lisp/xml.el

   1 ;;; xml.el --- XML parser
   2
   3 ;; Copyright (C) 2000, 2001, 2002, 2003, 2004,
   4 ;;   2005 Free Software Foundation, Inc.
   5
   6 ;; Author: Emmanuel Briot  <briot@gnat.com>
   7 ;; Maintainer: Mark A. Hershberger <mah@everybody.org>
   8 ;; Keywords: xml, data
   9
  10 ;; This file is part of GNU Emacs.
  11
  12 ;; GNU Emacs is free software; you can redistribute it and/or modify
  13 ;; it under the terms of the GNU General Public License as published by
  14 ;; the Free Software Foundation; either version 2, or (at your option)
  15 ;; any later version.
  16
  17 ;; GNU Emacs is distributed in the hope that it will be useful,
  18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20 ;; GNU General Public License for more details.
  21
  22 ;; You should have received a copy of the GNU General Public License
  23 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  24 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  25 ;; Boston, MA 02110-1301, USA.
  26
  27 ;;; Commentary:
  28
  29 ;; This file contains a somewhat incomplete non-validating XML parser.  It
  30 ;; parses a file, and returns a list that can be used internally by
  31 ;; any other Lisp libraries.
  32
  33 ;;; FILE FORMAT
  34
  35 ;; The document type declaration may either be ignored or (optionally)
  36 ;; parsed, but currently the parsing will only accept element
  37 ;; declarations.  The XML file is assumed to be well-formed.  In case
  38 ;; of error, the parsing stops and the XML file is shown where the
  39 ;; parsing stopped.
  40 ;;
  41 ;; It also knows how to ignore comments and processing instructions.
  42 ;;
  43 ;; The XML file should have the following format:
  44 ;;    <node1 attr1="name1" attr2="name2" ...>value
  45 ;;       <node2 attr3="name3" attr4="name4">value2</node2>
  46 ;;       <node3 attr5="name5" attr6="name6">value3</node3>
  47 ;;    </node1>
  48 ;; Of course, the name of the nodes and attributes can be anything.  There can
  49 ;; be any number of attributes (or none), as well as any number of children
  50 ;; below the nodes.
  51 ;;
  52 ;; There can be only top level node, but with any number of children below.
  53
  54 ;;; LIST FORMAT
  55
  56 ;; The functions `xml-parse-file', `xml-parse-region' and
  57 ;; `xml-parse-tag' return a list with the following format:
  58 ;;
  59 ;;    xml-list   ::= (node node ...)
  60 ;;    node       ::= (qname attribute-list . child_node_list)
  61 ;;    child_node_list ::= child_node child_node ...
  62 ;;    child_node ::= node | string
  63 ;;    qname      ::= (:namespace-uri . "name") | "name"
  64 ;;    attribute_list ::= ((qname . "value") (qname . "value") ...)
  65 ;;                       | nil
  66 ;;    string     ::= "..."
  67 ;;
  68 ;; Some macros are provided to ease the parsing of this list.
  69 ;; Whitespace is preserved.  Fixme: There should be a tree-walker that
  70 ;; can remove it.
  71
  72 ;; TODO:
  73 ;;  * xml:base, xml:space support
  74 ;;  * more complete DOCTYPE parsing
  75 ;;  * pi support
  76
  77 ;;; Code:
  78
  79 ;; Note that {buffer-substring,match-string}-no-properties were
  80 ;; formerly used in several places, but that removes composition info.
  81
  82 ;;*******************************************************************
  83 ;;**
  84 ;;**  Macros to parse the list
  85 ;;**
  86 ;;*******************************************************************
  87
  88 (defconst xml-undefined-entity "?"
  89   "What to substitute for undefined entities")
  90
  91 (defvar xml-entity-alist
  92   '(("lt"   . "<")
  93     ("gt"   . ">")
  94     ("apos" . "'")
  95     ("quot" . "\"")
  96     ("amp"  . "&"))
  97   "The defined entities.  Entities are added to this when the DTD is parsed.")
  98
  99 (defvar xml-sub-parser nil
 100   "Dynamically set this to a non-nil value if you want to parse an XML fragment.")
 101
 102 (defvar xml-validating-parser nil
 103   "Set to non-nil to get validity checking.")
 104
 105 (defsubst xml-node-name (node)
 106   "Return the tag associated with NODE.
 107 Without namespace-aware parsing, the tag is a symbol.
 108
 109 With namespace-aware parsing, the tag is a cons of a string
 110 representing the uri of the namespace with the local name of the
 111 tag.  For example,
 112
 113     <foo>
 114
 115 would be represented by
 116
 117     '(\"\" . \"foo\")."
 118
 119   (car node))
 120
 121 (defsubst xml-node-attributes (node)
 122   "Return the list of attributes of NODE.
 123 The list can be nil."
 124   (nth 1 node))
 125
 126 (defsubst xml-node-children (node)
 127   "Return the list of children of NODE.
 128 This is a list of nodes, and it can be nil."
 129   (cddr node))
 130
 131 (defun xml-get-children (node child-name)
 132   "Return the children of NODE whose tag is CHILD-NAME.
 133 CHILD-NAME should match the value returned by `xml-node-name'."
 134   (let ((match ()))
 135     (dolist (child (xml-node-children node))
 136       (if (and (listp child)
 137                (equal (xml-node-name child) child-name))
 138           (push child match)))
 139     (nreverse match)))
 140
 141 (defun xml-get-attribute-or-nil (node attribute)
 142   "Get from NODE the value of ATTRIBUTE.
 143 Return nil if the attribute was not found.
 144
 145 See also `xml-get-attribute'."
 146   (cdr (assoc attribute (xml-node-attributes node))))
 147
 148 (defsubst xml-get-attribute (node attribute)
 149   "Get from NODE the value of ATTRIBUTE.
 150 An empty string is returned if the attribute was not found.
 151
 152 See also `xml-get-attribute-or-nil'."
 153   (or (xml-get-attribute-or-nil node attribute) ""))
 154
 155 ;;*******************************************************************
 156 ;;**
 157 ;;**  Creating the list
 158 ;;**
 159 ;;*******************************************************************
 160
 161 ;;;###autoload
 162 (defun xml-parse-file (file &optional parse-dtd parse-ns)
 163   "Parse the well-formed XML file FILE.
 164 If FILE is already visited, use its buffer and don't kill it.
 165 Returns the top node with all its children.
 166 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped.
 167 If PARSE-NS is non-nil, then QNAMES are expanded."
 168   (let ((keep))
 169     (if (get-file-buffer file)
 170         (progn
 171           (set-buffer (get-file-buffer file))
 172           (setq keep (point)))
 173       (let (auto-mode-alist)            ; no need for xml-mode
 174         (find-file file)))
 175
 176     (let ((xml (xml-parse-region (point-min)
 177                                  (point-max)
 178                                  (current-buffer)
 179                                  parse-dtd parse-ns)))
 180       (if keep
 181           (goto-char keep)
 182         (kill-buffer (current-buffer)))
 183       xml)))
 184
 185
 186 (defvar xml-name-re)
 187 (defvar xml-entity-value-re)
 188 (defvar xml-att-def-re)
 189 (let* ((start-chars (concat "[:alpha:]:_"))
 190        (name-chars  (concat "-[:digit:]." start-chars))
 191 ;;[3]           S          ::=          (#x20 | #x9 | #xD | #xA)+
 192        (whitespace  "[ \t\n\r]"))
 193 ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6]
 194 ;;                      | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF]
 195 ;;                      | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF]
 196 ;;                      | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
 197   (defvar xml-name-start-char-re (concat "[" start-chars "]"))
 198 ;;[4a] NameChar ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
 199   (defvar xml-name-char-re       (concat "[" name-chars  "]"))
 200 ;;[5] Name     ::= NameStartChar (NameChar)*
 201   (defvar xml-name-re            (concat xml-name-start-char-re xml-name-char-re "*"))
 202 ;;[6] Names    ::= Name (#x20 Name)*
 203   (defvar xml-names-re           (concat xml-name-re "\\(?: " xml-name-re "\\)*"))
 204 ;;[7] Nmtoken ::= (NameChar)+
 205   (defvar xml-nmtoken-re         (concat xml-name-char-re "+"))
 206 ;;[8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
 207   (defvar xml-nmtokens-re        (concat xml-nmtoken-re "\\(?: " xml-name-re "\\)*"))
 208 ;;[66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
 209   (defvar xml-char-ref-re        "\\(?:&#[0-9]+;\\|&#x[0-9a-fA-F]+;\\)")
 210 ;;[68] EntityRef   ::= '&' Name ';'
 211   (defvar xml-entity-ref         (concat "&" xml-name-re ";"))
 212 ;;[69] PEReference ::= '%' Name ';'
 213   (defvar xml-pe-reference-re    (concat "%" xml-name-re ";"))
 214 ;;[67] Reference   ::= EntityRef | CharRef
 215   (defvar xml-reference-re       (concat "\\(?:" xml-entity-ref "\\|" xml-char-ref-re "\\)"))
 216 ;;[10]          AttValue           ::=          '"' ([^<&"] | Reference)* '"' |  "'" ([^<&'] | Reference)* "'"
 217   (defvar xml-att-value-re    (concat "\\(?:\"\\(?:[^&\"]\\|" xml-reference-re "\\)*\"\\|"
 218                                       "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)"))
 219 ;;[56]          TokenizedType      ::=          'ID'       [VC: ID] [VC: One ID per Element Type] [VC: ID Attribute Default]
 220 ;;                                            | 'IDREF'    [VC: IDREF]
 221 ;;                                            | 'IDREFS'   [VC: IDREF]
 222 ;;                                            | 'ENTITY'   [VC: Entity Name]
 223 ;;                                            | 'ENTITIES' [VC: Entity Name]
 224 ;;                                            | 'NMTOKEN'  [VC: Name Token]
 225 ;;                                            | 'NMTOKENS' [VC: Name Token]
 226   (defvar xml-tokenized-type-re "\\(?:ID\\|IDREF\\|IDREFS\\|ENTITY\\|ENTITIES\\|NMTOKEN\\|NMTOKENS\\)")
 227 ;;[58]          NotationType       ::=          'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 228   (defvar xml-notation-type-re (concat "\\(?:NOTATION" whitespace "(" whitespace "*" xml-name-re
 229                                        "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)"))
 230 ;;[59]          Enumeration        ::=          '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'      [VC: Enumeration] [VC: No Duplicate Tokens]
 231   (defvar xml-enumeration-re (concat "\\(?:(" whitespace "*" xml-nmtoken-re
 232                                      "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*"
 233                                      whitespace ")\\)"))
 234 ;;[57]          EnumeratedType     ::=          NotationType | Enumeration
 235   (defvar xml-enumerated-type-re (concat "\\(?:" xml-notation-type-re "\\|" xml-enumeration-re "\\)"))
 236 ;;[54]          AttType    ::=          StringType | TokenizedType | EnumeratedType
 237 ;;[55]          StringType         ::=          'CDATA'
 238   (defvar xml-att-type-re (concat "\\(?:CDATA\\|" xml-tokenized-type-re "\\|" xml-notation-type-re"\\|" xml-enumerated-type-re "\\)"))
 239 ;;[60]          DefaultDecl        ::=          '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
 240   (defvar xml-default-decl-re (concat "\\(?:#REQUIRED\\|#IMPLIED\\|\\(?:#FIXED" whitespace "\\)*" xml-att-value-re "\\)"))
 241 ;;[53]          AttDef     ::=          S Name S AttType S DefaultDecl
 242   (defvar xml-att-def-re         (concat "\\(?:" whitespace "*" xml-name-re
 243                                          whitespace "*" xml-att-type-re
 244                                          whitespace "*" xml-default-decl-re "\\)"))
 245 ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 246 ;;                 |  "'" ([^%&'] | PEReference | Reference)* "'"
 247   (defvar xml-entity-value-re    (concat "\\(?:\"\\(?:[^%&\"]\\|" xml-pe-reference-re
 248                                          "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|"
 249                                          xml-pe-reference-re "\\|" xml-reference-re "\\)*'\\)")))
 250 ;;[75] ExternalID ::= 'SYSTEM' S SystemLiteral
 251 ;;                 | 'PUBLIC' S PubidLiteral S SystemLiteral
 252 ;;[76] NDataDecl ::=    S 'NDATA' S
 253 ;;[73] EntityDef  ::= EntityValue| (ExternalID NDataDecl?)
 254 ;;[71] GEDecl     ::= '<!ENTITY' S Name S EntityDef S? '>'
 255 ;;[74] PEDef      ::= EntityValue | ExternalID
 256 ;;[72] PEDecl     ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 257 ;;[70] EntityDecl ::= GEDecl | PEDecl
 258
 259 ;; Note that this is setup so that we can do whitespace-skipping with
 260 ;; `(skip-syntax-forward " ")', inter alia.  Previously this was slow
 261 ;; compared with `re-search-forward', but that has been fixed.  Also
 262 ;; note that the standard syntax table contains other characters with
 263 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
 264 ;; which we might skip whitespace -- specifically, they're not
 265 ;; NameChars [XML 4].
 266
 267 (defvar xml-syntax-table
 268   (let ((table (make-syntax-table)))
 269     ;; Get space syntax correct per XML [3].
 270     (dotimes (c 31)
 271       (modify-syntax-entry c "." table)) ; all are space in standard table
 272     (dolist (c '(?\t ?\n ?\r))          ; these should be space
 273       (modify-syntax-entry c " " table))
 274     ;; For skipping attributes.
 275     (modify-syntax-entry ?\" "\"" table)
 276     (modify-syntax-entry ?' "\"" table)
 277     ;; Non-alnum name chars should be symbol constituents (`-' and `_'
 278     ;; are OK by default).
 279     (modify-syntax-entry ?. "_" table)
 280     (modify-syntax-entry ?: "_" table)
 281     ;; XML [89]
 282     (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005
 283                  #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC
 284                  #x30FD #x30FE))
 285       (modify-syntax-entry (decode-char 'ucs c) "w" table))
 286     ;; Fixme: rest of [4]
 287     table)
 288   "Syntax table used by `xml-parse-region'.")
 289
 290 ;; XML [5]
 291 ;; Note that [:alpha:] matches all multibyte chars with word syntax.
 292 (eval-and-compile
 293   (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*"))
 294
 295 ;; Fixme:  This needs re-writing to deal with the XML grammar properly, i.e.
 296 ;;   document    ::=    prolog element Misc*
 297 ;;   prolog    ::=    XMLDecl? Misc* (doctypedecl Misc*)?
 298
 299 ;;;###autoload
 300 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns)
 301   "Parse the region from BEG to END in BUFFER.
 302 If BUFFER is nil, it defaults to the current buffer.
 303 Returns the XML list for the region, or raises an error if the region
 304 is not well-formed XML.
 305 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped,
 306 and returned as the first element of the list.
 307 If PARSE-NS is non-nil, then QNAMES are expanded."
 308   (save-restriction
 309     (narrow-to-region beg end)
 310     ;; Use fixed syntax table to ensure regexp char classes and syntax
 311     ;; specs DTRT.
 312     (with-syntax-table (standard-syntax-table)
 313       (let ((case-fold-search nil)      ; XML is case-sensitive.
 314             xml result dtd)
 315         (save-excursion
 316           (if buffer
 317               (set-buffer buffer))
 318           (goto-char (point-min))
 319           (while (not (eobp))
 320             (if (search-forward "<" nil t)
 321                 (progn
 322                   (forward-char -1)
 323                   (setq result (xml-parse-tag parse-dtd parse-ns))
 324                   (if (and xml result (not xml-sub-parser))
 325                       ;;  translation of rule [1] of XML specifications
 326                       (error "XML: (Not Well-Formed) Only one root tag allowed")
 327                     (cond
 328                      ((null result))
 329                      ((and (listp (car result))
 330                            parse-dtd)
 331                       (setq dtd (car result))
 332                       (if (cdr result)  ; possible leading comment
 333                           (add-to-list 'xml (cdr result))))
 334                      (t
 335                       (add-to-list 'xml result)))))
 336               (goto-char (point-max))))
 337           (if parse-dtd
 338               (cons dtd (nreverse xml))
 339             (nreverse xml)))))))
 340
 341 (defun xml-maybe-do-ns (name default xml-ns)
 342   "Perform any namespace expansion.
 343 NAME is the name to perform the expansion on.
 344 DEFAULT is the default namespace.  XML-NS is a cons of namespace
 345 names to uris.  When namespace-aware parsing is off, then XML-NS
 346 is nil.
 347
 348 During namespace-aware parsing, any name without a namespace is
 349 put into the namespace identified by DEFAULT.  nil is used to
 350 specify that the name shouldn't be given a namespace."
 351   (if (consp xml-ns)
 352       (let* ((nsp (string-match ":" name))
 353              (lname (if nsp (substring name (match-end 0)) name))
 354              (prefix (if nsp (substring name 0 (match-beginning 0)) default))
 355              (special (and (string-equal lname "xmlns") (not prefix)))
 356              ;; Setting default to nil will insure that there is not
 357              ;; matching cons in xml-ns.  In which case we
 358              (ns (or (cdr (assoc (if special "xmlns" prefix)
 359                                  xml-ns))
 360                      "")))
 361         (cons ns (if special "" lname)))
 362     (intern name)))
 363
 364 (defun xml-parse-fragment (&optional parse-dtd parse-ns)
 365   "Parse xml-like fragments."
 366   (let ((xml-sub-parser t)
 367         children)
 368     (while (not (eobp))
 369       (let ((bit (xml-parse-tag
 370                   parse-dtd parse-ns)))
 371         (if children
 372             (setq children (append (list bit) children))
 373           (if (stringp bit)
 374               (setq children (list bit))
 375             (setq children bit)))))
 376     (reverse children)))
 377
 378 (defun xml-parse-tag (&optional parse-dtd parse-ns)
 379   "Parse the tag at point.
 380 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
 381 returned as the first element in the list.
 382 If PARSE-NS is non-nil, then QNAMES are expanded.
 383 Returns one of:
 384  - a list : the matching node
 385  - nil    : the point is not looking at a tag.
 386  - a pair : the first element is the DTD, the second is the node."
 387   (let ((xml-validating-parser (or parse-dtd xml-validating-parser))
 388         (xml-ns (if (consp parse-ns)
 389                     parse-ns
 390                   (if parse-ns
 391                       (list
 392                        ;; Default for empty prefix is no namespace
 393                        (cons ""      "")
 394                        ;; "xml" namespace
 395                        (cons "xml"   "http://www.w3.org/XML/1998/namespace")
 396                        ;; We need to seed the xmlns namespace
 397                        (cons "xmlns" "http://www.w3.org/2000/xmlns/"))))))
 398     (cond
 399      ;; Processing instructions (like the <?xml version="1.0"?> tag at the
 400      ;; beginning of a document).
 401      ((looking-at "<\\?")
 402       (search-forward "?>")
 403       (skip-syntax-forward " ")
 404       (xml-parse-tag parse-dtd xml-ns))
 405      ;;  Character data (CDATA) sections, in which no tag should be interpreted
 406      ((looking-at "<!\\[CDATA\\[")
 407       (let ((pos (match-end 0)))
 408         (unless (search-forward "]]>" nil t)
 409           (error "XML: (Not Well Formed) CDATA section does not end anywhere in the document"))
 410         (concat
 411          (buffer-substring pos (match-beginning 0))
 412          (xml-parse-string))))
 413      ;;  DTD for the document
 414      ((looking-at "<!DOCTYPE")
 415       (let ((dtd (xml-parse-dtd parse-ns)))
 416         (skip-syntax-forward " ")
 417         (if xml-validating-parser
 418             (cons dtd (xml-parse-tag nil xml-ns))
 419           (xml-parse-tag nil xml-ns))))
 420      ;;  skip comments
 421      ((looking-at "<!--")
 422       (search-forward "-->")
 423       nil)
 424      ;;  end tag
 425      ((looking-at "</")
 426       '())
 427      ;;  opening tag
 428      ((looking-at "<\\([^/>[:space:]]+\\)")
 429       (goto-char (match-end 1))
 430
 431       ;; Parse this node
 432       (let* ((node-name (match-string 1))
 433              ;; Parse the attribute list.
 434              (attrs (xml-parse-attlist xml-ns))
 435              children pos)
 436
 437         ;; add the xmlns:* attrs to our cache
 438         (when (consp xml-ns)
 439           (dolist (attr attrs)
 440             (when (and (consp (car attr))
 441                        (equal "http://www.w3.org/2000/xmlns/"
 442                               (caar attr)))
 443               (push (cons (cdar attr) (cdr attr))
 444                     xml-ns))))
 445
 446         (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns)))
 447
 448         ;; is this an empty element ?
 449         (if (looking-at "/>")
 450             (progn
 451               (forward-char 2)
 452               (nreverse children))
 453
 454           ;; is this a valid start tag ?
 455           (if (eq (char-after) ?>)
 456               (progn
 457                 (forward-char 1)
 458                 ;;  Now check that we have the right end-tag. Note that this
 459                 ;;  one might contain spaces after the tag name
 460                 (let ((end (concat "</" node-name "\\s-*>")))
 461                   (while (not (looking-at end))
 462                     (cond
 463                      ((looking-at "</")
 464                       (error "XML: (Not Well-Formed) Invalid end tag (expecting %s) at pos %d"
 465                              node-name (point)))
 466                      ((= (char-after) ?<)
 467                       (let ((tag (xml-parse-tag nil xml-ns)))
 468                         (when tag
 469                           (push tag children))))
 470                      (t
 471                       (let ((expansion (xml-parse-string)))
 472                         (setq children
 473                               (if (stringp expansion)
 474                                   (if (stringp (car children))
 475                                       ;; The two strings were separated by a comment.
 476                                       (setq children (append (concat (car children) expansion)
 477                                                              (cdr children)))
 478                                     (setq children (append (list expansion) children)))
 479                                 (setq children (append expansion children))))))))
 480
 481                   (goto-char (match-end 0))
 482                   (nreverse children)))
 483             ;;  This was an invalid start tag (Expected ">", but didn't see it.)
 484             (error "XML: (Well-Formed) Couldn't parse tag: %s"
 485                    (buffer-substring (- (point) 10) (+ (point) 1)))))))
 486      (t ;; (Not one of PI, CDATA, Comment, End tag, or Start tag)
 487       (unless xml-sub-parser            ; Usually, we error out.
 488         (error "XML: (Well-Formed) Invalid character"))
 489
 490       ;; However, if we're parsing incrementally, then we need to deal
 491       ;; with stray CDATA.
 492       (xml-parse-string)))))
 493
 494 (defun xml-parse-string ()
 495   "Parse the next whatever.  Could be a string, or an element."
 496     (let* ((pos (point))
 497            (string (progn (if (search-forward "<" nil t)
 498                               (forward-char -1)
 499                             (goto-char (point-max)))
 500                           (buffer-substring pos (point)))))
 501       ;; Clean up the string.  As per XML specifications, the XML
 502       ;; processor should always pass the whole string to the
 503       ;; application.  But \r's should be replaced:
 504       ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends
 505       (setq pos 0)
 506       (while (string-match "\r\n?" string pos)
 507         (setq string (replace-match "\n" t t string))
 508         (setq pos (1+ (match-beginning 0))))
 509
 510       (xml-substitute-special string)))
 511
 512 (defun xml-parse-attlist (&optional xml-ns)
 513   "Return the attribute-list after point.
 514 Leave point at the first non-blank character after the tag."
 515   (let ((attlist ())
 516         end-pos name)
 517     (skip-syntax-forward " ")
 518     (while (looking-at (eval-when-compile
 519                          (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*")))
 520       (setq end-pos (match-end 0))
 521       (setq name (xml-maybe-do-ns (match-string 1) nil xml-ns))
 522       (goto-char end-pos)
 523
 524       ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize
 525
 526       ;; Do we have a string between quotes (or double-quotes),
 527       ;;  or a simple word ?
 528       (if (looking-at "\"\\([^\"]*\\)\"")
 529           (setq end-pos (match-end 0))
 530         (if (looking-at "'\\([^']*\\)'")
 531             (setq end-pos (match-end 0))
 532           (error "XML: (Not Well-Formed) Attribute values must be given between quotes")))
 533
 534       ;; Each attribute must be unique within a given element
 535       (if (assoc name attlist)
 536           (error "XML: (Not Well-Formed) Each attribute must be unique within an element"))
 537
 538       ;; Multiple whitespace characters should be replaced with a single one
 539       ;; in the attributes
 540       (let ((string (match-string 1))
 541             (pos 0))
 542         (replace-regexp-in-string "\\s-\\{2,\\}" " " string)
 543         (let ((expansion (xml-substitute-special string)))
 544           (unless (stringp expansion)
 545             ; We say this is the constraint.  It is acctually that
 546             ; external entities nor "<" can be in an attribute value.
 547             (error "XML: (Not Well-Formed) Entities in attributes cannot expand into elements"))
 548           (push (cons name expansion) attlist)))
 549
 550       (goto-char end-pos)
 551       (skip-syntax-forward " "))
 552     (nreverse attlist)))
 553
 554 ;;*******************************************************************
 555 ;;**
 556 ;;**  The DTD (document type declaration)
 557 ;;**  The following functions know how to skip or parse the DTD of
 558 ;;**  a document
 559 ;;**
 560 ;;*******************************************************************
 561
 562 ;; Fixme: This fails at least if the DTD contains conditional sections.
 563
 564 (defun xml-skip-dtd ()
 565   "Skip the DTD at point.
 566 This follows the rule [28] in the XML specifications."
 567   (let ((xml-validating-parser nil))
 568     (xml-parse-dtd)))
 569
 570 (defun xml-parse-dtd (&optional parse-ns)
 571   "Parse the DTD at point."
 572   (forward-char (eval-when-compile (length "<!DOCTYPE")))
 573   (skip-syntax-forward " ")
 574   (if (and (looking-at ">")
 575            xml-validating-parser)
 576       (error "XML: (Validity) Invalid DTD (expecting name of the document)"))
 577
 578   ;;  Get the name of the document
 579   (looking-at xml-name-regexp)
 580   (let ((dtd (list (match-string 0) 'dtd))
 581         type element end-pos)
 582     (goto-char (match-end 0))
 583
 584     (skip-syntax-forward " ")
 585     ;; XML [75]
 586     (cond ((looking-at "PUBLIC\\s-+")
 587            (goto-char (match-end 0))
 588            (unless (or (re-search-forward
 589                         "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\""
 590                         nil t)
 591                        (re-search-forward
 592                         "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'"
 593                         nil t))
 594              (error "XML: Missing Public ID"))
 595            (let ((pubid (match-string 1)))
 596              (skip-syntax-forward " ")
 597              (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
 598                          (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
 599                (error "XML: Missing System ID"))
 600              (push (list pubid (match-string 1) 'public) dtd)))
 601           ((looking-at "SYSTEM\\s-+")
 602            (goto-char (match-end 0))
 603            (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
 604                        (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
 605              (error "XML: Missing System ID"))
 606            (push (list (match-string 1) 'system) dtd)))
 607     (skip-syntax-forward " ")
 608     (if (eq ?> (char-after))
 609         (forward-char)
 610       (if (not (eq (char-after) ?\[))
 611           (error "XML: Bad DTD")
 612         (forward-char)
 613         ;;  Parse the rest of the DTD
 614         ;;  Fixme: Deal with NOTATION, PIs.
 615         (while (not (looking-at "\\s-*\\]"))
 616           (skip-syntax-forward " ")
 617           (cond
 618
 619            ;;  Translation of rule [45] of XML specifications
 620            ((looking-at
 621              "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>")
 622
 623             (setq element (match-string 1)
 624                   type    (match-string-no-properties 2))
 625             (setq end-pos (match-end 0))
 626
 627             ;;  Translation of rule [46] of XML specifications
 628             (cond
 629              ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration
 630               (setq type 'empty))
 631              ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents
 632               (setq type 'any))
 633              ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47])
 634               (setq type (xml-parse-elem-type (match-string 1 type))))
 635              ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution
 636               nil)
 637              (t
 638               (if xml-validating-parser
 639                   (error "XML: (Validity) Invalid element type in the DTD"))))
 640
 641             ;;  rule [45]: the element declaration must be unique
 642             (if (and (assoc element dtd)
 643                      xml-validating-parser)
 644                 (error "XML: (Validity) Element declarations must be unique in a DTD (<%s>)"
 645                        element))
 646
 647             ;;  Store the element in the DTD
 648             (push (list element type) dtd)
 649             (goto-char end-pos))
 650
 651            ;; Translation of rule [52] of XML specifications
 652            ((looking-at (concat "<!ATTLIST[ \t\n\r]*\\(" xml-name-re
 653                                 "\\)[ \t\n\r]*\\(" xml-att-def-re
 654                                 "\\)*[ \t\n\r]*>"))
 655
 656             ;; We don't do anything with ATTLIST currently
 657             (goto-char (match-end 0)))
 658
 659            ((looking-at "<!--")
 660             (search-forward "-->"))
 661            ((looking-at (concat "<!ENTITY[ \t\n\r]*\\(" xml-name-re
 662                                 "\\)[ \t\n\r]*\\(" xml-entity-value-re
 663                                 "\\)[ \t\n\r]*>"))
 664             (let ((name  (match-string 1))
 665                   (value (substring (match-string 2) 1
 666                                     (- (length (match-string 2)) 1))))
 667               (goto-char (match-end 0))
 668               (setq xml-entity-alist
 669                     (append xml-entity-alist
 670                             (list (cons name
 671                                         (with-temp-buffer
 672                                           (insert value)
 673                                           (goto-char (point-min))
 674                                           (xml-parse-fragment
 675                                            xml-validating-parser
 676                                            parse-ns))))))))
 677            ((or (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re
 678                                     "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
 679                                     "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
 680                 (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re
 681                                     "\\)[ \t\n\r]+PUBLIC[ \t\n\r]+"
 682                                     "\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
 683                                     "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'"
 684                                     "[ \t\n\r]+\\(\"[^\"]*\"\\|'[^']*'\\)"
 685                                     "[ \t\n\r]*>")))
 686             (let ((name  (match-string 1))
 687                   (file  (substring (match-string 2) 1
 688                                     (- (length (match-string 2)) 1))))
 689               (goto-char (match-end 0))
 690               (setq xml-entity-alist
 691                     (append xml-entity-alist
 692                             (list (cons name (with-temp-buffer
 693                                                (insert-file-contents file)
 694                                                (goto-char (point-min))
 695                                                (xml-parse-fragment
 696                                                 xml-validating-parser
 697                                                 parse-ns))))))))
 698            ;; skip parameter entity declarations
 699            ((or (looking-at (concat "<!ENTITY[ \t\n\r]+%[ \t\n\r]+\\(" xml-name-re
 700                                     "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
 701                                     "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
 702                 (looking-at (concat "<!ENTITY[ \t\n\r]+"
 703                                     "%[ \t\n\r]+"
 704                                     "\\(" xml-name-re "\\)[ \t\n\r]+"
 705                                     "PUBLIC[ \t\n\r]+"
 706                                     "\\(\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
 707                                     "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'\\)[ \t\n\r]+"
 708                                     "\\(\"[^\"]+\"\\|'[^']+'\\)"
 709                                     "[ \t\n\r]*>")))
 710             (goto-char (match-end 0)))
 711            ;; skip parameter entities
 712            ((looking-at (concat "%" xml-name-re ";"))
 713             (goto-char (match-end 0)))
 714            (t
 715             (when xml-validating-parser
 716               (error "XML: (Validity) Invalid DTD item"))))))
 717       (if (looking-at "\\s-*]>")
 718           (goto-char (match-end 0))))
 719     (nreverse dtd)))
 720
 721 (defun xml-parse-elem-type (string)
 722   "Convert element type STRING into a Lisp structure."
 723
 724   (let (elem modifier)
 725     (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
 726         (progn
 727           (setq elem     (match-string 1 string)
 728                 modifier (match-string 2 string))
 729           (if (string-match "|" elem)
 730               (setq elem (cons 'choice
 731                                (mapcar 'xml-parse-elem-type
 732                                        (split-string elem "|"))))
 733             (if (string-match "," elem)
 734                 (setq elem (cons 'seq
 735                                  (mapcar 'xml-parse-elem-type
 736                                          (split-string elem ",")))))))
 737       (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string)
 738           (setq elem     (match-string 1 string)
 739                 modifier (match-string 2 string))))
 740
 741     (if (and (stringp elem) (string= elem "#PCDATA"))
 742         (setq elem 'pcdata))
 743
 744     (cond
 745      ((string= modifier "+")
 746       (list '+ elem))
 747      ((string= modifier "*")
 748       (list '* elem))
 749      ((string= modifier "?")
 750       (list '\? elem))
 751      (t
 752       elem))))
 753
 754 ;;*******************************************************************
 755 ;;**
 756 ;;**  Substituting special XML sequences
 757 ;;**
 758 ;;*******************************************************************
 759
 760 (defun xml-substitute-special (string)
 761   "Return STRING, after subsituting entity references."
 762   ;; This originally made repeated passes through the string from the
 763   ;; beginning, which isn't correct, since then either "&amp;amp;" or
 764   ;; "&#38;amp;" won't DTRT.
 765
 766   (let ((point 0)
 767         children end-point)
 768     (while (string-match "&\\([^;]*\\);" string point)
 769       (setq end-point (match-end 0))
 770       (let* ((this-part (match-string 1 string))
 771              (prev-part (substring string point (match-beginning 0)))
 772              (entity (assoc this-part xml-entity-alist))
 773              (expansion
 774               (cond ((string-match "#\\([0-9]+\\)" this-part)
 775                      (let ((c (decode-char
 776                                'ucs
 777                                (string-to-number (match-string 1 this-part)))))
 778                        (if c (string c))))
 779                     ((string-match "#x\\([[:xdigit:]]+\\)" this-part)
 780                      (let ((c (decode-char
 781                                'ucs
 782                                (string-to-number (match-string 1 this-part) 16))))
 783                        (if c (string c))))
 784                     (entity
 785                      (cdr entity))
 786                     ((eq (length this-part) 0)
 787                      (error "XML: (Not Well-Formed) No entity given"))
 788                     (t
 789                      (if xml-validating-parser
 790                          (error "XML: (Validity) Undefined entity `%s'"
 791                                 this-part)
 792                        xml-undefined-entity)))))
 793
 794         (cond ((null children)
 795                ;; FIXME: If we have an entity that expands into XML, this won't work.
 796                (setq children
 797                      (concat prev-part expansion)))
 798               ((stringp children)
 799                (if (stringp expansion)
 800                    (setq children (concat children prev-part expansion))
 801                  (setq children (list expansion (concat prev-part children)))))
 802               ((and (stringp expansion)
 803                     (stringp (car children)))
 804                (setcar children (concat prev-part expansion (car children))))
 805               ((stringp expansion)
 806                (setq children (append (concat prev-part expansion)
 807                                       children)))
 808               ((stringp (car children))
 809                (setcar children (concat (car children) prev-part))
 810                (setq children (append expansion children)))
 811               (t
 812                (setq children (list expansion
 813                                     prev-part
 814                                     children))))
 815         (setq point end-point)))
 816     (cond ((stringp children)
 817            (concat children (substring string point)))
 818           ((stringp (car (last children)))
 819            (concat (car (last children)) (substring string point)))
 820           ((null children)
 821            string)
 822           (t
 823            (concat (mapconcat 'identity
 824                               (nreverse children)
 825                               "")
 826                    (substring string point))))))
 827
 828 ;;*******************************************************************
 829 ;;**
 830 ;;**  Printing a tree.
 831 ;;**  This function is intended mainly for debugging purposes.
 832 ;;**
 833 ;;*******************************************************************
 834
 835 (defun xml-debug-print (xml &optional indent-string)
 836   "Outputs the XML in the current buffer.
 837 XML can be a tree or a list of nodes.
 838 The first line is indented with the optional INDENT-STRING."
 839   (setq indent-string (or indent-string ""))
 840   (dolist (node xml)
 841     (xml-debug-print-internal node indent-string)))
 842
 843 (defalias 'xml-print 'xml-debug-print)
 844
 845 (defun xml-debug-print-internal (xml indent-string)
 846   "Outputs the XML tree in the current buffer.
 847 The first line is indented with INDENT-STRING."
 848   (let ((tree xml)
 849         attlist)
 850     (insert indent-string ?< (symbol-name (xml-node-name tree)))
 851
 852     ;;  output the attribute list
 853     (setq attlist (xml-node-attributes tree))
 854     (while attlist
 855       (insert ?\  (symbol-name (caar attlist)) "=\"" (cdar attlist) ?\")
 856       (setq attlist (cdr attlist)))
 857
 858     (setq tree (xml-node-children tree))
 859
 860     (if (null tree)
 861         (insert ?/ ?>)
 862       (insert ?>)
 863
 864       ;;  output the children
 865       (dolist (node tree)
 866         (cond
 867          ((listp node)
 868           (insert ?\n)
 869           (xml-debug-print-internal node (concat indent-string "  ")))
 870          ((stringp node) (insert node))
 871          (t
 872           (error "Invalid XML tree"))))
 873
 874       (when (not (and (null (cdr tree))
 875                       (stringp (car tree))))
 876         (insert ?\n indent-string))
 877       (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>))))
 878
 879 (provide 'xml)
 880
 881 ;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b
 882 ;;; xml.el ends here