#include <protomset.h>

Collaboration diagram for ProtoMSet:

Classes
class	MCmpAdaptor
	Adapt MSetCmp to be usable with min_heap. More...

Public Member Functions
	ProtoMSet (Xapian::doccount first_, Xapian::doccount max_items, Xapian::doccount check_at_least_, MSetCmp mcmp_, Xapian::Enquire::Internal::sort_setting sort_by_, Xapian::termcount total_subqs_, PostListTree &pltree_, Xapian::valueno collapse_key, Xapian::doccount collapse_max, int percent_threshold_, double percent_threshold_factor_, double max_possible_, bool stop_once_full_, double time_limit)

	ProtoMSet (const ProtoMSet &)=delete

ProtoMSet &	operator= (const ProtoMSet &)=delete

Collapser &	get_collapser ()

bool	full () const

double	get_min_weight () const

void	update_max_weight (double weight)

bool	checked_enough ()

bool	handle_min_weight_pending (bool finalising=false)
	Resolve a pending min_weight change. More...

bool	early_reject (Result &new_item, bool calculated_weight, SpyMaster &spymaster, const Xapian::Document &doc)

bool	process (Result &&new_item, ValueStreamDocument &vsdoc)
	Process new_item. More...

Xapian::doccount	add (Result &&item)

void	replace (Xapian::doccount old_item, Result &&b)

void	set_new_min_weight (double min_wt)

void	finalise_percentages ()

Xapian::MSet	finalise (const Xapian::MatchDecider *mdecider, const std::vector< std::unique_ptr< LocalSubMatch >> &locals, const Xapian::VecUniquePtr< EstimateOp > &estimates)

Private Member Functions
Xapian::doccount	size () const

Private Attributes
Xapian::doccount	max_size
	Maximum size the ProtoMSet needs to grow to. More...

Xapian::doccount	check_at_least

Xapian::Enquire::Internal::sort_setting	sort_by

MSetCmp	mcmp

double	min_weight = 0.0
	Minimum threshold on the weight. More...

double	max_weight = 0.0
	The highest document weight seen. More...

bool	min_weight_pending = false

Xapian::doccount	known_matching_docs = 0
	Count of how many known matching documents have been processed so far. More...

std::vector< Result >	results
	The items in the proto-MSet. More...

std::vector< Xapian::doccount >	min_heap
	A heap of offsets into results. More...

Xapian::doccount	first
	First entry wanted in MSet. More...

Xapian::termcount	total_subqs
	How many weighted leaf subqueries there are. More...

Xapian::termcount	max_weight_subqs_matched = 0
	The number of subqueries which matched to give max_weight. More...

int	percent_threshold

double	percent_threshold_factor

double	percent_scale = 0.0

PostListTree &	pltree

Collapser	collapser

double	max_possible

bool	stop_once_full

TimeOut	timeout

Friends
class	MCmpAdaptor

Detailed Description

Definition at line 38 of file protomset.h.

Constructor & Destructor Documentation

◆ ProtoMSet() [1/2]

ProtoMSet::ProtoMSet	(	Xapian::doccount	first_,
		Xapian::doccount	max_items,
		Xapian::doccount	check_at_least_,
		MSetCmp	mcmp_,
		Xapian::Enquire::Internal::sort_setting	sort_by_,
		Xapian::termcount	total_subqs_,
		PostListTree &	pltree_,
		Xapian::valueno	collapse_key,
		Xapian::doccount	collapse_max,
		int	percent_threshold_,
		double	percent_threshold_factor_,
		double	max_possible_,
		bool	stop_once_full_,
		double	time_limit
	)

inline

Definition at line 135 of file protomset.h.

References max_size, and results.

◆ ProtoMSet() [2/2]

ProtoMSet::ProtoMSet ( const ProtoMSet & )

delete

Member Function Documentation

◆ add()

Xapian::doccount ProtoMSet::add ( Result && item )

inline

Definition at line 346 of file protomset.h.

References Assert, checked_enough(), full(), handle_min_weight_pending(), known_matching_docs, Heap::make(), max_weight, mcmp, min_heap, min_weight, min_weight_pending, Xapian::Enquire::Internal::REL, Xapian::Enquire::Internal::REL_VAL, Heap::replace(), results, size(), sort_by, and update_max_weight().

Referenced by process().

◆ checked_enough()

bool ProtoMSet::checked_enough ( )

inline

Definition at line 187 of file protomset.h.

References check_at_least, known_matching_docs, max_size, TimeOut::timed_out(), and timeout.

Referenced by add(), early_reject(), handle_min_weight_pending(), and process().

◆ early_reject()

bool ProtoMSet::early_reject	(	Result &	new_item,
		bool	calculated_weight,
		SpyMaster &	spymaster,
		const Xapian::Document &	doc
	)

inline

Definition at line 250 of file protomset.h.

References checked_enough(), collapser, Result::get_weight(), PostListTree::get_weight(), known_matching_docs, mcmp, min_heap, pltree, results, and update_max_weight().

Referenced by Matcher::get_local_mset().

◆ finalise()

Xapian::MSet ProtoMSet::finalise	(	const Xapian::MatchDecider *	mdecider,
		const std::vector< std::unique_ptr< LocalSubMatch >> &	locals,
		const Xapian::VecUniquePtr< EstimateOp > &	estimates
	)

inline

Definition at line 491 of file protomset.h.

References Assert, AssertEq, AssertRel, check_at_least, collapser, Estimates::est, Collapser::finalise(), finalise_percentages(), first, full(), Collapser::get_docs_considered(), Collapser::get_dups_ignored(), Collapser::get_matches_lower_bound(), known_matching_docs, Estimates::max, max_possible, max_weight, mcmp, Estimates::min, min_weight, percent_scale, percent_threshold, percent_threshold_factor, results, Xapian::Vec< T, COW, UNIQUEPTR, typename >::size(), size(), and Heap::sort().

Referenced by Matcher::get_local_mset().

◆ finalise_percentages()

void ProtoMSet::finalise_percentages ( )

inline

Definition at line 472 of file protomset.h.

References Assert, handle_min_weight_pending(), max_weight, max_weight_subqs_matched, min_weight_pending, percent_scale, percent_threshold, percent_threshold_factor, results, set_new_min_weight(), and total_subqs.

Referenced by finalise().

◆ full()

bool ProtoMSet::full ( ) const

inline

Definition at line 172 of file protomset.h.

References max_size, and size().

Referenced by add(), finalise(), and process().

◆ get_collapser()

Collapser& ProtoMSet::get_collapser ( )

inline

Definition at line 170 of file protomset.h.

References collapser.

◆ get_min_weight()

double ProtoMSet::get_min_weight ( ) const

inline

Definition at line 174 of file protomset.h.

References min_weight.

Referenced by Matcher::get_local_mset().

◆ handle_min_weight_pending()

bool ProtoMSet::handle_min_weight_pending ( bool finalising = false )

inline

Resolve a pending min_weight change.

Only called when there's a percentage weight cut-off.

Definition at line 202 of file protomset.h.

References Assert, check_at_least, checked_enough(), collapser, Xapian::Enquire::Internal::DOCID, known_matching_docs, min_weight, min_weight_pending, Xapian::Enquire::Internal::REL, Xapian::Enquire::Internal::REL_VAL, Collapser::result_has_moved(), results, size(), and sort_by.

Referenced by add(), and finalise_percentages().

◆ operator=()

ProtoMSet& ProtoMSet::operator= ( const ProtoMSet & )

delete

◆ process()

bool ProtoMSet::process	(	Result &&	new_item,
		ValueStreamDocument &	vsdoc
	)

inline

Process new_item.

Conceptually this is "add new_item", but taking into account collapsing.

Definition at line 301 of file protomset.h.

References add(), Assert, Collapser::check(), checked_enough(), collapser, EMPTY, full(), Collapser::old_item, Collapser::process(), REJECT, REPLACE, replace(), results, stop_once_full, and update_max_weight().

Referenced by Matcher::get_local_mset().

◆ replace()

void ProtoMSet::replace	(	Xapian::doccount	old_item,
		Result &&	b
	)

inline

Definition at line 408 of file protomset.h.

References Assert, min_heap, rare, results, and Heap::siftdown().

Referenced by process().

◆ set_new_min_weight()

void ProtoMSet::set_new_min_weight ( double min_wt )

inline

Definition at line 432 of file protomset.h.

References max_weight, min_heap, min_weight, min_weight_pending, and results.

Referenced by finalise_percentages(), Matcher::get_local_mset(), and update_max_weight().

◆ size()

Xapian::doccount ProtoMSet::size ( ) const

inlineprivate

Definition at line 132 of file protomset.h.

References results.

Referenced by add(), finalise(), full(), and handle_min_weight_pending().

◆ update_max_weight()

void ProtoMSet::update_max_weight ( double weight )

inline

Definition at line 176 of file protomset.h.

References PostListTree::count_matching_subqs(), max_weight, max_weight_subqs_matched, percent_threshold, percent_threshold_factor, pltree, and set_new_min_weight().

Referenced by add(), early_reject(), and process().

Friends And Related Function Documentation

◆ MCmpAdaptor

friend class MCmpAdaptor

friend

Definition at line 52 of file protomset.h.

Member Data Documentation

◆ check_at_least

Xapian::doccount ProtoMSet::check_at_least

private

Definition at line 60 of file protomset.h.

Referenced by checked_enough(), finalise(), and handle_min_weight_pending().

◆ collapser

Collapser ProtoMSet::collapser

private

Definition at line 124 of file protomset.h.

Referenced by early_reject(), finalise(), get_collapser(), handle_min_weight_pending(), and process().

◆ first

Xapian::doccount ProtoMSet::first

private

First entry wanted in MSet.

Definition at line 104 of file protomset.h.

Referenced by finalise().

◆ known_matching_docs

Xapian::doccount ProtoMSet::known_matching_docs = 0

private

Count of how many known matching documents have been processed so far.

Used to implement "check_at_least".

Definition at line 92 of file protomset.h.

Referenced by add(), checked_enough(), early_reject(), finalise(), and handle_min_weight_pending().

◆ max_possible

double ProtoMSet::max_possible

private

Definition at line 126 of file protomset.h.

Referenced by finalise().

◆ max_size

Xapian::doccount ProtoMSet::max_size

private

Maximum size the ProtoMSet needs to grow to.

This is the maximum rank we care about.

Definition at line 58 of file protomset.h.

Referenced by checked_enough(), full(), and ProtoMSet().

◆ max_weight

double ProtoMSet::max_weight = 0.0

private

The highest document weight seen.

This weight may not actually be present in results if we're not sorting primarily by relevance, or if min_weight > max_weight.

Definition at line 84 of file protomset.h.

Referenced by add(), finalise(), finalise_percentages(), set_new_min_weight(), and update_max_weight().

◆ max_weight_subqs_matched

Xapian::termcount ProtoMSet::max_weight_subqs_matched = 0

private

The number of subqueries which matched to give max_weight.

Definition at line 114 of file protomset.h.

Referenced by finalise_percentages(), and update_max_weight().

◆ mcmp

MSetCmp ProtoMSet::mcmp

private

Definition at line 64 of file protomset.h.

Referenced by add(), early_reject(), finalise(), and ProtoMSet::MCmpAdaptor::operator()().

◆ min_heap

std::vector<Xapian::doccount> ProtoMSet::min_heap

private

A heap of offsets into results.

Created lazily once we actually need it.

Definition at line 101 of file protomset.h.

Referenced by add(), early_reject(), replace(), and set_new_min_weight().

◆ min_weight

double ProtoMSet::min_weight = 0.0

private

Minimum threshold on the weight.

If the primary result ordering is by decreasing relevance (i.e. sort_by is REL or REL_VAL) then once the min_heap kicks in this threshold is raised to the lowest weight in the proto-mset.

Enquire::set_cutoff() can also affect min_weight - an absolute threshold determines the initial value; a percentage threshold raises the threshold each time max_weight increases (unless it's already higher than the value the percentage threshold results in).

Definition at line 77 of file protomset.h.

Referenced by add(), finalise(), get_min_weight(), handle_min_weight_pending(), and set_new_min_weight().

◆ min_weight_pending

bool ProtoMSet::min_weight_pending = false

private

Definition at line 86 of file protomset.h.

Referenced by add(), finalise_percentages(), handle_min_weight_pending(), and set_new_min_weight().

◆ percent_scale

double ProtoMSet::percent_scale = 0.0

private

Definition at line 120 of file protomset.h.

Referenced by finalise(), and finalise_percentages().

◆ percent_threshold

int ProtoMSet::percent_threshold

private

Definition at line 116 of file protomset.h.

Referenced by finalise(), finalise_percentages(), and update_max_weight().

◆ percent_threshold_factor

double ProtoMSet::percent_threshold_factor

private

Definition at line 118 of file protomset.h.

Referenced by finalise(), finalise_percentages(), and update_max_weight().

◆ pltree

PostListTree& ProtoMSet::pltree

private

Definition at line 122 of file protomset.h.

Referenced by early_reject(), and update_max_weight().

◆ results

std::vector<Result> ProtoMSet::results

private

The items in the proto-MSet.

Definition at line 95 of file protomset.h.

Referenced by add(), early_reject(), finalise(), finalise_percentages(), handle_min_weight_pending(), ProtoMSet::MCmpAdaptor::operator()(), process(), ProtoMSet(), replace(), set_new_min_weight(), and size().

◆ sort_by

Xapian::Enquire::Internal::sort_setting ProtoMSet::sort_by

private

Definition at line 62 of file protomset.h.

Referenced by add(), and handle_min_weight_pending().

◆ stop_once_full

bool ProtoMSet::stop_once_full

private

Definition at line 128 of file protomset.h.

Referenced by process().

◆ timeout

TimeOut ProtoMSet::timeout

private

Definition at line 130 of file protomset.h.

Referenced by checked_enough().

◆ total_subqs

Xapian::termcount ProtoMSet::total_subqs

private

How many weighted leaf subqueries there are.

Used for scaling percentages when the highest weighted document doesn't "match all terms".

Definition at line 111 of file protomset.h.

Referenced by finalise_percentages().

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

matcher/protomset.h

Classes

Public Member Functions

Private Member Functions

Private Attributes

Friends

Detailed Description

Constructor & Destructor Documentation

◆ ProtoMSet() [1/2]

◆ ProtoMSet() [2/2]

Member Function Documentation

◆ add()

◆ checked_enough()

◆ early_reject()

◆ finalise()

◆ finalise_percentages()

◆ full()

◆ get_collapser()

◆ get_min_weight()

◆ handle_min_weight_pending()

◆ operator=()

◆ process()

◆ replace()

◆ set_new_min_weight()

◆ size()

◆ update_max_weight()

Friends And Related Function Documentation

◆ MCmpAdaptor

Member Data Documentation

◆ check_at_least

◆ collapser

◆ first

◆ known_matching_docs

◆ max_possible

◆ max_size

◆ max_weight

◆ max_weight_subqs_matched

◆ mcmp

◆ min_heap

◆ min_weight

◆ min_weight_pending

◆ percent_scale

◆ percent_threshold

◆ percent_threshold_factor

◆ pltree

◆ results

◆ sort_by

◆ stop_once_full

◆ timeout

◆ total_subqs