#include <dawg.h>

Inheritance diagram for tesseract::SquishedDawg:

Collaboration diagram for tesseract::SquishedDawg:

Public Member Functions
	SquishedDawg (DawgType type, const STRING &lang, PermuterType perm, int debug_level)

	SquishedDawg (const char *filename, DawgType type, const STRING &lang, PermuterType perm, int debug_level)

	SquishedDawg (EDGE_ARRAY edges, int num_edges, DawgType type, const STRING &lang, PermuterType perm, int unicharset_size, int debug_level)

virtual	~SquishedDawg ()

bool	Load (TFile *fp)

int	NumEdges ()

EDGE_REF	edge_char_of (NODE_REF node, UNICHAR_ID unichar_id, bool word_end) const
	Returns the edge that corresponds to the letter out of this node. More...

void	unichar_ids_of (NODE_REF node, NodeChildVector *vec, bool word_end) const

NODE_REF	next_node (EDGE_REF edge) const

bool	end_of_word (EDGE_REF edge_ref) const

UNICHAR_ID	edge_letter (EDGE_REF edge_ref) const
	Returns UNICHAR_ID stored in the edge indicated by the given EDGE_REF. More...

void	print_node (NODE_REF node, int max_num_edges) const

bool	write_squished_dawg (TFile *file)
	Writes the squished/reduced Dawg to a file. More...

bool	write_squished_dawg (const char *filename)

Public Member Functions inherited from tesseract::Dawg
DawgType	type () const

const STRING &	lang () const

PermuterType	permuter () const

virtual	~Dawg ()

bool	word_in_dawg (const WERD_CHOICE &word) const
	Returns true if the given word is in the Dawg. More...

bool	prefix_in_dawg (const WERD_CHOICE &prefix, bool requires_complete) const

int	check_for_words (const char *filename, const UNICHARSET &unicharset, bool enable_wildcard) const

void	iterate_words (const UNICHARSET &unicharset, TessCallback1< const WERD_CHOICE > cb) const

void	iterate_words (const UNICHARSET &unicharset, TessCallback1< const char > cb) const

virtual void	unichar_id_to_patterns (UNICHAR_ID unichar_id, const UNICHARSET &unicharset, GenericVector< UNICHAR_ID > *vec) const

virtual EDGE_REF	pattern_loop_edge (EDGE_REF edge_ref, UNICHAR_ID unichar_id, bool word_end) const

Private Member Functions
void	set_next_node (EDGE_REF edge_ref, EDGE_REF value)
	Sets the next node link for this edge. More...

void	set_empty_edge (EDGE_REF edge_ref)
	Sets the edge to be empty. More...

void	clear_all_edges ()
	Goes through all the edges and clears each one out. More...

void	clear_marker_flag (EDGE_REF edge_ref)
	Clears the last flag of this edge. More...

bool	forward_edge (EDGE_REF edge_ref) const
	Returns true if this edge is in the forward direction. More...

bool	backward_edge (EDGE_REF edge_ref) const
	Returns true if this edge is in the backward direction. More...

bool	edge_occupied (EDGE_REF edge_ref) const
	Returns true if the edge spot in this location is occupied. More...

bool	last_edge (EDGE_REF edge_ref) const
	Returns true if this edge is the last edge in a sequence. More...

int32_t	num_forward_edges (NODE_REF node) const
	Counts and returns the number of forward edges in this node. More...

bool	read_squished_dawg (TFile *file)
	Reads SquishedDawg from a file. More...

void	print_edge (EDGE_REF edge) const
	Prints the contents of an edge indicated by the given EDGE_REF. More...

void	print_all (const char *msg)
	Prints the contents of the SquishedDawg. More...

std::unique_ptr< EDGE_REF[]>	build_node_map (int32_t *num_nodes) const
	Constructs a mapping from the memory node indices to disk node indices. More...

Private Attributes
EDGE_ARRAY	edges_

int32_t	num_edges_

int	num_forward_edges_in_node0

Additional Inherited Members
Static Public Attributes inherited from tesseract::Dawg
static const int16_t	kDawgMagicNumber = 42
	Magic number to determine endianness when reading the Dawg from file. More...

static const UNICHAR_ID	kPatternUnicharID = 0

Protected Member Functions inherited from tesseract::Dawg
	Dawg (DawgType type, const STRING &lang, PermuterType perm, int debug_level)

NODE_REF	next_node_from_edge_rec (const EDGE_RECORD &edge_rec) const
	Returns the next node visited by following this edge. More...

bool	marker_flag_from_edge_rec (const EDGE_RECORD &edge_rec) const
	Returns the marker flag of this edge. More...

int	direction_from_edge_rec (const EDGE_RECORD &edge_rec) const
	Returns the direction flag of this edge. More...

bool	end_of_word_from_edge_rec (const EDGE_RECORD &edge_rec) const
	Returns true if this edge marks the end of a word. More...

UNICHAR_ID	unichar_id_from_edge_rec (const EDGE_RECORD &edge_rec) const
	Returns UNICHAR_ID recorded in this edge. More...

void	set_next_node_in_edge_rec (EDGE_RECORD *edge_rec, EDGE_REF value)
	Sets the next node link for this edge in the Dawg. More...

void	set_marker_flag_in_edge_rec (EDGE_RECORD *edge_rec)
	Sets this edge record to be the last one in a sequence of edges. More...

int	given_greater_than_edge_rec (NODE_REF next_node, bool word_end, UNICHAR_ID unichar_id, const EDGE_RECORD &edge_rec) const

bool	edge_rec_match (NODE_REF next_node, bool word_end, UNICHAR_ID unichar_id, NODE_REF other_next_node, bool other_word_end, UNICHAR_ID other_unichar_id) const

void	init (int unicharset_size)

bool	match_words (WERD_CHOICE *word, int32_t index, NODE_REF node, UNICHAR_ID wildcard) const

void	iterate_words_rec (const WERD_CHOICE &word_so_far, NODE_REF to_explore, TessCallback1< const WERD_CHOICE > cb) const

Protected Attributes inherited from tesseract::Dawg
DawgType	type_

STRING	lang_

PermuterType	perm_
	Permuter code that should be used if the word is found in this Dawg. More...

int	unicharset_size_

int	flag_start_bit_

int	next_node_start_bit_

uint64_t	next_node_mask_

uint64_t	flags_mask_

uint64_t	letter_mask_

int	debug_level_

Detailed Description

Concrete class that can operate on a compacted (squished) Dawg (read, search and write to file). This class is read-only in the sense that new words can not be added to an instance of SquishedDawg. The underlying representation of the nodes and edges in SquishedDawg is stored as a contiguous EDGE_ARRAY (read from file or given as an argument to the constructor).

Constructor & Destructor Documentation

◆ SquishedDawg() [1/3]

tesseract::SquishedDawg::SquishedDawg	(	DawgType	type,
		const STRING &	lang,
		PermuterType	perm,
		int	debug_level
	)

inline

◆ SquishedDawg() [2/3]

tesseract::SquishedDawg::SquishedDawg	(	const char *	filename,
		DawgType	type,
		const STRING &	lang,
		PermuterType	perm,
		int	debug_level
	)

inline

◆ SquishedDawg() [3/3]

tesseract::SquishedDawg::SquishedDawg	(	EDGE_ARRAY	edges,
		int	num_edges,
		DawgType	type,
		const STRING &	lang,
		PermuterType	perm,
		int	unicharset_size,
		int	debug_level
	)

inline

◆ ~SquishedDawg()

tesseract::SquishedDawg::~SquishedDawg ( )

virtual

Member Function Documentation

◆ backward_edge()

bool tesseract::SquishedDawg::backward_edge ( EDGE_REF edge_ref ) const

inlineprivate

Returns true if this edge is in the backward direction.

◆ build_node_map()

std::unique_ptr< EDGE_REF[]> tesseract::SquishedDawg::build_node_map ( int32_t * num_nodes ) const

private

Constructs a mapping from the memory node indices to disk node indices.

◆ clear_all_edges()

void tesseract::SquishedDawg::clear_all_edges ( )

inlineprivate

Goes through all the edges and clears each one out.

◆ clear_marker_flag()

void tesseract::SquishedDawg::clear_marker_flag ( EDGE_REF edge_ref )

inlineprivate

Clears the last flag of this edge.

◆ edge_char_of()

EDGE_REF tesseract::SquishedDawg::edge_char_of	(	NODE_REF	node,
		UNICHAR_ID	unichar_id,
		bool	word_end
	)		const

virtual

Returns the edge that corresponds to the letter out of this node.

Implements tesseract::Dawg.

◆ edge_letter()

UNICHAR_ID tesseract::SquishedDawg::edge_letter ( EDGE_REF edge_ref ) const

inlinevirtual

Returns UNICHAR_ID stored in the edge indicated by the given EDGE_REF.

Implements tesseract::Dawg.

◆ edge_occupied()

bool tesseract::SquishedDawg::edge_occupied ( EDGE_REF edge_ref ) const

inlineprivate

Returns true if the edge spot in this location is occupied.

◆ end_of_word()

bool tesseract::SquishedDawg::end_of_word ( EDGE_REF edge_ref ) const

inlinevirtual

Returns true if the edge indicated by the given EDGE_REF marks the end of a word.

Implements tesseract::Dawg.

◆ forward_edge()

bool tesseract::SquishedDawg::forward_edge ( EDGE_REF edge_ref ) const

inlineprivate

Returns true if this edge is in the forward direction.

◆ last_edge()

bool tesseract::SquishedDawg::last_edge ( EDGE_REF edge_ref ) const

inlineprivate

Returns true if this edge is the last edge in a sequence.

◆ Load()

bool tesseract::SquishedDawg::Load ( TFile * fp )

inline

◆ next_node()

NODE_REF tesseract::SquishedDawg::next_node ( EDGE_REF edge ) const

inlinevirtual

Returns the next node visited by following the edge indicated by the given EDGE_REF.

Implements tesseract::Dawg.

◆ num_forward_edges()

int32_t tesseract::SquishedDawg::num_forward_edges ( NODE_REF node ) const

private

Counts and returns the number of forward edges in this node.

◆ NumEdges()

int tesseract::SquishedDawg::NumEdges ( )

inline

◆ print_all()

void tesseract::SquishedDawg::print_all ( const char * msg )

inlineprivate

Prints the contents of the SquishedDawg.

◆ print_edge()

void tesseract::SquishedDawg::print_edge ( EDGE_REF edge ) const

private

Prints the contents of an edge indicated by the given EDGE_REF.

◆ print_node()

void tesseract::SquishedDawg::print_node	(	NODE_REF	node,
		int	max_num_edges
	)		const

virtual

Prints the contents of the node indicated by the given NODE_REF. At most max_num_edges will be printed.

Implements tesseract::Dawg.

◆ read_squished_dawg()

bool tesseract::SquishedDawg::read_squished_dawg ( TFile * file )

private

Reads SquishedDawg from a file.

◆ set_empty_edge()

void tesseract::SquishedDawg::set_empty_edge ( EDGE_REF edge_ref )

inlineprivate

Sets the edge to be empty.

◆ set_next_node()

void tesseract::SquishedDawg::set_next_node	(	EDGE_REF	edge_ref,
		EDGE_REF	value
	)

inlineprivate

Sets the next node link for this edge.

◆ unichar_ids_of()

void tesseract::SquishedDawg::unichar_ids_of	(	NODE_REF	node,
		NodeChildVector *	vec,
		bool	word_end
	)		const

inlinevirtual

Fills the given NodeChildVector with all the unichar ids (and the corresponding EDGE_REFs) for which there is an edge out of this node.

Implements tesseract::Dawg.

◆ write_squished_dawg() [1/2]

bool tesseract::SquishedDawg::write_squished_dawg ( TFile * file )

Writes the squished/reduced Dawg to a file.

◆ write_squished_dawg() [2/2]

bool tesseract::SquishedDawg::write_squished_dawg ( const char * filename )

inline

Opens the file with the given filename and writes the squished/reduced Dawg to the file.

Member Data Documentation

◆ edges_

EDGE_ARRAY tesseract::SquishedDawg::edges_

private

◆ num_edges_

int32_t tesseract::SquishedDawg::num_edges_

private

◆ num_forward_edges_in_node0

int tesseract::SquishedDawg::num_forward_edges_in_node0

private

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

/home/stephane/src/tesseract/src/dict/dawg.h
/home/stephane/src/tesseract/src/dict/dawg.cpp

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ SquishedDawg() [1/3]

◆ SquishedDawg() [2/3]

◆ SquishedDawg() [3/3]

◆ ~SquishedDawg()

Member Function Documentation

◆ backward_edge()

◆ build_node_map()

◆ clear_all_edges()

◆ clear_marker_flag()

◆ edge_char_of()

◆ edge_letter()

◆ edge_occupied()

◆ end_of_word()

◆ forward_edge()

◆ last_edge()

◆ Load()

◆ next_node()

◆ num_forward_edges()

◆ NumEdges()

◆ print_all()

◆ print_edge()

◆ print_node()

◆ read_squished_dawg()

◆ set_empty_edge()

◆ set_next_node()

◆ unichar_ids_of()

◆ write_squished_dawg() [1/2]

◆ write_squished_dawg() [2/2]

Member Data Documentation

◆ edges_

◆ num_edges_

◆ num_forward_edges_in_node0