@Override
public void analyzeInternal(ASTNode ast) throws SemanticException {
isLocal = false;
isOverWrite = false;
Tree fromTree = ast.getChild(0);
Tree tableTree = ast.getChild(1);
if (ast.getChildCount() == 4) {
isLocal = true;
isOverWrite = true;
}
if (ast.getChildCount() == 3) {
if (ast.getChild(2).getText().toLowerCase().equals("local")) {
isLocal = true;
} else {
isOverWrite = true;
}
}
// initialize load path
URI fromURI;
try {
String fromPath = stripQuotes(fromTree.getText());
fromURI = initializeFromURI(fromPath);
} catch (IOException e) {
throw new SemanticException(ErrorMsg.INVALID_PATH.getMsg(fromTree, e.getMessage()), e);
} catch (URISyntaxException e) {
throw new SemanticException(ErrorMsg.INVALID_PATH.getMsg(fromTree, e.getMessage()), e);
}
// initialize destination table/partition
tableSpec ts = new tableSpec(db, conf, (ASTNode) tableTree);
if (ts.tableHandle.isOffline()) {
throw new SemanticException(
ErrorMsg.OFFLINE_TABLE_OR_PARTITION.getMsg(":Table " + ts.tableName));
}
if (ts.tableHandle.isView()) {
throw new SemanticException(ErrorMsg.DML_AGAINST_VIEW.getMsg());
}
if (ts.tableHandle.isNonNative()) {
throw new SemanticException(ErrorMsg.LOAD_INTO_NON_NATIVE.getMsg());
}
if (ts.tableHandle.isStoredAsSubDirectories()) {
throw new SemanticException(ErrorMsg.LOAD_INTO_STORED_AS_DIR.getMsg());
}
URI toURI =
(ts.partHandle != null)
? ts.partHandle.getDataLocation()
: ts.tableHandle.getDataLocation();
List<FieldSchema> parts = ts.tableHandle.getPartitionKeys();
if ((parts != null && parts.size() > 0) && (ts.partSpec == null || ts.partSpec.size() == 0)) {
throw new SemanticException(ErrorMsg.NEED_PARTITION_ERROR.getMsg());
}
// make sure the arguments make sense
applyConstraints(fromURI, toURI, fromTree, isLocal);
Task<? extends Serializable> rTask = null;
// create copy work
if (isLocal) {
// if the local keyword is specified - we will always make a copy. this
// might seem redundant in the case
// that the hive warehouse is also located in the local file system - but
// that's just a test case.
String copyURIStr = ctx.getExternalTmpFileURI(toURI);
URI copyURI = URI.create(copyURIStr);
rTask = TaskFactory.get(new CopyWork(fromURI.toString(), copyURIStr), conf);
fromURI = copyURI;
}
// create final load/move work
String loadTmpPath = ctx.getExternalTmpFileURI(toURI);
Map<String, String> partSpec = ts.getPartSpec();
if (partSpec == null) {
partSpec = new LinkedHashMap<String, String>();
outputs.add(new WriteEntity(ts.tableHandle));
} else {
try {
Partition part = Hive.get().getPartition(ts.tableHandle, partSpec, false);
if (part != null) {
if (part.isOffline()) {
throw new SemanticException(
ErrorMsg.OFFLINE_TABLE_OR_PARTITION.getMsg(ts.tableName + ":" + part.getName()));
}
outputs.add(new WriteEntity(part));
} else {
outputs.add(new WriteEntity(ts.tableHandle));
}
} catch (HiveException e) {
throw new SemanticException(e);
}
}
LoadTableDesc loadTableWork =
new LoadTableDesc(
fromURI.toString(),
loadTmpPath,
Utilities.getTableDesc(ts.tableHandle),
partSpec,
isOverWrite);
Task<? extends Serializable> childTask =
TaskFactory.get(new MoveWork(getInputs(), getOutputs(), loadTableWork, null, true), conf);
if (rTask != null) {
rTask.addDependentTask(childTask);
} else {
rTask = childTask;
}
rootTasks.add(rTask);
// The user asked for stats to be collected.
// Some stats like number of rows require a scan of the data
// However, some other stats, like number of files, do not require a complete scan
// Update the stats which do not require a complete scan.
Task<? extends Serializable> statTask = null;
if (conf.getBoolVar(HiveConf.ConfVars.HIVESTATSAUTOGATHER)) {
StatsWork statDesc = new StatsWork(loadTableWork);
statDesc.setNoStatsAggregator(true);
statDesc.setClearAggregatorStats(true);
statDesc.setStatsReliable(conf.getBoolVar(HiveConf.ConfVars.HIVE_STATS_RELIABLE));
statTask = TaskFactory.get(statDesc, conf);
}
// HIVE-3334 has been filed for load file with index auto update
if (HiveConf.getBoolVar(conf, HiveConf.ConfVars.HIVEINDEXAUTOUPDATE)) {
IndexUpdater indexUpdater = new IndexUpdater(loadTableWork, getInputs(), conf);
try {
List<Task<? extends Serializable>> indexUpdateTasks = indexUpdater.generateUpdateTasks();
for (Task<? extends Serializable> updateTask : indexUpdateTasks) {
// LOAD DATA will either have a copy & move or just a move,
// we always want the update to be dependent on the move
childTask.addDependentTask(updateTask);
if (statTask != null) {
updateTask.addDependentTask(statTask);
}
}
} catch (HiveException e) {
console.printInfo(
"WARNING: could not auto-update stale indexes, indexes are not out of sync");
}
} else if (statTask != null) {
childTask.addDependentTask(statTask);
}
}
private List<HiveLockObj> getLockObjects(
QueryPlan plan, Database db, Table t, Partition p, HiveLockMode mode) throws LockException {
List<HiveLockObj> locks = new LinkedList<HiveLockObj>();
HiveLockObject.HiveLockObjectData lockData =
new HiveLockObject.HiveLockObjectData(
plan.getQueryId(),
String.valueOf(System.currentTimeMillis()),
"IMPLICIT",
plan.getQueryStr());
if (db != null) {
locks.add(new HiveLockObj(new HiveLockObject(db.getName(), lockData), mode));
return locks;
}
if (t != null) {
locks.add(new HiveLockObj(new HiveLockObject(t, lockData), mode));
mode = HiveLockMode.SHARED;
locks.add(new HiveLockObj(new HiveLockObject(t.getDbName(), lockData), mode));
return locks;
}
if (p != null) {
if (!(p instanceof DummyPartition)) {
locks.add(new HiveLockObj(new HiveLockObject(p, lockData), mode));
}
// All the parents are locked in shared mode
mode = HiveLockMode.SHARED;
// For dummy partitions, only partition name is needed
String name = p.getName();
if (p instanceof DummyPartition) {
name = p.getName().split("@")[2];
}
String partialName = "";
String[] partns = name.split("/");
int len = p instanceof DummyPartition ? partns.length : partns.length - 1;
Map<String, String> partialSpec = new LinkedHashMap<String, String>();
for (int idx = 0; idx < len; idx++) {
String partn = partns[idx];
partialName += partn;
String[] nameValue = partn.split("=");
assert (nameValue.length == 2);
partialSpec.put(nameValue[0], nameValue[1]);
try {
locks.add(
new HiveLockObj(
new HiveLockObject(
new DummyPartition(
p.getTable(),
p.getTable().getDbName()
+ "/"
+ MetaStoreUtils.encodeTableName(p.getTable().getTableName())
+ "/"
+ partialName,
partialSpec),
lockData),
mode));
partialName += "/";
} catch (HiveException e) {
throw new LockException(e.getMessage());
}
}
locks.add(new HiveLockObj(new HiveLockObject(p.getTable(), lockData), mode));
locks.add(new HiveLockObj(new HiveLockObject(p.getTable().getDbName(), lockData), mode));
}
return locks;
}