1 | /** |
---|
2 | * This software is copyright (c) 2012-2013 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.client.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.XmlStreamReaderUtils; |
---|
27 | |
---|
28 | public class DataViewParserKWIC implements DataViewParser { |
---|
29 | private static final String FCS_KWIC_NS = "http://clarin.eu/fcs/1.0/kwic"; |
---|
30 | private static final String KWIC_LEGACY_TYPE = "kwic"; |
---|
31 | private static final String KWIC_TYPE = |
---|
32 | "application/x-clarin-fcs-kwic+xml"; |
---|
33 | private static final Logger logger = |
---|
34 | LoggerFactory.getLogger(DataViewParserKWIC.class); |
---|
35 | |
---|
36 | @Override |
---|
37 | public boolean acceptType(String type) { |
---|
38 | return KWIC_TYPE.equals(type) || KWIC_LEGACY_TYPE.equals(type); |
---|
39 | } |
---|
40 | |
---|
41 | @Override |
---|
42 | public int getPriority() { |
---|
43 | return 1000; |
---|
44 | } |
---|
45 | |
---|
46 | @Override |
---|
47 | public DataView parse(XMLStreamReader reader, String type, String pid, |
---|
48 | String ref) throws XMLStreamException, SRUClientException { |
---|
49 | if (KWIC_LEGACY_TYPE.equals(type)) { |
---|
50 | logger.warn("type '" + KWIC_LEGACY_TYPE + "' is deprecated " + |
---|
51 | "for a KWIC <DataView>, please use '" + |
---|
52 | KWIC_TYPE + "' instead"); |
---|
53 | } |
---|
54 | String left = null; |
---|
55 | String keyword = null; |
---|
56 | String right = null; |
---|
57 | |
---|
58 | XmlStreamReaderUtils.readStart(reader, FCS_KWIC_NS, "kwic", true); |
---|
59 | if (XmlStreamReaderUtils.readStart(reader, FCS_KWIC_NS, "c", false)) { |
---|
60 | left = XmlStreamReaderUtils.readString(reader, false); |
---|
61 | XmlStreamReaderUtils.readEnd(reader, FCS_KWIC_NS, "c"); |
---|
62 | } |
---|
63 | keyword = XmlStreamReaderUtils.readContent(reader, FCS_KWIC_NS, "kw", true); |
---|
64 | if (XmlStreamReaderUtils.readStart(reader, FCS_KWIC_NS, "c", false)) { |
---|
65 | right = XmlStreamReaderUtils.readString(reader, false); |
---|
66 | XmlStreamReaderUtils.readEnd(reader, FCS_KWIC_NS, "c"); |
---|
67 | } |
---|
68 | XmlStreamReaderUtils.readEnd(reader, FCS_KWIC_NS, "kwic"); |
---|
69 | |
---|
70 | logger.debug("left='{}' keyword='{}', right='{}'", |
---|
71 | left, keyword, right); |
---|
72 | |
---|
73 | logger.warn("Upgraded deprecated KWIC dataview to HITS dataview. " + |
---|
74 | "Please upgrade to the new CLARIN-FCS specification " + |
---|
75 | "as soon as possible."); |
---|
76 | final int[] offsets = new int[3]; |
---|
77 | final StringBuilder sb = new StringBuilder(); |
---|
78 | if (left != null) { |
---|
79 | sb.append(left); |
---|
80 | if (!Character.isWhitespace(sb.charAt(sb.length() - 1))) { |
---|
81 | sb.append(" "); |
---|
82 | } |
---|
83 | } |
---|
84 | offsets[0] = sb.length(); |
---|
85 | sb.append(keyword); |
---|
86 | offsets[1] = sb.length(); |
---|
87 | if (right != null) { |
---|
88 | if (!Character.isWhitespace(sb.charAt(sb.length() - 1))) { |
---|
89 | sb.append(" "); |
---|
90 | } |
---|
91 | sb.append(right); |
---|
92 | } |
---|
93 | return new DataViewHits(pid, ref, sb.toString(), offsets, 3); |
---|
94 | } |
---|
95 | |
---|
96 | } // class DataViewParserKWIC |
---|