lisp/xml.el

   1 ;;; xml.el --- XML parser
   2
   3 ;; Copyright (C) 2000, 2001, 2003 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 2, or (at your option)
  14 ;; 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; see the file COPYING.  If not, write to the
  23 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  24 ;; Boston, MA 02111-1307, USA.
  25
  26 ;;; Commentary:
  27
  28 ;; This file contains a somewhat incomplete non-validating XML parser.  It
  29 ;; parses a file, and returns a list that can be used internally by
  30 ;; any other lisp libraries.
  31
  32 ;;; FILE FORMAT
  33
  34 ;; The document type declaration may either be ignored or (optionally)
  35 ;; parsed, but currently the parsing will only accept element
  36 ;; declarations.  The XML file is assumed to be well-formed. In case
  37 ;; of error, the parsing stops and the XML file is shown where the
  38 ;; parsing stopped.
  39 ;;
  40 ;; It also knows how to ignore comments and processing instructions.
  41 ;;
  42 ;; The XML file should have the following format:
  43 ;;    <node1 attr1="name1" attr2="name2" ...>value
  44 ;;       <node2 attr3="name3" attr4="name4">value2</node2>
  45 ;;       <node3 attr5="name5" attr6="name6">value3</node3>
  46 ;;    </node1>
  47 ;; Of course, the name of the nodes and attributes can be anything. There can
  48 ;; be any number of attributes (or none), as well as any number of children
  49 ;; below the nodes.
  50 ;;
  51 ;; There can be only top level node, but with any number of children below.
  52
  53 ;;; LIST FORMAT
  54
  55 ;; The functions `xml-parse-file' and `xml-parse-tag' return a list with
  56 ;; the following format:
  57 ;;
  58 ;;    xml-list   ::= (node node ...)
  59 ;;    node       ::= (tag_name attribute-list . child_node_list)
  60 ;;    child_node_list ::= child_node child_node ...
  61 ;;    child_node ::= node | string
  62 ;;    tag_name   ::= string
  63 ;;    attribute_list ::= (("attribute" . "value") ("attribute" . "value") ...)
  64 ;;                       | nil
  65 ;;    string     ::= "..."
  66 ;;
  67 ;; Some macros are provided to ease the parsing of this list.
  68 ;; Whitespace is preserved.  Fixme: There should be a tree-walker that
  69 ;; can remove it.
  70
  71 ;;; Code:
  72
  73 ;; Note that {buffer-substring,match-string}-no-properties were
  74 ;; formerly used in several places, but that removes composition info.
  75
  76 ;;*******************************************************************
  77 ;;**
  78 ;;**  Macros to parse the list
  79 ;;**
  80 ;;*******************************************************************
  81
  82 (defsubst xml-node-name (node)
  83   "Return the tag associated with NODE.
  84 The tag is a lower-case symbol."
  85   (car node))
  86
  87 (defsubst xml-node-attributes (node)
  88   "Return the list of attributes of NODE.
  89 The list can be nil."
  90   (nth 1 node))
  91
  92 (defsubst xml-node-children (node)
  93   "Return the list of children of NODE.
  94 This is a list of nodes, and it can be nil."
  95   (cddr node))
  96
  97 (defun xml-get-children (node child-name)
  98   "Return the children of NODE whose tag is CHILD-NAME.
  99 CHILD-NAME should be a lower case symbol."
 100   (let ((match ()))
 101     (dolist (child (xml-node-children node))
 102       (if child
 103           (if (equal (xml-node-name child) child-name)
 104               (push child match))))
 105     (nreverse match)))
 106
 107 (defun xml-get-attribute (node attribute)
 108   "Get from NODE the value of ATTRIBUTE.
 109 An empty string is returned if the attribute was not found."
 110   (if (xml-node-attributes node)
 111       (let ((value (assoc attribute (xml-node-attributes node))))
 112         (if value
 113             (cdr value)
 114           ""))
 115     ""))
 116
 117 ;;*******************************************************************
 118 ;;**
 119 ;;**  Creating the list
 120 ;;**
 121 ;;*******************************************************************
 122
 123 ;;;###autoload
 124 (defun xml-parse-file (file &optional parse-dtd)
 125   "Parse the well-formed XML file FILE.
 126 If FILE is already visited, use its buffer and don't kill it.
 127 Returns the top node with all its children.
 128 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped."
 129   (let ((keep))
 130     (if (get-file-buffer file)
 131         (progn
 132           (set-buffer (get-file-buffer file))
 133           (setq keep (point)))
 134       (let (auto-mode-alist)            ; no need for xml-mode
 135         (find-file file)))
 136
 137     (let ((xml (xml-parse-region (point-min)
 138                                  (point-max)
 139                                  (current-buffer)
 140                                  parse-dtd)))
 141       (if keep
 142           (goto-char keep)
 143         (kill-buffer (current-buffer)))
 144       xml)))
 145
 146 ;; Note that this is setup so that we can do whitespace-skipping with
 147 ;; `(skip-syntax-forward " ")', inter alia.  Previously this was slow
 148 ;; compared with `re-search-forward', but that has been fixed.  Also
 149 ;; note that the standard syntax table contains other characters with
 150 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
 151 ;; which we might skip whitespace -- specifically, they're not
 152 ;; NameChars [XML 4].
 153
 154 (defvar xml-syntax-table
 155   (let ((table (make-syntax-table)))
 156     ;; Get space syntax correct per XML [3].
 157     (dotimes (c 31)
 158       (modify-syntax-entry c "." table)) ; all are space in standard table
 159     (dolist (c '(?\t ?\n ?\r))          ; these should be space
 160       (modify-syntax-entry c " " table))
 161     ;; For skipping attributes.
 162     (modify-syntax-entry ?\" "\"" table)
 163     (modify-syntax-entry ?' "\"" table)
 164     ;; Non-alnum name chars should be symbol constituents (`-' and `_'
 165     ;; are OK by default).
 166     (modify-syntax-entry ?. "_" table)
 167     (modify-syntax-entry ?: "_" table)
 168     ;; XML [89]
 169     (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005
 170                  #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC
 171                  #x30FD #x30FE))
 172       (modify-syntax-entry (decode-char 'ucs c) "w" table))
 173     ;; Fixme: rest of [4]
 174     table)
 175   "Syntax table used by `xml-parse-region'.")
 176
 177 ;; XML [5]
 178 ;; Note that [:alpha:] matches all multibyte chars with word syntax.
 179 (eval-and-compile
 180   (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*"))
 181
 182 ;; Fixme:  This needs re-writing to deal with the XML grammar properly, i.e.
 183 ;;   document    ::=    prolog element Misc*
 184 ;;   prolog    ::=    XMLDecl? Misc* (doctypedecl Misc*)?
 185
 186 ;;;###autoload
 187 (defun xml-parse-region (beg end &optional buffer parse-dtd)
 188   "Parse the region from BEG to END in BUFFER.
 189 If BUFFER is nil, it defaults to the current buffer.
 190 Returns the XML list for the region, or raises an error if the region
 191 is not a well-formed XML file.
 192 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped,
 193 and returned as the first element of the list."
 194   (save-restriction
 195     (narrow-to-region beg end)
 196     ;; Use fixed syntax table to ensure regexp char classes and syntax
 197     ;; specs DTRT.
 198     (with-syntax-table (standard-syntax-table)
 199       (let ((case-fold-search nil)      ; XML is case-sensitive.
 200             xml result dtd)
 201         (save-excursion
 202           (if buffer
 203               (set-buffer buffer))
 204           (goto-char (point-min))
 205           (while (not (eobp))
 206             (if (search-forward "<" nil t)
 207                 (progn
 208                   (forward-char -1)
 209                   (if xml
 210                       ;;  translation of rule [1] of XML specifications
 211                       (error "XML files can have only one toplevel tag")
 212                     (setq result (xml-parse-tag parse-dtd))
 213                     (cond
 214                      ((null result))
 215                      ((listp (car result))
 216                       (setq dtd (car result))
 217                       (if (cdr result)  ; possible leading comment
 218                           (add-to-list 'xml (cdr result))))
 219                      (t
 220                       (add-to-list 'xml result)))))
 221               (goto-char (point-max))))
 222           (if parse-dtd
 223               (cons dtd (nreverse xml))
 224             (nreverse xml)))))))
 225
 226
 227 (defun xml-parse-tag (&optional parse-dtd)
 228   "Parse the tag at point.
 229 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
 230 returned as the first element in the list.
 231 Returns one of:
 232  - a list : the matching node
 233  - nil    : the point is not looking at a tag.
 234  - a pair : the first element is the DTD, the second is the node."
 235   (cond
 236    ;; Processing instructions (like the <?xml version="1.0"?> tag at the
 237    ;; beginning of a document).
 238    ((looking-at "<\\?")
 239     (search-forward "?>")
 240     (skip-syntax-forward " ")
 241     (xml-parse-tag parse-dtd))
 242    ;;  Character data (CDATA) sections, in which no tag should be interpreted
 243    ((looking-at "<!\\[CDATA\\[")
 244     (let ((pos (match-end 0)))
 245       (unless (search-forward "]]>" nil t)
 246         (error "CDATA section does not end anywhere in the document"))
 247       (buffer-substring pos (match-beginning 0))))
 248    ;;  DTD for the document
 249    ((looking-at "<!DOCTYPE")
 250     (let (dtd)
 251       (if parse-dtd
 252           (setq dtd (xml-parse-dtd))
 253         (xml-skip-dtd))
 254       (skip-syntax-forward " ")
 255       (if dtd
 256           (cons dtd (xml-parse-tag))
 257         (xml-parse-tag))))
 258    ;;  skip comments
 259    ((looking-at "<!--")
 260     (search-forward "-->")
 261     nil)
 262    ;;  end tag
 263    ((looking-at "</")
 264     '())
 265    ;;  opening tag
 266    ((looking-at "<\\([^/>[:space:]]+\\)")
 267     (goto-char (match-end 1))
 268     (let* ((node-name (match-string 1))
 269            ;; Parse the attribute list.
 270            (children (list (xml-parse-attlist) (intern node-name)))
 271            pos)
 272
 273       ;; is this an empty element ?
 274       (if (looking-at "/>")
 275           (progn
 276             (forward-char 2)
 277             (nreverse children))
 278
 279         ;; is this a valid start tag ?
 280         (if (eq (char-after) ?>)
 281             (progn
 282               (forward-char 1)
 283               ;;  Now check that we have the right end-tag. Note that this
 284               ;;  one might contain spaces after the tag name
 285               (let ((end (concat "</" node-name "\\s-*>")))
 286                 (while (not (looking-at end))
 287                   (cond
 288                    ((looking-at "</")
 289                     (error "XML: Invalid end tag (expecting %s) at pos %d"
 290                            node-name (point)))
 291                    ((= (char-after) ?<)
 292                     (let ((tag (xml-parse-tag)))
 293                       (when tag
 294                         (push tag children))))
 295                    (t
 296                     (setq pos (point))
 297                     (search-forward "<")
 298                     (forward-char -1)
 299                     (let ((string (buffer-substring pos (point)))
 300                           (pos 0))
 301
 302                       ;; Clean up the string.  As per XML
 303                       ;; specifications, the XML processor should
 304                       ;; always pass the whole string to the
 305                       ;; application.  But \r's should be replaced:
 306                       ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends
 307                       (while (string-match "\r\n?" string pos)
 308                         (setq string (replace-match "\n" t t string))
 309                         (setq pos (1+ (match-beginning 0))))
 310
 311                       (setq string (xml-substitute-special string))
 312                       (setq children
 313                             (if (stringp (car children))
 314                                 ;; The two strings were separated by a comment.
 315                                 (cons (concat (car children) string)
 316                                       (cdr children))
 317                               (cons string children))))))))
 318
 319               (goto-char (match-end 0))
 320               (nreverse children))
 321           ;;  This was an invalid start tag
 322           (error "XML: Invalid attribute list")))))
 323    (t ;; This is not a tag.
 324     (error "XML: Invalid character"))))
 325
 326 (defun xml-parse-attlist ()
 327   "Return the attribute-list after point.
 328 Leave point at the first non-blank character after the tag."
 329   (let ((attlist ())
 330         start-pos name)
 331     (skip-syntax-forward " ")
 332     (while (looking-at (eval-when-compile
 333                          (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*")))
 334       (setq name (intern (match-string 1)))
 335       (goto-char (match-end 0))
 336
 337       ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize
 338
 339       ;; Do we have a string between quotes (or double-quotes),
 340       ;;  or a simple word ?
 341       (if (looking-at "\"\\([^\"]*\\)\"")
 342           (setq start-pos (match-beginning 0))
 343         (if (looking-at "'\\([^']*\\)'")
 344             (setq start-pos (match-beginning 0))
 345           (error "XML: Attribute values must be given between quotes")))
 346
 347       ;; Each attribute must be unique within a given element
 348       (if (assoc name attlist)
 349           (error "XML: each attribute must be unique within an element"))
 350
 351       ;; Multiple whitespace characters should be replaced with a single one
 352       ;; in the attributes
 353       (let ((string (match-string 1))
 354             (pos 0))
 355         (replace-regexp-in-string "\\s-\\{2,\\}" " " string)
 356         (push (cons name (xml-substitute-special string)) attlist))
 357
 358       (goto-char start-pos)
 359       (forward-sexp)                    ; we have string syntax
 360
 361       (skip-syntax-forward " "))
 362     (nreverse attlist)))
 363
 364 ;;*******************************************************************
 365 ;;**
 366 ;;**  The DTD (document type declaration)
 367 ;;**  The following functions know how to skip or parse the DTD of
 368 ;;**  a document
 369 ;;**
 370 ;;*******************************************************************
 371
 372 ;; Fixme: This fails at least if the DTD contains conditional sections.
 373
 374 (defun xml-skip-dtd ()
 375   "Skip the DTD at point.
 376 This follows the rule [28] in the XML specifications."
 377   (forward-char (length "<!DOCTYPE"))
 378   (if (looking-at "\\s-*>")
 379       (error "XML: invalid DTD (excepting name of the document)"))
 380   (condition-case nil
 381       (progn
 382         (forward-sexp)
 383         (skip-syntax-forward " ")
 384         (if (looking-at "\\[")
 385             (re-search-forward "]\\s-*>")
 386           (search-forward ">")))
 387     (error (error "XML: No end to the DTD"))))
 388
 389 (defun xml-parse-dtd ()
 390   "Parse the DTD at point."
 391   (forward-char (eval-when-compile (length "<!DOCTYPE")))
 392   (skip-syntax-forward " ")
 393   (if (looking-at ">")
 394       (error "XML: invalid DTD (excepting name of the document)"))
 395
 396   ;;  Get the name of the document
 397   (looking-at xml-name-regexp)
 398   (let ((dtd (list (match-string 0) 'dtd))
 399         type element end-pos)
 400     (goto-char (match-end 0))
 401
 402     (skip-syntax-forward " ")
 403     ;; XML [75]
 404     (cond ((looking-at "PUBLIC\\s-+")
 405            (goto-char (match-end 0))
 406            (unless (or (re-search-forward
 407                         "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\""
 408                         nil t)
 409                        (re-search-forward
 410                         "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'"
 411                         nil t))
 412              (error "XML: missing public id"))
 413            (let ((pubid (match-string 1)))
 414              (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
 415                          (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
 416                (error "XML: missing system id"))
 417              (push (list pubid (match-string 1) 'public) dtd)))
 418           ((looking-at "SYSTEM\\s-+")
 419            (goto-char (match-end 0))
 420            (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
 421                        (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
 422              (error "XML: missing system id"))
 423            (push (list (match-string 1) 'system) dtd)))
 424     (skip-syntax-forward " ")
 425     (if (eq ?> (char-after))
 426         (forward-char)
 427       (skip-syntax-forward " ")
 428       (if (not (eq (char-after) ?\[))
 429           (error "XML: bad DTD")
 430         (forward-char)
 431         ;;  Parse the rest of the DTD
 432         ;;  Fixme: Deal with ENTITY, ATTLIST, NOTATION, PIs.
 433         (while (not (looking-at "\\s-*\\]"))
 434           (skip-syntax-forward " ")
 435           (cond
 436
 437            ;;  Translation of rule [45] of XML specifications
 438            ((looking-at
 439              "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>")
 440
 441             (setq element (intern (match-string 1))
 442                   type    (match-string-no-properties 2))
 443             (setq end-pos (match-end 0))
 444
 445             ;;  Translation of rule [46] of XML specifications
 446             (cond
 447              ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration
 448               (setq type 'empty))
 449              ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents
 450               (setq type 'any))
 451              ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47])
 452               (setq type (xml-parse-elem-type (match-string 1 type))))
 453              ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution
 454               nil)
 455              (t
 456               (error "XML: Invalid element type in the DTD")))
 457
 458             ;;  rule [45]: the element declaration must be unique
 459             (if (assoc element dtd)
 460                 (error "XML: element declarations must be unique in a DTD (<%s>)"
 461                        (symbol-name element)))
 462
 463             ;;  Store the element in the DTD
 464             (push (list element type) dtd)
 465             (goto-char end-pos))
 466            ((looking-at "<!--")
 467             (search-forward "-->"))
 468
 469            (t
 470             (error "XML: Invalid DTD item")))
 471
 472           ;;  Skip the end of the DTD
 473           (search-forward ">"))))
 474     (nreverse dtd)))
 475
 476
 477 (defun xml-parse-elem-type (string)
 478   "Convert element type STRING into a Lisp structure."
 479
 480   (let (elem modifier)
 481     (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
 482         (progn
 483           (setq elem     (match-string 1 string)
 484                 modifier (match-string 2 string))
 485           (if (string-match "|" elem)
 486               (setq elem (cons 'choice
 487                                (mapcar 'xml-parse-elem-type
 488                                        (split-string elem "|"))))
 489             (if (string-match "," elem)
 490                 (setq elem (cons 'seq
 491                                  (mapcar 'xml-parse-elem-type
 492                                          (split-string elem ",")))))))
 493       (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string)
 494           (setq elem     (match-string 1 string)
 495                 modifier (match-string 2 string))))
 496
 497     (if (and (stringp elem) (string= elem "#PCDATA"))
 498         (setq elem 'pcdata))
 499
 500     (cond
 501      ((string= modifier "+")
 502       (list '+ elem))
 503      ((string= modifier "*")
 504       (list '* elem))
 505      ((string= modifier "?")
 506       (list '\? elem))
 507      (t
 508       elem))))
 509
 510 ;;*******************************************************************
 511 ;;**
 512 ;;**  Substituting special XML sequences
 513 ;;**
 514 ;;*******************************************************************
 515
 516 (eval-when-compile
 517   (defvar str))                ; dynamic from replace-regexp-in-string
 518
 519 ;; Fixme:  Take declared entities from the DTD when they're available.
 520 (defun xml-substitute-entity (match)
 521   "Subroutine of xml-substitute-special."
 522   (save-match-data
 523     (let ((match1 (match-string 1 str)))
 524       (cond ((string= match1 "lt") "<")
 525             ((string= match1 "gt") ">")
 526             ((string= match1 "apos") "'")
 527             ((string= match1 "quot") "\"")
 528             ((string= match1 "amp") "&")
 529             ((and (string-match "#\\([0-9]+\\)" match1)
 530                   (let ((c (decode-char
 531                             'ucs
 532                             (string-to-number (match-string 1 match1)))))
 533                     (if c (string c))))) ; else unrepresentable
 534             ((and (string-match "#x\\([[:xdigit:]]+\\)" match1)
 535                   (let ((c (decode-char
 536                             'ucs
 537                             (string-to-number (match-string 1 match1) 16))))
 538                     (if c (string c)))))
 539             ;; Default to asis.  Arguably, unrepresentable code points
 540             ;; might be best replaced with U+FFFD.
 541             (t match)))))
 542
 543 (defun xml-substitute-special (string)
 544   "Return STRING, after subsituting entity references."
 545   ;; This originally made repeated passes through the string from the
 546   ;; beginning, which isn't correct, since then either "&amp;amp;" or
 547   ;; "&#38;amp;" won't DTRT.
 548   (replace-regexp-in-string "&\\([^;]+\\);"
 549                             #'xml-substitute-entity string t t))
 550
 551 ;;*******************************************************************
 552 ;;**
 553 ;;**  Printing a tree.
 554 ;;**  This function is intended mainly for debugging purposes.
 555 ;;**
 556 ;;*******************************************************************
 557
 558 (defun xml-debug-print (xml)
 559   (dolist (node xml)
 560     (xml-debug-print-internal node "")))
 561
 562 (defun xml-debug-print-internal (xml indent-string)
 563   "Outputs the XML tree in the current buffer.
 564 The first line is indented with INDENT-STRING."
 565   (let ((tree xml)
 566         attlist)
 567     (insert indent-string ?< (symbol-name (xml-node-name tree)))
 568
 569     ;;  output the attribute list
 570     (setq attlist (xml-node-attributes tree))
 571     (while attlist
 572       (insert ?\  (symbol-name (caar attlist)) "=\"" (cdar attlist) ?\")
 573       (setq attlist (cdr attlist)))
 574
 575     (insert ?>)
 576
 577     (setq tree (xml-node-children tree))
 578
 579     ;;  output the children
 580     (dolist (node tree)
 581       (cond
 582        ((listp node)
 583         (insert ?\n)
 584         (xml-debug-print-internal node (concat indent-string "  ")))
 585        ((stringp node) (insert node))
 586        (t
 587         (error "Invalid XML tree"))))
 588
 589     (insert ?\n indent-string
 590             ?< ?/ (symbol-name (xml-node-name xml)) ?>)))
 591
 592 (provide 'xml)
 593
 594 ;;; xml.el ends here