The API to JASS's anytime seach engine. More...

#include <JASS_anytime_api.h>

Inheritance diagram for JASS_anytime_api:

Collaboration diagram for JASS_anytime_api:

Classes
class	thread_data

Public Member Functions
	JASS_anytime_api ()
	Constructor.

virtual	~JASS_anytime_api ()
	Destructor.

JASS_ERROR	load_index (size_t index_version, const std::string &directory="", bool verbose=false)
	Load a JASS index from the given directory. More...

uint32_t	get_document_count (void)
	Return the number of documents in the index. More...

JASS_ERROR	get_encoding_scheme (std::string &codex_name, int32_t &d_ness)
	Return the name of the compression algorithm and the delta (d-gap) size. More...

std::string	get_encoding_scheme_name (void)
	Return the name of the compression algorithm used on this index. More...

int32_t	get_encoding_scheme_d (void)
	Return the d-gap value used in this index. More...

void	set_accumulator_manager (const std::string &name)
	Set the name of the accumulator manager to use. More...

JASS_ERROR	set_postings_to_process_proportion (double percent)
	Set the number of postings to process as a proportion of the number of documents in the collection. More...

JASS_ERROR	set_postings_to_process_relative (double percent)
	Set the number of postings to process as a proportion of the number of postings for this query. More...

JASS_ERROR	set_postings_to_process (size_t count)
	Set the number of postings to process as an absolute number. More...

uint32_t	get_postings_to_process (void)
	Return the current maimum number of postings to process value. This might be very large. More...

uint32_t	get_max_top_k (void)
	Return the largest possible top_k value, which might be smaller than the number of documents in the collection. More...

JASS_ERROR	set_top_k (size_t k)
	Set the maximum numbre of documents to return in a resuts list. More...

size_t	get_top_k (void)
	Return the current top-k value. More...

JASS_ERROR	set_accumulator_width (size_t width)
	Set the accumulator page-table width (assuming a page-table like approach is being used) More...

JASS_ERROR	use_ascii_parser (void)
	Use the query parser that seperates tokens on whitespace along (this method is not normally needed). More...

JASS_ERROR	use_query_parser (void)
	Use the default query parser that understands alphanumnerics, spaces, unicode, and so on (this method is not normally needed). More...

JASS_anytime_result	search (const std::string &query)
	Search using the current index and the current parameters. More...

JASS_ERROR	search (std::vector< JASS_anytime_thread_result > &output, std::vector< JASS_anytime_query > &query_list, size_t thread_count)
	Search using the current index and the current parameters. More...

std::vector< JASS_anytime_thread_result >	threaded_search (std::vector< std::string > &query_list, size_t thread_count)
	Search using the current index and the current parameters. More...

Private Member Functions
	JASS_anytime_api (JASS_anytime_api &from)=delete

	JASS_anytime_api (JASS_anytime_api &&from)=delete

JASS_anytime_api &	operator= (JASS_anytime_api &from)=delete

JASS_anytime_api &	operator= (JASS_anytime_api &&from)=delete

void	anytime (JASS_anytime_thread_result &output, std::vector< JASS_anytime_query > &query_list, size_t thread_number=0)
	This method calls into the search engine with a set of queries and retrieves a set of results for each. More...

thread_data &	get_thread_local_data (size_t thread_number)
	Return the expensive-to-initialise thread local data that has already been allocated on index load. More...

Static Private Member Functions
static void	anytime_bootstrap (JASS_anytime_api *thiss, JASS_anytime_thread_result &output, std::vector< JASS_anytime_query > &query_list, size_t thread_number)
	Bootstrapping method for a thread to call into anytime() More...

Private Attributes
JASS::deserialised_jass_v1 *	index
	The index.

size_t	postings_to_process
	The maximunm number of postings to process.

double	relative_postings_to_process
	If not 1 then then this is the proportion of this query's postings that should be processed.

size_t	top_k
	The number of documents we want in the results list.

JASS::parser_query::parser_type	which_query_parser
	Use the simple ASCII parser or the regular query parser.

size_t	accumulator_width
	Width of the accumulator array.

JASS_anytime_stats	stats
	Stats for this "session".

std::map< size_t, thread_data >	thread_local_data
	Data needed by each thread (the accumulators array, etc)

std::string	accumulator_manager
	The name of the accumulator manager.

Static Private Attributes
static constexpr size_t	MAX_QUANTUM = 0x0FFF
	The maximum number of segments in a query.

static constexpr size_t	MAX_TERMS_PER_QUERY = 1024
	The maximum number of terms in a query.

Detailed Description

The API to JASS's anytime seach engine.

Member Function Documentation

◆ anytime()

void JASS_anytime_api::anytime	(	JASS_anytime_thread_result &	output,
		std::vector< JASS_anytime_query > &	query_list,
		size_t	thread_number = `0`
	)

private

This method calls into the search engine with a set of queries and retrieves a set of results for each.

Parameters

output	[out] The results for each query
query_list	[in] The list of queries to perform

◆ anytime_bootstrap()

void JASS_anytime_api::anytime_bootstrap	(	JASS_anytime_api *	thiss,
		JASS_anytime_thread_result &	output,
		std::vector< JASS_anytime_query > &	query_list,
		size_t	thread_number
	)

staticprivate

Bootstrapping method for a thread to call into anytime()

Parameters

thiss	[in] Pointer to the object to call into
output	[out] The results for each query
query_ist	[in] The list of queries to perform
thread_number	[in] The ID of this thread

◆ get_document_count()

JASS::query::DOCID_TYPE JASS_anytime_api::get_document_count ( void )

Return the number of documents in the index.

Returns: The number of documetns in the collection, or 0 if no index has been loaded

◆ get_encoding_scheme()

JASS_ERROR JASS_anytime_api::get_encoding_scheme	(	std::string &	codex_name,
		int32_t &	d_ness
	)

Return the name of the compression algorithm and the delta (d-gap) size.

This method might not be exposed to languages (such as Python) that do not support non-const references to strings, use get_encoding_scheme_name() or get_encoding_scheme_d() instead.

Parameters

codex_name	[out] The compression scheme
d_ness	[out] The d-gap size (normally 1)]

Returns: JASS_ERROR_OK or JASS_ERROR_NO_INDEX if an index has not yet been loaded

◆ get_encoding_scheme_d()

int32_t JASS_anytime_api::get_encoding_scheme_d ( void )

Return the d-gap value used in this index.

Returns: The d-gap value or JASS_ERROR_NO_INDEX if no index has been loaded.

◆ get_encoding_scheme_name()

std::string JASS_anytime_api::get_encoding_scheme_name ( void )

Return the name of the compression algorithm used on this index.

Returns: The name of the compression algorithm, or an empty string if no index has been loaded

◆ get_max_top_k()

JASS::query::DOCID_TYPE JASS_anytime_api::get_max_top_k ( void )

Return the largest possible top_k value, which might be smaller than the number of documents in the collection.

Returns: The maximum top-k value (normally 1000)

◆ get_postings_to_process()

uint32_t JASS_anytime_api::get_postings_to_process ( void )

Return the current maimum number of postings to process value. This might be very large.

Returns: The current maimum number of postings to process in a query.

◆ get_thread_local_data()

JASS_anytime_api::thread_data & JASS_anytime_api::get_thread_local_data ( size_t thread_number )

private

Return the expensive-to-initialise thread local data that has already been allocated on index load.

Parameters

thread_number [in] The therad number (counts from 0)

Returns: Thread local data

◆ get_top_k()

size_t JASS_anytime_api::get_top_k ( void )

Return the current top-k value.

Returns: The current top-k value, the maximum number of documents to put into the results list

◆ load_index()

JASS_ERROR JASS_anytime_api::load_index	(	size_t	index_version,
		const std::string &	directory = `""`,
		bool	verbose = `false`
	)

Load a JASS index from the given directory.

Parameters

index_version	[in] What verison of the index is this - normally 2.
directory[in]	The path to the index, default = "."
verbose	[in] if true, diagnostics are printed while the index is loading, default = false

Returns: JASS_ERROR_OK on success, else an error code.

◆ search() [1/2]

JASS_anytime_result JASS_anytime_api::search ( const std::string & query )

Search using the current index and the current parameters.

Parameters

query[in] The query. If the query starts with a numner that number is assumed to be the query_id and NOT a search term (so it is not searched for).

Returns: a JASS_anytime_result object with the results. If not index has been loaded then an empty JASS_anytime_result is returned.

◆ search() [2/2]

JASS_ERROR JASS_anytime_api::search	(	std::vector< JASS_anytime_thread_result > &	output,
		std::vector< JASS_anytime_query > &	query_list,
		size_t	thread_count
	)

Search using the current index and the current parameters.

This method might not be exposed to languages (such as Python) that do not support non-const references to objects. Search one query at a time using search() and do the threading in the calling laguage if need be.

Parameters

output	[out] a vector of results, one for each thread
query_list	[in] A vector of queries to be spread over all the threads
thread_count	[in] The number of threads to use for searching

Returns: JASS_ERROR_OK, or JASS_ERROR_NO_INDEX if no index has been loaded.

◆ set_accumulator_manager()

void JASS_anytime_api::set_accumulator_manager ( const std::string & name )

inline

Set the name of the accumulator manager to use.

Parameters

[in] name The name of the manager to use

◆ set_accumulator_width()

JASS_ERROR JASS_anytime_api::set_accumulator_width ( size_t width )

Set the accumulator page-table width (assuming a page-table like approach is being used)

Parameters

width [in[ The width of the table will be set to 2^width

Returns: JASS_ERROR_OK, or JASS_ERROR_TOO_LARGE if width is too large

◆ set_postings_to_process()

JASS_ERROR JASS_anytime_api::set_postings_to_process ( size_t count )

Set the number of postings to process as an absolute number.

An index does not need to be loaded first. Which ever of set_postings_to_process() and set_postings_to_process_proportion() is called last takes presidence. By default all postings are processed.

Parameters

count [in] The maximum number of postings to process

Returns: JASS_ERROR_OK

◆ set_postings_to_process_proportion()

JASS_ERROR JASS_anytime_api::set_postings_to_process_proportion ( double percent )

Set the number of postings to process as a proportion of the number of documents in the collection.

An index must be loaded before this method is called, if not it returns JASS_ERROR_NO_INDEX and has no effect Which ever of set_postings_to_process() and set_postings_to_process_proportion() is called last takes presidence. By default all postings are processed.

Parameters

percent [in] The percent to use (for example, 10 is use 10% of the postings)

Returns: JASS_ERROR_OK or JASS_ERROR_NO_INDEX

◆ set_postings_to_process_relative()

JASS_ERROR JASS_anytime_api::set_postings_to_process_relative ( double percent )

Set the number of postings to process as a proportion of the number of postings for this query.

An index does not need to be loaded first. This method takes precidence over set_postings_to_process() and set_postings_to_process_proportion(). By default all postings are processed.

Parameters

percent [in] The percent to use (for example, 10 is use 10% of the postings)

Returns: JASS_ERROR_OK or JASS_ERROR_NO_INDEX

◆ set_top_k()

JASS_ERROR JASS_anytime_api::set_top_k ( size_t k )

Set the maximum numbre of documents to return in a resuts list.

Parameters

k	[in] The k value for top-k search

Returns: JASS_ERROR_OK or JASS_ERROR_TOO_LARGE if k > get_max_top_k()

◆ threaded_search()

std::vector< JASS_anytime_thread_result > JASS_anytime_api::threaded_search	(	std::vector< std::string > &	query_list,
		size_t	thread_count
	)

Search using the current index and the current parameters.

Parameters

query_list	[in] A vector of queries to be spread over all the threads
thread_count	[in] The number of threads to use for searching

Returns: On success, a vector of results, one for each thread. On failure, an empty vvector or results

◆ use_ascii_parser()

JASS_ERROR JASS_anytime_api::use_ascii_parser ( void )

Use the query parser that seperates tokens on whitespace along (this method is not normally needed).

Returns: Always returns JASS_ERROR_OK

◆ use_query_parser()

JASS_ERROR JASS_anytime_api::use_query_parser ( void )

Use the default query parser that understands alphanumnerics, spaces, unicode, and so on (this method is not normally needed).

Returns: Always returns JASS_ERROR_OK

The documentation for this class was generated from the following files:

anytime/JASS_anytime_api.h
anytime/JASS_anytime_api.cpp

Classes

Public Member Functions

Private Member Functions

Static Private Member Functions

Private Attributes

Static Private Attributes

Detailed Description

Member Function Documentation

◆ anytime()

◆ anytime_bootstrap()

◆ get_document_count()

◆ get_encoding_scheme()

◆ get_encoding_scheme_d()

◆ get_encoding_scheme_name()

◆ get_max_top_k()

◆ get_postings_to_process()

◆ get_thread_local_data()

◆ get_top_k()

◆ load_index()

◆ search() [1/2]

◆ search() [2/2]

◆ set_accumulator_manager()

◆ set_accumulator_width()

◆ set_postings_to_process()

◆ set_postings_to_process_proportion()

◆ set_postings_to_process_relative()

◆ set_top_k()

◆ threaded_search()

◆ use_ascii_parser()

◆ use_query_parser()