Examples with CsvFileFormat com.hazelcast.jet.pipeline.file.CsvFileFormat used on opensource projects

Example 1 with CsvFileFormat

use of com.hazelcast.jet.pipeline.file.CsvFileFormat in project hazelcast by hazelcast.

the class CsvReadFileFnProvider method createReadFileFn.

@SuppressWarnings("unchecked")
@Nonnull
@Override
public <T> FunctionEx<Path, Stream<T>> createReadFileFn(@Nonnull FileFormat<T> format) {
    CsvFileFormat<T> csvFileFormat = (CsvFileFormat<T>) format;
    // Format is not Serializable
    Class<?> formatClazz = csvFileFormat.clazz();
    return path -> {
        FileInputStream fis = new FileInputStream(path.toFile());
        MappingIterator<T> iterator;
        Function<T, T> projection = identity();
        if (formatClazz == String[].class) {
            ObjectReader reader = new CsvMapper().enable(Feature.WRAP_AS_ARRAY).readerFor(String[].class).with(CsvSchema.emptySchema().withSkipFirstDataRow(false));
            iterator = reader.readValues(fis);
            if (!iterator.hasNext()) {
                throw new JetException("Header row missing in " + path);
            }
            String[] header = (String[]) iterator.next();
            List<String> fieldNames = csvFileFormat.fieldNames();
            if (fieldNames != null) {
                projection = (Function<T, T>) createFieldProjection(header, fieldNames);
            }
        } else {
            iterator = new CsvMapper().readerFor(formatClazz).withoutFeatures(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES).with(CsvSchema.emptySchema().withHeader()).readValues(fis);
        }
        return StreamSupport.stream(Spliterators.spliteratorUnknownSize(iterator, ORDERED), false).map(projection).onClose(() -> uncheckRun(fis::close));
    };
}