@Override public boolean transform(Media media, String docNewId, DocumentMetadata.Builder dmBuider,DocumentProtos.DocumentWrapper.Builder builder) { byte[] b = media.getContent().toByteArray(); byte[] c = Arrays.copyOf(b, b.length); ScholarRecordP record; BasicMetadata.Builder bmBuilder = BasicMetadata.newBuilder(); try { record = ScholarRecordP.parseFrom(c); } catch (InvalidProtocolBufferException ex) { java.util.logging.Logger.getLogger(GsMediaToBw2Metadata.class.getName()).log(Level.SEVERE, null, ex); return false; } AuthorData ad = new AuthorData(); ad.docId = docNewId; if (translateGoogleScholarToDocumentMetadata(record, dmBuider, bmBuilder, ad)) { dmBuider.setBasicMetadata(bmBuilder); return true; } return false; } }
kwdList.addKeywords(s); dm.addKeywords(kwdList); return true;
public Builder enrichDocumentMetadata(Builder dwb, Map<Integer, String> refsMap) { List<ReferenceMetadata.Builder> rml_in = dwb .getDocumentMetadataBuilder().getReferenceBuilderList(); List<ReferenceMetadata> rml_out = new ArrayList<ReferenceMetadata>(); dwb.getDocumentMetadataBuilder().clearReference(); dwb.getDocumentMetadataBuilder().addAllReference(rml_out); return dwb;
dm.addExtId(KeyValue.newBuilder().setKey(YConstants.EXT_SCHEME_PMID).setValue(value)); return true;
if(commonDocumentMetadata == null){ commonDocumentMetadata = DocumentMetadata.newBuilder(documentMetadata); commonBasicMetadata = BasicMetadata.newBuilder(commonDocumentMetadata.getBasicMetadata()); continue; commonDocumentMetadata.setBasicMetadata(commonBasicMetadata.build()); DocumentWrapper.Builder commonDocumentWrapper = DocumentWrapper.newBuilder(); commonDocumentWrapper.setDocumentMetadata(commonDocumentMetadata);
BasicMetadata.Builder bm = dwb.getDocumentMetadataBuilder().getBasicMetadataBuilder(); dmb.setBasicMetadata(bm); dwb.setDocumentMetadata(dmb);
documentMetadata.setKey(doi); documentMetadata.addCollection("orcid"); basicMetadata.addAuthor(author); documentMetadata.setBasicMetadata(basicMetadata);
public Tuple exec(Tuple input) throws IOException { if (input == null || input.size() != 3) { return null; } try{ String key = (String) input.get(0); DocumentWrapper dw = DocumentWrapper.parseFrom(((DataByteArray) input.get(1)).get()); String correctedDoi = (String) input.get(2); DocumentWrapper.Builder dwb = DocumentWrapper.newBuilder(dw); DocumentMetadata.Builder dmb = DocumentMetadata.newBuilder(dw.getDocumentMetadata()); BasicMetadata.Builder bmb = BasicMetadata.newBuilder(dmb.getBasicMetadata()); bmb.setDoi(correctedDoi); dmb.setBasicMetadata(bmb); dwb.setDocumentMetadata(dmb); Tuple ret = TupleFactory.getInstance().newTuple(); ret.append(key); ret.append(new DataByteArray(dwb.build().toByteArray())); return ret; }catch(Exception e){ logger.error("Error in processing input row:"+ StackTraceExtractor.getStackTrace(e), e); throw new IOException("Caught exception processing input row:\n" + StackTraceExtractor.getStackTrace(e)); } } }
private void incorporateDataToDWBuilder(Builder dwb, BasicMetadata.Builder bm, List<Author> myAuthorList) { bm.clearAuthor(); bm.addAllAuthor(myAuthorList); DocumentMetadata.Builder dmb = dwb.getDocumentMetadataBuilder(); dmb.setBasicMetadata(bm); dwb.setDocumentMetadata(dmb); }
@Override public void setIDFor(DocumentProtos.DocumentWrapper.Builder t, String id) { t.setRowId(id); t.getDocumentMetadataBuilder().setKey(id); for (DocumentProtos.Author.Builder build:t.getDocumentMetadataBuilder().getBasicMetadataBuilder().getAuthorBuilderList()) { build.setDocId(id); } }
private Builder buildDocumentMetadata(Builder dwb, DocumentSimilarityInfo dco) { DocumentMetadata.Builder dm = dwb.getDocumentMetadataBuilder(); dm.addSimilarDocumentInfo(dco); dwb.setDocumentMetadata(dm); return dwb; } }
@Override public Builder execute(Tuple t, int dataIndex, Builder dwb) throws IOException { DisambiguationAuthorOut dco = readInputData(t, dataIndex); boolean dataExists = dco!=null ? true : false; Map<String, String> contribIdToPersonId = extractContribToPersonMap(dco, dataExists); ProvenanceInfo.Builder pb = createProvenance(); BasicMetadata.Builder bm = dwb.getDocumentMetadataBuilder().getBasicMetadataBuilder(); List<Author> myAuthorList = updateAuthorList(dataExists,contribIdToPersonId, pb, bm); incorporateDataToDWBuilder(dwb, bm, myAuthorList); return dwb; }
private Builder buildDocumentMetadata(Builder dwb, KeywordsList kl) { DocumentMetadata.Builder dmb = dwb.getDocumentMetadataBuilder(); dmb.clearKeywords(); dmb.addKeywords(kl); dwb.setDocumentMetadata(dmb.build()); return dwb; } }
private Builder buildDocumentMetadata(Builder dwb, ParentModelProtos.ParentDisambiguationOutOrBuilder pmo) { DocumentMetadata.Builder dm = dwb.getDocumentMetadataBuilder(); dm.getBasicMetadataBuilder().addParentId(pmo.getParentId()); dwb.setDocumentMetadata(dm); return dwb; } }