source: vlo/trunk/vlo_webapp/src/main/resources/importerConfig.xml @ 994

Last change on this file since 994 was 994, checked in by patdui, 13 years ago
  • fixed property name
  • Property svn:executable set to *
File size: 15.3 KB
Line 
1<?xml version="1.0" encoding="UTF-8"?>
2<beans xmlns="http://www.springframework.org/schema/beans" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
3  xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-3.0.xsd">
4
5  <bean id="importerConfig" class="eu.clarin.cmdi.vlo.importer.ImporterConfig">
6    <!--    <property name="deleteFirst" value="true"/>-->
7    <property name="dataRoots">
8      <list>
9        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
10          <property name="originName" value="OLAC Metadata Providers" />
11          <property name="rootFile"
12            value="/Users/patdui/data/snapshots2/olac-cmdi-20101105/_corpusstructure/collection_root.cmdi" />
13          <property name="facetMapping" ref="olacMapping"></property>
14        </bean>
15        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
16          <property name="originName" value="Nijmegen corpora of casual speech" />
17          <property name="rootFile"
18
19            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/casual_speech/Corpusstructure/casual_speech.imdi.cmdi" />
20          <property name="facetMapping" ref="imdiMapping"></property>
21        </bean>
22        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
23          <property name="originName" value="MPI corpora" />
24          <property name="rootFile"
25            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/Corpusstructure/MPI.imdi.cmdi" />
26          <property name="facetMapping" ref="imdiMapping"></property>
27        </bean>
28        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
29          <property name="originName" value="CORP-ORAL" />
30          <property name="rootFile"
31            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/CORP_ORAL/Corpusstructure/CORP_ORAL.imdi.cmdi" />
32          <property name="facetMapping" ref="imdiMapping"></property>
33        </bean>
34        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
35          <property name="originName" value="DoBeS archive" />
36          <property name="rootFile"
37            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/dobes_data/Corpusstructure/dobes.imdi.cmdi" />
38          <property name="facetMapping" ref="imdiMapping"></property>
39        </bean>
40        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
41          <property name="originName" value="ECHO" />
42          <property name="rootFile"
43            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/echo_data/Corpusstructure/echo.imdi.cmdi" />
44          <property name="facetMapping" ref="imdiMapping"></property>
45        </bean>
46        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
47          <property name="originName" value="DBD" />
48          <property name="rootFile"
49            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/dbd_data/Corpusstructure/dbd.imdi.cmdi" />
50          <property name="facetMapping" ref="imdiMapping"></property>
51        </bean>
52        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
53          <property name="originName" value="Sign Language" />
54          <property name="rootFile"
55            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/Corpusstructure/sign_language.imdi.cmdi" />
56          <property name="facetMapping" ref="imdiMapping"></property>
57        </bean>
58        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
59          <property name="originName" value="Endagered Languages" />
60          <property name="rootFile"
61            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/Corpusstructure/endangered_languages.imdi.cmdi" />
62          <property name="facetMapping" ref="imdiMapping"></property>
63        </bean>
64        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
65          <property name="originName" value="ANDES" />
66          <property name="rootFile"
67            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/ANDES_data/Corpusstructure/ANDES.imdi.cmdi" />
68          <property name="facetMapping" ref="imdiMapping"></property>
69        </bean>
70        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
71          <property name="originName" value="Leiden Archives" />
72          <property name="rootFile"
73            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/LeidenArchives/Corpusstructure/LeidenArchives.imdi.cmdi" />
74          <property name="facetMapping" ref="imdiMapping"></property>
75        </bean>
76        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
77          <property name="originName" value="ILSP INTERA Contribution" />
78          <property name="rootFile"
79            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/ilsp_data/ILSP_INTERA.imdi.cmdi" />
80          <property name="facetMapping" ref="imdiMapping"></property>
81        </bean>
82        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
83          <property name="originName" value="MPI fÃŒr Bildungsforschung" />
84          <property name="rootFile"
85            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/Bildungsforschung/Corpusstructure/MPI_Bildungsforschung.imdi.cmdi" />
86          <property name="facetMapping" ref="imdiMapping"></property>
87        </bean>
88        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
89          <property name="originName" value="Humanethologisches Filmarchiv" />
90          <property name="rootFile"
91            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/humanethology/Corpusstructure/humanethology.imdi.cmdi" />
92          <property name="facetMapping" ref="imdiMapping"></property>
93        </bean>
94        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
95          <property name="originName" value="SUCA" />
96          <property name="rootFile"
97            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/suca_data/Corpusstructure/suca.imdi.cmdi" />
98          <property name="facetMapping" ref="imdiMapping"></property>
99        </bean>
100        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
101          <property name="originName" value="Nijmegen corpora of casual speech" />
102          <property name="rootFile"
103            value="/Users/patdui/data/snapshots2/data/corpora/qfs1/media-archive/casual_speech/Corpusstructure/casual_speech.imdi.cmdi" />
104          <property name="facetMapping" ref="imdiMapping"></property>
105        </bean>
106        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
107          <property name="originName" value="ESF corpus" />
108          <property name="rootFile" value="/Users/patdui/data/snapshots2/data/corpora/esf_conv/Corpusstructure/esf.imdi.cmdi" />
109          <property name="facetMapping" ref="imdiMapping"></property>
110        </bean>
111        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
112          <property name="originName" value="IFA corpus" />
113          <property name="rootFile" value="/Users/patdui/data/snapshots2/data/corpora/IFAcorpus/IMDI/IFAcorpus.imdi.cmdi" />
114          <property name="facetMapping" ref="imdiMapping"></property>
115        </bean>
116        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
117          <property name="originName" value="CGN corpus" />
118          <property name="rootFile"
119            value="/Users/patdui/data/snapshots2/data/corpora/CGN/COREX6/data/meta/imdi_3.0_eaf/corpora/cgn.imdi.cmdi" />
120          <property name="facetMapping" ref="imdiMapping"></property>
121        </bean>
122        <bean class="eu.clarin.cmdi.vlo.importer.DataRoot">
123          <property name="originName" value="CLARIN LRT" />
124          <property name="rootFile"
125            value="/Users/patdui/data/snapshots2/lrt-20101117/_corpusstructure/collection_lrt_inventory.cmdi" />
126          <property name="facetMapping" ref="lrtMapping"></property>
127        </bean>
128      </list>
129    </property>
130  </bean>
131
132  <!--
133    Mapping of facets to XPATH expressions. The facets should correspond to the facets described in the
134    solr/conf/schema.xml file of the SOLR Server installation
135  -->
136
137  <bean id="imdiMapping" class="eu.clarin.cmdi.vlo.importer.FacetMapping"> <!-- add year? -->
138    <property name="idMapping" value="CMD/Header/MdSelfLink/text()" />
139    <property name="facets">
140      <list>
141        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
142          <property name="name" value="name" />
143          <property name="pattern" value="CMD/Components/Session/Name/text()" />
144        </bean>
145        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
146          <property name="name" value="year" />
147          <property name="pattern" value="CMD/Components/Session/Date/text()" />
148        </bean>
149        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
150          <property name="name" value="continent" />
151          <property name="pattern" value="CMD/Components/Session/MDGroup/Location/Continent/text()" />
152        </bean>
153        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
154          <property name="name" value="country" />
155          <property name="pattern" value="CMD/Components/Session/MDGroup/Location/Country/text()" />
156        </bean>
157        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
158          <property name="name" value="language" />
159          <property name="pattern" value="CMD/Components/Session/MDGroup/Content/Content_Languages/Content_Language/Id/text()" />
160        </bean>
161        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
162          <property name="name" value="organisation" />
163          <property name="pattern" value="CMD/Components/Session/MDGroup/Project/Contact/Organisation/text()" />
164        </bean>
165        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
166          <property name="name" value="genre" />
167          <property name="pattern" value="CMD/Components/Session/MDGroup/Content/Genre/text()" />
168          <property name="caseInsensitive" value="true" />
169        </bean>
170        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
171          <property name="name" value="subject" />
172          <property name="pattern" value="CMD/Components/Session/MDGroup/Content/Subject/text()" />
173          <property name="caseInsensitive" value="true" />
174        </bean>
175        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
176          <property name="name" value="description" />
177          <property name="pattern" value="CMD/Components/Session/descriptions/Description/text()" />
178        </bean>
179      </list>
180    </property>
181  </bean>
182
183
184  <!--
185    /CMD/Components/OLAC-DcmiTerms/identifier (if starting with http://) -> open in original context (now: IMDI browser)
186  -->
187
188  <bean id="olacMapping" class="eu.clarin.cmdi.vlo.importer.FacetMapping">
189    <property name="idMapping" value="CMD/Header/MdSelfLink/text()" /> <!-- And some other example see http://trac.clarin.eu/wiki/CmdiVirtualLanguageObservatory -->
190    <property name="facets">
191      <list>
192        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
193          <property name="name" value="name" />
194          <property name="pattern" value="CMD/Components/OLAC-DcmiTerms/title/text()" />
195        </bean>
196        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
197          <property name="name" value="country" />
198          <property name="pattern" value="CMD/Components/OLAC-DcmiTerms/spatial/text()" />
199          <!--
200            /CMD/Components/OLAC-DcmiTerms/spatial[@dcterms-type="ISO3166"] -> country
201            /CMD/Components/OLAC-DcmiTerms/coverage[@dcterms-type="ISO3166"] -> country
202          -->
203        </bean>
204        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
205          <property name="name" value="language" />
206          <property name="pattern" value="CMD/Components/OLAC-DcmiTerms/language/@olac-language" />
207        </bean>
208        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
209          <property name="name" value="organisation" />
210          <property name="pattern" value="CMD/Components/OLAC-DcmiTerms/publisher/text()" />
211        </bean>
212        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
213          <property name="name" value="genre" />
214          <property name="pattern" value="CMD/Components/OLAC-DcmiTerms/type/@olac-linguistic-type" />
215          <property name="caseInsensitive" value="true" />
216        </bean>
217        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
218          <property name="name" value="description" />
219          <property name="pattern" value="CMD/Components/OLAC-DcmiTerms/description/text()" />
220        </bean>
221        <!--
222          <entry key="continent"> <value>CMD/Components/olac</value> </entry>-->
223        <!--        <entry key="subject">-->
224        <!--          <value>CMD/Components/OLAC-DcmiTerms/subject</value> -->
225        <!-- [@dcterms-type="LCSH"]  -->
226        <!--        </entry>-->
227      </list>
228    </property>
229  </bean>
230
231
232
233
234
235  <!--
236    /CMD/Components/LrtInventoryResource/LrtCommon/MetadataLink (if not existing: ReferenceLink?) -> open in original
237    context
238  -->
239  <!--/CMD/Components/LrtInventoryResource/LrtCommon/ResourceType -> resource type-->
240  <!--
241    /CMD/Components/LrtInventoryResource/LrtCommon/Format-> format Resource type and format need to talk to Dieter what
242    will the implications be? Also need to think about a way to get language and country codes, human readable?
243  -->
244
245  <bean id="lrtMapping" class="eu.clarin.cmdi.vlo.importer.FacetMapping">
246    <property name="idMapping" value="CMD/Header/MdSelfLink/text()" />
247    <property name="facets">
248      <list>
249        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
250          <property name="name" value="name" />
251          <property name="pattern" value="CMD/Components/LrtInventoryResource/LrtCommon/ResourceName/text()" />
252        </bean>
253        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
254          <property name="name" value="country" />
255          <property name="pattern" value="CMD/Components/LrtInventoryResource/LrtCommon/Countries/Country/code/text()" />
256        </bean>
257        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
258          <property name="name" value="language" />
259          <property name="pattern" value="CMD/Components/LrtInventoryResource/LrtCommon/Languages/ISO639/iso-639-3-code/text()" />
260        </bean>
261        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
262          <property name="name" value="organisation" />
263          <property name="pattern" value="CMD/Components/LrtInventoryResource/LrtCommon/Institute/text()" />
264        </bean>
265        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
266          <property name="name" value="year" />
267          <property name="pattern"
268            value="CMD/Components/LrtInventoryResource/LrtCommon/FinalizationYearResourceCreation/text()" />
269        </bean>
270        <bean class="eu.clarin.cmdi.vlo.importer.FacetConfiguration">
271          <property name="name" value="description" />
272          <property name="pattern" value="CMD/Components/LrtInventoryResource/LrtCommon/Description/text()" />
273        </bean>
274        <!--        <entry key="continent">-->
275        <!--          <value></value>-->
276        <!--        </entry>-->
277        <!--        <entry key="genre">-->
278        <!--          <value></value>-->
279        <!--        </entry>-->
280        <!--        <entry key="subject">-->
281        <!--          <value></value>-->
282        <!--        </entry>-->
283      </list>
284    </property>
285  </bean>
286
287</beans>
288
Note: See TracBrowser for help on using the repository browser.