public ObjectStream<SentenceSample> create(String[] args) {
Parameters params = ArgumentParser.parse(args, Parameters.class);
ObjectStream<POSSample> posSampleStream =
StreamFactoryRegistry.getFactory(POSSample.class, StreamFactoryRegistry.DEFAULT_FORMAT)
.create(ArgumentParser.filter(args, WordTagSampleStreamFactory.Parameters.class));
return new POSToSentenceSampleStream(createDetokenizer(params), posSampleStream, 30);
}
public ObjectStream<ChunkSample> create(String[] args) {
Parameters params = ArgumentParser.parse(args, Parameters.class);
language = params.getLang();
FileInputStream sampleDataIn = CmdLineUtil.openInFile(params.getData());
ObjectStream<String> lineStream =
new PlainTextByLineStream(sampleDataIn.getChannel(), params.getEncoding());
ADChunkBasedShallowParserSampleStream sampleStream =
new ADChunkBasedShallowParserSampleStream(
lineStream,
params.getFunctTags(),
params.getIsIncludePOSTags(),
params.getUseCGTags(),
params.getExpandME());
if (params.getStart() != null && params.getStart() > -1) {
sampleStream.setStart(params.getStart());
}
if (params.getEnd() != null && params.getEnd() > -1) {
sampleStream.setEnd(params.getEnd());
}
return sampleStream;
}
public ObjectStream<NameSample> create(String[] args) {
Parameters params = ArgumentParser.parse(args, Parameters.class);
int typesToGenerate = 0;
if (params.getTypes().contains("DNA")) {
typesToGenerate = typesToGenerate | BioNLP2004NameSampleStream.GENERATE_DNA_ENTITIES;
} else if (params.getTypes().contains("protein")) {
typesToGenerate = typesToGenerate | BioNLP2004NameSampleStream.GENERATE_PROTEIN_ENTITIES;
} else if (params.getTypes().contains("cell_type")) {
typesToGenerate = typesToGenerate | BioNLP2004NameSampleStream.GENERATE_CELLTYPE_ENTITIES;
} else if (params.getTypes().contains("cell_line")) {
typesToGenerate = typesToGenerate | BioNLP2004NameSampleStream.GENERATE_CELLLINE_ENTITIES;
} else if (params.getTypes().contains("RNA")) {
typesToGenerate = typesToGenerate | BioNLP2004NameSampleStream.GENERATE_RNA_ENTITIES;
}
return new BioNLP2004NameSampleStream(
CmdLineUtil.openInFile(new File(params.getData())), typesToGenerate);
}
public boolean validateArguments(String[] args) {
return ArgumentParser.validateArguments(args, Parameters.class);
}
public String getUsage() {
return ArgumentParser.createUsage(Parameters.class);
}