1 | /** |
---|
2 | * This software is copyright (c) 2011 by |
---|
3 | * - Institut fuer Deutsche Sprache (http://www.ids-mannheim.de) |
---|
4 | * This is free software. You can redistribute it |
---|
5 | * and/or modify it under the terms described in |
---|
6 | * the GNU General Public License v3 of which you |
---|
7 | * should have received a copy. Otherwise you can download |
---|
8 | * it from |
---|
9 | * |
---|
10 | * http://www.gnu.org/licenses/gpl-3.0.txt |
---|
11 | * |
---|
12 | * @copyright Institut fuer Deutsche Sprache (http://www.ids-mannheim.de) |
---|
13 | * |
---|
14 | * @license http://www.gnu.org/licenses/gpl-3.0.txt |
---|
15 | * GNU General Public License v3 |
---|
16 | */ |
---|
17 | package eu.clarin.sru.fcs; |
---|
18 | |
---|
19 | import javax.xml.stream.XMLStreamException; |
---|
20 | import javax.xml.stream.XMLStreamReader; |
---|
21 | |
---|
22 | import org.slf4j.Logger; |
---|
23 | import org.slf4j.LoggerFactory; |
---|
24 | |
---|
25 | import eu.clarin.sru.client.SRUClientException; |
---|
26 | import eu.clarin.sru.client.SRURecordData; |
---|
27 | import eu.clarin.sru.client.SRURecordDataParser; |
---|
28 | import eu.clarin.sru.client.XmlStreamReaderUtils; |
---|
29 | |
---|
30 | |
---|
31 | /** |
---|
32 | * A record for CLARIN FCS. |
---|
33 | */ |
---|
34 | public class ClarinFederatedContentSearchRecordParser implements |
---|
35 | SRURecordDataParser { |
---|
36 | private static final Logger logger = |
---|
37 | LoggerFactory.getLogger(ClarinFederatedContentSearchRecordParser.class); |
---|
38 | public static final String FCS_NS = |
---|
39 | ClarinFederatedContentSearchRecordData.RECORD_SCHEMA; |
---|
40 | public static final String FCS_KWIC_NS = "http://clarin.eu/fcs/1.0/kwic"; |
---|
41 | private static final String DATAVIEW_KWIC = "kwic"; |
---|
42 | |
---|
43 | |
---|
44 | @Override |
---|
45 | public String getRecordSchema() { |
---|
46 | return ClarinFederatedContentSearchRecordData.RECORD_SCHEMA; |
---|
47 | } |
---|
48 | |
---|
49 | |
---|
50 | @Override |
---|
51 | public SRURecordData parse(XMLStreamReader reader) |
---|
52 | throws XMLStreamException, SRUClientException { |
---|
53 | XmlStreamReaderUtils.readStart(reader, FCS_NS, "Resource", true, true); |
---|
54 | String pid = XmlStreamReaderUtils.readAttributeValue(reader, null, |
---|
55 | "pid"); |
---|
56 | XmlStreamReaderUtils.consumeStart(reader); |
---|
57 | |
---|
58 | String left = null; |
---|
59 | String keyword = null; |
---|
60 | String right = null; |
---|
61 | |
---|
62 | boolean first = true; |
---|
63 | boolean kwic = false; |
---|
64 | |
---|
65 | while (XmlStreamReaderUtils.readStart(reader, FCS_NS, "DataView", |
---|
66 | first, true)) { |
---|
67 | first = false; |
---|
68 | String type = XmlStreamReaderUtils.readAttributeValue(reader, null, |
---|
69 | "type"); |
---|
70 | XmlStreamReaderUtils.consumeStart(reader); |
---|
71 | if ((type == null) || type.isEmpty()) { |
---|
72 | throw new SRUClientException( |
---|
73 | "DataView element need a non-empty 'type' attribute"); |
---|
74 | } |
---|
75 | logger.debug("found DataView @type = {}", type); |
---|
76 | if (DATAVIEW_KWIC.equals(type)) { |
---|
77 | if (kwic) { |
---|
78 | throw new SRUClientException( |
---|
79 | "only one KWIC dataview is allowed"); |
---|
80 | } |
---|
81 | XmlStreamReaderUtils.readStart(reader, FCS_KWIC_NS, "kwic", |
---|
82 | true); |
---|
83 | if (XmlStreamReaderUtils.readStart(reader, FCS_KWIC_NS, "c", |
---|
84 | false)) { |
---|
85 | left = XmlStreamReaderUtils.readString(reader, false); |
---|
86 | XmlStreamReaderUtils.readEnd(reader, FCS_KWIC_NS, "c"); |
---|
87 | } |
---|
88 | keyword = XmlStreamReaderUtils.readContent(reader, FCS_KWIC_NS, |
---|
89 | "kw", true); |
---|
90 | if (XmlStreamReaderUtils.readStart(reader, FCS_KWIC_NS, "c", |
---|
91 | false)) { |
---|
92 | right = XmlStreamReaderUtils.readString(reader, false); |
---|
93 | XmlStreamReaderUtils.readEnd(reader, FCS_KWIC_NS, "c"); |
---|
94 | } |
---|
95 | XmlStreamReaderUtils.readEnd(reader, FCS_KWIC_NS, "kwic"); |
---|
96 | kwic = true; |
---|
97 | } else { |
---|
98 | logger.warn("skipping dataview of type '{}'", type); |
---|
99 | } |
---|
100 | XmlStreamReaderUtils.readEnd(reader, FCS_NS, "DataView", true); |
---|
101 | } // while |
---|
102 | |
---|
103 | XmlStreamReaderUtils.readEnd(reader, FCS_NS, "Resource"); |
---|
104 | |
---|
105 | if (kwic) { |
---|
106 | return new ClarinFederatedContentSearchRecordData(pid, left, |
---|
107 | keyword, right); |
---|
108 | } else { |
---|
109 | throw new SRUClientException("no mandatroy kwic dataview found"); |
---|
110 | } |
---|
111 | } |
---|
112 | |
---|
113 | } // class ClarinFederatedContentSearchRecordParser |
---|