|
38409
|
1 ;;; xml.el --- XML parser
|
|
30329
|
2
|
|
64762
|
3 ;; Copyright (C) 2000, 2001, 2002, 2003, 2004,
|
|
100908
|
4 ;; 2005, 2006, 2007, 2008, 2009 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
|
|
94678
|
12 ;; GNU Emacs is free software: you can redistribute it and/or modify
|
|
30329
|
13 ;; it under the terms of the GNU General Public License as published by
|
|
94678
|
14 ;; the Free Software Foundation, either version 3 of the License, or
|
|
|
15 ;; (at your option) any later version.
|
|
30329
|
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
|
|
94678
|
23 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
|
|
30329
|
24
|
|
|
25 ;;; Commentary:
|
|
|
26
|
|
51102
|
27 ;; This file contains a somewhat incomplete non-validating XML parser. It
|
|
|
28 ;; parses a file, and returns a list that can be used internally by
|
|
54937
|
29 ;; any other Lisp libraries.
|
|
30329
|
30
|
|
|
31 ;;; FILE FORMAT
|
|
|
32
|
|
51102
|
33 ;; The document type declaration may either be ignored or (optionally)
|
|
|
34 ;; parsed, but currently the parsing will only accept element
|
|
54937
|
35 ;; declarations. The XML file is assumed to be well-formed. In case
|
|
51102
|
36 ;; of error, the parsing stops and the XML file is shown where the
|
|
|
37 ;; parsing stopped.
|
|
30329
|
38 ;;
|
|
51102
|
39 ;; It also knows how to ignore comments and processing instructions.
|
|
30329
|
40 ;;
|
|
|
41 ;; The XML file should have the following format:
|
|
34825
|
42 ;; <node1 attr1="name1" attr2="name2" ...>value
|
|
|
43 ;; <node2 attr3="name3" attr4="name4">value2</node2>
|
|
|
44 ;; <node3 attr5="name5" attr6="name6">value3</node3>
|
|
30329
|
45 ;; </node1>
|
|
54937
|
46 ;; Of course, the name of the nodes and attributes can be anything. There can
|
|
30329
|
47 ;; be any number of attributes (or none), as well as any number of children
|
|
|
48 ;; below the nodes.
|
|
|
49 ;;
|
|
|
50 ;; There can be only top level node, but with any number of children below.
|
|
|
51
|
|
|
52 ;;; LIST FORMAT
|
|
|
53
|
|
54877
|
54 ;; The functions `xml-parse-file', `xml-parse-region' and
|
|
|
55 ;; `xml-parse-tag' return a list with the following format:
|
|
30329
|
56 ;;
|
|
|
57 ;; xml-list ::= (node node ...)
|
|
54877
|
58 ;; node ::= (qname attribute-list . child_node_list)
|
|
30329
|
59 ;; child_node_list ::= child_node child_node ...
|
|
|
60 ;; child_node ::= node | string
|
|
54877
|
61 ;; qname ::= (:namespace-uri . "name") | "name"
|
|
|
62 ;; attribute_list ::= ((qname . "value") (qname . "value") ...)
|
|
30329
|
63 ;; | nil
|
|
|
64 ;; string ::= "..."
|
|
|
65 ;;
|
|
51102
|
66 ;; Some macros are provided to ease the parsing of this list.
|
|
|
67 ;; Whitespace is preserved. Fixme: There should be a tree-walker that
|
|
|
68 ;; can remove it.
|
|
30329
|
69
|
|
54877
|
70 ;; TODO:
|
|
|
71 ;; * xml:base, xml:space support
|
|
|
72 ;; * more complete DOCTYPE parsing
|
|
|
73 ;; * pi support
|
|
|
74
|
|
30329
|
75 ;;; Code:
|
|
|
76
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
77 ;; 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
|
78 ;; 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
|
79 ;; 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
|
80 ;; 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
|
81 ;; 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
|
82 ;; instead of hand-crafted XML.
|
|
51102
|
83
|
|
30329
|
84 ;;*******************************************************************
|
|
|
85 ;;**
|
|
|
86 ;;** Macros to parse the list
|
|
|
87 ;;**
|
|
|
88 ;;*******************************************************************
|
|
|
89
|
|
62752
|
90 (defconst xml-undefined-entity "?"
|
|
|
91 "What to substitute for undefined entities")
|
|
|
92
|
|
56375
|
93 (defvar xml-entity-alist
|
|
|
94 '(("lt" . "<")
|
|
|
95 ("gt" . ">")
|
|
|
96 ("apos" . "'")
|
|
|
97 ("quot" . "\"")
|
|
|
98 ("amp" . "&"))
|
|
|
99 "The defined entities. Entities are added to this when the DTD is parsed.")
|
|
|
100
|
|
|
101 (defvar xml-sub-parser nil
|
|
|
102 "Dynamically set this to a non-nil value if you want to parse an XML fragment.")
|
|
|
103
|
|
|
104 (defvar xml-validating-parser nil
|
|
|
105 "Set to non-nil to get validity checking.")
|
|
|
106
|
|
42031
|
107 (defsubst xml-node-name (node)
|
|
30329
|
108 "Return the tag associated with NODE.
|
|
54937
|
109 Without namespace-aware parsing, the tag is a symbol.
|
|
|
110
|
|
|
111 With namespace-aware parsing, the tag is a cons of a string
|
|
|
112 representing the uri of the namespace with the local name of the
|
|
|
113 tag. For example,
|
|
|
114
|
|
|
115 <foo>
|
|
|
116
|
|
|
117 would be represented by
|
|
|
118
|
|
|
119 '(\"\" . \"foo\")."
|
|
|
120
|
|
42031
|
121 (car node))
|
|
30329
|
122
|
|
42031
|
123 (defsubst xml-node-attributes (node)
|
|
30329
|
124 "Return the list of attributes of NODE.
|
|
|
125 The list can be nil."
|
|
42031
|
126 (nth 1 node))
|
|
30329
|
127
|
|
42031
|
128 (defsubst xml-node-children (node)
|
|
30329
|
129 "Return the list of children of NODE.
|
|
|
130 This is a list of nodes, and it can be nil."
|
|
42031
|
131 (cddr node))
|
|
30329
|
132
|
|
|
133 (defun xml-get-children (node child-name)
|
|
|
134 "Return the children of NODE whose tag is CHILD-NAME.
|
|
54937
|
135 CHILD-NAME should match the value returned by `xml-node-name'."
|
|
42031
|
136 (let ((match ()))
|
|
|
137 (dolist (child (xml-node-children node))
|
|
54937
|
138 (if (and (listp child)
|
|
|
139 (equal (xml-node-name child) child-name))
|
|
|
140 (push child match)))
|
|
42031
|
141 (nreverse match)))
|
|
30329
|
142
|
|
53375
|
143 (defun xml-get-attribute-or-nil (node attribute)
|
|
30329
|
144 "Get from NODE the value of ATTRIBUTE.
|
|
54937
|
145 Return nil if the attribute was not found.
|
|
53375
|
146
|
|
|
147 See also `xml-get-attribute'."
|
|
54243
|
148 (cdr (assoc attribute (xml-node-attributes node))))
|
|
53375
|
149
|
|
|
150 (defsubst xml-get-attribute (node attribute)
|
|
|
151 "Get from NODE the value of ATTRIBUTE.
|
|
|
152 An empty string is returned if the attribute was not found.
|
|
|
153
|
|
|
154 See also `xml-get-attribute-or-nil'."
|
|
|
155 (or (xml-get-attribute-or-nil node attribute) ""))
|
|
30329
|
156
|
|
|
157 ;;*******************************************************************
|
|
|
158 ;;**
|
|
|
159 ;;** Creating the list
|
|
|
160 ;;**
|
|
|
161 ;;*******************************************************************
|
|
|
162
|
|
51102
|
163 ;;;###autoload
|
|
51930
|
164 (defun xml-parse-file (file &optional parse-dtd parse-ns)
|
|
51102
|
165 "Parse the well-formed XML file FILE.
|
|
|
166 If FILE is already visited, use its buffer and don't kill it.
|
|
30329
|
167 Returns the top node with all its children.
|
|
51930
|
168 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped.
|
|
|
169 If PARSE-NS is non-nil, then QNAMES are expanded."
|
|
72108
|
170 (if (get-file-buffer file)
|
|
|
171 (with-current-buffer (get-file-buffer file)
|
|
|
172 (save-excursion
|
|
|
173 (xml-parse-region (point-min)
|
|
|
174 (point-max)
|
|
|
175 (current-buffer)
|
|
|
176 parse-dtd parse-ns)))
|
|
|
177 (with-temp-buffer
|
|
|
178 (insert-file-contents file)
|
|
|
179 (xml-parse-region (point-min)
|
|
|
180 (point-max)
|
|
|
181 (current-buffer)
|
|
|
182 parse-dtd parse-ns))))
|
|
30329
|
183
|
|
56375
|
184
|
|
58939
|
185 (defvar xml-name-re)
|
|
|
186 (defvar xml-entity-value-re)
|
|
64794
|
187 (defvar xml-att-def-re)
|
|
58724
|
188 (let* ((start-chars (concat "[:alpha:]:_"))
|
|
56375
|
189 (name-chars (concat "-[:digit:]." start-chars))
|
|
68534
|
190 ;;[3] S ::= (#x20 | #x9 | #xD | #xA)+
|
|
56375
|
191 (whitespace "[ \t\n\r]"))
|
|
68534
|
192 ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6]
|
|
|
193 ;; | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF]
|
|
|
194 ;; | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF]
|
|
|
195 ;; | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF]
|
|
56375
|
196 (defvar xml-name-start-char-re (concat "[" start-chars "]"))
|
|
68534
|
197 ;;[4a] NameChar ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040]
|
|
56375
|
198 (defvar xml-name-char-re (concat "[" name-chars "]"))
|
|
68534
|
199 ;;[5] Name ::= NameStartChar (NameChar)*
|
|
56375
|
200 (defvar xml-name-re (concat xml-name-start-char-re xml-name-char-re "*"))
|
|
68534
|
201 ;;[6] Names ::= Name (#x20 Name)*
|
|
56375
|
202 (defvar xml-names-re (concat xml-name-re "\\(?: " xml-name-re "\\)*"))
|
|
68534
|
203 ;;[7] Nmtoken ::= (NameChar)+
|
|
56375
|
204 (defvar xml-nmtoken-re (concat xml-name-char-re "+"))
|
|
68534
|
205 ;;[8] Nmtokens ::= Nmtoken (#x20 Nmtoken)*
|
|
56375
|
206 (defvar xml-nmtokens-re (concat xml-nmtoken-re "\\(?: " xml-name-re "\\)*"))
|
|
68534
|
207 ;;[66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';'
|
|
56375
|
208 (defvar xml-char-ref-re "\\(?:&#[0-9]+;\\|&#x[0-9a-fA-F]+;\\)")
|
|
68534
|
209 ;;[68] EntityRef ::= '&' Name ';'
|
|
56375
|
210 (defvar xml-entity-ref (concat "&" xml-name-re ";"))
|
|
68534
|
211 ;;[69] PEReference ::= '%' Name ';'
|
|
56375
|
212 (defvar xml-pe-reference-re (concat "%" xml-name-re ";"))
|
|
68534
|
213 ;;[67] Reference ::= EntityRef | CharRef
|
|
56375
|
214 (defvar xml-reference-re (concat "\\(?:" xml-entity-ref "\\|" xml-char-ref-re "\\)"))
|
|
68534
|
215 ;;[10] AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'"
|
|
63282
|
216 (defvar xml-att-value-re (concat "\\(?:\"\\(?:[^&\"]\\|" xml-reference-re "\\)*\"\\|"
|
|
|
217 "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)"))
|
|
68534
|
218 ;;[56] TokenizedType ::= 'ID' [VC: ID] [VC: One ID per Element Type] [VC: ID Attribute Default]
|
|
|
219 ;; | 'IDREF' [VC: IDREF]
|
|
|
220 ;; | 'IDREFS' [VC: IDREF]
|
|
|
221 ;; | 'ENTITY' [VC: Entity Name]
|
|
|
222 ;; | 'ENTITIES' [VC: Entity Name]
|
|
|
223 ;; | 'NMTOKEN' [VC: Name Token]
|
|
|
224 ;; | 'NMTOKENS' [VC: Name Token]
|
|
63282
|
225 (defvar xml-tokenized-type-re "\\(?:ID\\|IDREF\\|IDREFS\\|ENTITY\\|ENTITIES\\|NMTOKEN\\|NMTOKENS\\)")
|
|
68534
|
226 ;;[58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
|
|
63282
|
227 (defvar xml-notation-type-re (concat "\\(?:NOTATION" whitespace "(" whitespace "*" xml-name-re
|
|
|
228 "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)"))
|
|
68534
|
229 ;;[59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' [VC: Enumeration] [VC: No Duplicate Tokens]
|
|
63282
|
230 (defvar xml-enumeration-re (concat "\\(?:(" whitespace "*" xml-nmtoken-re
|
|
|
231 "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*"
|
|
|
232 whitespace ")\\)"))
|
|
68534
|
233 ;;[57] EnumeratedType ::= NotationType | Enumeration
|
|
63282
|
234 (defvar xml-enumerated-type-re (concat "\\(?:" xml-notation-type-re "\\|" xml-enumeration-re "\\)"))
|
|
68534
|
235 ;;[54] AttType ::= StringType | TokenizedType | EnumeratedType
|
|
|
236 ;;[55] StringType ::= 'CDATA'
|
|
63282
|
237 (defvar xml-att-type-re (concat "\\(?:CDATA\\|" xml-tokenized-type-re "\\|" xml-notation-type-re"\\|" xml-enumerated-type-re "\\)"))
|
|
68534
|
238 ;;[60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue)
|
|
63282
|
239 (defvar xml-default-decl-re (concat "\\(?:#REQUIRED\\|#IMPLIED\\|\\(?:#FIXED" whitespace "\\)*" xml-att-value-re "\\)"))
|
|
68534
|
240 ;;[53] AttDef ::= S Name S AttType S DefaultDecl
|
|
63282
|
241 (defvar xml-att-def-re (concat "\\(?:" whitespace "*" xml-name-re
|
|
|
242 whitespace "*" xml-att-type-re
|
|
|
243 whitespace "*" xml-default-decl-re "\\)"))
|
|
68534
|
244 ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
|
|
|
245 ;; | "'" ([^%&'] | PEReference | Reference)* "'"
|
|
56375
|
246 (defvar xml-entity-value-re (concat "\\(?:\"\\(?:[^%&\"]\\|" xml-pe-reference-re
|
|
|
247 "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|"
|
|
|
248 xml-pe-reference-re "\\|" xml-reference-re "\\)*'\\)")))
|
|
|
249 ;;[75] ExternalID ::= 'SYSTEM' S SystemLiteral
|
|
|
250 ;; | 'PUBLIC' S PubidLiteral S SystemLiteral
|
|
|
251 ;;[76] NDataDecl ::= S 'NDATA' S
|
|
|
252 ;;[73] EntityDef ::= EntityValue| (ExternalID NDataDecl?)
|
|
|
253 ;;[71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
|
|
|
254 ;;[74] PEDef ::= EntityValue | ExternalID
|
|
|
255 ;;[72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
|
|
|
256 ;;[70] EntityDecl ::= GEDecl | PEDecl
|
|
|
257
|
|
51102
|
258 ;; Note that this is setup so that we can do whitespace-skipping with
|
|
|
259 ;; `(skip-syntax-forward " ")', inter alia. Previously this was slow
|
|
|
260 ;; compared with `re-search-forward', but that has been fixed. Also
|
|
|
261 ;; note that the standard syntax table contains other characters with
|
|
|
262 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
|
|
|
263 ;; which we might skip whitespace -- specifically, they're not
|
|
|
264 ;; NameChars [XML 4].
|
|
|
265
|
|
|
266 (defvar xml-syntax-table
|
|
|
267 (let ((table (make-syntax-table)))
|
|
|
268 ;; Get space syntax correct per XML [3].
|
|
|
269 (dotimes (c 31)
|
|
|
270 (modify-syntax-entry c "." table)) ; all are space in standard table
|
|
68534
|
271 (dolist (c '(?\t ?\n ?\r)) ; these should be space
|
|
51102
|
272 (modify-syntax-entry c " " table))
|
|
|
273 ;; For skipping attributes.
|
|
|
274 (modify-syntax-entry ?\" "\"" table)
|
|
|
275 (modify-syntax-entry ?' "\"" table)
|
|
|
276 ;; Non-alnum name chars should be symbol constituents (`-' and `_'
|
|
|
277 ;; are OK by default).
|
|
|
278 (modify-syntax-entry ?. "_" table)
|
|
|
279 (modify-syntax-entry ?: "_" table)
|
|
|
280 ;; XML [89]
|
|
66646
|
281 (unless (featurep 'xemacs)
|
|
|
282 (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005
|
|
|
283 #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC
|
|
|
284 #x30FD #x30FE))
|
|
|
285 (modify-syntax-entry (decode-char 'ucs c) "w" table)))
|
|
51102
|
286 ;; Fixme: rest of [4]
|
|
|
287 table)
|
|
|
288 "Syntax table used by `xml-parse-region'.")
|
|
|
289
|
|
|
290 ;; XML [5]
|
|
|
291 ;; Note that [:alpha:] matches all multibyte chars with word syntax.
|
|
51105
|
292 (eval-and-compile
|
|
|
293 (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*"))
|
|
51102
|
294
|
|
|
295 ;; Fixme: This needs re-writing to deal with the XML grammar properly, i.e.
|
|
|
296 ;; document ::= prolog element Misc*
|
|
|
297 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
|
|
|
298
|
|
|
299 ;;;###autoload
|
|
51930
|
300 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns)
|
|
30329
|
301 "Parse the region from BEG to END in BUFFER.
|
|
|
302 If BUFFER is nil, it defaults to the current buffer.
|
|
|
303 Returns the XML list for the region, or raises an error if the region
|
|
51930
|
304 is not well-formed XML.
|
|
30329
|
305 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped,
|
|
51930
|
306 and returned as the first element of the list.
|
|
|
307 If PARSE-NS is non-nil, then QNAMES are expanded."
|
|
68535
|
308 ;; Use fixed syntax table to ensure regexp char classes and syntax
|
|
|
309 ;; specs DTRT.
|
|
|
310 (with-syntax-table (standard-syntax-table)
|
|
|
311 (let ((case-fold-search nil) ; XML is case-sensitive.
|
|
|
312 xml result dtd)
|
|
|
313 (save-excursion
|
|
|
314 (if buffer
|
|
|
315 (set-buffer buffer))
|
|
|
316 (save-restriction
|
|
|
317 (narrow-to-region beg end)
|
|
51102
|
318 (goto-char (point-min))
|
|
|
319 (while (not (eobp))
|
|
|
320 (if (search-forward "<" nil t)
|
|
|
321 (progn
|
|
|
322 (forward-char -1)
|
|
52975
|
323 (setq result (xml-parse-tag parse-dtd parse-ns))
|
|
56375
|
324 (if (and xml result (not xml-sub-parser))
|
|
51102
|
325 ;; translation of rule [1] of XML specifications
|
|
56375
|
326 (error "XML: (Not Well-Formed) Only one root tag allowed")
|
|
30329
|
327 (cond
|
|
42031
|
328 ((null result))
|
|
52975
|
329 ((and (listp (car result))
|
|
|
330 parse-dtd)
|
|
42031
|
331 (setq dtd (car result))
|
|
51102
|
332 (if (cdr result) ; possible leading comment
|
|
|
333 (add-to-list 'xml (cdr result))))
|
|
30329
|
334 (t
|
|
51102
|
335 (add-to-list 'xml result)))))
|
|
|
336 (goto-char (point-max))))
|
|
|
337 (if parse-dtd
|
|
|
338 (cons dtd (nreverse xml))
|
|
|
339 (nreverse xml)))))))
|
|
30329
|
340
|
|
54877
|
341 (defun xml-maybe-do-ns (name default xml-ns)
|
|
54937
|
342 "Perform any namespace expansion.
|
|
|
343 NAME is the name to perform the expansion on.
|
|
54877
|
344 DEFAULT is the default namespace. XML-NS is a cons of namespace
|
|
|
345 names to uris. When namespace-aware parsing is off, then XML-NS
|
|
|
346 is nil.
|
|
52975
|
347
|
|
54877
|
348 During namespace-aware parsing, any name without a namespace is
|
|
|
349 put into the namespace identified by DEFAULT. nil is used to
|
|
|
350 specify that the name shouldn't be given a namespace."
|
|
|
351 (if (consp xml-ns)
|
|
|
352 (let* ((nsp (string-match ":" name))
|
|
|
353 (lname (if nsp (substring name (match-end 0)) name))
|
|
|
354 (prefix (if nsp (substring name 0 (match-beginning 0)) default))
|
|
|
355 (special (and (string-equal lname "xmlns") (not prefix)))
|
|
|
356 ;; Setting default to nil will insure that there is not
|
|
|
357 ;; matching cons in xml-ns. In which case we
|
|
|
358 (ns (or (cdr (assoc (if special "xmlns" prefix)
|
|
|
359 xml-ns))
|
|
56375
|
360 "")))
|
|
54877
|
361 (cons ns (if special "" lname)))
|
|
|
362 (intern name)))
|
|
30329
|
363
|
|
56375
|
364 (defun xml-parse-fragment (&optional parse-dtd parse-ns)
|
|
|
365 "Parse xml-like fragments."
|
|
|
366 (let ((xml-sub-parser t)
|
|
|
367 children)
|
|
|
368 (while (not (eobp))
|
|
|
369 (let ((bit (xml-parse-tag
|
|
|
370 parse-dtd parse-ns)))
|
|
|
371 (if children
|
|
|
372 (setq children (append (list bit) children))
|
|
|
373 (if (stringp bit)
|
|
|
374 (setq children (list bit))
|
|
|
375 (setq children bit)))))
|
|
|
376 (reverse children)))
|
|
|
377
|
|
51930
|
378 (defun xml-parse-tag (&optional parse-dtd parse-ns)
|
|
51102
|
379 "Parse the tag at point.
|
|
30329
|
380 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
|
|
|
381 returned as the first element in the list.
|
|
51930
|
382 If PARSE-NS is non-nil, then QNAMES are expanded.
|
|
30329
|
383 Returns one of:
|
|
51102
|
384 - a list : the matching node
|
|
|
385 - nil : the point is not looking at a tag.
|
|
|
386 - a pair : the first element is the DTD, the second is the node."
|
|
56375
|
387 (let ((xml-validating-parser (or parse-dtd xml-validating-parser))
|
|
|
388 (xml-ns (if (consp parse-ns)
|
|
51930
|
389 parse-ns
|
|
|
390 (if parse-ns
|
|
|
391 (list
|
|
68534
|
392 ;; Default for empty prefix is no namespace
|
|
56375
|
393 (cons "" "")
|
|
54877
|
394 ;; "xml" namespace
|
|
56375
|
395 (cons "xml" "http://www.w3.org/XML/1998/namespace")
|
|
51930
|
396 ;; We need to seed the xmlns namespace
|
|
56375
|
397 (cons "xmlns" "http://www.w3.org/2000/xmlns/"))))))
|
|
51930
|
398 (cond
|
|
|
399 ;; Processing instructions (like the <?xml version="1.0"?> tag at the
|
|
|
400 ;; beginning of a document).
|
|
|
401 ((looking-at "<\\?")
|
|
|
402 (search-forward "?>")
|
|
|
403 (skip-syntax-forward " ")
|
|
|
404 (xml-parse-tag parse-dtd xml-ns))
|
|
|
405 ;; Character data (CDATA) sections, in which no tag should be interpreted
|
|
|
406 ((looking-at "<!\\[CDATA\\[")
|
|
|
407 (let ((pos (match-end 0)))
|
|
|
408 (unless (search-forward "]]>" nil t)
|
|
56375
|
409 (error "XML: (Not Well Formed) CDATA section does not end anywhere in the document"))
|
|
58698
|
410 (concat
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
411 (buffer-substring-no-properties pos (match-beginning 0))
|
|
58698
|
412 (xml-parse-string))))
|
|
51930
|
413 ;; DTD for the document
|
|
|
414 ((looking-at "<!DOCTYPE")
|
|
56375
|
415 (let ((dtd (xml-parse-dtd parse-ns)))
|
|
|
416 (skip-syntax-forward " ")
|
|
|
417 (if xml-validating-parser
|
|
|
418 (cons dtd (xml-parse-tag nil xml-ns))
|
|
|
419 (xml-parse-tag nil xml-ns))))
|
|
51930
|
420 ;; skip comments
|
|
|
421 ((looking-at "<!--")
|
|
|
422 (search-forward "-->")
|
|
|
423 nil)
|
|
|
424 ;; end tag
|
|
|
425 ((looking-at "</")
|
|
|
426 '())
|
|
|
427 ;; opening tag
|
|
|
428 ((looking-at "<\\([^/>[:space:]]+\\)")
|
|
|
429 (goto-char (match-end 1))
|
|
52975
|
430
|
|
|
431 ;; Parse this node
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
432 (let* ((node-name (match-string-no-properties 1))
|
|
68534
|
433 ;; Parse the attribute list.
|
|
|
434 (attrs (xml-parse-attlist xml-ns))
|
|
|
435 children pos)
|
|
51930
|
436
|
|
68534
|
437 ;; add the xmlns:* attrs to our cache
|
|
|
438 (when (consp xml-ns)
|
|
54877
|
439 (dolist (attr attrs)
|
|
|
440 (when (and (consp (car attr))
|
|
56375
|
441 (equal "http://www.w3.org/2000/xmlns/"
|
|
|
442 (caar attr)))
|
|
|
443 (push (cons (cdar attr) (cdr attr))
|
|
54877
|
444 xml-ns))))
|
|
|
445
|
|
68534
|
446 (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns)))
|
|
54877
|
447
|
|
51930
|
448 ;; is this an empty element ?
|
|
|
449 (if (looking-at "/>")
|
|
56375
|
450 (progn
|
|
|
451 (forward-char 2)
|
|
|
452 (nreverse children))
|
|
|
453
|
|
|
454 ;; is this a valid start tag ?
|
|
|
455 (if (eq (char-after) ?>)
|
|
|
456 (progn
|
|
|
457 (forward-char 1)
|
|
|
458 ;; Now check that we have the right end-tag. Note that this
|
|
|
459 ;; one might contain spaces after the tag name
|
|
|
460 (let ((end (concat "</" node-name "\\s-*>")))
|
|
|
461 (while (not (looking-at end))
|
|
|
462 (cond
|
|
|
463 ((looking-at "</")
|
|
|
464 (error "XML: (Not Well-Formed) Invalid end tag (expecting %s) at pos %d"
|
|
|
465 node-name (point)))
|
|
|
466 ((= (char-after) ?<)
|
|
|
467 (let ((tag (xml-parse-tag nil xml-ns)))
|
|
|
468 (when tag
|
|
|
469 (push tag children))))
|
|
|
470 (t
|
|
|
471 (let ((expansion (xml-parse-string)))
|
|
|
472 (setq children
|
|
|
473 (if (stringp expansion)
|
|
|
474 (if (stringp (car children))
|
|
|
475 ;; The two strings were separated by a comment.
|
|
66646
|
476 (setq children (append (list (concat (car children) expansion))
|
|
56375
|
477 (cdr children)))
|
|
|
478 (setq children (append (list expansion) children)))
|
|
|
479 (setq children (append expansion children))))))))
|
|
30329
|
480
|
|
56375
|
481 (goto-char (match-end 0))
|
|
|
482 (nreverse children)))
|
|
|
483 ;; This was an invalid start tag (Expected ">", but didn't see it.)
|
|
|
484 (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
|
485 (buffer-substring-no-properties (- (point) 10) (+ (point) 1)))))))
|
|
56375
|
486 (t ;; (Not one of PI, CDATA, Comment, End tag, or Start tag)
|
|
|
487 (unless xml-sub-parser ; Usually, we error out.
|
|
|
488 (error "XML: (Well-Formed) Invalid character"))
|
|
|
489
|
|
|
490 ;; However, if we're parsing incrementally, then we need to deal
|
|
|
491 ;; with stray CDATA.
|
|
|
492 (xml-parse-string)))))
|
|
49036
|
493
|
|
56375
|
494 (defun xml-parse-string ()
|
|
|
495 "Parse the next whatever. Could be a string, or an element."
|
|
68534
|
496 (let* ((pos (point))
|
|
98529
|
497 (string (progn (skip-chars-forward "^<")
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
498 (buffer-substring-no-properties pos (point)))))
|
|
68534
|
499 ;; Clean up the string. As per XML specifications, the XML
|
|
|
500 ;; processor should always pass the whole string to the
|
|
|
501 ;; application. But \r's should be replaced:
|
|
|
502 ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends
|
|
|
503 (setq pos 0)
|
|
|
504 (while (string-match "\r\n?" string pos)
|
|
|
505 (setq string (replace-match "\n" t t string))
|
|
|
506 (setq pos (1+ (match-beginning 0))))
|
|
42031
|
507
|
|
68534
|
508 (xml-substitute-special string)))
|
|
30329
|
509
|
|
54877
|
510 (defun xml-parse-attlist (&optional xml-ns)
|
|
54937
|
511 "Return the attribute-list after point.
|
|
|
512 Leave point at the first non-blank character after the tag."
|
|
42031
|
513 (let ((attlist ())
|
|
52975
|
514 end-pos name)
|
|
51102
|
515 (skip-syntax-forward " ")
|
|
|
516 (while (looking-at (eval-when-compile
|
|
|
517 (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*")))
|
|
54877
|
518 (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
|
519 (setq name (xml-maybe-do-ns (match-string-no-properties 1) nil xml-ns))
|
|
54877
|
520 (goto-char end-pos)
|
|
30329
|
521
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
522 ;; 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
|
523
|
|
30329
|
524 ;; Do we have a string between quotes (or double-quotes),
|
|
|
525 ;; 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
|
526 (if (looking-at "\"\\([^\"]*\\)\"")
|
|
52975
|
527 (setq end-pos (match-end 0))
|
|
50210
|
528 (if (looking-at "'\\([^']*\\)'")
|
|
52975
|
529 (setq end-pos (match-end 0))
|
|
56375
|
530 (error "XML: (Not Well-Formed) Attribute values must be given between quotes")))
|
|
30329
|
531
|
|
|
532 ;; Each attribute must be unique within a given element
|
|
|
533 (if (assoc name attlist)
|
|
56375
|
534 (error "XML: (Not Well-Formed) Each attribute must be unique within an element"))
|
|
49036
|
535
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
536 ;; 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
|
537 ;; in the attributes
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
538 (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
|
539 (pos 0))
|
|
51102
|
540 (replace-regexp-in-string "\\s-\\{2,\\}" " " string)
|
|
56375
|
541 (let ((expansion (xml-substitute-special string)))
|
|
|
542 (unless (stringp expansion)
|
|
68534
|
543 ; We say this is the constraint. It is acctually that
|
|
|
544 ; external entities nor "<" can be in an attribute value.
|
|
56375
|
545 (error "XML: (Not Well-Formed) Entities in attributes cannot expand into elements"))
|
|
|
546 (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
|
547
|
|
52975
|
548 (goto-char end-pos)
|
|
51102
|
549 (skip-syntax-forward " "))
|
|
42031
|
550 (nreverse attlist)))
|
|
30329
|
551
|
|
|
552 ;;*******************************************************************
|
|
|
553 ;;**
|
|
|
554 ;;** The DTD (document type declaration)
|
|
|
555 ;;** The following functions know how to skip or parse the DTD of
|
|
|
556 ;;** a document
|
|
|
557 ;;**
|
|
|
558 ;;*******************************************************************
|
|
|
559
|
|
51102
|
560 ;; Fixme: This fails at least if the DTD contains conditional sections.
|
|
|
561
|
|
|
562 (defun xml-skip-dtd ()
|
|
|
563 "Skip the DTD at point.
|
|
30329
|
564 This follows the rule [28] in the XML specifications."
|
|
56375
|
565 (let ((xml-validating-parser nil))
|
|
|
566 (xml-parse-dtd)))
|
|
30329
|
567
|
|
56375
|
568 (defun xml-parse-dtd (&optional parse-ns)
|
|
51102
|
569 "Parse the DTD at point."
|
|
|
570 (forward-char (eval-when-compile (length "<!DOCTYPE")))
|
|
|
571 (skip-syntax-forward " ")
|
|
56375
|
572 (if (and (looking-at ">")
|
|
|
573 xml-validating-parser)
|
|
|
574 (error "XML: (Validity) Invalid DTD (expecting name of the document)"))
|
|
49036
|
575
|
|
42031
|
576 ;; Get the name of the document
|
|
51102
|
577 (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
|
578 (let ((dtd (list (match-string-no-properties 0) 'dtd))
|
|
42031
|
579 type element end-pos)
|
|
30329
|
580 (goto-char (match-end 0))
|
|
|
581
|
|
51102
|
582 (skip-syntax-forward " ")
|
|
|
583 ;; XML [75]
|
|
|
584 (cond ((looking-at "PUBLIC\\s-+")
|
|
|
585 (goto-char (match-end 0))
|
|
|
586 (unless (or (re-search-forward
|
|
|
587 "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\""
|
|
|
588 nil t)
|
|
|
589 (re-search-forward
|
|
|
590 "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'"
|
|
|
591 nil t))
|
|
56375
|
592 (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
|
593 (let ((pubid (match-string-no-properties 1)))
|
|
56375
|
594 (skip-syntax-forward " ")
|
|
51102
|
595 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
|
|
|
596 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
|
|
56375
|
597 (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
|
598 (push (list pubid (match-string-no-properties 1) 'public) dtd)))
|
|
51102
|
599 ((looking-at "SYSTEM\\s-+")
|
|
|
600 (goto-char (match-end 0))
|
|
|
601 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
|
|
|
602 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t))
|
|
56375
|
603 (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
|
604 (push (list (match-string-no-properties 1) 'system) dtd)))
|
|
51102
|
605 (skip-syntax-forward " ")
|
|
|
606 (if (eq ?> (char-after))
|
|
|
607 (forward-char)
|
|
|
608 (if (not (eq (char-after) ?\[))
|
|
56375
|
609 (error "XML: Bad DTD")
|
|
51102
|
610 (forward-char)
|
|
|
611 ;; Parse the rest of the DTD
|
|
63282
|
612 ;; Fixme: Deal with NOTATION, PIs.
|
|
51102
|
613 (while (not (looking-at "\\s-*\\]"))
|
|
|
614 (skip-syntax-forward " ")
|
|
|
615 (cond
|
|
49036
|
616
|
|
51102
|
617 ;; Translation of rule [45] of XML specifications
|
|
|
618 ((looking-at
|
|
|
619 "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>")
|
|
|
620
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
621 (setq element (match-string-no-properties 1)
|
|
51102
|
622 type (match-string-no-properties 2))
|
|
|
623 (setq end-pos (match-end 0))
|
|
30329
|
624
|
|
51102
|
625 ;; Translation of rule [46] of XML specifications
|
|
|
626 (cond
|
|
|
627 ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration
|
|
|
628 (setq type 'empty))
|
|
|
629 ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents
|
|
|
630 (setq type 'any))
|
|
|
631 ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47])
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
632 (setq type (xml-parse-elem-type (match-string-no-properties 1 type))))
|
|
51102
|
633 ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution
|
|
|
634 nil)
|
|
|
635 (t
|
|
56375
|
636 (if xml-validating-parser
|
|
57346
|
637 (error "XML: (Validity) Invalid element type in the DTD"))))
|
|
58723
|
638
|
|
51102
|
639 ;; rule [45]: the element declaration must be unique
|
|
56375
|
640 (if (and (assoc element dtd)
|
|
|
641 xml-validating-parser)
|
|
|
642 (error "XML: (Validity) Element declarations must be unique in a DTD (<%s>)"
|
|
53011
|
643 element))
|
|
30329
|
644
|
|
51102
|
645 ;; Store the element in the DTD
|
|
|
646 (push (list element type) dtd)
|
|
|
647 (goto-char end-pos))
|
|
63282
|
648
|
|
|
649 ;; Translation of rule [52] of XML specifications
|
|
|
650 ((looking-at (concat "<!ATTLIST[ \t\n\r]*\\(" xml-name-re
|
|
|
651 "\\)[ \t\n\r]*\\(" xml-att-def-re
|
|
|
652 "\\)*[ \t\n\r]*>"))
|
|
|
653
|
|
|
654 ;; We don't do anything with ATTLIST currently
|
|
|
655 (goto-char (match-end 0)))
|
|
|
656
|
|
51102
|
657 ((looking-at "<!--")
|
|
|
658 (search-forward "-->"))
|
|
56375
|
659 ((looking-at (concat "<!ENTITY[ \t\n\r]*\\(" xml-name-re
|
|
|
660 "\\)[ \t\n\r]*\\(" xml-entity-value-re
|
|
|
661 "\\)[ \t\n\r]*>"))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
662 (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
|
663 (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
|
664 (- (length (match-string-no-properties 2)) 1))))
|
|
63282
|
665 (goto-char (match-end 0))
|
|
56375
|
666 (setq xml-entity-alist
|
|
|
667 (append xml-entity-alist
|
|
|
668 (list (cons name
|
|
|
669 (with-temp-buffer
|
|
|
670 (insert value)
|
|
|
671 (goto-char (point-min))
|
|
|
672 (xml-parse-fragment
|
|
|
673 xml-validating-parser
|
|
|
674 parse-ns))))))))
|
|
|
675 ((or (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re
|
|
|
676 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
|
|
|
677 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
|
|
|
678 (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re
|
|
|
679 "\\)[ \t\n\r]+PUBLIC[ \t\n\r]+"
|
|
|
680 "\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
|
|
|
681 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'"
|
|
|
682 "[ \t\n\r]+\\(\"[^\"]*\"\\|'[^']*'\\)"
|
|
|
683 "[ \t\n\r]*>")))
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
684 (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
|
685 (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
|
686 (- (length (match-string-no-properties 2)) 1))))
|
|
63282
|
687 (goto-char (match-end 0))
|
|
56375
|
688 (setq xml-entity-alist
|
|
|
689 (append xml-entity-alist
|
|
|
690 (list (cons name (with-temp-buffer
|
|
|
691 (insert-file-contents file)
|
|
|
692 (goto-char (point-min))
|
|
|
693 (xml-parse-fragment
|
|
|
694 xml-validating-parser
|
|
|
695 parse-ns))))))))
|
|
58724
|
696 ;; skip parameter entity declarations
|
|
|
697 ((or (looking-at (concat "<!ENTITY[ \t\n\r]+%[ \t\n\r]+\\(" xml-name-re
|
|
|
698 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
|
|
|
699 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
|
|
|
700 (looking-at (concat "<!ENTITY[ \t\n\r]+"
|
|
|
701 "%[ \t\n\r]+"
|
|
|
702 "\\(" xml-name-re "\\)[ \t\n\r]+"
|
|
|
703 "PUBLIC[ \t\n\r]+"
|
|
|
704 "\\(\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
|
|
|
705 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'\\)[ \t\n\r]+"
|
|
|
706 "\\(\"[^\"]+\"\\|'[^']+'\\)"
|
|
|
707 "[ \t\n\r]*>")))
|
|
|
708 (goto-char (match-end 0)))
|
|
|
709 ;; skip parameter entities
|
|
|
710 ((looking-at (concat "%" xml-name-re ";"))
|
|
|
711 (goto-char (match-end 0)))
|
|
51102
|
712 (t
|
58721
ff6b2a793277
Ensure that validity messages only show when xml-validating-parser is set.
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
713 (when xml-validating-parser
|
|
58723
|
714 (error "XML: (Validity) Invalid DTD item"))))))
|
|
56375
|
715 (if (looking-at "\\s-*]>")
|
|
63282
|
716 (goto-char (match-end 0))))
|
|
53011
|
717 (nreverse dtd)))
|
|
30329
|
718
|
|
|
719 (defun xml-parse-elem-type (string)
|
|
51102
|
720 "Convert element type STRING into a Lisp structure."
|
|
30329
|
721
|
|
|
722 (let (elem modifier)
|
|
|
723 (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
|
|
|
724 (progn
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
725 (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
|
726 modifier (match-string-no-properties 2 string))
|
|
30329
|
727 (if (string-match "|" elem)
|
|
42031
|
728 (setq elem (cons 'choice
|
|
30329
|
729 (mapcar 'xml-parse-elem-type
|
|
|
730 (split-string elem "|"))))
|
|
|
731 (if (string-match "," elem)
|
|
42031
|
732 (setq elem (cons 'seq
|
|
30329
|
733 (mapcar 'xml-parse-elem-type
|
|
51102
|
734 (split-string elem ",")))))))
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
735 (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
|
736 (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
|
737 modifier (match-string-no-properties 2 string))))
|
|
30329
|
738
|
|
42031
|
739 (if (and (stringp elem) (string= elem "#PCDATA"))
|
|
|
740 (setq elem 'pcdata))
|
|
49036
|
741
|
|
42031
|
742 (cond
|
|
|
743 ((string= modifier "+")
|
|
|
744 (list '+ elem))
|
|
|
745 ((string= modifier "*")
|
|
|
746 (list '* elem))
|
|
|
747 ((string= modifier "?")
|
|
49787
|
748 (list '\? elem))
|
|
42031
|
749 (t
|
|
|
750 elem))))
|
|
30329
|
751
|
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
752 ;;*******************************************************************
|
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
diff
changeset
|
753 ;;**
|
|
30329
|
754 ;;** Substituting special XML sequences
|
|
|
755 ;;**
|
|
|
756 ;;*******************************************************************
|
|
|
757
|
|
|
758 (defun xml-substitute-special (string)
|
|
51102
|
759 "Return STRING, after subsituting entity references."
|
|
|
760 ;; This originally made repeated passes through the string from the
|
|
|
761 ;; beginning, which isn't correct, since then either "&amp;" or
|
|
|
762 ;; "&amp;" won't DTRT.
|
|
56375
|
763
|
|
|
764 (let ((point 0)
|
|
|
765 children end-point)
|
|
58698
|
766 (while (string-match "&\\([^;]*\\);" string point)
|
|
56375
|
767 (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
|
768 (let* ((this-part (match-string-no-properties 1 string))
|
|
56375
|
769 (prev-part (substring string point (match-beginning 0)))
|
|
|
770 (entity (assoc this-part xml-entity-alist))
|
|
|
771 (expansion
|
|
|
772 (cond ((string-match "#\\([0-9]+\\)" this-part)
|
|
|
773 (let ((c (decode-char
|
|
|
774 'ucs
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
775 (string-to-number (match-string-no-properties 1 this-part)))))
|
|
56375
|
776 (if c (string c))))
|
|
|
777 ((string-match "#x\\([[:xdigit:]]+\\)" this-part)
|
|
|
778 (let ((c (decode-char
|
|
|
779 'ucs
|
76507
d1760553962c
* xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
diff
changeset
|
780 (string-to-number (match-string-no-properties 1 this-part) 16))))
|
|
56375
|
781 (if c (string c))))
|
|
|
782 (entity
|
|
|
783 (cdr entity))
|
|
58698
|
784 ((eq (length this-part) 0)
|
|
58723
|
785 (error "XML: (Not Well-Formed) No entity given"))
|
|
56375
|
786 (t
|
|
62752
|
787 (if xml-validating-parser
|
|
56375
|
788 (error "XML: (Validity) Undefined entity `%s'"
|
|
62752
|
789 this-part)
|
|
|
790 xml-undefined-entity)))))
|
|
30329
|
791
|
|
56375
|
792 (cond ((null children)
|
|
57367
|
793 ;; FIXME: If we have an entity that expands into XML, this won't work.
|
|
|
794 (setq children
|
|
|
795 (concat prev-part expansion)))
|
|
56375
|
796 ((stringp children)
|
|
|
797 (if (stringp expansion)
|
|
|
798 (setq children (concat children prev-part expansion))
|
|
|
799 (setq children (list expansion (concat prev-part children)))))
|
|
|
800 ((and (stringp expansion)
|
|
|
801 (stringp (car children)))
|
|
|
802 (setcar children (concat prev-part expansion (car children))))
|
|
|
803 ((stringp expansion)
|
|
|
804 (setq children (append (concat prev-part expansion)
|
|
|
805 children)))
|
|
|
806 ((stringp (car children))
|
|
|
807 (setcar children (concat (car children) prev-part))
|
|
|
808 (setq children (append expansion children)))
|
|
|
809 (t
|
|
|
810 (setq children (list expansion
|
|
|
811 prev-part
|
|
|
812 children))))
|
|
|
813 (setq point end-point)))
|
|
|
814 (cond ((stringp children)
|
|
|
815 (concat children (substring string point)))
|
|
|
816 ((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
|
817 (concat (car (last children)) (substring string point)))
|
|
56375
|
818 ((null children)
|
|
|
819 string)
|
|
|
820 (t
|
57344
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
821 (concat (mapconcat 'identity
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
822 (nreverse children)
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
823 "")
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
824 (substring string point))))))
|
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
825
|
|
30329
|
826 ;;*******************************************************************
|
|
|
827 ;;**
|
|
|
828 ;;** Printing a tree.
|
|
|
829 ;;** This function is intended mainly for debugging purposes.
|
|
|
830 ;;**
|
|
|
831 ;;*******************************************************************
|
|
|
832
|
|
55253
|
833 (defun xml-debug-print (xml &optional indent-string)
|
|
|
834 "Outputs the XML in the current buffer.
|
|
|
835 XML can be a tree or a list of nodes.
|
|
|
836 The first line is indented with the optional INDENT-STRING."
|
|
|
837 (setq indent-string (or indent-string ""))
|
|
42031
|
838 (dolist (node xml)
|
|
55253
|
839 (xml-debug-print-internal node indent-string)))
|
|
|
840
|
|
|
841 (defalias 'xml-print 'xml-debug-print)
|
|
30329
|
842
|
|
87317
|
843 (defun xml-escape-string (string)
|
|
87318
|
844 "Return the string with entity substitutions made from
|
|
|
845 xml-entity-alist."
|
|
87317
|
846 (mapconcat (lambda (byte)
|
|
|
847 (let ((char (char-to-string byte)))
|
|
|
848 (if (rassoc char xml-entity-alist)
|
|
|
849 (concat "&" (car (rassoc char xml-entity-alist)) ";")
|
|
|
850 char)))
|
91342
3ec2fe8547cf
* xml.el (xml-escape-string): Don't do any encoding changes on the string.
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
851 ;; This differs from the non-unicode branch. Just
|
3ec2fe8547cf
* xml.el (xml-escape-string): Don't do any encoding changes on the string.
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
852 ;; grabbing the string works here.
|
3ec2fe8547cf
* xml.el (xml-escape-string): Don't do any encoding changes on the string.
Mark A. Hershberger <mah@everybody.org>
diff
changeset
|
853 string ""))
|
|
87317
|
854
|
|
42031
|
855 (defun xml-debug-print-internal (xml indent-string)
|
|
30329
|
856 "Outputs the XML tree in the current buffer.
|
|
51102
|
857 The first line is indented with INDENT-STRING."
|
|
30329
|
858 (let ((tree xml)
|
|
|
859 attlist)
|
|
51102
|
860 (insert indent-string ?< (symbol-name (xml-node-name tree)))
|
|
49036
|
861
|
|
30329
|
862 ;; output the attribute list
|
|
42031
|
863 (setq attlist (xml-node-attributes tree))
|
|
30329
|
864 (while attlist
|
|
87317
|
865 (insert ?\ (symbol-name (caar attlist)) "=\""
|
|
|
866 (xml-escape-string (cdar attlist)) ?\")
|
|
42031
|
867 (setq attlist (cdr attlist)))
|
|
49036
|
868
|
|
42031
|
869 (setq tree (xml-node-children tree))
|
|
30329
|
870
|
|
55253
|
871 (if (null tree)
|
|
|
872 (insert ?/ ?>)
|
|
|
873 (insert ?>)
|
|
30329
|
874
|
|
55253
|
875 ;; output the children
|
|
|
876 (dolist (node tree)
|
|
|
877 (cond
|
|
|
878 ((listp node)
|
|
|
879 (insert ?\n)
|
|
|
880 (xml-debug-print-internal node (concat indent-string " ")))
|
|
87317
|
881 ((stringp node)
|
|
|
882 (insert (xml-escape-string node)))
|
|
55253
|
883 (t
|
|
|
884 (error "Invalid XML tree"))))
|
|
|
885
|
|
|
886 (when (not (and (null (cdr tree))
|
|
|
887 (stringp (car tree))))
|
|
|
888 (insert ?\n indent-string))
|
|
|
889 (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>))))
|
|
30329
|
890
|
|
|
891 (provide 'xml)
|
|
|
892
|
|
55316
|
893 ;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b
|
|
30329
|
894 ;;; xml.el ends here
|
