lisp/xml.el

   1 ;; @(#) xml.el --- XML parser
   2
   3 ;; Copyright (C) 2000, 2001 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    (t ;; This is not a tag.
 276     (error "XML: Invalid character."))
 277    ))
 278
 279 (defun xml-parse-attlist (end)
 280   "Return the attribute-list that point is looking at.
 281 The search for attributes end at the position END in the current buffer.
 282 Leaves the point on the first non-blank character after the tag."
 283   (let ((attlist '())
 284         name)
 285     (skip-chars-forward " \t\n")
 286     (while (looking-at "\\([a-zA-Z_:][-a-zA-Z0-9._:]*\\)[ \t\n]*=[ \t\n]*")
 287       (set 'name (intern (match-string 1)))
 288       (goto-char (match-end 0))
 289
 290       ;; Do we have a string between quotes (or double-quotes),
 291       ;;  or a simple word ?
 292       (unless (looking-at "\"\\([^\"]+\\)\"")
 293         (unless (looking-at "'\\([^\"]+\\)'")
 294           (error "XML: Attribute values must be given between quotes.")))
 295
 296       ;; Each attribute must be unique within a given element
 297       (if (assoc name attlist)
 298           (error "XML: each attribute must be unique within an element."))
 299
 300       (set 'attlist (append attlist
 301                             (list (cons name (match-string-no-properties 1)))))
 302       (goto-char (match-end 0))
 303       (skip-chars-forward " \t\n")
 304       (if (> (point) end)
 305           (error "XML: end of attribute list not found before end of region."))
 306       )
 307     attlist
 308     ))
 309
 310 ;;*******************************************************************
 311 ;;**
 312 ;;**  The DTD (document type declaration)
 313 ;;**  The following functions know how to skip or parse the DTD of
 314 ;;**  a document
 315 ;;**
 316 ;;*******************************************************************
 317
 318 (defun xml-skip-dtd (end)
 319   "Skip the DTD that point is looking at.
 320 The DTD must end before the position END in the current buffer.
 321 The point must be just before the starting tag of the DTD.
 322 This follows the rule [28] in the XML specifications."
 323   (forward-char (length "<!DOCTYPE"))
 324   (if (looking-at "[ \t\n]*>")
 325       (error "XML: invalid DTD (excepting name of the document)"))
 326   (condition-case nil
 327       (progn
 328         (forward-word 1)  ;; name of the document
 329         (skip-chars-forward " \t\n")
 330         (if (looking-at "\\[")
 331             (re-search-forward "\\][ \t\n]*>" end)
 332           (search-forward ">" end)))
 333     (error (error "XML: No end to the DTD"))))
 334
 335 (defun xml-parse-dtd (end)
 336   "Parse the DTD that point is looking at.
 337 The DTD must end before the position END in the current buffer."
 338   (let (dtd type element end-pos)
 339     (forward-char (length "<!DOCTYPE"))
 340     (skip-chars-forward " \t\n")
 341     (if (looking-at ">")
 342         (error "XML: invalid DTD (excepting name of the document)"))
 343
 344     ;;  Get the name of the document
 345     (looking-at "\\sw+")
 346     (set 'dtd (list 'dtd (match-string-no-properties 0)))
 347     (goto-char (match-end 0))
 348
 349     (skip-chars-forward " \t\n")
 350
 351     ;;  External DTDs => don't know how to handle them yet
 352     (if (looking-at "SYSTEM")
 353         (error "XML: Don't know how to handle external DTDs."))
 354
 355     (if (not (= (char-after) ?\[))
 356         (error "XML: Unknown declaration in the DTD."))
 357
 358     ;;  Parse the rest of the DTD
 359     (forward-char 1)
 360     (while (and (not (looking-at "[ \t\n]*\\]"))
 361                 (<= (point) end))
 362       (cond
 363
 364        ;;  Translation of rule [45] of XML specifications
 365        ((looking-at
 366          "[\t \n]*<!ELEMENT[ \t\n]+\\([a-zA-Z0-9.%;]+\\)[ \t\n]+\\([^>]+\\)>")
 367
 368         (setq element (intern (match-string-no-properties 1))
 369               type    (match-string-no-properties 2))
 370         (set 'end-pos (match-end 0))
 371
 372         ;;  Translation of rule [46] of XML specifications
 373         (cond
 374          ((string-match "^EMPTY[ \t\n]*$" type)     ;; empty declaration
 375           (set 'type 'empty))
 376          ((string-match "^ANY[ \t\n]*$" type)       ;; any type of contents
 377           (set 'type 'any))
 378          ((string-match "^(\\(.*\\))[ \t\n]*$" type) ;; children ([47])
 379           (set 'type (xml-parse-elem-type (match-string-no-properties 1 type))))
 380          ((string-match "^%[^;]+;[ \t\n]*$" type)   ;; substitution
 381           nil)
 382          (t
 383           (error "XML: Invalid element type in the DTD")))
 384
 385         ;;  rule [45]: the element declaration must be unique
 386         (if (assoc element dtd)
 387             (error "XML: elements declaration must be unique in a DTD (<%s>)."
 388                    (symbol-name element)))
 389
 390         ;;  Store the element in the DTD
 391         (set 'dtd (append dtd (list (list element type))))
 392         (goto-char end-pos)
 393         )
 394
 395
 396        (t
 397         (error "XML: Invalid DTD item"))
 398        )
 399       )
 400
 401     ;;  Skip the end of the DTD
 402     (search-forward ">" end)
 403   dtd
 404   ))
 405
 406
 407 (defun xml-parse-elem-type (string)
 408   "Convert a STRING for an element type into an elisp structure."
 409
 410   (let (elem modifier)
 411     (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
 412         (progn
 413           (setq elem     (match-string 1 string)
 414                 modifier (match-string 2 string))
 415           (if (string-match "|" elem)
 416               (set 'elem (append '(choice)
 417                                (mapcar 'xml-parse-elem-type
 418                                        (split-string elem "|"))))
 419             (if (string-match "," elem)
 420                 (set 'elem (append '(seq)
 421                                  (mapcar 'xml-parse-elem-type
 422                                          (split-string elem ","))))
 423               )))
 424       (if (string-match "[ \t\n]*\\([^+*?]+\\)\\([+*?]?\\)" string)
 425           (setq elem     (match-string 1 string)
 426                 modifier (match-string 2 string))))
 427
 428       (if (and (stringp elem)
 429                (string= elem "#PCDATA"))
 430           (set 'elem 'pcdata))
 431
 432       (cond
 433        ((string= modifier "+")
 434         (list '+ elem))
 435        ((string= modifier "*")
 436         (list '* elem))
 437        ((string= modifier "?")
 438         (list '? elem))
 439        (t
 440         elem))))
 441
 442
 443 ;;*******************************************************************
 444 ;;**
 445 ;;**  Substituting special XML sequences
 446 ;;**
 447 ;;*******************************************************************
 448
 449 (defun xml-substitute-special (string)
 450   "Return STRING, after subsituting special XML sequences."
 451   (while (string-match "&amp;" string)
 452     (set 'string (replace-match "&"  t nil string)))
 453   (while (string-match "&lt;" string)
 454     (set 'string (replace-match "<"  t nil string)))
 455   (while (string-match "&gt;" string)
 456     (set 'string (replace-match ">"  t nil string)))
 457   (while (string-match "&apos;" string)
 458     (set 'string (replace-match "'"  t nil string)))
 459   (while (string-match "&quot;" string)
 460     (set 'string (replace-match "\"" t nil string)))
 461   string)
 462
 463 ;;*******************************************************************
 464 ;;**
 465 ;;**  Printing a tree.
 466 ;;**  This function is intended mainly for debugging purposes.
 467 ;;**
 468 ;;*******************************************************************
 469
 470 (defun xml-debug-print (xml)
 471   (while xml
 472     (xml-debug-print-internal (car xml) "")
 473     (set 'xml (cdr xml)))
 474   )
 475
 476 (defun xml-debug-print-internal (xml &optional indent-string)
 477   "Outputs the XML tree in the current buffer.
 478 The first line indented with INDENT-STRING."
 479   (let ((tree xml)
 480         attlist)
 481     (unless indent-string
 482       (set 'indent-string ""))
 483
 484     (insert indent-string "<" (symbol-name (xml-node-name tree)))
 485
 486     ;;  output the attribute list
 487     (set 'attlist (xml-node-attributes tree))
 488     (while attlist
 489       (insert " ")
 490       (insert (symbol-name (caar attlist)) "=\"" (cdar attlist) "\"")
 491       (set 'attlist (cdr attlist)))
 492
 493     (insert ">")
 494
 495     (set 'tree (xml-node-children tree))
 496
 497     ;;  output the children
 498     (while tree
 499       (cond
 500        ((listp (car tree))
 501         (insert "\n")
 502         (xml-debug-print-internal (car tree) (concat indent-string "  "))
 503         )
 504        ((stringp (car tree))
 505         (insert (car tree))
 506         )
 507        (t
 508         (error "Invalid XML tree")))
 509       (set 'tree (cdr tree))
 510      )
 511
 512     (insert "\n" indent-string
 513             "</" (symbol-name (xml-node-name xml)) ">")
 514     ))
 515
 516 (provide 'xml)
 517
 518 ;;; xml.el ends here