Mercurial > emacs
annotate lisp/xml.el @ 54694:253149f265f2
*** empty log message ***
author | Andreas Schwab <schwab@suse.de> |
---|---|
date | Sat, 03 Apr 2004 20:02:16 +0000 |
parents | 586ffda6e9f9 |
children | 1bf7ef48f54f |
rev | line source |
---|---|
38409
153f1b1f2efd
Emacs lisp coding convention fixes.
Pavel Janík <Pavel@Janik.cz>
parents:
37958
diff
changeset
|
1 ;;; xml.el --- XML parser |
30329 | 2 |
54243
586ffda6e9f9
(xml-get-attribute-or-nil): Simplify.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
53384
diff
changeset
|
3 ;; Copyright (C) 2000, 01, 03, 2004 Free Software Foundation, Inc. |
30329 | 4 |
5 ;; Author: Emmanuel Briot <briot@gnat.com> | |
51697 | 6 ;; Maintainer: Mark A. Hershberger <mah@everybody.org> |
51102 | 7 ;; Keywords: xml, data |
30329 | 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 | |
51102 | 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. | |
30329 | 31 |
32 ;;; FILE FORMAT | |
33 | |
51102 | 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. | |
30329 | 39 ;; |
51102 | 40 ;; It also knows how to ignore comments and processing instructions. |
30329 | 41 ;; |
42 ;; The XML file should have the following format: | |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
43 ;; <node1 attr1="name1" attr2="name2" ...>value |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
44 ;; <node2 attr3="name3" attr4="name4">value2</node2> |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
45 ;; <node3 attr5="name5" attr6="name6">value3</node3> |
30329 | 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 ;; | |
51102 | 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. | |
30329 | 70 |
71 ;;; Code: | |
72 | |
51102 | 73 ;; Note that {buffer-substring,match-string}-no-properties were |
74 ;; formerly used in several places, but that removes composition info. | |
75 | |
30329 | 76 ;;******************************************************************* |
77 ;;** | |
78 ;;** Macros to parse the list | |
79 ;;** | |
80 ;;******************************************************************* | |
81 | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
82 (defsubst xml-node-name (node) |
30329 | 83 "Return the tag associated with NODE. |
84 The tag is a lower-case symbol." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
85 (car node)) |
30329 | 86 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
87 (defsubst xml-node-attributes (node) |
30329 | 88 "Return the list of attributes of NODE. |
89 The list can be nil." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
90 (nth 1 node)) |
30329 | 91 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
92 (defsubst xml-node-children (node) |
30329 | 93 "Return the list of children of NODE. |
94 This is a list of nodes, and it can be nil." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
95 (cddr node)) |
30329 | 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." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
100 (let ((match ())) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
101 (dolist (child (xml-node-children node)) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
102 (if child |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
103 (if (equal (xml-node-name child) child-name) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
104 (push child match)))) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
105 (nreverse match))) |
30329 | 106 |
53375
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
107 (defun xml-get-attribute-or-nil (node attribute) |
30329 | 108 "Get from NODE the value of ATTRIBUTE. |
53384
6eee911caf68
(xml-get-attribute-or-nil): Doc fix.
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53377
diff
changeset
|
109 Return `nil' if the attribute was not found. |
53375
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
110 |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
111 See also `xml-get-attribute'." |
54243
586ffda6e9f9
(xml-get-attribute-or-nil): Simplify.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
53384
diff
changeset
|
112 (cdr (assoc attribute (xml-node-attributes node)))) |
53375
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
113 |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
114 (defsubst xml-get-attribute (node attribute) |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
115 "Get from NODE the value of ATTRIBUTE. |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
116 An empty string is returned if the attribute was not found. |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
117 |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
118 See also `xml-get-attribute-or-nil'." |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
119 (or (xml-get-attribute-or-nil node attribute) "")) |
30329 | 120 |
121 ;;******************************************************************* | |
122 ;;** | |
123 ;;** Creating the list | |
124 ;;** | |
125 ;;******************************************************************* | |
126 | |
51102 | 127 ;;;###autoload |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
128 (defun xml-parse-file (file &optional parse-dtd parse-ns) |
51102 | 129 "Parse the well-formed XML file FILE. |
130 If FILE is already visited, use its buffer and don't kill it. | |
30329 | 131 Returns the top node with all its children. |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
132 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped. |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
133 If PARSE-NS is non-nil, then QNAMES are expanded." |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
134 (let ((keep)) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
135 (if (get-file-buffer file) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
136 (progn |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
137 (set-buffer (get-file-buffer file)) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
138 (setq keep (point))) |
51102 | 139 (let (auto-mode-alist) ; no need for xml-mode |
140 (find-file file))) | |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
141 |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
142 (let ((xml (xml-parse-region (point-min) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
143 (point-max) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
144 (current-buffer) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
145 parse-dtd parse-ns))) |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
146 (if keep |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
147 (goto-char keep) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
148 (kill-buffer (current-buffer))) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
149 xml))) |
30329 | 150 |
51102 | 151 ;; Note that this is setup so that we can do whitespace-skipping with |
152 ;; `(skip-syntax-forward " ")', inter alia. Previously this was slow | |
153 ;; compared with `re-search-forward', but that has been fixed. Also | |
154 ;; note that the standard syntax table contains other characters with | |
155 ;; whitespace syntax, like NBSP, but they are invalid in contexts in | |
156 ;; which we might skip whitespace -- specifically, they're not | |
157 ;; NameChars [XML 4]. | |
158 | |
159 (defvar xml-syntax-table | |
160 (let ((table (make-syntax-table))) | |
161 ;; Get space syntax correct per XML [3]. | |
162 (dotimes (c 31) | |
163 (modify-syntax-entry c "." table)) ; all are space in standard table | |
164 (dolist (c '(?\t ?\n ?\r)) ; these should be space | |
165 (modify-syntax-entry c " " table)) | |
166 ;; For skipping attributes. | |
167 (modify-syntax-entry ?\" "\"" table) | |
168 (modify-syntax-entry ?' "\"" table) | |
169 ;; Non-alnum name chars should be symbol constituents (`-' and `_' | |
170 ;; are OK by default). | |
171 (modify-syntax-entry ?. "_" table) | |
172 (modify-syntax-entry ?: "_" table) | |
173 ;; XML [89] | |
174 (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005 | |
175 #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC | |
176 #x30FD #x30FE)) | |
177 (modify-syntax-entry (decode-char 'ucs c) "w" table)) | |
178 ;; Fixme: rest of [4] | |
179 table) | |
180 "Syntax table used by `xml-parse-region'.") | |
181 | |
182 ;; XML [5] | |
183 ;; Note that [:alpha:] matches all multibyte chars with word syntax. | |
51105
aac5eaf1454e
(xml-name-regexp): Wrap in `eval-and-compile'.
John Paul Wallington <jpw@pobox.com>
parents:
51102
diff
changeset
|
184 (eval-and-compile |
aac5eaf1454e
(xml-name-regexp): Wrap in `eval-and-compile'.
John Paul Wallington <jpw@pobox.com>
parents:
51102
diff
changeset
|
185 (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*")) |
51102 | 186 |
187 ;; Fixme: This needs re-writing to deal with the XML grammar properly, i.e. | |
188 ;; document ::= prolog element Misc* | |
189 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? | |
190 | |
191 ;;;###autoload | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
192 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns) |
30329 | 193 "Parse the region from BEG to END in BUFFER. |
194 If BUFFER is nil, it defaults to the current buffer. | |
195 Returns the XML list for the region, or raises an error if the region | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
196 is not well-formed XML. |
30329 | 197 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped, |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
198 and returned as the first element of the list. |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
199 If PARSE-NS is non-nil, then QNAMES are expanded." |
51102 | 200 (save-restriction |
201 (narrow-to-region beg end) | |
202 ;; Use fixed syntax table to ensure regexp char classes and syntax | |
203 ;; specs DTRT. | |
204 (with-syntax-table (standard-syntax-table) | |
205 (let ((case-fold-search nil) ; XML is case-sensitive. | |
206 xml result dtd) | |
207 (save-excursion | |
208 (if buffer | |
209 (set-buffer buffer)) | |
210 (goto-char (point-min)) | |
211 (while (not (eobp)) | |
212 (if (search-forward "<" nil t) | |
213 (progn | |
214 (forward-char -1) | |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
215 (setq result (xml-parse-tag parse-dtd parse-ns)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
216 (if (and xml result) |
51102 | 217 ;; translation of rule [1] of XML specifications |
218 (error "XML files can have only one toplevel tag") | |
30329 | 219 (cond |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
220 ((null result)) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
221 ((and (listp (car result)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
222 parse-dtd) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
223 (setq dtd (car result)) |
51102 | 224 (if (cdr result) ; possible leading comment |
225 (add-to-list 'xml (cdr result)))) | |
30329 | 226 (t |
51102 | 227 (add-to-list 'xml result))))) |
228 (goto-char (point-max)))) | |
229 (if parse-dtd | |
230 (cons dtd (nreverse xml)) | |
231 (nreverse xml))))))) | |
30329 | 232 |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
233 (defun xml-ns-parse-ns-attrs (attr-list &optional xml-ns) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
234 "Parse the namespace attributes and return a list of cons in the form: |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
235 \(namespace . prefix)" |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
236 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
237 (mapcar |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
238 (lambda (attr) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
239 (let* ((splitup (split-string (car attr) ":")) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
240 (prefix (nth 0 splitup)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
241 (lname (nth 1 splitup))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
242 (when (string= "xmlns" prefix) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
243 (push (cons (if lname |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
244 lname |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
245 "") |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
246 (cdr attr)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
247 xml-ns)))) attr-list) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
248 xml-ns) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
249 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
250 ;; expand element names |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
251 (defun xml-ns-expand-el (el xml-ns) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
252 "Expand the XML elements from \"prefix:local-name\" to a cons in the form |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
253 \"(namespace . local-name)\"." |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
254 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
255 (let* ((splitup (split-string el ":")) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
256 (lname (or (nth 1 splitup) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
257 (nth 0 splitup))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
258 (prefix (if (nth 1 splitup) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
259 (nth 0 splitup) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
260 (if (string= lname "xmlns") |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
261 "xmlns" |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
262 ""))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
263 (ns (cdr (assoc-string prefix xml-ns)))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
264 (if (string= "" ns) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
265 lname |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
266 (cons (intern (concat ":" ns)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
267 lname)))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
268 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
269 ;; expand attribute names |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
270 (defun xml-ns-expand-attr (attr-list xml-ns) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
271 "Expand the attribute list for a particular element from the form |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
272 \"prefix:local-name\" to the form \"{namespace}:local-name\"." |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
273 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
274 (mapcar |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
275 (lambda (attr) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
276 (let* ((splitup (split-string (car attr) ":")) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
277 (lname (or (nth 1 splitup) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
278 (nth 0 splitup))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
279 (prefix (if (nth 1 splitup) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
280 (nth 0 splitup) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
281 (if (string= (car attr) "xmlns") |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
282 "xmlns" |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
283 ""))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
284 (ns (cdr (assoc-string prefix xml-ns)))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
285 (setcar attr |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
286 (if (string= "" ns) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
287 lname |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
288 (cons (intern (concat ":" ns)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
289 lname))))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
290 attr-list) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
291 attr-list) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
292 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
293 (defun xml-intern-attrlist (attr-list) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
294 "Convert attribute names to symbols for backward compatibility." |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
295 (mapcar (lambda (attr) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
296 (setcar attr (intern (car attr)))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
297 attr-list) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
298 attr-list) |
30329 | 299 |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
300 (defun xml-parse-tag (&optional parse-dtd parse-ns) |
51102 | 301 "Parse the tag at point. |
30329 | 302 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and |
303 returned as the first element in the list. | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
304 If PARSE-NS is non-nil, then QNAMES are expanded. |
30329 | 305 Returns one of: |
51102 | 306 - a list : the matching node |
307 - nil : the point is not looking at a tag. | |
308 - a pair : the first element is the DTD, the second is the node." | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
309 (let ((xml-ns (if (consp parse-ns) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
310 parse-ns |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
311 (if parse-ns |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
312 (list |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
313 ;; Default no namespace |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
314 (cons "" "") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
315 ;; We need to seed the xmlns namespace |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
316 (cons "xmlns" "http://www.w3.org/2000/xmlns/")))))) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
317 (cond |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
318 ;; Processing instructions (like the <?xml version="1.0"?> tag at the |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
319 ;; beginning of a document). |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
320 ((looking-at "<\\?") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
321 (search-forward "?>") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
322 (skip-syntax-forward " ") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
323 (xml-parse-tag parse-dtd xml-ns)) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
324 ;; Character data (CDATA) sections, in which no tag should be interpreted |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
325 ((looking-at "<!\\[CDATA\\[") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
326 (let ((pos (match-end 0))) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
327 (unless (search-forward "]]>" nil t) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
328 (error "CDATA section does not end anywhere in the document")) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
329 (buffer-substring pos (match-beginning 0)))) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
330 ;; DTD for the document |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
331 ((looking-at "<!DOCTYPE") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
332 (let (dtd) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
333 (if parse-dtd |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
334 (setq dtd (xml-parse-dtd)) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
335 (xml-skip-dtd)) |
51102 | 336 (skip-syntax-forward " ") |
30329 | 337 (if dtd |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
338 (cons dtd (xml-parse-tag nil xml-ns)) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
339 (xml-parse-tag nil xml-ns)))) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
340 ;; skip comments |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
341 ((looking-at "<!--") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
342 (search-forward "-->") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
343 nil) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
344 ;; end tag |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
345 ((looking-at "</") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
346 '()) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
347 ;; opening tag |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
348 ((looking-at "<\\([^/>[:space:]]+\\)") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
349 (goto-char (match-end 1)) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
350 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
351 ;; Parse this node |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
352 (let* ((node-name (match-string 1)) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
353 (attr-list (xml-parse-attlist)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
354 (children (if (consp xml-ns) ;; take care of namespace parsing |
53377
e4e98d69d87a
Fix previous change. Remove redundant trailing whitespace.
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53375
diff
changeset
|
355 (progn |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
356 (setq xml-ns (xml-ns-parse-ns-attrs |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
357 attr-list xml-ns)) |
53377
e4e98d69d87a
Fix previous change. Remove redundant trailing whitespace.
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53375
diff
changeset
|
358 (list (xml-ns-expand-attr |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
359 attr-list xml-ns) |
53377
e4e98d69d87a
Fix previous change. Remove redundant trailing whitespace.
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53375
diff
changeset
|
360 (xml-ns-expand-el |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
361 node-name xml-ns))) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
362 (list (xml-intern-attrlist attr-list) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
363 (intern node-name)))) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
364 pos) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
365 |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
366 ;; is this an empty element ? |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
367 (if (looking-at "/>") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
368 (progn |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
369 (forward-char 2) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
370 (nreverse children)) |
30329 | 371 |
372 ;; is this a valid start tag ? | |
40030
7507bd185307
(xml-parse-tag): Use eq on char-after's return value.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
39407
diff
changeset
|
373 (if (eq (char-after) ?>) |
30329 | 374 (progn |
375 (forward-char 1) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
376 ;; Now check that we have the right end-tag. Note that this |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
377 ;; one might contain spaces after the tag name |
51102 | 378 (let ((end (concat "</" node-name "\\s-*>"))) |
379 (while (not (looking-at end)) | |
380 (cond | |
381 ((looking-at "</") | |
382 (error "XML: Invalid end tag (expecting %s) at pos %d" | |
383 node-name (point))) | |
384 ((= (char-after) ?<) | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
385 (let ((tag (xml-parse-tag nil xml-ns))) |
51102 | 386 (when tag |
387 (push tag children)))) | |
388 (t | |
389 (setq pos (point)) | |
390 (search-forward "<") | |
391 (forward-char -1) | |
392 (let ((string (buffer-substring pos (point))) | |
393 (pos 0)) | |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
394 |
51102 | 395 ;; Clean up the string. As per XML |
396 ;; specifications, the XML processor should | |
397 ;; always pass the whole string to the | |
398 ;; application. But \r's should be replaced: | |
399 ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends | |
400 (while (string-match "\r\n?" string pos) | |
401 (setq string (replace-match "\n" t t string)) | |
402 (setq pos (1+ (match-beginning 0)))) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
403 |
51102 | 404 (setq string (xml-substitute-special string)) |
405 (setq children | |
406 (if (stringp (car children)) | |
407 ;; The two strings were separated by a comment. | |
408 (cons (concat (car children) string) | |
409 (cdr children)) | |
410 (cons string children)))))))) | |
411 | |
30329 | 412 (goto-char (match-end 0)) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
413 (nreverse children)) |
30329 | 414 ;; This was an invalid start tag |
51102 | 415 (error "XML: Invalid attribute list"))))) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
416 (t ;; This is not a tag. |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
417 (error "XML: Invalid character"))))) |
30329 | 418 |
51102 | 419 (defun xml-parse-attlist () |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
420 "Return the attribute-list after point. Leave point at the |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
421 first non-blank character after the tag." |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
422 (let ((attlist ()) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
423 end-pos name) |
51102 | 424 (skip-syntax-forward " ") |
425 (while (looking-at (eval-when-compile | |
426 (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*"))) | |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
427 (setq name (match-string 1)) |
30329 | 428 (goto-char (match-end 0)) |
429 | |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
430 ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
431 |
30329 | 432 ;; Do we have a string between quotes (or double-quotes), |
433 ;; or a simple word ? | |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
434 (if (looking-at "\"\\([^\"]*\\)\"") |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
435 (setq end-pos (match-end 0)) |
50210
575aa6820adc
(xml-parse-attlist): typo in attribute parsing.
Juanma Barranquero <lekktu@gmail.com>
parents:
50144
diff
changeset
|
436 (if (looking-at "'\\([^']*\\)'") |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
437 (setq end-pos (match-end 0)) |
38409
153f1b1f2efd
Emacs lisp coding convention fixes.
Pavel Janík <Pavel@Janik.cz>
parents:
37958
diff
changeset
|
438 (error "XML: Attribute values must be given between quotes"))) |
30329 | 439 |
440 ;; Each attribute must be unique within a given element | |
441 (if (assoc name attlist) | |
38409
153f1b1f2efd
Emacs lisp coding convention fixes.
Pavel Janík <Pavel@Janik.cz>
parents:
37958
diff
changeset
|
442 (error "XML: each attribute must be unique within an element")) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
443 |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
444 ;; Multiple whitespace characters should be replaced with a single one |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
445 ;; in the attributes |
51102 | 446 (let ((string (match-string 1)) |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
447 (pos 0)) |
51102 | 448 (replace-regexp-in-string "\\s-\\{2,\\}" " " string) |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
449 (push (cons name (xml-substitute-special string)) attlist)) |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
450 |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
451 (goto-char end-pos) |
51102 | 452 (skip-syntax-forward " ")) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
453 (nreverse attlist))) |
30329 | 454 |
455 ;;******************************************************************* | |
456 ;;** | |
457 ;;** The DTD (document type declaration) | |
458 ;;** The following functions know how to skip or parse the DTD of | |
459 ;;** a document | |
460 ;;** | |
461 ;;******************************************************************* | |
462 | |
51102 | 463 ;; Fixme: This fails at least if the DTD contains conditional sections. |
464 | |
465 (defun xml-skip-dtd () | |
466 "Skip the DTD at point. | |
30329 | 467 This follows the rule [28] in the XML specifications." |
468 (forward-char (length "<!DOCTYPE")) | |
51102 | 469 (if (looking-at "\\s-*>") |
30329 | 470 (error "XML: invalid DTD (excepting name of the document)")) |
471 (condition-case nil | |
472 (progn | |
51102 | 473 (forward-sexp) |
474 (skip-syntax-forward " ") | |
30329 | 475 (if (looking-at "\\[") |
51102 | 476 (re-search-forward "]\\s-*>") |
477 (search-forward ">"))) | |
30329 | 478 (error (error "XML: No end to the DTD")))) |
479 | |
51102 | 480 (defun xml-parse-dtd () |
481 "Parse the DTD at point." | |
482 (forward-char (eval-when-compile (length "<!DOCTYPE"))) | |
483 (skip-syntax-forward " ") | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
484 (if (looking-at ">") |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
485 (error "XML: invalid DTD (excepting name of the document)")) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
486 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
487 ;; Get the name of the document |
51102 | 488 (looking-at xml-name-regexp) |
489 (let ((dtd (list (match-string 0) 'dtd)) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
490 type element end-pos) |
30329 | 491 (goto-char (match-end 0)) |
492 | |
51102 | 493 (skip-syntax-forward " ") |
494 ;; XML [75] | |
495 (cond ((looking-at "PUBLIC\\s-+") | |
496 (goto-char (match-end 0)) | |
497 (unless (or (re-search-forward | |
498 "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\"" | |
499 nil t) | |
500 (re-search-forward | |
501 "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'" | |
502 nil t)) | |
503 (error "XML: missing public id")) | |
504 (let ((pubid (match-string 1))) | |
505 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t) | |
506 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t)) | |
507 (error "XML: missing system id")) | |
508 (push (list pubid (match-string 1) 'public) dtd))) | |
509 ((looking-at "SYSTEM\\s-+") | |
510 (goto-char (match-end 0)) | |
511 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t) | |
512 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t)) | |
513 (error "XML: missing system id")) | |
514 (push (list (match-string 1) 'system) dtd))) | |
515 (skip-syntax-forward " ") | |
516 (if (eq ?> (char-after)) | |
517 (forward-char) | |
518 (skip-syntax-forward " ") | |
519 (if (not (eq (char-after) ?\[)) | |
520 (error "XML: bad DTD") | |
521 (forward-char) | |
522 ;; Parse the rest of the DTD | |
523 ;; Fixme: Deal with ENTITY, ATTLIST, NOTATION, PIs. | |
524 (while (not (looking-at "\\s-*\\]")) | |
525 (skip-syntax-forward " ") | |
526 (cond | |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
527 |
51102 | 528 ;; Translation of rule [45] of XML specifications |
529 ((looking-at | |
530 "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>") | |
531 | |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
532 (setq element (match-string 1) |
51102 | 533 type (match-string-no-properties 2)) |
534 (setq end-pos (match-end 0)) | |
30329 | 535 |
51102 | 536 ;; Translation of rule [46] of XML specifications |
537 (cond | |
538 ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration | |
539 (setq type 'empty)) | |
540 ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents | |
541 (setq type 'any)) | |
542 ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47]) | |
543 (setq type (xml-parse-elem-type (match-string 1 type)))) | |
544 ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution | |
545 nil) | |
546 (t | |
547 (error "XML: Invalid element type in the DTD"))) | |
30329 | 548 |
51102 | 549 ;; rule [45]: the element declaration must be unique |
550 (if (assoc element dtd) | |
551 (error "XML: element declarations must be unique in a DTD (<%s>)" | |
53011
45bd3ad34572
(xml-parse-dtd): Fix misplaced paren.
Andreas Schwab <schwab@suse.de>
parents:
52975
diff
changeset
|
552 element)) |
30329 | 553 |
51102 | 554 ;; Store the element in the DTD |
555 (push (list element type) dtd) | |
556 (goto-char end-pos)) | |
557 ((looking-at "<!--") | |
558 (search-forward "-->")) | |
30329 | 559 |
51102 | 560 (t |
561 (error "XML: Invalid DTD item"))) | |
562 | |
563 ;; Skip the end of the DTD | |
564 (search-forward ">")))) | |
53011
45bd3ad34572
(xml-parse-dtd): Fix misplaced paren.
Andreas Schwab <schwab@suse.de>
parents:
52975
diff
changeset
|
565 (nreverse dtd))) |
30329 | 566 |
567 (defun xml-parse-elem-type (string) | |
51102 | 568 "Convert element type STRING into a Lisp structure." |
30329 | 569 |
570 (let (elem modifier) | |
571 (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string) | |
572 (progn | |
573 (setq elem (match-string 1 string) | |
574 modifier (match-string 2 string)) | |
575 (if (string-match "|" elem) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
576 (setq elem (cons 'choice |
30329 | 577 (mapcar 'xml-parse-elem-type |
578 (split-string elem "|")))) | |
579 (if (string-match "," elem) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
580 (setq elem (cons 'seq |
30329 | 581 (mapcar 'xml-parse-elem-type |
51102 | 582 (split-string elem ","))))))) |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
583 (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string) |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
584 (setq elem (match-string 1 string) |
30329 | 585 modifier (match-string 2 string)))) |
586 | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
587 (if (and (stringp elem) (string= elem "#PCDATA")) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
588 (setq elem 'pcdata)) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
589 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
590 (cond |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
591 ((string= modifier "+") |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
592 (list '+ elem)) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
593 ((string= modifier "*") |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
594 (list '* elem)) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
595 ((string= modifier "?") |
49787
6269b5c10aec
(xml-parse-elem-type): Fix use of character constant.
Juanma Barranquero <lekktu@gmail.com>
parents:
49133
diff
changeset
|
596 (list '\? elem)) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
597 (t |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
598 elem)))) |
30329 | 599 |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
600 ;;******************************************************************* |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
601 ;;** |
30329 | 602 ;;** Substituting special XML sequences |
603 ;;** | |
604 ;;******************************************************************* | |
605 | |
51102 | 606 (eval-when-compile |
607 (defvar str)) ; dynamic from replace-regexp-in-string | |
608 | |
609 ;; Fixme: Take declared entities from the DTD when they're available. | |
610 (defun xml-substitute-entity (match) | |
611 "Subroutine of xml-substitute-special." | |
612 (save-match-data | |
613 (let ((match1 (match-string 1 str))) | |
614 (cond ((string= match1 "lt") "<") | |
615 ((string= match1 "gt") ">") | |
616 ((string= match1 "apos") "'") | |
617 ((string= match1 "quot") "\"") | |
618 ((string= match1 "amp") "&") | |
619 ((and (string-match "#\\([0-9]+\\)" match1) | |
620 (let ((c (decode-char | |
621 'ucs | |
622 (string-to-number (match-string 1 match1))))) | |
623 (if c (string c))))) ; else unrepresentable | |
624 ((and (string-match "#x\\([[:xdigit:]]+\\)" match1) | |
625 (let ((c (decode-char | |
626 'ucs | |
627 (string-to-number (match-string 1 match1) 16)))) | |
628 (if c (string c))))) | |
629 ;; Default to asis. Arguably, unrepresentable code points | |
630 ;; might be best replaced with U+FFFD. | |
631 (t match))))) | |
632 | |
30329 | 633 (defun xml-substitute-special (string) |
51102 | 634 "Return STRING, after subsituting entity references." |
635 ;; This originally made repeated passes through the string from the | |
636 ;; beginning, which isn't correct, since then either "&amp;" or | |
637 ;; "&amp;" won't DTRT. | |
638 (replace-regexp-in-string "&\\([^;]+\\);" | |
639 #'xml-substitute-entity string t t)) | |
30329 | 640 |
641 ;;******************************************************************* | |
642 ;;** | |
643 ;;** Printing a tree. | |
644 ;;** This function is intended mainly for debugging purposes. | |
645 ;;** | |
646 ;;******************************************************************* | |
647 | |
648 (defun xml-debug-print (xml) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
649 (dolist (node xml) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
650 (xml-debug-print-internal node ""))) |
30329 | 651 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
652 (defun xml-debug-print-internal (xml indent-string) |
30329 | 653 "Outputs the XML tree in the current buffer. |
51102 | 654 The first line is indented with INDENT-STRING." |
30329 | 655 (let ((tree xml) |
656 attlist) | |
51102 | 657 (insert indent-string ?< (symbol-name (xml-node-name tree))) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
658 |
30329 | 659 ;; output the attribute list |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
660 (setq attlist (xml-node-attributes tree)) |
30329 | 661 (while attlist |
51102 | 662 (insert ?\ (symbol-name (caar attlist)) "=\"" (cdar attlist) ?\") |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
663 (setq attlist (cdr attlist))) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
664 |
51102 | 665 (insert ?>) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
666 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
667 (setq tree (xml-node-children tree)) |
30329 | 668 |
669 ;; output the children | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
670 (dolist (node tree) |
30329 | 671 (cond |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
672 ((listp node) |
51102 | 673 (insert ?\n) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
674 (xml-debug-print-internal node (concat indent-string " "))) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
675 ((stringp node) (insert node)) |
30329 | 676 (t |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
677 (error "Invalid XML tree")))) |
30329 | 678 |
51102 | 679 (insert ?\n indent-string |
680 ?< ?/ (symbol-name (xml-node-name xml)) ?>))) | |
30329 | 681 |
682 (provide 'xml) | |
683 | |
52401 | 684 ;;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b |
30329 | 685 ;;; xml.el ends here |