Changeset 989


Ignore:
Timestamp:
12/14/10 10:32:42 (13 years ago)
Author:
paucas
Message:
  • implemented lowercasing
Location:
vlo/trunk/vlo_webapp/src
Files:
3 edited

Legend:

Unmodified
Added
Removed
  • vlo/trunk/vlo_webapp/src/main/java/eu/clarin/cmdi/vlo/importer/CMDIData.java

    r801 r989  
    1717    }
    1818
    19     public void addDocField(String name, String value) {
     19    public void addDocField(String name, String value, boolean caseSensitive) {
    2020        if (doc == null) {
    2121            doc = new SolrInputDocument();
    2222        }
    2323        if (value != null && !value.isEmpty()) {
    24             doc.addField(name, value);
     24            if (caseSensitive) {
     25                doc.addField(name, value.toLowerCase());
     26            } else {
     27                doc.addField(name, value);
     28            }
    2529        }
    2630    }
     
    4347        return id;
    4448    }
    45 
    4649}
  • vlo/trunk/vlo_webapp/src/main/java/eu/clarin/cmdi/vlo/importer/CMDIDigester.java

    r988 r989  
    7878        List<FacetConfiguration> facetList = facetMapping.getFacets();
    7979        for (FacetConfiguration facetConfiguration : facetList) {
    80             matchDocumentField(result, facetConfiguration.getPattern(), facetConfiguration.getName(), doc, xpath);
     80            matchDocumentField(result, facetConfiguration, doc, xpath);
    8181        }
    8282        return result;
    8383    }
    8484
    85     private void matchDocumentField(CMDIData result, String pattern, String fieldName, Document doc, XPath xpath)
     85    private void matchDocumentField(CMDIData result, FacetConfiguration facetConfig, Document doc, XPath xpath)
    8686            throws XPathExpressionException {
    87         NodeList nodes = (NodeList) xpath.evaluate(pattern, doc, XPathConstants.NODESET);
     87        NodeList nodes = (NodeList) xpath.evaluate(facetConfig.getPattern(), doc, XPathConstants.NODESET);
    8888        if (nodes != null) {
    8989            for (int i = 0; i < nodes.getLength(); i++) {
    90                 result.addDocField(fieldName, nodes.item(i).getNodeValue());
     90                result.addDocField(facetConfig.getName(), nodes.item(i).getNodeValue(), facetConfig.isCaseSensitive());
    9191            }
    9292        } // else do nothing it is perfectly acceptable that not all data is in a cmdi file so not everything will be matched. E.G xpath expression evaluation CMDI session files will never match on CMD corpus files.
  • vlo/trunk/vlo_webapp/src/test/java/eu/clarin/cmdi/vlo/importer/CMDIDigesterTest.java

    r988 r989  
    375375        assertEquals("Netherlands", doc.getFieldValue("country"));
    376376        assertEquals("Max Planck Institute for Psycholinguistics", doc.getFieldValue("organisation"));
    377         assertEquals("Unspecified", doc.getFieldValue("genre"));
     377        assertEquals("unspecified", doc.getFieldValue("genre"));
    378378        assertEquals(
    379379                "This  recording was made to generate a freely available test resource including speech and gestures. The annotations were created by Peter and Kita who is gesture researcher at the MPI for Psycholinguistics.",
     
    466466        assertEquals("Europe", doc.getFieldValue("continent"));
    467467        assertEquals("Netherlands", doc.getFieldValue("country"));
    468         assertEquals("Unspecified", doc.getFieldValue("genre"));
     468        assertEquals("unspecified", doc.getFieldValue("genre"));
    469469        assertEquals("Test.", doc.getFieldValue("description"));
    470470        assertEquals("Should be null not empty string", null, doc.getFieldValue("organisation"));
     
    533533        assertEquals(null, doc.getFieldValue("country"));
    534534        assertEquals(null, doc.getFieldValue("organisation"));
    535         assertEquals("Transcription", doc.getFieldValue("genre"));
     535        assertEquals("transcription", doc.getFieldValue("genre"));
    536536      //  assertEquals("Kuna", doc.getFieldValue("subject"));
    537537        assertEquals(2, doc.getFieldValues("description").size());
Note: See TracChangeset for help on using the changeset viewer.