Mercurial > emacs
annotate lisp/xml.el @ 71538:bc2ebc6f8c85
*** empty log message ***
author | Juanma Barranquero <lekktu@gmail.com> |
---|---|
date | Fri, 30 Jun 2006 08:39:28 +0000 |
parents | 3bd95f4f2941 |
children | ee72501a64b4 c5406394f567 |
rev | line source |
---|---|
38409
153f1b1f2efd
Emacs lisp coding convention fixes.
Pavel Janík <Pavel@Janik.cz>
parents:
37958
diff
changeset
|
1 ;;; xml.el --- XML parser |
30329 | 2 |
64762
41bb365f41c4
Update years in copyright notice; nfc.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
64091
diff
changeset
|
3 ;; Copyright (C) 2000, 2001, 2002, 2003, 2004, |
68651
3bd95f4f2941
Update years in copyright notice; nfc.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
68535
diff
changeset
|
4 ;; 2005, 2006 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 | |
14 ;; the Free Software Foundation; either version 2, or (at your option) | |
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
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
44 ;; <node1 attr1="name1" attr2="name2" ...>value |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
45 ;; <node2 attr3="name3" attr4="name4">value2</node2> |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
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
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
56 ;; The functions `xml-parse-file', `xml-parse-region' and |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
57 ;; `xml-parse-tag' return a list with the following format: |
30329 | 58 ;; |
59 ;; xml-list ::= (node node ...) | |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
60 ;; node ::= (qname attribute-list . child_node_list) |
30329 | 61 ;; child_node_list ::= child_node child_node ... |
62 ;; child_node ::= node | string | |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
63 ;; qname ::= (:namespace-uri . "name") | "name" |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
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
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
72 ;; TODO: |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
73 ;; * xml:base, xml:space support |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
74 ;; * more complete DOCTYPE parsing |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
75 ;; * pi support |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
76 |
30329 | 77 ;;; Code: |
78 | |
51102 | 79 ;; Note that {buffer-substring,match-string}-no-properties were |
80 ;; formerly used in several places, but that removes composition info. | |
81 | |
30329 | 82 ;;******************************************************************* |
83 ;;** | |
84 ;;** Macros to parse the list | |
85 ;;** | |
86 ;;******************************************************************* | |
87 | |
62752
d0f8033496b1
2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
58939
diff
changeset
|
88 (defconst xml-undefined-entity "?" |
d0f8033496b1
2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
58939
diff
changeset
|
89 "What to substitute for undefined entities") |
d0f8033496b1
2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
58939
diff
changeset
|
90 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
91 (defvar xml-entity-alist |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
92 '(("lt" . "<") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
93 ("gt" . ">") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
94 ("apos" . "'") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
95 ("quot" . "\"") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
96 ("amp" . "&")) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
97 "The defined entities. Entities are added to this when the DTD is parsed.") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
98 |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
99 (defvar xml-sub-parser nil |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
100 "Dynamically set this to a non-nil value if you want to parse an XML fragment.") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
101 |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
102 (defvar xml-validating-parser nil |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
103 "Set to non-nil to get validity checking.") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
104 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
105 (defsubst xml-node-name (node) |
30329 | 106 "Return the tag associated with NODE. |
54937 | 107 Without namespace-aware parsing, the tag is a symbol. |
108 | |
109 With namespace-aware parsing, the tag is a cons of a string | |
110 representing the uri of the namespace with the local name of the | |
111 tag. For example, | |
112 | |
113 <foo> | |
114 | |
115 would be represented by | |
116 | |
117 '(\"\" . \"foo\")." | |
118 | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
119 (car node)) |
30329 | 120 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
121 (defsubst xml-node-attributes (node) |
30329 | 122 "Return the list of attributes of NODE. |
123 The list can be nil." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
124 (nth 1 node)) |
30329 | 125 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
126 (defsubst xml-node-children (node) |
30329 | 127 "Return the list of children of NODE. |
128 This is a list of nodes, and it can be nil." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
129 (cddr node)) |
30329 | 130 |
131 (defun xml-get-children (node child-name) | |
132 "Return the children of NODE whose tag is CHILD-NAME. | |
54937 | 133 CHILD-NAME should match the value returned by `xml-node-name'." |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
134 (let ((match ())) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
135 (dolist (child (xml-node-children node)) |
54937 | 136 (if (and (listp child) |
137 (equal (xml-node-name child) child-name)) | |
138 (push child match))) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
139 (nreverse match))) |
30329 | 140 |
53375
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
141 (defun xml-get-attribute-or-nil (node attribute) |
30329 | 142 "Get from NODE the value of ATTRIBUTE. |
54937 | 143 Return nil if the attribute was not found. |
53375
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
144 |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
145 See also `xml-get-attribute'." |
54243
586ffda6e9f9
(xml-get-attribute-or-nil): Simplify.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
53384
diff
changeset
|
146 (cdr (assoc attribute (xml-node-attributes node)))) |
53375
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
147 |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
148 (defsubst xml-get-attribute (node attribute) |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
149 "Get from NODE the value of ATTRIBUTE. |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
150 An empty string is returned if the attribute was not found. |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
151 |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
152 See also `xml-get-attribute-or-nil'." |
e085973399ee
(xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents:
53011
diff
changeset
|
153 (or (xml-get-attribute-or-nil node attribute) "")) |
30329 | 154 |
155 ;;******************************************************************* | |
156 ;;** | |
157 ;;** Creating the list | |
158 ;;** | |
159 ;;******************************************************************* | |
160 | |
51102 | 161 ;;;###autoload |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
162 (defun xml-parse-file (file &optional parse-dtd parse-ns) |
51102 | 163 "Parse the well-formed XML file FILE. |
164 If FILE is already visited, use its buffer and don't kill it. | |
30329 | 165 Returns the top node with all its children. |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
166 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped. |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
167 If PARSE-NS is non-nil, then QNAMES are expanded." |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
168 (let ((keep)) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
169 (if (get-file-buffer file) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
170 (progn |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
171 (set-buffer (get-file-buffer file)) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
172 (setq keep (point))) |
51102 | 173 (let (auto-mode-alist) ; no need for xml-mode |
174 (find-file file))) | |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
175 |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
176 (let ((xml (xml-parse-region (point-min) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
177 (point-max) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
178 (current-buffer) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
179 parse-dtd parse-ns))) |
34825
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
180 (if keep |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
181 (goto-char keep) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
182 (kill-buffer (current-buffer))) |
2cad4cde52bd
(top level comment): Updated to reflect the fact that
Gerd Moellmann <gerd@gnu.org>
parents:
33977
diff
changeset
|
183 xml))) |
30329 | 184 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
185 |
58939
e696b8c143a0
(xml-name-re, xml-entity-value-re): Add defvars.
Richard M. Stallman <rms@gnu.org>
parents:
58724
diff
changeset
|
186 (defvar xml-name-re) |
e696b8c143a0
(xml-name-re, xml-entity-value-re): Add defvars.
Richard M. Stallman <rms@gnu.org>
parents:
58724
diff
changeset
|
187 (defvar xml-entity-value-re) |
64794
950e3342b9d7
(xml-att-def-re): Add defvar.
Richard M. Stallman <rms@gnu.org>
parents:
64762
diff
changeset
|
188 (defvar xml-att-def-re) |
58724
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
189 (let* ((start-chars (concat "[:alpha:]:_")) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
190 (name-chars (concat "-[:digit:]." start-chars)) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
191 ;;[3] S ::= (#x20 | #x9 | #xD | #xA)+ |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
192 (whitespace "[ \t\n\r]")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
193 ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
194 ;; | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
195 ;; | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
196 ;; | [#xF900-#xFDCF] | [#xFDF0-#xFFFD] | [#x10000-#xEFFFF] |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
197 (defvar xml-name-start-char-re (concat "[" start-chars "]")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
198 ;;[4a] NameChar ::= NameStartChar | "-" | "." | [0-9] | #xB7 | [#x0300-#x036F] | [#x203F-#x2040] |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
199 (defvar xml-name-char-re (concat "[" name-chars "]")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
200 ;;[5] Name ::= NameStartChar (NameChar)* |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
201 (defvar xml-name-re (concat xml-name-start-char-re xml-name-char-re "*")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
202 ;;[6] Names ::= Name (#x20 Name)* |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
203 (defvar xml-names-re (concat xml-name-re "\\(?: " xml-name-re "\\)*")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
204 ;;[7] Nmtoken ::= (NameChar)+ |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
205 (defvar xml-nmtoken-re (concat xml-name-char-re "+")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
206 ;;[8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
207 (defvar xml-nmtokens-re (concat xml-nmtoken-re "\\(?: " xml-name-re "\\)*")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
208 ;;[66] CharRef ::= '&#' [0-9]+ ';' | '&#x' [0-9a-fA-F]+ ';' |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
209 (defvar xml-char-ref-re "\\(?:&#[0-9]+;\\|&#x[0-9a-fA-F]+;\\)") |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
210 ;;[68] EntityRef ::= '&' Name ';' |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
211 (defvar xml-entity-ref (concat "&" xml-name-re ";")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
212 ;;[69] PEReference ::= '%' Name ';' |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
213 (defvar xml-pe-reference-re (concat "%" xml-name-re ";")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
214 ;;[67] Reference ::= EntityRef | CharRef |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
215 (defvar xml-reference-re (concat "\\(?:" xml-entity-ref "\\|" xml-char-ref-re "\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
216 ;;[10] AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'" |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
217 (defvar xml-att-value-re (concat "\\(?:\"\\(?:[^&\"]\\|" xml-reference-re "\\)*\"\\|" |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
218 "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
219 ;;[56] TokenizedType ::= 'ID' [VC: ID] [VC: One ID per Element Type] [VC: ID Attribute Default] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
220 ;; | 'IDREF' [VC: IDREF] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
221 ;; | 'IDREFS' [VC: IDREF] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
222 ;; | 'ENTITY' [VC: Entity Name] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
223 ;; | 'ENTITIES' [VC: Entity Name] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
224 ;; | 'NMTOKEN' [VC: Name Token] |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
225 ;; | 'NMTOKENS' [VC: Name Token] |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
226 (defvar xml-tokenized-type-re "\\(?:ID\\|IDREF\\|IDREFS\\|ENTITY\\|ENTITIES\\|NMTOKEN\\|NMTOKENS\\)") |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
227 ;;[58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
228 (defvar xml-notation-type-re (concat "\\(?:NOTATION" whitespace "(" whitespace "*" xml-name-re |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
229 "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
230 ;;[59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' [VC: Enumeration] [VC: No Duplicate Tokens] |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
231 (defvar xml-enumeration-re (concat "\\(?:(" whitespace "*" xml-nmtoken-re |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
232 "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*" |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
233 whitespace ")\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
234 ;;[57] EnumeratedType ::= NotationType | Enumeration |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
235 (defvar xml-enumerated-type-re (concat "\\(?:" xml-notation-type-re "\\|" xml-enumeration-re "\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
236 ;;[54] AttType ::= StringType | TokenizedType | EnumeratedType |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
237 ;;[55] StringType ::= 'CDATA' |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
238 (defvar xml-att-type-re (concat "\\(?:CDATA\\|" xml-tokenized-type-re "\\|" xml-notation-type-re"\\|" xml-enumerated-type-re "\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
239 ;;[60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
240 (defvar xml-default-decl-re (concat "\\(?:#REQUIRED\\|#IMPLIED\\|\\(?:#FIXED" whitespace "\\)*" xml-att-value-re "\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
241 ;;[53] AttDef ::= S Name S AttType S DefaultDecl |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
242 (defvar xml-att-def-re (concat "\\(?:" whitespace "*" xml-name-re |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
243 whitespace "*" xml-att-type-re |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
244 whitespace "*" xml-default-decl-re "\\)")) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
245 ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
246 ;; | "'" ([^%&'] | PEReference | Reference)* "'" |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
247 (defvar xml-entity-value-re (concat "\\(?:\"\\(?:[^%&\"]\\|" xml-pe-reference-re |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
248 "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
249 xml-pe-reference-re "\\|" xml-reference-re "\\)*'\\)"))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
250 ;;[75] ExternalID ::= 'SYSTEM' S SystemLiteral |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
251 ;; | 'PUBLIC' S PubidLiteral S SystemLiteral |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
252 ;;[76] NDataDecl ::= S 'NDATA' S |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
253 ;;[73] EntityDef ::= EntityValue| (ExternalID NDataDecl?) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
254 ;;[71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
255 ;;[74] PEDef ::= EntityValue | ExternalID |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
256 ;;[72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
257 ;;[70] EntityDecl ::= GEDecl | PEDecl |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
258 |
51102 | 259 ;; Note that this is setup so that we can do whitespace-skipping with |
260 ;; `(skip-syntax-forward " ")', inter alia. Previously this was slow | |
261 ;; compared with `re-search-forward', but that has been fixed. Also | |
262 ;; note that the standard syntax table contains other characters with | |
263 ;; whitespace syntax, like NBSP, but they are invalid in contexts in | |
264 ;; which we might skip whitespace -- specifically, they're not | |
265 ;; NameChars [XML 4]. | |
266 | |
267 (defvar xml-syntax-table | |
268 (let ((table (make-syntax-table))) | |
269 ;; Get space syntax correct per XML [3]. | |
270 (dotimes (c 31) | |
271 (modify-syntax-entry c "." table)) ; all are space in standard table | |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
272 (dolist (c '(?\t ?\n ?\r)) ; these should be space |
51102 | 273 (modify-syntax-entry c " " table)) |
274 ;; For skipping attributes. | |
275 (modify-syntax-entry ?\" "\"" table) | |
276 (modify-syntax-entry ?' "\"" table) | |
277 ;; Non-alnum name chars should be symbol constituents (`-' and `_' | |
278 ;; are OK by default). | |
279 (modify-syntax-entry ?. "_" table) | |
280 (modify-syntax-entry ?: "_" table) | |
281 ;; XML [89] | |
66646 | 282 (unless (featurep 'xemacs) |
283 (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005 | |
284 #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC | |
285 #x30FD #x30FE)) | |
286 (modify-syntax-entry (decode-char 'ucs c) "w" table))) | |
51102 | 287 ;; Fixme: rest of [4] |
288 table) | |
289 "Syntax table used by `xml-parse-region'.") | |
290 | |
291 ;; XML [5] | |
292 ;; Note that [:alpha:] matches all multibyte chars with word syntax. | |
51105
aac5eaf1454e
(xml-name-regexp): Wrap in `eval-and-compile'.
John Paul Wallington <jpw@pobox.com>
parents:
51102
diff
changeset
|
293 (eval-and-compile |
aac5eaf1454e
(xml-name-regexp): Wrap in `eval-and-compile'.
John Paul Wallington <jpw@pobox.com>
parents:
51102
diff
changeset
|
294 (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*")) |
51102 | 295 |
296 ;; Fixme: This needs re-writing to deal with the XML grammar properly, i.e. | |
297 ;; document ::= prolog element Misc* | |
298 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? | |
299 | |
300 ;;;###autoload | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
301 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns) |
30329 | 302 "Parse the region from BEG to END in BUFFER. |
303 If BUFFER is nil, it defaults to the current buffer. | |
304 Returns the XML list for the region, or raises an error if the region | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
305 is not well-formed XML. |
30329 | 306 If PARSE-DTD is non-nil, the DTD is parsed rather than skipped, |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
307 and returned as the first element of the list. |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
308 If PARSE-NS is non-nil, then QNAMES are expanded." |
68535
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
309 ;; Use fixed syntax table to ensure regexp char classes and syntax |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
310 ;; specs DTRT. |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
311 (with-syntax-table (standard-syntax-table) |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
312 (let ((case-fold-search nil) ; XML is case-sensitive. |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
313 xml result dtd) |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
314 (save-excursion |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
315 (if buffer |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
316 (set-buffer buffer)) |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
317 (save-restriction |
6fef36718f37
use provided patch
Mark A. Hershberger <mah@everybody.org>
parents:
68534
diff
changeset
|
318 (narrow-to-region beg end) |
51102 | 319 (goto-char (point-min)) |
320 (while (not (eobp)) | |
321 (if (search-forward "<" nil t) | |
322 (progn | |
323 (forward-char -1) | |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
324 (setq result (xml-parse-tag parse-dtd parse-ns)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
325 (if (and xml result (not xml-sub-parser)) |
51102 | 326 ;; translation of rule [1] of XML specifications |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
327 (error "XML: (Not Well-Formed) Only one root tag allowed") |
30329 | 328 (cond |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
329 ((null result)) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
330 ((and (listp (car result)) |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
331 parse-dtd) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
332 (setq dtd (car result)) |
51102 | 333 (if (cdr result) ; possible leading comment |
334 (add-to-list 'xml (cdr result)))) | |
30329 | 335 (t |
51102 | 336 (add-to-list 'xml result))))) |
337 (goto-char (point-max)))) | |
338 (if parse-dtd | |
339 (cons dtd (nreverse xml)) | |
340 (nreverse xml))))))) | |
30329 | 341 |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
342 (defun xml-maybe-do-ns (name default xml-ns) |
54937 | 343 "Perform any namespace expansion. |
344 NAME is the name to perform the expansion on. | |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
345 DEFAULT is the default namespace. XML-NS is a cons of namespace |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
346 names to uris. When namespace-aware parsing is off, then XML-NS |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
347 is nil. |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
348 |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
349 During namespace-aware parsing, any name without a namespace is |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
350 put into the namespace identified by DEFAULT. nil is used to |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
351 specify that the name shouldn't be given a namespace." |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
352 (if (consp xml-ns) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
353 (let* ((nsp (string-match ":" name)) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
354 (lname (if nsp (substring name (match-end 0)) name)) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
355 (prefix (if nsp (substring name 0 (match-beginning 0)) default)) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
356 (special (and (string-equal lname "xmlns") (not prefix))) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
357 ;; Setting default to nil will insure that there is not |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
358 ;; matching cons in xml-ns. In which case we |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
359 (ns (or (cdr (assoc (if special "xmlns" prefix) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
360 xml-ns)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
361 ""))) |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
362 (cons ns (if special "" lname))) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
363 (intern name))) |
30329 | 364 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
365 (defun xml-parse-fragment (&optional parse-dtd parse-ns) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
366 "Parse xml-like fragments." |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
367 (let ((xml-sub-parser t) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
368 children) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
369 (while (not (eobp)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
370 (let ((bit (xml-parse-tag |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
371 parse-dtd parse-ns))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
372 (if children |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
373 (setq children (append (list bit) children)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
374 (if (stringp bit) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
375 (setq children (list bit)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
376 (setq children bit))))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
377 (reverse children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
378 |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
379 (defun xml-parse-tag (&optional parse-dtd parse-ns) |
51102 | 380 "Parse the tag at point. |
30329 | 381 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and |
382 returned as the first element in the list. | |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
383 If PARSE-NS is non-nil, then QNAMES are expanded. |
30329 | 384 Returns one of: |
51102 | 385 - a list : the matching node |
386 - nil : the point is not looking at a tag. | |
387 - a pair : the first element is the DTD, the second is the node." | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
388 (let ((xml-validating-parser (or parse-dtd xml-validating-parser)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
389 (xml-ns (if (consp parse-ns) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
390 parse-ns |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
391 (if parse-ns |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
392 (list |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
393 ;; Default for empty prefix is no namespace |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
394 (cons "" "") |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
395 ;; "xml" namespace |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
396 (cons "xml" "http://www.w3.org/XML/1998/namespace") |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
397 ;; We need to seed the xmlns namespace |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
398 (cons "xmlns" "http://www.w3.org/2000/xmlns/")))))) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
399 (cond |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
400 ;; Processing instructions (like the <?xml version="1.0"?> tag at the |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
401 ;; beginning of a document). |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
402 ((looking-at "<\\?") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
403 (search-forward "?>") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
404 (skip-syntax-forward " ") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
405 (xml-parse-tag parse-dtd xml-ns)) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
406 ;; Character data (CDATA) sections, in which no tag should be interpreted |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
407 ((looking-at "<!\\[CDATA\\[") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
408 (let ((pos (match-end 0))) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
409 (unless (search-forward "]]>" nil t) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
410 (error "XML: (Not Well Formed) CDATA section does not end anywhere in the document")) |
58698
10ecc7ffdc5a
2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57367
diff
changeset
|
411 (concat |
10ecc7ffdc5a
2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57367
diff
changeset
|
412 (buffer-substring pos (match-beginning 0)) |
10ecc7ffdc5a
2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57367
diff
changeset
|
413 (xml-parse-string)))) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
414 ;; DTD for the document |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
415 ((looking-at "<!DOCTYPE") |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
416 (let ((dtd (xml-parse-dtd parse-ns))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
417 (skip-syntax-forward " ") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
418 (if xml-validating-parser |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
419 (cons dtd (xml-parse-tag nil xml-ns)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
420 (xml-parse-tag nil xml-ns)))) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
421 ;; skip comments |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
422 ((looking-at "<!--") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
423 (search-forward "-->") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
424 nil) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
425 ;; end tag |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
426 ((looking-at "</") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
427 '()) |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
428 ;; opening tag |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
429 ((looking-at "<\\([^/>[:space:]]+\\)") |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
430 (goto-char (match-end 1)) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
431 |
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
432 ;; Parse this node |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
433 (let* ((node-name (match-string 1)) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
434 ;; Parse the attribute list. |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
435 (attrs (xml-parse-attlist xml-ns)) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
436 children pos) |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
437 |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
438 ;; add the xmlns:* attrs to our cache |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
439 (when (consp xml-ns) |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
440 (dolist (attr attrs) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
441 (when (and (consp (car attr)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
442 (equal "http://www.w3.org/2000/xmlns/" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
443 (caar attr))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
444 (push (cons (cdar attr) (cdr attr)) |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
445 xml-ns)))) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
446 |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
447 (setq children (list attrs (xml-maybe-do-ns node-name "" xml-ns))) |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
448 |
51930
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
449 ;; is this an empty element ? |
608552c82ffc
(xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents:
51697
diff
changeset
|
450 (if (looking-at "/>") |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
451 (progn |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
452 (forward-char 2) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
453 (nreverse children)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
454 |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
455 ;; is this a valid start tag ? |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
456 (if (eq (char-after) ?>) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
457 (progn |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
458 (forward-char 1) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
459 ;; Now check that we have the right end-tag. Note that this |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
460 ;; one might contain spaces after the tag name |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
461 (let ((end (concat "</" node-name "\\s-*>"))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
462 (while (not (looking-at end)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
463 (cond |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
464 ((looking-at "</") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
465 (error "XML: (Not Well-Formed) Invalid end tag (expecting %s) at pos %d" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
466 node-name (point))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
467 ((= (char-after) ?<) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
468 (let ((tag (xml-parse-tag nil xml-ns))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
469 (when tag |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
470 (push tag children)))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
471 (t |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
472 (let ((expansion (xml-parse-string))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
473 (setq children |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
474 (if (stringp expansion) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
475 (if (stringp (car children)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
476 ;; The two strings were separated by a comment. |
66646 | 477 (setq children (append (list (concat (car children) expansion)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
478 (cdr children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
479 (setq children (append (list expansion) children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
480 (setq children (append expansion children)))))))) |
30329 | 481 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
482 (goto-char (match-end 0)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
483 (nreverse children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
484 ;; This was an invalid start tag (Expected ">", but didn't see it.) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
485 (error "XML: (Well-Formed) Couldn't parse tag: %s" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
486 (buffer-substring (- (point) 10) (+ (point) 1))))))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
487 (t ;; (Not one of PI, CDATA, Comment, End tag, or Start tag) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
488 (unless xml-sub-parser ; Usually, we error out. |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
489 (error "XML: (Well-Formed) Invalid character")) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
490 |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
491 ;; However, if we're parsing incrementally, then we need to deal |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
492 ;; with stray CDATA. |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
493 (xml-parse-string))))) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
494 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
495 (defun xml-parse-string () |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
496 "Parse the next whatever. Could be a string, or an element." |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
497 (let* ((pos (point)) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
498 (string (progn (if (search-forward "<" nil t) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
499 (forward-char -1) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
500 (goto-char (point-max))) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
501 (buffer-substring pos (point))))) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
502 ;; Clean up the string. As per XML specifications, the XML |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
503 ;; processor should always pass the whole string to the |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
504 ;; application. But \r's should be replaced: |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
505 ;; http://www.w3.org/TR/2000/REC-xml-20001006#sec-line-ends |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
506 (setq pos 0) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
507 (while (string-match "\r\n?" string pos) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
508 (setq string (replace-match "\n" t t string)) |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
509 (setq pos (1+ (match-beginning 0)))) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
510 |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
511 (xml-substitute-special string))) |
30329 | 512 |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
513 (defun xml-parse-attlist (&optional xml-ns) |
54937 | 514 "Return the attribute-list after point. |
515 Leave point at the first non-blank character after the tag." | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
516 (let ((attlist ()) |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
517 end-pos name) |
51102 | 518 (skip-syntax-forward " ") |
519 (while (looking-at (eval-when-compile | |
520 (concat "\\(" xml-name-regexp "\\)\\s-*=\\s-*"))) | |
54877
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
521 (setq end-pos (match-end 0)) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
522 (setq name (xml-maybe-do-ns (match-string 1) nil xml-ns)) |
1bf7ef48f54f
(xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
54243
diff
changeset
|
523 (goto-char end-pos) |
30329 | 524 |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
525 ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
526 |
30329 | 527 ;; Do we have a string between quotes (or double-quotes), |
528 ;; or a simple word ? | |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
529 (if (looking-at "\"\\([^\"]*\\)\"") |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
530 (setq end-pos (match-end 0)) |
50210
575aa6820adc
(xml-parse-attlist): typo in attribute parsing.
Juanma Barranquero <lekktu@gmail.com>
parents:
50144
diff
changeset
|
531 (if (looking-at "'\\([^']*\\)'") |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
532 (setq end-pos (match-end 0)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
533 (error "XML: (Not Well-Formed) Attribute values must be given between quotes"))) |
30329 | 534 |
535 ;; Each attribute must be unique within a given element | |
536 (if (assoc name attlist) | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
537 (error "XML: (Not Well-Formed) Each attribute must be unique within an element")) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
538 |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
539 ;; Multiple whitespace characters should be replaced with a single one |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
540 ;; in the attributes |
51102 | 541 (let ((string (match-string 1)) |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
542 (pos 0)) |
51102 | 543 (replace-regexp-in-string "\\s-\\{2,\\}" " " string) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
544 (let ((expansion (xml-substitute-special string))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
545 (unless (stringp expansion) |
68534
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
546 ; We say this is the constraint. It is acctually that |
bf640be8a692
fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents:
66646
diff
changeset
|
547 ; external entities nor "<" can be in an attribute value. |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
548 (error "XML: (Not Well-Formed) Entities in attributes cannot expand into elements")) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
549 (push (cons name expansion) attlist))) |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
550 |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
551 (goto-char end-pos) |
51102 | 552 (skip-syntax-forward " ")) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
553 (nreverse attlist))) |
30329 | 554 |
555 ;;******************************************************************* | |
556 ;;** | |
557 ;;** The DTD (document type declaration) | |
558 ;;** The following functions know how to skip or parse the DTD of | |
559 ;;** a document | |
560 ;;** | |
561 ;;******************************************************************* | |
562 | |
51102 | 563 ;; Fixme: This fails at least if the DTD contains conditional sections. |
564 | |
565 (defun xml-skip-dtd () | |
566 "Skip the DTD at point. | |
30329 | 567 This follows the rule [28] in the XML specifications." |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
568 (let ((xml-validating-parser nil)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
569 (xml-parse-dtd))) |
30329 | 570 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
571 (defun xml-parse-dtd (&optional parse-ns) |
51102 | 572 "Parse the DTD at point." |
573 (forward-char (eval-when-compile (length "<!DOCTYPE"))) | |
574 (skip-syntax-forward " ") | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
575 (if (and (looking-at ">") |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
576 xml-validating-parser) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
577 (error "XML: (Validity) Invalid DTD (expecting name of the document)")) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
578 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
579 ;; Get the name of the document |
51102 | 580 (looking-at xml-name-regexp) |
581 (let ((dtd (list (match-string 0) 'dtd)) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
582 type element end-pos) |
30329 | 583 (goto-char (match-end 0)) |
584 | |
51102 | 585 (skip-syntax-forward " ") |
586 ;; XML [75] | |
587 (cond ((looking-at "PUBLIC\\s-+") | |
588 (goto-char (match-end 0)) | |
589 (unless (or (re-search-forward | |
590 "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\"" | |
591 nil t) | |
592 (re-search-forward | |
593 "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'" | |
594 nil t)) | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
595 (error "XML: Missing Public ID")) |
51102 | 596 (let ((pubid (match-string 1))) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
597 (skip-syntax-forward " ") |
51102 | 598 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t) |
599 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t)) | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
600 (error "XML: Missing System ID")) |
51102 | 601 (push (list pubid (match-string 1) 'public) dtd))) |
602 ((looking-at "SYSTEM\\s-+") | |
603 (goto-char (match-end 0)) | |
604 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t) | |
605 (re-search-forward "\\=\"\\([^\"]*\\)\"" nil t)) | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
606 (error "XML: Missing System ID")) |
51102 | 607 (push (list (match-string 1) 'system) dtd))) |
608 (skip-syntax-forward " ") | |
609 (if (eq ?> (char-after)) | |
610 (forward-char) | |
611 (if (not (eq (char-after) ?\[)) | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
612 (error "XML: Bad DTD") |
51102 | 613 (forward-char) |
614 ;; Parse the rest of the DTD | |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
615 ;; Fixme: Deal with NOTATION, PIs. |
51102 | 616 (while (not (looking-at "\\s-*\\]")) |
617 (skip-syntax-forward " ") | |
618 (cond | |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
619 |
51102 | 620 ;; Translation of rule [45] of XML specifications |
621 ((looking-at | |
622 "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>") | |
623 | |
52975
6958c2be0aa9
Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents:
52401
diff
changeset
|
624 (setq element (match-string 1) |
51102 | 625 type (match-string-no-properties 2)) |
626 (setq end-pos (match-end 0)) | |
30329 | 627 |
51102 | 628 ;; Translation of rule [46] of XML specifications |
629 (cond | |
630 ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration | |
631 (setq type 'empty)) | |
632 ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents | |
633 (setq type 'any)) | |
634 ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47]) | |
635 (setq type (xml-parse-elem-type (match-string 1 type)))) | |
636 ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution | |
637 nil) | |
638 (t | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
639 (if xml-validating-parser |
57346
8e110e726b85
(xml-parse-dtd): Fix `error' call.
John Paul Wallington <jpw@pobox.com>
parents:
57344
diff
changeset
|
640 (error "XML: (Validity) Invalid element type in the DTD")))) |
58723
9104d032d2fa
Change existence of &; to not-well-formed.
Mark A. Hershberger <mah@everybody.org>
parents:
58721
diff
changeset
|
641 |
51102 | 642 ;; rule [45]: the element declaration must be unique |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
643 (if (and (assoc element dtd) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
644 xml-validating-parser) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
645 (error "XML: (Validity) Element declarations must be unique in a DTD (<%s>)" |
53011
45bd3ad34572
(xml-parse-dtd): Fix misplaced paren.
Andreas Schwab <schwab@suse.de>
parents:
52975
diff
changeset
|
646 element)) |
30329 | 647 |
51102 | 648 ;; Store the element in the DTD |
649 (push (list element type) dtd) | |
650 (goto-char end-pos)) | |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
651 |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
652 ;; Translation of rule [52] of XML specifications |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
653 ((looking-at (concat "<!ATTLIST[ \t\n\r]*\\(" xml-name-re |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
654 "\\)[ \t\n\r]*\\(" xml-att-def-re |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
655 "\\)*[ \t\n\r]*>")) |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
656 |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
657 ;; We don't do anything with ATTLIST currently |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
658 (goto-char (match-end 0))) |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
659 |
51102 | 660 ((looking-at "<!--") |
661 (search-forward "-->")) | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
662 ((looking-at (concat "<!ENTITY[ \t\n\r]*\\(" xml-name-re |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
663 "\\)[ \t\n\r]*\\(" xml-entity-value-re |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
664 "\\)[ \t\n\r]*>")) |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
665 (let ((name (match-string 1)) |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
666 (value (substring (match-string 2) 1 |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
667 (- (length (match-string 2)) 1)))) |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
668 (goto-char (match-end 0)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
669 (setq xml-entity-alist |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
670 (append xml-entity-alist |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
671 (list (cons name |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
672 (with-temp-buffer |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
673 (insert value) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
674 (goto-char (point-min)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
675 (xml-parse-fragment |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
676 xml-validating-parser |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
677 parse-ns)))))))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
678 ((or (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
679 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
680 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>")) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
681 (looking-at (concat "<!ENTITY[ \t\n\r]+\\(" xml-name-re |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
682 "\\)[ \t\n\r]+PUBLIC[ \t\n\r]+" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
683 "\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\"" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
684 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
685 "[ \t\n\r]+\\(\"[^\"]*\"\\|'[^']*'\\)" |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
686 "[ \t\n\r]*>"))) |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
687 (let ((name (match-string 1)) |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
688 (file (substring (match-string 2) 1 |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
689 (- (length (match-string 2)) 1)))) |
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
690 (goto-char (match-end 0)) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
691 (setq xml-entity-alist |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
692 (append xml-entity-alist |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
693 (list (cons name (with-temp-buffer |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
694 (insert-file-contents file) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
695 (goto-char (point-min)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
696 (xml-parse-fragment |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
697 xml-validating-parser |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
698 parse-ns)))))))) |
58724
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
699 ;; skip parameter entity declarations |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
700 ((or (looking-at (concat "<!ENTITY[ \t\n\r]+%[ \t\n\r]+\\(" xml-name-re |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
701 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
702 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>")) |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
703 (looking-at (concat "<!ENTITY[ \t\n\r]+" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
704 "%[ \t\n\r]+" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
705 "\\(" xml-name-re "\\)[ \t\n\r]+" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
706 "PUBLIC[ \t\n\r]+" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
707 "\\(\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\"" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
708 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'\\)[ \t\n\r]+" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
709 "\\(\"[^\"]+\"\\|'[^']+'\\)" |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
710 "[ \t\n\r]*>"))) |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
711 (goto-char (match-end 0))) |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
712 ;; skip parameter entities |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
713 ((looking-at (concat "%" xml-name-re ";")) |
99943ffd0774
Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents:
58723
diff
changeset
|
714 (goto-char (match-end 0))) |
51102 | 715 (t |
58721
ff6b2a793277
Ensure that validity messages only show when xml-validating-parser is set.
Mark A. Hershberger <mah@everybody.org>
parents:
58698
diff
changeset
|
716 (when xml-validating-parser |
58723
9104d032d2fa
Change existence of &; to not-well-formed.
Mark A. Hershberger <mah@everybody.org>
parents:
58721
diff
changeset
|
717 (error "XML: (Validity) Invalid DTD item")))))) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
718 (if (looking-at "\\s-*]>") |
63282
3609dc754369
eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents:
62752
diff
changeset
|
719 (goto-char (match-end 0)))) |
53011
45bd3ad34572
(xml-parse-dtd): Fix misplaced paren.
Andreas Schwab <schwab@suse.de>
parents:
52975
diff
changeset
|
720 (nreverse dtd))) |
30329 | 721 |
722 (defun xml-parse-elem-type (string) | |
51102 | 723 "Convert element type STRING into a Lisp structure." |
30329 | 724 |
725 (let (elem modifier) | |
726 (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string) | |
727 (progn | |
728 (setq elem (match-string 1 string) | |
729 modifier (match-string 2 string)) | |
730 (if (string-match "|" elem) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
731 (setq elem (cons 'choice |
30329 | 732 (mapcar 'xml-parse-elem-type |
733 (split-string elem "|")))) | |
734 (if (string-match "," elem) | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
735 (setq elem (cons 'seq |
30329 | 736 (mapcar 'xml-parse-elem-type |
51102 | 737 (split-string elem ","))))))) |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
738 (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string) |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
739 (setq elem (match-string 1 string) |
30329 | 740 modifier (match-string 2 string)))) |
741 | |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
742 (if (and (stringp elem) (string= elem "#PCDATA")) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
743 (setq elem 'pcdata)) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
744 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
745 (cond |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
746 ((string= modifier "+") |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
747 (list '+ elem)) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
748 ((string= modifier "*") |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
749 (list '* elem)) |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
750 ((string= modifier "?") |
49787
6269b5c10aec
(xml-parse-elem-type): Fix use of character constant.
Juanma Barranquero <lekktu@gmail.com>
parents:
49133
diff
changeset
|
751 (list '\? elem)) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
752 (t |
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
753 elem)))) |
30329 | 754 |
50144
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
755 ;;******************************************************************* |
ff1b1d15e1f2
(xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents:
50081
diff
changeset
|
756 ;;** |
30329 | 757 ;;** Substituting special XML sequences |
758 ;;** | |
759 ;;******************************************************************* | |
760 | |
761 (defun xml-substitute-special (string) | |
51102 | 762 "Return STRING, after subsituting entity references." |
763 ;; This originally made repeated passes through the string from the | |
764 ;; beginning, which isn't correct, since then either "&amp;" or | |
765 ;; "&amp;" won't DTRT. | |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
766 |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
767 (let ((point 0) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
768 children end-point) |
58698
10ecc7ffdc5a
2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57367
diff
changeset
|
769 (while (string-match "&\\([^;]*\\);" string point) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
770 (setq end-point (match-end 0)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
771 (let* ((this-part (match-string 1 string)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
772 (prev-part (substring string point (match-beginning 0))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
773 (entity (assoc this-part xml-entity-alist)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
774 (expansion |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
775 (cond ((string-match "#\\([0-9]+\\)" this-part) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
776 (let ((c (decode-char |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
777 'ucs |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
778 (string-to-number (match-string 1 this-part))))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
779 (if c (string c)))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
780 ((string-match "#x\\([[:xdigit:]]+\\)" this-part) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
781 (let ((c (decode-char |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
782 'ucs |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
783 (string-to-number (match-string 1 this-part) 16)))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
784 (if c (string c)))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
785 (entity |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
786 (cdr entity)) |
58698
10ecc7ffdc5a
2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57367
diff
changeset
|
787 ((eq (length this-part) 0) |
58723
9104d032d2fa
Change existence of &; to not-well-formed.
Mark A. Hershberger <mah@everybody.org>
parents:
58721
diff
changeset
|
788 (error "XML: (Not Well-Formed) No entity given")) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
789 (t |
62752
d0f8033496b1
2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
58939
diff
changeset
|
790 (if xml-validating-parser |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
791 (error "XML: (Validity) Undefined entity `%s'" |
62752
d0f8033496b1
2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
58939
diff
changeset
|
792 this-part) |
d0f8033496b1
2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
58939
diff
changeset
|
793 xml-undefined-entity))))) |
30329 | 794 |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
795 (cond ((null children) |
57367
8f9302e4a35f
2004-10-07 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57346
diff
changeset
|
796 ;; FIXME: If we have an entity that expands into XML, this won't work. |
8f9302e4a35f
2004-10-07 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57346
diff
changeset
|
797 (setq children |
8f9302e4a35f
2004-10-07 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
57346
diff
changeset
|
798 (concat prev-part expansion))) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
799 ((stringp children) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
800 (if (stringp expansion) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
801 (setq children (concat children prev-part expansion)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
802 (setq children (list expansion (concat prev-part children))))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
803 ((and (stringp expansion) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
804 (stringp (car children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
805 (setcar children (concat prev-part expansion (car children)))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
806 ((stringp expansion) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
807 (setq children (append (concat prev-part expansion) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
808 children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
809 ((stringp (car children)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
810 (setcar children (concat (car children) prev-part)) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
811 (setq children (append expansion children))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
812 (t |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
813 (setq children (list expansion |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
814 prev-part |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
815 children)))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
816 (setq point end-point))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
817 (cond ((stringp children) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
818 (concat children (substring string point))) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
819 ((stringp (car (last children))) |
57344
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents:
56375
diff
changeset
|
820 (concat (car (last children)) (substring string point))) |
56375
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
821 ((null children) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
822 string) |
2e4e974fa50b
2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents:
55316
diff
changeset
|
823 (t |
57344
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents:
56375
diff
changeset
|
824 (concat (mapconcat 'identity |
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents:
56375
diff
changeset
|
825 (nreverse children) |
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents:
56375
diff
changeset
|
826 "") |
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents:
56375
diff
changeset
|
827 (substring string point)))))) |
95ee82562c2a
fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents:
56375
diff
changeset
|
828 |
30329 | 829 ;;******************************************************************* |
830 ;;** | |
831 ;;** Printing a tree. | |
832 ;;** This function is intended mainly for debugging purposes. | |
833 ;;** | |
834 ;;******************************************************************* | |
835 | |
55253
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
836 (defun xml-debug-print (xml &optional indent-string) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
837 "Outputs the XML in the current buffer. |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
838 XML can be a tree or a list of nodes. |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
839 The first line is indented with the optional INDENT-STRING." |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
840 (setq indent-string (or indent-string "")) |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
841 (dolist (node xml) |
55253
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
842 (xml-debug-print-internal node indent-string))) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
843 |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
844 (defalias 'xml-print 'xml-debug-print) |
30329 | 845 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
846 (defun xml-debug-print-internal (xml indent-string) |
30329 | 847 "Outputs the XML tree in the current buffer. |
51102 | 848 The first line is indented with INDENT-STRING." |
30329 | 849 (let ((tree xml) |
850 attlist) | |
51102 | 851 (insert indent-string ?< (symbol-name (xml-node-name tree))) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
852 |
30329 | 853 ;; output the attribute list |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
854 (setq attlist (xml-node-attributes tree)) |
30329 | 855 (while attlist |
51102 | 856 (insert ?\ (symbol-name (caar attlist)) "=\"" (cdar attlist) ?\") |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
857 (setq attlist (cdr attlist))) |
49036
466922eb2b8d
(xml-substitute-special): Move "&" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents:
48869
diff
changeset
|
858 |
42031
54db4085a7df
Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
40030
diff
changeset
|
859 (setq tree (xml-node-children tree)) |
30329 | 860 |
55253
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
861 (if (null tree) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
862 (insert ?/ ?>) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
863 (insert ?>) |
30329 | 864 |
55253
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
865 ;; output the children |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
866 (dolist (node tree) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
867 (cond |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
868 ((listp node) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
869 (insert ?\n) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
870 (xml-debug-print-internal node (concat indent-string " "))) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
871 ((stringp node) (insert node)) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
872 (t |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
873 (error "Invalid XML tree")))) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
874 |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
875 (when (not (and (null (cdr tree)) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
876 (stringp (car tree)))) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
877 (insert ?\n indent-string)) |
d33879ad154a
(xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents:
54937
diff
changeset
|
878 (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>)))) |
30329 | 879 |
880 (provide 'xml) | |
881 | |
55316
7ac80356d84c
Arch-tags shouldn't be outline headers.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
55253
diff
changeset
|
882 ;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b |
30329 | 883 ;;; xml.el ends here |