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;
}
public static void SeedData(HBaseConfiguration conf) throws IOException {
HTable table = new HTable(conf, "people");
Put put = new Put(Bytes.toBytes("doe-john-m-12345"));
put.add(Bytes.toBytes("personal"), Bytes.toBytes("givenName"), Bytes.toBytes("John"));
put.add(Bytes.toBytes("personal"), Bytes.toBytes("mi"), Bytes.toBytes("M"));
put.add(Bytes.toBytes("personal"), Bytes.toBytes("surame"), Bytes.toBytes("Doe"));
put.add(
Bytes.toBytes("contactinfo"),
Bytes.toBytes("email"),
Bytes.toBytes("*****@*****.**"));
table.put(put);
table.flushCommits();
table.close();
}
public void putBatch(Optional<List<Request>> putRequests, boolean optimize) {
if (!valid) {
Logger.error("CANNOT PUT! NO VALID CONNECTION");
return;
}
List<Put> puts = new ArrayList<>();
if (putRequests.isPresent() && !putRequests.get().isEmpty()) {
String tableName = putRequests.get().get(0).table;
putRequests
.get()
.forEach(
pr ->
pr.getPut()
.ifPresent(
p -> {
if (optimize) {
p.setDurability(Durability.SKIP_WAL);
}
puts.add(p);
}));
try {
final Table table = connection.getTable(TableName.valueOf(tableName));
if (optimize && table instanceof HTable) {
((HTable) table).setAutoFlush(false, true);
}
table.put(puts);
table.close();
} catch (IOException e) {
e.printStackTrace();
}
}
}
@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);
}
}
}