38409
|
1 ;;; xml.el --- XML parser
|
30329
|
2
|
64762
|
3 ;; Copyright (C) 2000, 2001, 2002, 2003, 2004,
|
75347
|
4 ;; 2005, 2006, 2007 Free Software Foundation, Inc.
|
30329
|
5
|
|
6 ;; Author: Emmanuel Briot <briot@gnat.com>
|
51697
|
7 ;; Maintainer: Mark A. Hershberger <mah@everybody.org>
|
51102
|
8 ;; Keywords: xml, data
|
30329
|
9
|
|
10 ;; This file is part of GNU Emacs.
|
|
11
|
|
12 ;; GNU Emacs is free software; you can redistribute it and/or modify
|
|
13 ;; it under the terms of the GNU General Public License as published by
|
78236
|
14 ;; the Free Software Foundation; either version 3, or (at your option)
|
30329
|
15 ;; any later version.
|
|
16
|
|
17 ;; GNU Emacs is distributed in the hope that it will be useful,
|
|
18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
20 ;; GNU General Public License for more details.
|
|
21
|
|
22 ;; You should have received a copy of the GNU General Public License
|
|
23 ;; along with GNU Emacs; see the file COPYING. If not, write to the
|
64091
|
24 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
|
|
25 ;; Boston, MA 02110-1301, USA.
|
30329
|
26
|
|
27 ;;; Commentary:
|
|
28
|
51102
|
29 ;; This file contains a somewhat incomplete non-validating XML parser. It
|
|
30 ;; parses a file, and returns a list that can be used internally by
|
54937
|
31 ;; any other Lisp libraries.
|
30329
|
32
|
|
33 ;;; FILE FORMAT
|
|
34
|
51102
|
35 ;; The document type declaration may either be ignored or (optionally)
|
|
36 ;; parsed, but currently the parsing will only accept element
|
54937
|
37 ;; declarations. The XML file is assumed to be well-formed. In case
|
51102
|
38 ;; of error, the parsing stops and the XML file is shown where the
|
|
39 ;; parsing stopped.
|
30329
|
40 ;;
|
51102
|
41 ;; It also knows how to ignore comments and processing instructions.
|
30329
|
42 ;;
|
|
43 ;; The XML file should have the following format:
|
34825
|
44 ;; <node1 attr1="name1" attr2="name2" ...>value
|
|
45 ;; <node2 attr3="name3" attr4="name4">value2</node2>
|
|
46 ;; <node3 attr5="name5" attr6="name6">value3</node3>
|
30329
|
47 ;; </node1>
|
54937
|
48 ;; Of course, the name of the nodes and attributes can be anything. There can
|
30329
|
49 ;; be any number of attributes (or none), as well as any number of children
|
|
50 ;; below the nodes.
|
|
51 ;;
|
|
52 ;; There can be only top level node, but with any number of children below.
|
|
53
|
|
54 ;;; LIST FORMAT
|
|
55
|
54877
|
56 ;; The functions `xml-parse-file', `xml-parse-region' and
|
|
57 ;; `xml-parse-tag' return a list with the following format:
|
30329
|
58 ;;
|
|
59 ;; xml-list ::= (node node ...)
|
54877
|
60 ;; node ::= (qname attribute-list . child_node_list)
|
30329
|
61 ;; child_node_list ::= child_node child_node ...
|
|
62 ;; child_node ::= node | string
|
54877
|
63 ;; qname ::= (:namespace-uri . "name") | "name"
|
|
64 ;; attribute_list ::= ((qname . "value") (qname . "value") ...)
|
30329
|
65 ;; | nil
|
|
66 ;; string ::= "..."
|
|
67 ;;
|
51102
|
68 ;; Some macros are provided to ease the parsing of this list.
|
|
69 ;; Whitespace is preserved. Fixme: There should be a tree-walker that
|
|
70 ;; can remove it.
|
30329
|
71
|
54877
|
72 ;; TODO:
|
|
73 ;; * xml:base, xml:space support
|
|
74 ;; * more complete DOCTYPE parsing
|
|
75 ;; * pi support
|
|
76
|
30329
|
77 ;;; Code:
|
|
78
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
79 ;; Note that buffer-substring and match-string were formerly used in
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
80 ;; several places, because the -no-properties variants remove
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
81 ;; composition info. However, after some discussion on emacs-devel,
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
82 ;; the consensus was that the speed of the -no-properties variants was
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
83 ;; a worthwhile tradeoff especially since we're usually parsing files
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
84 ;; instead of hand-crafted XML.
|
51102
|
85
|
30329
|
86 ;;*******************************************************************
|
|
87 ;;**
|
|
88 ;;** Macros to parse the list
|
|
89 ;;**
|
|
90 ;;*******************************************************************
|
|
91
|
62752
|
92 (defconst xml-undefined-entity "?"
|
|
93 "What to substitute for undefined entities")
|
|
94
|
56375
|
95 (defvar xml-entity-alist
|
|
96 '(("lt" . "<")
|
|
97 ("gt" . ">")
|
|
98 ("apos" . "'")
|
|
99 ("quot" . "\"")
|
|
100 ("amp" . "&"))
|
|
101 "The defined entities. Entities are added to this when the DTD is parsed.")
|
|
102
|
|
103 (defvar xml-sub-parser nil
|
|
104 "Dynamically set this to a non-nil value if you want to parse an XML fragment.")
|
|
105
|
|
106 (defvar xml-validating-parser nil
|
|
107 "Set to non-nil to get validity checking.")
|
|
108
|
42031
|
109 (defsubst xml-node-name (node)
|
30329
|
110 "Return the tag associated with NODE.
|
54937
|
111 Without namespace-aware parsing, the tag is a symbol.
|
|
112
|
|
113 With namespace-aware parsing, the tag is a cons of a string
|
|
114 representing the uri of the namespace with the local name of the
|
|
115 tag. For example,
|
|
116
|
|
117 <foo>
|
|
118
|
|
119 would be represented by
|
|
120
|
|
121 '(\"\" . \"foo\")."
|
|
122
|
42031
|
123 (car node))
|
30329
|
124
|
42031
|
125 (defsubst xml-node-attributes (node)
|
30329
|
126 "Return the list of attributes of NODE.
|
|
127 The list can be nil."
|
42031
|
128 (nth 1 node))
|
30329
|
129
|
42031
|
130 (defsubst xml-node-children (node)
|
30329
|
131 "Return the list of children of NODE.
|
|
132 This is a list of nodes, and it can be nil."
|
42031
|
133 (cddr node))
|
30329
|
134
|
|
135 (defun xml-get-children (node child-name)
|
|
136 "Return the children of NODE whose tag is CHILD-NAME.
|
54937
|
137 CHILD-NAME should match the value returned by `xml-node-name'."
|
42031
|
138 (let ((match ()))
|
|
139 (dolist (child (xml-node-children node))
|
54937
|
140 (if (and (listp child)
|
|
141 (equal (xml-node-name child) child-name))
|
|
142 (push child match)))
|
42031
|
143 (nreverse match)))
|
30329
|
144
|
53375
|
145 (defun xml-get-attribute-or-nil (node attribute)
|
30329
|
146 "Get from NODE the value of ATTRIBUTE.
|
54937
|
147 Return nil if the attribute was not found.
|
53375
|
148
|
|
149 See also `xml-get-attribute'."
|
54243
|
150 (cdr (assoc attribute (xml-node-attributes node))))
|
53375
|
151
|
|
152 (defsubst xml-get-attribute (node attribute)
|
|
153 "Get from NODE the value of ATTRIBUTE.
|
|
154 An empty string is returned if the attribute was not found.
|
|
155
|
|
156 See also `xml-get-attribute-or-nil'."
|
|
157 (or (xml-get-attribute-or-nil node attribute) ""))
|
30329
|
158
|
|
159 ;;*******************************************************************
|
|
160 ;;**
|
|
161 ;;** Creating the list
|
|
162 ;;**
|
|
163 ;;*******************************************************************
|
|
164
|
51102
|
165 ;;;###autoload
|
51930
|
166 (defun xml-parse-file (file &optional parse-dtd parse-ns)
|
51102
|
167 "Parse the well-formed XML file FILE.
|
|
168 If FILE is already visited, use its buffer and don't kill it.
|
30329
|
169 Returns the top node with all its children.
|
51930
|
170 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped.
|
|
171 If PARSE-NS is non-nil, then QNAMES are expanded."
|
72108
|
172 (if (get-file-buffer file)
|
|
173 (with-current-buffer (get-file-buffer file)
|
|
174 (save-excursion
|
|
175 (xml-parse-region (point-min)
|
|
176 (point-max)
|
|
177 (current-buffer)
|
|
178 parse-dtd parse-ns)))
|
|
179 (with-temp-buffer
|
|
180 (insert-file-contents file)
|
|
181 (xml-parse-region (point-min)
|
|
182 (point-max)
|
|
183 (current-buffer)
|
|
184 parse-dtd parse-ns))))
|
30329
|
185
|
56375
|
186
|
58939
|
187 (defvar xml-name-re)
|
|
188 (defvar xml-entity-value-re)
|
64794
|
189 (defvar xml-att-def-re)
|
58724
|
190 (let* ((start-chars (concat "[:alpha:]:_"))
|
56375
|
191 (name-chars (concat "-[:digit:]." start-chars))
|
68534
|
192 ;;[3] S ::= (#x20 | #x9 | #xD | #xA)+
|
56375
|
193 (whitespace "[ \t\n\r]"))
|
68534
|
194 ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6]
|
|
195 ;; | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF]
|
|
196 ;; | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF]
|
|
197 ;; | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
|
56375
|
198 (defvar xml-name-start-char-re (concat "[" start-chars "]"))
|
68534
|
199 ;;[4a] NameChar ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
|
56375
|
200 (defvar xml-name-char-re (concat "[" name-chars "]"))
|
68534
|
201 ;;[5] Name ::= NameStartChar (NameChar)*
|
56375
|
202 (defvar xml-name-re (concat xml-name-start-char-re xml-name-char-re "*"))
|
68534
|
203 ;;[6] Names ::= Name (#x20 Name)*
|
56375
|
204 (defvar xml-names-re (concat xml-name-re "\\(?: " xml-name-re "\\)*"))
|
68534
|
205 ;;[7] Nmtoken ::= (NameChar)+
|
56375
|
206 (defvar xml-nmtoken-re (concat xml-name-char-re "+"))
|
68534
|
207 ;;[8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
|
56375
|
208 (defvar xml-nmtokens-re (concat xml-nmtoken-re "\\(?: " xml-name-re "\\)*"))
|
68534
|
209 ;;[66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
|
56375
|
210 (defvar xml-char-ref-re "\\(?:&#[0-9]+;\\|&#x[0-9a-fA-F]+;\\)")
|
68534
|
211 ;;[68] EntityRef ::= '&' Name ';'
|
56375
|
212 (defvar xml-entity-ref (concat "&" xml-name-re ";"))
|
68534
|
213 ;;[69] PEReference ::= '%' Name ';'
|
56375
|
214 (defvar xml-pe-reference-re (concat "%" xml-name-re ";"))
|
68534
|
215 ;;[67] Reference ::= EntityRef | CharRef
|
56375
|
216 (defvar xml-reference-re (concat "\\(?:" xml-entity-ref "\\|" xml-char-ref-re "\\)"))
|
68534
|
217 ;;[10] AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'"
|
63282
|
218 (defvar xml-att-value-re (concat "\\(?:\"\\(?:[^&\"]\\|" xml-reference-re "\\)*\"\\|"
|
|
219 "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)"))
|
68534
|
220 ;;[56] TokenizedType ::= 'ID' [VC: ID] [VC: One ID per Element Type] [VC: ID Attribute Default]
|
|
221 ;; | 'IDREF' [VC: IDREF]
|
|
222 ;; | 'IDREFS' [VC: IDREF]
|
|
223 ;; | 'ENTITY' [VC: Entity Name]
|
|
224 ;; | 'ENTITIES' [VC: Entity Name]
|
|
225 ;; | 'NMTOKEN' [VC: Name Token]
|
|
226 ;; | 'NMTOKENS' [VC: Name Token]
|
63282
|
227 (defvar xml-tokenized-type-re "\\(?:ID\\|IDREF\\|IDREFS\\|ENTITY\\|ENTITIES\\|NMTOKEN\\|NMTOKENS\\)")
|
68534
|
228 ;;[58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
|
63282
|
229 (defvar xml-notation-type-re (concat "\\(?:NOTATION" whitespace "(" whitespace "*" xml-name-re
|
|
230 "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)"))
|
68534
|
231 ;;[59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' [VC: Enumeration] [VC: No Duplicate Tokens]
|
63282
|
232 (defvar xml-enumeration-re (concat "\\(?:(" whitespace "*" xml-nmtoken-re
|
|
233 "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*"
|
|
234 whitespace ")\\)"))
|
68534
|
235 ;;[57] EnumeratedType ::= NotationType | Enumeration
|
63282
|
236 (defvar xml-enumerated-type-re (concat "\\(?:" xml-notation-type-re "\\|" xml-enumeration-re "\\)"))
|
68534
|
237 ;;[54] AttType ::= StringType | TokenizedType | EnumeratedType
|
|
238 ;;[55] StringType ::= 'CDATA'
|
63282
|
239 (defvar xml-att-type-re (concat "\\(?:CDATA\\|" xml-tokenized-type-re "\\|" xml-notation-type-re"\\|" xml-enumerated-type-re "\\)"))
|
68534
|
240 ;;[60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
|
63282
|
241 (defvar xml-default-decl-re (concat "\\(?:#REQUIRED\\|#IMPLIED\\|\\(?:#FIXED" whitespace "\\)*" xml-att-value-re "\\)"))
|
68534
|
242 ;;[53] AttDef ::= S Name S AttType S DefaultDecl
|
63282
|
243 (defvar xml-att-def-re (concat "\\(?:" whitespace "*" xml-name-re
|
|
244 whitespace "*" xml-att-type-re
|
|
245 whitespace "*" xml-default-decl-re "\\)"))
|
68534
|
246 ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
|
|
247 ;; | "'" ([^%&'] | PEReference | Reference)* "'"
|
56375
|
248 (defvar xml-entity-value-re (concat "\\(?:\"\\(?:[^%&\"]\\|" xml-pe-reference-re
|
|
249 "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|"
|
|
250 xml-pe-reference-re "\\|" xml-reference-re "\\)*'\\)")))
|
|
251 ;;[75] ExternalID ::= 'SYSTEM' S SystemLiteral
|
|
252 ;; | 'PUBLIC' S PubidLiteral S SystemLiteral
|
|
253 ;;[76] NDataDecl ::= S 'NDATA' S
|
|
254 ;;[73] EntityDef ::= EntityValue| (ExternalID NDataDecl?)
|
|
255 ;;[71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
|
|
256 ;;[74] PEDef ::= EntityValue | ExternalID
|
|
257 ;;[72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
|
|
258 ;;[70] EntityDecl ::= GEDecl | PEDecl
|
|
259
|
51102
|
260 ;; Note that this is setup so that we can do whitespace-skipping with
|
|
261 ;; `(skip-syntax-forward " ")', inter alia. Previously this was slow
|
|
262 ;; compared with `re-search-forward', but that has been fixed. Also
|
|
263 ;; note that the standard syntax table contains other characters with
|
|
264 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
|
|
265 ;; which we might skip whitespace -- specifically, they're not
|
|
266 ;; NameChars [XML 4].
|
|
267
|
|
268 (defvar xml-syntax-table
|
|
269 (let ((table (make-syntax-table)))
|
|
270 ;; Get space syntax correct per XML [3].
|
|
271 (dotimes (c 31)
|
|
272 (modify-syntax-entry c "." table)) ; all are space in standard table
|
68534
|
273 (dolist (c '(?\t ?\n ?\r)) ; these should be space
|
51102
|
274 (modify-syntax-entry c " " table))
|
|
275 ;; For skipping attributes.
|
|
276 (modify-syntax-entry ?\" "\"" table)
|
|
277 (modify-syntax-entry ?' "\"" table)
|
|
278 ;; Non-alnum name chars should be symbol constituents (`-' and `_'
|
|
279 ;; are OK by default).
|
|
280 (modify-syntax-entry ?. "_" table)
|
|
281 (modify-syntax-entry ?: "_" table)
|
|
282 ;; XML [89]
|
66646
|
283 (unless (featurep 'xemacs)
|
|
284 (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005
|
|
285 #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC
|
|
286 #x30FD #x30FE))
|
|
287 (modify-syntax-entry (decode-char 'ucs c) "w" table)))
|
51102
|
288 ;; Fixme: rest of [4]
|
|
289 table)
|
|
290 "Syntax table used by `xml-parse-region'.")
|
|
291
|
|
292 ;; XML [5]
|
|
293 ;; Note that [:alpha:] matches all multibyte chars with word syntax.
|
51105
|
294 (eval-and-compile
|
|
295 (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*"))
|
51102
|
296
|
|
297 ;; Fixme: This needs re-writing to deal with the XML grammar properly, i.e.
|
|
298 ;; document ::= prolog element Misc*
|
|
299 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
|
|
300
|
|
301 ;;;###autoload
|
51930
|
302 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns)
|
30329
|
303 "Parse the region from BEG to END in BUFFER.
|
|
304 If BUFFER is nil, it defaults to the current buffer.
|
|
305 Returns the XML list for the region, or raises an error if the region
|
51930
|
306 is not well-formed XML.
|
30329
|
307 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped,
|
51930
|
308 and returned as the first element of the list.
|
|
309 If PARSE-NS is non-nil, then QNAMES are expanded."
|
68535
|
310 ;; Use fixed syntax table to ensure regexp char classes and syntax
|
|
311 ;; specs DTRT.
|
|
312 (with-syntax-table (standard-syntax-table)
|
|
313 (let ((case-fold-search nil) ; XML is case-sensitive.
|
|
314 xml result dtd)
|
|
315 (save-excursion
|
|
316 (if buffer
|
|
317 (set-buffer buffer))
|
|
318 (save-restriction
|
|
319 (narrow-to-region beg end)
|
51102
|
320 (goto-char (point-min))
|
|
321 (while (not (eobp))
|
|
322 (if (search-forward "<" nil t)
|
|
323 (progn
|
|
324 (forward-char -1)
|
52975
|
325 (setq result (xml-parse-tag parse-dtd parse-ns))
|
56375
|
326 (if (and xml result (not xml-sub-parser))
|
51102
|
327 ;; translation of rule [1] of XML specifications
|
56375
|
328 (error "XML: (Not Well-Formed) Only one root tag allowed")
|
30329
|
329 (cond
|
42031
|
330 ((null result))
|
52975
|
331 ((and (listp (car result))
|
|
332 parse-dtd)
|
42031
|
333 (setq dtd (car result))
|
51102
|
334 (if (cdr result) ; possible leading comment
|
|
335 (add-to-list 'xml (cdr result))))
|
30329
|
336 (t
|
51102
|
337 (add-to-list 'xml result)))))
|
|
338 (goto-char (point-max))))
|
|
339 (if parse-dtd
|
|
340 (cons dtd (nreverse xml))
|
|
341 (nreverse xml)))))))
|
30329
|
342
|
54877
|
343 (defun xml-maybe-do-ns (name default xml-ns)
|
54937
|
344 "Perform any namespace expansion.
|
|
345 NAME is the name to perform the expansion on.
|
54877
|
346 DEFAULT is the default namespace. XML-NS is a cons of namespace
|
|
347 names to uris. When namespace-aware parsing is off, then XML-NS
|
|
348 is nil.
|
52975
|
349
|
54877
|
350 During namespace-aware parsing, any name without a namespace is
|
|
351 put into the namespace identified by DEFAULT. nil is used to
|
|
352 specify that the name shouldn't be given a namespace."
|
|
353 (if (consp xml-ns)
|
|
354 (let* ((nsp (string-match ":" name))
|
|
355 (lname (if nsp (substring name (match-end 0)) name))
|
|
356 (prefix (if nsp (substring name 0 (match-beginning 0)) default))
|
|
357 (special (and (string-equal lname "xmlns") (not prefix)))
|
|
358 ;; Setting default to nil will insure that there is not
|
|
359 ;; matching cons in xml-ns. In which case we
|
|
360 (ns (or (cdr (assoc (if special "xmlns" prefix)
|
|
361 xml-ns))
|
56375
|
362 "")))
|
54877
|
363 (cons ns (if special "" lname)))
|
|
364 (intern name)))
|
30329
|
365
|
56375
|
366 (defun xml-parse-fragment (&optional parse-dtd parse-ns)
|
|
367 "Parse xml-like fragments."
|
|
368 (let ((xml-sub-parser t)
|
|
369 children)
|
|
370 (while (not (eobp))
|
|
371 (let ((bit (xml-parse-tag
|
|
372 parse-dtd parse-ns)))
|
|
373 (if children
|
|
374 (setq children (append (list bit) children))
|
|
375 (if (stringp bit)
|
|
376 (setq children (list bit))
|
|
377 (setq children bit)))))
|
|
378 (reverse children)))
|
|
379
|
51930
|
380 (defun xml-parse-tag (&optional parse-dtd parse-ns)
|
51102
|
381 "Parse the tag at point.
|
30329
|
382 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
|
|
383 returned as the first element in the list.
|
51930
|
384 If PARSE-NS is non-nil, then QNAMES are expanded.
|
30329
|
385 Returns one of:
|
51102
|
386 - a list : the matching node
|
|
387 - nil : the point is not looking at a tag.
|
|
388 - a pair : the first element is the DTD, the second is the node."
|
56375
|
389 (let ((xml-validating-parser (or parse-dtd xml-validating-parser))
|
|
390 (xml-ns (if (consp parse-ns)
|
51930
|
391 parse-ns
|
|
392 (if parse-ns
|
|
393 (list
|
68534
|
394 ;; Default for empty prefix is no namespace
|
56375
|
395 (cons "" "")
|
54877
|
396 ;; "xml" namespace
|
56375
|
397 (cons "xml" "http://www.w3.org/XML/1998/namespace")
|
51930
|
398 ;; We need to seed the xmlns namespace
|
56375
|
399 (cons "xmlns" "http://www.w3.org/2000/xmlns/"))))))
|
51930
|
400 (cond
|
|
401 ;; Processing instructions (like the <?xml version="1.0"?> tag at the
|
|
402 ;; beginning of a document).
|
|
403 ((looking-at "<\\?")
|
|
404 (search-forward "?>")
|
|
405 (skip-syntax-forward " ")
|
|
406 (xml-parse-tag parse-dtd xml-ns))
|
|
407 ;; Character data (CDATA) sections, in which no tag should be interpreted
|
|
408 ((looking-at "<!\\[CDATA\\[")
|
|
409 (let ((pos (match-end 0)))
|
|
410 (unless (search-forward "]]>" nil t)
|
56375
|
411 (error "XML: (Not Well Formed) CDATA section does not end anywhere in the document"))
|
58698
|
412 (concat
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
413 (buffer-substring-no-properties pos (match-beginning 0))
|
58698
|
414 (xml-parse-string))))
|
51930
|
415 ;; DTD for the document
|
|
416 ((looking-at "<!DOCTYPE")
|
56375
|
417 (let ((dtd (xml-parse-dtd parse-ns)))
|
|
418 (skip-syntax-forward " ")
|
|
419 (if xml-validating-parser
|
|
420 (cons dtd (xml-parse-tag nil xml-ns))
|
|
421 (xml-parse-tag nil xml-ns))))
|
51930
|
422 ;; skip comments
|
|
423 ((looking-at "<!--")
|
|
424 (search-forward "-->")
|
|
425 nil)
|
|
426 ;; end tag
|
|
427 ((looking-at "</")
|
|
428 '())
|
|
429 ;; opening tag
|
|
430 ((looking-at "<\\([^/>[:space:]]+\\)")
|
|
431 (goto-char (match-end 1))
|
52975
|
432
|
|
433 ;; Parse this node
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
434 (let* ((node-name (match-string-no-properties 1))
|
68534
|
435 ;; Parse the attribute list.
|
|
436 (attrs (xml-parse-attlist xml-ns))
|
|
437 children pos)
|
51930
|
438
|
68534
|
439 ;; add the xmlns:* attrs to our cache
|
|
440 (when (consp xml-ns)
|
54877
|
441 (dolist (attr attrs)
|
|
442 (when (and (consp (car attr))
|
56375
|
443 (equal "http://www.w3.org/2000/xmlns/"
|
|
444 (caar attr)))
|
|
445 (push (cons (cdar attr) (cdr attr))
|
54877
|
446 xml-ns))))
|
|
447
|
68534
|
448 (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns)))
|
54877
|
449
|
51930
|
450 ;; is this an empty element ?
|
|
451 (if (looking-at "/>")
|
56375
|
452 (progn
|
|
453 (forward-char 2)
|
|
454 (nreverse children))
|
|
455
|
|
456 ;; is this a valid start tag ?
|
|
457 (if (eq (char-after) ?>)
|
|
458 (progn
|
|
459 (forward-char 1)
|
|
460 ;; Now check that we have the right end-tag. Note that this
|
|
461 ;; one might contain spaces after the tag name
|
|
462 (let ((end (concat "</" node-name "\\s-*>")))
|
|
463 (while (not (looking-at end))
|
|
464 (cond
|
|
465 ((looking-at "</")
|
|
466 (error "XML: (Not Well-Formed) Invalid end tag (expecting %s) at pos %d"
|
|
467 node-name (point)))
|
|
468 ((= (char-after) ?<)
|
|
469 (let ((tag (xml-parse-tag nil xml-ns)))
|
|
470 (when tag
|
|
471 (push tag children))))
|
|
472 (t
|
|
473 (let ((expansion (xml-parse-string)))
|
|
474 (setq children
|
|
475 (if (stringp expansion)
|
|
476 (if (stringp (car children))
|
|
477 ;; The two strings were separated by a comment.
|
66646
|
478 (setq children (append (list (concat (car children) expansion))
|
56375
|
479 (cdr children)))
|
|
480 (setq children (append (list expansion) children)))
|
|
481 (setq children (append expansion children))))))))
|
30329
|
482
|
56375
|
483 (goto-char (match-end 0))
|
|
484 (nreverse children)))
|
|
485 ;; This was an invalid start tag (Expected ">", but didn't see it.)
|
|
486 (error "XML: (Well-Formed) Couldn't parse tag: %s"
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
487 (buffer-substring-no-properties (- (point) 10) (+ (point) 1)))))))
|
56375
|
488 (t ;; (Not one of PI, CDATA, Comment, End tag, or Start tag)
|
|
489 (unless xml-sub-parser ; Usually, we error out.
|
|
490 (error "XML: (Well-Formed) Invalid character"))
|
|
491
|
|
492 ;; However, if we're parsing incrementally, then we need to deal
|
|
493 ;; with stray CDATA.
|
|
494 (xml-parse-string)))))
|
49036
|
495
|
56375
|
496 (defun xml-parse-string ()
|
|
497 "Parse the next whatever. Could be a string, or an element."
|
68534
|
498 (let* ((pos (point))
|
|
499 (string (progn (if (search-forward "<" nil t)
|
|
500 (forward-char -1)
|
|
501 (goto-char (point-max)))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
502 (buffer-substring-no-properties pos (point)))))
|
68534
|
503 ;; Clean up the string. As per XML specifications, the XML
|
|
504 ;; processor should always pass the whole string to the
|
|
505 ;; application. But \r's should be replaced:
|
|
506 ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends
|
|
507 (setq pos 0)
|
|
508 (while (string-match "\r\n?" string pos)
|
|
509 (setq string (replace-match "\n" t t string))
|
|
510 (setq pos (1+ (match-beginning 0))))
|
42031
|
511
|
68534
|
512 (xml-substitute-special string)))
|
30329
|
513
|
54877
|
514 (defun xml-parse-attlist (&optional xml-ns)
|
54937
|
515 "Return the attribute-list after point.
|
|
516 Leave point at the first non-blank character after the tag."
|
42031
|
517 (let ((attlist ())
|
52975
|
518 end-pos name)
|
51102
|
519 (skip-syntax-forward " ")
|
|
520 (while (looking-at (eval-when-compile
|
|
521 (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*")))
|
54877
|
522 (setq end-pos (match-end 0))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
523 (setq name (xml-maybe-do-ns (match-string-no-properties 1) nil xml-ns))
|
54877
|
524 (goto-char end-pos)
|
30329
|
525
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
526 ;; 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>
diff
changeset
|
527
|
30329
|
528 ;; Do we have a string between quotes (or double-quotes),
|
|
529 ;; or a simple word ?
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
530 (if (looking-at "\"\\([^\"]*\\)\"")
|
52975
|
531 (setq end-pos (match-end 0))
|
50210
|
532 (if (looking-at "'\\([^']*\\)'")
|
52975
|
533 (setq end-pos (match-end 0))
|
56375
|
534 (error "XML: (Not Well-Formed) Attribute values must be given between quotes")))
|
30329
|
535
|
|
536 ;; Each attribute must be unique within a given element
|
|
537 (if (assoc name attlist)
|
56375
|
538 (error "XML: (Not Well-Formed) Each attribute must be unique within an element"))
|
49036
|
539
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
540 ;; 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>
diff
changeset
|
541 ;; in the attributes
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
542 (let ((string (match-string-no-properties 1))
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
543 (pos 0))
|
51102
|
544 (replace-regexp-in-string "\\s-\\{2,\\}" " " string)
|
56375
|
545 (let ((expansion (xml-substitute-special string)))
|
|
546 (unless (stringp expansion)
|
68534
|
547 ; We say this is the constraint. It is acctually that
|
|
548 ; external entities nor "<" can be in an attribute value.
|
56375
|
549 (error "XML: (Not Well-Formed) Entities in attributes cannot expand into elements"))
|
|
550 (push (cons name expansion) attlist)))
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
551
|
52975
|
552 (goto-char end-pos)
|
51102
|
553 (skip-syntax-forward " "))
|
42031
|
554 (nreverse attlist)))
|
30329
|
555
|
|
556 ;;*******************************************************************
|
|
557 ;;**
|
|
558 ;;** The DTD (document type declaration)
|
|
559 ;;** The following functions know how to skip or parse the DTD of
|
|
560 ;;** a document
|
|
561 ;;**
|
|
562 ;;*******************************************************************
|
|
563
|
51102
|
564 ;; Fixme: This fails at least if the DTD contains conditional sections.
|
|
565
|
|
566 (defun xml-skip-dtd ()
|
|
567 "Skip the DTD at point.
|
30329
|
568 This follows the rule [28] in the XML specifications."
|
56375
|
569 (let ((xml-validating-parser nil))
|
|
570 (xml-parse-dtd)))
|
30329
|
571
|
56375
|
572 (defun xml-parse-dtd (&optional parse-ns)
|
51102
|
573 "Parse the DTD at point."
|
|
574 (forward-char (eval-when-compile (length "<!DOCTYPE")))
|
|
575 (skip-syntax-forward " ")
|
56375
|
576 (if (and (looking-at ">")
|
|
577 xml-validating-parser)
|
|
578 (error "XML: (Validity) Invalid DTD (expecting name of the document)"))
|
49036
|
579
|
42031
|
580 ;; Get the name of the document
|
51102
|
581 (looking-at xml-name-regexp)
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
582 (let ((dtd (list (match-string-no-properties 0) 'dtd))
|
42031
|
583 type element end-pos)
|
30329
|
584 (goto-char (match-end 0))
|
|
585
|
51102
|
586 (skip-syntax-forward " ")
|
|
587 ;; XML [75]
|
|
588 (cond ((looking-at "PUBLIC\\s-+")
|
|
589 (goto-char (match-end 0))
|
|
590 (unless (or (re-search-forward
|
|
591 "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\""
|
|
592 nil t)
|
|
593 (re-search-forward
|
|
594 "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'"
|
|
595 nil t))
|
56375
|
596 (error "XML: Missing Public ID"))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
597 (let ((pubid (match-string-no-properties 1)))
|
56375
|
598 (skip-syntax-forward " ")
|
51102
|
599 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
|
|
600 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
|
56375
|
601 (error "XML: Missing System ID"))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
602 (push (list pubid (match-string-no-properties 1) 'public) dtd)))
|
51102
|
603 ((looking-at "SYSTEM\\s-+")
|
|
604 (goto-char (match-end 0))
|
|
605 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
|
|
606 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
|
56375
|
607 (error "XML: Missing System ID"))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
608 (push (list (match-string-no-properties 1) 'system) dtd)))
|
51102
|
609 (skip-syntax-forward " ")
|
|
610 (if (eq ?> (char-after))
|
|
611 (forward-char)
|
|
612 (if (not (eq (char-after) ?\[))
|
56375
|
613 (error "XML: Bad DTD")
|
51102
|
614 (forward-char)
|
|
615 ;; Parse the rest of the DTD
|
63282
|
616 ;; Fixme: Deal with NOTATION, PIs.
|
51102
|
617 (while (not (looking-at "\\s-*\\]"))
|
|
618 (skip-syntax-forward " ")
|
|
619 (cond
|
49036
|
620
|
51102
|
621 ;; Translation of rule [45] of XML specifications
|
|
622 ((looking-at
|
|
623 "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>")
|
|
624
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
625 (setq element (match-string-no-properties 1)
|
51102
|
626 type (match-string-no-properties 2))
|
|
627 (setq end-pos (match-end 0))
|
30329
|
628
|
51102
|
629 ;; Translation of rule [46] of XML specifications
|
|
630 (cond
|
|
631 ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration
|
|
632 (setq type 'empty))
|
|
633 ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents
|
|
634 (setq type 'any))
|
|
635 ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47])
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
636 (setq type (xml-parse-elem-type (match-string-no-properties 1 type))))
|
51102
|
637 ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution
|
|
638 nil)
|
|
639 (t
|
56375
|
640 (if xml-validating-parser
|
57346
|
641 (error "XML: (Validity) Invalid element type in the DTD"))))
|
58723
|
642
|
51102
|
643 ;; rule [45]: the element declaration must be unique
|
56375
|
644 (if (and (assoc element dtd)
|
|
645 xml-validating-parser)
|
|
646 (error "XML: (Validity) Element declarations must be unique in a DTD (<%s>)"
|
53011
|
647 element))
|
30329
|
648
|
51102
|
649 ;; Store the element in the DTD
|
|
650 (push (list element type) dtd)
|
|
651 (goto-char end-pos))
|
63282
|
652
|
|
653 ;; Translation of rule [52] of XML specifications
|
|
654 ((looking-at (concat "<!ATTLIST[ \t\n\r]*\\(" xml-name-re
|
|
655 "\\)[ \t\n\r]*\\(" xml-att-def-re
|
|
656 "\\)*[ \t\n\r]*>"))
|
|
657
|
|
658 ;; We don't do anything with ATTLIST currently
|
|
659 (goto-char (match-end 0)))
|
|
660
|
51102
|
661 ((looking-at "<!--")
|
|
662 (search-forward "-->"))
|
56375
|
663 ((looking-at (concat "<!ENTITY[ \t\n\r]*\\(" xml-name-re
|
|
664 "\\)[ \t\n\r]*\\(" xml-entity-value-re
|
|
665 "\\)[ \t\n\r]*>"))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
666 (let ((name (match-string-no-properties 1))
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
667 (value (substring (match-string-no-properties 2) 1
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
668 (- (length (match-string-no-properties 2)) 1))))
|
63282
|
669 (goto-char (match-end 0))
|
56375
|
670 (setq xml-entity-alist
|
|
671 (append xml-entity-alist
|
|
672 (list (cons name
|
|
673 (with-temp-buffer
|
|
674 (insert value)
|
|
675 (goto-char (point-min))
|
|
676 (xml-parse-fragment
|
|
677 xml-validating-parser
|
|
678 parse-ns))))))))
|
|
679 ((or (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re
|
|
680 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
|
|
681 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
|
|
682 (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re
|
|
683 "\\)[ \t\n\r]+PUBLIC[ \t\n\r]+"
|
|
684 "\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
|
|
685 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'"
|
|
686 "[ \t\n\r]+\\(\"[^\"]*\"\\|'[^']*'\\)"
|
|
687 "[ \t\n\r]*>")))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
688 (let ((name (match-string-no-properties 1))
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
689 (file (substring (match-string-no-properties 2) 1
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
690 (- (length (match-string-no-properties 2)) 1))))
|
63282
|
691 (goto-char (match-end 0))
|
56375
|
692 (setq xml-entity-alist
|
|
693 (append xml-entity-alist
|
|
694 (list (cons name (with-temp-buffer
|
|
695 (insert-file-contents file)
|
|
696 (goto-char (point-min))
|
|
697 (xml-parse-fragment
|
|
698 xml-validating-parser
|
|
699 parse-ns))))))))
|
58724
|
700 ;; skip parameter entity declarations
|
|
701 ((or (looking-at (concat "<!ENTITY[ \t\n\r]+%[ \t\n\r]+\\(" xml-name-re
|
|
702 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
|
|
703 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
|
|
704 (looking-at (concat "<!ENTITY[ \t\n\r]+"
|
|
705 "%[ \t\n\r]+"
|
|
706 "\\(" xml-name-re "\\)[ \t\n\r]+"
|
|
707 "PUBLIC[ \t\n\r]+"
|
|
708 "\\(\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
|
|
709 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'\\)[ \t\n\r]+"
|
|
710 "\\(\"[^\"]+\"\\|'[^']+'\\)"
|
|
711 "[ \t\n\r]*>")))
|
|
712 (goto-char (match-end 0)))
|
|
713 ;; skip parameter entities
|
|
714 ((looking-at (concat "%" xml-name-re ";"))
|
|
715 (goto-char (match-end 0)))
|
51102
|
716 (t
|
58721
ff6b2a793277
Ensure that validity messages only show when xml-validating-parser is set.
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
717 (when xml-validating-parser
|
58723
|
718 (error "XML: (Validity) Invalid DTD item"))))))
|
56375
|
719 (if (looking-at "\\s-*]>")
|
63282
|
720 (goto-char (match-end 0))))
|
53011
|
721 (nreverse dtd)))
|
30329
|
722
|
|
723 (defun xml-parse-elem-type (string)
|
51102
|
724 "Convert element type STRING into a Lisp structure."
|
30329
|
725
|
|
726 (let (elem modifier)
|
|
727 (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
|
|
728 (progn
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
729 (setq elem (match-string-no-properties 1 string)
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
730 modifier (match-string-no-properties 2 string))
|
30329
|
731 (if (string-match "|" elem)
|
42031
|
732 (setq elem (cons 'choice
|
30329
|
733 (mapcar 'xml-parse-elem-type
|
|
734 (split-string elem "|"))))
|
|
735 (if (string-match "," elem)
|
42031
|
736 (setq elem (cons 'seq
|
30329
|
737 (mapcar 'xml-parse-elem-type
|
51102
|
738 (split-string elem ",")))))))
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
739 (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string)
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
740 (setq elem (match-string-no-properties 1 string)
|
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
741 modifier (match-string-no-properties 2 string))))
|
30329
|
742
|
42031
|
743 (if (and (stringp elem) (string= elem "#PCDATA"))
|
|
744 (setq elem 'pcdata))
|
49036
|
745
|
42031
|
746 (cond
|
|
747 ((string= modifier "+")
|
|
748 (list '+ elem))
|
|
749 ((string= modifier "*")
|
|
750 (list '* elem))
|
|
751 ((string= modifier "?")
|
49787
|
752 (list '\? elem))
|
42031
|
753 (t
|
|
754 elem))))
|
30329
|
755
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
756 ;;*******************************************************************
|
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
757 ;;**
|
30329
|
758 ;;** Substituting special XML sequences
|
|
759 ;;**
|
|
760 ;;*******************************************************************
|
|
761
|
|
762 (defun xml-substitute-special (string)
|
51102
|
763 "Return STRING, after subsituting entity references."
|
|
764 ;; This originally made repeated passes through the string from the
|
|
765 ;; beginning, which isn't correct, since then either "&amp;" or
|
|
766 ;; "&amp;" won't DTRT.
|
56375
|
767
|
|
768 (let ((point 0)
|
|
769 children end-point)
|
58698
|
770 (while (string-match "&\\([^;]*\\);" string point)
|
56375
|
771 (setq end-point (match-end 0))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
772 (let* ((this-part (match-string-no-properties 1 string))
|
56375
|
773 (prev-part (substring string point (match-beginning 0)))
|
|
774 (entity (assoc this-part xml-entity-alist))
|
|
775 (expansion
|
|
776 (cond ((string-match "#\\([0-9]+\\)" this-part)
|
|
777 (let ((c (decode-char
|
|
778 'ucs
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
779 (string-to-number (match-string-no-properties 1 this-part)))))
|
56375
|
780 (if c (string c))))
|
|
781 ((string-match "#x\\([[:xdigit:]]+\\)" this-part)
|
|
782 (let ((c (decode-char
|
|
783 'ucs
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
784 (string-to-number (match-string-no-properties 1 this-part) 16))))
|
56375
|
785 (if c (string c))))
|
|
786 (entity
|
|
787 (cdr entity))
|
58698
|
788 ((eq (length this-part) 0)
|
58723
|
789 (error "XML: (Not Well-Formed) No entity given"))
|
56375
|
790 (t
|
62752
|
791 (if xml-validating-parser
|
56375
|
792 (error "XML: (Validity) Undefined entity `%s'"
|
62752
|
793 this-part)
|
|
794 xml-undefined-entity)))))
|
30329
|
795
|
56375
|
796 (cond ((null children)
|
57367
|
797 ;; FIXME: If we have an entity that expands into XML, this won't work.
|
|
798 (setq children
|
|
799 (concat prev-part expansion)))
|
56375
|
800 ((stringp children)
|
|
801 (if (stringp expansion)
|
|
802 (setq children (concat children prev-part expansion))
|
|
803 (setq children (list expansion (concat prev-part children)))))
|
|
804 ((and (stringp expansion)
|
|
805 (stringp (car children)))
|
|
806 (setcar children (concat prev-part expansion (car children))))
|
|
807 ((stringp expansion)
|
|
808 (setq children (append (concat prev-part expansion)
|
|
809 children)))
|
|
810 ((stringp (car children))
|
|
811 (setcar children (concat (car children) prev-part))
|
|
812 (setq children (append expansion children)))
|
|
813 (t
|
|
814 (setq children (list expansion
|
|
815 prev-part
|
|
816 children))))
|
|
817 (setq point end-point)))
|
|
818 (cond ((stringp children)
|
|
819 (concat children (substring string point)))
|
|
820 ((stringp (car (last children)))
|
57344
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
821 (concat (car (last children)) (substring string point)))
|
56375
|
822 ((null children)
|
|
823 string)
|
|
824 (t
|
57344
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
825 (concat (mapconcat 'identity
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
826 (nreverse children)
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
827 "")
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
828 (substring string point))))))
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
829
|
30329
|
830 ;;*******************************************************************
|
|
831 ;;**
|
|
832 ;;** Printing a tree.
|
|
833 ;;** This function is intended mainly for debugging purposes.
|
|
834 ;;**
|
|
835 ;;*******************************************************************
|
|
836
|
55253
|
837 (defun xml-debug-print (xml &optional indent-string)
|
|
838 "Outputs the XML in the current buffer.
|
|
839 XML can be a tree or a list of nodes.
|
|
840 The first line is indented with the optional INDENT-STRING."
|
|
841 (setq indent-string (or indent-string ""))
|
42031
|
842 (dolist (node xml)
|
55253
|
843 (xml-debug-print-internal node indent-string)))
|
|
844
|
|
845 (defalias 'xml-print 'xml-debug-print)
|
30329
|
846
|
42031
|
847 (defun xml-debug-print-internal (xml indent-string)
|
30329
|
848 "Outputs the XML tree in the current buffer.
|
51102
|
849 The first line is indented with INDENT-STRING."
|
30329
|
850 (let ((tree xml)
|
|
851 attlist)
|
51102
|
852 (insert indent-string ?< (symbol-name (xml-node-name tree)))
|
49036
|
853
|
30329
|
854 ;; output the attribute list
|
42031
|
855 (setq attlist (xml-node-attributes tree))
|
30329
|
856 (while attlist
|
51102
|
857 (insert ?\ (symbol-name (caar attlist)) "=\"" (cdar attlist) ?\")
|
42031
|
858 (setq attlist (cdr attlist)))
|
49036
|
859
|
42031
|
860 (setq tree (xml-node-children tree))
|
30329
|
861
|
55253
|
862 (if (null tree)
|
|
863 (insert ?/ ?>)
|
|
864 (insert ?>)
|
30329
|
865
|
55253
|
866 ;; output the children
|
|
867 (dolist (node tree)
|
|
868 (cond
|
|
869 ((listp node)
|
|
870 (insert ?\n)
|
|
871 (xml-debug-print-internal node (concat indent-string " ")))
|
|
872 ((stringp node) (insert node))
|
|
873 (t
|
|
874 (error "Invalid XML tree"))))
|
|
875
|
|
876 (when (not (and (null (cdr tree))
|
|
877 (stringp (car tree))))
|
|
878 (insert ?\n indent-string))
|
|
879 (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>))))
|
30329
|
880
|
|
881 (provide 'xml)
|
|
882
|
55316
|
883 ;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b
|
30329
|
884 ;;; xml.el ends here
|