Search in sources :

Example 1 with ResettableReaderIteratorFactory

use of edu.stanford.nlp.objectbank.ResettableReaderIteratorFactory in project CoreNLP by stanfordnlp.

the class AbstractSequenceClassifier method makeObjectBankFromFiles.

public ObjectBank<List<IN>> makeObjectBankFromFiles(String baseDir, String filePattern, DocumentReaderAndWriter<IN> readerAndWriter) {
    File path = new File(baseDir);
    FileFilter filter = new RegExFileFilter(Pattern.compile(filePattern));
    File[] origFiles = path.listFiles(filter);
    Collection<File> files = new ArrayList<>();
    for (File file : origFiles) {
        if (file.isFile()) {
            if (flags.announceObjectBankEntries) {
                log.info("Getting data from " + file + " (" + flags.inputEncoding + " encoding)");
            }
            files.add(file);
        }
    }
    if (files.isEmpty()) {
        throw new RuntimeException("No matching files: " + baseDir + '\t' + filePattern);
    }
    // TODO get rid of ObjectBankWrapper
    return new ObjectBankWrapper<>(flags, new ObjectBank<>(new ResettableReaderIteratorFactory(files, flags.inputEncoding), readerAndWriter), knownLCWords);
}
Also used : RegExFileFilter(edu.stanford.nlp.io.RegExFileFilter) RegExFileFilter(edu.stanford.nlp.io.RegExFileFilter) ResettableReaderIteratorFactory(edu.stanford.nlp.objectbank.ResettableReaderIteratorFactory)

Aggregations

RegExFileFilter (edu.stanford.nlp.io.RegExFileFilter)1 ResettableReaderIteratorFactory (edu.stanford.nlp.objectbank.ResettableReaderIteratorFactory)1