[SOLR-3395] FieldStreamDataSource should handle null fields

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Won't Fix
• Affects Version/s: 3.5
• Component/s: contrib - DataImportHandler
• Assignee: Unassigned
• Reporter: Andreas W
• Created: 23/Apr/12 07:20
• Updated: 29/Aug/20 20:23
• Resolved: 29/Aug/20 20:23

Description

The FieldStreamDataSource currently throws a DataImportHandlerException if a field value is null.
IMHO this is not appropriate: It is legal for field values to be null (like no Blob exists in a particular row).
I suggest to return an empty InputStream rather than throwing a DataImportHandlerException.

  public InputStream getData(String query) {
    Object o = wrapper.getVariableResolver().resolve(dataField);
    if (o == null) {
//better: return new ByteArrayInputStream(new byte[0]);
      throw new DataImportHandlerException(SEVERE, "No field available for name : " + dataField);
    }
    if (o instanceof Blob) {
      Blob blob = (Blob) o;
      try {
        //Most of the JDBC drivers have getBinaryStream defined as public
        // so let us just check it
        Method m = blob.getClass().getDeclaredMethod("getBinaryStream");
        if (Modifier.isPublic(m.getModifiers())) {
          return (InputStream) m.invoke(blob);
        } else {
          // force invoke
          m.setAccessible(true);
          return (InputStream) m.invoke(blob);
        }
      } catch (Exception e) {
        LOG.info("Unable to get data from BLOB");
        return null;

      }
    } else if (o instanceof byte[]) {
      byte[] bytes = (byte[]) o;
      return new ByteArrayInputStream(bytes);
    } else {
      throw new RuntimeException("unsupported type : " + o.getClass());
    } 

  }

Maybe the best implementation would be to distinguish between a none-existing field and a null or empty field.

Issue Links

• is superceded by: SOLR-14783 Remove DIH from 9.0