/**
* @param args
* @throws Exception
*/
public static void main(String[] args) throws Exception {
/* Parse command line arguments */
Getopt g = new Getopt("gateExtractor", args, "i:r:te");
g.setOpterr(false);
String inputPath = "";
String outputPath = "";
boolean train = false;
boolean eval = false;
boolean run = false;
int c;
String arg;
while ((c = g.getopt()) != -1) {
switch (c) {
case 'i':
arg = g.getOptarg();
if (arg == null || arg.isEmpty()) {
usage("Please provide an input path");
}
inputPath = arg;
break;
case 'r':
run = true;
arg = g.getOptarg();
if (arg == null || arg.isEmpty()) {
usage("Please provide an output path");
}
outputPath = arg;
break;
case 't':
train = true;
break;
case 'e':
eval = true;
break;
case '?':
default:
usage(null);
}
}
if (args.length == 0 || (!run && !train && !eval)) {
usage("Nothing to do.");
}
if (inputPath == null || inputPath.isEmpty()) {
usage("Please provide an input path");
}
if (run && (outputPath == null || outputPath.isEmpty())) {
usage("Please provide an output directory!");
}
if (train && eval) {
usage("Only one mode allowed at a time");
}
if (train && run) {
usage("Only one mode allowed at a time");
}
if (eval && run) {
usage("Only one mode allowed at a time");
}
/* Initialize GATE */
String location =
new File(Main.class.getProtectionDomain().getCodeSource().getLocation().getPath())
.getParent();
String resourcesFolder = location + "/resources";
Gate.setGateHome(new File(resourcesFolder));
/* Create ml-config.xml with threads */
createConfig(resourcesFolder + File.separator);
Gate.init();
/* Load Corpus */
log.info("Loading Corpus ... ");
Corpus corpus = Factory.newCorpus("Training Corpus");
File directory = new File(inputPath);
URL url = directory.toURI().toURL();
corpus.populate(url, null, null, true);
log.info("Done loading Corpus!");
Pipeline pipeline = null;
/* Do Tagging */
pipeline = new Tagger();
pipeline.run(corpus, resourcesFolder);
/* Train */
if (train) {
pipeline = new Trainer();
pipeline.run(corpus, resourcesFolder);
}
/* Apply learned rules */
if (run) {
pipeline = new Extractor();
pipeline.run(corpus, resourcesFolder);
ExecutorService executorService = Executors.newFixedThreadPool(20);
for (int i = 0; i < corpus.size(); i++) {
executorService.execute(new OutputGenerator(outputPath, corpus.get(i)));
}
executorService.shutdown();
executorService.awaitTermination(Long.MAX_VALUE, TimeUnit.DAYS);
}
/* Evaluate results */
if (eval) {
pipeline = new Evaluator();
pipeline.run(corpus, resourcesFolder);
}
/* Clean up */
Factory.deleteResource(corpus);
outputFile_mlConfigThreads.delete();
}
@Override
public String execute(String uuid, String... params) throws RemoteException {
try {
LongOpt[] longopts = new LongOpt[4];
longopts[0] = new LongOpt("help", LongOpt.NO_ARGUMENT, null, 'h');
longopts[1] = new LongOpt("project", LongOpt.REQUIRED_ARGUMENT, null, 'p');
longopts[2] = new LongOpt("file", LongOpt.REQUIRED_ARGUMENT, null, 'f');
longopts[3] = new LongOpt("dest", LongOpt.REQUIRED_ARGUMENT, null, 'd');
Getopt g = new Getopt("put", params, "", longopts);
g.setOpterr(false);
String project = null;
String path = null;
String[] parents = null;
String dest = null;
int c;
while ((c = g.getopt()) != -1) {
switch (c) {
case 'p':
project = g.getOptarg();
break;
case 'f':
path = g.getOptarg();
break;
case 'd':
dest = g.getOptarg();
if (!".".equals(dest)) {
parents = CLIUtils.convertAbsolutePathToArray(dest, false, false);
}
break;
case 'h':
return usage(uuid);
default:
return CLIResultCodes.INTERNAL_SERVER_ERROR
+ "::Could not execute the command \n "
+ usage(uuid);
}
}
if (uuid == null) {
return CLIResultCodes.USER_NOT_AUTHORIZED + "::--uuid parameter is required!";
}
if (project == null) {
return CLIResultCodes.USER_ERROR + "::--project parameter is required!";
}
if (path == null) {
return CLIResultCodes.USER_ERROR + "::--file parameter is required!";
}
if (dest == null) {
return CLIResultCodes.USER_ERROR + "::--dest parameter is required!";
}
if (!fileService.exists(uuid, project, null, null)) {
return CLIResultCodes.PROJECT_NOT_FOUND
+ "::Could not find project ["
+ project
+ "] for the current user";
}
File src = new File(path);
if (!src.exists()) {
return CLIResultCodes.FILE_DOES_NOT_EXIST
+ "::File ["
+ src.getName()
+ "] does not exist!";
}
String fileName = src.getName();
if (src.isDirectory()) {
fileService.create(uuid, project, parents, fileName, true);
String childrenPath = fileName;
if (parents != null) {
childrenPath = dest + "/" + fileName;
}
createDir(uuid, project, childrenPath, src.listFiles());
return CLIResultCodes.OK
+ "::Put the directory ["
+ src.getName()
+ "] to the destination ["
+ dest
+ "]";
} else {
fileService.upload(uuid, project, parents, fileName, src, true);
return CLIResultCodes.OK
+ "::Put the file ["
+ src.getName()
+ "] to the destination ["
+ dest
+ "]";
}
} catch (UserNotAuthorizedException e) {
CLIUtils.handleException(e, logger);
return CLIResultCodes.USER_NOT_AUTHORIZED + "::User not authorized!";
} catch (Exception e) {
CLIUtils.handleException(e, logger);
return CLIResultCodes.OK + "::Could not execute the command \n " + usage(uuid);
}
}