Commit | Line | Data |
---|---|---|
1cd7adc6 | 1 | ;;; xml.el --- XML parser |
47db06aa | 2 | |
2e9bdf15 | 3 | ;; Copyright (C) 2000, 01, 03, 2004 Free Software Foundation, Inc. |
47db06aa GM |
4 | |
5 | ;; Author: Emmanuel Briot <briot@gnat.com> | |
720058f2 | 6 | ;; Maintainer: Mark A. Hershberger <mah@everybody.org> |
a98e819b | 7 | ;; Keywords: xml, data |
47db06aa GM |
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 | ||
a98e819b DL |
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. | |
47db06aa GM |
31 | |
32 | ;;; FILE FORMAT | |
33 | ||
a98e819b DL |
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. | |
47db06aa | 39 | ;; |
a98e819b | 40 | ;; It also knows how to ignore comments and processing instructions. |
47db06aa GM |
41 | ;; |
42 | ;; The XML file should have the following format: | |
653558a1 GM |
43 | ;; <node1 attr1="name1" attr2="name2" ...>value |
44 | ;; <node2 attr3="name3" attr4="name4">value2</node2> | |
45 | ;; <node3 attr5="name5" attr6="name6">value3</node3> | |
47db06aa GM |
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 | ||
c7f8d055 SM |
55 | ;; The functions `xml-parse-file', `xml-parse-region' and |
56 | ;; `xml-parse-tag' return a list with the following format: | |
47db06aa GM |
57 | ;; |
58 | ;; xml-list ::= (node node ...) | |
c7f8d055 | 59 | ;; node ::= (qname attribute-list . child_node_list) |
47db06aa GM |
60 | ;; child_node_list ::= child_node child_node ... |
61 | ;; child_node ::= node | string | |
c7f8d055 SM |
62 | ;; qname ::= (:namespace-uri . "name") | "name" |
63 | ;; attribute_list ::= ((qname . "value") (qname . "value") ...) | |
47db06aa GM |
64 | ;; | nil |
65 | ;; string ::= "..." | |
66 | ;; | |
a98e819b DL |
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. | |
47db06aa | 70 | |
c7f8d055 SM |
71 | ;; TODO: |
72 | ;; * xml:base, xml:space support | |
73 | ;; * more complete DOCTYPE parsing | |
74 | ;; * pi support | |
75 | ||
47db06aa GM |
76 | ;;; Code: |
77 | ||
a98e819b DL |
78 | ;; Note that {buffer-substring,match-string}-no-properties were |
79 | ;; formerly used in several places, but that removes composition info. | |
80 | ||
47db06aa GM |
81 | ;;******************************************************************* |
82 | ;;** | |
83 | ;;** Macros to parse the list | |
84 | ;;** | |
85 | ;;******************************************************************* | |
86 | ||
971489ea | 87 | (defsubst xml-node-name (node) |
47db06aa GM |
88 | "Return the tag associated with NODE. |
89 | The tag is a lower-case symbol." | |
971489ea | 90 | (car node)) |
47db06aa | 91 | |
971489ea | 92 | (defsubst xml-node-attributes (node) |
47db06aa GM |
93 | "Return the list of attributes of NODE. |
94 | The list can be nil." | |
971489ea | 95 | (nth 1 node)) |
47db06aa | 96 | |
971489ea | 97 | (defsubst xml-node-children (node) |
47db06aa GM |
98 | "Return the list of children of NODE. |
99 | This is a list of nodes, and it can be nil." | |
971489ea | 100 | (cddr node)) |
47db06aa GM |
101 | |
102 | (defun xml-get-children (node child-name) | |
103 | "Return the children of NODE whose tag is CHILD-NAME. | |
104 | CHILD-NAME should be a lower case symbol." | |
971489ea SM |
105 | (let ((match ())) |
106 | (dolist (child (xml-node-children node)) | |
107 | (if child | |
108 | (if (equal (xml-node-name child) child-name) | |
109 | (push child match)))) | |
110 | (nreverse match))) | |
47db06aa | 111 | |
9bcd6a7e | 112 | (defun xml-get-attribute-or-nil (node attribute) |
47db06aa | 113 | "Get from NODE the value of ATTRIBUTE. |
65f3f600 | 114 | Return `nil' if the attribute was not found. |
9bcd6a7e EZ |
115 | |
116 | See also `xml-get-attribute'." | |
2e9bdf15 | 117 | (cdr (assoc attribute (xml-node-attributes node)))) |
9bcd6a7e EZ |
118 | |
119 | (defsubst xml-get-attribute (node attribute) | |
120 | "Get from NODE the value of ATTRIBUTE. | |
121 | An empty string is returned if the attribute was not found. | |
122 | ||
123 | See also `xml-get-attribute-or-nil'." | |
124 | (or (xml-get-attribute-or-nil node attribute) "")) | |
47db06aa GM |
125 | |
126 | ;;******************************************************************* | |
127 | ;;** | |
128 | ;;** Creating the list | |
129 | ;;** | |
130 | ;;******************************************************************* | |
131 | ||
a98e819b | 132 | ;;;###autoload |
2d42509a | 133 | (defun xml-parse-file (file &optional parse-dtd parse-ns) |
a98e819b DL |
134 | "Parse the well-formed XML file FILE. |
135 | If FILE is already visited, use its buffer and don't kill it. | |
47db06aa | 136 | Returns the top node with all its children. |
2d42509a JB |
137 | If PARSE-DTD is non-nil, the DTD is parsed rather than skipped. |
138 | If PARSE-NS is non-nil, then QNAMES are expanded." | |
653558a1 GM |
139 | (let ((keep)) |
140 | (if (get-file-buffer file) | |
141 | (progn | |
142 | (set-buffer (get-file-buffer file)) | |
143 | (setq keep (point))) | |
a98e819b DL |
144 | (let (auto-mode-alist) ; no need for xml-mode |
145 | (find-file file))) | |
524425ae | 146 | |
653558a1 GM |
147 | (let ((xml (xml-parse-region (point-min) |
148 | (point-max) | |
149 | (current-buffer) | |
2d42509a | 150 | parse-dtd parse-ns))) |
653558a1 GM |
151 | (if keep |
152 | (goto-char keep) | |
153 | (kill-buffer (current-buffer))) | |
154 | xml))) | |
47db06aa | 155 | |
a98e819b DL |
156 | ;; Note that this is setup so that we can do whitespace-skipping with |
157 | ;; `(skip-syntax-forward " ")', inter alia. Previously this was slow | |
158 | ;; compared with `re-search-forward', but that has been fixed. Also | |
159 | ;; note that the standard syntax table contains other characters with | |
160 | ;; whitespace syntax, like NBSP, but they are invalid in contexts in | |
161 | ;; which we might skip whitespace -- specifically, they're not | |
162 | ;; NameChars [XML 4]. | |
163 | ||
164 | (defvar xml-syntax-table | |
165 | (let ((table (make-syntax-table))) | |
166 | ;; Get space syntax correct per XML [3]. | |
167 | (dotimes (c 31) | |
168 | (modify-syntax-entry c "." table)) ; all are space in standard table | |
169 | (dolist (c '(?\t ?\n ?\r)) ; these should be space | |
170 | (modify-syntax-entry c " " table)) | |
171 | ;; For skipping attributes. | |
172 | (modify-syntax-entry ?\" "\"" table) | |
173 | (modify-syntax-entry ?' "\"" table) | |
174 | ;; Non-alnum name chars should be symbol constituents (`-' and `_' | |
175 | ;; are OK by default). | |
176 | (modify-syntax-entry ?. "_" table) | |
177 | (modify-syntax-entry ?: "_" table) | |
178 | ;; XML [89] | |
179 | (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005 | |
180 | #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC | |
181 | #x30FD #x30FE)) | |
182 | (modify-syntax-entry (decode-char 'ucs c) "w" table)) | |
183 | ;; Fixme: rest of [4] | |
184 | table) | |
185 | "Syntax table used by `xml-parse-region'.") | |
186 | ||
187 | ;; XML [5] | |
188 | ;; Note that [:alpha:] matches all multibyte chars with word syntax. | |
ab161457 JPW |
189 | (eval-and-compile |
190 | (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*")) | |
a98e819b DL |
191 | |
192 | ;; Fixme: This needs re-writing to deal with the XML grammar properly, i.e. | |
193 | ;; document ::= prolog element Misc* | |
194 | ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? | |
195 | ||
196 | ;;;###autoload | |
2d42509a | 197 | (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns) |
47db06aa GM |
198 | "Parse the region from BEG to END in BUFFER. |
199 | If BUFFER is nil, it defaults to the current buffer. | |
200 | Returns the XML list for the region, or raises an error if the region | |
2d42509a | 201 | is not well-formed XML. |
47db06aa | 202 | If PARSE-DTD is non-nil, the DTD is parsed rather than skipped, |
2d42509a JB |
203 | and returned as the first element of the list. |
204 | If PARSE-NS is non-nil, then QNAMES are expanded." | |
a98e819b DL |
205 | (save-restriction |
206 | (narrow-to-region beg end) | |
207 | ;; Use fixed syntax table to ensure regexp char classes and syntax | |
208 | ;; specs DTRT. | |
209 | (with-syntax-table (standard-syntax-table) | |
210 | (let ((case-fold-search nil) ; XML is case-sensitive. | |
211 | xml result dtd) | |
212 | (save-excursion | |
213 | (if buffer | |
214 | (set-buffer buffer)) | |
215 | (goto-char (point-min)) | |
216 | (while (not (eobp)) | |
217 | (if (search-forward "<" nil t) | |
218 | (progn | |
219 | (forward-char -1) | |
34638996 EZ |
220 | (setq result (xml-parse-tag parse-dtd parse-ns)) |
221 | (if (and xml result) | |
a98e819b DL |
222 | ;; translation of rule [1] of XML specifications |
223 | (error "XML files can have only one toplevel tag") | |
47db06aa | 224 | (cond |
971489ea | 225 | ((null result)) |
34638996 EZ |
226 | ((and (listp (car result)) |
227 | parse-dtd) | |
971489ea | 228 | (setq dtd (car result)) |
a98e819b DL |
229 | (if (cdr result) ; possible leading comment |
230 | (add-to-list 'xml (cdr result)))) | |
47db06aa | 231 | (t |
a98e819b DL |
232 | (add-to-list 'xml result))))) |
233 | (goto-char (point-max)))) | |
234 | (if parse-dtd | |
235 | (cons dtd (nreverse xml)) | |
236 | (nreverse xml))))))) | |
47db06aa | 237 | |
c7f8d055 SM |
238 | (defun xml-maybe-do-ns (name default xml-ns) |
239 | "Perform any namespace expansion. NAME is the name to perform the expansion on. | |
240 | DEFAULT is the default namespace. XML-NS is a cons of namespace | |
241 | names to uris. When namespace-aware parsing is off, then XML-NS | |
242 | is nil. | |
243 | ||
244 | During namespace-aware parsing, any name without a namespace is | |
245 | put into the namespace identified by DEFAULT. nil is used to | |
246 | specify that the name shouldn't be given a namespace." | |
247 | (if (consp xml-ns) | |
248 | (let* ((nsp (string-match ":" name)) | |
249 | (lname (if nsp (substring name (match-end 0)) name)) | |
250 | (prefix (if nsp (substring name 0 (match-beginning 0)) default)) | |
251 | (special (and (string-equal lname "xmlns") (not prefix))) | |
252 | ;; Setting default to nil will insure that there is not | |
253 | ;; matching cons in xml-ns. In which case we | |
254 | (ns (or (cdr (assoc (if special "xmlns" prefix) | |
255 | xml-ns)) | |
256 | :))) | |
257 | (cons ns (if special "" lname))) | |
258 | (intern name))) | |
47db06aa | 259 | |
2d42509a | 260 | (defun xml-parse-tag (&optional parse-dtd parse-ns) |
a98e819b | 261 | "Parse the tag at point. |
47db06aa GM |
262 | If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and |
263 | returned as the first element in the list. | |
2d42509a | 264 | If PARSE-NS is non-nil, then QNAMES are expanded. |
47db06aa | 265 | Returns one of: |
a98e819b DL |
266 | - a list : the matching node |
267 | - nil : the point is not looking at a tag. | |
268 | - a pair : the first element is the DTD, the second is the node." | |
2d42509a JB |
269 | (let ((xml-ns (if (consp parse-ns) |
270 | parse-ns | |
271 | (if parse-ns | |
272 | (list | |
c7f8d055 SM |
273 | ;; Default for empty prefix is no namespace |
274 | (cons "" :) | |
275 | ;; "xml" namespace | |
276 | (cons "xml" :http://www.w3.org/XML/1998/namespace) | |
2d42509a | 277 | ;; We need to seed the xmlns namespace |
c7f8d055 | 278 | (cons "xmlns" :http://www.w3.org/2000/xmlns/)))))) |
2d42509a JB |
279 | (cond |
280 | ;; Processing instructions (like the <?xml version="1.0"?> tag at the | |
281 | ;; beginning of a document). | |
282 | ((looking-at "<\\?") | |
283 | (search-forward "?>") | |
284 | (skip-syntax-forward " ") | |
285 | (xml-parse-tag parse-dtd xml-ns)) | |
286 | ;; Character data (CDATA) sections, in which no tag should be interpreted | |
287 | ((looking-at "<!\\[CDATA\\[") | |
288 | (let ((pos (match-end 0))) | |
289 | (unless (search-forward "]]>" nil t) | |
290 | (error "CDATA section does not end anywhere in the document")) | |
291 | (buffer-substring pos (match-beginning 0)))) | |
292 | ;; DTD for the document | |
293 | ((looking-at "<!DOCTYPE") | |
294 | (let (dtd) | |
295 | (if parse-dtd | |
296 | (setq dtd (xml-parse-dtd)) | |
297 | (xml-skip-dtd)) | |
a98e819b | 298 | (skip-syntax-forward " ") |
47db06aa | 299 | (if dtd |
2d42509a JB |
300 | (cons dtd (xml-parse-tag nil xml-ns)) |
301 | (xml-parse-tag nil xml-ns)))) | |
302 | ;; skip comments | |
303 | ((looking-at "<!--") | |
304 | (search-forward "-->") | |
305 | nil) | |
306 | ;; end tag | |
307 | ((looking-at "</") | |
308 | '()) | |
309 | ;; opening tag | |
310 | ((looking-at "<\\([^/>[:space:]]+\\)") | |
311 | (goto-char (match-end 1)) | |
34638996 EZ |
312 | |
313 | ;; Parse this node | |
2d42509a | 314 | (let* ((node-name (match-string 1)) |
c7f8d055 SM |
315 | ;; Parse the attribute list. |
316 | (attrs (xml-parse-attlist xml-ns)) | |
317 | children pos) | |
318 | ||
319 | ;; add the xmlns:* attrs to our cache | |
320 | (when (consp xml-ns) | |
321 | (dolist (attr attrs) | |
322 | (when (and (consp (car attr)) | |
323 | (eq :http://www.w3.org/2000/xmlns/ | |
324 | (caar attr))) | |
325 | (push (cons (cdar attr) (intern (concat ":" (cdr attr)))) | |
326 | xml-ns)))) | |
327 | ||
43b5fd81 | 328 | (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns))) |
c7f8d055 | 329 | |
2d42509a JB |
330 | ;; is this an empty element ? |
331 | (if (looking-at "/>") | |
332 | (progn | |
333 | (forward-char 2) | |
334 | (nreverse children)) | |
47db06aa GM |
335 | |
336 | ;; is this a valid start tag ? | |
e54030af | 337 | (if (eq (char-after) ?>) |
47db06aa GM |
338 | (progn |
339 | (forward-char 1) | |
971489ea SM |
340 | ;; Now check that we have the right end-tag. Note that this |
341 | ;; one might contain spaces after the tag name | |
a98e819b DL |
342 | (let ((end (concat "</" node-name "\\s-*>"))) |
343 | (while (not (looking-at end)) | |
344 | (cond | |
345 | ((looking-at "</") | |
346 | (error "XML: Invalid end tag (expecting %s) at pos %d" | |
347 | node-name (point))) | |
348 | ((= (char-after) ?<) | |
2d42509a | 349 | (let ((tag (xml-parse-tag nil xml-ns))) |
a98e819b DL |
350 | (when tag |
351 | (push tag children)))) | |
352 | (t | |
353 | (setq pos (point)) | |
354 | (search-forward "<") | |
355 | (forward-char -1) | |
356 | (let ((string (buffer-substring pos (point))) | |
357 | (pos 0)) | |
358 | ||
359 | ;; Clean up the string. As per XML | |
360 | ;; specifications, the XML processor should | |
361 | ;; always pass the whole string to the | |
362 | ;; application. But \r's should be replaced: | |
363 | ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends | |
364 | (while (string-match "\r\n?" string pos) | |
365 | (setq string (replace-match "\n" t t string)) | |
366 | (setq pos (1+ (match-beginning 0)))) | |
367 | ||
368 | (setq string (xml-substitute-special string)) | |
369 | (setq children | |
370 | (if (stringp (car children)) | |
371 | ;; The two strings were separated by a comment. | |
372 | (cons (concat (car children) string) | |
373 | (cdr children)) | |
374 | (cons string children)))))))) | |
375 | ||
47db06aa | 376 | (goto-char (match-end 0)) |
971489ea | 377 | (nreverse children)) |
47db06aa | 378 | ;; This was an invalid start tag |
a98e819b | 379 | (error "XML: Invalid attribute list"))))) |
2d42509a JB |
380 | (t ;; This is not a tag. |
381 | (error "XML: Invalid character"))))) | |
47db06aa | 382 | |
c7f8d055 | 383 | (defun xml-parse-attlist (&optional xml-ns) |
34638996 EZ |
384 | "Return the attribute-list after point. Leave point at the |
385 | first non-blank character after the tag." | |
971489ea | 386 | (let ((attlist ()) |
34638996 | 387 | end-pos name) |
a98e819b DL |
388 | (skip-syntax-forward " ") |
389 | (while (looking-at (eval-when-compile | |
390 | (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*"))) | |
c7f8d055 SM |
391 | (setq end-pos (match-end 0)) |
392 | (setq name (xml-maybe-do-ns (match-string 1) nil xml-ns)) | |
393 | (goto-char end-pos) | |
47db06aa | 394 | |
a158ff81 JB |
395 | ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize |
396 | ||
47db06aa GM |
397 | ;; Do we have a string between quotes (or double-quotes), |
398 | ;; or a simple word ? | |
a158ff81 | 399 | (if (looking-at "\"\\([^\"]*\\)\"") |
34638996 | 400 | (setq end-pos (match-end 0)) |
f0ec1711 | 401 | (if (looking-at "'\\([^']*\\)'") |
34638996 | 402 | (setq end-pos (match-end 0)) |
1cd7adc6 | 403 | (error "XML: Attribute values must be given between quotes"))) |
47db06aa GM |
404 | |
405 | ;; Each attribute must be unique within a given element | |
406 | (if (assoc name attlist) | |
1cd7adc6 | 407 | (error "XML: each attribute must be unique within an element")) |
524425ae | 408 | |
a158ff81 JB |
409 | ;; Multiple whitespace characters should be replaced with a single one |
410 | ;; in the attributes | |
a98e819b | 411 | (let ((string (match-string 1)) |
a158ff81 | 412 | (pos 0)) |
a98e819b | 413 | (replace-regexp-in-string "\\s-\\{2,\\}" " " string) |
a158ff81 JB |
414 | (push (cons name (xml-substitute-special string)) attlist)) |
415 | ||
34638996 | 416 | (goto-char end-pos) |
a98e819b | 417 | (skip-syntax-forward " ")) |
971489ea | 418 | (nreverse attlist))) |
47db06aa GM |
419 | |
420 | ;;******************************************************************* | |
421 | ;;** | |
422 | ;;** The DTD (document type declaration) | |
423 | ;;** The following functions know how to skip or parse the DTD of | |
424 | ;;** a document | |
425 | ;;** | |
426 | ;;******************************************************************* | |
427 | ||
a98e819b DL |
428 | ;; Fixme: This fails at least if the DTD contains conditional sections. |
429 | ||
430 | (defun xml-skip-dtd () | |
431 | "Skip the DTD at point. | |
47db06aa GM |
432 | This follows the rule [28] in the XML specifications." |
433 | (forward-char (length "<!DOCTYPE")) | |
a98e819b | 434 | (if (looking-at "\\s-*>") |
47db06aa GM |
435 | (error "XML: invalid DTD (excepting name of the document)")) |
436 | (condition-case nil | |
437 | (progn | |
a98e819b DL |
438 | (forward-sexp) |
439 | (skip-syntax-forward " ") | |
47db06aa | 440 | (if (looking-at "\\[") |
a98e819b DL |
441 | (re-search-forward "]\\s-*>") |
442 | (search-forward ">"))) | |
47db06aa GM |
443 | (error (error "XML: No end to the DTD")))) |
444 | ||
a98e819b DL |
445 | (defun xml-parse-dtd () |
446 | "Parse the DTD at point." | |
447 | (forward-char (eval-when-compile (length "<!DOCTYPE"))) | |
448 | (skip-syntax-forward " ") | |
971489ea SM |
449 | (if (looking-at ">") |
450 | (error "XML: invalid DTD (excepting name of the document)")) | |
524425ae | 451 | |
971489ea | 452 | ;; Get the name of the document |
a98e819b DL |
453 | (looking-at xml-name-regexp) |
454 | (let ((dtd (list (match-string 0) 'dtd)) | |
971489ea | 455 | type element end-pos) |
47db06aa GM |
456 | (goto-char (match-end 0)) |
457 | ||
a98e819b DL |
458 | (skip-syntax-forward " ") |
459 | ;; XML [75] | |
460 | (cond ((looking-at "PUBLIC\\s-+") | |
461 | (goto-char (match-end 0)) | |
462 | (unless (or (re-search-forward | |
463 | "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\"" | |
464 | nil t) | |
465 | (re-search-forward | |
466 | "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'" | |
467 | nil t)) | |
468 | (error "XML: missing public id")) | |
469 | (let ((pubid (match-string 1))) | |
470 | (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t) | |
471 | (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t)) | |
472 | (error "XML: missing system id")) | |
473 | (push (list pubid (match-string 1) 'public) dtd))) | |
474 | ((looking-at "SYSTEM\\s-+") | |
475 | (goto-char (match-end 0)) | |
476 | (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t) | |
477 | (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t)) | |
478 | (error "XML: missing system id")) | |
479 | (push (list (match-string 1) 'system) dtd))) | |
480 | (skip-syntax-forward " ") | |
481 | (if (eq ?> (char-after)) | |
482 | (forward-char) | |
483 | (skip-syntax-forward " ") | |
484 | (if (not (eq (char-after) ?\[)) | |
485 | (error "XML: bad DTD") | |
486 | (forward-char) | |
487 | ;; Parse the rest of the DTD | |
488 | ;; Fixme: Deal with ENTITY, ATTLIST, NOTATION, PIs. | |
489 | (while (not (looking-at "\\s-*\\]")) | |
490 | (skip-syntax-forward " ") | |
491 | (cond | |
492 | ||
493 | ;; Translation of rule [45] of XML specifications | |
494 | ((looking-at | |
495 | "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>") | |
496 | ||
34638996 | 497 | (setq element (match-string 1) |
a98e819b DL |
498 | type (match-string-no-properties 2)) |
499 | (setq end-pos (match-end 0)) | |
500 | ||
501 | ;; Translation of rule [46] of XML specifications | |
502 | (cond | |
503 | ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration | |
504 | (setq type 'empty)) | |
505 | ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents | |
506 | (setq type 'any)) | |
507 | ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47]) | |
508 | (setq type (xml-parse-elem-type (match-string 1 type)))) | |
509 | ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution | |
510 | nil) | |
511 | (t | |
512 | (error "XML: Invalid element type in the DTD"))) | |
513 | ||
514 | ;; rule [45]: the element declaration must be unique | |
515 | (if (assoc element dtd) | |
516 | (error "XML: element declarations must be unique in a DTD (<%s>)" | |
461f3ad0 | 517 | element)) |
a98e819b DL |
518 | |
519 | ;; Store the element in the DTD | |
520 | (push (list element type) dtd) | |
521 | (goto-char end-pos)) | |
522 | ((looking-at "<!--") | |
523 | (search-forward "-->")) | |
524 | ||
525 | (t | |
526 | (error "XML: Invalid DTD item"))) | |
527 | ||
528 | ;; Skip the end of the DTD | |
529 | (search-forward ">")))) | |
461f3ad0 | 530 | (nreverse dtd))) |
47db06aa GM |
531 | |
532 | (defun xml-parse-elem-type (string) | |
a98e819b | 533 | "Convert element type STRING into a Lisp structure." |
47db06aa GM |
534 | |
535 | (let (elem modifier) | |
536 | (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string) | |
537 | (progn | |
538 | (setq elem (match-string 1 string) | |
539 | modifier (match-string 2 string)) | |
540 | (if (string-match "|" elem) | |
971489ea | 541 | (setq elem (cons 'choice |
47db06aa GM |
542 | (mapcar 'xml-parse-elem-type |
543 | (split-string elem "|")))) | |
544 | (if (string-match "," elem) | |
971489ea | 545 | (setq elem (cons 'seq |
47db06aa | 546 | (mapcar 'xml-parse-elem-type |
a98e819b | 547 | (split-string elem ","))))))) |
a158ff81 JB |
548 | (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string) |
549 | (setq elem (match-string 1 string) | |
47db06aa GM |
550 | modifier (match-string 2 string)))) |
551 | ||
971489ea SM |
552 | (if (and (stringp elem) (string= elem "#PCDATA")) |
553 | (setq elem 'pcdata)) | |
524425ae | 554 | |
971489ea SM |
555 | (cond |
556 | ((string= modifier "+") | |
557 | (list '+ elem)) | |
558 | ((string= modifier "*") | |
559 | (list '* elem)) | |
560 | ((string= modifier "?") | |
0fa6f70c | 561 | (list '\? elem)) |
971489ea SM |
562 | (t |
563 | elem)))) | |
47db06aa | 564 | |
47db06aa GM |
565 | ;;******************************************************************* |
566 | ;;** | |
567 | ;;** Substituting special XML sequences | |
568 | ;;** | |
569 | ;;******************************************************************* | |
570 | ||
a98e819b DL |
571 | (eval-when-compile |
572 | (defvar str)) ; dynamic from replace-regexp-in-string | |
573 | ||
574 | ;; Fixme: Take declared entities from the DTD when they're available. | |
575 | (defun xml-substitute-entity (match) | |
576 | "Subroutine of xml-substitute-special." | |
577 | (save-match-data | |
578 | (let ((match1 (match-string 1 str))) | |
579 | (cond ((string= match1 "lt") "<") | |
580 | ((string= match1 "gt") ">") | |
581 | ((string= match1 "apos") "'") | |
582 | ((string= match1 "quot") "\"") | |
583 | ((string= match1 "amp") "&") | |
584 | ((and (string-match "#\\([0-9]+\\)" match1) | |
585 | (let ((c (decode-char | |
586 | 'ucs | |
587 | (string-to-number (match-string 1 match1))))) | |
588 | (if c (string c))))) ; else unrepresentable | |
589 | ((and (string-match "#x\\([[:xdigit:]]+\\)" match1) | |
590 | (let ((c (decode-char | |
591 | 'ucs | |
592 | (string-to-number (match-string 1 match1) 16)))) | |
593 | (if c (string c))))) | |
594 | ;; Default to asis. Arguably, unrepresentable code points | |
595 | ;; might be best replaced with U+FFFD. | |
596 | (t match))))) | |
597 | ||
47db06aa | 598 | (defun xml-substitute-special (string) |
a98e819b DL |
599 | "Return STRING, after subsituting entity references." |
600 | ;; This originally made repeated passes through the string from the | |
601 | ;; beginning, which isn't correct, since then either "&amp;" or | |
602 | ;; "&amp;" won't DTRT. | |
603 | (replace-regexp-in-string "&\\([^;]+\\);" | |
604 | #'xml-substitute-entity string t t)) | |
47db06aa GM |
605 | |
606 | ;;******************************************************************* | |
607 | ;;** | |
608 | ;;** Printing a tree. | |
609 | ;;** This function is intended mainly for debugging purposes. | |
610 | ;;** | |
611 | ;;******************************************************************* | |
612 | ||
613 | (defun xml-debug-print (xml) | |
971489ea SM |
614 | (dolist (node xml) |
615 | (xml-debug-print-internal node ""))) | |
47db06aa | 616 | |
971489ea | 617 | (defun xml-debug-print-internal (xml indent-string) |
47db06aa | 618 | "Outputs the XML tree in the current buffer. |
a98e819b | 619 | The first line is indented with INDENT-STRING." |
47db06aa GM |
620 | (let ((tree xml) |
621 | attlist) | |
a98e819b | 622 | (insert indent-string ?< (symbol-name (xml-node-name tree))) |
524425ae | 623 | |
47db06aa | 624 | ;; output the attribute list |
971489ea | 625 | (setq attlist (xml-node-attributes tree)) |
47db06aa | 626 | (while attlist |
a98e819b | 627 | (insert ?\ (symbol-name (caar attlist)) "=\"" (cdar attlist) ?\") |
971489ea | 628 | (setq attlist (cdr attlist))) |
524425ae | 629 | |
a98e819b | 630 | (insert ?>) |
524425ae | 631 | |
971489ea | 632 | (setq tree (xml-node-children tree)) |
47db06aa GM |
633 | |
634 | ;; output the children | |
971489ea | 635 | (dolist (node tree) |
47db06aa | 636 | (cond |
971489ea | 637 | ((listp node) |
a98e819b | 638 | (insert ?\n) |
971489ea SM |
639 | (xml-debug-print-internal node (concat indent-string " "))) |
640 | ((stringp node) (insert node)) | |
47db06aa | 641 | (t |
971489ea | 642 | (error "Invalid XML tree")))) |
47db06aa | 643 | |
a98e819b DL |
644 | (insert ?\n indent-string |
645 | ?< ?/ (symbol-name (xml-node-name xml)) ?>))) | |
47db06aa GM |
646 | |
647 | (provide 'xml) | |
648 | ||
ab5796a9 | 649 | ;;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b |
47db06aa | 650 | ;;; xml.el ends here |