@Test
public void test() throws Exception {
log.info("You can ignore exceptions and scary stack traces from this test");
IdService idService = new CachingIdService(5, new MapIdService(), "test");
Configuration conf = getTestUtil().getConfiguration();
HTablePool pool = new HTablePool(conf, Integer.MAX_VALUE);
DbHarness<LongOp> dbHarness =
new HBaseDbHarness<LongOp>(
pool,
"XY".getBytes(),
"nonexistentTable".getBytes(),
"nonexistentCf".getBytes(),
LongOp.DESERIALIZER,
idService,
CommitType.INCREMENT,
5,
2,
2,
null);
DataCube<LongOp> cube;
Dimension<String> zipcode =
new Dimension<String>("zipcode", new StringToBytesBucketer(), true, 5);
Rollup zipRollup = new Rollup(zipcode);
List<Dimension<?>> dimensions = ImmutableList.<Dimension<?>>of(zipcode);
List<Rollup> rollups = ImmutableList.of(zipRollup);
cube = new DataCube<LongOp>(dimensions, rollups);
DataCubeIo<LongOp> dataCubeIo =
new DataCubeIo<LongOp>(cube, dbHarness, 1, Long.MAX_VALUE, SyncLevel.BATCH_ASYNC);
dataCubeIo.writeAsync(new LongOp(1), new WriteBuilder(cube).at(zipcode, "97212"));
dataCubeIo.writeAsync(new LongOp(1), new WriteBuilder(cube).at(zipcode, "97212"));
dataCubeIo.flush();
try {
dataCubeIo.writeAsync(new LongOp(1), new WriteBuilder(cube).at(zipcode, "97212"));
Assert.fail("Cube should not have accepted more writes after an error!");
} catch (AsyncException e) {
// This exception *should* happen. Because we wrote to a nonexistent table.
Assert.assertTrue(e.getCause().getCause().getCause() instanceof TableNotFoundException);
}
}
/** Get the number of hashtag occurrences in the given time bucket. */
protected long getTagTimeCount(String hashtag, BucketType timeBucketType, DateTime dateTime)
throws IOException, InterruptedException {
Optional<LongOp> optCount =
dataCubeIo.get(
new ReadBuilder(dataCube)
.at(tagsDimension, hashtag)
.at(timeDimension, timeBucketType, dateTime));
return unpackOrZero(optCount);
}
/** Get the number of times that retweeterUser retweeted a tweet by sourceUser. */
public long getRetweetsOfBy(String sourceUser, String retweeterUser)
throws IOException, InterruptedException {
Optional<LongOp> optCount =
dataCubeIo.get(
new ReadBuilder(dataCube)
.at(retweetedFromDimension, sourceUser)
.at(userDimension, retweeterUser));
return unpackOrZero(optCount);
}
/** Get the number of tweets sent by the given user on the given day. */
public long getUserDayCount(String userName, DateTime day)
throws InterruptedException, IOException {
Optional<LongOp> optCount =
dataCubeIo.get(
new ReadBuilder(dataCube)
.at(userDimension, userName)
.at(timeDimension, HourDayMonthBucketer.days, day));
return unpackOrZero(optCount);
}
/**
* Do all the increments necessary to add a tweet to the datacube. May not immediately flush to
* the DB.
*/
public void countTweet(Tweet tweet) throws IOException, InterruptedException, AsyncException {
WriteBuilder writeBuilder =
new WriteBuilder(dataCube)
.at(timeDimension, tweet.time)
.at(userDimension, tweet.username)
.at(retweetedFromDimension, tweet.retweetedFrom.or(""))
.at(tagsDimension, tweet.hashTags);
Batch<LongOp> cubeUpdates = dataCube.getWrites(writeBuilder, new LongOp(1));
dataCubeIo.writeAsync(cubeUpdates);
}
/** Write all batched/cached changes to the backing database. */
public void flush() throws InterruptedException {
dataCubeIo.flush();
}
/** Get the number of tweets that included the given hashtag. */
public long getTagCount(String hashtag) throws IOException, InterruptedException {
Optional<LongOp> optCount =
dataCubeIo.get(new ReadBuilder(dataCube).at(tagsDimension, hashtag));
return unpackOrZero(optCount);
}
/** Get the total number of tweets sent by the given user. */
public long getUserCount(String userName) throws InterruptedException, IOException {
Optional<LongOp> optCount =
dataCubeIo.get(new ReadBuilder(dataCube).at(userDimension, userName));
return unpackOrZero(optCount);
}
/** Get the total number of tweets. */
public long getCount() throws InterruptedException, IOException {
return dataCubeIo.get(new ReadBuilder(dataCube)).or(new LongOp(0)).getLong();
}