@Override
public AtomicLong next() {
if (done) {
return fileCounter;
}
try {
URI uri = input.poll();
if (uri != null) {
push(uri);
} else {
done = true;
}
fileCounter.incrementAndGet();
} catch (Exception e) {
logger.error(e.getMessage(), e);
done = true;
}
return fileCounter;
}
public static void main(String[] args) {
int exitcode = 0;
try {
OptionParser parser =
new OptionParser() {
{
accepts("elasticsearch").withRequiredArg().ofType(String.class).required();
accepts("index").withRequiredArg().ofType(String.class).required();
accepts("type").withRequiredArg().ofType(String.class).required();
accepts("maxbulkactions").withRequiredArg().ofType(Integer.class).defaultsTo(1000);
accepts("maxconcurrentbulkrequests")
.withRequiredArg()
.ofType(Integer.class)
.defaultsTo(4 * Runtime.getRuntime().availableProcessors());
accepts("mock").withOptionalArg().ofType(Boolean.class).defaultsTo(Boolean.FALSE);
accepts("path").withRequiredArg().ofType(String.class).required();
accepts("pattern")
.withRequiredArg()
.ofType(String.class)
.required()
.defaultsTo("*.txt");
accepts("threads").withRequiredArg().ofType(Integer.class).defaultsTo(1);
accepts("help");
}
};
final OptionSet options = parser.parse(args);
if (options.hasArgument("help")) {
System.err.println(
"Help for "
+ Medline.class.getCanonicalName()
+ lf
+ " --help print this help message"
+ lf
+ " --elasticsearch <uri> Elasticesearch URI"
+ lf
+ " --index <index> Elasticsearch index name"
+ lf
+ " --type <type> Elasticsearch type name"
+ lf
+ " --maxbulkactions <n> the number of bulk actions per request (optional, default: 1000)"
+ " --maxconcurrentbulkrequests <n>the number of concurrent bulk requests (optional, default: 4 * cpu cores)"
+ " --path <path> a file path from where the input files are recursively collected (required)"
+ lf
+ " --pattern <pattern> a regex for selecting matching file names for input (default: *.txt)"
+ lf
+ " --threads <n> the number of threads (optional, default: <num-of=cpus)");
System.exit(1);
}
input =
new Finder((String) options.valueOf("pattern"))
.find((String) options.valueOf("path"))
.getURIs();
final Integer threads = (Integer) options.valueOf("threads");
logger.info("found {} input files", input.size());
URI esURI = URI.create((String) options.valueOf("elasticsearch"));
index = (String) options.valueOf("index");
type = (String) options.valueOf("type");
int maxbulkactions = (Integer) options.valueOf("maxbulkactions");
int maxconcurrentbulkrequests = (Integer) options.valueOf("maxconcurrentbulkrequests");
boolean mock = (Boolean) options.valueOf("mock");
final IngestClient es = mock ? new MockIngestClient() : new IngestClient();
es.maxBulkActions(maxbulkactions)
.maxConcurrentBulkRequests(maxconcurrentbulkrequests)
.newClient(esURI)
.waitForCluster(ClusterHealthStatus.YELLOW, TimeValue.timeValueSeconds(30));
logger.info("creating new index ...");
es.setIndex(index).setType(type).newIndex();
logger.info("... new index created");
final ResourceSink sink = new ResourceSink(es);
ImportService service =
new ImportService()
.threads(threads)
.factory(
new ImporterFactory() {
@Override
public Importer newImporter() {
return new SpringerCitations(sink);
}
})
.execute();
logger.info(
"finished, number of files = {}, resources indexed = {}", fileCounter, sink.getCounter());
service.shutdown();
logger.info("service shutdown");
es.shutdown();
logger.info("elasticsearch client shutdown");
} catch (IOException | InterruptedException | ExecutionException e) {
logger.error(e.getMessage(), e);
exitcode = 1;
}
System.exit(exitcode);
}
