|
1
|
|
package net.sf.flock.parser;
|
|
2
|
|
|
|
3
|
|
import java.text.SimpleDateFormat;
|
|
4
|
|
import java.util.Date;
|
|
5
|
|
import java.util.Iterator;
|
|
6
|
|
import java.util.List;
|
|
7
|
|
|
|
8
|
|
import net.sf.flock.FeedFactoryI;
|
|
9
|
|
import net.sf.flock.FeedI;
|
|
10
|
|
import net.sf.flock.FlockResourceException;
|
|
11
|
|
import net.sf.flock.SubscriptionInfoI;
|
|
12
|
|
|
|
13
|
|
import org.apache.log4j.LogManager;
|
|
14
|
|
import org.apache.log4j.Logger;
|
|
15
|
|
import org.jdom.Attribute;
|
|
16
|
|
import org.jdom.Document;
|
|
17
|
|
import org.jdom.Element;
|
|
18
|
|
import org.jdom.Namespace;
|
|
19
|
|
|
|
20
|
|
public class Rss20Parser implements FeedParserI {
|
|
21
|
|
|
|
22
|
|
private final static Logger LOGGER = LogManager.getLogger(Rss20Parser.class);
|
|
23
|
|
|
|
24
|
|
private SimpleDateFormat timeFormat = new SimpleDateFormat("yyyy.MM.dd H:mm");
|
|
25
|
|
|
|
26
|
|
/**
|
|
27
|
|
* @see net.sf.flock.parser.FeedParserI#isSuitable(Document)
|
|
28
|
|
*/
|
|
29
|
0
|
public boolean isSuitable(Document doc) {
|
|
30
|
0
|
Element root = doc.getRootElement();
|
|
31
|
0
|
if (!"rss".equals(root.getName())) {
|
|
32
|
0
|
return false;
|
|
33
|
|
}
|
|
34
|
0
|
Attribute rssVersion = root.getAttribute("version");
|
|
35
|
0
|
if (rssVersion==null || rssVersion.getValue()==null) {
|
|
36
|
0
|
return false;
|
|
37
|
|
}
|
|
38
|
0
|
LOGGER.debug("Found RSS version " + rssVersion.getValue());
|
|
39
|
0
|
return rssVersion.getValue().startsWith("2.0");
|
|
40
|
|
}
|
|
41
|
|
|
|
42
|
|
/**
|
|
43
|
|
* @see net.sf.flock.parser.FeedParserI#parse(FeedFactoryI,Document)
|
|
44
|
|
*/
|
|
45
|
0
|
public FeedI parse(SubscriptionInfoI subscriptionInfoI,FeedFactoryI feedFactory, Document doc) throws FlockResourceException {
|
|
46
|
|
|
|
47
|
0
|
Date dateParsed = new Date();
|
|
48
|
0
|
LOGGER.debug("start parsing.");
|
|
49
|
|
|
|
50
|
0
|
Element root = doc.getRootElement();
|
|
51
|
|
|
|
52
|
|
// Get the channel element (only one occurs)
|
|
53
|
0
|
Element channel = root.getChild("channel");
|
|
54
|
|
|
|
55
|
|
// 1 title element
|
|
56
|
0
|
FeedI feed = feedFactory.createFeed(subscriptionInfoI);
|
|
57
|
|
|
|
58
|
0
|
feed.setTitle( HTMLUtil.unescape( channel.getChildTextTrim("title") ) );
|
|
59
|
|
|
|
60
|
|
// 1 description element
|
|
61
|
|
// !!! feed.setDescription(channel.getChildTextTrim("description"));
|
|
62
|
|
|
|
63
|
|
// 1 link element
|
|
64
|
0
|
feed.setSite(ParserUtil.getURL(channel.getChildTextTrim("link")));
|
|
65
|
|
|
|
66
|
|
// 1 language element
|
|
67
|
|
// !!! feed.setLanguage(channel.getChildTextTrim("language"));
|
|
68
|
|
|
|
69
|
|
// 1..n item elements
|
|
70
|
0
|
List items = channel.getChildren("item");
|
|
71
|
0
|
Iterator i = items.iterator();
|
|
72
|
0
|
while (i.hasNext()) {
|
|
73
|
0
|
Element item = (Element) i.next();
|
|
74
|
|
|
|
75
|
|
// get description element
|
|
76
|
0
|
Element elDesc = item.getChild("description");
|
|
77
|
0
|
String strDesc = "";
|
|
78
|
0
|
if (elDesc != null) {
|
|
79
|
0
|
strDesc = elDesc.getTextTrim();
|
|
80
|
|
}
|
|
81
|
|
|
|
82
|
|
// get link element
|
|
83
|
0
|
Element elLink = item.getChild("link");
|
|
84
|
0
|
String strLink = "";
|
|
85
|
0
|
if (elLink != null) {
|
|
86
|
0
|
strLink = elLink.getTextTrim();
|
|
87
|
0
|
LOGGER.debug("found link '"+strLink+"'");
|
|
88
|
|
} else {
|
|
89
|
|
// try 'guid'
|
|
90
|
|
// ... http://radio.weblogs.com/0112015/rss.xml
|
|
91
|
0
|
elLink = item.getChild("guid");
|
|
92
|
0
|
if (elLink != null) {
|
|
93
|
0
|
strLink = elLink.getTextTrim();
|
|
94
|
0
|
LOGGER.debug("found guid '"+strLink+"'");
|
|
95
|
|
} else {
|
|
96
|
0
|
LOGGER.debug("no link/guid");
|
|
97
|
|
}
|
|
98
|
|
|
|
99
|
|
}
|
|
100
|
|
|
|
101
|
|
|
|
102
|
|
|
|
103
|
0
|
Date pubDate = ParserUtil.parseDate(item.getChild("pubDate"));
|
|
104
|
|
|
|
105
|
0
|
if (pubDate==null) {
|
|
106
|
|
// try dc namespace
|
|
107
|
0
|
Namespace dcNS = ParserUtil.getNamespace(root, "dc");
|
|
108
|
0
|
if (dcNS!=null) {
|
|
109
|
0
|
pubDate = ParserUtil.parseDate( item.getChild("date", dcNS) );
|
|
110
|
|
}
|
|
111
|
|
}
|
|
112
|
|
|
|
113
|
0
|
LOGGER.debug("date :"+pubDate);
|
|
114
|
|
|
|
115
|
|
|
|
116
|
0
|
Date creationTime = pubDate!=null ? pubDate : dateParsed;
|
|
117
|
|
// !!! item.setFound(dateParsed);
|
|
118
|
|
|
|
119
|
0
|
Element elTitle = item.getChild("title");
|
|
120
|
0
|
LOGGER.debug("title '"+(elTitle==null ? "null" : elTitle.getTextTrim()) +"'");
|
|
121
|
0
|
String title =
|
|
122
|
|
elTitle!=null ? HTMLUtil.unescape( elTitle.getTextTrim() ) :
|
|
123
|
|
feed.getTitle() + " " + this.timeFormat.format(creationTime);
|
|
124
|
|
|
|
125
|
0
|
feed.newItem(creationTime, title, strDesc, ParserUtil.getURL(strLink));
|
|
126
|
|
}
|
|
127
|
|
|
|
128
|
|
// 0..1 rating element
|
|
129
|
|
|
|
130
|
|
// 0..1 image element
|
|
131
|
|
/*
|
|
132
|
|
Element image = channel.getChild("image");
|
|
133
|
|
if (image != null) {
|
|
134
|
|
ImageIF rssImage =
|
|
135
|
|
cBuilder.makeImage(image.getChildTextTrim("title"),
|
|
136
|
|
getURL(image.getChildTextTrim("url")));
|
|
137
|
|
Element imgWidth = image.getChild("width");
|
|
138
|
|
if (imgWidth != null) {
|
|
139
|
|
try {
|
|
140
|
|
rssImage.setWidth(Integer.parseInt(imgWidth.getTextTrim()));
|
|
141
|
|
} catch (NumberFormatException e) {
|
|
142
|
|
logger.warn(e);
|
|
143
|
|
}
|
|
144
|
|
}
|
|
145
|
|
Element imgHeight = image.getChild("height");
|
|
146
|
|
if (imgHeight != null) {
|
|
147
|
|
try {
|
|
148
|
|
rssImage.setHeight(Integer.parseInt(imgHeight.getTextTrim()));
|
|
149
|
|
} catch (NumberFormatException e) {
|
|
150
|
|
logger.warn(e);
|
|
151
|
|
}
|
|
152
|
|
}
|
|
153
|
|
Element imgDescr = image.getChild("description");
|
|
154
|
|
if (imgDescr != null) {
|
|
155
|
|
rssImage.setDescription(imgDescr.getTextTrim());
|
|
156
|
|
}
|
|
157
|
|
chnl.setImage(rssImage);
|
|
158
|
|
}
|
|
159
|
|
|
|
160
|
|
// 0..1 textinput element
|
|
161
|
|
Element txtinp = channel.getChild("textinput");
|
|
162
|
|
if (txtinp != null) {
|
|
163
|
|
TextInputIF rssTextInput =
|
|
164
|
|
cBuilder.makeTextInput(txtinp.getChild("title").getTextTrim(),
|
|
165
|
|
txtinp.getChild("description").getTextTrim(),
|
|
166
|
|
getURL(txtinp.getChild("link").getTextTrim()));
|
|
167
|
|
chnl.setTextInput(rssTextInput);
|
|
168
|
|
}
|
|
169
|
|
|
|
170
|
|
// 0..1 copyright element
|
|
171
|
|
Element copyright = channel.getChild("copyright");
|
|
172
|
|
if (copyright != null) {
|
|
173
|
|
chnl.setCopyright(copyright.getTextTrim());
|
|
174
|
|
}
|
|
175
|
|
|
|
176
|
|
// 0..1 pubDate element
|
|
177
|
|
Element pubDate = channel.getChild("pubDate");
|
|
178
|
|
// if (pubDate != null) {
|
|
179
|
|
// chnl.setPubDate(pubDate.getTextTrim());
|
|
180
|
|
// }
|
|
181
|
|
|
|
182
|
|
// 0..1 lastBuildDate element
|
|
183
|
|
Element lastBuildDate = channel.getChild("lastBuildDate");
|
|
184
|
|
// if (lastBuildDate != null) {
|
|
185
|
|
// chnl.setLastBuildDate(lastBuildDate.getTextTrim());
|
|
186
|
|
// }
|
|
187
|
|
|
|
188
|
|
// 0..1 docs element
|
|
189
|
|
|
|
190
|
|
// 0..1 managingEditor element
|
|
191
|
|
Element managingEditor = channel.getChild("managingEditor");
|
|
192
|
|
if (managingEditor != null) {
|
|
193
|
|
chnl.setCreator(managingEditor.getTextTrim());
|
|
194
|
|
}
|
|
195
|
|
|
|
196
|
|
// 0..1 webMaster element
|
|
197
|
|
Element webMaster = channel.getChild("webMaster");
|
|
198
|
|
if (webMaster != null) {
|
|
199
|
|
chnl.setPublisher(webMaster.getTextTrim());
|
|
200
|
|
}
|
|
201
|
|
|
|
202
|
|
// 0..1 skipHours element
|
|
203
|
|
// 0..1 skipDays element
|
|
204
|
|
*/
|
|
205
|
0
|
return feed;
|
|
206
|
|
}
|
|
207
|
|
|
|
208
|
|
}
|
|
209
|
|
|
