lisp/xml.el

   1 ;;; xml.el --- XML parser
   2
   3 ;; Copyright (C) 2000, 2001, 2002, 2003, 2004,
   4 ;;   2005, 2006, 2007, 2008, 2009, 2010, 2011 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 3 of the License, or
  15 ;; (at your option) 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.  If not, see <http://www.gnu.org/licenses/>.
  24
  25 ;;; Commentary:
  26
  27 ;; This file contains a somewhat incomplete non-validating XML parser.  It
  28 ;; parses a file, and returns a list that can be used internally by
  29 ;; any other Lisp libraries.
  30
  31 ;;; FILE FORMAT
  32
  33 ;; The document type declaration may either be ignored or (optionally)
  34 ;; parsed, but currently the parsing will only accept element
  35 ;; declarations.  The XML file is assumed to be well-formed.  In case
  36 ;; of error, the parsing stops and the XML file is shown where the
  37 ;; parsing stopped.
  38 ;;
  39 ;; It also knows how to ignore comments and processing instructions.
  40 ;;
  41 ;; The XML file should have the following format:
  42 ;;    <node1 attr1="name1" attr2="name2" ...>value
  43 ;;       <node2 attr3="name3" attr4="name4">value2</node2>
  44 ;;       <node3 attr5="name5" attr6="name6">value3</node3>
  45 ;;    </node1>
  46 ;; Of course, the name of the nodes and attributes can be anything.  There can
  47 ;; be any number of attributes (or none), as well as any number of children
  48 ;; below the nodes.
  49 ;;
  50 ;; There can be only top level node, but with any number of children below.
  51
  52 ;;; LIST FORMAT
  53
  54 ;; The functions `xml-parse-file', `xml-parse-region' and
  55 ;; `xml-parse-tag' return a list with the following format:
  56 ;;
  57 ;;    xml-list   ::= (node node ...)
  58 ;;    node       ::= (qname attribute-list . child_node_list)
  59 ;;    child_node_list ::= child_node child_node ...
  60 ;;    child_node ::= node | string
  61 ;;    qname      ::= (:namespace-uri . "name") | "name"
  62 ;;    attribute_list ::= ((qname . "value") (qname . "value") ...)
  63 ;;                       | nil
  64 ;;    string     ::= "..."
  65 ;;
  66 ;; Some macros are provided to ease the parsing of this list.
  67 ;; Whitespace is preserved.  Fixme: There should be a tree-walker that
  68 ;; can remove it.
  69
  70 ;; TODO:
  71 ;;  * xml:base, xml:space support
  72 ;;  * more complete DOCTYPE parsing
  73 ;;  * pi support
  74
  75 ;;; Code:
  76
  77 ;; Note that buffer-substring and match-string were formerly used in
  78 ;; several places, because the -no-properties variants remove
  79 ;; composition info.  However, after some discussion on emacs-devel,
  80 ;; the consensus was that the speed of the -no-properties variants was
  81 ;; a worthwhile tradeoff especially since we're usually parsing files
  82 ;; instead of hand-crafted XML.
  83
  84 ;;*******************************************************************
  85 ;;**
  86 ;;**  Macros to parse the list
  87 ;;**
  88 ;;*******************************************************************
  89
  90 (defconst xml-undefined-entity "?"
  91   "What to substitute for undefined entities")
  92
  93 (defvar xml-entity-alist
  94   '(("lt"   . "<")
  95     ("gt"   . ">")
  96     ("apos" . "'")
  97     ("quot" . "\"")
  98     ("amp"  . "&"))
  99   "The defined entities.  Entities are added to this when the DTD is parsed.")
 100
 101 (defvar xml-sub-parser nil
 102   "Dynamically set this to a non-nil value if you want to parse an XML fragment.")
 103
 104 (defvar xml-validating-parser nil
 105   "Set to non-nil to get validity checking.")
 106
 107 (defsubst xml-node-name (node)
 108   "Return the tag associated with NODE.
 109 Without namespace-aware parsing, the tag is a symbol.
 110
 111 With namespace-aware parsing, the tag is a cons of a string
 112 representing the uri of the namespace with the local name of the
 113 tag.  For example,
 114
 115     <foo>
 116
 117 would be represented by
 118
 119     '(\"\" . \"foo\")."
 120
 121   (car node))
 122
 123 (defsubst xml-node-attributes (node)
 124   "Return the list of attributes of NODE.
 125 The list can be nil."
 126   (nth 1 node))
 127
 128 (defsubst xml-node-children (node)
 129   "Return the list of children of NODE.
 130 This is a list of nodes, and it can be nil."
 131   (cddr node))
 132
 133 (defun xml-get-children (node child-name)
 134   "Return the children of NODE whose tag is CHILD-NAME.
 135 CHILD-NAME should match the value returned by `xml-node-name'."
 136   (let ((match ()))
 137     (dolist (child (xml-node-children node))
 138       (if (and (listp child)
 139                (equal (xml-node-name child) child-name))
 140           (push child match)))
 141     (nreverse match)))
 142
 143 (defun xml-get-attribute-or-nil (node attribute)
 144   "Get from NODE the value of ATTRIBUTE.
 145 Return nil if the attribute was not found.
 146
 147 See also `xml-get-attribute'."
 148   (cdr (assoc attribute (xml-node-attributes node))))
 149
 150 (defsubst xml-get-attribute (node attribute)
 151   "Get from NODE the value of ATTRIBUTE.
 152 An empty string is returned if the attribute was not found.
 153
 154 See also `xml-get-attribute-or-nil'."
 155   (or (xml-get-attribute-or-nil node attribute) ""))
 156
 157 ;;*******************************************************************
 158 ;;**
 159 ;;**  Creating the list
 160 ;;**
 161 ;;*******************************************************************
 162
 163 ;;;###autoload
 164 (defun xml-parse-file (file &optional parse-dtd parse-ns)
 165   "Parse the well-formed XML file FILE.
 166 If FILE is already visited, use its buffer and don't kill it.
 167 Returns the top node with all its children.
 168 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped.
 169 If PARSE-NS is non-nil, then QNAMES are expanded."
 170   (if (get-file-buffer file)
 171       (with-current-buffer (get-file-buffer file)
 172         (save-excursion
 173           (xml-parse-region (point-min)
 174                             (point-max)
 175                             (current-buffer)
 176                             parse-dtd parse-ns)))
 177     (with-temp-buffer
 178       (insert-file-contents file)
 179       (xml-parse-region (point-min)
 180                         (point-max)
 181                         (current-buffer)
 182                         parse-dtd parse-ns))))
 183
 184
 185 (defvar xml-name-re)
 186 (defvar xml-entity-value-re)
 187 (defvar xml-att-def-re)
 188 (let* ((start-chars (concat "[:alpha:]:_"))
 189        (name-chars  (concat "-[:digit:]." start-chars))
 190        ;;[3]    S          ::=          (#x20 | #x9 | #xD | #xA)+
 191        (whitespace  "[ \t\n\r]"))
 192   ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6]
 193   ;;                      | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF]
 194   ;;                      | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF]
 195   ;;                      | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
 196   (defvar xml-name-start-char-re (concat "[" start-chars "]"))
 197   ;;[4a] NameChar       ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
 198   (defvar xml-name-char-re       (concat "[" name-chars  "]"))
 199   ;;[5] Name     ::= NameStartChar (NameChar)*
 200   (defvar xml-name-re            (concat xml-name-start-char-re xml-name-char-re "*"))
 201   ;;[6] Names    ::= Name (#x20 Name)*
 202   (defvar xml-names-re           (concat xml-name-re "\\(?: " xml-name-re "\\)*"))
 203   ;;[7] Nmtoken ::= (NameChar)+
 204   (defvar xml-nmtoken-re         (concat xml-name-char-re "+"))
 205   ;;[8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
 206   (defvar xml-nmtokens-re        (concat xml-nmtoken-re "\\(?: " xml-name-re "\\)*"))
 207   ;;[66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
 208   (defvar xml-char-ref-re        "\\(?:&#[0-9]+;\\|&#x[0-9a-fA-F]+;\\)")
 209   ;;[68] EntityRef   ::= '&' Name ';'
 210   (defvar xml-entity-ref         (concat "&" xml-name-re ";"))
 211   ;;[69] PEReference ::= '%' Name ';'
 212   (defvar xml-pe-reference-re    (concat "%" xml-name-re ";"))
 213   ;;[67] Reference   ::= EntityRef | CharRef
 214   (defvar xml-reference-re       (concat "\\(?:" xml-entity-ref "\\|" xml-char-ref-re "\\)"))
 215   ;;[10]        AttValue           ::=          '"' ([^<&"] | Reference)* '"' |  "'" ([^<&'] | Reference)* "'"
 216   (defvar xml-att-value-re    (concat "\\(?:\"\\(?:[^&\"]\\|" xml-reference-re "\\)*\"\\|"
 217                                       "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)"))
 218   ;;[56]        TokenizedType      ::=          'ID'       [VC: ID] [VC: One ID per Element Type] [VC: ID Attribute Default]
 219   ;;                                            | 'IDREF'    [VC: IDREF]
 220   ;;                                                  | 'IDREFS'   [VC: IDREF]
 221   ;;                                            | 'ENTITY'   [VC: Entity Name]
 222   ;;                                            | 'ENTITIES' [VC: Entity Name]
 223   ;;                                            | 'NMTOKEN'  [VC: Name Token]
 224   ;;                                            | 'NMTOKENS' [VC: Name Token]
 225   (defvar xml-tokenized-type-re "\\(?:ID\\|IDREF\\|IDREFS\\|ENTITY\\|ENTITIES\\|NMTOKEN\\|NMTOKENS\\)")
 226   ;;[58]        NotationType       ::=          'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
 227   (defvar xml-notation-type-re (concat "\\(?:NOTATION" whitespace "(" whitespace "*" xml-name-re
 228                                        "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)"))
 229   ;;[59]        Enumeration        ::=          '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')'      [VC: Enumeration] [VC: No Duplicate Tokens]
 230   (defvar xml-enumeration-re (concat "\\(?:(" whitespace "*" xml-nmtoken-re
 231                                      "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*"
 232                                      whitespace ")\\)"))
 233   ;;[57]        EnumeratedType     ::=          NotationType | Enumeration
 234   (defvar xml-enumerated-type-re (concat "\\(?:" xml-notation-type-re "\\|" xml-enumeration-re "\\)"))
 235   ;;[54]        AttType    ::=          StringType | TokenizedType | EnumeratedType
 236   ;;[55]        StringType         ::=          'CDATA'
 237   (defvar xml-att-type-re (concat "\\(?:CDATA\\|" xml-tokenized-type-re "\\|" xml-notation-type-re"\\|" xml-enumerated-type-re "\\)"))
 238   ;;[60]        DefaultDecl        ::=          '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
 239   (defvar xml-default-decl-re (concat "\\(?:#REQUIRED\\|#IMPLIED\\|\\(?:#FIXED" whitespace "\\)*" xml-att-value-re "\\)"))
 240   ;;[53]        AttDef     ::=          S Name S AttType S DefaultDecl
 241   (defvar xml-att-def-re         (concat "\\(?:" whitespace "*" xml-name-re
 242                                          whitespace "*" xml-att-type-re
 243                                          whitespace "*" xml-default-decl-re "\\)"))
 244   ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
 245   ;;               |  "'" ([^%&'] | PEReference | Reference)* "'"
 246   (defvar xml-entity-value-re    (concat "\\(?:\"\\(?:[^%&\"]\\|" xml-pe-reference-re
 247                                          "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|"
 248                                          xml-pe-reference-re "\\|" xml-reference-re "\\)*'\\)")))
 249 ;;[75] ExternalID ::= 'SYSTEM' S SystemLiteral
 250 ;;                 | 'PUBLIC' S PubidLiteral S SystemLiteral
 251 ;;[76] NDataDecl ::=    S 'NDATA' S
 252 ;;[73] EntityDef  ::= EntityValue| (ExternalID NDataDecl?)
 253 ;;[71] GEDecl     ::= '<!ENTITY' S Name S EntityDef S? '>'
 254 ;;[74] PEDef      ::= EntityValue | ExternalID
 255 ;;[72] PEDecl     ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
 256 ;;[70] EntityDecl ::= GEDecl | PEDecl
 257
 258 ;; Note that this is setup so that we can do whitespace-skipping with
 259 ;; `(skip-syntax-forward " ")', inter alia.  Previously this was slow
 260 ;; compared with `re-search-forward', but that has been fixed.  Also
 261 ;; note that the standard syntax table contains other characters with
 262 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
 263 ;; which we might skip whitespace -- specifically, they're not
 264 ;; NameChars [XML 4].
 265
 266 (defvar xml-syntax-table
 267   (let ((table (make-syntax-table)))
 268     ;; Get space syntax correct per XML [3].
 269     (dotimes (c 31)
 270       (modify-syntax-entry c "." table)) ; all are space in standard table
 271     (dolist (c '(?\t ?\n ?\r))           ; these should be space
 272       (modify-syntax-entry c " " table))
 273     ;; For skipping attributes.
 274     (modify-syntax-entry ?\" "\"" table)
 275     (modify-syntax-entry ?' "\"" table)
 276     ;; Non-alnum name chars should be symbol constituents (`-' and `_'
 277     ;; are OK by default).
 278     (modify-syntax-entry ?. "_" table)
 279     (modify-syntax-entry ?: "_" table)
 280     ;; XML [89]
 281     (unless (featurep 'xemacs)
 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   ;; Use fixed syntax table to ensure regexp char classes and syntax
 309   ;; specs DTRT.
 310   (with-syntax-table (standard-syntax-table)
 311     (let ((case-fold-search nil)        ; XML is case-sensitive.
 312           xml result dtd)
 313       (save-excursion
 314         (if buffer
 315             (set-buffer buffer))
 316         (save-restriction
 317           (narrow-to-region beg end)
 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                   (cond
 325                    ((null result)
 326                     ;; Not looking at an xml start tag.
 327                     (forward-char 1))
 328                    ((and xml (not xml-sub-parser))
 329                     ;; Translation of rule [1] of XML specifications
 330                     (error "XML: (Not Well-Formed) Only one root tag allowed"))
 331                    ((and (listp (car result))
 332                          parse-dtd)
 333                     (setq dtd (car result))
 334                     (if (cdr result)    ; possible leading comment
 335                         (add-to-list 'xml (cdr result))))
 336                    (t
 337                     (add-to-list 'xml result))))
 338               (goto-char (point-max))))
 339           (if parse-dtd
 340               (cons dtd (nreverse xml))
 341             (nreverse xml)))))))
 342
 343 (defun xml-maybe-do-ns (name default xml-ns)
 344   "Perform any namespace expansion.
 345 NAME is the name to perform the expansion on.
 346 DEFAULT is the default namespace.  XML-NS is a cons of namespace
 347 names to uris.  When namespace-aware parsing is off, then XML-NS
 348 is nil.
 349
 350 During namespace-aware parsing, any name without a namespace is
 351 put into the namespace identified by DEFAULT.  nil is used to
 352 specify that the name shouldn't be given a namespace."
 353   (if (consp xml-ns)
 354       (let* ((nsp (string-match ":" name))
 355              (lname (if nsp (substring name (match-end 0)) name))
 356              (prefix (if nsp (substring name 0 (match-beginning 0)) default))
 357              (special (and (string-equal lname "xmlns") (not prefix)))
 358              ;; Setting default to nil will insure that there is not
 359              ;; matching cons in xml-ns.  In which case we
 360              (ns (or (cdr (assoc (if special "xmlns" prefix)
 361                                  xml-ns))
 362                      "")))
 363         (cons ns (if special "" lname)))
 364     (intern name)))
 365
 366 (defun xml-parse-fragment (&optional parse-dtd parse-ns)
 367   "Parse xml-like fragments."
 368   (let ((xml-sub-parser t)
 369         children)
 370     (while (not (eobp))
 371       (let ((bit (xml-parse-tag
 372                   parse-dtd parse-ns)))
 373         (if children
 374             (setq children (append (list bit) children))
 375           (if (stringp bit)
 376               (setq children (list bit))
 377             (setq children bit)))))
 378     (reverse children)))
 379
 380 (defun xml-parse-tag (&optional parse-dtd parse-ns)
 381   "Parse the tag at point.
 382 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
 383 returned as the first element in the list.
 384 If PARSE-NS is non-nil, then QNAMES are expanded.
 385 Returns one of:
 386  - a list : the matching node
 387  - nil    : the point is not looking at a tag.
 388  - a pair : the first element is the DTD, the second is the node."
 389   (let ((xml-validating-parser (or parse-dtd xml-validating-parser))
 390         (xml-ns (if (consp parse-ns)
 391                     parse-ns
 392                   (if parse-ns
 393                       (list
 394                        ;; Default for empty prefix is no namespace
 395                        (cons ""      "")
 396                        ;; "xml" namespace
 397                        (cons "xml"   "http://www.w3.org/XML/1998/namespace")
 398                        ;; We need to seed the xmlns namespace
 399                        (cons "xmlns" "http://www.w3.org/2000/xmlns/"))))))
 400     (cond
 401      ;; Processing instructions (like the <?xml version="1.0"?> tag at the
 402      ;; beginning of a document).
 403      ((looking-at "<\\?")
 404       (search-forward "?>")
 405       (skip-syntax-forward " ")
 406       (xml-parse-tag parse-dtd xml-ns))
 407      ;;  Character data (CDATA) sections, in which no tag should be interpreted
 408      ((looking-at "<!\\[CDATA\\[")
 409       (let ((pos (match-end 0)))
 410         (unless (search-forward "]]>" nil t)
 411           (error "XML: (Not Well Formed) CDATA section does not end anywhere in the document"))
 412         (concat
 413          (buffer-substring-no-properties pos (match-beginning 0))
 414          (xml-parse-string))))
 415      ;;  DTD for the document
 416      ((looking-at "<!DOCTYPE")
 417       (let ((dtd (xml-parse-dtd parse-ns)))
 418         (skip-syntax-forward " ")
 419         (if xml-validating-parser
 420             (cons dtd (xml-parse-tag nil xml-ns))
 421           (xml-parse-tag nil xml-ns))))
 422      ;;  skip comments
 423      ((looking-at "<!--")
 424       (search-forward "-->")
 425       nil)
 426      ;;  end tag
 427      ((looking-at "</")
 428       '())
 429      ;;  opening tag
 430      ((looking-at "<\\([^/>[:space:]]+\\)")
 431       (goto-char (match-end 1))
 432
 433       ;; Parse this node
 434       (let* ((node-name (match-string-no-properties 1))
 435              ;; Parse the attribute list.
 436              (attrs (xml-parse-attlist xml-ns))
 437              children pos)
 438
 439         ;; add the xmlns:* attrs to our cache
 440         (when (consp xml-ns)
 441           (dolist (attr attrs)
 442             (when (and (consp (car attr))
 443                        (equal "http://www.w3.org/2000/xmlns/"
 444                               (caar attr)))
 445               (push (cons (cdar attr) (cdr attr))
 446                     xml-ns))))
 447
 448         (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns)))
 449
 450         ;; is this an empty element ?
 451         (if (looking-at "/>")
 452             (progn
 453               (forward-char 2)
 454               (nreverse children))
 455
 456           ;; is this a valid start tag ?
 457           (if (eq (char-after) ?>)
 458               (progn
 459                 (forward-char 1)
 460                 ;;  Now check that we have the right end-tag. Note that this
 461                 ;;  one might contain spaces after the tag name
 462                 (let ((end (concat "</" node-name "\\s-*>")))
 463                   (while (not (looking-at end))
 464                     (cond
 465                      ((looking-at "</")
 466                       (error "XML: (Not Well-Formed) Invalid end tag (expecting %s) at pos %d"
 467                              node-name (point)))
 468                      ((= (char-after) ?<)
 469                       (let ((tag (xml-parse-tag nil xml-ns)))
 470                         (when tag
 471                           (push tag children))))
 472                      (t
 473                       (let ((expansion (xml-parse-string)))
 474                         (setq children
 475                               (if (stringp expansion)
 476                                   (if (stringp (car children))
 477                                       ;; The two strings were separated by a comment.
 478                                       (setq children (append (list (concat (car children) expansion))
 479                                                              (cdr children)))
 480                                     (setq children (append (list expansion) children)))
 481                                 (setq children (append expansion children))))))))
 482
 483                   (goto-char (match-end 0))
 484                   (nreverse children)))
 485             ;;  This was an invalid start tag (Expected ">", but didn't see it.)
 486             (error "XML: (Well-Formed) Couldn't parse tag: %s"
 487                    (buffer-substring-no-properties (- (point) 10) (+ (point) 1)))))))
 488      (t ;; (Not one of PI, CDATA, Comment, End tag, or Start tag)
 489       (unless xml-sub-parser            ; Usually, we error out.
 490         (error "XML: (Well-Formed) Invalid character"))
 491
 492       ;; However, if we're parsing incrementally, then we need to deal
 493       ;; with stray CDATA.
 494       (xml-parse-string)))))
 495
 496 (defun xml-parse-string ()
 497   "Parse the next whatever.  Could be a string, or an element."
 498   (let* ((pos (point))
 499          (string (progn (skip-chars-forward "^<")
 500                         (buffer-substring-no-properties 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-no-properties 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-no-properties 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-no-properties 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-no-properties 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-no-properties 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-no-properties 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-no-properties 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]*$" type) ;; children ([47])
 634               (setq type (xml-parse-elem-type (match-string-no-properties 1 type))))
 635              ((string-match "^%[^;]+;[ \t\n\r]*$" 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-no-properties 1))
 665                   (value (substring (match-string-no-properties 2) 1
 666                                     (- (length (match-string-no-properties 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-no-properties 1))
 687                   (file  (substring (match-string-no-properties 2) 1
 688                                     (- (length (match-string-no-properties 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-no-properties 1 string)
 728                 modifier (match-string-no-properties 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-no-properties 1 string)
 739                 modifier (match-string-no-properties 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-no-properties 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-no-properties 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-no-properties 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 (defun xml-substitute-numeric-entities (string)
 829   "Substitute SGML numeric entities by their respective utf characters.
 830 This function replaces numeric entities in the input STRING and
 831 returns the modified string.  For example \"&#42;\" gets replaced
 832 by \"*\"."
 833   (if (and string (stringp string))
 834       (let ((start 0))
 835         (while (string-match "&#\\([0-9]+\\);" string start)
 836           (condition-case nil
 837               (setq string (replace-match
 838                             (string (read (substring string
 839                                                      (match-beginning 1)
 840                                                      (match-end 1))))
 841                             nil nil string))
 842             (error nil))
 843           (setq start (1+ (match-beginning 0))))
 844         string)
 845     nil))
 846
 847 ;;*******************************************************************
 848 ;;**
 849 ;;**  Printing a tree.
 850 ;;**  This function is intended mainly for debugging purposes.
 851 ;;**
 852 ;;*******************************************************************
 853
 854 (defun xml-debug-print (xml &optional indent-string)
 855   "Outputs the XML in the current buffer.
 856 XML can be a tree or a list of nodes.
 857 The first line is indented with the optional INDENT-STRING."
 858   (setq indent-string (or indent-string ""))
 859   (dolist (node xml)
 860     (xml-debug-print-internal node indent-string)))
 861
 862 (defalias 'xml-print 'xml-debug-print)
 863
 864 (defun xml-escape-string (string)
 865   "Return the string with entity substitutions made from
 866 xml-entity-alist."
 867   (mapconcat (lambda (byte)
 868                (let ((char (char-to-string byte)))
 869                  (if (rassoc char xml-entity-alist)
 870                      (concat "&" (car (rassoc char xml-entity-alist)) ";")
 871                    char)))
 872              ;; This differs from the non-unicode branch.  Just
 873              ;; grabbing the string works here.
 874              string ""))
 875
 876 (defun xml-debug-print-internal (xml indent-string)
 877   "Outputs the XML tree in the current buffer.
 878 The first line is indented with INDENT-STRING."
 879   (let ((tree xml)
 880         attlist)
 881     (insert indent-string ?< (symbol-name (xml-node-name tree)))
 882
 883     ;;  output the attribute list
 884     (setq attlist (xml-node-attributes tree))
 885     (while attlist
 886       (insert ?\  (symbol-name (caar attlist)) "=\""
 887               (xml-escape-string (cdar attlist)) ?\")
 888       (setq attlist (cdr attlist)))
 889
 890     (setq tree (xml-node-children tree))
 891
 892     (if (null tree)
 893         (insert ?/ ?>)
 894       (insert ?>)
 895
 896       ;;  output the children
 897       (dolist (node tree)
 898         (cond
 899          ((listp node)
 900           (insert ?\n)
 901           (xml-debug-print-internal node (concat indent-string "  ")))
 902          ((stringp node)
 903           (insert (xml-escape-string node)))
 904          (t
 905           (error "Invalid XML tree"))))
 906
 907       (when (not (and (null (cdr tree))
 908                       (stringp (car tree))))
 909         (insert ?\n indent-string))
 910       (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>))))
 911
 912 (provide 'xml)
 913
 914 ;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b
 915 ;;; xml.el ends here