1 | package eu.clarin.sru.server.fcs.utils; |
---|
2 | |
---|
3 | import java.io.IOException; |
---|
4 | import java.net.URL; |
---|
5 | import java.util.ArrayList; |
---|
6 | import java.util.Arrays; |
---|
7 | import java.util.HashMap; |
---|
8 | import java.util.HashSet; |
---|
9 | import java.util.Iterator; |
---|
10 | import java.util.LinkedList; |
---|
11 | import java.util.List; |
---|
12 | import java.util.Map; |
---|
13 | import java.util.Set; |
---|
14 | |
---|
15 | import javax.xml.XMLConstants; |
---|
16 | import javax.xml.namespace.NamespaceContext; |
---|
17 | import javax.xml.parsers.DocumentBuilder; |
---|
18 | import javax.xml.parsers.DocumentBuilderFactory; |
---|
19 | import javax.xml.parsers.ParserConfigurationException; |
---|
20 | import javax.xml.xpath.XPath; |
---|
21 | import javax.xml.xpath.XPathConstants; |
---|
22 | import javax.xml.xpath.XPathExpression; |
---|
23 | import javax.xml.xpath.XPathExpressionException; |
---|
24 | import javax.xml.xpath.XPathFactory; |
---|
25 | |
---|
26 | import org.slf4j.Logger; |
---|
27 | import org.slf4j.LoggerFactory; |
---|
28 | import org.w3c.dom.Document; |
---|
29 | import org.w3c.dom.Element; |
---|
30 | import org.w3c.dom.Node; |
---|
31 | import org.w3c.dom.NodeList; |
---|
32 | import org.xml.sax.SAXException; |
---|
33 | |
---|
34 | import eu.clarin.sru.server.SRUConfigException; |
---|
35 | import eu.clarin.sru.server.fcs.ResourceInfoInventory; |
---|
36 | import eu.clarin.sru.server.fcs.ResourceInfo; |
---|
37 | |
---|
38 | |
---|
39 | /** |
---|
40 | * A parser, that parses an XML file and produces a static list of resource info |
---|
41 | * records. The resulting list can be used to construct a |
---|
42 | * {@link SimpleResourceInfoInventory} instance. |
---|
43 | * |
---|
44 | * @see ResourceInfo |
---|
45 | * @see SimpleResourceInfoInventory |
---|
46 | */ |
---|
47 | public class SimpleResourceInfoInventoryParser { |
---|
48 | private static final String NS = "http://clarin.eu/fcs/endpoint-description"; |
---|
49 | private static final String LANG_EN = "en"; |
---|
50 | private static final Logger logger = |
---|
51 | LoggerFactory.getLogger(SimpleResourceInfoInventoryParser.class); |
---|
52 | |
---|
53 | |
---|
54 | /** |
---|
55 | * Parse an XML file and return a static list of resource info records. |
---|
56 | * |
---|
57 | * @param url |
---|
58 | * the URI pointing to the file to be parsed |
---|
59 | * @return a list of resource info records represented as |
---|
60 | * {@link ResourceInfo} instances |
---|
61 | * @throws SRUConfigException |
---|
62 | * if an error occurred |
---|
63 | */ |
---|
64 | public static ResourceInfoInventory parse(URL url) throws SRUConfigException { |
---|
65 | if (url == null) { |
---|
66 | throw new NullPointerException("url == null"); |
---|
67 | } |
---|
68 | |
---|
69 | logger.debug("parsing resource-info from: {}", url); |
---|
70 | |
---|
71 | final Set<String> ids = new HashSet<String>(); |
---|
72 | try { |
---|
73 | DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance(); |
---|
74 | dbf.setNamespaceAware(true); |
---|
75 | dbf.setCoalescing(true); |
---|
76 | DocumentBuilder db = dbf.newDocumentBuilder(); |
---|
77 | Document doc = db.parse(url.openStream()); |
---|
78 | |
---|
79 | XPathFactory factory = XPathFactory.newInstance(); |
---|
80 | XPath xpath = factory.newXPath(); |
---|
81 | xpath.setNamespaceContext(new NamespaceContext() { |
---|
82 | @Override |
---|
83 | public Iterator<?> getPrefixes(String namespaceURI) { |
---|
84 | throw new UnsupportedOperationException(); |
---|
85 | } |
---|
86 | |
---|
87 | @Override |
---|
88 | public String getPrefix(String namespaceURI) { |
---|
89 | throw new UnsupportedOperationException(); |
---|
90 | } |
---|
91 | |
---|
92 | @Override |
---|
93 | public String getNamespaceURI(String prefix) { |
---|
94 | if (prefix == null) { |
---|
95 | throw new NullPointerException("prefix == null"); |
---|
96 | } |
---|
97 | if (prefix.equals("ri")) { |
---|
98 | return NS; |
---|
99 | } else if (prefix.equals(XMLConstants.XML_NS_PREFIX)) { |
---|
100 | return XMLConstants.XML_NS_URI; |
---|
101 | } else { |
---|
102 | return XMLConstants.NULL_NS_URI; |
---|
103 | } |
---|
104 | } |
---|
105 | }); |
---|
106 | XPathExpression expression = |
---|
107 | xpath.compile("/ri:Resources/ri:Resource"); |
---|
108 | NodeList list = |
---|
109 | (NodeList) expression.evaluate(doc, XPathConstants.NODESET); |
---|
110 | |
---|
111 | List<ResourceInfo> entries = parseResourceInfo(xpath, list, ids); |
---|
112 | return new SimpleResourceInfoInventory(entries, false); |
---|
113 | } catch (ParserConfigurationException e) { |
---|
114 | e.printStackTrace(); |
---|
115 | } catch (SAXException e) { |
---|
116 | e.printStackTrace(); |
---|
117 | } catch (IOException e) { |
---|
118 | e.printStackTrace(); |
---|
119 | } catch (XPathExpressionException e) { |
---|
120 | e.printStackTrace(); |
---|
121 | } |
---|
122 | return null; |
---|
123 | } |
---|
124 | |
---|
125 | |
---|
126 | private static List<ResourceInfo> parseResourceInfo(XPath xpath, |
---|
127 | NodeList nodes, Set<String> ids) throws SRUConfigException, |
---|
128 | XPathExpressionException { |
---|
129 | logger.debug("parsing 'ResourceInfo' ({} nodes) ...", |
---|
130 | nodes.getLength()); |
---|
131 | |
---|
132 | List<ResourceInfo> ris = null; |
---|
133 | for (int k = 0; k < nodes.getLength(); k++) { |
---|
134 | final Element node = (Element) nodes.item(k); |
---|
135 | String pid = null; |
---|
136 | int resourceCount = -1; |
---|
137 | Map<String, String> titles = null; |
---|
138 | Map<String, String> descrs = null; |
---|
139 | String link = null; |
---|
140 | List<String> langs = null; |
---|
141 | String[] availableDataViews = null; |
---|
142 | List<ResourceInfo> sub = null; |
---|
143 | |
---|
144 | pid = node.getAttribute("pid"); |
---|
145 | if (pid != null) { |
---|
146 | pid = pid.trim(); |
---|
147 | if (pid.isEmpty()) { |
---|
148 | pid = null; |
---|
149 | } |
---|
150 | } |
---|
151 | if (pid == null) { |
---|
152 | throw new SRUConfigException("Element <ResourceInfo> " + |
---|
153 | "must carry a proper 'pid' attribute"); |
---|
154 | } |
---|
155 | if (ids.contains(pid)) { |
---|
156 | throw new SRUConfigException("Another element <ResourceInfo> " + |
---|
157 | "with pid '" + pid + "' already exists"); |
---|
158 | } |
---|
159 | ids.add(pid); |
---|
160 | |
---|
161 | XPathExpression x1 = xpath.compile("ri:Title"); |
---|
162 | NodeList l1 = (NodeList) x1.evaluate(node, XPathConstants.NODESET); |
---|
163 | if (l1 != null) { |
---|
164 | for (int i = 0; i < l1.getLength(); i++) { |
---|
165 | final Element n = (Element) l1.item(i); |
---|
166 | |
---|
167 | final String lang = getLangAttribute(n); |
---|
168 | if (lang == null) { |
---|
169 | throw new SRUConfigException("Element <Title> must " + |
---|
170 | "carry a proper 'xml:lang' attribute"); |
---|
171 | } |
---|
172 | |
---|
173 | final String title = cleanString(n.getTextContent()); |
---|
174 | if (title == null) { |
---|
175 | throw new SRUConfigException("Element <Title> must " + |
---|
176 | "carry a non-empty 'xml:lang' attribute"); |
---|
177 | } |
---|
178 | |
---|
179 | if (titles == null) { |
---|
180 | titles = new HashMap<String, String>(); |
---|
181 | } |
---|
182 | if (titles.containsKey(lang)) { |
---|
183 | logger.warn("title with language '{}' already exists", |
---|
184 | lang); |
---|
185 | } else { |
---|
186 | logger.debug("title: '{}' '{}'", lang, title); |
---|
187 | titles.put(lang, title); |
---|
188 | } |
---|
189 | } |
---|
190 | if ((titles != null) && !titles.containsKey(LANG_EN)) { |
---|
191 | throw new SRUConfigException( |
---|
192 | "A <Title> with language 'en' is mandatory"); |
---|
193 | } |
---|
194 | } |
---|
195 | XPathExpression x2 = xpath.compile("ri:Description"); |
---|
196 | NodeList l2 = (NodeList) x2.evaluate(node, XPathConstants.NODESET); |
---|
197 | if (l2 != null) { |
---|
198 | for (int i = 0; i < l2.getLength(); i++) { |
---|
199 | Element n = (Element) l2.item(i); |
---|
200 | |
---|
201 | String lang = getLangAttribute(n); |
---|
202 | if (lang == null) { |
---|
203 | throw new SRUConfigException("Element <Description> " + |
---|
204 | "must carry a proper 'xml:lang' attribute"); |
---|
205 | |
---|
206 | } |
---|
207 | String desc = cleanString(n.getTextContent()); |
---|
208 | |
---|
209 | if (descrs == null) { |
---|
210 | descrs = new HashMap<String, String>(); |
---|
211 | } |
---|
212 | |
---|
213 | if (descrs.containsKey(lang)) { |
---|
214 | logger.warn("description with language '{}' " |
---|
215 | + "already exists", lang); |
---|
216 | } else { |
---|
217 | logger.debug("description: '{}' '{}'", lang, desc); |
---|
218 | descrs.put(lang, desc); |
---|
219 | } |
---|
220 | } |
---|
221 | if ((descrs != null) && !descrs.containsKey(LANG_EN)) { |
---|
222 | throw new SRUConfigException( |
---|
223 | "A <Description> with language 'en' is mandatory"); |
---|
224 | } |
---|
225 | } |
---|
226 | |
---|
227 | XPathExpression x3 = xpath.compile("ri:LandingPageURI"); |
---|
228 | NodeList l3 = (NodeList) x3.evaluate(node, XPathConstants.NODESET); |
---|
229 | if (l3 != null) { |
---|
230 | for (int i = 0; i < l3.getLength(); i++) { |
---|
231 | Element n = (Element) l3.item(i); |
---|
232 | link = cleanString(n.getTextContent()); |
---|
233 | logger.debug("link: \"{}\"", n.getTextContent()); |
---|
234 | } |
---|
235 | } |
---|
236 | |
---|
237 | XPathExpression x4 = xpath.compile("ri:Languages/ri:Language"); |
---|
238 | NodeList l4 = (NodeList) x4.evaluate(node, XPathConstants.NODESET); |
---|
239 | if (l4 != null) { |
---|
240 | for (int i = 0; i < l4.getLength(); i++) { |
---|
241 | Element n = (Element) l4.item(i); |
---|
242 | |
---|
243 | String s = n.getTextContent(); |
---|
244 | if (s != null) { |
---|
245 | s = s.trim(); |
---|
246 | if (s.isEmpty()) { |
---|
247 | s = null; |
---|
248 | } |
---|
249 | } |
---|
250 | |
---|
251 | /* |
---|
252 | * enforce three letter codes |
---|
253 | */ |
---|
254 | if ((s == null) || (s.length() != 3)) { |
---|
255 | throw new SRUConfigException("Element <Language> " + |
---|
256 | "must use ISO-632-3 three letter " + |
---|
257 | "language codes"); |
---|
258 | } |
---|
259 | |
---|
260 | if (langs == null) { |
---|
261 | langs = new ArrayList<String>(); |
---|
262 | } |
---|
263 | logger.debug("language: '{}'", n.getTextContent()); |
---|
264 | langs.add(s); |
---|
265 | } |
---|
266 | } |
---|
267 | |
---|
268 | XPathExpression x6 = xpath.compile("ri:AvailableDataViews/@ref"); |
---|
269 | String ref = (String) x6.evaluate(node, XPathConstants.STRING); |
---|
270 | |
---|
271 | if (ref == null || ref.isEmpty()){ |
---|
272 | throw new SRUConfigException("Element <AvailableDataViews> " + |
---|
273 | "must have a non-empty attribute ref."); |
---|
274 | } |
---|
275 | |
---|
276 | availableDataViews = ref.split("\\s+"); |
---|
277 | |
---|
278 | |
---|
279 | XPathExpression x5 = |
---|
280 | xpath.compile("ri:Resources/ri:Resource"); |
---|
281 | NodeList l5 = (NodeList) x5.evaluate(node, XPathConstants.NODESET); |
---|
282 | if ((l5 != null) && (l5.getLength() > 0)) { |
---|
283 | sub = parseResourceInfo(xpath, l5, ids); |
---|
284 | } |
---|
285 | |
---|
286 | if (ris == null) { |
---|
287 | ris = new LinkedList<ResourceInfo>(); |
---|
288 | } |
---|
289 | ris.add(new ResourceInfo(pid, resourceCount, titles, descrs, link, |
---|
290 | langs, Arrays.asList(availableDataViews),sub)); |
---|
291 | } |
---|
292 | return ris; |
---|
293 | } |
---|
294 | |
---|
295 | |
---|
296 | private static String getLangAttribute(Element el) { |
---|
297 | String lang = el.getAttributeNS(XMLConstants.XML_NS_URI, "lang"); |
---|
298 | if (lang != null) { |
---|
299 | lang = lang.trim(); |
---|
300 | if (!lang.isEmpty()) { |
---|
301 | return lang; |
---|
302 | } |
---|
303 | } |
---|
304 | return null; |
---|
305 | } |
---|
306 | |
---|
307 | |
---|
308 | private static String cleanString(String s) { |
---|
309 | if (s != null) { |
---|
310 | s = s.trim(); |
---|
311 | if (!s.isEmpty()) { |
---|
312 | StringBuilder sb = new StringBuilder(); |
---|
313 | for (String z : s.split("\\s*\\n+\\s*")) { |
---|
314 | z = z.trim(); |
---|
315 | if (!z.isEmpty()) { |
---|
316 | if (sb.length() > 0) { |
---|
317 | sb.append(' '); |
---|
318 | } |
---|
319 | sb.append(z); |
---|
320 | } |
---|
321 | } |
---|
322 | if (sb.length() > 0) { |
---|
323 | return sb.toString(); |
---|
324 | } |
---|
325 | } |
---|
326 | } |
---|
327 | return null; |
---|
328 | } |
---|
329 | |
---|
330 | } // class SimpleResourceInfoInventoryParser |
---|