Project

General

Profile

« Previous | Next » 

Revision 30986

Added by Marek Horst over 9 years ago

#757 fixing pmid and doi matching, fixing sourceDocumentId and destinationDocumentId generation

View differences:

CitationExtractorFunction.java
1 1 
package eu.dnetlib.iis.ingest.pmc.citations;
2 2 

  		




  
  3
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.DST_FIELD;


  		




  
  4
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.DST_IDS_FIELD;


  		




  
  5
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.DST_PMID_FIELD;


  		




  
  6
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.DST_DOI_FIELD;


  		




  
  7
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.POSITION_FIELD;


  		




  
  8
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.RAW_TEXT_FIELD;


  		




  
  9
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.SRC_FIELD;


  		




  3
  10
  
    
import cascading.flow.FlowProcess;


  		




  4
  11
  
    
import cascading.operation.BaseOperation;


  		




  5
  12
  
    
import cascading.operation.Function;


  		




  ......




  7
  14
  
    
import cascading.tuple.Fields;


  		




  8
  15
  
    
import cascading.tuple.Tuple;


  		




  9
  16
  
    

  		




  10
  
  
    
import static eu.dnetlib.iis.ingest.pmc.citations.Constants.*;


  		




  11
  
  
    

  		




  12
  17
  
    
/**


  		




  13
  
  
    
 * Extracts Resolved citations from an NLM file. Input pipe should contain a field named TEXT_FIELD. Result stream will


  		




  14
  
  
    
 * contain fields SRC_FIELD, POSITION_FIELD, DST_FIELD and DST_IDS_FIELD


  		




  
  18
  
    
 * Extracts Resolved citations from an NLM file. 


  		




  
  19
  
    
 * Input pipe should contain a field named TEXT_FIELD and ID_FIELD.


  		




  15
  20
  
    
 *


  		




  16
  21
  
    
 * @author Mateusz Fedoryszak (m.fedoryszak@icm.edu.pl)


  		




  17
  22
  
    
 */


  		




  18
  23
  
    
public class CitationExtractorFunction extends BaseOperation implements Function {


  		




  19
  24
  
    
    private final static Fields fields =


  		




  20
  
  
    
            new Fields(SRC_FIELD, POSITION_FIELD, RAW_TEXT_FIELD, DST_FIELD, DST_IDS_FIELD);


  		




  21
  
  
    
    //private final Logger logger = Logger.getLogger(CitationExtractorFunction.class);


  		




  
  25
  
    
            new Fields(SRC_FIELD, POSITION_FIELD, RAW_TEXT_FIELD, 


  		




  
  26
  
    
            		DST_FIELD, DST_IDS_FIELD, DST_DOI_FIELD, DST_PMID_FIELD);


  		




  22
  27
  
    

  		




  23
  28
  
    
    public  CitationExtractorFunction() {


  		




  24
  29
  
    
        super(2, fields);


  		




  ......




  27
  32
  
    
    @Override


  		




  28
  33
  
    
    public void operate(FlowProcess flowProcess, FunctionCall functionCall) {


  		




  29
  34
  
    
        String text = functionCall.getArguments().getString(Constants.TEXT_FIELD);


  		




  
  35
  
    
        String oaId = functionCall.getArguments().getString(Constants.ID_FIELD);


  		




  30
  36
  
    

  		




  31
  37
  
    
        try {


  		




  32
  
  
    
            for (ResolvedCitation cit : ResolvedCitation.extractFromNlm(text)) {


  		




  
  38
  
    
            for (ResolvedCitation cit : ResolvedCitation.extractFromNlm(oaId, text)) {


  		




  33
  39
  
    
                Tuple result = new Tuple();


  		




  34
  40
  
    
                result.addString(cit.getSourceOaid());


  		




  35
  41
  
    
                result.addInteger(cit.getPosition());


  		




  36
  42
  
    
                result.addString(cit.getRawText());


  		




  37
  
  
    
                result.addString(cit.getTargetOaid());


  		




  
  43
  
    
                result.addString(null);


  		




  38
  44
  
    
                result.addString(cit.getTargetIdsJson());


  		




  
  45
  
    
                result.addString(cit.getTargetDoi());


  		




  
  46
  
    
                result.addString(cit.getTargetPmid());


  		




  39
  47
  
    
                functionCall.getOutputCollector().add(result);


  		




  40
  48
  
    
            }


  		




  41
  49
  
    
        } catch (Exception e) {


  		




  42
  
  
    
            //in case of parse exception


  		




  43
  
  
    
            //logger.error("Error while parsing NLM\n"+text, e);


  		




  
  50
  
    
        	throw new RuntimeException(e);


  		




  44
  51
  
    
        }


  		




  45
  52
  
    
    }


  		




  46
  53
  
    
}


  		












Also available in:  Unified diff