1 | /** |
---|
2 | * This software is copyright (c) 2012-2014 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 java.util.Arrays; |
---|
20 | |
---|
21 | import javax.xml.stream.XMLStreamException; |
---|
22 | import javax.xml.stream.XMLStreamReader; |
---|
23 | |
---|
24 | import org.slf4j.Logger; |
---|
25 | import org.slf4j.LoggerFactory; |
---|
26 | |
---|
27 | import eu.clarin.sru.client.SRUClientException; |
---|
28 | import eu.clarin.sru.client.XmlStreamReaderUtils; |
---|
29 | |
---|
30 | |
---|
31 | /** |
---|
32 | * An implementation of a Data View parser that parses HITS Data Views. This |
---|
33 | * parser expects input that conforms to the CLARIN-FCS specification for the |
---|
34 | * HITS Data View. |
---|
35 | * |
---|
36 | * @see DataViewHits |
---|
37 | */ |
---|
38 | public final class DataViewParserHits implements DataViewParser { |
---|
39 | private static final int OFFSET_CHUNK_SIZE = 8; |
---|
40 | private static final String FCS_HITS_NS = |
---|
41 | "http://clarin.eu/fcs/dataview/hits"; |
---|
42 | private static final Logger logger = |
---|
43 | LoggerFactory.getLogger(DataViewParserHits.class); |
---|
44 | |
---|
45 | |
---|
46 | @Override |
---|
47 | public boolean acceptType(String type) { |
---|
48 | return DataViewHits.TYPE.equals(type); |
---|
49 | } |
---|
50 | |
---|
51 | |
---|
52 | @Override |
---|
53 | public int getPriority() { |
---|
54 | return 1000; |
---|
55 | } |
---|
56 | |
---|
57 | |
---|
58 | @Override |
---|
59 | public DataView parse(XMLStreamReader reader, String type, String pid, |
---|
60 | String ref) throws XMLStreamException, SRUClientException { |
---|
61 | int offsets[] = new int[OFFSET_CHUNK_SIZE]; |
---|
62 | int offsets_idx = 0; |
---|
63 | StringBuilder buffer = new StringBuilder(); |
---|
64 | XmlStreamReaderUtils.readStart(reader, FCS_HITS_NS, "Result", true); |
---|
65 | |
---|
66 | int idx = 0; |
---|
67 | while (!XmlStreamReaderUtils.peekEnd(reader, FCS_HITS_NS, "Result")) { |
---|
68 | if (buffer.length() > 0) { |
---|
69 | if (!Character.isWhitespace(buffer.charAt(buffer.length() - 1))) { |
---|
70 | buffer.append(' '); |
---|
71 | } |
---|
72 | idx = buffer.length(); |
---|
73 | } |
---|
74 | |
---|
75 | if (XmlStreamReaderUtils.readStart(reader, FCS_HITS_NS, "Hit", false)) { |
---|
76 | String hit = XmlStreamReaderUtils.readString(reader, false); |
---|
77 | XmlStreamReaderUtils.readEnd(reader, FCS_HITS_NS, "Hit"); |
---|
78 | if (hit.length() > 0) { |
---|
79 | buffer.append(hit); |
---|
80 | if (offsets_idx == offsets.length) { |
---|
81 | offsets = Arrays.copyOf(offsets, offsets.length + 8); |
---|
82 | } |
---|
83 | /* |
---|
84 | * add pair of offsets and simultaneously increase index |
---|
85 | */ |
---|
86 | offsets[offsets_idx++] = idx; |
---|
87 | offsets[offsets_idx++] = idx + hit.length(); |
---|
88 | } else { |
---|
89 | logger.warn("skipping empty <Hit> element within <Result> element"); |
---|
90 | } |
---|
91 | } else { |
---|
92 | buffer.append(XmlStreamReaderUtils.readString(reader, false)); |
---|
93 | } |
---|
94 | } // while |
---|
95 | XmlStreamReaderUtils.readEnd(reader, FCS_HITS_NS, "Result"); |
---|
96 | |
---|
97 | final String text = buffer.toString(); |
---|
98 | return new DataViewHits(pid, ref, text, offsets, offsets_idx); |
---|
99 | } |
---|
100 | |
---|
101 | } // class DataViewParserHits |
---|