com.yahoo.bard.webservice.data.dimension.impl

Class LuceneSearchProvider

java.lang.Object

com.yahoo.bard.webservice.data.dimension.impl.LuceneSearchProvider

All Implemented Interfaces:

SearchProvider

public class LuceneSearchProvider
extends Object
implements SearchProvider

LuceneSearchProvider. Search provider which uses lucene.

Field Summary

Fields

Modifier and Type	Field and Description
static int	LUCENE_SEARCH_TIMEOUT_MS

Constructor Summary

Constructors

Constructor and Description
LuceneSearchProvider(String luceneIndexPath, int maxResults) Constructor.
LuceneSearchProvider(String luceneIndexPath, int maxResults, int searchTimeout) Constructor.

Method Summary

Modifier and Type	Method and Description
void	clearDimension() Clears the dimension cache, and resets the indices, effectively resetting the SearchProvider to a clean state.
Pagination<DimensionRow>	findAllDimensionRowsPaged(PaginationParameters paginationParameters) Return the desired page of dimension rows.
TreeSet<DimensionRow>	findAllOrderedDimensionRows() Getter for dimension rows in tree set for consistent order.
Pagination<DimensionRow>	findFilteredDimensionRowsPaged(Set<ApiFilter> filters, PaginationParameters paginationParameters) Return the desired page of dimension rows that match the specified filters.
int	getDimensionCardinality() Gets the number of distinct dimension rows (assuming the key field is unique) in the index.
boolean	isHealthy() Method to check if search provider is healthy.
void	refreshIndex(Map<String,HashDataCache.Pair<DimensionRow,DimensionRow>> changedRows) Method to add / update indexes.
void	refreshIndex(String rowId, DimensionRow dimensionRow, DimensionRow dimensionRowOld) Refresh lucene index
void	replaceIndex(String newLuceneIndexPathString) Replaces index with a new index in hot-swap manner.
void	setDimension(Dimension dimension) Setter for dimension.
void	setKeyValueStore(KeyValueStore keyValueStore) Setter for store.

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.yahoo.bard.webservice.data.dimension.SearchProvider

findAllDimensionRows, findFilteredDimensionRows, hasAnyRows

Field Detail

LUCENE_SEARCH_TIMEOUT_MS

public static final int LUCENE_SEARCH_TIMEOUT_MS

Constructor Detail

LuceneSearchProvider

public LuceneSearchProvider(String luceneIndexPath,
                            int maxResults,
                            int searchTimeout)

Constructor.

Parameters:

luceneIndexPath - Path to the lucene index files

maxResults - Maximum number of allowed results in a page

searchTimeout - Maximum time in milliseconds that a lucene search can run

LuceneSearchProvider

public LuceneSearchProvider(String luceneIndexPath,
                            int maxResults)

Constructor. The search timeout is initialized to the default (or configured) value.

Parameters:

luceneIndexPath - Path to the lucene index files

maxResults - Maximum number of allowed results in a page

Method Detail

setDimension

public void setDimension(Dimension dimension)

Description copied from interface: SearchProvider

Setter for dimension.

Specified by:

setDimension in interface SearchProvider

Parameters:

dimension - Dimension the SearchProvider is searching

setKeyValueStore

public void setKeyValueStore(KeyValueStore keyValueStore)

Description copied from interface: SearchProvider

Setter for store.

Specified by:

setKeyValueStore in interface SearchProvider

Parameters:

keyValueStore - KeyValueStore that holds the data rows indexed by the Search Provider

getDimensionCardinality

public int getDimensionCardinality()

Description copied from interface: SearchProvider

Gets the number of distinct dimension rows (assuming the key field is unique) in the index.

Specified by:

getDimensionCardinality in interface SearchProvider

Returns:

The number of dimension rows for this dimension

findAllDimensionRowsPaged

public Pagination<DimensionRow> findAllDimensionRowsPaged(PaginationParameters paginationParameters)

Description copied from interface: SearchProvider

Return the desired page of dimension rows.

Specified by:

findAllDimensionRowsPaged in interface SearchProvider

Parameters:

paginationParameters - The parameters that define a page (i.e. the number of rows per page, and the page of results desired)

Returns:

A page of the dimension rows contained in the Search Provider's indices Throws TimeoutException If the search times out.

findAllOrderedDimensionRows

public TreeSet<DimensionRow> findAllOrderedDimensionRows()

Description copied from interface: SearchProvider

Getter for dimension rows in tree set for consistent order. Throws TimeoutException If the search times out.

Specified by:

findAllOrderedDimensionRows in interface SearchProvider

Returns:

tree set of dimension rows the Search Provider has in it's indexes

isHealthy

public boolean isHealthy()

Description copied from interface: SearchProvider

Method to check if search provider is healthy.

Specified by:

isHealthy in interface SearchProvider

Returns:

true if healthy

refreshIndex

public void refreshIndex(String rowId,
                         DimensionRow dimensionRow,
                         DimensionRow dimensionRowOld)

Refresh lucene index

Query the lucene indexes on key column to see if there are any existing indexes.

If yes, then we need to update the indexes. Update is done by first deleting the existing documents and then adding new ones.

If not, then just add the documents to indexes.

Specified by:

refreshIndex in interface SearchProvider

Parameters:

rowId - key for id column

dimensionRow - the new dimension row

dimensionRowOld - the dimension row with which the index needs to be added / updated

refreshIndex

public void refreshIndex(Map<String,HashDataCache.Pair<DimensionRow,DimensionRow>> changedRows)

Description copied from interface: SearchProvider

Method to add / update indexes.

Specified by:

refreshIndex in interface SearchProvider

Parameters:

changedRows - Collection of newRow / oldRow pairs keyed by rowId to update the index with. Each row is a pair of dimension rows (newRow is the key, oldRow is the value), with the rowId (ie. unique key in the dimension) as the top-level Map key.

replaceIndex

public void replaceIndex(String newLuceneIndexPathString)

Description copied from interface: SearchProvider

Replaces index with a new index in hot-swap manner.

Specified by:

replaceIndex in interface SearchProvider

Parameters:

newLuceneIndexPathString - The location of directory that contains the the new index.

clearDimension

public void clearDimension()

Clears the dimension cache, and resets the indices, effectively resetting the SearchProvider to a clean state.

Note that this method attempts to acquire a write lock before clearing the index.

Specified by:

clearDimension in interface SearchProvider

findFilteredDimensionRowsPaged

public Pagination<DimensionRow> findFilteredDimensionRowsPaged(Set<ApiFilter> filters,
                                                               PaginationParameters paginationParameters)

Description copied from interface: SearchProvider

Return the desired page of dimension rows that match the specified filters.

Specified by:

findFilteredDimensionRowsPaged in interface SearchProvider

Parameters:

filters - ApiFilters to use for finding matching dimension rows

paginationParameters - The parameters that define a page (i.e. the number of rows per page, and the page of results desired) Throws TimeoutException If the search times out.

Returns:

A page of the dimension rows contained in the Search Provider's indices and matched by the specified filters.