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 java.util.LinkedList; |
---|
20 | import java.util.List; |
---|
21 | |
---|
22 | import javax.xml.stream.XMLStreamException; |
---|
23 | import javax.xml.stream.XMLStreamReader; |
---|
24 | import org.slf4j.Logger; |
---|
25 | import org.slf4j.LoggerFactory; |
---|
26 | import eu.clarin.sru.client.SRUClientException; |
---|
27 | import eu.clarin.sru.client.SRURecordData; |
---|
28 | import eu.clarin.sru.client.SRURecordDataParser; |
---|
29 | import eu.clarin.sru.client.XmlStreamReaderUtils; |
---|
30 | |
---|
31 | |
---|
32 | /** |
---|
33 | * A record parse to parse records conforming to CLARIN-FCS specification. |
---|
34 | */ |
---|
35 | public class ClarinFCSRecordParser implements SRURecordDataParser { |
---|
36 | private static final Logger logger = |
---|
37 | LoggerFactory.getLogger(ClarinFCSRecordParser.class); |
---|
38 | private static final String FCS_NS = |
---|
39 | ClarinFCSRecordData.RECORD_SCHEMA; |
---|
40 | // TODO: make this configurable |
---|
41 | private final DataViewParser[] parsers = new DataViewParser[] { |
---|
42 | new DataViewParserGenericDOM(), |
---|
43 | // new DataViewParserGenericString(), |
---|
44 | new DataViewParserKWIC() |
---|
45 | }; |
---|
46 | |
---|
47 | |
---|
48 | @Override |
---|
49 | public String getRecordSchema() { |
---|
50 | return ClarinFCSRecordData.RECORD_SCHEMA; |
---|
51 | } |
---|
52 | |
---|
53 | |
---|
54 | @Override |
---|
55 | public SRURecordData parse(XMLStreamReader reader) |
---|
56 | throws XMLStreamException, SRUClientException { |
---|
57 | logger.debug("parsing CLARIN-FCS record"); |
---|
58 | |
---|
59 | // Resource |
---|
60 | XmlStreamReaderUtils.readStart(reader, FCS_NS, "Resource", true, true); |
---|
61 | String pid = XmlStreamReaderUtils.readAttributeValue(reader, null, "pid"); |
---|
62 | String ref = XmlStreamReaderUtils.readAttributeValue(reader, null, "ref"); |
---|
63 | XmlStreamReaderUtils.consumeStart(reader); |
---|
64 | |
---|
65 | // Resource/Resource (optional) |
---|
66 | if (XmlStreamReaderUtils.readStart(reader, FCS_NS, "Resource", false)) { |
---|
67 | logger.info("skipping nested <Resource> element"); |
---|
68 | XmlStreamReaderUtils.readEnd(reader, FCS_NS, "Resource", true); |
---|
69 | } |
---|
70 | |
---|
71 | // Resource/DataView |
---|
72 | final List<DataView> dataviews = parseDataViews(reader); |
---|
73 | |
---|
74 | // Resource/ResourceFragment |
---|
75 | final List<Resource.ResourceFragment> resourceFragments = |
---|
76 | parseResourceFragments(reader); |
---|
77 | |
---|
78 | XmlStreamReaderUtils.readEnd(reader, FCS_NS, "Resource", true); |
---|
79 | |
---|
80 | return new ClarinFCSRecordData(pid, ref, dataviews, resourceFragments); |
---|
81 | } |
---|
82 | |
---|
83 | |
---|
84 | private List<DataView> parseDataViews(XMLStreamReader reader) |
---|
85 | throws XMLStreamException, SRUClientException { |
---|
86 | List<DataView> dataviews = null; |
---|
87 | |
---|
88 | while (XmlStreamReaderUtils.readStart(reader, FCS_NS, "DataView", false, true)) { |
---|
89 | String pid = XmlStreamReaderUtils.readAttributeValue(reader, null, "pid"); |
---|
90 | String ref = XmlStreamReaderUtils.readAttributeValue(reader, null, "ref"); |
---|
91 | String type = XmlStreamReaderUtils.readAttributeValue(reader, null, "type"); |
---|
92 | if ((type == null) || type.isEmpty()) { |
---|
93 | throw new SRUClientException("element <DataView> needs a " |
---|
94 | + "non-empty 'type' attribute"); |
---|
95 | } |
---|
96 | |
---|
97 | // consume start element and get rid of any whitespace |
---|
98 | XmlStreamReaderUtils.consumeStart(reader); |
---|
99 | XmlStreamReaderUtils.consumeWhitespace(reader); |
---|
100 | |
---|
101 | logger.debug("processing <DataView> of type = {}", type); |
---|
102 | |
---|
103 | DataViewParser parser = null; |
---|
104 | for (int i = 0; i < parsers.length; i++) { |
---|
105 | if (parsers[i].acceptType(type) && |
---|
106 | ((parser == null) || |
---|
107 | (parser.getPriority() < parsers[i].getPriority()))) { |
---|
108 | parser = parsers[i]; |
---|
109 | } |
---|
110 | } |
---|
111 | |
---|
112 | DataView dataview = null; |
---|
113 | if (parser != null) { |
---|
114 | dataview = parser.parse(reader, type, pid, ref); |
---|
115 | } else { |
---|
116 | logger.warn("no parser found for <DataView> of type = {}", type); |
---|
117 | } |
---|
118 | |
---|
119 | XmlStreamReaderUtils.readEnd(reader, FCS_NS, "DataView", true); |
---|
120 | |
---|
121 | if (dataview != null) { |
---|
122 | if (dataviews == null) { |
---|
123 | dataviews = new LinkedList<DataView>(); |
---|
124 | } |
---|
125 | dataviews.add(dataview); |
---|
126 | } else { |
---|
127 | logger.warn("skipped <DataView> of type = {}", type); |
---|
128 | } |
---|
129 | } // while |
---|
130 | return dataviews; |
---|
131 | } |
---|
132 | |
---|
133 | |
---|
134 | private List<Resource.ResourceFragment> parseResourceFragments( |
---|
135 | XMLStreamReader reader) throws XMLStreamException, |
---|
136 | SRUClientException { |
---|
137 | List<Resource.ResourceFragment> resourceFragments = null; |
---|
138 | while (XmlStreamReaderUtils.readStart(reader, FCS_NS, "ResourceFragment", false, true)) { |
---|
139 | logger.debug("found ResourceFragment"); |
---|
140 | String pid = XmlStreamReaderUtils.readAttributeValue(reader, null, "pid"); |
---|
141 | String ref = XmlStreamReaderUtils.readAttributeValue(reader, null, "ref"); |
---|
142 | XmlStreamReaderUtils.consumeStart(reader); |
---|
143 | final List<DataView> dataviews = parseDataViews(reader); |
---|
144 | XmlStreamReaderUtils.readEnd(reader, FCS_NS, "ResourceFragment", true); |
---|
145 | |
---|
146 | if (resourceFragments == null) { |
---|
147 | resourceFragments = new LinkedList<Resource.ResourceFragment>(); |
---|
148 | } |
---|
149 | resourceFragments.add(new Resource.ResourceFragment(pid, ref, dataviews)); |
---|
150 | } // while |
---|
151 | return resourceFragments; |
---|
152 | } |
---|
153 | |
---|
154 | } // class ClarinFCSRecordParser |
---|