17 |
ilm |
1 |
/*
|
|
|
2 |
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
|
|
|
3 |
*
|
|
|
4 |
* Copyright 2011 OpenConcerto, by ILM Informatique. All rights reserved.
|
|
|
5 |
*
|
|
|
6 |
* The contents of this file are subject to the terms of the GNU General Public License Version 3
|
|
|
7 |
* only ("GPL"). You may not use this file except in compliance with the License. You can obtain a
|
|
|
8 |
* copy of the License at http://www.gnu.org/licenses/gpl-3.0.html See the License for the specific
|
|
|
9 |
* language governing permissions and limitations under the License.
|
|
|
10 |
*
|
|
|
11 |
* When distributing the software, include this License Header Notice in each file.
|
|
|
12 |
*/
|
|
|
13 |
|
|
|
14 |
/*
|
|
|
15 |
* Créé le 28 oct. 2004
|
|
|
16 |
*/
|
|
|
17 |
package org.openconcerto.openoffice;
|
|
|
18 |
|
|
|
19 |
import org.openconcerto.xml.JDOMUtils;
|
|
|
20 |
import org.openconcerto.xml.Validator;
|
|
|
21 |
|
|
|
22 |
import java.util.ArrayList;
|
|
|
23 |
import java.util.HashMap;
|
|
|
24 |
import java.util.Iterator;
|
|
|
25 |
import java.util.List;
|
|
|
26 |
import java.util.Map;
|
|
|
27 |
import java.util.Map.Entry;
|
|
|
28 |
import java.util.regex.Matcher;
|
|
|
29 |
import java.util.regex.Pattern;
|
|
|
30 |
|
|
|
31 |
import javax.xml.XMLConstants;
|
|
|
32 |
import javax.xml.validation.Schema;
|
|
|
33 |
import javax.xml.validation.SchemaFactory;
|
|
|
34 |
|
|
|
35 |
import org.apache.commons.collections.Transformer;
|
|
|
36 |
import org.apache.commons.collections.map.LazyMap;
|
|
|
37 |
import org.jdom.Content;
|
|
|
38 |
import org.jdom.Document;
|
|
|
39 |
import org.jdom.Element;
|
|
|
40 |
import org.jdom.JDOMException;
|
|
|
41 |
import org.jdom.Namespace;
|
|
|
42 |
import org.jdom.Parent;
|
|
|
43 |
import org.jdom.Text;
|
|
|
44 |
import org.jdom.xpath.XPath;
|
|
|
45 |
import org.xml.sax.SAXException;
|
|
|
46 |
|
|
|
47 |
/**
|
|
|
48 |
* Various bits of OpenDocument XML.
|
|
|
49 |
*
|
|
|
50 |
* @author Sylvain CUAZ
|
|
|
51 |
* @see #get(XMLVersion)
|
|
|
52 |
*/
|
|
|
53 |
public class OOXML {
|
|
|
54 |
|
|
|
55 |
private static final Map instances = LazyMap.decorate(new HashMap(), new Transformer() {
|
|
|
56 |
public Object transform(Object input) {
|
|
|
57 |
return new OOXML((XMLVersion) input);
|
|
|
58 |
}
|
|
|
59 |
});
|
|
|
60 |
|
|
|
61 |
/**
|
|
|
62 |
* Returns the instance that match the requested version.
|
|
|
63 |
*
|
|
|
64 |
* @param version the version.
|
|
|
65 |
* @return the corresponding instance.
|
|
|
66 |
*/
|
|
|
67 |
public static OOXML get(XMLVersion version) {
|
|
|
68 |
return (OOXML) instances.get(version);
|
|
|
69 |
}
|
|
|
70 |
|
|
|
71 |
static public final String getLineBreakS() {
|
|
|
72 |
return "<text:line-break/>";
|
|
|
73 |
}
|
|
|
74 |
|
|
|
75 |
static private final String rt2oo(String content, String tagName, String styleName) {
|
|
|
76 |
return content.replaceAll("\\[" + tagName + "\\]", "<text:span text:style-name=\"" + styleName + "\">").replaceAll("\\[/" + tagName + "\\]", "</text:span>");
|
|
|
77 |
}
|
|
|
78 |
|
|
|
79 |
/**
|
|
|
80 |
* Encode spaces for OpenOffice 1, and escape characters for XML.
|
|
|
81 |
*
|
|
|
82 |
* @param s a string to encode, eg "hi\n 3<4".
|
|
|
83 |
* @return the string encoded in XML, eg "hi<text:line-break/><text:s text:c="2"/>3<4".
|
|
|
84 |
* @deprecated see {@link #encodeWS(String)}
|
|
|
85 |
*/
|
|
|
86 |
static public final String encodeOOWS(final String s) {
|
|
|
87 |
String tmp = JDOMUtils.OUTPUTTER.escapeElementEntities(s).replaceAll("\n", getLineBreakS()).replaceAll("\t", OOXML.get(XMLVersion.OOo).getTabS());
|
|
|
88 |
|
|
|
89 |
String res = "";
|
|
|
90 |
// les séries de plus d'un espace consécutifs
|
|
|
91 |
final Pattern p = Pattern.compile(" +");
|
|
|
92 |
final Matcher m = p.matcher(tmp);
|
|
|
93 |
int lastEnd = 0;
|
|
|
94 |
while (m.find()) {
|
|
|
95 |
// c == le nombre d'espaces
|
|
|
96 |
res += tmp.substring(lastEnd, m.start()) + "<text:s text:c=\"" + (m.group().length()) + "\"/>";
|
|
|
97 |
lastEnd = m.end();
|
|
|
98 |
}
|
|
|
99 |
res += tmp.substring(lastEnd);
|
|
|
100 |
|
|
|
101 |
return res;
|
|
|
102 |
}
|
|
|
103 |
|
|
|
104 |
// *** instances
|
|
|
105 |
|
|
|
106 |
private final XMLVersion version;
|
|
|
107 |
private Schema schema = null;
|
|
|
108 |
|
|
|
109 |
private OOXML(XMLVersion version) {
|
|
|
110 |
this.version = version;
|
|
|
111 |
}
|
|
|
112 |
|
|
|
113 |
public final XMLVersion getVersion() {
|
|
|
114 |
return this.version;
|
|
|
115 |
}
|
|
|
116 |
|
|
|
117 |
private Schema getSchema() throws SAXException {
|
|
|
118 |
if (this.schema == null) {
|
|
|
119 |
this.schema = SchemaFactory.newInstance(XMLConstants.RELAXNG_NS_URI).newSchema(getClass().getResource("oofficeDTDs/OpenDocument-strict-schema-v1.1.rng"));
|
|
|
120 |
}
|
|
|
121 |
assert this.schema != null;
|
|
|
122 |
return this.schema;
|
|
|
123 |
}
|
|
|
124 |
|
|
|
125 |
/**
|
|
|
126 |
* Verify that the passed document is a valid OpenOffice.org 1 or ODF document.
|
|
|
127 |
*
|
|
|
128 |
* @param doc the xml to test.
|
|
|
129 |
* @return a validator on <code>doc</code>.
|
|
|
130 |
*/
|
|
|
131 |
public Validator getValidator(Document doc) {
|
|
|
132 |
if (this.getVersion() == XMLVersion.OD) {
|
|
|
133 |
final Schema schema;
|
|
|
134 |
try {
|
|
|
135 |
schema = this.getSchema();
|
|
|
136 |
} catch (SAXException e) {
|
|
|
137 |
throw new IllegalStateException("relaxNG schemas pb", e);
|
|
|
138 |
}
|
|
|
139 |
return new Validator.JAXPValidator(doc, schema);
|
|
|
140 |
} else {
|
|
|
141 |
// DTDs are stubborn, xmlns have to be exactly where they want
|
|
|
142 |
// in this case the root element
|
|
|
143 |
for (final Namespace n : getVersion().getALL())
|
|
|
144 |
doc.getRootElement().addNamespaceDeclaration(n);
|
|
|
145 |
return new Validator.DTDValidator(doc, OOUtils.getBuilderLoadDTD());
|
|
|
146 |
}
|
|
|
147 |
}
|
|
|
148 |
|
|
|
149 |
/**
|
|
|
150 |
* Return the names of font face declarations.
|
|
|
151 |
*
|
|
|
152 |
* @return at index 0 the name of the container element, at 1 the qualified name of its
|
|
|
153 |
* children.
|
|
|
154 |
*/
|
|
|
155 |
public String[] getFontDecls() {
|
|
|
156 |
if (this.getVersion() == XMLVersion.OOo)
|
|
|
157 |
return new String[] { "font-decls", "style:font-decl" };
|
|
|
158 |
else
|
|
|
159 |
return new String[] { "font-face-decls", "style:font-face" };
|
|
|
160 |
}
|
|
|
161 |
|
|
|
162 |
public final Element getLineBreak() {
|
|
|
163 |
return new Element("line-break", getVersion().getTEXT());
|
|
|
164 |
}
|
|
|
165 |
|
|
|
166 |
public final Element getTab() {
|
|
|
167 |
return new Element(this.getVersion().equals(XMLVersion.OD) ? "tab" : "tab-stop", getVersion().getTEXT());
|
|
|
168 |
}
|
|
|
169 |
|
|
|
170 |
/**
|
|
|
171 |
* How to encode a tab.
|
|
|
172 |
*
|
|
|
173 |
* @return the xml string to encode a tab.
|
|
|
174 |
* @deprecated use {@link #getTab()}
|
|
|
175 |
*/
|
|
|
176 |
public final String getTabS() {
|
|
|
177 |
return this.getVersion().equals(XMLVersion.OD) ? "<text:tab/>" : "<text:tab-stop/>";
|
|
|
178 |
}
|
|
|
179 |
|
|
|
180 |
protected final List encodeRT_L(String content, Map styles) {
|
|
|
181 |
String res = JDOMUtils.OUTPUTTER.escapeElementEntities(content);
|
|
|
182 |
final Iterator iter = styles.entrySet().iterator();
|
|
|
183 |
while (iter.hasNext()) {
|
|
|
184 |
final Entry e = (Entry) iter.next();
|
|
|
185 |
res = rt2oo(res, (String) e.getKey(), (String) e.getValue());
|
|
|
186 |
}
|
|
|
187 |
try {
|
|
|
188 |
return JDOMUtils.parseString(res, getVersion().getALL());
|
|
|
189 |
} catch (JDOMException e) {
|
|
|
190 |
// should not happpen as we did escapeElementEntities which gives valid xml and then
|
|
|
191 |
// rt2oo which introduce only static xml
|
|
|
192 |
throw new IllegalStateException("could not parse " + res, e);
|
|
|
193 |
}
|
|
|
194 |
}
|
|
|
195 |
|
|
|
196 |
/**
|
|
|
197 |
* Convert rich text (with [] tags) into XML.
|
|
|
198 |
*
|
|
|
199 |
* @param content the string to convert, eg "texte [b]gras[/b]".
|
|
|
200 |
* @param styles the mapping from tagname (eg "b") to the name of the character style (eg
|
|
|
201 |
* "Gras").
|
|
|
202 |
* @return the corresponding element.
|
|
|
203 |
*/
|
|
|
204 |
public final Element encodeRT(String content, Map styles) {
|
|
|
205 |
return new Element("span", getVersion().getTEXT()).addContent(encodeRT_L(content, styles));
|
|
|
206 |
}
|
|
|
207 |
|
|
|
208 |
// create the necessary <text:s c="n"/>
|
|
|
209 |
private Element createSpaces(String spacesS) {
|
|
|
210 |
return new Element("s", getVersion().getTEXT()).setAttribute("c", spacesS.length() + "", getVersion().getTEXT());
|
|
|
211 |
}
|
|
|
212 |
|
|
|
213 |
/**
|
|
|
214 |
* Encode a String to OO XML. Handles substition of whitespaces to their OO equivalent.
|
|
|
215 |
*
|
|
|
216 |
* @param s a plain ole String, eg "term\tdefinition".
|
|
|
217 |
* @return an Element suitable to be inserted in an OO XML document, eg
|
|
|
218 |
*
|
|
|
219 |
* <pre>
|
|
|
220 |
* <text:span>term<text:tab-stop/>definition</text:span>
|
|
|
221 |
* </pre>
|
|
|
222 |
*
|
|
|
223 |
* .
|
|
|
224 |
*/
|
|
|
225 |
public final Element encodeWS(final String s) {
|
|
|
226 |
return new Element("span", getVersion().getTEXT()).setContent(encodeWSasList(s));
|
|
|
227 |
}
|
|
|
228 |
|
|
|
229 |
private final List<Content> encodeWSasList(final String s) {
|
|
|
230 |
final List<Content> res = new ArrayList<Content>();
|
|
|
231 |
final Matcher m = Pattern.compile("\n|\t| {2,}").matcher(s);
|
|
|
232 |
int last = 0;
|
|
|
233 |
while (m.find()) {
|
|
|
234 |
res.add(new Text(s.substring(last, m.start())));
|
|
|
235 |
switch (m.group().charAt(0)) {
|
|
|
236 |
case '\n':
|
|
|
237 |
res.add(getLineBreak());
|
|
|
238 |
break;
|
|
|
239 |
case '\t':
|
|
|
240 |
res.add(getTab());
|
|
|
241 |
break;
|
|
|
242 |
case ' ':
|
|
|
243 |
res.add(createSpaces(m.group()));
|
|
|
244 |
break;
|
|
|
245 |
|
|
|
246 |
default:
|
|
|
247 |
throw new IllegalStateException("unknown item: " + m.group());
|
|
|
248 |
}
|
|
|
249 |
last = m.end();
|
|
|
250 |
}
|
|
|
251 |
res.add(new Text(s.substring(last)));
|
|
|
252 |
return res;
|
|
|
253 |
}
|
|
|
254 |
|
|
|
255 |
@SuppressWarnings("unchecked")
|
|
|
256 |
public final void encodeWS(final Text t) {
|
|
|
257 |
final Parent parent = t.getParent();
|
|
|
258 |
final int ind = parent.indexOf(t);
|
|
|
259 |
t.detach();
|
|
|
260 |
parent.getContent().addAll(ind, encodeWSasList(t.getText()));
|
|
|
261 |
}
|
|
|
262 |
|
|
|
263 |
@SuppressWarnings("unchecked")
|
|
|
264 |
public final Element encodeWS(final Element elem) {
|
|
|
265 |
final XPath path;
|
|
|
266 |
try {
|
|
|
267 |
path = OOUtils.getXPath(".//text()", XMLVersion.getVersion(elem));
|
|
|
268 |
} catch (JDOMException e) {
|
|
|
269 |
// static path, hence always valid
|
|
|
270 |
throw new IllegalStateException("cannot create XPath", e);
|
|
|
271 |
}
|
|
|
272 |
try {
|
|
|
273 |
final Iterator iter = new ArrayList(path.selectNodes(elem)).iterator();
|
|
|
274 |
while (iter.hasNext()) {
|
|
|
275 |
final Text t = (Text) iter.next();
|
|
|
276 |
encodeWS(t);
|
|
|
277 |
}
|
|
|
278 |
} catch (JDOMException e) {
|
|
|
279 |
throw new IllegalArgumentException("cannot find text nodes of " + elem, e);
|
|
|
280 |
}
|
|
|
281 |
return elem;
|
|
|
282 |
}
|
|
|
283 |
|
|
|
284 |
}
|