|
1
|
|
package net.sf.flock.parser;
|
|
2
|
|
|
|
3
|
|
import java.util.Date;
|
|
4
|
|
import java.util.Iterator;
|
|
5
|
|
import java.util.List;
|
|
6
|
|
|
|
7
|
|
import net.sf.flock.FeedFactoryI;
|
|
8
|
|
import net.sf.flock.FeedI;
|
|
9
|
|
import net.sf.flock.FlockResourceException;
|
|
10
|
|
import net.sf.flock.SubscriptionInfoI;
|
|
11
|
|
|
|
12
|
|
import org.apache.log4j.LogManager;
|
|
13
|
|
import org.apache.log4j.Logger;
|
|
14
|
|
import org.jdom.Attribute;
|
|
15
|
|
import org.jdom.Document;
|
|
16
|
|
import org.jdom.Element;
|
|
17
|
|
import org.jdom.Namespace;
|
|
18
|
|
import org.jdom.output.XMLOutputter;
|
|
19
|
|
|
|
20
|
|
public class Rss10Parser implements FeedParserI {
|
|
21
|
|
|
|
22
|
|
private final static Logger LOGGER = LogManager.getLogger(Rss10Parser.class);
|
|
23
|
|
|
|
24
|
|
/**
|
|
25
|
|
* @see net.sf.flock.parser.FeedParserI#isSuitable(Document)
|
|
26
|
|
*/
|
|
27
|
0
|
public boolean isSuitable(Document doc) {
|
|
28
|
0
|
Element root = doc.getRootElement();
|
|
29
|
|
|
|
30
|
0
|
LOGGER.debug("root element "+root.getName());
|
|
31
|
|
|
|
32
|
0
|
Namespace defNS = ParserUtil.getDefaultNS(root);
|
|
33
|
0
|
if (defNS == null) {
|
|
34
|
0
|
Namespace rss10 = ParserUtil.getNamespaceFromURI(root,"http://purl.org/rss/1.0/");
|
|
35
|
0
|
if (rss10==null)
|
|
36
|
0
|
return false;
|
|
37
|
|
}
|
|
38
|
0
|
if ("RDF".equals(root.getName())) {
|
|
39
|
0
|
return true;
|
|
40
|
|
}
|
|
41
|
0
|
return false;
|
|
42
|
|
}
|
|
43
|
|
|
|
44
|
|
/**
|
|
45
|
|
* @see net.sf.flock.parser.FeedParserI#parse(FeedFactoryI, Document)
|
|
46
|
|
*/
|
|
47
|
0
|
public FeedI parse(SubscriptionInfoI subscriptionInfoI,FeedFactoryI feedFactory, Document doc) throws FlockResourceException {
|
|
48
|
|
|
|
49
|
0
|
LOGGER.debug("start parsing.");
|
|
50
|
|
|
|
51
|
|
// Get the root element (must be rss)
|
|
52
|
0
|
Element root = doc.getRootElement();
|
|
53
|
0
|
Namespace defNS = ParserUtil.getDefaultNS(root);
|
|
54
|
0
|
if (defNS == null) {
|
|
55
|
0
|
defNS = ParserUtil.getNamespaceFromURI(root,"http://purl.org/rss/1.0/");
|
|
56
|
0
|
if (defNS==null)
|
|
57
|
0
|
throw new FlockResourceException("No default namespace found.");
|
|
58
|
|
}
|
|
59
|
0
|
Namespace dcNS = ParserUtil.getNamespace(root, "dc");
|
|
60
|
|
// fall back to default name space
|
|
61
|
0
|
if (dcNS == null) {
|
|
62
|
0
|
dcNS = defNS;
|
|
63
|
|
}
|
|
64
|
|
|
|
65
|
0
|
FeedI feed = feedFactory.createFeed(subscriptionInfoI);
|
|
66
|
|
|
|
67
|
|
// Get the channel element (only one occurs)
|
|
68
|
0
|
Element channel = root.getChild("channel", defNS);
|
|
69
|
|
|
|
70
|
|
// title element
|
|
71
|
0
|
feed.setTitle( HTMLUtil.unescape( channel.getChildTextTrim("title", defNS) ) );
|
|
72
|
|
|
|
73
|
|
// description element
|
|
74
|
|
// !!! feed.setDescription(channel.getChildTextTrim("description", defNS));
|
|
75
|
|
|
|
76
|
|
// link element
|
|
77
|
0
|
String link = channel.getChildTextTrim("link", defNS);
|
|
78
|
0
|
if ((link==null) || (link.length()==0)) {
|
|
79
|
|
|
|
80
|
0
|
for (Iterator iterator = channel.getAttributes().iterator();iterator.hasNext();) {
|
|
81
|
0
|
Attribute attr = (Attribute) iterator.next();
|
|
82
|
0
|
if ("about".equals(attr.getName())) {
|
|
83
|
0
|
LOGGER.debug("channel.about attribute found:"+attr.getNamespace()+":"+attr.getName()+" = "+attr.getValue());
|
|
84
|
0
|
link = attr.getValue();
|
|
85
|
0
|
break;
|
|
86
|
|
}
|
|
87
|
0
|
LOGGER.debug("channel attribute :"+attr.getNamespace()+":"+attr.getName()+" = "+attr.getValue());
|
|
88
|
|
}
|
|
89
|
|
}
|
|
90
|
0
|
feed.setSite(ParserUtil.getURL(link));
|
|
91
|
|
|
|
92
|
|
/*
|
|
93
|
|
// !!!
|
|
94
|
|
|
|
95
|
|
// creator element
|
|
96
|
|
Element creator = channel.getChild("creator", dcNS);
|
|
97
|
|
if (creator != null) {
|
|
98
|
|
feed.setCreator(creator.getTextTrim());
|
|
99
|
|
}
|
|
100
|
|
|
|
101
|
|
// publisher element
|
|
102
|
|
Element publisher = channel.getChild("publisher", dcNS);
|
|
103
|
|
if (publisher != null) {
|
|
104
|
|
feed.setPublisher(publisher.getTextTrim());
|
|
105
|
|
}
|
|
106
|
|
|
|
107
|
|
// language element
|
|
108
|
|
Element language = channel.getChild("language", dcNS);
|
|
109
|
|
if (language != null) {
|
|
110
|
|
feed.setLanguage(language.getTextTrim());
|
|
111
|
|
}
|
|
112
|
|
|
|
113
|
|
// rights element
|
|
114
|
|
Element copyright = channel.getChild("copyright", defNS);
|
|
115
|
|
if (copyright != null) {
|
|
116
|
|
feed.setCopyright(copyright.getTextTrim());
|
|
117
|
|
}
|
|
118
|
|
*/
|
|
119
|
|
|
|
120
|
0
|
long parseDate = System.currentTimeMillis();
|
|
121
|
|
|
|
122
|
|
// item elements
|
|
123
|
0
|
List items = root.getChildren("item", defNS);
|
|
124
|
0
|
Iterator i = items.iterator();
|
|
125
|
0
|
while (i.hasNext()) {
|
|
126
|
0
|
Element item = (Element) i.next();
|
|
127
|
|
// get title element
|
|
128
|
0
|
Element elTitle = item.getChild("title", defNS);
|
|
129
|
0
|
String strTitle = "<No Title>";
|
|
130
|
0
|
if (elTitle != null) {
|
|
131
|
0
|
strTitle = HTMLUtil.unescape( elTitle.getTextTrim() );
|
|
132
|
|
}
|
|
133
|
|
|
|
134
|
0
|
LOGGER.debug("Item element found (" + strTitle + ").");
|
|
135
|
|
|
|
136
|
|
// get link element
|
|
137
|
0
|
Element elLink = item.getChild("link", defNS);
|
|
138
|
0
|
String strLink = "";
|
|
139
|
0
|
if (elLink != null) {
|
|
140
|
0
|
strLink = elLink.getTextTrim();
|
|
141
|
|
}
|
|
142
|
|
|
|
143
|
|
// get description element
|
|
144
|
0
|
Element elDesc = item.getChild("description", dcNS);
|
|
145
|
0
|
if (elDesc==null) {
|
|
146
|
|
// try to get it w/o namespace
|
|
147
|
0
|
elDesc = item.getChild("description", defNS);
|
|
148
|
0
|
LOGGER.debug("description without namespace is "+ParserUtil.elementValue(elDesc,20));
|
|
149
|
|
} else {
|
|
150
|
0
|
LOGGER.debug("description with namespace is not null");
|
|
151
|
|
}
|
|
152
|
|
|
|
153
|
0
|
if (elDesc==null) {
|
|
154
|
0
|
elDesc = item.getChild("subtitle", defNS);
|
|
155
|
0
|
LOGGER.debug("subtitle is "+ParserUtil.elementValue(elDesc,20));
|
|
156
|
|
}
|
|
157
|
0
|
String strDesc = "";
|
|
158
|
0
|
if (elDesc != null) {
|
|
159
|
|
|
|
160
|
0
|
strDesc = elDesc.getTextTrim();
|
|
161
|
|
|
|
162
|
0
|
if (strDesc.length()==0) {
|
|
163
|
0
|
LOGGER.debug("text empty, use XMLOutputter");
|
|
164
|
0
|
XMLOutputter output = new XMLOutputter("",false);
|
|
165
|
0
|
strDesc = output.outputString(elDesc);
|
|
166
|
|
}
|
|
167
|
|
}
|
|
168
|
|
|
|
169
|
|
// generate new RSS item (link to article)
|
|
170
|
|
|
|
171
|
|
/*
|
|
172
|
|
article.setFound(dateParsed);
|
|
173
|
|
|
|
174
|
|
// get creator element
|
|
175
|
|
Element elCreator = item.getChild("creator", dcNS);
|
|
176
|
|
if (elCreator != null) {
|
|
177
|
|
article.setCreator(elCreator.getTextTrim());
|
|
178
|
|
}
|
|
179
|
|
// get subject element
|
|
180
|
|
Element elSubject = item.getChild("subject", dcNS);
|
|
181
|
|
if (elSubject != null) {
|
|
182
|
|
// !!! Mulitple subject elements not handled currently
|
|
183
|
|
article.setSubject(elSubject.getTextTrim());
|
|
184
|
|
}
|
|
185
|
|
*/
|
|
186
|
|
|
|
187
|
|
// get date element
|
|
188
|
0
|
Date creationTime = ParserUtil.parseDate( item.getChild("date", dcNS) );
|
|
189
|
|
|
|
190
|
|
// TODO: this is bullshit :)
|
|
191
|
0
|
if (creationTime==null) {
|
|
192
|
0
|
creationTime = new Date(parseDate--);
|
|
193
|
|
}
|
|
194
|
|
|
|
195
|
0
|
feed.newItem(creationTime, strTitle, strDesc, ParserUtil.getURL(strLink));
|
|
196
|
|
}
|
|
197
|
|
|
|
198
|
|
/*
|
|
199
|
|
// image element
|
|
200
|
|
Element image = root.getChild("image", defNS);
|
|
201
|
|
if (image != null) {
|
|
202
|
|
ImageIF rssImage =
|
|
203
|
|
cBuilder.makeImage(image.getChildTextTrim("title", defNS),
|
|
204
|
|
getURL(image.getChildTextTrim("url", defNS)));
|
|
205
|
|
Element imgWidth = image.getChild("width", defNS);
|
|
206
|
|
if (imgWidth != null) {
|
|
207
|
|
try {
|
|
208
|
|
rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim()));
|
|
209
|
|
} catch (NumberFormatException e) {
|
|
210
|
|
logger.warn(e);
|
|
211
|
|
}
|
|
212
|
|
}
|
|
213
|
|
Element imgHeight = image.getChild("height", defNS);
|
|
214
|
|
if (imgHeight != null) {
|
|
215
|
|
try {
|
|
216
|
|
rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim()));
|
|
217
|
|
} catch (NumberFormatException e) {
|
|
218
|
|
logger.warn(e);
|
|
219
|
|
}
|
|
220
|
|
}
|
|
221
|
|
Element imgDescr = image.getChild("description", dcNS);
|
|
222
|
|
if (imgDescr != null) {
|
|
223
|
|
rssImage.setDescription(imgDescr.getTextTrim());
|
|
224
|
|
}
|
|
225
|
|
feed.setImage(rssImage);
|
|
226
|
|
}
|
|
227
|
|
|
|
228
|
|
// textinput element
|
|
229
|
|
Element txtinp = root.getChild("textinput", defNS);
|
|
230
|
|
if (txtinp != null) {
|
|
231
|
|
String tiTitle = null;
|
|
232
|
|
if (txtinp.getChild("title", defNS) != null) {
|
|
233
|
|
tiTitle = txtinp.getChild("title", defNS).getTextTrim();
|
|
234
|
|
}
|
|
235
|
|
String tiDescr = null;
|
|
236
|
|
if (txtinp.getChild("description", dcNS) != null) {
|
|
237
|
|
tiDescr = txtinp.getChild("description", dcNS).getTextTrim();
|
|
238
|
|
}
|
|
239
|
|
URL tiLink = null;
|
|
240
|
|
if (txtinp.getChild("link", defNS) != null) {
|
|
241
|
|
tiLink = getURL(txtinp.getChild("link", defNS).getTextTrim());
|
|
242
|
|
}
|
|
243
|
|
TextInputIF rssTextInput =
|
|
244
|
|
cBuilder.makeTextInput(tiTitle, tiDescr, tiLink);
|
|
245
|
|
feed.setTextInput(rssTextInput);
|
|
246
|
|
}
|
|
247
|
|
*/
|
|
248
|
|
|
|
249
|
0
|
return feed;
|
|
250
|
|
}
|
|
251
|
|
|
|
252
|
|
}
|
|
253
|
|
|
