net.opentsdb.tree

Class TreeBuilder

java.lang.Object

net.opentsdb.tree.TreeBuilder

public final class TreeBuilder
extends Object

Contains the logic and methods for building a branch from a tree definition and a TSMeta object. Use the class by loading a tree, passing it to the builder constructor, and call processTimeseriesMeta(net.opentsdb.meta.TSMeta) with a TSMeta object.

When processing, the builder runs the meta data through each of the rules in the rule set and recursively builds a tree. After running through all of the rules, if valid results were obtained, each branch is saved to storage if they haven't been processed before (in the processed_branches map). If a leaf was found, it will be saved. If any collisions or not-matched reports occurred, they will be saved to storage.

If processTimeseriesMeta(net.opentsdb.meta.TSMeta) is called with the testing flag, the tree will be built but none of the branches will be stored. This is used for RPC calls to display the results to a user and test_messages will contain a detailed description of the processing results.

Warning: This class is not thread safe. It should only be used by a single thread to process a TSMeta at a time. If processing multiple TSMetas you can create the builder and run all of the meta objects through the process methods.

Since:

2.0

Constructor Summary

Constructors

Constructor and Description
TreeBuilder(TSDB tsdb, Tree tree) Constructor to initialize the builder.

Method Summary

Methods

Modifier and Type	Method and Description
Branch	getRootBranch()
ArrayList<String>	getTestMessage()
Tree	getTree()
static com.stumbleupon.async.Deferred<Branch>	loadOrInitializeRoot(TSDB tsdb, int tree_id, boolean is_testing) Attempts to retrieve or initialize the root branch for the configured tree.
static com.stumbleupon.async.Deferred<Boolean>	processAllTrees(TSDB tsdb, TSMeta meta) Attempts to run the given TSMeta object through all of the trees in the system.
com.stumbleupon.async.Deferred<ArrayList<Boolean>>	processTimeseriesMeta(TSMeta meta) Convenience overload of processTimeseriesMeta(TSMeta, boolean) that sets the testing flag to false.
com.stumbleupon.async.Deferred<ArrayList<Boolean>>	processTimeseriesMeta(TSMeta meta, boolean is_testing) Runs the TSMeta object through the Trees rule set, optionally storing the resulting branches, leaves and meta data.
void	setTree(Tree tree)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

TreeBuilder

public TreeBuilder(TSDB tsdb,
           Tree tree)

Constructor to initialize the builder. Also calculates the max_rule_level based on the tree's rules

Parameters:

tsdb - The TSDB to use for access

tree - A tree with rules configured and ready for parsing

Method Detail

processTimeseriesMeta

public com.stumbleupon.async.Deferred<ArrayList<Boolean>> processTimeseriesMeta(TSMeta meta)

Convenience overload of processTimeseriesMeta(TSMeta, boolean) that sets the testing flag to false. Any changes processed from this method will be saved to storage

Parameters:

meta - The timeseries meta object to process

Returns:

A list of deferreds to wait on for storage completion

Throws:

IllegalArgumentException - if the tree has not been set or is invalid

processTimeseriesMeta

public com.stumbleupon.async.Deferred<ArrayList<Boolean>> processTimeseriesMeta(TSMeta meta,
                                                                       boolean is_testing)

Runs the TSMeta object through the Trees rule set, optionally storing the resulting branches, leaves and meta data. If the testing flag is set, no results will be saved but the caller can fetch the root branch from this object as it will contain the tree that would result from the processing. Also, the test_messages list will contain details about the process for debugging purposes.

Parameters:

meta - The timeseries meta object to process

is_testing - Whether or not changes should be written to storage. If false, resulting branches and leaves will be saved. If true, results will not be flushed to storage.

Returns:

A list of deferreds to wait on for storage completion

Throws:

IllegalArgumentException - if the tree has not been set or is invalid

org.hbase.async.HBaseException - if a storage exception occurred

loadOrInitializeRoot

public static com.stumbleupon.async.Deferred<Branch> loadOrInitializeRoot(TSDB tsdb,
                                                          int tree_id,
                                                          boolean is_testing)

Attempts to retrieve or initialize the root branch for the configured tree. If the is_testing flag is false, the root will be saved if it has to be created. The new or existing root branch will be stored to the local root object. Note: This will also cache the root in the local store since we don't want to keep loading on every TSMeta during real-time processing

Parameters:

tsdb - The tsdb to use for storage calls

tree_id - ID of the tree the root should be fetched/initialized for

is_testing - Whether or not the root should be written to storage if initialized.

Returns:

True if loading or initialization was successful.

processAllTrees

public static com.stumbleupon.async.Deferred<Boolean> processAllTrees(TSDB tsdb,
                                                      TSMeta meta)

Attempts to run the given TSMeta object through all of the trees in the system.

Parameters:

tsdb - The TSDB to use for access

meta - The timeseries meta object to process

Returns:

A meaningless deferred to wait on for all trees to process the meta object

Throws:

IllegalArgumentException - if the tree has not been set or is invalid

org.hbase.async.HBaseException - if a storage exception occurred

getTree

public Tree getTree()

Returns:

the local tree object

getRootBranch

public Branch getRootBranch()

Returns:

the root object

getTestMessage

public ArrayList<String> getTestMessage()

Returns:

the list of test message results

setTree

public void setTree(Tree tree)

Parameters:

tree - The tree to store locally