)
1073       (setq char (1+ char)))
1074     ;; printable character: a-z
1075     (setq char ?\141)
1076     (while (< char ?\173)
1077       (aset ebnf-dtd-token-table char 'name)
1078       (setq char (1+ char)))
1079     ;; European 8-bit accentuated characters:
1080     (setq char ?\240)
1081     (while (< char ?\400)
1082       (aset ebnf-dtd-token-table char 'name)
1083       (setq char (1+ char)))
1084     ;; Override name characters:
1085     (aset ebnf-dtd-token-table ?_ 'name)
1086     (aset ebnf-dtd-token-table ?: 'name)
1087     (aset ebnf-dtd-token-table ?. 'name-char)
1088     (aset ebnf-dtd-token-table ?- 'name-char)
1089     ;; Override space characters:
1090     (aset ebnf-dtd-token-table ?\n 'space) ; [NL] linefeed
1091     (aset ebnf-dtd-token-table ?\r 'space) ; [CR] carriage return
1092     (aset ebnf-dtd-token-table ?\t 'space) ; [HT] horizontal tab
1093     (aset ebnf-dtd-token-table ?\  'space) ; [SP] space
1094     ;; Override other lexical characters:
1095     (aset ebnf-dtd-token-table ?=  'equal)
1096     (aset ebnf-dtd-token-table ?,  'comma)
1097     (aset ebnf-dtd-token-table ?*  'zero-or-more)
1098     (aset ebnf-dtd-token-table ?+  'one-or-more)
1099     (aset ebnf-dtd-token-table ?|  'alternative)
1100     (aset ebnf-dtd-token-table ?%  'percent)
1101     (aset ebnf-dtd-token-table ?&  'ampersand)
1102     (aset ebnf-dtd-token-table ?#  'hash)
1103     (aset ebnf-dtd-token-table ?\? 'interrogation)
1104     (aset ebnf-dtd-token-table ?\" 'double-quote)
1105     (aset ebnf-dtd-token-table ?\' 'single-quote)
1106     (aset ebnf-dtd-token-table ?<  'less-than)
1107     (aset ebnf-dtd-token-table ?>  'end-decl)
1108     (aset ebnf-dtd-token-table ?\( 'begin-group)
1109     (aset ebnf-dtd-token-table ?\) 'end-group)
1110     (aset ebnf-dtd-token-table ?\[ 'begin-subset)
1111     (aset ebnf-dtd-token-table ?\] 'end-subset)))
1112
1113
1114 ;; replace the range "\240-\377" (see `ebnf-range-regexp').
1115 (defconst ebnf-dtd-name-chars
1116   (ebnf-range-regexp "-._:0-9A-Za-z" ?\240 ?\377))
1117
1118
1119 (defconst ebnf-dtd-decl-alist
1120   '(("ATTLIST"  . attlist-decl)
1121     ("DOCTYPE"  . doctype-decl)
1122     ("ELEMENT"  . element-decl)
1123     ("ENTITY"   . entity-decl)
1124     ("NOTATION" . notation-decl)))
1125
1126
1127 (defconst ebnf-dtd-element-alist
1128   '(("#FIXED"    . fixed)
1129     ("#IMPLIED"  . implied)
1130     ("#PCDATA"   . pcdata)
1131     ("#REQUIRED" . required)))
1132
1133
1134 (defconst ebnf-dtd-name-alist
1135   '(("ANY"        . any)
1136     ("CDATA"      . cdata)
1137     ("EMPTY"      . empty)
1138     ("ENTITIES"   . entities)
1139     ("ENTITY"     . entity)
1140     ("ID"         . id)
1141     ("IDREF"      . idref)
1142     ("IDREFS"     . idrefs)
1143     ("NDATA"      . ndata)
1144     ("NMTOKEN"    . nmtoken)
1145     ("NMTOKENS"   . nmtokens)
1146     ("NOTATION"   . notation)
1147     ("PUBLIC"     . public)
1148     ("SYSTEM"     . system)
1149     ("encoding"   . encoding-attr)
1150     ("standalone" . standalone-attr)
1151     ("version"    . version-attr)))
1152
1153
1154 (defun ebnf-dtd-lex ()
1155   "Lexical analyzer for DTD.
1156
1157 Return a lexical token.
1158
1159 See documentation for variable `ebnf-dtd-lex'."
1160   (if (>= (point) ebnf-limit)
1161       'end-of-input
1162     (let (token)
1163       ;; skip spaces and comments
1164       (while (if (> (following-char) 255)
1165                  (progn
1166                    (setq token 'error)
1167                    nil)
1168                (setq token (aref ebnf-dtd-token-table (following-char)))
1169                (cond
1170                 ((eq token 'space)
1171                  (skip-chars-forward " \n\r\t" ebnf-limit)
1172                  (< (point) ebnf-limit))
1173                 ((and (eq token 'less-than)
1174                       (looking-at "<!--"))
1175                  (ebnf-dtd-skip-comment))
1176                 (t nil)
1177                 )))
1178       (cond
1179        ;; end of input
1180        ((>= (point) ebnf-limit)
1181         'end-of-input)
1182        ;; error
1183        ((eq token 'error)
1184         (error "Invalid character"))
1185        ;; beginning of declaration:
1186        ;; <?name, <!ATTLIST, <!DOCTYPE, <!ELEMENT, <!ENTITY, <!NOTATION
1187        ((eq token 'less-than)
1188         (forward-char)
1189         (let ((char (following-char)))
1190           (cond ((= char ?\?)           ; <?
1191                  (forward-char)
1192                  (setq ebnf-dtd-lex (ebnf-buffer-substring ebnf-dtd-name-chars))
1193                  'begin-pi)
1194                 ((= char ?!)            ; <!
1195                  (forward-char)
1196                  (let ((decl (ebnf-buffer-substring ebnf-dtd-name-chars)))
1197                    (or (cdr (assoc decl ebnf-dtd-decl-alist))
1198                        (error "Invalid declaration name `%s'" decl))))
1199                 (t                      ; <x
1200                  (error "Invalid declaration `<%c'" char)))))
1201        ;; name, namechar
1202        ((memq  token '(name name-char))
1203         (setq ebnf-dtd-lex (ebnf-buffer-substring ebnf-dtd-name-chars))
1204         (or (cdr (assoc ebnf-dtd-lex ebnf-dtd-name-alist))
1205             token))
1206        ;; ?, ?>
1207        ((eq token 'interrogation)
1208         (forward-char)
1209         (if (/= (following-char) ?>)
1210             'optional
1211           (forward-char)
1212           'end-pi))
1213        ;; #FIXED, #IMPLIED, #PCDATA, #REQUIRED
1214        ((eq token 'hash)
1215         (forward-char)
1216         (setq ebnf-dtd-lex
1217               (concat "#" (ebnf-buffer-substring ebnf-dtd-name-chars)))
1218         (or (cdr (assoc ebnf-dtd-lex ebnf-dtd-element-alist))
1219             (error "Invalid element `%s'" ebnf-dtd-lex)))
1220        ;; "string"
1221        ((eq token 'double-quote)
1222         (setq ebnf-dtd-lex (ebnf-dtd-string ?\"))
1223         'string)
1224        ;; 'string'
1225        ((eq token 'single-quote)
1226         (setq ebnf-dtd-lex (ebnf-dtd-string ?\'))
1227         'string)
1228        ;; %, %name;
1229        ((eq token 'percent)
1230         (forward-char)
1231         (if (looking-at "[ \n\r\t]")
1232             'percent
1233           (setq ebnf-dtd-lex (ebnf-dtd-name-ref "%"))
1234           'pe-ref))
1235        ;; &#...;, &#x...;, &name;
1236        ((eq token 'ampersand)
1237         (forward-char)
1238         (if (/= (following-char) ?#)
1239             (progn
1240               ;; &name;
1241               (setq ebnf-dtd-lex (ebnf-dtd-name-ref "&"))
1242               'entity-ref)
1243           ;; &#...;, &#x...;
1244           (forward-char)
1245           (setq ebnf-dtd-lex (if (/= (following-char) ?x)
1246                                  (ebnf-dtd-char-ref "&#" "0-9")
1247                                (forward-char)
1248                                (ebnf-dtd-char-ref "&#x" "0-9a-fA-F")))
1249           'char-ref))
1250        ;; miscellaneous: (, ), [, ], =, |, *, +, >, `,'
1251        (t
1252         (forward-char)
1253         token)
1254        ))))
1255
1256
1257 (defun ebnf-dtd-name-ref (start)
1258   (ebnf-dtd-char-ref start ebnf-dtd-name-chars))
1259
1260
1261 (defun ebnf-dtd-char-ref (start chars)
1262   (let ((char (ebnf-buffer-substring chars)))
1263     (or (= (following-char) ?\;)
1264         (error "Invalid element `%s%s%c'" start char (following-char)))
1265     (forward-char)
1266     (format "%s%s;" start char)))
1267
1268
1269 ;; replace the range "\240-\377" (see `ebnf-range-regexp').
1270 (defconst ebnf-dtd-double-string-chars
1271   (ebnf-range-regexp "\t -!#-~" ?\240 ?\377))
1272 (defconst ebnf-dtd-single-string-chars
1273   (ebnf-range-regexp "\t -&(-~" ?\240 ?\377))
1274
1275
1276 (defun ebnf-dtd-string (delim)
1277   (buffer-substring-no-properties
1278    (progn
1279      (forward-char)
1280      (point))
1281    (progn
1282      (skip-chars-forward (if (= delim ?\")
1283                              ebnf-dtd-double-string-chars
1284                            ebnf-dtd-single-string-chars)
1285                          ebnf-limit)
1286      (or (= (following-char) delim)
1287          (error "Missing string delimiter `%c'" delim))
1288      (prog1
1289          (point)
1290        (forward-char)))))
1291
1292
1293 ;; replace the range "\177-\237" (see `ebnf-range-regexp').
1294 (defconst ebnf-dtd-comment-chars
1295   (ebnf-range-regexp "^-\000-\010\013\014\016-\037" ?\177 ?\237))
1296 (defconst ebnf-dtd-filename-chars
1297   (ebnf-range-regexp "^-\000-\037" ?\177 ?\237))
1298
1299
1300 (defun ebnf-dtd-skip-comment ()
1301   (forward-char 4)                      ; <!--
1302   (cond
1303    ;; open EPS file
1304    ((and ebnf-eps-executing (= (following-char) ?\[))
1305     (ebnf-eps-add-context (ebnf-dtd-eps-filename)))
1306    ;; close EPS file
1307    ((and ebnf-eps-executing (= (following-char) ?\]))
1308     (ebnf-eps-remove-context (ebnf-dtd-eps-filename)))
1309    ;; EPS header
1310    ((and ebnf-eps-executing (= (following-char) ?H))
1311     (ebnf-eps-header-comment (ebnf-dtd-eps-filename)))
1312    ;; EPS footer
1313    ((and ebnf-eps-executing (= (following-char) ?F))
1314     (ebnf-eps-footer-comment (ebnf-dtd-eps-filename)))
1315    ;; any other action in comment
1316    (t
1317     (setq ebnf-action (aref ebnf-comment-table (following-char))))
1318    )
1319   (while (progn
1320            (skip-chars-forward ebnf-dtd-comment-chars ebnf-limit)
1321            (and (< (point) ebnf-limit)
1322                 (not (looking-at "-->"))))
1323     (skip-chars-forward "-" ebnf-limit))
1324   ;; check for a valid end of comment
1325   (cond ((>= (point) ebnf-limit)
1326          nil)
1327         ((looking-at "-->")
1328          (forward-char 3)
1329          t)
1330         (t
1331          (error "Invalid character"))
1332         ))
1333
1334
1335 (defun ebnf-dtd-eps-filename ()
1336   (forward-char)
1337   (let (fname)
1338     (while (progn
1339              (setq fname
1340                    (concat fname
1341                            (ebnf-buffer-substring ebnf-dtd-filename-chars)))
1342              (and (< (point) ebnf-limit)
1343                   (= (following-char) ?-)       ; may be \n, \t, \r
1344                   (not (looking-at "-->"))))
1345       (setq fname (concat fname (ebnf-buffer-substring "-"))))
1346     fname))
1347
1348 \f
1349 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1350
1351
1352 (provide 'ebnf-dtd)
1353
1354 ;;; arch-tag: c21bb640-135f-4afa-8712-fa11d86301c4
1355 ;;; ebnf-dtd.el ends here