1 | /** |
---|
2 | * This software is copyright (c) 2012-2022 by |
---|
3 | * - Leibniz-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 Leibniz-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.ArrayList; |
---|
20 | import java.util.Collections; |
---|
21 | import java.util.LinkedList; |
---|
22 | import java.util.List; |
---|
23 | |
---|
24 | import javax.xml.stream.XMLStreamException; |
---|
25 | import javax.xml.stream.XMLStreamReader; |
---|
26 | |
---|
27 | import org.slf4j.Logger; |
---|
28 | import org.slf4j.LoggerFactory; |
---|
29 | |
---|
30 | import eu.clarin.sru.client.SRUClientException; |
---|
31 | import eu.clarin.sru.client.SRURecordData; |
---|
32 | import eu.clarin.sru.client.SRURecordDataParser; |
---|
33 | import eu.clarin.sru.client.XmlStreamReaderUtils; |
---|
34 | |
---|
35 | |
---|
36 | /** |
---|
37 | * An abstract base class for CLARIN-FCS record data parsers. |
---|
38 | */ |
---|
39 | abstract class AbstractClarinFCSRecordDataParser implements SRURecordDataParser { |
---|
40 | protected static final Logger logger = |
---|
41 | LoggerFactory.getLogger(AbstractClarinFCSRecordDataParser.class); |
---|
42 | protected final List<DataViewParser> parsers; |
---|
43 | |
---|
44 | |
---|
45 | /** |
---|
46 | * Constructor. |
---|
47 | * |
---|
48 | * @param parsers |
---|
49 | * the list of data view parsers to be used by this record data |
---|
50 | * parser. This list should contain one |
---|
51 | * {@link DataViewParserGenericDOM} or |
---|
52 | * {@link DataViewParserGenericString} instance. |
---|
53 | * @throws NullPointerException |
---|
54 | * if parsers is <code>null</code> |
---|
55 | * @throws IllegalArgumentException |
---|
56 | * if parsers is empty or contains duplicate entries |
---|
57 | */ |
---|
58 | protected AbstractClarinFCSRecordDataParser(List<DataViewParser> parsers) { |
---|
59 | if (parsers == null) { |
---|
60 | throw new NullPointerException("parsers == null"); |
---|
61 | } |
---|
62 | if (parsers.isEmpty()) { |
---|
63 | throw new IllegalArgumentException("parsers is empty"); |
---|
64 | } |
---|
65 | |
---|
66 | ArrayList<DataViewParser> list = new ArrayList<DataViewParser>(); |
---|
67 | boolean foundGeneric = false; |
---|
68 | for (DataViewParser parser : parsers) { |
---|
69 | if ((parser instanceof DataViewParserGenericDOM) || |
---|
70 | (parser instanceof DataViewParserGenericString)) { |
---|
71 | if (foundGeneric) { |
---|
72 | throw new IllegalAccessError("parser list should contain " + |
---|
73 | "only one instance of DataViewParserGenericDOM " + |
---|
74 | "or DataViewParserGenericString"); |
---|
75 | } |
---|
76 | foundGeneric = true; |
---|
77 | } |
---|
78 | if (list.contains(parser)) { |
---|
79 | throw new IllegalArgumentException("parser list should not " + |
---|
80 | "contain douplicates: " + parser.getClass().getName()); |
---|
81 | } |
---|
82 | list.add(parser); |
---|
83 | } |
---|
84 | |
---|
85 | if (!foundGeneric) { |
---|
86 | logger.warn("No generic type data view parser found. You should " + |
---|
87 | "make sure that the parser list contains one " + |
---|
88 | "DataViewParserGenericDOM or DataViewParserGenericString " + |
---|
89 | "instance"); |
---|
90 | } |
---|
91 | |
---|
92 | this.parsers = Collections.unmodifiableList(list); |
---|
93 | } |
---|
94 | |
---|
95 | |
---|
96 | @SuppressWarnings("deprecation") |
---|
97 | protected SRURecordData parse(XMLStreamReader reader, String ns, |
---|
98 | boolean fullLegacyCompatMode) |
---|
99 | throws XMLStreamException, SRUClientException { |
---|
100 | // Resource |
---|
101 | XmlStreamReaderUtils.readStart(reader, ns, "Resource", true, true); |
---|
102 | final String pid = |
---|
103 | XmlStreamReaderUtils.readAttributeValue(reader, null, "pid"); |
---|
104 | final String ref = |
---|
105 | XmlStreamReaderUtils.readAttributeValue(reader, null, "ref"); |
---|
106 | XmlStreamReaderUtils.consumeStart(reader); |
---|
107 | |
---|
108 | // Resource/Resource (optional) |
---|
109 | if (XmlStreamReaderUtils.readStart(reader, ns, "Resource", false)) { |
---|
110 | logger.info("skipping nested <Resource> element"); |
---|
111 | XmlStreamReaderUtils.readEnd(reader, ns, "Resource", true); |
---|
112 | } |
---|
113 | |
---|
114 | // Resource/DataView |
---|
115 | final List<DataView> dataviews = parseDataViews(reader, ns); |
---|
116 | |
---|
117 | // Resource/ResourceFragment |
---|
118 | final List<Resource.ResourceFragment> resourceFragments = |
---|
119 | parseResourceFragments(reader, ns); |
---|
120 | |
---|
121 | XmlStreamReaderUtils.readEnd(reader, ns, "Resource", true); |
---|
122 | if (fullLegacyCompatMode) { |
---|
123 | return new LegacyClarinFCSRecordData(pid, ref, dataviews, |
---|
124 | resourceFragments); |
---|
125 | } else { |
---|
126 | return new ClarinFCSRecordData(pid, ref, dataviews, |
---|
127 | resourceFragments); |
---|
128 | } |
---|
129 | } |
---|
130 | |
---|
131 | |
---|
132 | private List<DataView> parseDataViews(XMLStreamReader reader, String ns) |
---|
133 | throws XMLStreamException, SRUClientException { |
---|
134 | List<DataView> dataviews = null; |
---|
135 | |
---|
136 | while (XmlStreamReaderUtils.readStart(reader, ns, "DataView", false, true)) { |
---|
137 | final String pid = |
---|
138 | XmlStreamReaderUtils.readAttributeValue(reader, null, "pid"); |
---|
139 | final String ref = |
---|
140 | XmlStreamReaderUtils.readAttributeValue(reader, null, "ref"); |
---|
141 | final String type = |
---|
142 | XmlStreamReaderUtils.readAttributeValue(reader, null, "type"); |
---|
143 | if ((type == null) || type.isEmpty()) { |
---|
144 | throw new SRUClientException("element <DataView> needs a " |
---|
145 | + "non-empty 'type' attribute"); |
---|
146 | } |
---|
147 | |
---|
148 | // consume start element and get rid of any whitespace |
---|
149 | XmlStreamReaderUtils.consumeStart(reader); |
---|
150 | XmlStreamReaderUtils.consumeWhitespace(reader); |
---|
151 | |
---|
152 | logger.debug("processing <DataView> of type = {}", type); |
---|
153 | |
---|
154 | DataViewParser selectedParser = null; |
---|
155 | for (DataViewParser parser : parsers) { |
---|
156 | if (parser.acceptType(type) && |
---|
157 | ((selectedParser == null) || |
---|
158 | (selectedParser.getPriority() < parser.getPriority()))) { |
---|
159 | selectedParser = parser; |
---|
160 | } |
---|
161 | } |
---|
162 | |
---|
163 | DataView dataview = null; |
---|
164 | if (selectedParser != null) { |
---|
165 | dataview = selectedParser.parse(reader, type, pid, ref); |
---|
166 | } else { |
---|
167 | logger.warn("no parser found for <DataView> of type = {}", type); |
---|
168 | } |
---|
169 | |
---|
170 | XmlStreamReaderUtils.readEnd(reader, ns, "DataView", true); |
---|
171 | |
---|
172 | if (dataview != null) { |
---|
173 | if (dataviews == null) { |
---|
174 | dataviews = new LinkedList<DataView>(); |
---|
175 | } |
---|
176 | dataviews.add(dataview); |
---|
177 | } else { |
---|
178 | logger.warn("skipped <DataView> of type = {}", type); |
---|
179 | } |
---|
180 | } // while |
---|
181 | return dataviews; |
---|
182 | } |
---|
183 | |
---|
184 | |
---|
185 | private List<Resource.ResourceFragment> parseResourceFragments( |
---|
186 | XMLStreamReader reader, String ns) throws XMLStreamException, |
---|
187 | SRUClientException { |
---|
188 | List<Resource.ResourceFragment> resourceFragments = null; |
---|
189 | while (XmlStreamReaderUtils.readStart(reader, ns, "ResourceFragment", false, true)) { |
---|
190 | logger.debug("found ResourceFragment"); |
---|
191 | String pid = XmlStreamReaderUtils.readAttributeValue(reader, null, "pid"); |
---|
192 | String ref = XmlStreamReaderUtils.readAttributeValue(reader, null, "ref"); |
---|
193 | XmlStreamReaderUtils.consumeStart(reader); |
---|
194 | final List<DataView> dataviews = parseDataViews(reader, ns); |
---|
195 | XmlStreamReaderUtils.readEnd(reader, ns, "ResourceFragment", true); |
---|
196 | |
---|
197 | if (resourceFragments == null) { |
---|
198 | resourceFragments = new LinkedList<Resource.ResourceFragment>(); |
---|
199 | } |
---|
200 | resourceFragments.add(new Resource.ResourceFragment(pid, ref, dataviews)); |
---|
201 | } // while |
---|
202 | return resourceFragments; |
---|
203 | } |
---|
204 | |
---|
205 | } // class AbstractClarinFCSRecordDataParser |
---|