lisp/xml.el

   1 ;;; xml.el --- XML parser
   2
   3 ;; Copyright (C) 2000-2012 Free Software Foundation, Inc.
   4
   5 ;; Author: Emmanuel Briot  <briot@gnat.com>
   6 ;; Maintainer: Mark A. Hershberger <mah@everybody.org>
   7 ;; Keywords: xml, data
   8
   9 ;; This file is part of GNU Emacs.
  10
  11 ;; GNU Emacs is free software: you can redistribute it and/or modify
  12 ;; it under the terms of the GNU General Public License as published by
  13 ;; the Free Software Foundation, either version 3 of the License, or
  14 ;; (at your option) any later version.
  15
  16 ;; GNU Emacs is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;; GNU General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
  23
  24 ;;; Commentary:
  25
  26 ;; This file contains a somewhat incomplete non-validating XML parser.  It
  27 ;; parses a file, and returns a list that can be used internally by
  28 ;; any other Lisp libraries.
  29
  30 ;;; FILE FORMAT
  31
  32 ;; The document type declaration may either be ignored or (optionally)
  33 ;; parsed, but currently the parsing will only accept element
  34 ;; declarations.  The XML file is assumed to be well-formed.  In case
  35 ;; of error, the parsing stops and the XML file is shown where the
  36 ;; parsing stopped.
  37 ;;
  38 ;; It also knows how to ignore comments and processing instructions.
  39 ;;
  40 ;; The XML file should have the following format:
  41 ;;    <node1 attr1="name1" attr2="name2" ...>value
  42 ;;       <node2 attr3="name3" attr4="name4">value2</node2>
  43 ;;       <node3 attr5="name5" attr6="name6">value3</node3>
  44 ;;    </node1>
  45 ;; Of course, the name of the nodes and attributes can be anything.  There can
  46 ;; be any number of attributes (or none), as well as any number of children
  47 ;; below the nodes.
  48 ;;
  49 ;; There can be only top level node, but with any number of children below.
  50
  51 ;;; LIST FORMAT
  52
  53 ;; The functions `xml-parse-file', `xml-parse-region' and
  54 ;; `xml-parse-tag' return a list with the following format:
  55 ;;
  56 ;;    xml-list   ::= (node node ...)
  57 ;;    node       ::= (qname attribute-list . child_node_list)
  58 ;;    child_node_list ::= child_node child_node ...
  59 ;;    child_node ::= node | string
  60 ;;    qname      ::= (:namespace-uri . "name") | "name"
  61 ;;    attribute_list ::= ((qname . "value") (qname . "value") ...)
  62 ;;                       | nil
  63 ;;    string     ::= "..."
  64 ;;
  65 ;; Some macros are provided to ease the parsing of this list.
  66 ;; Whitespace is preserved.  Fixme: There should be a tree-walker that
  67 ;; can remove it.
  68
  69 ;; TODO:
  70 ;;  * xml:base, xml:space support
  71 ;;  * more complete DOCTYPE parsing
  72 ;;  * pi support
  73
  74 ;;; Code:
  75
  76 ;; Note that buffer-substring and match-string were formerly used in
  77 ;; several places, because the -no-properties variants remove
  78 ;; composition info.  However, after some discussion on emacs-devel,
  79 ;; the consensus was that the speed of the -no-properties variants was
  80 ;; a worthwhile tradeoff especially since we're usually parsing files
  81 ;; instead of hand-crafted XML.
  82
  83 ;;*******************************************************************
  84 ;;**
  85 ;;**  Macros to parse the list
  86 ;;**
  87 ;;*******************************************************************
  88
  89 (defconst xml-undefined-entity "?"
  90   "What to substitute for undefined entities")
  91
  92 (defvar xml-entity-alist
  93   '(("lt"   . "<")
  94     ("gt"   . ">")
  95     ("apos" . "'")
  96     ("quot" . "\"")
  97     ("amp"  . "&"))
  98   "Alist of defined XML entities.")
  99
 100 (defvar xml-parameter-entity-alist nil
 101   "Alist of defined XML parametric entities.")
 102
 103 (defvar xml-sub-parser nil
 104   "Non-nil when the XML parser is parsing an XML fragment.")
 105
 106 (defvar xml-validating-parser nil
 107   "Set to non-nil to get validity checking.")
 108
 109 (defsubst xml-node-name (node)
 110   "Return the tag associated with NODE.
 111 Without namespace-aware parsing, the tag is a symbol.
 112
 113 With namespace-aware parsing, the tag is a cons of a string
 114 representing the uri of the namespace with the local name of the
 115 tag.  For example,
 116
 117     <foo>
 118
 119 would be represented by
 120
 121     '(\"\" . \"foo\")."
 122
 123   (car node))
 124
 125 (defsubst xml-node-attributes (node)
 126   "Return the list of attributes of NODE.
 127 The list can be nil."
 128   (nth 1 node))
 129
 130 (defsubst xml-node-children (node)
 131   "Return the list of children of NODE.
 132 This is a list of nodes, and it can be nil."
 133   (cddr node))
 134
 135 (defun xml-get-children (node child-name)
 136   "Return the children of NODE whose tag is CHILD-NAME.
 137 CHILD-NAME should match the value returned by `xml-node-name'."
 138   (let ((match ()))
 139     (dolist (child (xml-node-children node))
 140       (if (and (listp child)
 141                (equal (xml-node-name child) child-name))
 142           (push child match)))
 143     (nreverse match)))
 144
 145 (defun xml-get-attribute-or-nil (node attribute)
 146   "Get from NODE the value of ATTRIBUTE.
 147 Return nil if the attribute was not found.
 148
 149 See also `xml-get-attribute'."
 150   (cdr (assoc attribute (xml-node-attributes node))))
 151
 152 (defsubst xml-get-attribute (node attribute)
 153   "Get from NODE the value of ATTRIBUTE.
 154 An empty string is returned if the attribute was not found.
 155
 156 See also `xml-get-attribute-or-nil'."
 157   (or (xml-get-attribute-or-nil node attribute) ""))
 158
 159 ;;*******************************************************************
 160 ;;**
 161 ;;**  Creating the list
 162 ;;**
 163 ;;*******************************************************************
 164
 165 ;;;###autoload
 166 (defun xml-parse-file (file &optional parse-dtd parse-ns)
 167   "Parse the well-formed XML file FILE.
 168 If FILE is already visited, use its buffer and don't kill it.
 169 Returns the top node with all its children.
 170 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped.
 171 If PARSE-NS is non-nil, then QNAMES are expanded."
 172   (if (get-file-buffer file)
 173       (with-current-buffer (get-file-buffer file)
 174         (save-excursion
 175           (xml-parse-region (point-min)
 176                             (point-max)
 177                             (current-buffer)
 178                             parse-dtd parse-ns)))
 179     (with-temp-buffer
 180       (insert-file-contents file)
 181       (xml-parse-region (point-min)
 182                         (point-max)
 183                         (current-buffer)
 184                         parse-dtd parse-ns))))
 185
 186
 187 (defvar xml-name-re)
 188 (defvar xml-entity-value-re)
 189 (defvar xml-att-def-re)
 190 (let* ((start-chars (concat "[:alpha:]:_"))
 191        (name-chars  (concat "-[:digit:]." start-chars))
 192        ;;[3]    S          ::=          (#x20 | #x9 | #xD | #xA)+
 193        (whitespace  "[ \t\n\r]"))
 194   ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6]
 195   ;;                      | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF]
 196   ;;                      | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF]
 197   ;;                      | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
 198   (defvar xml-name-start-char-re (concat "[" start-chars "]"))
 199   ;;[4a] NameChar       ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
 200   (defvar xml-name-char-re       (concat "[" name-chars  "]"))
 201   ;;[5] Name     ::= NameStartChar (NameChar)*
 202   (defvar xml-name-re            (concat xml-name-start-char-re xml-name-char-re "*"))
 203   ;;[6] Names    ::= Name (#x20 Name)*
 204   (defvar xml-names-re           (concat xml-name-re "\\(?: " xml-name-re "\\)*"))
 205   ;;[7] Nmtoken ::= (NameChar)+
 206   (defvar xml-nmtoken-re         (concat xml-name-char-re "+"))
 207   ;;[8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
 208   (defvar xml-nmtokens-re        (concat xml-nmtoken-re "\\(?: " xml-name-re "\\)*"))
 209   ;;[66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
 210   (defvar xml-char-ref-re        "\\(?:&#[0-9]+;\\|&#x[0-9a-fA-F]+;\\)")
 211   ;;[68] EntityRef   ::= '&' Name ';'
 212   (defvar xml-entity-ref         (concat "&" xml-name-re ";"))
 213   ;;[69] PEReference ::= '%' Name ';'
 214   (defvar xml-pe-reference-re    (concat "%" xml-name-re ";"))
 215   ;;[67] Reference   ::= EntityRef | CharRef
 216   (defvar xml-reference-re       (concat "\\(?:" xml-entity-ref "\\|" xml-char-ref-re "\\)"))
 217   ;;[10]        AttValue           ::=          '"' ([^<&"] | Reference)* '"' |  "'" ([^<&'] | Reference)* "'"
 218   (defvar xml-att-value-re    (concat "\\(?:\"\\(?:[^&\"]\\|" xml-reference-re "\\)*\"\\|"
 219                                       "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)"))
 220   ;;[56]        TokenizedType      ::=          'ID'       [VC: ID] [VC: One ID per Element Type] [VC: ID Attribute Default]
 221   ;;                                            | 'IDREF'    [VC: IDREF]
 222   ;;                                                  | 'IDREFS'   [VC: IDREF]
 223   ;;                                            | 'ENTITY'   [VC: Entity Name]
 224   ;;                                            | 'ENTITIES' [VC: Entity Name]
 225   ;;                                            | 'NMTOKEN'  [VC: Name Token]
 226   ;;                                            | 'NMTOKENS' [VC: Name Token]
 227   (defvar xml-tokenized-type-re "\\(?:ID\\|IDREF\\|IDREFS\\|ENTITY\\|ENTITIES\\|NMTOKEN\\|NMTOKENS\\)")
 228   ;;[58]        NotationType       ::=          'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 229   (defvar xml-notation-type-re (concat "\\(?:NOTATION" whitespace "(" whitespace "*" xml-name-re
 230                                        "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)"))
 231   ;;[59]        Enumeration        ::=          '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'      [VC: Enumeration] [VC: No Duplicate Tokens]
 232   (defvar xml-enumeration-re (concat "\\(?:(" whitespace "*" xml-nmtoken-re
 233                                      "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*"
 234                                      whitespace ")\\)"))
 235   ;;[57]        EnumeratedType     ::=          NotationType | Enumeration
 236   (defvar xml-enumerated-type-re (concat "\\(?:" xml-notation-type-re "\\|" xml-enumeration-re "\\)"))
 237   ;;[54]        AttType    ::=          StringType | TokenizedType | EnumeratedType
 238   ;;[55]        StringType         ::=          'CDATA'
 239   (defvar xml-att-type-re (concat "\\(?:CDATA\\|" xml-tokenized-type-re "\\|" xml-notation-type-re"\\|" xml-enumerated-type-re "\\)"))
 240   ;;[60]        DefaultDecl        ::=          '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
 241   (defvar xml-default-decl-re (concat "\\(?:#REQUIRED\\|#IMPLIED\\|\\(?:#FIXED" whitespace "\\)*" xml-att-value-re "\\)"))
 242   ;;[53]        AttDef     ::=          S Name S AttType S DefaultDecl
 243   (defvar xml-att-def-re         (concat "\\(?:" whitespace "*" xml-name-re
 244                                          whitespace "*" xml-att-type-re
 245                                          whitespace "*" xml-default-decl-re "\\)"))
 246   ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 247   ;;               |  "'" ([^%&'] | PEReference | Reference)* "'"
 248   (defvar xml-entity-value-re    (concat "\\(?:\"\\(?:[^%&\"]\\|" xml-pe-reference-re
 249                                          "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|"
 250                                          xml-pe-reference-re "\\|" xml-reference-re "\\)*'\\)")))
 251 ;;[75] ExternalID ::= 'SYSTEM' S SystemLiteral
 252 ;;                 | 'PUBLIC' S PubidLiteral S SystemLiteral
 253 ;;[76] NDataDecl ::=    S 'NDATA' S
 254 ;;[73] EntityDef  ::= EntityValue| (ExternalID NDataDecl?)
 255 ;;[71] GEDecl     ::= '<!ENTITY' S Name S EntityDef S? '>'
 256 ;;[74] PEDef      ::= EntityValue | ExternalID
 257 ;;[72] PEDecl     ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 258 ;;[70] EntityDecl ::= GEDecl | PEDecl
 259
 260 ;; Note that this is setup so that we can do whitespace-skipping with
 261 ;; `(skip-syntax-forward " ")', inter alia.  Previously this was slow
 262 ;; compared with `re-search-forward', but that has been fixed.  Also
 263 ;; note that the standard syntax table contains other characters with
 264 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
 265 ;; which we might skip whitespace -- specifically, they're not
 266 ;; NameChars [XML 4].
 267
 268 (defvar xml-syntax-table
 269   (let ((table (make-syntax-table)))
 270     ;; Get space syntax correct per XML [3].
 271     (dotimes (c 31)
 272       (modify-syntax-entry c "." table)) ; all are space in standard table
 273     (dolist (c '(?\t ?\n ?\r))           ; these should be space
 274       (modify-syntax-entry c " " table))
 275     ;; For skipping attributes.
 276     (modify-syntax-entry ?\" "\"" table)
 277     (modify-syntax-entry ?' "\"" table)
 278     ;; Non-alnum name chars should be symbol constituents (`-' and `_'
 279     ;; are OK by default).
 280     (modify-syntax-entry ?. "_" table)
 281     (modify-syntax-entry ?: "_" table)
 282     ;; XML [89]
 283     (unless (featurep 'xemacs)
 284       (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005
 285                           #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC
 286                           #x30FD #x30FE))
 287         (modify-syntax-entry (decode-char 'ucs c) "w" table)))
 288     ;; Fixme: rest of [4]
 289     table)
 290   "Syntax table used by `xml-parse-region'.")
 291
 292 ;; XML [5]
 293 ;; Note that [:alpha:] matches all multibyte chars with word syntax.
 294 (eval-and-compile
 295   (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*"))
 296
 297 ;; Fixme:  This needs re-writing to deal with the XML grammar properly, i.e.
 298 ;;   document    ::=    prolog element Misc*
 299 ;;   prolog    ::=    XMLDecl? Misc* (doctypedecl Misc*)?
 300
 301 ;;;###autoload
 302 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns)
 303   "Parse the region from BEG to END in BUFFER.
 304 If BUFFER is nil, it defaults to the current buffer.
 305 Returns the XML list for the region, or raises an error if the region
 306 is not well-formed XML.
 307 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped,
 308 and returned as the first element of the list.
 309 If PARSE-NS is non-nil, then QNAMES are expanded."
 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           ;; Prevent entity definitions from changing the defaults
 315           (xml-entity-alist xml-entity-alist)
 316           (xml-parameter-entity-alist xml-parameter-entity-alist)
 317           xml result dtd)
 318       (save-excursion
 319         (if buffer
 320             (set-buffer buffer))
 321         (save-restriction
 322           (narrow-to-region beg end)
 323           (goto-char (point-min))
 324           (while (not (eobp))
 325             (if (search-forward "<" nil t)
 326                 (progn
 327                   (forward-char -1)
 328                   (setq result (xml-parse-tag parse-dtd parse-ns))
 329                   (cond
 330                    ((null result)
 331                     ;; Not looking at an xml start tag.
 332                     (unless (eobp)
 333                       (forward-char 1)))
 334                    ((and xml (not xml-sub-parser))
 335                     ;; Translation of rule [1] of XML specifications
 336                     (error "XML: (Not Well-Formed) Only one root tag allowed"))
 337                    ((and (listp (car result))
 338                          parse-dtd)
 339                     (setq dtd (car result))
 340                     (if (cdr result)    ; possible leading comment
 341                         (add-to-list 'xml (cdr result))))
 342                    (t
 343                     (add-to-list 'xml result))))
 344               (goto-char (point-max))))
 345           (if parse-dtd
 346               (cons dtd (nreverse xml))
 347             (nreverse xml)))))))
 348
 349 (defun xml-maybe-do-ns (name default xml-ns)
 350   "Perform any namespace expansion.
 351 NAME is the name to perform the expansion on.
 352 DEFAULT is the default namespace.  XML-NS is a cons of namespace
 353 names to uris.  When namespace-aware parsing is off, then XML-NS
 354 is nil.
 355
 356 During namespace-aware parsing, any name without a namespace is
 357 put into the namespace identified by DEFAULT.  nil is used to
 358 specify that the name shouldn't be given a namespace."
 359   (if (consp xml-ns)
 360       (let* ((nsp (string-match ":" name))
 361              (lname (if nsp (substring name (match-end 0)) name))
 362              (prefix (if nsp (substring name 0 (match-beginning 0)) default))
 363              (special (and (string-equal lname "xmlns") (not prefix)))
 364              ;; Setting default to nil will insure that there is not
 365              ;; matching cons in xml-ns.  In which case we
 366              (ns (or (cdr (assoc (if special "xmlns" prefix)
 367                                  xml-ns))
 368                      "")))
 369         (cons ns (if special "" lname)))
 370     (intern name)))
 371
 372 (defun xml-parse-fragment (&optional parse-dtd parse-ns)
 373   "Parse xml-like fragments."
 374   (let ((xml-sub-parser t)
 375         ;; Prevent entity definitions from changing the defaults
 376         (xml-entity-alist xml-entity-alist)
 377         (xml-parameter-entity-alist xml-parameter-entity-alist)
 378         children)
 379     (while (not (eobp))
 380       (let ((bit (xml-parse-tag
 381                   parse-dtd parse-ns)))
 382         (if children
 383             (setq children (append (list bit) children))
 384           (if (stringp bit)
 385               (setq children (list bit))
 386             (setq children bit)))))
 387     (reverse children)))
 388
 389 (defun xml-parse-tag (&optional parse-dtd parse-ns)
 390   "Parse the tag at point.
 391 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
 392 returned as the first element in the list.
 393 If PARSE-NS is non-nil, then QNAMES are expanded.
 394 Returns one of:
 395  - a list : the matching node
 396  - nil    : the point is not looking at a tag.
 397  - a pair : the first element is the DTD, the second is the node."
 398   (let ((xml-validating-parser (or parse-dtd xml-validating-parser))
 399         (xml-ns (if (consp parse-ns)
 400                     parse-ns
 401                   (if parse-ns
 402                       (list
 403                        ;; Default for empty prefix is no namespace
 404                        (cons ""      "")
 405                        ;; "xml" namespace
 406                        (cons "xml"   "http://www.w3.org/XML/1998/namespace")
 407                        ;; We need to seed the xmlns namespace
 408                        (cons "xmlns" "http://www.w3.org/2000/xmlns/"))))))
 409     (cond
 410      ;; Processing instructions (like the <?xml version="1.0"?> tag at the
 411      ;; beginning of a document).
 412      ((looking-at "<\\?")
 413       (search-forward "?>")
 414       (skip-syntax-forward " ")
 415       (xml-parse-tag parse-dtd xml-ns))
 416      ;;  Character data (CDATA) sections, in which no tag should be interpreted
 417      ((looking-at "<!\\[CDATA\\[")
 418       (let ((pos (match-end 0)))
 419         (unless (search-forward "]]>" nil t)
 420           (error "XML: (Not Well Formed) CDATA section does not end anywhere in the document"))
 421         (concat
 422          (buffer-substring-no-properties pos (match-beginning 0))
 423          (xml-parse-string))))
 424      ;;  DTD for the document
 425      ((looking-at "<!DOCTYPE[ \t\n\r]")
 426       (let ((dtd (xml-parse-dtd parse-ns)))
 427         (skip-syntax-forward " ")
 428         (if xml-validating-parser
 429             (cons dtd (xml-parse-tag nil xml-ns))
 430           (xml-parse-tag nil xml-ns))))
 431      ;;  skip comments
 432      ((looking-at "<!--")
 433       (search-forward "-->")
 434       (skip-syntax-forward " ")
 435       (unless (eobp)
 436         (let ((xml-sub-parser t))
 437           (xml-parse-tag parse-dtd xml-ns))))
 438      ;;  end tag
 439      ((looking-at "</")
 440       '())
 441      ;;  opening tag
 442      ((looking-at "<\\([^/>[:space:]]+\\)")
 443       (goto-char (match-end 1))
 444
 445       ;; Parse this node
 446       (let* ((node-name (match-string-no-properties 1))
 447              ;; Parse the attribute list.
 448              (attrs (xml-parse-attlist xml-ns))
 449              children)
 450
 451         ;; add the xmlns:* attrs to our cache
 452         (when (consp xml-ns)
 453           (dolist (attr attrs)
 454             (when (and (consp (car attr))
 455                        (equal "http://www.w3.org/2000/xmlns/"
 456                               (caar attr)))
 457               (push (cons (cdar attr) (cdr attr))
 458                     xml-ns))))
 459
 460         (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns)))
 461
 462         ;; is this an empty element ?
 463         (if (looking-at "/>")
 464             (progn
 465               (forward-char 2)
 466               (nreverse children))
 467
 468           ;; is this a valid start tag ?
 469           (if (eq (char-after) ?>)
 470               (progn
 471                 (forward-char 1)
 472                 ;;  Now check that we have the right end-tag. Note that this
 473                 ;;  one might contain spaces after the tag name
 474                 (let ((end (concat "</" node-name "\\s-*>")))
 475                   (while (not (looking-at end))
 476                     (cond
 477                      ((looking-at "</")
 478                       (error "XML: (Not Well-Formed) Invalid end tag (expecting %s) at pos %d"
 479                              node-name (point)))
 480                      ((= (char-after) ?<)
 481                       (let ((tag (xml-parse-tag nil xml-ns)))
 482                         (when tag
 483                           (push tag children))))
 484                      (t
 485                       (let ((expansion (xml-parse-string)))
 486                         (setq children
 487                               (if (stringp expansion)
 488                                   (if (stringp (car children))
 489                                       ;; The two strings were separated by a comment.
 490                                       (setq children (append (list (concat (car children) expansion))
 491                                                              (cdr children)))
 492                                     (setq children (append (list expansion) children)))
 493                                 (setq children (append expansion children))))))))
 494
 495                   (goto-char (match-end 0))
 496                   (nreverse children)))
 497             ;;  This was an invalid start tag (Expected ">", but didn't see it.)
 498             (error "XML: (Well-Formed) Couldn't parse tag: %s"
 499                    (buffer-substring-no-properties (- (point) 10) (+ (point) 1)))))))
 500      (t ;; (Not one of PI, CDATA, Comment, End tag, or Start tag)
 501       (unless xml-sub-parser            ; Usually, we error out.
 502         (error "XML: (Well-Formed) Invalid character"))
 503
 504       ;; However, if we're parsing incrementally, then we need to deal
 505       ;; with stray CDATA.
 506       (xml-parse-string)))))
 507
 508 (defun xml-parse-string ()
 509   "Parse the next whatever.  Could be a string, or an element."
 510   (let* ((pos (point))
 511          (string (progn (skip-chars-forward "^<")
 512                         (buffer-substring-no-properties pos (point)))))
 513     ;; Clean up the string.  As per XML specifications, the XML
 514     ;; processor should always pass the whole string to the
 515     ;; application.  But \r's should be replaced:
 516     ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends
 517     (setq pos 0)
 518     (while (string-match "\r\n?" string pos)
 519       (setq string (replace-match "\n" t t string))
 520       (setq pos (1+ (match-beginning 0))))
 521
 522     (xml-substitute-special string)))
 523
 524 (defun xml-parse-attlist (&optional xml-ns)
 525   "Return the attribute-list after point.
 526 Leave point at the first non-blank character after the tag."
 527   (let ((attlist ())
 528         end-pos name)
 529     (skip-syntax-forward " ")
 530     (while (looking-at (eval-when-compile
 531                          (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*")))
 532       (setq end-pos (match-end 0))
 533       (setq name (xml-maybe-do-ns (match-string-no-properties 1) nil xml-ns))
 534       (goto-char end-pos)
 535
 536       ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize
 537
 538       ;; Do we have a string between quotes (or double-quotes),
 539       ;;  or a simple word ?
 540       (if (looking-at "\"\\([^\"]*\\)\"")
 541           (setq end-pos (match-end 0))
 542         (if (looking-at "'\\([^']*\\)'")
 543             (setq end-pos (match-end 0))
 544           (error "XML: (Not Well-Formed) Attribute values must be given between quotes")))
 545
 546       ;; Each attribute must be unique within a given element
 547       (if (assoc name attlist)
 548           (error "XML: (Not Well-Formed) Each attribute must be unique within an element"))
 549
 550       ;; Multiple whitespace characters should be replaced with a single one
 551       ;; in the attributes
 552       (let ((string (match-string-no-properties 1)))
 553         (replace-regexp-in-string "\\s-\\{2,\\}" " " string)
 554         (let ((expansion (xml-substitute-special string)))
 555           (unless (stringp expansion)
 556                                         ; We say this is the constraint.  It is actually that neither
 557                                         ; external entities nor "<" can be in an attribute value.
 558             (error "XML: (Not Well-Formed) Entities in attributes cannot expand into elements"))
 559           (push (cons name expansion) attlist)))
 560
 561       (goto-char end-pos)
 562       (skip-syntax-forward " "))
 563     (nreverse attlist)))
 564
 565 ;;*******************************************************************
 566 ;;**
 567 ;;**  The DTD (document type declaration)
 568 ;;**  The following functions know how to skip or parse the DTD of
 569 ;;**  a document
 570 ;;**
 571 ;;*******************************************************************
 572
 573 ;; Fixme: This fails at least if the DTD contains conditional sections.
 574
 575 (defun xml-skip-dtd ()
 576   "Skip the DTD at point.
 577 This follows the rule [28] in the XML specifications."
 578   (let ((xml-validating-parser nil))
 579     (xml-parse-dtd)))
 580
 581 (defun xml-parse-dtd (&optional parse-ns)
 582   "Parse the DTD at point."
 583   (forward-char (eval-when-compile (length "<!DOCTYPE")))
 584   (skip-syntax-forward " ")
 585   (if (and (looking-at ">")
 586            xml-validating-parser)
 587       (error "XML: (Validity) Invalid DTD (expecting name of the document)"))
 588
 589   ;;  Get the name of the document
 590   (looking-at xml-name-regexp)
 591   (let ((dtd (list (match-string-no-properties 0) 'dtd))
 592         (xml-parameter-entity-alist xml-parameter-entity-alist))
 593     (goto-char (match-end 0))
 594     (skip-syntax-forward " ")
 595
 596     ;; External subset (XML [75])
 597     (cond ((looking-at "PUBLIC\\s-+")
 598            (goto-char (match-end 0))
 599            (unless (or (re-search-forward
 600                         "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\""
 601                         nil t)
 602                        (re-search-forward
 603                         "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'"
 604                         nil t))
 605              (error "XML: Missing Public ID"))
 606            (let ((pubid (match-string-no-properties 1)))
 607              (skip-syntax-forward " ")
 608              (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
 609                          (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
 610                (error "XML: Missing System ID"))
 611              (push (list pubid (match-string-no-properties 1) 'public) dtd)))
 612           ((looking-at "SYSTEM\\s-+")
 613            (goto-char (match-end 0))
 614            (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
 615                        (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
 616              (error "XML: Missing System ID"))
 617            (push (list (match-string-no-properties 1) 'system) dtd)))
 618     (skip-syntax-forward " ")
 619
 620     (if (eq (char-after) ?>)
 621
 622         ;; No internal subset
 623         (forward-char)
 624
 625       ;; Internal subset (XML [28b])
 626       (unless (eq (char-after) ?\[)
 627         (error "XML: Bad DTD"))
 628       (forward-char)
 629
 630       ;; Parse the rest of the DTD
 631       ;; Fixme: Deal with NOTATION, PIs.
 632       (while (not (looking-at "\\s-*\\]"))
 633         (skip-syntax-forward " ")
 634         (cond
 635          ;; Element declaration [45]:
 636          ((looking-at "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>")
 637           (let ((element (match-string-no-properties 1))
 638                 (type    (match-string-no-properties 2))
 639                 (end-pos (match-end 0)))
 640             ;; Translation of rule [46] of XML specifications
 641             (cond
 642              ((string-match "^EMPTY[ \t\n\r]*$" type)       ; empty declaration
 643               (setq type 'empty))
 644              ((string-match "^ANY[ \t\n\r]*$" type)         ; any type of contents
 645               (setq type 'any))
 646              ((string-match "^(\\(.*\\))[ \t\n\r]*$" type)  ; children ([47])
 647               (setq type (xml-parse-elem-type (match-string-no-properties 1 type))))
 648              ((string-match "^%[^;]+;[ \t\n\r]*$" type)   ; substitution
 649               nil)
 650              (xml-validating-parser
 651               (error "XML: (Validity) Invalid element type in the DTD")))
 652
 653             ;; rule [45]: the element declaration must be unique
 654             (and (assoc element dtd)
 655                  xml-validating-parser
 656                  (error "XML: (Validity) DTD element declarations must be unique (<%s>)"
 657                         element))
 658
 659             ;;  Store the element in the DTD
 660             (push (list element type) dtd)
 661             (goto-char end-pos)))
 662
 663          ;; Attribute-list declaration [52] (currently unsupported):
 664          ((looking-at (concat "<!ATTLIST[ \t\n\r]*\\(" xml-name-re
 665                               "\\)[ \t\n\r]*\\(" xml-att-def-re
 666                               "\\)*[ \t\n\r]*>"))
 667           (goto-char (match-end 0)))
 668
 669          ;; Comments (skip to end):
 670          ((looking-at "<!--")
 671           (search-forward "-->"))
 672
 673          ;; Internal entity declarations:
 674          ((looking-at (concat "<!ENTITY[ \t\n\r]+\\(%[ \t\n\r]+\\)?\\("
 675                               xml-name-re "\\)[ \t\n\r]*\\("
 676                               xml-entity-value-re "\\)[ \t\n\r]*>"))
 677           (let* ((name (prog1 (match-string-no-properties 2)
 678                          (goto-char (match-end 0))))
 679                  (alist (if (match-string 1)
 680                             'xml-parameter-entity-alist
 681                           'xml-entity-alist))
 682                  ;; Retrieve the deplacement text:
 683                  (value (xml--entity-replacement-text
 684                          ;; Entity value, sans quotation marks:
 685                          (substring (match-string-no-properties 3) 1 -1))))
 686             ;; If the same entity is declared more than once, the
 687             ;; first declaration is binding.
 688             (unless (assoc name (symbol-value alist))
 689               (set alist (cons (cons name value) (symbol-value alist))))))
 690
 691          ;; External entity declarations (currently unsupported):
 692          ((or (looking-at (concat "<!ENTITY[ \t\n\r]+\\(%[ \t\n\r]+\\)?\\("
 693                                   xml-name-re "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
 694                                   "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
 695               (looking-at (concat "<!ENTITY[ \t\n\r]+\\(%[ \t\n\r]+\\)?\\("
 696                                   xml-name-re "\\)[ \t\n\r]+PUBLIC[ \t\n\r]+"
 697                                   "\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
 698                                   "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'"
 699                                   "[ \t\n\r]+\\(\"[^\"]*\"\\|'[^']*'\\)"
 700                                   "[ \t\n\r]*>")))
 701           (goto-char (match-end 0)))
 702
 703          ;; Parameter entity:
 704          ((looking-at (concat "%\\(" xml-name-re "\\);"))
 705           (goto-char (match-end 0))
 706           (let* ((entity (match-string 1))
 707                  (end (point-marker))
 708                  (elt (assoc entity xml-parameter-entity-alist)))
 709             (when elt
 710               (replace-match (cdr elt) t t)
 711               (goto-char end))))
 712
 713          ;; Anything else:
 714          (xml-validating-parser
 715           (error "XML: (Validity) Invalid DTD item"))))
 716
 717       (if (looking-at "\\s-*]>")
 718           (goto-char (match-end 0))))
 719     (nreverse dtd)))
 720
 721 (defun xml--entity-replacement-text (string)
 722   "Return the replacement text for the entity value STRING.
 723 The replacement text is obtained by replacing character
 724 references and parameter-entity references."
 725   (let ((ref-re (concat "\\(?:&#\\([0-9]+\\)\\|&#x\\([0-9a-fA-F]+\\)\\|%\\("
 726                         xml-name-re "\\)\\);"))
 727         children)
 728     (while (string-match ref-re string)
 729       (push (substring string 0 (match-beginning 0)) children)
 730       (let ((remainder (substring string (match-end 0)))
 731             ref val)
 732         (cond ((setq ref (match-string 1 string))
 733                ;; Decimal character reference
 734                (setq val (decode-char 'ucs (string-to-number ref)))
 735                (if val (push (string val) children)))
 736               ;; Hexadecimal character reference
 737               ((setq ref (match-string 2 string))
 738                (setq val (decode-char 'ucs (string-to-number ref 16)))
 739                (if val (push (string val) children)))
 740               ;; Parameter entity reference
 741               ((setq ref (match-string 3 string))
 742                (setq val (assoc ref xml-parameter-entity-alist))
 743                (if val
 744                    (push (cdr val) children)
 745                  (push (concat "%" ref ";") children))))
 746         (setq string remainder)))
 747     (mapconcat 'identity (nreverse (cons string children)) "")))
 748
 749 (defun xml-parse-elem-type (string)
 750   "Convert element type STRING into a Lisp structure."
 751
 752   (let (elem modifier)
 753     (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
 754         (progn
 755           (setq elem     (match-string-no-properties 1 string)
 756                 modifier (match-string-no-properties 2 string))
 757           (if (string-match "|" elem)
 758               (setq elem (cons 'choice
 759                                (mapcar 'xml-parse-elem-type
 760                                        (split-string elem "|"))))
 761             (if (string-match "," elem)
 762                 (setq elem (cons 'seq
 763                                  (mapcar 'xml-parse-elem-type
 764                                          (split-string elem ",")))))))
 765       (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string)
 766           (setq elem     (match-string-no-properties 1 string)
 767                 modifier (match-string-no-properties 2 string))))
 768
 769     (if (and (stringp elem) (string= elem "#PCDATA"))
 770         (setq elem 'pcdata))
 771
 772     (cond
 773      ((string= modifier "+")
 774       (list '+ elem))
 775      ((string= modifier "*")
 776       (list '* elem))
 777      ((string= modifier "?")
 778       (list '\? elem))
 779      (t
 780       elem))))
 781
 782 ;;*******************************************************************
 783 ;;**
 784 ;;**  Substituting special XML sequences
 785 ;;**
 786 ;;*******************************************************************
 787
 788 (defun xml-substitute-special (string)
 789   "Return STRING, after substituting entity references."
 790   ;; This originally made repeated passes through the string from the
 791   ;; beginning, which isn't correct, since then either "&amp;amp;" or
 792   ;; "&#38;amp;" won't DTRT.
 793
 794   (let ((point 0)
 795         children end-point)
 796     (while (string-match "&\\([^;]*\\);" string point)
 797       (setq end-point (match-end 0))
 798       (let* ((this-part (match-string-no-properties 1 string))
 799              (prev-part (substring string point (match-beginning 0)))
 800              (entity (assoc this-part xml-entity-alist))
 801              (expansion
 802               (cond ((string-match "#\\([0-9]+\\)" this-part)
 803                      (let ((c (decode-char
 804                                'ucs
 805                                (string-to-number (match-string-no-properties 1 this-part)))))
 806                        (if c (string c))))
 807                     ((string-match "#x\\([[:xdigit:]]+\\)" this-part)
 808                      (let ((c (decode-char
 809                                'ucs
 810                                (string-to-number (match-string-no-properties 1 this-part) 16))))
 811                        (if c (string c))))
 812                     (entity
 813                      (cdr entity))
 814                     ((eq (length this-part) 0)
 815                      (error "XML: (Not Well-Formed) No entity given"))
 816                     (t
 817                      (if xml-validating-parser
 818                          (error "XML: (Validity) Undefined entity `%s'"
 819                                 this-part)
 820                        xml-undefined-entity)))))
 821
 822         (cond ((null children)
 823                ;; FIXME: If we have an entity that expands into XML, this won't work.
 824                (setq children
 825                      (concat prev-part expansion)))
 826               ((stringp children)
 827                (if (stringp expansion)
 828                    (setq children (concat children prev-part expansion))
 829                  (setq children (list expansion (concat prev-part children)))))
 830               ((and (stringp expansion)
 831                     (stringp (car children)))
 832                (setcar children (concat prev-part expansion (car children))))
 833               ((stringp expansion)
 834                (setq children (append (concat prev-part expansion)
 835                                       children)))
 836               ((stringp (car children))
 837                (setcar children (concat (car children) prev-part))
 838                (setq children (append expansion children)))
 839               (t
 840                (setq children (list expansion
 841                                     prev-part
 842                                     children))))
 843         (setq point end-point)))
 844     (cond ((stringp children)
 845            (concat children (substring string point)))
 846           ((stringp (car (last children)))
 847            (concat (car (last children)) (substring string point)))
 848           ((null children)
 849            string)
 850           (t
 851            (concat (mapconcat 'identity
 852                               (nreverse children)
 853                               "")
 854                    (substring string point))))))
 855
 856 (defun xml-substitute-numeric-entities (string)
 857   "Substitute SGML numeric entities by their respective utf characters.
 858 This function replaces numeric entities in the input STRING and
 859 returns the modified string.  For example \"&#42;\" gets replaced
 860 by \"*\"."
 861   (if (and string (stringp string))
 862       (let ((start 0))
 863         (while (string-match "&#\\([0-9]+\\);" string start)
 864           (condition-case nil
 865               (setq string (replace-match
 866                             (string (read (substring string
 867                                                      (match-beginning 1)
 868                                                      (match-end 1))))
 869                             nil nil string))
 870             (error nil))
 871           (setq start (1+ (match-beginning 0))))
 872         string)
 873     nil))
 874
 875 ;;*******************************************************************
 876 ;;**
 877 ;;**  Printing a tree.
 878 ;;**  This function is intended mainly for debugging purposes.
 879 ;;**
 880 ;;*******************************************************************
 881
 882 (defun xml-debug-print (xml &optional indent-string)
 883   "Outputs the XML in the current buffer.
 884 XML can be a tree or a list of nodes.
 885 The first line is indented with the optional INDENT-STRING."
 886   (setq indent-string (or indent-string ""))
 887   (dolist (node xml)
 888     (xml-debug-print-internal node indent-string)))
 889
 890 (defalias 'xml-print 'xml-debug-print)
 891
 892 (defun xml-escape-string (string)
 893   "Return STRING with entity substitutions made from `xml-entity-alist'."
 894   (mapconcat (lambda (byte)
 895                (let ((char (char-to-string byte)))
 896                  (if (rassoc char xml-entity-alist)
 897                      (concat "&" (car (rassoc char xml-entity-alist)) ";")
 898                    char)))
 899              string ""))
 900
 901 (defun xml-debug-print-internal (xml indent-string)
 902   "Outputs the XML tree in the current buffer.
 903 The first line is indented with INDENT-STRING."
 904   (let ((tree xml)
 905         attlist)
 906     (insert indent-string ?< (symbol-name (xml-node-name tree)))
 907
 908     ;;  output the attribute list
 909     (setq attlist (xml-node-attributes tree))
 910     (while attlist
 911       (insert ?\  (symbol-name (caar attlist)) "=\""
 912               (xml-escape-string (cdar attlist)) ?\")
 913       (setq attlist (cdr attlist)))
 914
 915     (setq tree (xml-node-children tree))
 916
 917     (if (null tree)
 918         (insert ?/ ?>)
 919       (insert ?>)
 920
 921       ;;  output the children
 922       (dolist (node tree)
 923         (cond
 924          ((listp node)
 925           (insert ?\n)
 926           (xml-debug-print-internal node (concat indent-string "  ")))
 927          ((stringp node)
 928           (insert (xml-escape-string node)))
 929          (t
 930           (error "Invalid XML tree"))))
 931
 932       (when (not (and (null (cdr tree))
 933                       (stringp (car tree))))
 934         (insert ?\n indent-string))
 935       (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>))))
 936
 937 (provide 'xml)
 938
 939 ;;; xml.el ends here