Search in sources :

Example 31 with Partition

use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.

the class HCatClientHMSImpl method getPartitions.

@Override
public List<HCatPartition> getPartitions(String dbName, String tblName) throws HCatException {
    List<HCatPartition> hcatPtns = new ArrayList<HCatPartition>();
    try {
        HCatTable hcatTable = getTable(dbName, tblName);
        List<Partition> hivePtns = hmsClient.listPartitions(checkDB(dbName), tblName, (short) -1);
        for (Partition ptn : hivePtns) {
            hcatPtns.add(new HCatPartition(hcatTable, ptn));
        }
    } catch (NoSuchObjectException e) {
        throw new ObjectNotFoundException("NoSuchObjectException while retrieving partition.", e);
    } catch (MetaException e) {
        throw new HCatException("MetaException while retrieving partition.", e);
    } catch (TException e) {
        throw new ConnectionFailureException("TException while retrieving partition.", e);
    }
    return hcatPtns;
}
Also used : TException(org.apache.thrift.TException) Partition(org.apache.hadoop.hive.metastore.api.Partition) ArrayList(java.util.ArrayList) HCatException(org.apache.hive.hcatalog.common.HCatException) NoSuchObjectException(org.apache.hadoop.hive.metastore.api.NoSuchObjectException) MetaException(org.apache.hadoop.hive.metastore.api.MetaException)

Example 32 with Partition

use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.

the class TestMetastoreAuthorizationProvider method testSimplePrivileges.

public void testSimplePrivileges() throws Exception {
    if (!isTestEnabled()) {
        System.out.println("Skipping test " + this.getClass().getName());
        return;
    }
    String dbName = getTestDbName();
    String tblName = getTestTableName();
    String userName = setupUser();
    allowCreateDatabase(userName);
    CommandProcessorResponse ret = driver.run("create database " + dbName);
    assertEquals(0, ret.getResponseCode());
    Database db = msc.getDatabase(dbName);
    String dbLocn = db.getLocationUri();
    validateCreateDb(db, dbName);
    disallowCreateInDb(dbName, userName, dbLocn);
    disallowCreateDatabase(userName);
    driver.run("use " + dbName);
    ret = driver.run(String.format("create table %s (a string) partitioned by (b string)", tblName));
    assertEquals(1, ret.getResponseCode());
    // Even if table location is specified table creation should fail
    String tblNameLoc = tblName + "_loc";
    String tblLocation = new Path(dbLocn).getParent().toUri() + "/" + tblNameLoc;
    driver.run("use " + dbName);
    ret = driver.run(String.format("create table %s (a string) partitioned by (b string) location '" + tblLocation + "'", tblNameLoc));
    assertEquals(1, ret.getResponseCode());
    // failure from not having permissions to create table
    ArrayList<FieldSchema> fields = new ArrayList<FieldSchema>(2);
    fields.add(new FieldSchema("a", serdeConstants.STRING_TYPE_NAME, ""));
    Table ttbl = new Table();
    ttbl.setDbName(dbName);
    ttbl.setTableName(tblName);
    StorageDescriptor sd = new StorageDescriptor();
    ttbl.setSd(sd);
    sd.setCols(fields);
    sd.setParameters(new HashMap<String, String>());
    sd.getParameters().put("test_param_1", "Use this for comments etc");
    sd.setSerdeInfo(new SerDeInfo());
    sd.getSerdeInfo().setName(ttbl.getTableName());
    sd.getSerdeInfo().setParameters(new HashMap<String, String>());
    sd.getSerdeInfo().getParameters().put(org.apache.hadoop.hive.serde.serdeConstants.SERIALIZATION_FORMAT, "1");
    sd.getSerdeInfo().setSerializationLib(org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe.class.getName());
    sd.setInputFormat(HiveInputFormat.class.getName());
    sd.setOutputFormat(HiveOutputFormat.class.getName());
    ttbl.setPartitionKeys(new ArrayList<FieldSchema>());
    MetaException me = null;
    try {
        msc.createTable(ttbl);
    } catch (MetaException e) {
        me = e;
    }
    assertNoPrivileges(me);
    allowCreateInDb(dbName, userName, dbLocn);
    driver.run("use " + dbName);
    ret = driver.run(String.format("create table %s (a string) partitioned by (b string)", tblName));
    // now it succeeds.
    assertEquals(0, ret.getResponseCode());
    Table tbl = msc.getTable(dbName, tblName);
    validateCreateTable(tbl, tblName, dbName);
    // Table creation should succeed even if location is specified
    driver.run("use " + dbName);
    ret = driver.run(String.format("create table %s (a string) partitioned by (b string) location '" + tblLocation + "'", tblNameLoc));
    assertEquals(0, ret.getResponseCode());
    Table tblLoc = msc.getTable(dbName, tblNameLoc);
    validateCreateTable(tblLoc, tblNameLoc, dbName);
    String fakeUser = "mal";
    List<String> fakeGroupNames = new ArrayList<String>();
    fakeGroupNames.add("groupygroup");
    InjectableDummyAuthenticator.injectUserName(fakeUser);
    InjectableDummyAuthenticator.injectGroupNames(fakeGroupNames);
    InjectableDummyAuthenticator.injectMode(true);
    ret = driver.run(String.format("create table %s (a string) partitioned by (b string)", tblName + "mal"));
    assertEquals(1, ret.getResponseCode());
    ttbl.setTableName(tblName + "mal");
    me = null;
    try {
        msc.createTable(ttbl);
    } catch (MetaException e) {
        me = e;
    }
    assertNoPrivileges(me);
    disallowCreateInTbl(tbl.getTableName(), userName, tbl.getSd().getLocation());
    ret = driver.run("alter table " + tblName + " add partition (b='2011')");
    assertEquals(1, ret.getResponseCode());
    List<String> ptnVals = new ArrayList<String>();
    ptnVals.add("b=2011");
    Partition tpart = new Partition();
    tpart.setDbName(dbName);
    tpart.setTableName(tblName);
    tpart.setValues(ptnVals);
    tpart.setParameters(new HashMap<String, String>());
    tpart.setSd(tbl.getSd().deepCopy());
    tpart.getSd().setSerdeInfo(tbl.getSd().getSerdeInfo().deepCopy());
    tpart.getSd().setLocation(tbl.getSd().getLocation() + "/tpart");
    me = null;
    try {
        msc.add_partition(tpart);
    } catch (MetaException e) {
        me = e;
    }
    assertNoPrivileges(me);
    InjectableDummyAuthenticator.injectMode(false);
    allowCreateInTbl(tbl.getTableName(), userName, tbl.getSd().getLocation());
    ret = driver.run("alter table " + tblName + " add partition (b='2011')");
    assertEquals(0, ret.getResponseCode());
    allowDropOnTable(tblName, userName, tbl.getSd().getLocation());
    allowDropOnDb(dbName, userName, db.getLocationUri());
    ret = driver.run("drop database if exists " + getTestDbName() + " cascade");
    assertEquals(0, ret.getResponseCode());
    InjectableDummyAuthenticator.injectUserName(userName);
    InjectableDummyAuthenticator.injectGroupNames(Arrays.asList(ugi.getGroupNames()));
    InjectableDummyAuthenticator.injectMode(true);
    allowCreateDatabase(userName);
    driver.run("create database " + dbName);
    allowCreateInDb(dbName, userName, dbLocn);
    tbl.setTableType("EXTERNAL_TABLE");
    msc.createTable(tbl);
    disallowDropOnTable(tblName, userName, tbl.getSd().getLocation());
    ret = driver.run("drop table " + tbl.getTableName());
    assertEquals(1, ret.getResponseCode());
}
Also used : Path(org.apache.hadoop.fs.Path) Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) CommandProcessorResponse(org.apache.hadoop.hive.ql.processors.CommandProcessorResponse) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) HiveOutputFormat(org.apache.hadoop.hive.ql.io.HiveOutputFormat) HiveInputFormat(org.apache.hadoop.hive.ql.io.HiveInputFormat) Database(org.apache.hadoop.hive.metastore.api.Database) MetaException(org.apache.hadoop.hive.metastore.api.MetaException)

Example 33 with Partition

use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.

the class TestHBaseStoreIntegration method partitionStatistics.

@Test
public void partitionStatistics() throws Exception {
    long now = System.currentTimeMillis();
    String dbname = "default";
    String tableName = "statspart";
    String[] partNames = { "ds=today", "ds=yesterday" };
    String[] partVals = { "today", "yesterday" };
    String boolcol = "boolcol";
    String longcol = "longcol";
    String doublecol = "doublecol";
    String stringcol = "stringcol";
    String binarycol = "bincol";
    String decimalcol = "deccol";
    long trues = 37;
    long falses = 12;
    long booleanNulls = 2;
    long strMaxLen = 1234;
    double strAvgLen = 32.3;
    long strNulls = 987;
    long strDVs = 906;
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema(boolcol, "boolean", "nocomment"));
    cols.add(new FieldSchema(longcol, "long", "nocomment"));
    cols.add(new FieldSchema(doublecol, "double", "nocomment"));
    cols.add(new FieldSchema(stringcol, "varchar(32)", "nocomment"));
    cols.add(new FieldSchema(binarycol, "binary", "nocomment"));
    cols.add(new FieldSchema(decimalcol, "decimal(5, 3)", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
    List<FieldSchema> partCols = new ArrayList<FieldSchema>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbname, "me", (int) now / 1000, (int) now / 1000, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    for (String partVal : partVals) {
        Partition part = new Partition(Arrays.asList(partVal), dbname, tableName, (int) now / 1000, (int) now / 1000, sd, emptyParameters);
        store.addPartition(part);
    }
    for (int i = 0; i < partNames.length; i++) {
        ColumnStatistics stats = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc();
        desc.setLastAnalyzed(now);
        desc.setDbName(dbname);
        desc.setTableName(tableName);
        desc.setIsTblLevel(false);
        desc.setPartName(partNames[i]);
        stats.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName(boolcol);
        obj.setColType("boolean");
        ColumnStatisticsData data = new ColumnStatisticsData();
        BooleanColumnStatsData boolData = new BooleanColumnStatsData();
        boolData.setNumTrues(trues);
        boolData.setNumFalses(falses);
        boolData.setNumNulls(booleanNulls);
        data.setBooleanStats(boolData);
        obj.setStatsData(data);
        stats.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(stats, Arrays.asList(partVals[i]));
    }
    List<ColumnStatistics> statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(1, statsList.get(i).getStatsObjSize());
    }
    for (int i = 0; i < partNames.length; i++) {
        ColumnStatistics stats = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc();
        desc.setLastAnalyzed(now);
        desc.setDbName(dbname);
        desc.setTableName(tableName);
        desc.setIsTblLevel(false);
        desc.setPartName(partNames[i]);
        stats.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName(stringcol);
        obj.setColType("string");
        ColumnStatisticsData data = new ColumnStatisticsData();
        StringColumnStatsData strData = new StringColumnStatsData();
        strData.setMaxColLen(strMaxLen);
        strData.setAvgColLen(strAvgLen);
        strData.setNumNulls(strNulls);
        strData.setNumDVs(strDVs);
        data.setStringStats(strData);
        obj.setStatsData(data);
        stats.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(stats, Arrays.asList(partVals[i]));
    }
    // Make sure when we ask for one we only get one
    statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(1, statsList.get(i).getStatsObjSize());
    }
    statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol, stringcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(2, statsList.get(i).getStatsObjSize());
        // Just check one piece of the data, I don't need to check it all again
        Assert.assertEquals(booleanNulls, statsList.get(i).getStatsObj().get(0).getStatsData().getBooleanStats().getNumNulls());
        Assert.assertEquals(strDVs, statsList.get(i).getStatsObj().get(1).getStatsData().getStringStats().getNumDVs());
    }
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) BooleanColumnStatsData(org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData) Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) StringColumnStatsData(org.apache.hadoop.hive.metastore.api.StringColumnStatsData) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 34 with Partition

use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.

the class TestHBaseStoreIntegration method listPartitions.

@Test
public void listPartitions() throws Exception {
    String dbName = "default";
    String tableName = "listParts";
    int startTime = (int) (System.currentTimeMillis() / 1000);
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema("col1", "int", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
    List<FieldSchema> partCols = new ArrayList<FieldSchema>();
    partCols.add(new FieldSchema("pc", "string", ""));
    partCols.add(new FieldSchema("region", "string", ""));
    Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    String[][] partVals = new String[][] { { "today", "north america" }, { "tomorrow", "europe" } };
    for (String[] pv : partVals) {
        List<String> vals = new ArrayList<String>();
        for (String v : pv) vals.add(v);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/pc=" + pv[0] + "/region=" + pv[1]);
        Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
        store.addPartition(part);
    }
    List<String> names = store.listPartitionNames(dbName, tableName, (short) -1);
    Assert.assertEquals(2, names.size());
    String[] resultNames = names.toArray(new String[names.size()]);
    Arrays.sort(resultNames);
    Assert.assertArrayEquals(resultNames, new String[] { "pc=today/region=north america", "pc=tomorrow/region=europe" });
    List<Partition> parts = store.getPartitionsByNames(dbName, tableName, names);
    Assert.assertArrayEquals(partVals[0], parts.get(0).getValues().toArray(new String[2]));
    Assert.assertArrayEquals(partVals[1], parts.get(1).getValues().toArray(new String[2]));
    store.dropPartitions(dbName, tableName, names);
    List<Partition> afterDropParts = store.getPartitions(dbName, tableName, -1);
    Assert.assertEquals(0, afterDropParts.size());
}
Also used : Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) Test(org.junit.Test)

Example 35 with Partition

use of org.apache.hadoop.hive.metastore.api.Partition in project hive by apache.

the class TestHBaseStoreIntegration method listPartitionsWithPs.

@Test
public void listPartitionsWithPs() throws Exception {
    String dbName = "default";
    String tableName = "listPartitionsWithPs";
    int startTime = (int) (System.currentTimeMillis() / 1000);
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema("col1", "int", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
    List<FieldSchema> partCols = new ArrayList<FieldSchema>();
    partCols.add(new FieldSchema("ds", "string", ""));
    partCols.add(new FieldSchema("region", "string", ""));
    Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    String[][] partVals = new String[][] { { "today", "north america" }, { "today", "europe" }, { "tomorrow", "north america" }, { "tomorrow", "europe" } };
    for (String[] pv : partVals) {
        List<String> vals = new ArrayList<String>();
        for (String v : pv) vals.add(v);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/ds=" + pv[0] + "/region=" + pv[1]);
        Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
        store.addPartition(part);
    }
    // We only test listPartitionNamesPs since it calls listPartitionsPsWithAuth anyway.
    // Test the case where we completely specify the partition
    List<String> partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList(partVals[0]), (short) -1);
    Assert.assertEquals(1, partitionNames.size());
    Assert.assertEquals("ds=today/region=north america", partitionNames.get(0));
    // Leave off the last value of the partition
    partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList(partVals[0][0]), (short) -1);
    Assert.assertEquals(2, partitionNames.size());
    String[] names = partitionNames.toArray(new String[partitionNames.size()]);
    Arrays.sort(names);
    Assert.assertArrayEquals(new String[] { "ds=today/region=europe", "ds=today/region=north america" }, names);
    // Put a star in the last value of the partition
    partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList("today", "*"), (short) -1);
    Assert.assertEquals(2, partitionNames.size());
    names = partitionNames.toArray(new String[partitionNames.size()]);
    Arrays.sort(names);
    Assert.assertArrayEquals(new String[] { "ds=today/region=europe", "ds=today/region=north america" }, names);
    // Put a star in the first value of the partition
    partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList("*", "europe"), (short) -1);
    Assert.assertEquals(2, partitionNames.size());
    names = partitionNames.toArray(new String[partitionNames.size()]);
    Arrays.sort(names);
    Assert.assertArrayEquals(new String[] { "ds=today/region=europe", "ds=tomorrow/region=europe" }, names);
}
Also used : Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) Test(org.junit.Test)

Aggregations

Partition (org.apache.hadoop.hive.metastore.api.Partition)466 Test (org.junit.Test)311 MetastoreCheckinTest (org.apache.hadoop.hive.metastore.annotation.MetastoreCheckinTest)213 Table (org.apache.hadoop.hive.metastore.api.Table)207 ArrayList (java.util.ArrayList)202 FieldSchema (org.apache.hadoop.hive.metastore.api.FieldSchema)97 StorageDescriptor (org.apache.hadoop.hive.metastore.api.StorageDescriptor)88 MetaException (org.apache.hadoop.hive.metastore.api.MetaException)72 SerDeInfo (org.apache.hadoop.hive.metastore.api.SerDeInfo)71 Path (org.apache.hadoop.fs.Path)56 List (java.util.List)55 NoSuchObjectException (org.apache.hadoop.hive.metastore.api.NoSuchObjectException)51 PartitionSpecProxy (org.apache.hadoop.hive.metastore.partition.spec.PartitionSpecProxy)41 Database (org.apache.hadoop.hive.metastore.api.Database)38 HashMap (java.util.HashMap)36 IOException (java.io.IOException)34 TException (org.apache.thrift.TException)34 ColumnStatistics (org.apache.hadoop.hive.metastore.api.ColumnStatistics)33 ColumnStatisticsObj (org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj)33 ColumnStatisticsDesc (org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc)31