@Test
public void testSkippingHeaderWithText() throws IOException {
TableMeta meta = CatalogUtil.newTableMeta("TEXT");
meta.putOption(StorageConstants.TEXT_SKIP_HEADER_LINE, "1");
meta.putOption(StorageConstants.TEXT_DELIMITER, ",");
FileFragment fragment = getFileFragment("testSkip.txt");
Scanner scanner = TablespaceManager.getLocalFs().getScanner(meta, schema, fragment);
scanner.init();
int lines = 0;
try {
while (true) {
Tuple tuple = scanner.next();
if (tuple != null) {
assertEquals(17 + lines, tuple.getInt2(2));
lines++;
} else break;
}
} finally {
assertEquals(6, lines);
scanner.close();
}
}
@Test
public void testIgnoreTruncatedValueErrorTolerance() throws IOException {
TajoConf conf = new TajoConf();
TableMeta meta = CatalogUtil.newTableMeta("JSON");
meta.putOption(StorageUtil.TEXT_ERROR_TOLERANCE_MAXNUM, "1");
FileFragment fragment = getFileFragment("testErrorTolerance3.json");
Scanner scanner = TablespaceManager.getLocalFs().getScanner(meta, schema, fragment);
scanner.init();
try {
Tuple tuple = scanner.next();
assertNull(tuple);
} finally {
scanner.close();
}
}
@Test
public void testNoErrorTolerance() throws IOException {
TajoConf conf = new TajoConf();
TableMeta meta = CatalogUtil.newTableMeta("JSON");
meta.putOption(StorageUtil.TEXT_ERROR_TOLERANCE_MAXNUM, "0");
FileFragment fragment = getFileFragment("testErrorTolerance2.json");
Scanner scanner = TablespaceManager.getLocalFs().getScanner(meta, schema, fragment);
scanner.init();
try {
scanner.next();
} catch (IOException ioe) {
return;
} finally {
scanner.close();
}
fail();
}
@Test
public void testIgnoreAllErrors() throws IOException {
TajoConf conf = new TajoConf();
TableMeta meta = CatalogUtil.newTableMeta("JSON");
meta.putOption(StorageUtil.TEXT_ERROR_TOLERANCE_MAXNUM, "-1");
FileFragment fragment = getFileFragment("testErrorTolerance1.json");
Scanner scanner = TablespaceManager.getLocalFs().getScanner(meta, schema, fragment);
scanner.init();
Tuple tuple;
int i = 0;
while ((tuple = scanner.next()) != null) {
assertEquals(baseTuple, tuple);
i++;
}
assertEquals(3, i);
scanner.close();
}
@Before
public void setUp() throws Exception {
util = new TajoTestingCluster();
util.initTestDir();
catalog = util.startCatalogCluster().getCatalog();
testDir = CommonTestingUtil.getTestDir(TEST_PATH);
catalog.createTablespace(DEFAULT_TABLESPACE_NAME, testDir.toUri().toString());
catalog.createDatabase(DEFAULT_DATABASE_NAME, DEFAULT_TABLESPACE_NAME);
conf = util.getConfiguration();
// ----------------- dep3 ------------------------------
// dep_id | dep_name | loc_id
// --------------------------------
// 0 | dep_0 | 1000
// 1 | dep_1 | 1001
// 2 | dep_2 | 1002
// 3 | dep_3 | 1003
// 4 | dep_4 | 1004
// 5 | dep_5 | 1005
// 6 | dep_6 | 1006
// 7 | dep_7 | 1007
// 8 | dep_8 | 1008
// 9 | dep_9 | 1009
Schema dep3Schema = new Schema();
dep3Schema.addColumn("dep_id", Type.INT4);
dep3Schema.addColumn("dep_name", Type.TEXT);
dep3Schema.addColumn("loc_id", Type.INT4);
TableMeta dep3Meta = CatalogUtil.newTableMeta("CSV");
Path dep3Path = new Path(testDir, "dep3.csv");
Appender appender1 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(dep3Meta, dep3Schema, dep3Path);
appender1.init();
VTuple tuple = new VTuple(dep3Schema.size());
for (int i = 0; i < 10; i++) {
tuple.put(
new Datum[] {
DatumFactory.createInt4(i),
DatumFactory.createText("dept_" + i),
DatumFactory.createInt4(1000 + i)
});
appender1.addTuple(tuple);
}
appender1.flush();
appender1.close();
dep3 = CatalogUtil.newTableDesc(DEP3_NAME, dep3Schema, dep3Meta, dep3Path);
catalog.createTable(dep3);
// ----------------- dep4 ------------------------------
// dep_id | dep_name | loc_id
// --------------------------------
// 0 | dep_0 | 1000
// 1 | dep_1 | 1001
// 2 | dep_2 | 1002
// 3 | dep_3 | 1003
// 4 | dep_4 | 1004
// 5 | dep_5 | 1005
// 6 | dep_6 | 1006
// 7 | dep_7 | 1007
// 8 | dep_8 | 1008
// 9 | dep_9 | 1009
// 10 | dep_10 | 1010
Schema dep4Schema = new Schema();
dep4Schema.addColumn("dep_id", Type.INT4);
dep4Schema.addColumn("dep_name", Type.TEXT);
dep4Schema.addColumn("loc_id", Type.INT4);
TableMeta dep4Meta = CatalogUtil.newTableMeta("CSV");
Path dep4Path = new Path(testDir, "dep4.csv");
Appender appender4 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(dep4Meta, dep4Schema, dep4Path);
appender4.init();
VTuple tuple4 = new VTuple(dep4Schema.size());
for (int i = 0; i < 11; i++) {
tuple4.put(
new Datum[] {
DatumFactory.createInt4(i),
DatumFactory.createText("dept_" + i),
DatumFactory.createInt4(1000 + i)
});
appender4.addTuple(tuple4);
}
appender4.flush();
appender4.close();
dep4 = CatalogUtil.newTableDesc(DEP4_NAME, dep4Schema, dep4Meta, dep4Path);
catalog.createTable(dep4);
// ----------------- job3 ------------------------------
// job_id | job_title
// ----------------------
// 101 | job_101
// 102 | job_102
// 103 | job_103
Schema job3Schema = new Schema();
job3Schema.addColumn("job_id", Type.INT4);
job3Schema.addColumn("job_title", Type.TEXT);
TableMeta job3Meta = CatalogUtil.newTableMeta("CSV");
Path job3Path = new Path(testDir, "job3.csv");
Appender appender2 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(job3Meta, job3Schema, job3Path);
appender2.init();
VTuple tuple2 = new VTuple(job3Schema.size());
for (int i = 1; i < 4; i++) {
int x = 100 + i;
tuple2.put(
new Datum[] {DatumFactory.createInt4(100 + i), DatumFactory.createText("job_" + x)});
appender2.addTuple(tuple2);
}
appender2.flush();
appender2.close();
job3 = CatalogUtil.newTableDesc(JOB3_NAME, job3Schema, job3Meta, job3Path);
catalog.createTable(job3);
// ---------------------emp3 --------------------
// emp_id | first_name | last_name | dep_id | salary | job_id
// ------------------------------------------------------------
// 11 | fn_11 | ln_11 | 1 | 123 | 101
// 13 | fn_13 | ln_13 | 3 | 369 | 103
// 15 | fn_15 | ln_15 | 5 | 615 | null
// 17 | fn_17 | ln_17 | 7 | 861 | null
// 19 | fn_19 | ln_19 | 9 | 1107 | null
// 21 | fn_21 | ln_21 | 1 | 123 | 101
// 23 | fn_23 | ln_23 | 3 | 369 | 103
Schema emp3Schema = new Schema();
emp3Schema.addColumn("emp_id", Type.INT4);
emp3Schema.addColumn("first_name", Type.TEXT);
emp3Schema.addColumn("last_name", Type.TEXT);
emp3Schema.addColumn("dep_id", Type.INT4);
emp3Schema.addColumn("salary", Type.FLOAT4);
emp3Schema.addColumn("job_id", Type.INT4);
TableMeta emp3Meta = CatalogUtil.newTableMeta("CSV");
Path emp3Path = new Path(testDir, "emp3.csv");
Appender appender3 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(emp3Meta, emp3Schema, emp3Path);
appender3.init();
VTuple tuple3 = new VTuple(emp3Schema.size());
for (int i = 1; i < 4; i += 2) {
int x = 10 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(10 + i),
DatumFactory.createText("firstname_" + x),
DatumFactory.createText("lastname_" + x),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createInt4(100 + i)
});
appender3.addTuple(tuple3);
int y = 20 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(20 + i),
DatumFactory.createText("firstname_" + y),
DatumFactory.createText("lastname_" + y),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createInt4(100 + i)
});
appender3.addTuple(tuple3);
}
for (int i = 5; i < 10; i += 2) {
int x = 10 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(10 + i),
DatumFactory.createText("firstname_" + x),
DatumFactory.createText("lastname_" + x),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createNullDatum()
});
appender3.addTuple(tuple3);
}
appender3.flush();
appender3.close();
emp3 = CatalogUtil.newTableDesc(EMP3_NAME, emp3Schema, emp3Meta, emp3Path);
catalog.createTable(emp3);
// ---------------------phone3 --------------------
// emp_id | phone_number
// -----------------------------------------------
// this table is empty, no rows
Schema phone3Schema = new Schema();
phone3Schema.addColumn("emp_id", Type.INT4);
phone3Schema.addColumn("phone_number", Type.TEXT);
TableMeta phone3Meta = CatalogUtil.newTableMeta("CSV");
Path phone3Path = new Path(testDir, "phone3.csv");
Appender appender5 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(phone3Meta, phone3Schema, phone3Path);
appender5.init();
appender5.flush();
appender5.close();
phone3 = CatalogUtil.newTableDesc(PHONE3_NAME, phone3Schema, phone3Meta, phone3Path);
catalog.createTable(phone3);
analyzer = new SQLAnalyzer();
planner = new LogicalPlanner(catalog, TablespaceManager.getInstance());
defaultContext = LocalTajoTestingUtility.createDummyContext(conf);
}