annotate lisp/xml.el @ 102299:668bf00f127d

Fix last entry.
author Chong Yidong <cyd@stupidchicken.com>
date Fri, 27 Feb 2009 00:29:59 +0000
parents a9dc0e7c3f2b
children 3b1d7962549d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
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,
100908
a9dc0e7c3f2b Add 2009 to copyright years.
Glenn Morris <rgm@gnu.org>
parents: 98529
diff changeset
4 ;; 2005, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
5
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
6 ;; Author: Emmanuel Briot <briot@gnat.com>
51697
79747466c43b Record new maintainer.
Richard M. Stallman <rms@gnu.org>
parents: 51334
diff changeset
7 ;; Maintainer: Mark A. Hershberger <mah@everybody.org>
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
8 ;; Keywords: xml, data
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
9
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
10 ;; This file is part of GNU Emacs.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
11
94678
ee5932bf781d Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents: 91342
diff changeset
12 ;; GNU Emacs is free software: you can redistribute it and/or modify
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
13 ;; it under the terms of the GNU General Public License as published by
94678
ee5932bf781d Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents: 91342
diff changeset
14 ;; the Free Software Foundation, either version 3 of the License, or
ee5932bf781d Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents: 91342
diff changeset
15 ;; (at your option) any later version.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
16
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
17 ;; GNU Emacs is distributed in the hope that it will be useful,
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
20 ;; GNU General Public License for more details.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
21
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
22 ;; You should have received a copy of the GNU General Public License
94678
ee5932bf781d Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents: 91342
diff changeset
23 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
24
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
25 ;;; Commentary:
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
26
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
27 ;; This file contains a somewhat incomplete non-validating XML parser. It
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
28 ;; parses a file, and returns a list that can be used internally by
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
29 ;; any other Lisp libraries.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
30
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
31 ;;; FILE FORMAT
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
32
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
33 ;; The document type declaration may either be ignored or (optionally)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
34 ;; parsed, but currently the parsing will only accept element
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
35 ;; declarations. The XML file is assumed to be well-formed. In case
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
36 ;; of error, the parsing stops and the XML file is shown where the
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
37 ;; parsing stopped.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
38 ;;
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
39 ;; It also knows how to ignore comments and processing instructions.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
40 ;;
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
41 ;; 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
42 ;; <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
43 ;; <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
44 ;; <node3 attr5="name5" attr6="name6">value3</node3>
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
45 ;; </node1>
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
46 ;; Of course, the name of the nodes and attributes can be anything. There can
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
47 ;; be any number of attributes (or none), as well as any number of children
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
48 ;; below the nodes.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
49 ;;
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
50 ;; There can be only top level node, but with any number of children below.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
51
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
52 ;;; LIST FORMAT
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
53
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
54 ;; 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
55 ;; `xml-parse-tag' return a list with the following format:
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
56 ;;
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
57 ;; 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
58 ;; node ::= (qname attribute-list . child_node_list)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
59 ;; child_node_list ::= child_node child_node ...
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
60 ;; 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
61 ;; 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
62 ;; attribute_list ::= ((qname . "value") (qname . "value") ...)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
63 ;; | nil
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
64 ;; string ::= "..."
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
65 ;;
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
66 ;; Some macros are provided to ease the parsing of this list.
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
67 ;; Whitespace is preserved. Fixme: There should be a tree-walker that
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
68 ;; can remove it.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
69
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
70 ;; TODO:
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
71 ;; * 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
72 ;; * more complete DOCTYPE parsing
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
73 ;; * pi support
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
74
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
75 ;;; Code:
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
76
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
77 ;; Note that buffer-substring and match-string were formerly used in
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
78 ;; several places, because the -no-properties variants remove
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
79 ;; composition info. However, after some discussion on emacs-devel,
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
80 ;; the consensus was that the speed of the -no-properties variants was
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
81 ;; a worthwhile tradeoff especially since we're usually parsing files
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
82 ;; instead of hand-crafted XML.
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
83
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
84 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
85 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
86 ;;** Macros to parse the list
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
87 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
88 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
89
62752
d0f8033496b1 2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 58939
diff changeset
90 (defconst xml-undefined-entity "?"
d0f8033496b1 2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 58939
diff changeset
91 "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
92
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
93 (defvar xml-entity-alist
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
94 '(("lt" . "<")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
95 ("gt" . ">")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
96 ("apos" . "'")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
97 ("quot" . "\"")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
98 ("amp" . "&"))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
99 "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
100
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
101 (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
102 "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
103
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
104 (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
105 "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
106
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
107 (defsubst xml-node-name (node)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
108 "Return the tag associated with NODE.
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
109 Without namespace-aware parsing, the tag is a symbol.
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
110
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
111 With namespace-aware parsing, the tag is a cons of a string
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
112 representing the uri of the namespace with the local name of the
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
113 tag. For example,
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
114
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
115 <foo>
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
116
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
117 would be represented by
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
118
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
119 '(\"\" . \"foo\")."
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
120
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
121 (car node))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
122
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
123 (defsubst xml-node-attributes (node)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
124 "Return the list of attributes of NODE.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
125 The list can be nil."
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
126 (nth 1 node))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
127
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
128 (defsubst xml-node-children (node)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
129 "Return the list of children of NODE.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
130 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
131 (cddr node))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
132
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
133 (defun xml-get-children (node child-name)
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
134 "Return the children of NODE whose tag is CHILD-NAME.
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
135 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
136 (let ((match ()))
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
137 (dolist (child (xml-node-children node))
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
138 (if (and (listp child)
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
139 (equal (xml-node-name child) child-name))
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
140 (push child match)))
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
141 (nreverse match)))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
142
53375
e085973399ee (xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents: 53011
diff changeset
143 (defun xml-get-attribute-or-nil (node attribute)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
144 "Get from NODE the value of ATTRIBUTE.
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
145 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
146
e085973399ee (xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents: 53011
diff changeset
147 See also `xml-get-attribute'."
54243
586ffda6e9f9 (xml-get-attribute-or-nil): Simplify.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 53384
diff changeset
148 (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
149
e085973399ee (xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents: 53011
diff changeset
150 (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
151 "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
152 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
153
e085973399ee (xml-get-attribute-or-nil): New function, like
Eli Zaretskii <eliz@is.elta.co.il>
parents: 53011
diff changeset
154 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
155 (or (xml-get-attribute-or-nil node attribute) ""))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
156
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
157 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
158 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
159 ;;** Creating the list
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
160 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
161 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
162
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
163 ;;;###autoload
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
164 (defun xml-parse-file (file &optional parse-dtd parse-ns)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
165 "Parse the well-formed XML file FILE.
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
166 If FILE is already visited, use its buffer and don't kill it.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
167 Returns the top node with all its children.
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
168 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
169 If PARSE-NS is non-nil, then QNAMES are expanded."
72108
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
170 (if (get-file-buffer file)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
171 (with-current-buffer (get-file-buffer file)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
172 (save-excursion
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
173 (xml-parse-region (point-min)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
174 (point-max)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
175 (current-buffer)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
176 parse-dtd parse-ns)))
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
177 (with-temp-buffer
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
178 (insert-file-contents file)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
179 (xml-parse-region (point-min)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
180 (point-max)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
181 (current-buffer)
ee72501a64b4 (xml-parse-file): Clean up, and use with-temp-buffer.
Richard M. Stallman <rms@gnu.org>
parents: 68651
diff changeset
182 parse-dtd parse-ns))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
183
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
184
58939
e696b8c143a0 (xml-name-re, xml-entity-value-re): Add defvars.
Richard M. Stallman <rms@gnu.org>
parents: 58724
diff changeset
185 (defvar xml-name-re)
e696b8c143a0 (xml-name-re, xml-entity-value-re): Add defvars.
Richard M. Stallman <rms@gnu.org>
parents: 58724
diff changeset
186 (defvar xml-entity-value-re)
64794
950e3342b9d7 (xml-att-def-re): Add defvar.
Richard M. Stallman <rms@gnu.org>
parents: 64762
diff changeset
187 (defvar xml-att-def-re)
58724
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
188 (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
189 (name-chars (concat "-[:digit:]." start-chars))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
190 ;;[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
191 (whitespace "[ \t\n\r]"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
192 ;;[4] NameStartChar ::= ":" | [A-Z] | "_" | [a-z] | [#xC0-#xD6]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
193 ;; | [#xD8-#xF6] | [#xF8-#x2FF] | [#x370-#x37D] | [#x37F-#x1FFF]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
194 ;; | [#x200C-#x200D] | [#x2070-#x218F] | [#x2C00-#x2FEF] | [#x3001-#xD7FF]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
195 ;; | [#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
196 (defvar xml-name-start-char-re (concat "[" start-chars "]"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
197 ;;[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
198 (defvar xml-name-char-re (concat "[" name-chars "]"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
199 ;;[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
200 (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
201 ;;[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
202 (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
203 ;;[7] Nmtoken ::= (NameChar)+
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
204 (defvar xml-nmtoken-re (concat xml-name-char-re "+"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
205 ;;[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
206 (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
207 ;;[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
208 (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
209 ;;[68] EntityRef ::= '&' Name ';'
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
210 (defvar xml-entity-ref (concat "&" xml-name-re ";"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
211 ;;[69] PEReference ::= '%' Name ';'
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
212 (defvar xml-pe-reference-re (concat "%" xml-name-re ";"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
213 ;;[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
214 (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
215 ;;[10] AttValue ::= '"' ([^<&"] | Reference)* '"' | "'" ([^<&'] | Reference)* "'"
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
216 (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
217 "'\\(?:[^&']\\|" xml-reference-re "\\)*'\\)"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
218 ;;[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
219 ;; | 'IDREF' [VC: IDREF]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
220 ;; | 'IDREFS' [VC: IDREF]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
221 ;; | 'ENTITY' [VC: Entity Name]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
222 ;; | 'ENTITIES' [VC: Entity Name]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
223 ;; | 'NMTOKEN' [VC: Name Token]
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
224 ;; | 'NMTOKENS' [VC: Name Token]
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
225 (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
226 ;;[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
227 (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
228 "\\(?:" whitespace "*|" whitespace "*" xml-name-re "\\)*" whitespace "*)\\)"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
229 ;;[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
230 (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
231 "\\(?:" whitespace "*|" whitespace "*" xml-nmtoken-re "\\)*"
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
232 whitespace ")\\)"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
233 ;;[57] EnumeratedType ::= NotationType | Enumeration
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
234 (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
235 ;;[54] AttType ::= StringType | TokenizedType | EnumeratedType
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
236 ;;[55] StringType ::= 'CDATA'
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
237 (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
238 ;;[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
239 (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
240 ;;[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
241 (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
242 whitespace "*" xml-att-type-re
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
243 whitespace "*" xml-default-decl-re "\\)"))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
244 ;;[9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"'
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
245 ;; | "'" ([^%&'] | PEReference | Reference)* "'"
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
246 (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
247 "\\|" xml-reference-re "\\)*\"\\|'\\(?:[^%&']\\|"
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
248 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
249 ;;[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
250 ;; | '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
251 ;;[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
252 ;;[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
253 ;;[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
254 ;;[74] PEDef ::= EntityValue | ExternalID
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
255 ;;[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
256 ;;[70] EntityDecl ::= GEDecl | PEDecl
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
257
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
258 ;; Note that this is setup so that we can do whitespace-skipping with
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
259 ;; `(skip-syntax-forward " ")', inter alia. Previously this was slow
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
260 ;; compared with `re-search-forward', but that has been fixed. Also
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
261 ;; note that the standard syntax table contains other characters with
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
262 ;; whitespace syntax, like NBSP, but they are invalid in contexts in
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
263 ;; which we might skip whitespace -- specifically, they're not
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
264 ;; NameChars [XML 4].
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
265
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
266 (defvar xml-syntax-table
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
267 (let ((table (make-syntax-table)))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
268 ;; Get space syntax correct per XML [3].
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
269 (dotimes (c 31)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
270 (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
271 (dolist (c '(?\t ?\n ?\r)) ; these should be space
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
272 (modify-syntax-entry c " " table))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
273 ;; For skipping attributes.
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
274 (modify-syntax-entry ?\" "\"" table)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
275 (modify-syntax-entry ?' "\"" table)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
276 ;; Non-alnum name chars should be symbol constituents (`-' and `_'
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
277 ;; are OK by default).
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
278 (modify-syntax-entry ?. "_" table)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
279 (modify-syntax-entry ?: "_" table)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
280 ;; XML [89]
66646
173cd676da78 xml.el fixes
Mark A. Hershberger <mah@everybody.org>
parents: 64794
diff changeset
281 (unless (featurep 'xemacs)
173cd676da78 xml.el fixes
Mark A. Hershberger <mah@everybody.org>
parents: 64794
diff changeset
282 (dolist (c '(#x00B7 #x02D0 #x02D1 #x0387 #x0640 #x0E46 #x0EC6 #x3005
173cd676da78 xml.el fixes
Mark A. Hershberger <mah@everybody.org>
parents: 64794
diff changeset
283 #x3031 #x3032 #x3033 #x3034 #x3035 #x309D #x309E #x30FC
173cd676da78 xml.el fixes
Mark A. Hershberger <mah@everybody.org>
parents: 64794
diff changeset
284 #x30FD #x30FE))
173cd676da78 xml.el fixes
Mark A. Hershberger <mah@everybody.org>
parents: 64794
diff changeset
285 (modify-syntax-entry (decode-char 'ucs c) "w" table)))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
286 ;; Fixme: rest of [4]
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
287 table)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
288 "Syntax table used by `xml-parse-region'.")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
289
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
290 ;; XML [5]
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
291 ;; 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
292 (eval-and-compile
aac5eaf1454e (xml-name-regexp): Wrap in `eval-and-compile'.
John Paul Wallington <jpw@pobox.com>
parents: 51102
diff changeset
293 (defconst xml-name-regexp "[[:alpha:]_:][[:alnum:]._:-]*"))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
294
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
295 ;; Fixme: This needs re-writing to deal with the XML grammar properly, i.e.
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
296 ;; document ::= prolog element Misc*
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
297 ;; prolog ::= XMLDecl? Misc* (doctypedecl Misc*)?
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
298
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
299 ;;;###autoload
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
300 (defun xml-parse-region (beg end &optional buffer parse-dtd parse-ns)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
301 "Parse the region from BEG to END in BUFFER.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
302 If BUFFER is nil, it defaults to the current buffer.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
303 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
304 is not well-formed XML.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
305 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
306 and returned as the first element of the list.
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
307 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
308 ;; 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
309 ;; specs DTRT.
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
310 (with-syntax-table (standard-syntax-table)
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
311 (let ((case-fold-search nil) ; XML is case-sensitive.
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
312 xml result dtd)
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
313 (save-excursion
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
314 (if buffer
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
315 (set-buffer buffer))
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
316 (save-restriction
6fef36718f37 use provided patch
Mark A. Hershberger <mah@everybody.org>
parents: 68534
diff changeset
317 (narrow-to-region beg end)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
318 (goto-char (point-min))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
319 (while (not (eobp))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
320 (if (search-forward "<" nil t)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
321 (progn
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
322 (forward-char -1)
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
323 (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
324 (if (and xml result (not xml-sub-parser))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
325 ;; 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
326 (error "XML: (Not Well-Formed) Only one root tag allowed")
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
327 (cond
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
328 ((null result))
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
329 ((and (listp (car result))
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
330 parse-dtd)
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
331 (setq dtd (car result))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
332 (if (cdr result) ; possible leading comment
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
333 (add-to-list 'xml (cdr result))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
334 (t
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
335 (add-to-list 'xml result)))))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
336 (goto-char (point-max))))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
337 (if parse-dtd
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
338 (cons dtd (nreverse xml))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
339 (nreverse xml)))))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
340
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
341 (defun xml-maybe-do-ns (name default xml-ns)
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
342 "Perform any namespace expansion.
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
343 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
344 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
345 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
346 is nil.
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
347
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
348 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
349 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
350 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
351 (if (consp xml-ns)
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
352 (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
353 (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
354 (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
355 (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
356 ;; 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
357 ;; 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
358 (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
359 xml-ns))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
360 "")))
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
361 (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
362 (intern name)))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
363
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
364 (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
365 "Parse xml-like fragments."
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
366 (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
367 children)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
368 (while (not (eobp))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
369 (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
370 parse-dtd parse-ns)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
371 (if children
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
372 (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
373 (if (stringp bit)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
374 (setq children (list bit))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
375 (setq children bit)))))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
376 (reverse children)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
377
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
378 (defun xml-parse-tag (&optional parse-dtd parse-ns)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
379 "Parse the tag at point.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
380 If PARSE-DTD is non-nil, the DTD of the document, if any, is parsed and
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
381 returned as the first element in the list.
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
382 If PARSE-NS is non-nil, then QNAMES are expanded.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
383 Returns one of:
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
384 - a list : the matching node
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
385 - nil : the point is not looking at a tag.
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
386 - 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
387 (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
388 (xml-ns (if (consp parse-ns)
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
389 parse-ns
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
390 (if parse-ns
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
391 (list
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
392 ;; 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
393 (cons "" "")
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
394 ;; "xml" namespace
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
395 (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
396 ;; 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
397 (cons "xmlns" "http://www.w3.org/2000/xmlns/"))))))
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
398 (cond
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
399 ;; 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
400 ;; beginning of a document).
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
401 ((looking-at "<\\?")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
402 (search-forward "?>")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
403 (skip-syntax-forward " ")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
404 (xml-parse-tag parse-dtd xml-ns))
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
405 ;; 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
406 ((looking-at "<!\\[CDATA\\[")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
407 (let ((pos (match-end 0)))
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
408 (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
409 (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
410 (concat
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
411 (buffer-substring-no-properties pos (match-beginning 0))
58698
10ecc7ffdc5a 2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 57367
diff changeset
412 (xml-parse-string))))
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
413 ;; DTD for the document
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
414 ((looking-at "<!DOCTYPE")
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
415 (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
416 (skip-syntax-forward " ")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
417 (if xml-validating-parser
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
418 (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
419 (xml-parse-tag nil xml-ns))))
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
420 ;; skip comments
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
421 ((looking-at "<!--")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
422 (search-forward "-->")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
423 nil)
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
424 ;; end tag
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
425 ((looking-at "</")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
426 '())
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
427 ;; opening tag
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
428 ((looking-at "<\\([^/>[:space:]]+\\)")
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
429 (goto-char (match-end 1))
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
430
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
431 ;; Parse this node
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
432 (let* ((node-name (match-string-no-properties 1))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
433 ;; Parse the attribute list.
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
434 (attrs (xml-parse-attlist xml-ns))
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
435 children pos)
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
436
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
437 ;; add the xmlns:* attrs to our cache
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
438 (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
439 (dolist (attr attrs)
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
440 (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
441 (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
442 (caar attr)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
443 (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
444 xml-ns))))
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
445
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
446 (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
447
51930
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
448 ;; is this an empty element ?
608552c82ffc (xml-parse-tag): Namespace support.
Juanma Barranquero <lekktu@gmail.com>
parents: 51697
diff changeset
449 (if (looking-at "/>")
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
450 (progn
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
451 (forward-char 2)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
452 (nreverse children))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
453
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
454 ;; 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
455 (if (eq (char-after) ?>)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
456 (progn
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
457 (forward-char 1)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
458 ;; 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
459 ;; 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
460 (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
461 (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
462 (cond
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
463 ((looking-at "</")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
464 (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
465 node-name (point)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
466 ((= (char-after) ?<)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
467 (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
468 (when tag
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
469 (push tag children))))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
470 (t
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
471 (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
472 (setq children
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
473 (if (stringp expansion)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
474 (if (stringp (car children))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
475 ;; The two strings were separated by a comment.
66646
173cd676da78 xml.el fixes
Mark A. Hershberger <mah@everybody.org>
parents: 64794
diff changeset
476 (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
477 (cdr children)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
478 (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
479 (setq children (append expansion children))))))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
480
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
481 (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
482 (nreverse children)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
483 ;; 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
484 (error "XML: (Well-Formed) Couldn't parse tag: %s"
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
485 (buffer-substring-no-properties (- (point) 10) (+ (point) 1)))))))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
486 (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
487 (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
488 (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
489
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
490 ;; 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
491 ;; with stray CDATA.
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
492 (xml-parse-string)))))
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
493
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
494 (defun xml-parse-string ()
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
495 "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
496 (let* ((pos (point))
98529
2a05b2390d45 (xml-parse-string): Use skip-chars-forward.
Chong Yidong <cyd@stupidchicken.com>
parents: 94678
diff changeset
497 (string (progn (skip-chars-forward "^<")
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
498 (buffer-substring-no-properties pos (point)))))
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
499 ;; Clean up the string. As per XML specifications, the XML
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
500 ;; processor should always pass the whole string to the
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
501 ;; application. But \r's should be replaced:
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
502 ;; 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
503 (setq pos 0)
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
504 (while (string-match "\r\n?" string pos)
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
505 (setq string (replace-match "\n" t t string))
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
506 (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
507
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
508 (xml-substitute-special string)))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
509
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
510 (defun xml-parse-attlist (&optional xml-ns)
54937
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
511 "Return the attribute-list after point.
42293f0612cc Doc fixes.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54936
diff changeset
512 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
513 (let ((attlist ())
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
514 end-pos name)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
515 (skip-syntax-forward " ")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
516 (while (looking-at (eval-when-compile
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
517 (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
518 (setq end-pos (match-end 0))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
519 (setq name (xml-maybe-do-ns (match-string-no-properties 1) nil xml-ns))
54877
1bf7ef48f54f (xml-maybe-do-ns): New function to handle namespace
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 54243
diff changeset
520 (goto-char end-pos)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
521
50144
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
522 ;; See also: http://www.w3.org/TR/2000/REC-xml-20001006#AVNormalize
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
523
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
524 ;; Do we have a string between quotes (or double-quotes),
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
525 ;; 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
526 (if (looking-at "\"\\([^\"]*\\)\"")
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
527 (setq end-pos (match-end 0))
50210
575aa6820adc (xml-parse-attlist): typo in attribute parsing.
Juanma Barranquero <lekktu@gmail.com>
parents: 50144
diff changeset
528 (if (looking-at "'\\([^']*\\)'")
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
529 (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
530 (error "XML: (Not Well-Formed) Attribute values must be given between quotes")))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
531
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
532 ;; Each attribute must be unique within a given element
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
533 (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
534 (error "XML: (Not Well-Formed) Each attribute must be unique within an element"))
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
535
50144
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
536 ;; Multiple whitespace characters should be replaced with a single one
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
537 ;; in the attributes
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
538 (let ((string (match-string-no-properties 1))
50144
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
539 (pos 0))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
540 (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
541 (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
542 (unless (stringp expansion)
68534
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
543 ; We say this is the constraint. It is acctually that
bf640be8a692 fixing bug report
Mark A. Hershberger <mah@everybody.org>
parents: 66646
diff changeset
544 ; 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
545 (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
546 (push (cons name expansion) attlist)))
50144
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
547
52975
6958c2be0aa9 Allow comments following the top-level element.
Eli Zaretskii <eliz@gnu.org>
parents: 52401
diff changeset
548 (goto-char end-pos)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
549 (skip-syntax-forward " "))
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
550 (nreverse attlist)))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
551
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
552 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
553 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
554 ;;** The DTD (document type declaration)
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
555 ;;** The following functions know how to skip or parse the DTD of
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
556 ;;** a document
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
557 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
558 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
559
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
560 ;; Fixme: This fails at least if the DTD contains conditional sections.
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
561
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
562 (defun xml-skip-dtd ()
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
563 "Skip the DTD at point.
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
564 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
565 (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
566 (xml-parse-dtd)))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
567
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
568 (defun xml-parse-dtd (&optional parse-ns)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
569 "Parse the DTD at point."
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
570 (forward-char (eval-when-compile (length "<!DOCTYPE")))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
571 (skip-syntax-forward " ")
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
572 (if (and (looking-at ">")
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
573 xml-validating-parser)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
574 (error "XML: (Validity) Invalid DTD (expecting name of the document)"))
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
575
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
576 ;; Get the name of the document
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
577 (looking-at xml-name-regexp)
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
578 (let ((dtd (list (match-string-no-properties 0) 'dtd))
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
579 type element end-pos)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
580 (goto-char (match-end 0))
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
581
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
582 (skip-syntax-forward " ")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
583 ;; XML [75]
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
584 (cond ((looking-at "PUBLIC\\s-+")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
585 (goto-char (match-end 0))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
586 (unless (or (re-search-forward
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
587 "\\=\"\\([[:space:][:alnum:]-'()+,./:=?;!*#@$_%]*\\)\""
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
588 nil t)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
589 (re-search-forward
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
590 "\\='\\([[:space:][:alnum:]-()+,./:=?;!*#@$_%]*\\)'"
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
591 nil t))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
592 (error "XML: Missing Public ID"))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
593 (let ((pubid (match-string-no-properties 1)))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
594 (skip-syntax-forward " ")
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
595 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
596 (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
597 (error "XML: Missing System ID"))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
598 (push (list pubid (match-string-no-properties 1) 'public) dtd)))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
599 ((looking-at "SYSTEM\\s-+")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
600 (goto-char (match-end 0))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
601 (unless (or (re-search-forward "\\='\\([^']*\\)'" nil t)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
602 (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
603 (error "XML: Missing System ID"))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
604 (push (list (match-string-no-properties 1) 'system) dtd)))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
605 (skip-syntax-forward " ")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
606 (if (eq ?> (char-after))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
607 (forward-char)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
608 (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
609 (error "XML: Bad DTD")
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
610 (forward-char)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
611 ;; Parse the rest of the DTD
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
612 ;; Fixme: Deal with NOTATION, PIs.
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
613 (while (not (looking-at "\\s-*\\]"))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
614 (skip-syntax-forward " ")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
615 (cond
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
616
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
617 ;; Translation of rule [45] of XML specifications
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
618 ((looking-at
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
619 "<!ELEMENT\\s-+\\([[:alnum:].%;]+\\)\\s-+\\([^>]+\\)>")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
620
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
621 (setq element (match-string-no-properties 1)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
622 type (match-string-no-properties 2))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
623 (setq end-pos (match-end 0))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
624
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
625 ;; Translation of rule [46] of XML specifications
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
626 (cond
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
627 ((string-match "^EMPTY[ \t\n\r]*$" type) ;; empty declaration
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
628 (setq type 'empty))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
629 ((string-match "^ANY[ \t\n\r]*$" type) ;; any type of contents
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
630 (setq type 'any))
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
631 ((string-match "^(\\(.*\\))[ \t\n\r]*$" type) ;; children ([47])
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
632 (setq type (xml-parse-elem-type (match-string-no-properties 1 type))))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
633 ((string-match "^%[^;]+;[ \t\n\r]*$" type) ;; substitution
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
634 nil)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
635 (t
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
636 (if xml-validating-parser
57346
8e110e726b85 (xml-parse-dtd): Fix `error' call.
John Paul Wallington <jpw@pobox.com>
parents: 57344
diff changeset
637 (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
638
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
639 ;; 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
640 (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
641 xml-validating-parser)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
642 (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
643 element))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
644
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
645 ;; Store the element in the DTD
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
646 (push (list element type) dtd)
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
647 (goto-char end-pos))
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
648
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
649 ;; Translation of rule [52] of XML specifications
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
650 ((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
651 "\\)[ \t\n\r]*\\(" xml-att-def-re
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
652 "\\)*[ \t\n\r]*>"))
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
653
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
654 ;; 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
655 (goto-char (match-end 0)))
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
656
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
657 ((looking-at "<!--")
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
658 (search-forward "-->"))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
659 ((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
660 "\\)[ \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
661 "\\)[ \t\n\r]*>"))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
662 (let ((name (match-string-no-properties 1))
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
663 (value (substring (match-string-no-properties 2) 1
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
664 (- (length (match-string-no-properties 2)) 1))))
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
665 (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
666 (setq xml-entity-alist
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
667 (append xml-entity-alist
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
668 (list (cons name
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
669 (with-temp-buffer
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
670 (insert value)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
671 (goto-char (point-min))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
672 (xml-parse-fragment
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
673 xml-validating-parser
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
674 parse-ns))))))))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
675 ((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
676 "\\)[ \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
677 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
678 (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]+PUBLIC[ \t\n\r]+"
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
680 "\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
681 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'"
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
682 "[ \t\n\r]+\\(\"[^\"]*\"\\|'[^']*'\\)"
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
683 "[ \t\n\r]*>")))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
684 (let ((name (match-string-no-properties 1))
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
685 (file (substring (match-string-no-properties 2) 1
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
686 (- (length (match-string-no-properties 2)) 1))))
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
687 (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
688 (setq xml-entity-alist
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
689 (append xml-entity-alist
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
690 (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
691 (insert-file-contents file)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
692 (goto-char (point-min))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
693 (xml-parse-fragment
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
694 xml-validating-parser
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
695 parse-ns))))))))
58724
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
696 ;; skip parameter entity declarations
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
697 ((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
698 "\\)[ \t\n\r]+SYSTEM[ \t\n\r]+"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
699 "\\(\"[^\"]*\"\\|'[^']*'\\)[ \t\n\r]*>"))
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
700 (looking-at (concat "<!ENTITY[ \t\n\r]+"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
701 "%[ \t\n\r]+"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
702 "\\(" xml-name-re "\\)[ \t\n\r]+"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
703 "PUBLIC[ \t\n\r]+"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
704 "\\(\"[- \r\na-zA-Z0-9'()+,./:=?;!*#@$_%]*\""
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
705 "\\|'[- \r\na-zA-Z0-9()+,./:=?;!*#@$_%]*'\\)[ \t\n\r]+"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
706 "\\(\"[^\"]+\"\\|'[^']+'\\)"
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
707 "[ \t\n\r]*>")))
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
708 (goto-char (match-end 0)))
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
709 ;; skip parameter entities
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
710 ((looking-at (concat "%" xml-name-re ";"))
99943ffd0774 Skip parameter entity declarations.
Mark A. Hershberger <mah@everybody.org>
parents: 58723
diff changeset
711 (goto-char (match-end 0)))
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
712 (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
713 (when xml-validating-parser
58723
9104d032d2fa Change existence of &; to not-well-formed.
Mark A. Hershberger <mah@everybody.org>
parents: 58721
diff changeset
714 (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
715 (if (looking-at "\\s-*]>")
63282
3609dc754369 eliminate use of inefficient match-data
Mark A. Hershberger <mah@everybody.org>
parents: 62752
diff changeset
716 (goto-char (match-end 0))))
53011
45bd3ad34572 (xml-parse-dtd): Fix misplaced paren.
Andreas Schwab <schwab@suse.de>
parents: 52975
diff changeset
717 (nreverse dtd)))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
718
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
719 (defun xml-parse-elem-type (string)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
720 "Convert element type STRING into a Lisp structure."
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
721
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
722 (let (elem modifier)
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
723 (if (string-match "(\\([^)]+\\))\\([+*?]?\\)" string)
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
724 (progn
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
725 (setq elem (match-string-no-properties 1 string)
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
726 modifier (match-string-no-properties 2 string))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
727 (if (string-match "|" elem)
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
728 (setq elem (cons 'choice
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
729 (mapcar 'xml-parse-elem-type
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
730 (split-string elem "|"))))
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
731 (if (string-match "," elem)
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
732 (setq elem (cons 'seq
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
733 (mapcar 'xml-parse-elem-type
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
734 (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
735 (if (string-match "[ \t\n\r]*\\([^+*?]+\\)\\([+*?]?\\)" string)
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
736 (setq elem (match-string-no-properties 1 string)
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
737 modifier (match-string-no-properties 2 string))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
738
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
739 (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
740 (setq elem 'pcdata))
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
741
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
742 (cond
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
743 ((string= modifier "+")
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
744 (list '+ elem))
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
745 ((string= modifier "*")
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
746 (list '* elem))
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
747 ((string= modifier "?")
49787
6269b5c10aec (xml-parse-elem-type): Fix use of character constant.
Juanma Barranquero <lekktu@gmail.com>
parents: 49133
diff changeset
748 (list '\? elem))
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
749 (t
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
750 elem))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
751
50144
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
752 ;;*******************************************************************
ff1b1d15e1f2 (xml-ucs-to-string): New function to convert Unicode codepoints to strings.
Juanma Barranquero <lekktu@gmail.com>
parents: 50081
diff changeset
753 ;;**
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
754 ;;** Substituting special XML sequences
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
755 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
756 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
757
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
758 (defun xml-substitute-special (string)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
759 "Return STRING, after subsituting entity references."
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
760 ;; This originally made repeated passes through the string from the
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
761 ;; beginning, which isn't correct, since then either "&amp;amp;" or
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
762 ;; "&#38;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
763
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
764 (let ((point 0)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
765 children end-point)
58698
10ecc7ffdc5a 2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 57367
diff changeset
766 (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
767 (setq end-point (match-end 0))
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
768 (let* ((this-part (match-string-no-properties 1 string))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
769 (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
770 (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
771 (expansion
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
772 (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
773 (let ((c (decode-char
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
774 'ucs
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
775 (string-to-number (match-string-no-properties 1 this-part)))))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
776 (if c (string c))))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
777 ((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
778 (let ((c (decode-char
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
779 'ucs
76507
d1760553962c * xml.el (xml-parse-tag, xml-parse-string, xml-parse-attlist)
Chong Yidong <cyd@stupidchicken.com>
parents: 75347
diff changeset
780 (string-to-number (match-string-no-properties 1 this-part) 16))))
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
781 (if c (string c))))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
782 (entity
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
783 (cdr entity))
58698
10ecc7ffdc5a 2004-11-30 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 57367
diff changeset
784 ((eq (length this-part) 0)
58723
9104d032d2fa Change existence of &; to not-well-formed.
Mark A. Hershberger <mah@everybody.org>
parents: 58721
diff changeset
785 (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
786 (t
62752
d0f8033496b1 2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 58939
diff changeset
787 (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
788 (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
789 this-part)
d0f8033496b1 2005-05-26 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 58939
diff changeset
790 xml-undefined-entity)))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
791
56375
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
792 (cond ((null children)
57367
8f9302e4a35f 2004-10-07 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 57346
diff changeset
793 ;; 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
794 (setq children
8f9302e4a35f 2004-10-07 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 57346
diff changeset
795 (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
796 ((stringp children)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
797 (if (stringp expansion)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
798 (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
799 (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
800 ((and (stringp expansion)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
801 (stringp (car children)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
802 (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
803 ((stringp expansion)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
804 (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
805 children)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
806 ((stringp (car children))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
807 (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
808 (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
809 (t
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
810 (setq children (list expansion
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
811 prev-part
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
812 children))))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
813 (setq point end-point)))
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
814 (cond ((stringp children)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
815 (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
816 ((stringp (car (last children)))
57344
95ee82562c2a fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents: 56375
diff changeset
817 (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
818 ((null children)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
819 string)
2e4e974fa50b 2004-07-09 Mark A. Hershberger <mah@everybody.org>
Mark A. Hershberger <mah@everybody.org>
parents: 55316
diff changeset
820 (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
821 (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
822 (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
823 "")
95ee82562c2a fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents: 56375
diff changeset
824 (substring string point))))))
95ee82562c2a fix to for xml-substitute-special to produce a single string instead
Mark A. Hershberger <mah@everybody.org>
parents: 56375
diff changeset
825
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
826 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
827 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
828 ;;** Printing a tree.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
829 ;;** This function is intended mainly for debugging purposes.
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
830 ;;**
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
831 ;;*******************************************************************
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
832
55253
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
833 (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
834 "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
835 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
836 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
837 (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
838 (dolist (node xml)
55253
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
839 (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
840
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
841 (defalias 'xml-print 'xml-debug-print)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
842
87317
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
843 (defun xml-escape-string (string)
87318
3b14ea2b1d8a Forgot docstring on xml-escape-string
Mark A. Hershberger <mah@everybody.org>
parents: 87317
diff changeset
844 "Return the string with entity substitutions made from
3b14ea2b1d8a Forgot docstring on xml-escape-string
Mark A. Hershberger <mah@everybody.org>
parents: 87317
diff changeset
845 xml-entity-alist."
87317
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
846 (mapconcat (lambda (byte)
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
847 (let ((char (char-to-string byte)))
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
848 (if (rassoc char xml-entity-alist)
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
849 (concat "&" (car (rassoc char xml-entity-alist)) ";")
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
850 char)))
91342
3ec2fe8547cf * xml.el (xml-escape-string): Don't do any encoding changes on the string.
Mark A. Hershberger <mah@everybody.org>
parents: 91327
diff changeset
851 ;; This differs from the non-unicode branch. Just
3ec2fe8547cf * xml.el (xml-escape-string): Don't do any encoding changes on the string.
Mark A. Hershberger <mah@everybody.org>
parents: 91327
diff changeset
852 ;; grabbing the string works here.
3ec2fe8547cf * xml.el (xml-escape-string): Don't do any encoding changes on the string.
Mark A. Hershberger <mah@everybody.org>
parents: 91327
diff changeset
853 string ""))
87317
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
854
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
855 (defun xml-debug-print-internal (xml indent-string)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
856 "Outputs the XML tree in the current buffer.
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
857 The first line is indented with INDENT-STRING."
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
858 (let ((tree xml)
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
859 attlist)
51102
d604b76d3bbd Doc fixes.
Dave Love <fx@gnu.org>
parents: 50210
diff changeset
860 (insert indent-string ?< (symbol-name (xml-node-name tree)))
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
861
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
862 ;; output the attribute list
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
863 (setq attlist (xml-node-attributes tree))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
864 (while attlist
87317
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
865 (insert ?\ (symbol-name (caar attlist)) "=\""
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
866 (xml-escape-string (cdar attlist)) ?\")
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
867 (setq attlist (cdr attlist)))
49036
466922eb2b8d (xml-substitute-special): Move "&amp;" -> "&" last.
Thien-Thi Nguyen <ttn@gnuvola.org>
parents: 48869
diff changeset
868
42031
54db4085a7df Use setq rather than (set 'foo bar).
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 40030
diff changeset
869 (setq tree (xml-node-children tree))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
870
55253
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
871 (if (null tree)
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
872 (insert ?/ ?>)
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
873 (insert ?>)
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
874
55253
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
875 ;; output the children
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
876 (dolist (node tree)
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
877 (cond
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
878 ((listp node)
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
879 (insert ?\n)
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
880 (xml-debug-print-internal node (concat indent-string " ")))
87317
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
881 ((stringp node)
e8a485c87523 fix up xml-debug-print
Mark A. Hershberger <mah@everybody.org>
parents: 78236
diff changeset
882 (insert (xml-escape-string node)))
55253
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
883 (t
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
884 (error "Invalid XML tree"))))
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
885
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
886 (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
887 (stringp (car tree))))
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
888 (insert ?\n indent-string))
d33879ad154a (xml-debug-print-internal): Don't add newline and
Alex Schroeder <alex@gnu.org>
parents: 54937
diff changeset
889 (insert ?< ?/ (symbol-name (xml-node-name xml)) ?>))))
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
890
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
891 (provide 'xml)
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
892
55316
7ac80356d84c Arch-tags shouldn't be outline headers.
Stefan Monnier <monnier@iro.umontreal.ca>
parents: 55253
diff changeset
893 ;; arch-tag: 5864b283-5a68-4b59-a20d-36a72b353b9b
30329
1ea701655cf8 *** empty log message ***
Gerd Moellmann <gerd@gnu.org>
parents:
diff changeset
894 ;;; xml.el ends here