#include <ratngs.h>

Inheritance diagram for WERD_CHOICE:

Collaboration diagram for WERD_CHOICE:

[legend]

Public Member Functions
	WERD_CHOICE (const UNICHARSET *unicharset)

	WERD_CHOICE (const UNICHARSET *unicharset, int reserved)

	WERD_CHOICE (const char src_string, const char src_lengths, float src_rating, float src_certainty, uint8_t src_permuter, const UNICHARSET &unicharset)

	WERD_CHOICE (const char *src_string, const UNICHARSET &unicharset)

	WERD_CHOICE (const WERD_CHOICE &word)

	~WERD_CHOICE ()

const UNICHARSET *	unicharset () const

int	length () const

float	adjust_factor () const

void	set_adjust_factor (float factor)

const UNICHAR_ID *	unichar_ids () const

UNICHAR_ID	unichar_id (int index) const

int	state (int index) const

tesseract::ScriptPos	BlobPosition (int index) const

float	rating () const

float	certainty () const

float	certainty (int index) const

float	min_x_height () const

float	max_x_height () const

void	set_x_heights (float min_height, float max_height)

uint8_t	permuter () const

const char *	permuter_name () const

BLOB_CHOICE_LIST *	blob_choices (int index, MATRIX *ratings) const

MATRIX_COORD	MatrixCoord (int index) const

void	set_unichar_id (UNICHAR_ID unichar_id, int index)

bool	dangerous_ambig_found () const

void	set_dangerous_ambig_found_ (bool value)

void	set_rating (float new_val)

void	set_certainty (float new_val)

void	set_permuter (uint8_t perm)

void	set_length (int len)

void	double_the_size ()
	Make more space in unichar_id_ and fragment_lengths_ arrays. More...

void	init (int reserved)

void	init (const char src_string, const char src_lengths, float src_rating, float src_certainty, uint8_t src_permuter)

void	make_bad ()
	Set the fields in this choice to be default (bad) values. More...

void	append_unichar_id_space_allocated (UNICHAR_ID unichar_id, int blob_count, float rating, float certainty)

void	append_unichar_id (UNICHAR_ID unichar_id, int blob_count, float rating, float certainty)

void	set_unichar_id (UNICHAR_ID unichar_id, int blob_count, float rating, float certainty, int index)

void	set_blob_choice (int index, int blob_count, const BLOB_CHOICE *blob_choice)

bool	contains_unichar_id (UNICHAR_ID unichar_id) const

void	remove_unichar_ids (int index, int num)

void	remove_last_unichar_id ()

void	remove_unichar_id (int index)

bool	has_rtl_unichar_id () const

void	reverse_and_mirror_unichar_ids ()

void	punct_stripped (int start_core, int end_core) const

void	GetNonSuperscriptSpan (int start, int end) const

WERD_CHOICE	shallow_copy (int start, int end) const

void	string_and_lengths (STRING word_str, STRING word_lengths_str) const

const STRING	debug_string () const

bool	ContainsAnyNonSpaceDelimited () const

bool	IsAllSpaces () const

bool	set_unichars_in_script_order (bool in_script_order)

bool	unichars_in_script_order () const

const STRING &	unichar_string () const

const STRING &	unichar_lengths () const

void	SetScriptPositions (bool small_caps, TWERD *word, int debug=0)

void	SetScriptPositions (const tesseract::ScriptPos *positions, int length)

void	SetAllScriptPositions (tesseract::ScriptPos position)

int	GetTopScriptID () const

void	UpdateStateForSplit (int blob_position)

int	TotalOfStates () const

void	print () const

void	print (const char *msg) const

void	print_state (const char *msg) const

void	DisplaySegmentation (TWERD *word)

WERD_CHOICE &	operator+= (const WERD_CHOICE &second)

WERD_CHOICE &	operator= (const WERD_CHOICE &source)

Public Member Functions inherited from ELIST_LINK
	ELIST_LINK ()

	ELIST_LINK (const ELIST_LINK &)

void	operator= (const ELIST_LINK &)

Static Public Member Functions
static const char *	permuter_name (uint8_t permuter)

static tesseract::ScriptPos	ScriptPositionOf (bool print_debug, const UNICHARSET &unicharset, const TBOX &blob_box, UNICHAR_ID unichar_id)

Static Public Attributes
static const float	kBadRating = 100000.0

Private Attributes
const UNICHARSET *	unicharset_

UNICHAR_ID *	unichar_ids_

tesseract::ScriptPos *	script_pos_

int *	state_

float *	certainties_

int	reserved_

int	length_

float	adjust_factor_

float	rating_

float	certainty_

float	min_x_height_

float	max_x_height_

uint8_t	permuter_

bool	unichars_in_script_order_

bool	dangerous_ambig_found_

STRING	unichar_string_

STRING	unichar_lengths_

Constructor & Destructor Documentation

◆ WERD_CHOICE() [1/5]

WERD_CHOICE::WERD_CHOICE ( const UNICHARSET * unicharset )

inline

◆ WERD_CHOICE() [2/5]

WERD_CHOICE::WERD_CHOICE	(	const UNICHARSET *	unicharset,
		int	reserved
	)

inline

◆ WERD_CHOICE() [3/5]

WERD_CHOICE::WERD_CHOICE	(	const char *	src_string,
		const char *	src_lengths,
		float	src_rating,
		float	src_certainty,
		uint8_t	src_permuter,
		const UNICHARSET &	unicharset
	)

inline

◆ WERD_CHOICE() [4/5]

WERD_CHOICE::WERD_CHOICE	(	const char *	src_string,
		const UNICHARSET &	unicharset
	)

WERD_CHOICE::WERD_CHOICE

Constructor to build a WERD_CHOICE from the given string. The function assumes that src_string is not nullptr.

◆ WERD_CHOICE() [5/5]

WERD_CHOICE::WERD_CHOICE ( const WERD_CHOICE & word )

inline

◆ ~WERD_CHOICE()

WERD_CHOICE::~WERD_CHOICE ( )

WERD_CHOICE::~WERD_CHOICE

Member Function Documentation

◆ adjust_factor()

float WERD_CHOICE::adjust_factor ( ) const

inline

◆ append_unichar_id()

void WERD_CHOICE::append_unichar_id	(	UNICHAR_ID	unichar_id,
		int	blob_count,
		float	rating,
		float	certainty
	)

append_unichar_id

Make sure there is enough space in the word for the new unichar id and call append_unichar_id_space_allocated().

◆ append_unichar_id_space_allocated()

void WERD_CHOICE::append_unichar_id_space_allocated	(	UNICHAR_ID	unichar_id,
		int	blob_count,
		float	rating,
		float	certainty
	)

inline

This function assumes that there is enough space reserved in the WERD_CHOICE for adding another unichar. This is an efficient alternative to append_unichar_id().

◆ blob_choices()

BLOB_CHOICE_LIST * WERD_CHOICE::blob_choices	(	int	index,
		MATRIX *	ratings
	)		const

◆ BlobPosition()

tesseract::ScriptPos WERD_CHOICE::BlobPosition ( int index ) const

inline

◆ certainty() [1/2]

float WERD_CHOICE::certainty ( ) const

inline

◆ certainty() [2/2]

float WERD_CHOICE::certainty ( int index ) const

inline

◆ contains_unichar_id()

bool WERD_CHOICE::contains_unichar_id ( UNICHAR_ID unichar_id ) const

contains_unichar_id

Returns true if unichar_ids_ contain the given unichar_id, false otherwise.

◆ ContainsAnyNonSpaceDelimited()

bool WERD_CHOICE::ContainsAnyNonSpaceDelimited ( ) const

inline

◆ dangerous_ambig_found()

bool WERD_CHOICE::dangerous_ambig_found ( ) const

inline

◆ debug_string()

const STRING WERD_CHOICE::debug_string ( ) const

inline

◆ DisplaySegmentation()

void WERD_CHOICE::DisplaySegmentation ( TWERD * word )

◆ double_the_size()

void WERD_CHOICE::double_the_size ( )

inline

Make more space in unichar_id_ and fragment_lengths_ arrays.

◆ GetNonSuperscriptSpan()

void WERD_CHOICE::GetNonSuperscriptSpan	(	int *	start,
		int *	end
	)		const

◆ GetTopScriptID()

int WERD_CHOICE::GetTopScriptID ( ) const

◆ has_rtl_unichar_id()

bool WERD_CHOICE::has_rtl_unichar_id ( ) const

has_rtl_unichar_id

Returns true if unichar_ids contain at least one "strongly" RTL unichar.

◆ init() [1/2]

void WERD_CHOICE::init ( int reserved )

inline

Initializes WERD_CHOICE - reserves length slots in unichar_ids_ and fragment_length_ arrays. Sets other values to default (blank) values.

◆ init() [2/2]

void WERD_CHOICE::init	(	const char *	src_string,
		const char *	src_lengths,
		float	src_rating,
		float	src_certainty,
		uint8_t	src_permuter
	)

Helper function to build a WERD_CHOICE from the given string, fragment lengths, rating, certainty and permuter. The function assumes that src_string is not nullptr. src_lengths argument could be nullptr, in which case the unichars in src_string are assumed to all be of length 1.

WERD_CHOICE::init

Helper function to build a WERD_CHOICE from the given string, fragment lengths, rating, certainty and permuter.

The function assumes that src_string is not nullptr. src_lengths argument could be nullptr, in which case the unichars in src_string are assumed to all be of length 1.

◆ IsAllSpaces()

bool WERD_CHOICE::IsAllSpaces ( ) const

inline

◆ length()

int WERD_CHOICE::length ( ) const

inline

◆ make_bad()

void WERD_CHOICE::make_bad ( )

inline

Set the fields in this choice to be default (bad) values.

◆ MatrixCoord()

MATRIX_COORD WERD_CHOICE::MatrixCoord ( int index ) const

◆ max_x_height()

float WERD_CHOICE::max_x_height ( ) const

inline

◆ min_x_height()

float WERD_CHOICE::min_x_height ( ) const

inline

◆ operator+=()

WERD_CHOICE & WERD_CHOICE::operator+= ( const WERD_CHOICE & second )

WERD_CHOICE::operator+=

Cat a second word rating on the end of this current one. The ratings are added and the confidence is the min. If the permuters are NOT the same the permuter is set to COMPOUND_PERM

◆ operator=()

WERD_CHOICE & WERD_CHOICE::operator= ( const WERD_CHOICE & source )

WERD_CHOICE::operator=

Allocate enough memory to hold a copy of source and copy over all the information from source to this WERD_CHOICE.

◆ permuter()

uint8_t WERD_CHOICE::permuter ( ) const

inline

◆ permuter_name() [1/2]

const char * WERD_CHOICE::permuter_name ( uint8_t permuter )

static

◆ permuter_name() [2/2]

const char * WERD_CHOICE::permuter_name ( ) const

◆ print() [1/2]

void WERD_CHOICE::print ( ) const

inline

◆ print() [2/2]

void WERD_CHOICE::print ( const char * msg ) const

WERD_CHOICE::print

Print WERD_CHOICE to stdout.

◆ print_state()

void WERD_CHOICE::print_state ( const char * msg ) const

◆ punct_stripped()

void WERD_CHOICE::punct_stripped	(	int *	start,
		int *	end
	)		const

punct_stripped

Returns the half-open interval of unichar_id indices [start, end) which enclose the core portion of this word – the part after stripping punctuation from the left and right.

◆ rating()

float WERD_CHOICE::rating ( ) const

inline

◆ remove_last_unichar_id()

void WERD_CHOICE::remove_last_unichar_id ( )

inline

◆ remove_unichar_id()

void WERD_CHOICE::remove_unichar_id ( int index )

inline

◆ remove_unichar_ids()

void WERD_CHOICE::remove_unichar_ids	(	int	start,
		int	num
	)

remove_unichar_ids

Removes num unichar ids starting from index start from unichar_ids_ and updates length_ and fragment_lengths_ to reflect this change. Note: this function does not modify rating_ and certainty_.

◆ reverse_and_mirror_unichar_ids()

void WERD_CHOICE::reverse_and_mirror_unichar_ids ( )

reverse_and_mirror_unichar_ids

Reverses and mirrors unichars in unichar_ids.

◆ ScriptPositionOf()

ScriptPos WERD_CHOICE::ScriptPositionOf	(	bool	print_debug,
		const UNICHARSET &	unicharset,
		const TBOX &	blob_box,
		UNICHAR_ID	unichar_id
	)

static

◆ set_adjust_factor()

void WERD_CHOICE::set_adjust_factor ( float factor )

inline

◆ set_blob_choice()

void WERD_CHOICE::set_blob_choice	(	int	index,
		int	blob_count,
		const BLOB_CHOICE *	blob_choice
	)

◆ set_certainty()

void WERD_CHOICE::set_certainty ( float new_val )

inline

◆ set_dangerous_ambig_found_()

void WERD_CHOICE::set_dangerous_ambig_found_ ( bool value )

inline

◆ set_length()

void WERD_CHOICE::set_length ( int len )

inline

◆ set_permuter()

void WERD_CHOICE::set_permuter ( uint8_t perm )

inline

◆ set_rating()

void WERD_CHOICE::set_rating ( float new_val )

inline

◆ set_unichar_id() [1/2]

void WERD_CHOICE::set_unichar_id	(	UNICHAR_ID	unichar_id,
		int	index
	)

inline

◆ set_unichar_id() [2/2]

void WERD_CHOICE::set_unichar_id	(	UNICHAR_ID	unichar_id,
		int	blob_count,
		float	rating,
		float	certainty,
		int	index
	)

inline

◆ set_unichars_in_script_order()

bool WERD_CHOICE::set_unichars_in_script_order ( bool in_script_order )

inline

◆ set_x_heights()

void WERD_CHOICE::set_x_heights	(	float	min_height,
		float	max_height
	)

inline

◆ SetAllScriptPositions()

void WERD_CHOICE::SetAllScriptPositions ( tesseract::ScriptPos position )

◆ SetScriptPositions() [1/2]

void WERD_CHOICE::SetScriptPositions	(	bool	small_caps,
		TWERD *	word,
		int	debug = `0`
	)

◆ SetScriptPositions() [2/2]

void WERD_CHOICE::SetScriptPositions	(	const tesseract::ScriptPos *	positions,
		int	length
	)

◆ shallow_copy()

WERD_CHOICE WERD_CHOICE::shallow_copy	(	int	start,
		int	end
	)		const

◆ state()

int WERD_CHOICE::state ( int index ) const

inline

◆ string_and_lengths()

void WERD_CHOICE::string_and_lengths	(	STRING *	word_str,
		STRING *	word_lengths_str
	)		const

string_and_lengths

Populates the given word_str with unichars from unichar_ids and and word_lengths_str with the corresponding unichar lengths.

◆ TotalOfStates()

int WERD_CHOICE::TotalOfStates ( ) const

◆ unichar_id()

UNICHAR_ID WERD_CHOICE::unichar_id ( int index ) const

inline

◆ unichar_ids()

const UNICHAR_ID* WERD_CHOICE::unichar_ids ( ) const

inline

◆ unichar_lengths()

const STRING& WERD_CHOICE::unichar_lengths ( ) const

inline

◆ unichar_string()

const STRING& WERD_CHOICE::unichar_string ( ) const

inline

◆ unichars_in_script_order()

bool WERD_CHOICE::unichars_in_script_order ( ) const

inline

◆ unicharset()

const UNICHARSET* WERD_CHOICE::unicharset ( ) const

inline

◆ UpdateStateForSplit()

void WERD_CHOICE::UpdateStateForSplit ( int blob_position )

Member Data Documentation

◆ adjust_factor_

float WERD_CHOICE::adjust_factor_

private

◆ certainties_

float* WERD_CHOICE::certainties_

private

◆ certainty_

float WERD_CHOICE::certainty_

private

◆ dangerous_ambig_found_

bool WERD_CHOICE::dangerous_ambig_found_

private

◆ kBadRating

const float WERD_CHOICE::kBadRating = 100000.0

static

◆ length_

int WERD_CHOICE::length_

private

◆ max_x_height_

float WERD_CHOICE::max_x_height_

private

◆ min_x_height_

float WERD_CHOICE::min_x_height_

private

◆ permuter_

uint8_t WERD_CHOICE::permuter_

private

◆ rating_

float WERD_CHOICE::rating_

private

◆ reserved_

int WERD_CHOICE::reserved_

private

◆ script_pos_

tesseract::ScriptPos* WERD_CHOICE::script_pos_

private

◆ state_

int* WERD_CHOICE::state_

private

◆ unichar_ids_

UNICHAR_ID* WERD_CHOICE::unichar_ids_

private

◆ unichar_lengths_

STRING WERD_CHOICE::unichar_lengths_

mutableprivate

◆ unichar_string_

STRING WERD_CHOICE::unichar_string_

mutableprivate

◆ unichars_in_script_order_

bool WERD_CHOICE::unichars_in_script_order_

private

◆ unicharset_

const UNICHARSET* WERD_CHOICE::unicharset_

private

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

/home/stephane/src/tesseract/src/ccstruct/ratngs.h
/home/stephane/src/tesseract/src/ccstruct/ratngs.cpp

Public Member Functions

Static Public Member Functions

Static Public Attributes

Private Attributes

Constructor & Destructor Documentation

◆ WERD_CHOICE() [1/5]

◆ WERD_CHOICE() [2/5]

◆ WERD_CHOICE() [3/5]

◆ WERD_CHOICE() [4/5]

◆ WERD_CHOICE() [5/5]

◆ ~WERD_CHOICE()

Member Function Documentation

◆ adjust_factor()

◆ append_unichar_id()

◆ append_unichar_id_space_allocated()

◆ blob_choices()

◆ BlobPosition()

◆ certainty() [1/2]

◆ certainty() [2/2]

◆ contains_unichar_id()

◆ ContainsAnyNonSpaceDelimited()

◆ dangerous_ambig_found()

◆ debug_string()

◆ DisplaySegmentation()

◆ double_the_size()

◆ GetNonSuperscriptSpan()

◆ GetTopScriptID()

◆ has_rtl_unichar_id()

◆ init() [1/2]

◆ init() [2/2]

◆ IsAllSpaces()

◆ length()

◆ make_bad()

◆ MatrixCoord()

◆ max_x_height()

◆ min_x_height()

◆ operator+=()

◆ operator=()

◆ permuter()

◆ permuter_name() [1/2]

◆ permuter_name() [2/2]

◆ print() [1/2]

◆ print() [2/2]

◆ print_state()

◆ punct_stripped()

◆ rating()

◆ remove_last_unichar_id()

◆ remove_unichar_id()

◆ remove_unichar_ids()

◆ reverse_and_mirror_unichar_ids()

◆ ScriptPositionOf()

◆ set_adjust_factor()

◆ set_blob_choice()

◆ set_certainty()

◆ set_dangerous_ambig_found_()

◆ set_length()

◆ set_permuter()

◆ set_rating()

◆ set_unichar_id() [1/2]

◆ set_unichar_id() [2/2]

◆ set_unichars_in_script_order()

◆ set_x_heights()

◆ SetAllScriptPositions()

◆ SetScriptPositions() [1/2]

◆ SetScriptPositions() [2/2]

◆ shallow_copy()

◆ state()

◆ string_and_lengths()

◆ TotalOfStates()

◆ unichar_id()

◆ unichar_ids()

◆ unichar_lengths()

◆ unichar_string()

◆ unichars_in_script_order()

◆ unicharset()

◆ UpdateStateForSplit()

Member Data Documentation

◆ adjust_factor_

◆ certainties_

◆ certainty_