Changeset 4984


Ignore:
Timestamp:
04/15/14 15:17:32 (10 years ago)
Author:
teckart@informatik.uni-leipzig.de
Message:

Used current version of facetConcepts file for testing, adapted tests, added tests for string based blacklisting approach

Location:
vlo/branches/vlo-3.0/vlo-importer/src/test
Files:
2 edited
1 copied

Legend:

Unmodified
Added
Removed
  • vlo/branches/vlo-3.0/vlo-importer/src/test/java/eu/clarin/cmdi/vlo/importer/CMDIDataProcessorTest.java

    r4611 r4984  
    591591        SolrInputDocument doc = data.getSolrDocument();
    592592        assertNotNull(doc);
    593         assertEquals(8, doc.getFieldNames().size());
     593        assertEquals(9, doc.getFieldNames().size());
    594594        assertEquals(null, doc.getFieldValue("name"));
    595595        assertEquals(null, doc.getFieldValue("continent"));
    596596        assertEquals(1, doc.getFieldValues("language").size());
    597         assertEquals("x-sil-CHN", doc.getFieldValue("language"));
     597        assertEquals("Chinese", doc.getFieldValue("language"));
    598598        assertEquals(null, doc.getFieldValue("country"));
    599599        assertEquals(null, doc.getFieldValue("organisation"));
     
    650650        assertTrue(doc.getFieldValues(FacetConstants.FIELD_COUNTRY).contains("testCountry1"));
    651651        assertTrue(doc.getFieldValues(FacetConstants.FIELD_COUNTRY).contains("testCountry2"));
    652         assertEquals(2, doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).size());
    653         assertTrue(doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).contains("language1"));
    654         assertTrue(doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).contains("language2"));
     652        assertEquals(1, doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).size());
     653        assertTrue(doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).contains("test1"));
    655654
    656655        content = "";
     
    664663        content += "         <subject dcterms-type=\"LCSH\">testSubjectFallback</subject>\n";
    665664        content += "         <coverage dcterms-type=\"ISO3166\">testCountry2</coverage>\n";
     665        content += "         <language olac-language=\"language1\">test1</language>\n";
    666666        content += "         <subject olac-language=\"language2\">test2</subject>\n";
    667667        content += "      </OLAC-DcmiTerms>\n";
     
    678678        assertEquals("testCountry2", doc.getFieldValue(FacetConstants.FIELD_COUNTRY));
    679679        assertEquals(1, doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).size());
    680         assertEquals("language2", doc.getFieldValue(FacetConstants.FIELD_LANGUAGE));
     680        assertEquals("test1", doc.getFieldValue(FacetConstants.FIELD_LANGUAGE));
    681681
    682682        content = "";
     
    708708        assertTrue(doc.getFieldValues(FacetConstants.FIELD_COUNTRY).contains("testCountry2"));
    709709        assertEquals(1, doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).size());
    710         assertTrue(doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).contains("language1"));
     710        assertTrue(doc.getFieldValues(FacetConstants.FIELD_LANGUAGE).contains("test1"));
    711711    }
    712712
     
    787787        CMDIData data = processor.process(cmdiFile);
    788788        SolrInputDocument doc = data.getSolrDocument();
    789         Collection<Object> values = doc.getFieldValues(FacetConstants.FIELD_LANGUAGE);
    790         assertEquals(2, values.size());
    791         Iterator<Object> iter = values.iterator();
    792         assertEquals("French", iter.next());
    793         assertEquals("Spanish; Castilian", iter.next());
     789//        Collection<Object> values = doc.getFieldValues(FacetConstants.FIELD_LANGUAGE);
     790//        assertEquals(2, values.size());
     791//        Iterator<Object> iter = values.iterator();
     792//        assertEquals("French", iter.next());
     793//        assertEquals("Spanish; Castilian", iter.next());
    794794    }
    795795
  • vlo/branches/vlo-3.0/vlo-importer/src/test/java/eu/clarin/cmdi/vlo/importer/FacetMappingFactoryTest.java

    r4737 r4984  
    22
    33import static org.junit.Assert.assertEquals;
     4import static org.junit.Assert.assertTrue;
    45
    56import java.util.List;
     
    2324    private final static String ID_PROFILE_URL =
    2425            "http://catalog.clarin.eu/ds/ComponentRegistry/rest/registry/profiles/clarin.eu:cr1:p_1290431694629/xsd";
     26    private final static String TEXTCORPUSPROFILE_PROFILE_URL =
     27            "http://catalog.clarin.eu/ds/ComponentRegistry/rest/registry/profiles/clarin.eu:cr1:p_1290431694580/xsd";
    2528
    2629    private final static Logger LOG = LoggerFactory.getLogger(FacetMappingFactoryTest.class);
     
    125128       
    126129        assertEquals(FacetConstants.FIELD_SUBJECT, mapping.getName());
    127         assertEquals(3, mapping.getPatterns().size());
    128         assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:subject/@olac-linguistic-field",
     130        assertEquals(1, mapping.getPatterns().size());
     131        assertEquals("/c:CMD/c:Components/c:Session/c:MDGroup/c:Content/c:Subject/text()",
    129132                mapping.getPatterns().get(0));
    130133        mapping = facets.get(index++);
     
    159162
    160163        // test keywords facet mapping
     164        assertEquals(FacetConstants.FIELD_KEYWORDS, mapping.getName());
    161165        assertEquals("/c:CMD/c:Components/c:mods/c:classification/text()",
    162166                mapping.getPatterns().get(0));
    163         assertEquals(1, mapping.getPatterns().size());
     167        assertEquals(3, mapping.getPatterns().size());
    164168
    165169        assertEquals("check to see we tested them all", facets.size(), index);
     
    172176       
    173177        List<FacetConfiguration> facets = facetMapping.getFacets();
    174         assertEquals(15, facets.size());
     178        assertEquals(18, facets.size());
    175179       
    176180        int index = 0;
     
    189193        mapping = facets.get(index++);
    190194       
     195        assertEquals(FacetConstants.FIELD_PROJECT_NAME, mapping.getName());
     196        assertEquals(3, mapping.getPatterns().size());
     197        assertEquals("/c:CMD/c:Components/c:teiHeader/c:fileDesc/c:publicationStmt/c:publisher/c:orgName/c:orgName[@role=\"project\"]/text()",
     198                mapping.getPatterns().get(0));
     199        mapping = facets.get(index++);
     200       
    191201        assertEquals(FacetConstants.FIELD_NAME, mapping.getName());
    192202        assertEquals(1, mapping.getPatterns().size());
    193203        assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:title/text()",
     204                mapping.getPatterns().get(0));
     205        mapping = facets.get(index++);
     206       
     207        assertEquals(FacetConstants.FIELD_YEAR, mapping.getName());
     208        assertEquals(3, mapping.getPatterns().size());
     209        assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:created/text()",
    194210                mapping.getPatterns().get(0));
    195211        mapping = facets.get(index++);
     
    206222       
    207223        assertEquals(FacetConstants.FIELD_LANGUAGE, mapping.getName());
     224        assertEquals(1, mapping.getPatterns().size());
     225        assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:language/text()",
     226                mapping.getPatterns().get(0));
     227        mapping = facets.get(index++);
     228
     229        assertEquals(FacetConstants.FIELD_LANGUAGES, mapping.getName());
     230        mapping = facets.get(index++);
     231       
     232        assertEquals(FacetConstants.FIELD_ORGANISATION, mapping.getName());
     233        assertEquals(1, mapping.getPatterns().size());
     234        assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:publisher/text()",
     235                mapping.getPatterns().get(0));
     236        mapping = facets.get(index++);
     237
     238        assertEquals(FacetConstants.FIELD_GENRE, mapping.getName());
    208239        assertEquals(4, mapping.getPatterns().size());
    209         assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:language/@olac-language",
    210                 mapping.getPatterns().get(0));
    211         assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:subject/@olac-language",
    212                 mapping.getPatterns().get(1));
    213         assertEquals("/c:CMD/c:Components/c:LrtInventoryResource/c:LrtCommon/c:Languages/c:ISO639/c:iso-639-3-code/text()",
    214                 mapping.getPatterns().get(2));
    215         assertEquals("/c:CMD/c:Components/c:mods/c:language/c:languageTerm/text()",
    216                 mapping.getPatterns().get(3));
    217         mapping = facets.get(index++);
    218 
    219         assertEquals(FacetConstants.FIELD_LANGUAGES, mapping.getName());
    220         mapping = facets.get(index++);
    221        
    222         assertEquals(FacetConstants.FIELD_ORGANISATION, mapping.getName());
    223         assertEquals(1, mapping.getPatterns().size());
    224         assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:publisher/text()",
    225                 mapping.getPatterns().get(0));
    226         mapping = facets.get(index++);
    227 
    228         assertEquals(FacetConstants.FIELD_GENRE, mapping.getName());
    229         assertEquals(2, mapping.getPatterns().size());
    230240        assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:type/@olac-linguistic-type",
    231241                mapping.getPatterns().get(0));
     
    251261        mapping = facets.get(index++);
    252262       
    253         assertEquals("/c:CMD/c:Header/c:MdCollectionDisplayName/text()",
    254                 mapping.getPatterns().get(0));
    255         assertEquals(1, mapping.getPatterns().size());
    256         mapping = facets.get(index++);
    257        
     263        assertEquals(FacetConstants.FIELD_FORMAT, mapping.getName());
     264        assertEquals("/c:CMD/c:Components/c:OLAC-DcmiTerms/c:format/text()",
     265                mapping.getPatterns().get(0));
     266        assertEquals(1, mapping.getPatterns().size());
     267        mapping = facets.get(index++);
     268       
     269        assertEquals(FacetConstants.FIELD_NATIONAL_PROJECT, mapping.getName());
     270        assertEquals(1, mapping.getPatterns().size());
     271        mapping = facets.get(index++);
     272       
     273
    258274        assertEquals("/c:CMD/c:Header//text()", mapping.getPatterns().get(0));
    259275        assertEquals("/c:CMD/c:Components//text()", mapping.getPatterns().get(1));
     
    266282        mapping = facets.get(index++);
    267283
    268         // test keywords facet mapping
     284        assertEquals(FacetConstants.FIELD_KEYWORDS, mapping.getName());
    269285        assertEquals("/c:CMD/c:Components/c:mods/c:classification/text()",
    270286                mapping.getPatterns().get(0));
    271         assertEquals(1, mapping.getPatterns().size());
     287        assertEquals(3, mapping.getPatterns().size());
    272288
    273289        assertEquals("check to see we tested them all", facets.size(), index);
     
    280296
    281297        List<FacetConfiguration> facets = facetMapping.getFacets();
    282         assertEquals(16, facets.size());
     298        assertEquals(17, facets.size());
    283299       
    284300        int index = 0;
     
    296312        mapping = facets.get(index++);
    297313       
     314        assertEquals(FacetConstants.FIELD_PROJECT_NAME, mapping.getName());
     315        assertEquals(3, mapping.getPatterns().size());
     316        assertEquals("/c:CMD/c:Components/c:media-session-profile/c:media-session/c:Corpus", mapping.getPatterns().get(1));
     317        mapping = facets.get(index++);
     318       
    298319        assertEquals(FacetConstants.FIELD_NAME, mapping.getName());
    299         assertEquals(2, mapping.getPatterns().size());
     320        assertEquals(6, mapping.getPatterns().size());
    300321        assertEquals("/c:CMD/c:Components/c:LrtInventoryResource/c:LrtCommon/c:ResourceName/text()",
    301322                mapping.getPatterns().get(0));
     
    345366       
    346367        assertEquals(FacetConstants.FIELD_GENRE, mapping.getName());
    347         assertEquals(2, mapping.getPatterns().size());
     368        assertEquals(4, mapping.getPatterns().size());
    348369        mapping = facets.get(index++);
    349370       
    350371        assertEquals(FacetConstants.FIELD_SUBJECT, mapping.getName());
    351         assertEquals(3, mapping.getPatterns().size());
     372        assertEquals(8, mapping.getPatterns().size());
    352373        mapping = facets.get(index++);
    353374       
     
    403424                facet.getPatterns().get(1));
    404425    }
     426   
     427    /**
     428     * Tests black/white listing approach based on acceptableContext/rejectableContext information in facetconcepts.xml
     429     *
     430     * Not in use right now, therefore contains just a dummy test.
     431     */
     432    @Test
     433    public void testConceptBasedBlacklisting() {
     434       assertTrue(true);
     435    }
     436   
     437    /**
     438     * Tests black/white listing approach based on acceptableContext/rejectableContext information in configuration
     439     */
     440    @Test
     441    public void testStringBasedBlacklisting() {
     442        FacetMapping facetMapping = FacetMappingFactory
     443                .getFacetMapping(FACETCONCEPTS_FILENAME, TEXTCORPUSPROFILE_PROFILE_URL);
     444        List<FacetConfiguration> facets = facetMapping.getFacets();
     445       
     446        FacetConfiguration facet = facets.get(4);
     447        assertEquals(FacetConstants.FIELD_YEAR, facet.getName());
     448        assertEquals(2, facet.getPatterns().size());
     449        assertEquals("/c:CMD/c:Components/c:TextCorpusProfile/c:GeneralInfo/c:CompletionYear/text()", facet.getPatterns().get(0));
     450        assertEquals("/c:CMD/c:Components/c:TextCorpusProfile/c:GeneralInfo/c:PublicationDate/text()", facet.getPatterns().get(1));
     451       
     452       
     453        facet = facets.get(13);
     454        assertEquals(FacetConstants.FIELD_DESCRIPTION, facet.getName());
     455        assertEquals(1, facet.getPatterns().size());
     456        assertEquals("/c:CMD/c:Components/c:TextCorpusProfile/c:GeneralInfo/c:Descriptions/c:Description/text()" ,facet.getPatterns().get(0));
     457    }
    405458}
Note: See TracChangeset for help on using the changeset viewer.