private static Long latestVersion(Hashtable<String, String> config, dbutil db_util)
throws Exception {
if (!config.get("timestamp_stop").equals(Integer.toString(Integer.MAX_VALUE))) {
return new Long(config.get("timestamp_stop"));
}
String rowName = config.get("file_id") + config.get("run_id") + "_";
if (config.get("task_id") != "") {
try {
rowName = rowName + String.format("%04d", new Integer(config.get("task_id")));
} catch (NumberFormatException E) {
rowName = rowName + config.get("task_id");
}
}
Get timestampGet = new Get(rowName.getBytes());
timestampGet.addColumn("d".getBytes(), "update".getBytes());
Result timestampResult = db_util.doGet(config.get("db_name_updates"), timestampGet);
KeyValue tsKv = timestampResult.getColumnLatest("d".getBytes(), "update".getBytes());
if (tsKv == null) {
rowName = config.get("file_id") + "_";
timestampGet = new Get(rowName.getBytes());
timestampGet.addColumn("d".getBytes(), "update".getBytes());
timestampResult = db_util.doGet(config.get("db_name_updates"), timestampGet);
tsKv = timestampResult.getColumnLatest("d".getBytes(), "update".getBytes());
}
if (tsKv == null) {
return new Long(Integer.MAX_VALUE);
}
Long latestVersion = new Long(tsKv.getTimestamp());
return latestVersion;
}
public Optional<byte[]> get(
Optional<String> table,
Optional<String> family,
Optional<String> qualifier,
Optional<String> key) {
if (!valid) {
Logger.error("CANNOT GET! NO VALID CONNECTION");
return Optional.empty();
}
if (table.isPresent()
&& family.isPresent()
&& qualifier.isPresent()
&& key.isPresent()
&& !key.get().isEmpty()) {
try {
final Table htable = connection.getTable(TableName.valueOf(table.get()));
Result result = htable.get(new Get(key.get().getBytes("UTF8")));
return Optional.ofNullable(
result.getValue(family.get().getBytes("UTF8"), qualifier.get().getBytes("UTF8")));
} catch (IOException e) {
e.printStackTrace();
}
}
return Optional.empty();
}
public Optional<Response> get(Optional<Request> request) {
if (!valid) {
Logger.error("CANNOT GET! NO VALID CONNECTION");
return Optional.empty();
}
Response response = new Response();
if (request.isPresent()) {
Request r = request.get();
response.key = r.key;
response.table = r.table;
try {
final Table htable = connection.getTable(TableName.valueOf(r.table));
Result result = htable.get(new Get(r.key));
if (result == null || result.isEmpty()) {
return Optional.empty();
}
r.columns.forEach(
c ->
response.columns.add(
new Request.Column(
c.family,
c.qualifier,
result.getValue(c.family.getBytes(), c.qualifier.getBytes()))));
} catch (IOException e) {
e.printStackTrace();
}
}
return Optional.of(response);
}
private static boolean hasFile(
dbutil db_util, FileSystem fs, String db_name, String file_id, String file_path)
throws Exception {
Get file_id_get = new Get(file_id.getBytes());
Result file_result = db_util.doGet(db_name, file_id_get);
KeyValue file_names = file_result.getColumnLatest("d".getBytes(), "filenames".getBytes());
if (file_names == null) {
return false;
}
String all_files = new String(file_names.getValue());
String[] files = all_files.split("\n");
for (String line : files) {
if (line.equals(file_path)) {
if (fs.globStatus(new Path(line + "*")).length == 0) {
Put new_put = new Put(file_id.getBytes());
new_put.add(
"d".getBytes(),
"filenames".getBytes(),
all_files.replace(file_path + "\n", "").getBytes());
db_util.doPut(db_name, new_put);
return false;
}
return true;
}
}
return false;
}
private static boolean putFileEntry(
dbutil db_util,
FileSystem fs,
String db_name,
String file_id,
String file_path,
String source)
throws Exception {
String all_paths = file_path;
if (hasFile(db_util, fs, db_name, file_id, file_path)) {
logger.debug("File already found, putFileEntry aborting");
return false;
} else {
Get file_id_get = new Get(file_id.getBytes());
Result file_result = db_util.doGet(db_name, file_id_get);
KeyValue file_names = file_result.getColumnLatest("d".getBytes(), "filenames".getBytes());
if (file_names != null) {
String paths = new String(file_names.getValue());
all_paths = paths + "\n" + file_path;
}
}
Put file_id_put = new Put(file_id.getBytes());
file_id_put.add("d".getBytes(), "source".getBytes(), source.getBytes());
if (!source.equals("fullfile")) {
file_id_put.add("d".getBytes(), "filenames".getBytes(), all_paths.getBytes());
}
db_util.doPut(db_name, file_id_put);
return true;
}
public int run(String[] args) throws Exception {
Configuration argConf = getConf();
// JobConf conf = new JobConf(diffdb.class);
Configuration config = HBaseConfiguration.create();
HBaseAdmin hbAdmin = new HBaseAdmin(config);
dbutil db_util = new dbutil(config);
HTable runTable = new HTable(config, "gestore_runs");
Get runGet = new Get(argConf.get("id").getBytes());
Result pipeline = runTable.get(runGet);
NavigableMap<byte[], byte[]> pipeMap = pipeline.getFamilyMap("d".getBytes());
Map.Entry<byte[], byte[]> results = pipeMap.pollFirstEntry();
HashMap<String, HashMap<String, String>> resultMap =
new HashMap<String, HashMap<String, String>>();
while (results != null) {
String resultKey = new String(results.getKey());
String resultValue = new String(results.getValue());
String field = "type";
HashMap<String, String> tempMap = new HashMap<String, String>();
String entry = resultKey;
if (resultKey.endsWith("_db_timestamp")) {
field = "db_timestamp";
entry = resultKey.substring(0, resultKey.lastIndexOf("_db_timestamp"));
} else if (resultKey.endsWith("_filename")) {
field = "filename";
entry = resultKey.substring(0, resultKey.lastIndexOf("_filename"));
} else if (resultKey.endsWith("_regex")) {
field = "regex";
entry = resultKey.substring(0, resultKey.lastIndexOf("_regex"));
}
if (resultMap.containsKey(entry)) {
tempMap = resultMap.get(entry);
}
tempMap.put(field, resultValue);
resultMap.put(entry, tempMap);
// System.out.println("Key: " + resultKey + " Value: " + resultValue);
results = pipeMap.pollFirstEntry();
}
for (String key : resultMap.keySet()) {
System.out.println("File ID: " + key);
for (String subKey : resultMap.get(key).keySet()) {
// System.out.println("\t " + subKey + "\t\t" + resultMap.get(key).get(subKey));
System.out.format(" %1$-20s %2$s\n", subKey, resultMap.get(key).get(subKey));
}
}
return 0;
}
private static String getSource(dbutil db_util, String db_name, String file_id) throws Exception {
Get file_id_get = new Get(file_id.getBytes());
Result file_result = db_util.doGet(db_name, file_id_get);
KeyValue file_source = file_result.getColumnLatest("d".getBytes(), "source".getBytes());
if (file_source == null) {
return "local";
}
return new String(file_source.getValue());
}
public static void main(String[] args) throws IOException {
Configuration conf = HBaseConfiguration.create();
HTableInterface literaturesTable = new HTable(conf, "pb_literatures");
Scan s = new Scan();
s.addFamily(Bytes.toBytes("info"));
ResultScanner rs = literaturesTable.getScanner(s);
for (Result r : rs) {
byte[] b = r.getValue(Bytes.toBytes("info"), Bytes.toBytes("title"));
System.out.println(Bytes.toString(b));
}
}
private static String isDatabase(dbutil db_util, String db_name, String file_id)
throws Exception {
Get file_id_get = new Get(file_id.getBytes());
Result file_result = db_util.doGet(db_name, file_id_get);
KeyValue file_db = file_result.getColumnLatest("d".getBytes(), "database".getBytes());
if (file_db == null) {
return "n";
}
String db = new String(file_db.getValue());
if (db.equals("y")) {
return "y";
} else {
return "n";
}
}
/** Return the number of rows in the given table. */
public static int countMobRows(final Table table) throws IOException {
Scan scan = new Scan();
ResultScanner results = table.getScanner(scan);
int count = 0;
for (Result res : results) {
count++;
List<Cell> cells = res.listCells();
for (Cell cell : cells) {
// Verify the value
Assert.assertTrue(CellUtil.cloneValue(cell).length > 0);
}
}
results.close();
return count;
}
private static String getFileNames(
dbutil db_util, String db_name, String file_id, String file_path) throws Exception {
Get file_id_get = new Get(file_id.getBytes());
Result file_result = db_util.doGet(db_name, file_id_get);
KeyValue file_names = file_result.getColumnLatest("d".getBytes(), "filenames".getBytes());
if (file_names == null) {
return "";
}
String all_files = new String(file_names.getValue());
String[] files = all_files.split("\n");
for (String line : files) {
if (line.equals(file_path)) {
return all_files;
}
}
return all_files + "\n" + file_path;
}
@Test
public void testHostRank() throws Exception {
if (System.getProperty("prop.mapred.job.tracker") != null) {
if (LOG.isInfoEnabled())
LOG.info("testHBaseInputOutput: Ignore this test if not local mode.");
return;
}
File jarTest = new File(System.getProperty("prop.jarLocation"));
if (!jarTest.exists()) {
fail(
"Could not find Giraph jar at "
+ "location specified by 'prop.jarLocation'. "
+ "Make sure you built the main Giraph artifact?.");
}
MiniHBaseCluster cluster = null;
MiniZooKeeperCluster zkCluster = null;
FileSystem fs = null;
try {
// using the restart method allows us to avoid having the hbase
// root directory overwritten by /home/$username
zkCluster = testUtil.startMiniZKCluster();
testUtil.restartHBaseCluster(2);
cluster = testUtil.getMiniHBaseCluster();
final byte[] OL_BYTES = Bytes.toBytes("ol");
final byte[] S_BYTES = Bytes.toBytes("s");
final byte[] METADATA_BYTES = Bytes.toBytes("mtdt");
final byte[] HR_BYTES = Bytes.toBytes("_hr_");
final byte[] TAB = Bytes.toBytes(TABLE_NAME);
Configuration conf = cluster.getConfiguration();
HTableDescriptor desc = new HTableDescriptor(TAB);
desc.addFamily(new HColumnDescriptor(OL_BYTES));
desc.addFamily(new HColumnDescriptor(S_BYTES));
desc.addFamily(new HColumnDescriptor(METADATA_BYTES));
HBaseAdmin hbaseAdmin = new HBaseAdmin(conf);
if (hbaseAdmin.isTableAvailable(TABLE_NAME)) {
hbaseAdmin.disableTable(TABLE_NAME);
hbaseAdmin.deleteTable(TABLE_NAME);
}
hbaseAdmin.createTable(desc);
/**
* Enter the initial data (a,b), (b,c), (a,c) a = 1.0 - google b = 1.0 - yahoo c = 1.0 - bing
*/
HTable table = new HTable(conf, TABLE_NAME);
Put p1 = new Put(Bytes.toBytes("com.google.www"));
p1.add(OL_BYTES, Bytes.toBytes("www.yahoo.com"), Bytes.toBytes("ab"));
Put p2 = new Put(Bytes.toBytes("com.google.www"));
p2.add(OL_BYTES, Bytes.toBytes("www.bing.com"), Bytes.toBytes("ac"));
p2.add(OL_BYTES, Bytes.toBytes("www.bing.com"), Bytes.toBytes("invalid1"));
p2.add(OL_BYTES, Bytes.toBytes("www.google.com"), Bytes.toBytes("invalid2"));
Put p3 = new Put(Bytes.toBytes("com.yahoo.www"));
p3.add(OL_BYTES, Bytes.toBytes("www.bing.com"), Bytes.toBytes("bc"));
// p3.add(OL_BYTES, Bytes.toBytes(""), Bytes.toBytes("invalid4"));
Put p4 = new Put(Bytes.toBytes("com.bing.www"));
// TODO: Handle below case. use apache isValid method.
p4.add(OL_BYTES, Bytes.toBytes("http://invalidurl"), Bytes.toBytes("invalid5"));
p4.add(S_BYTES, S_BYTES, Bytes.toBytes(10.0d));
Put p5 = new Put(Bytes.toBytes("dummy"));
p5.add(S_BYTES, S_BYTES, Bytes.toBytes(10.0d));
table.put(p1);
table.put(p2);
table.put(p3);
table.put(p4);
table.put(p5);
// Set Giraph configuration
// now operate over HBase using Vertex I/O formats
conf.set(TableInputFormat.INPUT_TABLE, TABLE_NAME);
conf.set(TableOutputFormat.OUTPUT_TABLE, TABLE_NAME);
// Start the giraph job
GiraphJob giraphJob = new GiraphJob(conf, BspCase.getCallingMethodName());
GiraphConfiguration giraphConf = giraphJob.getConfiguration();
giraphConf.setZooKeeperConfiguration(cluster.getMaster().getZooKeeper().getQuorum());
setupConfiguration(giraphJob);
giraphConf.setComputationClass(LinkRankComputation.class);
giraphConf.setMasterComputeClass(LinkRankVertexMasterCompute.class);
giraphConf.setOutEdgesClass(ByteArrayEdges.class);
giraphConf.setVertexInputFormatClass(Nutch2HostInputFormat.class);
giraphConf.setVertexOutputFormatClass(Nutch2HostOutputFormat.class);
giraphConf.setInt("giraph.linkRank.superstepCount", 10);
giraphConf.setInt("giraph.linkRank.scale", 10);
giraphConf.set("giraph.linkRank.family", "mtdt");
giraphConf.set("giraph.linkRank.qualifier", "_hr_");
giraphConf.setVertexInputFilterClass(HostRankVertexFilter.class);
assertTrue(giraphJob.run(true));
if (LOG.isInfoEnabled()) LOG.info("Giraph job successful. Checking output qualifier.");
/** Check the results * */
Result result;
String key;
byte[] calculatedScoreByte;
HashMap expectedValues = new HashMap<String, Double>();
expectedValues.put("com.google.www", 1.3515060339386287d);
expectedValues.put("com.yahoo.www", 4.144902009567587d);
expectedValues.put("com.bing.www", 9.063893290511482d);
for (Object keyObject : expectedValues.keySet()) {
key = keyObject.toString();
result = table.get(new Get(key.getBytes()));
calculatedScoreByte = result.getValue(METADATA_BYTES, HR_BYTES);
assertNotNull(calculatedScoreByte);
assertTrue(calculatedScoreByte.length > 0);
Assert.assertEquals(
"Scores are not the same",
(Double) expectedValues.get(key),
Bytes.toDouble(calculatedScoreByte),
DELTA);
}
} finally {
if (cluster != null) {
cluster.shutdown();
}
if (zkCluster != null) {
zkCluster.shutdown();
}
// clean test files
if (fs != null) {
fs.delete(hbaseRootdir);
}
}
}
public int run(String[] args) throws Exception {
// printUsage();
/*
* SETUP
*/
Configuration argConf = getConf();
Hashtable<String, String> confArg = new Hashtable<String, String>();
setup(confArg, argConf);
Date currentTime = new Date();
Date endDate = new Date(new Long(confArg.get("timestamp_stop")));
Boolean full_run = confArg.get("intermediate").matches("(?i).*true.*");
Boolean quick_add = confArg.get("quick_add").matches("(?i).*true.*");
logger.info("Running GeStore");
// ZooKeeper setup
Configuration config = HBaseConfiguration.create();
zkWatcher = new ZooKeeperWatcher(config, "Testing", new HBaseAdmin(config));
zkInstance =
new ZooKeeper(
ZKConfig.getZKQuorumServersString(config),
config.getInt("zookeeper.session.timeout", -1),
zkWatcher);
if (!confArg.get("task_id").isEmpty()) {
confArg.put("temp_path", confArg.get("temp_path") + confArg.get("task_id"));
}
String lockRequest = confArg.get("file_id");
if (!confArg.get("run_id").isEmpty())
lockRequest = lockRequest + "_" + confArg.get("run_id") + "_";
if (!confArg.get("task_id").isEmpty())
lockRequest = lockRequest + "_" + confArg.get("task_id") + "_";
// Get type of movement
toFrom type_move = checkArgs(confArg);
if (type_move == toFrom.LOCAL2REMOTE && !confArg.get("format").equals("unknown")) {
List<String> arguments = new ArrayList<String>();
arguments.add("-Dinput=" + confArg.get("local_path"));
arguments.add("-Dtable=" + confArg.get("file_id"));
arguments.add("-Dtimestamp=" + confArg.get("timestamp_stop"));
arguments.add("-Dtype=" + confArg.get("format"));
arguments.add("-Dtarget_dir=" + confArg.get("base_path") + "_" + confArg.get("file_id"));
arguments.add("-Dtemp_hdfs_path=" + confArg.get("temp_path"));
arguments.add("-Drun_id=" + confArg.get("run_id"));
if (!confArg.get("run_id").isEmpty()) arguments.add("-Drun_id=" + confArg.get("run_id"));
if (!confArg.get("task_id").isEmpty()) arguments.add("-Dtask_id=" + confArg.get("task_id"));
if (quick_add) arguments.add("-Dquick_add=" + confArg.get("quick_add"));
String lockName = lock(lockRequest);
String[] argumentString = arguments.toArray(new String[arguments.size()]);
adddb.main(argumentString);
unlock(lockName);
System.exit(0);
}
// Database registration
dbutil db_util = new dbutil(config);
db_util.register_database(confArg.get("db_name_files"), true);
db_util.register_database(confArg.get("db_name_runs"), true);
db_util.register_database(confArg.get("db_name_updates"), true);
FileSystem hdfs = FileSystem.get(config);
FileSystem localFS = FileSystem.getLocal(config);
// Get source type
confArg.put("source", getSource(db_util, confArg.get("db_name_files"), confArg.get("file_id")));
confArg.put(
"database", isDatabase(db_util, confArg.get("db_name_files"), confArg.get("file_id")));
if (!confArg.get("source").equals("local")
&& type_move == toFrom.REMOTE2LOCAL
&& !confArg.get("timestamp_stop").equals(Integer.toString(Integer.MAX_VALUE))) {
confArg.put("timestamp_stop", Long.toString(latestVersion(confArg, db_util)));
}
/*
* Get previous timestamp
*/
Get run_id_get = new Get(confArg.get("run_id").getBytes());
Result run_get = db_util.doGet(confArg.get("db_name_runs"), run_id_get);
KeyValue run_file_prev =
run_get.getColumnLatest(
"d".getBytes(), (confArg.get("file_id") + "_db_timestamp").getBytes());
String last_timestamp = new String("0");
if (null != run_file_prev && !confArg.get("source").equals("local")) {
long last_timestamp_real = run_file_prev.getTimestamp();
Long current_timestamp = new Long(confArg.get("timestamp_real"));
if ((current_timestamp - last_timestamp_real) > 36000) {
last_timestamp = new String(run_file_prev.getValue());
Integer lastTimestamp = new Integer(last_timestamp);
lastTimestamp += 1;
last_timestamp = lastTimestamp.toString();
logger.info("Last timestamp: " + last_timestamp + " End data: " + endDate);
Date last_run = new Date(run_file_prev.getTimestamp());
if (last_run.before(endDate) && !full_run) {
confArg.put("timestamp_start", last_timestamp);
}
}
}
Integer tse = new Integer(confArg.get("timestamp_stop"));
Integer tss = new Integer(confArg.get("timestamp_start"));
if (tss > tse) {
logger.info("No new version of requested file.");
return 0;
}
/*
* Generate file
*/
String lockName = lock(lockRequest);
Get file_id_get = new Get(confArg.get("file_id").getBytes());
Result file_get = db_util.doGet(confArg.get("db_name_files"), file_id_get);
if (!file_get.isEmpty()) {
boolean found =
hasFile(
db_util,
hdfs,
confArg.get("db_name_files"),
confArg.get("file_id"),
getFullPath(confArg));
if (confArg.get("source").equals("fullfile")) {
found = false;
}
String filenames_put =
getFileNames(
db_util, confArg.get("db_name_files"), confArg.get("file_id"), getFullPath(confArg));
// Filename not found in file database
if (!found && type_move == toFrom.REMOTE2LOCAL) {
if (!confArg.get("source").equals("local")) {
// Generate intermediate file
if (getFile(hdfs, confArg, db_util) == null) {
unlock(lockName);
return 1;
}
// Put generated file into file database
if (!confArg.get("format").equals("fullfile")) {
putFileEntry(
db_util,
hdfs,
confArg.get("db_name_files"),
confArg.get("file_id"),
confArg.get("full_file_name"),
confArg.get("source"));
}
} else {
logger.warn("Remote file not found, and cannot be generated! File: " + confArg);
unlock(lockName);
return 1;
}
}
} else {
if (type_move == toFrom.REMOTE2LOCAL) {
logger.warn("Remote file not found, and cannot be generated.");
unlock(lockName);
return 1;
}
}
/*
* Copy file
* Update tables
*/
if (type_move == toFrom.LOCAL2REMOTE) {
if (!confArg.get("format").equals("fullfile")) {
putFileEntry(
db_util,
hdfs,
confArg.get("db_name_files"),
confArg.get("file_id"),
getFullPath(confArg),
confArg.get("source"));
}
putRunEntry(
db_util,
confArg.get("db_name_runs"),
confArg.get("run_id"),
confArg.get("file_id"),
confArg.get("type"),
confArg.get("timestamp_real"),
confArg.get("timestamp_stop"),
getFullPath(confArg),
confArg.get("delimiter"));
hdfs.copyFromLocalFile(new Path(confArg.get("local_path")), new Path(getFullPath(confArg)));
} else if (type_move == toFrom.REMOTE2LOCAL) {
FileStatus[] files = hdfs.globStatus(new Path(getFullPath(confArg) + "*"));
putRunEntry(
db_util,
confArg.get("db_name_runs"),
confArg.get("run_id"),
confArg.get("file_id"),
confArg.get("type"),
confArg.get("timestamp_real"),
confArg.get("timestamp_stop"),
getFullPath(confArg),
confArg.get("delimiter"));
unlock(lockName);
for (FileStatus file : files) {
Path cur_file = file.getPath();
Path cur_local_path =
new Path(new String(confArg.get("local_path") + confArg.get("file_id")));
String suffix = getSuffix(getFileName(confArg), cur_file.getName());
if (suffix.length() > 0) {
cur_local_path = cur_local_path.suffix(new String("." + suffix));
}
if (confArg.get("copy").equals("true")) {
String crc = hdfs.getFileChecksum(cur_file).toString();
if (checksumLocalTest(cur_local_path, crc)) {
continue;
} else {
hdfs.copyToLocalFile(cur_file, cur_local_path);
writeChecksum(cur_local_path, crc);
}
} else {
System.out.println(cur_local_path + "\t" + cur_file);
}
}
}
unlock(lockName);
return 0;
}