lisp/xml.el

   1 ;; @(#) xml.el --- XML parser
   2
   3 ;; Copyright (C) 2000 Free Software Foundation, Inc.
   4
   5 ;; Author: Emmanuel Briot  <briot@gnat.com>
   6 ;; Maintainer: Emmanuel Briot <briot@gnat.com>
   7 ;; Keywords: xml
   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 full XML parser. It parses a file, and returns a list
  29 ;; that can be used internally by any other lisp file.
  30 ;; See some example in todo.el
  31
  32 ;;; FILE FORMAT
  33
  34 ;; It does not parse the DTD, if present in the XML file, but knows how to
  35 ;; ignore it. The XML file is assumed to be well-formed. In case of error, the
  36 ;; parsing stops and the XML file is shown where the parsing stopped.
  37 ;;
  38 ;; It also knows how to ignore comments, as well as the special ?xml? tag
  39 ;; in the XML file.
  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' and `xml-parse-tag' return a list with
  55 ;; the following format:
  56 ;;
  57 ;;    xml-list   ::= (node node ...)
  58 ;;    node       ::= (tag_name attribute-list . child_node_list)
  59 ;;    child_node_list ::= child_node child_node ...
  60 ;;    child_node ::= node | string
  61 ;;    tag_name   ::= string
  62 ;;    attribute_list ::= (("attribute" . "value") ("attribute" . "value") ...)
  63 ;;                       | nil
  64 ;;    string     ::= "..."
  65 ;;
  66 ;; Some macros are provided to ease the parsing of this list
  67
  68 ;;; Code:
  69
  70 ;;*******************************************************************
  71 ;;**
  72 ;;**  Macros to parse the list
  73 ;;**
  74 ;;*******************************************************************
  75
  76 (defmacro xml-node-name       (node)
  77   "Return the tag associated with NODE.
  78 The tag is a lower-case symbol."
  79   (list 'car node))
  80
  81 (defmacro xml-node-attributes (node)
  82   "Return the list of attributes of NODE.
  83 The list can be nil."
  84   (list 'nth 1 node))
  85
  86 (defmacro xml-node-children   (node)
  87   "Return the list of children of NODE.
  88 This is a list of nodes, and it can be nil."
  89   (list 'cddr node))
  90
  91 (defun xml-get-children (node child-name)
  92   "Return the children of NODE whose tag is CHILD-NAME.
  93 CHILD-NAME should be a lower case symbol."
  94   (let ((children (xml-node-children node))
  95         match)
  96     (while children
  97       (if (car children)
  98           (if (equal (xml-node-name (car children)) child-name)
  99               (set 'match (append match (list (car children))))))
 100       (set 'children (cdr children)))
 101     match))
 102
 103 (defun xml-get-attribute (node attribute)
 104   "Get from NODE the value of ATTRIBUTE.
 105 An empty string is returned if the attribute was not found."
 106   (if (xml-node-attributes node)
 107       (let ((value (assoc attribute (xml-node-attributes node))))
 108         (if value
 109             (cdr value)
 110           ""))
 111     ""))
 112
 113 ;;*******************************************************************
 114 ;;**
 115 ;;**  Creating the list
 116 ;;**
 117 ;;*******************************************************************
 118
 119 (defun xml-parse-file (file &optional parse-dtd)
 120   "Parse the well-formed XML FILE.
 121 If FILE is already edited, this will keep the buffer alive.
 122 Returns the top node with all its children.
 123 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped."
 124   (let ((keep))
 125     (if (get-file-buffer file)
 126         (progn
 127           (set-buffer (get-file-buffer file))
 128           (setq keep (point)))
 129       (find-file file))
 130
 131     (let ((xml (xml-parse-region (point-min)
 132                                  (point-max)
 133                                  (current-buffer)
 134                                  parse-dtd)))
 135       (if keep
 136           (goto-char keep)
 137         (kill-buffer (current-buffer)))
 138       xml)))
 139
 140 (defun xml-parse-region (beg end &optional buffer parse-dtd)
 141   "Parse the region from BEG to END in BUFFER.
 142 If BUFFER is nil, it defaults to the current buffer.
 143 Returns the XML list for the region, or raises an error if the region
 144 is not a well-formed XML file.
 145 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped,
 146 and returned as the first element of the list"
 147   (let (xml result dtd)
 148     (save-excursion
 149       (if buffer
 150           (set-buffer buffer))
 151       (goto-char beg)
 152       (while (< (point) end)
 153         (if (search-forward "<" end t)
 154             (progn
 155               (forward-char -1)
 156               (if (null xml)
 157                   (progn
 158                     (set 'result (xml-parse-tag end parse-dtd))
 159                     (cond
 160                      ((listp (car result))
 161                       (set 'dtd (car result))
 162                       (add-to-list 'xml (cdr result)))
 163                      (t
 164                       (add-to-list 'xml result))))
 165
 166                 ;;  translation of rule [1] of XML specifications
 167                 (error "XML files can have only one toplevel tag.")))
 168           (goto-char end)))
 169       (if parse-dtd
 170           (cons dtd (reverse xml))
 171         (reverse xml)))))
 172
 173
 174 (defun xml-parse-tag (end &optional parse-dtd)
 175   "Parse the tag that is just in front of point.
 176 The end tag must be found before the position END in the current buffer.
 177 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
 178 returned as the first element in the list.
 179 Returns one of:
 180    - a list : the matching node
 181    - nil    : the point is not looking at a tag.
 182    - a cons cell: the first element is the DTD, the second is the node"
 183   (cond
 184    ;; Processing instructions (like the <?xml version="1.0"?> tag at the
 185    ;; beginning of a document)
 186    ((looking-at "<\\?")
 187     (search-forward "?>" end)
 188     (skip-chars-forward " \t\n")
 189     (xml-parse-tag end))
 190    ;;  Character data (CDATA) sections, in which no tag should be interpreted
 191    ((looking-at "<!\\[CDATA\\[")
 192     (let ((pos (match-end 0)))
 193       (unless (search-forward "]]>" end t)
 194         (error "CDATA section does not end anywhere in the document"))
 195       (buffer-substring-no-properties pos (match-beginning 0))))
 196    ;;  DTD for the document
 197    ((looking-at "<!DOCTYPE")
 198     (let (dtd)
 199       (if parse-dtd
 200           (set 'dtd (xml-parse-dtd end))
 201         (xml-skip-dtd end))
 202       (skip-chars-forward " \t\n")
 203       (if dtd
 204           (cons dtd (xml-parse-tag end))
 205         (xml-parse-tag end))))
 206    ;;  skip comments
 207    ((looking-at "<!--")
 208     (search-forward "-->" end)
 209     (skip-chars-forward " \t\n")
 210     (xml-parse-tag end))
 211    ;;  end tag
 212    ((looking-at "</")
 213     '())
 214    ;;  opening tag
 215    ((looking-at "<\\([^/> \t\n]+\\)")
 216     (let* ((node-name (match-string 1))
 217            (children (list (intern node-name)))
 218            (case-fold-search nil) ;; XML is case-sensitive
 219            pos)
 220       (goto-char (match-end 1))
 221
 222       ;; parses the attribute list
 223       (set 'children (append children (list (xml-parse-attlist end))))
 224
 225       ;; is this an empty element ?
 226       (if (looking-at "/>")
 227           (progn
 228             (forward-char 2)
 229             (skip-chars-forward " \t\n")
 230             (append children '("")))
 231
 232         ;; is this a valid start tag ?
 233         (if (= (char-after) ?>)
 234             (progn
 235               (forward-char 1)
 236               (skip-chars-forward " \t\n")
 237               ;;  Now check that we have the right end-tag. Note that this one might
 238               ;;  contain spaces after the tag name
 239               (while (not (looking-at (concat "</" node-name "[ \t\n]*>")))
 240                 (cond
 241                  ((looking-at "</")
 242                   (error (concat
 243                           "XML: invalid syntax -- invalid end tag (expecting "
 244                           node-name
 245                           ") at pos " (number-to-string (point)))))
 246                  ((= (char-after) ?<)
 247                   (set 'children (append children (list (xml-parse-tag end)))))
 248                  (t
 249                   (set 'pos (point))
 250                   (search-forward "<" end)
 251                   (forward-char -1)
 252                   (let ((string (buffer-substring-no-properties pos (point)))
 253                         (pos 0))
 254
 255                     ;; Clean up the string (no newline characters)
 256                     ;; Not done, since as per XML specifications, the XML processor
 257                     ;; should always pass the whole string to the application.
 258                     ;;      (while (string-match "\\s +" string pos)
 259                     ;;        (set 'string (replace-match " " t t string))
 260                     ;;        (set 'pos (1+ (match-beginning 0))))
 261
 262                     (set 'children (append children
 263                                            (list (xml-substitute-special string))))))))
 264               (goto-char (match-end 0))
 265               (skip-chars-forward " \t\n")
 266               (if (> (point) end)
 267                   (error "XML: End tag for %s not found before end of region."
 268                          node-name))
 269               children
 270               )
 271
 272           ;;  This was an invalid start tag
 273           (error "XML: Invalid attribute list")
 274           ))))
 275    ))
 276
 277 (defun xml-parse-attlist (end)
 278   "Return the attribute-list that point is looking at.
 279 The search for attributes end at the position END in the current buffer.
 280 Leaves the point on the first non-blank character after the tag."
 281   (let ((attlist '())
 282         name)
 283     (skip-chars-forward " \t\n")
 284     (while (looking-at "\\([a-zA-Z_:][-a-zA-Z0-9._:]*\\)[ \t\n]*=[ \t\n]*")
 285       (set 'name (intern (match-string 1)))
 286       (goto-char (match-end 0))
 287
 288       ;; Do we have a string between quotes (or double-quotes),
 289       ;;  or a simple word ?
 290       (unless (looking-at "\"\\([^\"]+\\)\"")
 291         (unless (looking-at "'\\([^\"]+\\)'")
 292           (error "XML: Attribute values must be given between quotes.")))
 293
 294       ;; Each attribute must be unique within a given element
 295       (if (assoc name attlist)
 296           (error "XML: each attribute must be unique within an element."))
 297
 298       (set 'attlist (append attlist
 299                             (list (cons name (match-string-no-properties 1)))))
 300       (goto-char (match-end 0))
 301       (skip-chars-forward " \t\n")
 302       (if (> (point) end)
 303           (error "XML: end of attribute list not found before end of region."))
 304       )
 305     attlist
 306     ))
 307
 308 ;;*******************************************************************
 309 ;;**
 310 ;;**  The DTD (document type declaration)
 311 ;;**  The following functions know how to skip or parse the DTD of
 312 ;;**  a document
 313 ;;**
 314 ;;*******************************************************************
 315
 316 (defun xml-skip-dtd (end)
 317   "Skip the DTD that point is looking at.
 318 The DTD must end before the position END in the current buffer.
 319 The point must be just before the starting tag of the DTD.
 320 This follows the rule [28] in the XML specifications."
 321   (forward-char (length "<!DOCTYPE"))
 322   (if (looking-at "[ \t\n]*>")
 323       (error "XML: invalid DTD (excepting name of the document)"))
 324   (condition-case nil
 325       (progn
 326         (forward-word 1)  ;; name of the document
 327         (skip-chars-forward " \t\n")
 328         (if (looking-at "\\[")
 329             (re-search-forward "\\][ \t\n]*>" end)
 330           (search-forward ">" end)))
 331     (error (error "XML: No end to the DTD"))))
 332
 333 (defun xml-parse-dtd (end)
 334   "Parse the DTD that point is looking at.
 335 The DTD must end before the position END in the current buffer."
 336   (let (dtd type element end-pos)
 337     (forward-char (length "<!DOCTYPE"))
 338     (skip-chars-forward " \t\n")
 339     (if (looking-at ">")
 340         (error "XML: invalid DTD (excepting name of the document)"))
 341
 342     ;;  Get the name of the document
 343     (looking-at "\\sw+")
 344     (set 'dtd (list 'dtd (match-string-no-properties 0)))
 345     (goto-char (match-end 0))
 346
 347     (skip-chars-forward " \t\n")
 348
 349     ;;  External DTDs => don't know how to handle them yet
 350     (if (looking-at "SYSTEM")
 351         (error "XML: Don't know how to handle external DTDs."))
 352
 353     (if (not (= (char-after) ?\[))
 354         (error "XML: Unknown declaration in the DTD."))
 355
 356     ;;  Parse the rest of the DTD
 357     (forward-char 1)
 358     (while (and (not (looking-at "[ \t\n]*\\]"))
 359                 (<= (point) end))
 360       (cond
 361
 362        ;;  Translation of rule [45] of XML specifications
 363        ((looking-at
 364          "[\t \n]*<!ELEMENT[ \t\n]+\\([a-zA-Z0-9.%;]+\\)[ \t\n]+\\([^>]+\\)>")
 365
 366         (setq element (intern (match-string-no-properties 1))
 367               type    (match-string-no-properties 2))
 368         (set 'end-pos (match-end 0))
 369
 370         ;;  Translation of rule [46] of XML specifications
 371         (cond
 372          ((string-match "^EMPTY[ \t\n]*$" type)     ;; empty declaration
 373           (set 'type 'empty))
 374          ((string-match "^ANY[ \t\n]*$" type)       ;; any type of contents
 375           (set 'type 'any))
 376          ((string-match "^(\\(.*\\))[ \t\n]*$" type) ;; children ([47])
 377           (set 'type (xml-parse-elem-type (match-string-no-properties 1 type))))
 378          ((string-match "^%[^;]+;[ \t\n]*$" type)   ;; substitution
 379           nil)
 380          (t
 381           (error "XML: Invalid element type in the DTD")))
 382
 383         ;;  rule [45]: the element declaration must be unique
 384         (if (assoc element dtd)
 385             (error "XML: elements declaration must be unique in a DTD (<%s>)."
 386                    (symbol-name element)))
 387
 388         ;;  Store the element in the DTD
 389         (set 'dtd (append dtd (list (list element type))))
 390         (goto-char end-pos)
 391         )
 392
 393
 394        (t
 395         (error "XML: Invalid DTD item"))
 396        )
 397       )
 398
 399     ;;  Skip the end of the DTD
 400     (search-forward ">" end)
 401   dtd
 402   ))
 403
 404
 405 (defun xml-parse-elem-type (string)
 406   "Convert a STRING for an element type into an elisp structure."
 407
 408   (let (elem modifier)
 409     (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
 410         (progn
 411           (setq elem     (match-string 1 string)
 412                 modifier (match-string 2 string))
 413           (if (string-match "|" elem)
 414               (set 'elem (append '(choice)
 415                                (mapcar 'xml-parse-elem-type
 416                                        (split-string elem "|"))))
 417             (if (string-match "," elem)
 418                 (set 'elem (append '(seq)
 419                                  (mapcar 'xml-parse-elem-type
 420                                          (split-string elem ","))))
 421               )))
 422       (if (string-match "[ \t\n]*\\([^+*?]+\\)\\([+*?]?\\)" string)
 423           (setq elem     (match-string 1 string)
 424                 modifier (match-string 2 string))))
 425
 426       (if (and (stringp elem)
 427                (string= elem "#PCDATA"))
 428           (set 'elem 'pcdata))
 429
 430       (cond
 431        ((string= modifier "+")
 432         (list '+ elem))
 433        ((string= modifier "*")
 434         (list '* elem))
 435        ((string= modifier "?")
 436         (list '? elem))
 437        (t
 438         elem))))
 439
 440
 441 ;;*******************************************************************
 442 ;;**
 443 ;;**  Substituting special XML sequences
 444 ;;**
 445 ;;*******************************************************************
 446
 447 (defun xml-substitute-special (string)
 448   "Return STRING, after subsituting special XML sequences."
 449   (while (string-match "&amp;" string)
 450     (set 'string (replace-match "&"  t nil string)))
 451   (while (string-match "&lt;" string)
 452     (set 'string (replace-match "<"  t nil string)))
 453   (while (string-match "&gt;" string)
 454     (set 'string (replace-match ">"  t nil string)))
 455   (while (string-match "&apos;" string)
 456     (set 'string (replace-match "'"  t nil string)))
 457   (while (string-match "&quot;" string)
 458     (set 'string (replace-match "\"" t nil string)))
 459   string)
 460
 461 ;;*******************************************************************
 462 ;;**
 463 ;;**  Printing a tree.
 464 ;;**  This function is intended mainly for debugging purposes.
 465 ;;**
 466 ;;*******************************************************************
 467
 468 (defun xml-debug-print (xml)
 469   (while xml
 470     (xml-debug-print-internal (car xml) "")
 471     (set 'xml (cdr xml)))
 472   )
 473
 474 (defun xml-debug-print-internal (xml &optional indent-string)
 475   "Outputs the XML tree in the current buffer.
 476 The first line indented with INDENT-STRING."
 477   (let ((tree xml)
 478         attlist)
 479     (unless indent-string
 480       (set 'indent-string ""))
 481
 482     (insert indent-string "<" (symbol-name (xml-node-name tree)))
 483
 484     ;;  output the attribute list
 485     (set 'attlist (xml-node-attributes tree))
 486     (while attlist
 487       (insert " ")
 488       (insert (symbol-name (caar attlist)) "=\"" (cdar attlist) "\"")
 489       (set 'attlist (cdr attlist)))
 490
 491     (insert ">")
 492
 493     (set 'tree (xml-node-children tree))
 494
 495     ;;  output the children
 496     (while tree
 497       (cond
 498        ((listp (car tree))
 499         (insert "\n")
 500         (xml-debug-print-internal (car tree) (concat indent-string "  "))
 501         )
 502        ((stringp (car tree))
 503         (insert (car tree))
 504         )
 505        (t
 506         (error "Invalid XML tree")))
 507       (set 'tree (cdr tree))
 508      )
 509
 510     (insert "\n" indent-string
 511             "</" (symbol-name (xml-node-name xml)) ">")
 512     ))
 513
 514 (provide 'xml)
 515
 516 ;;; xml.el ends here