@Override
@SuppressWarnings("unchecked")
protected void cleanup(Reducer.Context context) throws IOException, InterruptedException {
if (!Project.getProject().isMetadataCollectStandard()) {
// write summary headers with all metadata, but for standard metadata don't write the last
// line
context.write(new Text("Hash"), new Text(columnMetadata.delimiterSeparatedHeaders()));
}
zipFileWriter.closeZip();
if (Project.getProject().isLuceneIndexEnabled()) {
mergeLuceneIndex();
}
Project project = Project.getProject();
if (project.isEnvHadoop()) {
String outputPath = Project.getProject().getProperty(ParameterProcessing.OUTPUT_DIR_HADOOP);
String zipFileName = zipFileWriter.getZipFileName();
if (project.isFsHdfs()) {
String cmd =
"hadoop fs -copyFromLocal "
+ zipFileName
+ " "
+ outputPath
+ File.separator
+ context.getTaskAttemptID()
+ ".zip";
OsUtil.runCommand(cmd);
} else if (project.isFsS3()) {
S3Agent s3agent = new S3Agent();
String run = project.getRun();
if (!run.isEmpty()) {
run = run + "/";
}
String s3key =
project.getProjectCode()
+ File.separator
+ "output/"
+ run
+ "results/"
+ context.getTaskAttemptID()
+ ".zip";
// Keep updating the hadoop progress
int refreshInterval = 60000;
Timer timer = new Timer(refreshInterval, this);
timer.start();
s3agent.putFileInS3(zipFileName, s3key);
timer.stop();
}
}
Stats.getInstance().setJobFinished();
}
public static synchronized SolrIndex getInstance() {
if (__instance == null) {
if (Project.getProject().isSendIndexToSolrEnabled()) {
__instance = new HttpSolrIndex();
} else {
__instance = new DisabledSolrIndex();
}
}
return __instance;
}
@Override
@SuppressWarnings("unchecked")
protected void setup(Reducer.Context context) throws IOException, InterruptedException {
this.context = context;
String settingsStr = context.getConfiguration().get(ParameterProcessing.SETTINGS_STR);
Settings settings = Settings.loadFromString(settingsStr);
Settings.setSettings(settings);
String projectStr = context.getConfiguration().get(ParameterProcessing.PROJECT);
Project project = Project.loadFromString(projectStr);
if (project.isEnvHadoop()) {
String metadataFileContents =
context.getConfiguration().get(ParameterProcessing.METADATA_FILE);
new File(ColumnMetadata.metadataNamesFile).getParentFile().mkdirs();
Files.write(metadataFileContents.getBytes(), new File(ColumnMetadata.metadataNamesFile));
}
columnMetadata = new ColumnMetadata();
String fileSeparatorStr = project.getFieldSeparator();
char fieldSeparatorChar = Delimiter.getDelim(fileSeparatorStr);
columnMetadata.setFieldSeparator(String.valueOf(fieldSeparatorChar));
columnMetadata.setAllMetadata(project.getMetadataCollect());
// write standard metadata fields
context.write(null, new Text(columnMetadata.delimiterSeparatedHeaders()));
zipFileWriter.setup();
zipFileWriter.openZipForWriting();
luceneIndex = new LuceneIndex(settings.getLuceneIndexDir(), project.getProjectCode(), null);
luceneIndex.init();
}
private void mergeLuceneIndex() throws IOException {
String luceneDir = Settings.getSettings().getLuceneIndexDir();
String hdfsLuceneDir =
"/" + luceneDir + File.separator + Project.getProject().getProjectCode() + File.separator;
String localLuceneTempDir = luceneDir + File.separator + "tmp" + File.separator;
File localLuceneTempDirFile = new File(localLuceneTempDir);
if (localLuceneTempDirFile.exists()) {
Util.deleteDirectory(localLuceneTempDirFile);
}
localLuceneTempDirFile.mkdir();
// copy all zip lucene indexes, created by maps to local hd
String cmd = "hadoop fs -copyToLocal " + hdfsLuceneDir + "* " + localLuceneTempDir;
OsUtil.runCommand(cmd);
// remove the map indexes as they are now copied to local
String removeOldZips = "hadoop fs -rm " + hdfsLuceneDir + "*";
OsUtil.runCommand(removeOldZips);
logger.trace("Lucene index files collected to: {}", localLuceneTempDirFile.getAbsolutePath());
String[] zipFilesArr = localLuceneTempDirFile.list();
for (String indexZipFileStr : zipFilesArr) {
String indexZipFileName = localLuceneTempDir + indexZipFileStr;
String unzipToDir = localLuceneTempDir + indexZipFileStr.replace(".zip", "");
ZipUtil.unzipFile(indexZipFileName, unzipToDir);
File indexDir = new File(unzipToDir);
FSDirectory fsDir = FSDirectory.open(indexDir);
luceneIndex.addToIndex(fsDir);
}
// TODO check if we need to push the index to S3 or somewhere else
luceneIndex.destroy();
}
@Override
public void init() {
try {
String endpoint = getSolrEndpoint();
if (supportMultipleProjects) {
String projectCode = Project.getProject().getProjectCode();
String command =
endpoint
+ "solr/admin/cores?action=CREATE&name="
+ SOLR_INSTANCE_DIR
+ "_"
+ projectCode
+ "&instanceDir="
+ SOLR_INSTANCE_DIR
+ "&config=solrconfig.xml&dataDir=data_"
+ projectCode
+ "&schema=schema.xml";
sendGetCommand(command);
this.updateUrl = endpoint + "solr/" + SOLR_INSTANCE_DIR + "_" + projectCode + "/update";
} else {
sendGetCommand(endpoint + "solr/admin/ping");
this.updateUrl = endpoint + "solr/update";
}
String deleteAll = "<delete><query>id:[*TO *]</query></delete>";
sendPostCommand(updateUrl, deleteAll);
sendPostCommand(updateUrl, "<commit/>");
} catch (SolrException se) {
History.appendToHistory("Problem with SOLR init: " + se.getMessage());
// se.printStackTrace();
}
}
