A postlist in a glass database. More...

#include <glass_postlist.h>

Inheritance diagram for GlassPostList:

Collaboration diagram for GlassPostList:

Public Member Functions
	GlassPostList (Xapian::Internal::intrusive_ptr< const GlassDatabase > this_db_, std::string_view term, bool keep_reference)
	Default constructor. More...

	~GlassPostList ()
	Destructor. More...

bool	open_nearby_postlist (std::string_view term_, bool need_read_pos, LeafPostList *&pl) const
	Open another postlist from the same database. More...

bool	jump_to (Xapian::docid desired_did)
	Used for looking up doclens. More...

Xapian::docid	get_docid () const
	Returns the current docid. More...

Xapian::termcount	get_wdf () const
	Returns the Within Document Frequency of the term in the current document. More...

PositionList *	read_position_list ()
	Get the list of positions of the term in the current document. More...

PositionList *	open_position_list () const
	Get the list of positions of the term in the current document. More...

PostList *	next (double w_min)
	Move to the next document. More...

PostList *	skip_to (Xapian::docid desired_did, double w_min)
	Skip to next document with docid >= docid. More...

bool	at_end () const
	Return true if and only if we're off the end of the list. More...

Xapian::termcount	get_wdf_upper_bound () const

void	get_docid_range (Xapian::docid &first, Xapian::docid &last) const
	Get the bounds on the range of docids this PostList can return. More...

std::string	get_description () const
	Get a description of the document. More...

Public Member Functions inherited from LeafPostList
	~LeafPostList ()

Xapian::termcount	get_collfreq () const
	Get the collection frequency of the term. More...

void	set_termweight (const Xapian::Weight *weight_)
	Set the weighting scheme to use during matching. More...

double	resolve_lazy_termweight (Xapian::Weight weight_, Xapian::Weight::Internal stats, Xapian::termcount qlen, Xapian::termcount wqf, double factor, const Xapian::Database::Internal *shard)

double	get_weight (Xapian::termcount doclen, Xapian::termcount unique_terms, Xapian::termcount wdfdocmax) const
	Return the weight contribution for the current position. More...

double	recalc_maxweight ()
	Recalculate the upper bound on what get_weight() can return. More...

Xapian::termcount	count_matching_subqs () const
	Count the number of leaf subqueries which match at the current position. More...

void	gather_position_lists (OrPositionList *orposlist)
	Gather PositionList* objects for a subtree. More...

const std::string &	get_term () const
	Get the term name. More...

void	set_term (std::string_view term_)
	Set the term name. More...

Public Member Functions inherited from Xapian::Internal::PostList
virtual	~PostList ()
	We have virtual methods and want to be able to delete derived classes using a pointer to the base class, so we need a virtual destructor. More...

Xapian::doccount	get_termfreq () const
	Get an estimate of the number of documents this PostList will return. More...

virtual PostList *	check (Xapian::docid did, double w_min, bool &valid)
	Check if the specified docid occurs in this postlist. More...

PostList *	next ()
	Advance the current position to the next document in the postlist. More...

PostList *	skip_to (Xapian::docid did)
	Skip forward to the specified docid. More...

Static Public Member Functions
static void	read_freqs (const char *posptr, const char end, Xapian::doccount number_of_entries_ptr, Xapian::termcount collection_freq_ptr)
	Read the term frequency and collection frequency. More...

Private Member Functions
	GlassPostList (const GlassPostList &)
	Copying is not allowed. More...

void	operator= (const GlassPostList &)
	Assignment is not allowed. More...

bool	next_in_chunk ()
	Move to the next item in the chunk, if possible. More...

void	next_chunk ()
	Move to the next chunk. More...

bool	current_chunk_contains (Xapian::docid desired_did)
	Return true if the given document ID lies in the range covered by the current chunk. More...

void	move_to_chunk_containing (Xapian::docid desired_did)
	Move to chunk containing the specified document ID. More...

bool	move_forward_in_chunk_to_at_least (Xapian::docid desired_did)
	Scan forward in the current chunk for the specified document ID. More...

	GlassPostList (Xapian::Internal::intrusive_ptr< const GlassDatabase > this_db_, std::string_view term, GlassCursor *cursor_)

void	init ()

Private Attributes
Xapian::Internal::intrusive_ptr< const GlassDatabase >	this_db
	The database we are searching. More...

GlassRePositionList *	positionlist = NULL
	The position list object for this posting list. More...

bool	have_started
	Whether we've started reading the list yet. More...

bool	is_last_chunk
	True if this is the last chunk. More...

bool	is_at_end
	Whether we've run off the end of the list yet. More...

std::unique_ptr< GlassCursor >	cursor
	Cursor pointing to current chunk of postlist. More...

Xapian::docid	first_did_in_chunk
	The first document id in this chunk. More...

Xapian::docid	last_did_in_chunk
	The last document id in this chunk. More...

const char *	pos
	Position of iteration through current chunk. More...

const char *	end
	Pointer to byte after end of current chunk. More...

Xapian::docid	did
	Document id we're currently at. More...

Xapian::termcount	wdf
	The wdf of the current document. More...

Xapian::termcount	wdf_upper_bound
	Upper bound on wdf for this postlist. More...

Additional Inherited Members
Protected Member Functions inherited from LeafPostList
	LeafPostList (std::string_view term_)
	Only constructable as a base class for derived classes. More...

Protected Member Functions inherited from Xapian::Internal::PostList
	PostList ()
	Only constructable as a base class for derived classes. More...

Protected Attributes inherited from LeafPostList
const Xapian::Weight *	weight = nullptr

std::string	term
	The term name for this postlist (empty for an alldocs postlist). More...

Xapian::termcount	collfreq
	The collection frequency of the term. More...

Protected Attributes inherited from Xapian::Internal::PostList
Xapian::doccount	termfreq
	Estimate of the number of documents this PostList will return. More...

Detailed Description

A postlist in a glass database.

Definition at line 53 of file glass_postlist.h.

Constructor & Destructor Documentation

◆ GlassPostList() [1/3]

GlassPostList::GlassPostList ( const GlassPostList & )

private

Copying is not allowed.

Referenced by open_nearby_postlist().

◆ GlassPostList() [2/3]

GlassPostList::GlassPostList	(	Xapian::Internal::intrusive_ptr< const GlassDatabase >	this_db_,
		std::string_view	term,
		GlassCursor *	cursor_
	)

private

Definition at line 709 of file glass_postlist.cc.

References Xapian::Internal::intrusive_ptr< T >::get(), init(), and LOGCALL_CTOR.

◆ GlassPostList() [3/3]

GlassPostList::GlassPostList	(	Xapian::Internal::intrusive_ptr< const GlassDatabase >	this_db_,
		std::string_view	term,
		bool	keep_reference
	)

Default constructor.

The format of a postlist is:

Split into chunks. Key for first chunk is the termname (encoded as length : name). Key for subsequent chunks is the same, followed by the document ID of the first document in the chunk (encoded as length of representation in first byte, and then docid).

A chunk (except for the first chunk) contains:

1) bool - true if this is the last chunk. 2) difference between final docid in chunk and first docid. 3) wdf for the first item. 4) increment in docid to next item, followed by wdf for the item. 5) (4) repeatedly.

The first chunk begins with the number of entries, the collection frequency, then the docid of the first document, then has the header of a standard chunk.

Definition at line 696 of file glass_postlist.cc.

References Xapian::Internal::intrusive_ptr< T >::get(), init(), and LOGCALL_CTOR.

◆ ~GlassPostList()

GlassPostList::~GlassPostList ( )

Destructor.

Definition at line 757 of file glass_postlist.cc.

References LOGCALL_DTOR, and positionlist.

Member Function Documentation

◆ at_end()

bool GlassPostList::at_end ( ) const

inlinevirtual

Return true if and only if we're off the end of the list.

Implements Xapian::Internal::PostList.

Definition at line 195 of file glass_postlist.h.

References is_at_end.

Referenced by GlassAllDocsPostList::get_wdf().

◆ current_chunk_contains()

bool GlassPostList::current_chunk_contains ( Xapian::docid desired_did )

private

Return true if the given document ID lies in the range covered by the current chunk.

This does not say whether the document ID is actually present. It will return false if the document ID is greater than the last document ID in the chunk, even if it is less than the first document ID in the next chunk: it is possible for no chunk to contain a particular document ID.

Definition at line 888 of file glass_postlist.cc.

References first_did_in_chunk, last_did_in_chunk, LOGCALL, and RETURN.

Referenced by jump_to(), and skip_to().

◆ get_description()

string GlassPostList::get_description ( ) const

virtual

Get a description of the document.

Implements Xapian::Internal::PostList.

Definition at line 1050 of file glass_postlist.cc.

References description_append(), Xapian::Internal::str(), LeafPostList::term, and Xapian::Internal::PostList::termfreq.

◆ get_docid()

Xapian::docid GlassPostList::get_docid ( ) const

inlinevirtual

Returns the current docid.

Implements Xapian::Internal::PostList.

Definition at line 173 of file glass_postlist.h.

References Assert, did, and have_started.

◆ get_docid_range()

void GlassPostList::get_docid_range	(	Xapian::docid &	first,
		Xapian::docid &	last
	)		const

virtual

Get the bounds on the range of docids this PostList can return.

Parameters

[out]	first	Set to a lower bound on the docids that can be returned, or not changed if there's no known lower bound (other than 1).
[out]	last	Set to an upper bound on the docids that can be returned, or not changed if there's no known upper bound (other than the highest used docid).

The default implementation (PostList::get_docid_range()) does nothing, which is suitable when there's no known lower or upper bound.

Reimplemented from Xapian::Internal::PostList.

Definition at line 1037 of file glass_postlist.cc.

References first_did_in_chunk, is_last_chunk, last_did_in_chunk, and pos.

◆ get_wdf()

Xapian::termcount GlassPostList::get_wdf ( ) const

inlinevirtual

Returns the Within Document Frequency of the term in the current document.

Reimplemented from Xapian::Internal::PostList.

Definition at line 178 of file glass_postlist.h.

References Assert, have_started, and wdf.

◆ get_wdf_upper_bound()

Xapian::termcount GlassPostList::get_wdf_upper_bound ( ) const

virtual

Implements LeafPostList.

Definition at line 1379 of file glass_postlist.cc.

References Assert, LeafPostList::term, and wdf_upper_bound.

◆ init()

void GlassPostList::init ( )

private

Definition at line 723 of file glass_postlist.cc.

References LeafPostList::collfreq, cursor, did, end, first_did_in_chunk, is_at_end, is_last_chunk, last_did_in_chunk, LOGLINE, GlassPostListTable::make_key(), pos, read_start_of_chunk(), read_start_of_first_chunk(), read_wdf(), LeafPostList::term, Xapian::Internal::PostList::termfreq, wdf, and wdf_upper_bound.

Referenced by GlassPostList().

◆ jump_to()

bool GlassPostList::jump_to ( Xapian::docid desired_did )

Used for looking up doclens.

Returns: true if docid desired_did has a document length.

Definition at line 1008 of file glass_postlist.cc.

References current_chunk_contains(), did, have_started, is_at_end, LOGCALL, move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), pos, and RETURN.

◆ move_forward_in_chunk_to_at_least()

bool GlassPostList::move_forward_in_chunk_to_at_least ( Xapian::docid desired_did )

private

Scan forward in the current chunk for the specified document ID.

This is particularly efficient if the desired document ID is greater than the last in the chunk - it then skips straight to the end.

Returns: true if we moved to a valid document, false if we reached the end of the chunk.

Definition at line 947 of file glass_postlist.cc.

References Assert, did, end, last_did_in_chunk, LOGCALL, pos, read_did_increase(), read_wdf(), RETURN, and wdf.

Referenced by jump_to(), and skip_to().

◆ move_to_chunk_containing()

void GlassPostList::move_to_chunk_containing ( Xapian::docid desired_did )

private

Move to chunk containing the specified document ID.

This moves to the chunk whose starting document ID is <= desired_did, but such that the next chunk's starting document ID is > desired_did.

It is thus possible that current_chunk_contains(desired_did) will return false after this call, since the document ID might lie after the end of this chunk, but before the start of the next chunk.

Definition at line 899 of file glass_postlist.cc.

References Assert, check_tname_in_key_lite(), cursor, did, end, first_did_in_chunk, is_at_end, is_last_chunk, last_did_in_chunk, LOGCALL_VOID, GlassPostListTable::make_key(), next_chunk(), pos, read_start_of_chunk(), read_start_of_first_chunk(), read_wdf(), report_read_error(), LeafPostList::term, Xapian::Internal::PostList::termfreq, unpack_uint_preserving_sort(), and wdf.

Referenced by jump_to(), and skip_to().

◆ next()

PostList * GlassPostList::next ( double w_min )

virtual

Move to the next document.

Implements Xapian::Internal::PostList.

Definition at line 867 of file glass_postlist.cc.

References did, have_started, is_at_end, LOGCALL, LOGLINE, next_chunk(), next_in_chunk(), RETURN, and wdf.

◆ next_chunk()

void GlassPostList::next_chunk ( )

private

Move to the next chunk.

If there are no more chunks in this postlist, this will set is_at_end to true.

Definition at line 800 of file glass_postlist.cc.

References check_tname_in_key_lite(), cursor, did, end, first_did_in_chunk, is_at_end, is_last_chunk, last_did_in_chunk, LOGCALL_VOID, pos, read_start_of_chunk(), read_wdf(), report_read_error(), Xapian::Internal::str(), LeafPostList::term, unpack_uint_preserving_sort(), and wdf.

Referenced by move_to_chunk_containing(), and next().

◆ next_in_chunk()

bool GlassPostList::next_in_chunk ( )

private

Move to the next item in the chunk, if possible.

If already at the end of the chunk, returns false.

Definition at line 783 of file glass_postlist.cc.

References Assert, did, end, last_did_in_chunk, LOGCALL, pos, read_did_increase(), read_wdf(), RETURN, and wdf.

Referenced by next().

◆ open_nearby_postlist()

bool GlassPostList::open_nearby_postlist	(	std::string_view	term_,
		bool	need_read_pos,
		LeafPostList *&	pl
	)		const

virtual

Open another postlist from the same database.

Parameters

	term_	The term to open a postlist for (must not be an empty string). If term_ is near to this postlist's term, then this can be a lot more efficient (and if it isn't very near, there's not much of a penalty). Using this method can make a wildcard expansion much more memory efficient.
	need_read_pos	Does the postlist need to support read_position_list()? Note that open_position_list() may still be called even if need_read_pos is false.
[out]	pl	If true is returned, set to a new LeafPostList object (or may be set to NULL if the term doesn't index any documents). The caller takes ownership of the returned object.

Returns: true if successful (and pl has been set); false if not (in which case the caller should probably open the postlist via the database instead).

Reimplemented from LeafPostList.

Definition at line 764 of file glass_postlist.cc.

References cursor, Xapian::Internal::PostList::get_termfreq(), GlassPostList(), GlassTable::is_writable(), LOGCALL, GlassDatabase::postlist_table, RETURN, and this_db.

◆ open_position_list()

PositionList * GlassPostList::open_position_list ( ) const

virtual

Get the list of positions of the term in the current document.

Reimplemented from Xapian::Internal::PostList.

Definition at line 859 of file glass_postlist.cc.

References Assert, did, LOGCALL, GlassDatabase::open_position_list(), RETURN, LeafPostList::term, and this_db.

◆ operator=()

void GlassPostList::operator= ( const GlassPostList & )

private

Assignment is not allowed.

◆ read_freqs()

void GlassPostList::read_freqs	(	const char **	posptr,
		const char *	end,
		Xapian::doccount *	termfreq_ptr,
		Xapian::termcount *	collection_freq_ptr
	)

static

Read the term frequency and collection frequency.

Read the number of entries in the posting list.

This must only be called when *posptr is pointing to the start of the first chunk of the posting list.

Definition at line 666 of file glass_postlist.cc.

References report_read_error(), and unpack_uint().

Referenced by GlassPostListTable::get_freqs(), read_start_of_first_chunk(), and GlassAllTermsList::read_termfreq().

◆ read_position_list()

PositionList * GlassPostList::read_position_list ( )

virtual

Get the list of positions of the term in the current document.

Reimplemented from Xapian::Internal::PostList.

Definition at line 845 of file glass_postlist.cc.

References Assert, did, LOGCALL, GlassDatabase::position_table, positionlist, rare, GlassDatabase::read_position_list(), RETURN, LeafPostList::term, and this_db.

◆ skip_to()

PostList * GlassPostList::skip_to	(	Xapian::docid	desired_did,
		double	w_min
	)

virtual

Skip to next document with docid >= docid.

Implements Xapian::Internal::PostList.

Definition at line 973 of file glass_postlist.cc.

References Assert, current_chunk_contains(), did, have_started, is_at_end, LOGCALL, LOGLINE, move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), RETURN, and wdf.

Member Data Documentation

◆ cursor

std::unique_ptr<GlassCursor> GlassPostList::cursor

private

Cursor pointing to current chunk of postlist.

Definition at line 73 of file glass_postlist.h.

Referenced by init(), move_to_chunk_containing(), next_chunk(), and open_nearby_postlist().

◆ did

Xapian::docid GlassPostList::did

private

Document id we're currently at.

Definition at line 88 of file glass_postlist.h.

Referenced by get_docid(), init(), jump_to(), move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), next(), next_chunk(), next_in_chunk(), open_position_list(), read_position_list(), and skip_to().

◆ end

const char* GlassPostList::end

private

Pointer to byte after end of current chunk.

Definition at line 85 of file glass_postlist.h.

Referenced by init(), move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), next_chunk(), and next_in_chunk().

◆ first_did_in_chunk

Xapian::docid GlassPostList::first_did_in_chunk

private

The first document id in this chunk.

Definition at line 76 of file glass_postlist.h.

Referenced by current_chunk_contains(), get_docid_range(), init(), move_to_chunk_containing(), and next_chunk().

◆ have_started

bool GlassPostList::have_started

private

Whether we've started reading the list yet.

Definition at line 64 of file glass_postlist.h.

Referenced by get_docid(), get_wdf(), jump_to(), next(), and skip_to().

◆ is_at_end

bool GlassPostList::is_at_end

private

Whether we've run off the end of the list yet.

Definition at line 70 of file glass_postlist.h.

Referenced by at_end(), init(), jump_to(), move_to_chunk_containing(), next(), next_chunk(), and skip_to().

◆ is_last_chunk

bool GlassPostList::is_last_chunk

private

True if this is the last chunk.

Definition at line 67 of file glass_postlist.h.

Referenced by get_docid_range(), init(), move_to_chunk_containing(), and next_chunk().

◆ last_did_in_chunk

Xapian::docid GlassPostList::last_did_in_chunk

private

The last document id in this chunk.

Definition at line 79 of file glass_postlist.h.

Referenced by current_chunk_contains(), get_docid_range(), init(), move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), next_chunk(), and next_in_chunk().

◆ pos

const char* GlassPostList::pos

private

Position of iteration through current chunk.

Definition at line 82 of file glass_postlist.h.

Referenced by get_docid_range(), init(), jump_to(), move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), next_chunk(), and next_in_chunk().

◆ positionlist

GlassRePositionList* GlassPostList::positionlist = NULL

private

The position list object for this posting list.

Definition at line 61 of file glass_postlist.h.

Referenced by read_position_list(), and ~GlassPostList().

◆ this_db

Xapian::Internal::intrusive_ptr<const GlassDatabase> GlassPostList::this_db

private

The database we are searching.

This pointer is held so that the database doesn't get deleted before us, and also to give us access to the position_table.

Definition at line 58 of file glass_postlist.h.

Referenced by open_nearby_postlist(), open_position_list(), and read_position_list().

◆ wdf

Xapian::termcount GlassPostList::wdf

private

The wdf of the current document.

Definition at line 91 of file glass_postlist.h.

Referenced by get_wdf(), init(), move_forward_in_chunk_to_at_least(), move_to_chunk_containing(), next(), next_chunk(), next_in_chunk(), and skip_to().

◆ wdf_upper_bound

Xapian::termcount GlassPostList::wdf_upper_bound

private

Upper bound on wdf for this postlist.

Definition at line 94 of file glass_postlist.h.

Referenced by get_wdf_upper_bound(), and init().

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

backends/glass/glass_postlist.h
backends/glass/glass_postlist.cc

Public Member Functions

Static Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ GlassPostList() [1/3]

◆ GlassPostList() [2/3]

◆ GlassPostList() [3/3]

◆ ~GlassPostList()

Member Function Documentation

◆ at_end()

◆ current_chunk_contains()

◆ get_description()

◆ get_docid()

◆ get_docid_range()

◆ get_wdf()

◆ get_wdf_upper_bound()

◆ init()

◆ jump_to()

◆ move_forward_in_chunk_to_at_least()

◆ move_to_chunk_containing()

◆ next()

◆ next_chunk()

◆ next_in_chunk()

◆ open_nearby_postlist()

◆ open_position_list()

◆ operator=()

◆ read_freqs()

◆ read_position_list()

◆ skip_to()

Member Data Documentation

◆ cursor

◆ did

◆ end

◆ first_did_in_chunk

◆ have_started

◆ is_at_end

◆ is_last_chunk

◆ last_did_in_chunk

◆ pos

◆ positionlist

◆ this_db

◆ wdf

◆ wdf_upper_bound