WERD_CHOICE Class Reference

#include <ratngs.h>

Inheritance diagram for WERD_CHOICE:

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 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	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 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 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)
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 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

Detailed Description

Definition at line 269 of file ratngs.h.

Constructor & Destructor Documentation

WERD_CHOICE() [1/5]

WERD_CHOICE::WERD_CHOICE ( const UNICHARSET *  unicharset )

inline

Definition at line 274 of file ratngs.h.

    : unicharset_(unicharset) { this->init(8); }

WERD_CHOICE() [2/5]

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

inline

Definition at line 276 of file ratngs.h.

    : unicharset_(unicharset) { this->init(reserved); }

WERD_CHOICE() [3/5]

WERD_CHOICE::WERD_CHOICE ( const char *  src_string, const char *  src_lengths, float  src_rating, float  src_certainty, uinT8  src_permuter, const UNICHARSET &  unicharset  )

inline

Definition at line 278 of file ratngs.h.

    : unicharset_(&unicharset) {
    this->init(src_string, src_lengths, src_rating,
               src_certainty, src_permuter);
  }

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 NULL.

Definition at line 199 of file ratngs.cpp.

    : unicharset_(&unicharset){
  GenericVector<UNICHAR_ID> encoding;
  GenericVector<char> lengths;
  string cleaned = unicharset.CleanupString(src_string);
  if (unicharset.encode_string(cleaned.c_str(), true, &encoding, &lengths,
                               NULL)) {
    lengths.push_back('\0');
    STRING src_lengths = &lengths[0];
    this->init(cleaned.c_str(), src_lengths.string(), 0.0, 0.0, NO_PERM);
  } else {  // There must have been an invalid unichar in the string.
    this->init(8);
    this->make_bad();
  }
}

WERD_CHOICE() [5/5]

WERD_CHOICE::WERD_CHOICE ( const WERD_CHOICE &  word )

inline

Definition at line 289 of file ratngs.h.

      : ELIST_LINK(word), unicharset_(word.unicharset_) {
    this->init(word.length());
    this->operator=(word);
  }

~WERD_CHOICE()

WERD_CHOICE::~WERD_CHOICE

(

)

WERD_CHOICE::~WERD_CHOICE

Definition at line 257 of file ratngs.cpp.

                          {
  delete[] unichar_ids_;
  delete[] script_pos_;
  delete[] state_;
  delete[] certainties_;
}

Member Function Documentation

adjust_factor()

float WERD_CHOICE::adjust_factor ( ) const

inline

Definition at line 302 of file ratngs.h.

                              {
    return adjust_factor_;
  }

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().

Definition at line 449 of file ratngs.cpp.

                                   {
  if (length_ == reserved_) {
    this->double_the_size();
  }
  this->append_unichar_id_space_allocated(unichar_id, blob_count,
                                          rating, certainty);
}

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().

Definition at line 448 of file ratngs.h.

                                     {
    assert(reserved_ > length_);
    length_++;
    this->set_unichar_id(unichar_id, blob_count,
                         rating, certainty, length_-1);
  }

blob_choices()

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

Definition at line 271 of file ratngs.cpp.

                                                                            {
  MATRIX_COORD coord = MatrixCoord(index);
  BLOB_CHOICE_LIST* result = ratings->get(coord.col, coord.row);
  if (result == NULL) {
    result = new BLOB_CHOICE_LIST;
    ratings->put(coord.col, coord.row, result);
  }
  return result;
}

BlobPosition()

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

inline

Definition at line 318 of file ratngs.h.

                                                   {
    if (index < 0 || index >= length_)
      return tesseract::SP_NORMAL;
    return script_pos_[index];
  }

certainty() [1/2]

float WERD_CHOICE::certainty ( ) const

inline

Definition at line 326 of file ratngs.h.

                                 {
    return certainty_;
  }

certainty() [2/2]

float WERD_CHOICE::certainty ( int  index ) const

inline

Definition at line 329 of file ratngs.h.

                                          {
    return certainties_[index];
  }

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.

Definition at line 307 of file ratngs.cpp.

                                                                 {
  for (int i = 0; i < length_; ++i) {
    if (unichar_ids_[i] == unichar_id) {
      return true;
    }
  }
  return false;
}

ContainsAnyNonSpaceDelimited()

bool WERD_CHOICE::ContainsAnyNonSpaceDelimited ( ) const

inline

Definition at line 510 of file ratngs.h.

                                            {
    for (int i = 0; i < length_; ++i) {
      if (!unicharset_->IsSpaceDelimited(unichar_ids_[i])) return true;
    }
    return false;
  }

dangerous_ambig_found()

bool WERD_CHOICE::dangerous_ambig_found ( ) const

inline

Definition at line 359 of file ratngs.h.

                                     {
    return dangerous_ambig_found_;
  }

debug_string()

const STRING WERD_CHOICE::debug_string ( ) const

inline

Definition at line 501 of file ratngs.h.

                                    {
    STRING word_str;
    for (int i = 0; i < length_; ++i) {
      word_str += unicharset_->debug_str(unichar_ids_[i]);
      word_str += " ";
    }
    return word_str;
  }

DisplaySegmentation()

void WERD_CHOICE::DisplaySegmentation

(

TWERD *

word

)

Definition at line 750 of file ratngs.cpp.

                                                 {
#ifndef GRAPHICS_DISABLED
  // Number of different colors to draw with.
  const int kNumColors = 6;
  static ScrollView *segm_window = NULL;
  // Check the state against the static prev_drawn_state.
  static GenericVector<int> prev_drawn_state;
  bool already_done = prev_drawn_state.size() == length_;
  if (!already_done) prev_drawn_state.init_to_size(length_, 0);
  for (int i = 0; i < length_; ++i) {
    if (prev_drawn_state[i] != state_[i]) {
      already_done = false;
    }
    prev_drawn_state[i] = state_[i];
  }
  if (already_done || word->blobs.empty()) return;

  // Create the window if needed.
  if (segm_window == NULL) {
    segm_window = new ScrollView("Segmentation", 5, 10, 500, 256,
                                 2000.0, 256.0, true);
  } else {
    segm_window->Clear();
  }

  TBOX bbox;
  int blob_index = 0;
  for (int c = 0; c < length_; ++c) {
    ScrollView::Color color =
        static_cast<ScrollView::Color>(c % kNumColors + 3);
    for (int i = 0; i < state_[c]; ++i, ++blob_index) {
      TBLOB* blob = word->blobs[blob_index];
      bbox += blob->bounding_box();
      blob->plot(segm_window, color, color);
    }
  }
  segm_window->ZoomToRectangle(bbox.left(), bbox.top(),
                               bbox.right(), bbox.bottom());
  segm_window->Update();
  window_wait(segm_window);
#endif
}

double_the_size()

void WERD_CHOICE::double_the_size ( )

inline

Make more space in unichar_id_ and fragment_lengths_ arrays.

Definition at line 383 of file ratngs.h.

                                {
    if (reserved_ > 0) {
      unichar_ids_ = GenericVector<UNICHAR_ID>::double_the_size_memcpy(
          reserved_, unichar_ids_);
      script_pos_ = GenericVector<tesseract::ScriptPos>::double_the_size_memcpy(
          reserved_, script_pos_);
      state_ = GenericVector<int>::double_the_size_memcpy(
          reserved_, state_);
      certainties_ = GenericVector<float>::double_the_size_memcpy(
          reserved_, certainties_);
      reserved_ *= 2;
    } else {
      unichar_ids_ = new UNICHAR_ID[1];
      script_pos_ = new tesseract::ScriptPos[1];
      state_ = new int[1];
      certainties_ = new float[1];
      reserved_ = 1;
    }
  }

GetNonSuperscriptSpan()

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

Definition at line 378 of file ratngs.cpp.

                                                                    {
  int end = length();
  while (end > 0 &&
         unicharset_->get_isdigit(unichar_ids_[end - 1]) &&
         BlobPosition(end - 1) == tesseract::SP_SUPERSCRIPT) {
    end--;
  }
  int start = 0;
  while (start < end &&
         unicharset_->get_isdigit(unichar_ids_[start]) &&
         BlobPosition(start) == tesseract::SP_SUPERSCRIPT) {
    start++;
  }
  *pstart = start;
  *pend = end;
}

GetTopScriptID()

int WERD_CHOICE::GetTopScriptID

(

)

const

Definition at line 656 of file ratngs.cpp.

                                      {
  int max_script = unicharset_->get_script_table_size();
  int *sid = new int[max_script];
  int x;
  for (x = 0; x < max_script; x++) sid[x] = 0;
  for (x = 0; x < length_; ++x) {
    int script_id = unicharset_->get_script(unichar_id(x));
    sid[script_id]++;
  }
  if (unicharset_->han_sid() != unicharset_->null_sid()) {
    // Add the Hiragana & Katakana counts to Han and zero them out.
    if (unicharset_->hiragana_sid() != unicharset_->null_sid()) {
      sid[unicharset_->han_sid()] += sid[unicharset_->hiragana_sid()];
      sid[unicharset_->hiragana_sid()] = 0;
    }
    if (unicharset_->katakana_sid() != unicharset_->null_sid()) {
      sid[unicharset_->han_sid()] += sid[unicharset_->katakana_sid()];
      sid[unicharset_->katakana_sid()] = 0;
    }
  }
  // Note that high script ID overrides lower one on a tie, thus biasing
  // towards non-Common script (if sorted that way in unicharset file).
  int max_sid = 0;
  for (x = 1; x < max_script; x++)
    if (sid[x] >= sid[max_sid]) max_sid = x;
  if (sid[max_sid] < length_ / 2)
    max_sid = unicharset_->null_sid();
  delete[] sid;
  return max_sid;
}

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.

Definition at line 412 of file ratngs.cpp.

                                           {
  int i;
  for (i = 0; i < length_; ++i) {
    UNICHARSET::Direction dir = unicharset_->get_direction(unichar_ids_[i]);
    if (dir == UNICHARSET::U_RIGHT_TO_LEFT ||
        dir == UNICHARSET::U_RIGHT_TO_LEFT_ARABIC) {
      return true;
    }
  }
  return false;
}

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.

Definition at line 405 of file ratngs.h.

                                 {
    reserved_ = reserved;
    if (reserved > 0) {
      unichar_ids_ = new UNICHAR_ID[reserved];
      script_pos_ = new tesseract::ScriptPos[reserved];
      state_ = new int[reserved];
      certainties_ = new float[reserved];
    } else {
      unichar_ids_ = NULL;
      script_pos_ = NULL;
      state_ = NULL;
      certainties_ = NULL;
    }
    length_ = 0;
    adjust_factor_ = 1.0f;
    rating_ = 0.0;
    certainty_ = MAX_FLOAT32;
    min_x_height_ = 0.0f;
    max_x_height_ = MAX_FLOAT32;
    permuter_ = NO_PERM;
    unichars_in_script_order_ = false;  // Tesseract is strict left-to-right.
    dangerous_ambig_found_ = false;
  }

init() [2/2]

void WERD_CHOICE::init	(	const char *	src_string,
		const char *	src_lengths,
		float	src_rating,
		float	src_certainty,
		uinT8	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 NULL. src_lengths argument could be NULL, 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 NULL. src_lengths argument could be NULL, in which case the unichars in src_string are assumed to all be of length 1.

Definition at line 226 of file ratngs.cpp.

                                           {
  int src_string_len = strlen(src_string);
  if (src_string_len == 0) {
    this->init(8);
  } else {
    this->init(src_lengths ? strlen(src_lengths): src_string_len);
    length_ = reserved_;
    int offset = 0;
    for (int i = 0; i < length_; ++i) {
      int unichar_length = src_lengths ? src_lengths[i] : 1;
      unichar_ids_[i] =
          unicharset_->unichar_to_id(src_string+offset, unichar_length);
      state_[i] = 1;
      certainties_[i] = src_certainty;
      offset += unichar_length;
    }
  }
  adjust_factor_ = 1.0f;
  rating_ = src_rating;
  certainty_ = src_certainty;
  permuter_ = src_permuter;
  dangerous_ambig_found_ = false;
}

IsAllSpaces()

bool WERD_CHOICE::IsAllSpaces ( ) const

inline

Definition at line 517 of file ratngs.h.

                           {
    for (int i = 0; i < length_; ++i) {
      if (unichar_ids_[i] != UNICHAR_SPACE) return false;
    }
    return true;
  }

length()

int WERD_CHOICE::length ( ) const

inline

Definition at line 299 of file ratngs.h.

                            {
    return length_;
  }

make_bad()

void WERD_CHOICE::make_bad ( )

inline

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

Definition at line 439 of file ratngs.h.

                         {
    length_ = 0;
    rating_ = kBadRating;
    certainty_ = -MAX_FLOAT32;
  }

MatrixCoord()

MATRIX_COORD WERD_CHOICE::MatrixCoord

(

int

index

)

const

Definition at line 283 of file ratngs.cpp.

                                                     {
  int col = 0;
  for (int i = 0; i < index; ++i)
    col += state_[i];
  int row = col + state_[index] - 1;
  return MATRIX_COORD(col, row);
}

max_x_height()

float WERD_CHOICE::max_x_height ( ) const

inline

Definition at line 335 of file ratngs.h.

                                    {
    return max_x_height_;
  }

min_x_height()

float WERD_CHOICE::min_x_height ( ) const

inline

Definition at line 332 of file ratngs.h.

                                    {
    return min_x_height_;
  }

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

Definition at line 466 of file ratngs.cpp.

                                                                 {
  ASSERT_HOST(unicharset_ == second.unicharset_);
  while (reserved_ < length_ + second.length()) {
    this->double_the_size();
  }
  const UNICHAR_ID *other_unichar_ids = second.unichar_ids();
  for (int i = 0; i < second.length(); ++i) {
    unichar_ids_[length_ + i] = other_unichar_ids[i];
    state_[length_ + i] = second.state_[i];
    certainties_[length_ + i] = second.certainties_[i];
    script_pos_[length_ + i] = second.BlobPosition(i);
  }
  length_ += second.length();
  if (second.adjust_factor_ > adjust_factor_)
    adjust_factor_ = second.adjust_factor_;
  rating_ += second.rating();  // add ratings
  if (second.certainty() < certainty_) // take min
    certainty_ = second.certainty();
  if (second.dangerous_ambig_found_)
    dangerous_ambig_found_ = true;
  if (permuter_ == NO_PERM) {
    permuter_ = second.permuter();
  } else if (second.permuter() != NO_PERM &&
             second.permuter() != permuter_) {
    permuter_ = COMPOUND_PERM;
  }
  return *this;
}

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.

Definition at line 502 of file ratngs.cpp.

                                                             {
  while (reserved_ < source.length()) {
    this->double_the_size();
  }

  unicharset_ = source.unicharset_;
  const UNICHAR_ID *other_unichar_ids = source.unichar_ids();
  for (int i = 0; i < source.length(); ++i) {
    unichar_ids_[i] = other_unichar_ids[i];
    state_[i] = source.state_[i];
    certainties_[i] = source.certainties_[i];
    script_pos_[i] = source.BlobPosition(i);
  }
  length_ = source.length();
  adjust_factor_ = source.adjust_factor_;
  rating_ = source.rating();
  certainty_ = source.certainty();
  min_x_height_ = source.min_x_height();
  max_x_height_ = source.max_x_height();
  permuter_ = source.permuter();
  dangerous_ambig_found_ = source.dangerous_ambig_found_;
  return *this;
}

permuter()

uinT8 WERD_CHOICE::permuter ( ) const

inline

Definition at line 342 of file ratngs.h.

                                {
    return permuter_;
  }

permuter_name() [1/2]

const char * WERD_CHOICE::permuter_name ( uinT8  permuter )

static

Definition at line 175 of file ratngs.cpp.

                                                     {
  return kPermuterTypeNames[permuter];
}

permuter_name() [2/2]

const char * WERD_CHOICE::permuter_name

(

)

const

Definition at line 264 of file ratngs.cpp.

                                             {
  return kPermuterTypeNames[permuter_];
}

print() [1/2]

void WERD_CHOICE::print ( ) const

inline

Definition at line 576 of file ratngs.h.

{ this->print(""); }

print() [2/2]

void WERD_CHOICE::print

(

const char *

msg

)

const

WERD_CHOICE::print

Print WERD_CHOICE to stdout.

Definition at line 713 of file ratngs.cpp.

                                             {
  tprintf("%s : ", msg);
  for (int i = 0; i < length_; ++i) {
    tprintf("%s", unicharset_->id_to_unichar(unichar_ids_[i]));
  }
  tprintf(" : R=%g, C=%g, F=%g, Perm=%d, xht=[%g,%g], ambig=%d\n",
          rating_, certainty_, adjust_factor_, permuter_,
          min_x_height_, max_x_height_, dangerous_ambig_found_);
  tprintf("pos");
  for (int i = 0; i < length_; ++i) {
    tprintf("\t%s", ScriptPosToString(script_pos_[i]));
  }
  tprintf("\nstr");
  for (int i = 0; i < length_; ++i) {
    tprintf("\t%s", unicharset_->id_to_unichar(unichar_ids_[i]));
  }
  tprintf("\nstate:");
  for (int i = 0; i < length_; ++i) {
    tprintf("\t%d ", state_[i]);
  }
  tprintf("\nC");
  for (int i = 0; i < length_; ++i) {
    tprintf("\t%.3f", certainties_[i]);
  }
  tprintf("\n");
}

print_state()

void WERD_CHOICE::print_state

(

const char *

msg

)

const

Definition at line 741 of file ratngs.cpp.

                                                   {
  tprintf("%s", msg);
  for (int i = 0; i < length_; ++i)
    tprintf(" %d", state_[i]);
  tprintf("\n");
}

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.

Definition at line 364 of file ratngs.cpp.

                                                           {
  *start = 0;
  *end = length() - 1;
  while (*start < length() &&
         unicharset()->get_ispunctuation(unichar_id(*start))) {
    (*start)++;
  }
  while (*end > -1 &&
         unicharset()->get_ispunctuation(unichar_id(*end))) {
    (*end)--;
  }
  (*end)++;
}

rating()

float WERD_CHOICE::rating ( ) const

inline

Definition at line 323 of file ratngs.h.

                              {
    return rating_;
  }

remove_last_unichar_id()

void WERD_CHOICE::remove_last_unichar_id ( )

inline

Definition at line 479 of file ratngs.h.

{ --length_; }

remove_unichar_id()

void WERD_CHOICE::remove_unichar_id ( int  index )

inline

Definition at line 480 of file ratngs.h.

                                           {
    this->remove_unichar_ids(index, 1);
  }

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_.

Definition at line 323 of file ratngs.cpp.

                                                       {
  ASSERT_HOST(start >= 0 && start + num <= length_);
  // Accumulate the states to account for the merged blobs.
  for (int i = 0; i < num; ++i) {
    if (start > 0)
      state_[start - 1] += state_[start + i];
    else if (start + num < length_)
      state_[start + num] += state_[start + i];
  }
  for (int i = start; i + num < length_; ++i) {
    unichar_ids_[i] = unichar_ids_[i + num];
    script_pos_[i] = script_pos_[i + num];
    state_[i] = state_[i + num];
    certainties_[i] = certainties_[i + num];
  }
  length_ -= num;
}

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.

Definition at line 346 of file ratngs.cpp.

                                                 {
  for (int i = 0; i < length_ / 2; ++i) {
    UNICHAR_ID tmp_id = unichar_ids_[i];
    unichar_ids_[i] = unicharset_->get_mirror(unichar_ids_[length_-1-i]);
    unichar_ids_[length_-1-i] = unicharset_->get_mirror(tmp_id);
  }
  if (length_ % 2 != 0) {
    unichar_ids_[length_/2] = unicharset_->get_mirror(unichar_ids_[length_/2]);
  }
}

ScriptPositionOf()

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

static

Definition at line 618 of file ratngs.cpp.

                                                               {
  ScriptPos retval = tesseract::SP_NORMAL;
  int top = blob_box.top();
  int bottom = blob_box.bottom();
  int min_bottom, max_bottom, min_top, max_top;
  unicharset.get_top_bottom(unichar_id,
                            &min_bottom, &max_bottom,
                            &min_top, &max_top);

  int sub_thresh_top = min_top - kMinSubscriptOffset;
  int sub_thresh_bot = kBlnBaselineOffset - kMinSubscriptOffset;
  int sup_thresh_bot = max_bottom + kMinSuperscriptOffset;
  if (bottom <= kMaxDropCapBottom) {
    retval = tesseract::SP_DROPCAP;
  } else if (top < sub_thresh_top && bottom < sub_thresh_bot) {
    retval = tesseract::SP_SUBSCRIPT;
  } else if (bottom > sup_thresh_bot) {
    retval = tesseract::SP_SUPERSCRIPT;
  }

  if (print_debug) {
    const char *pos = ScriptPosToString(retval);
    tprintf("%s Character %s[bot:%d top: %d]  "
            "bot_range[%d,%d]  top_range[%d, %d] "
            "sub_thresh[bot:%d top:%d]  sup_thresh_bot %d\n",
            pos, unicharset.id_to_unichar(unichar_id),
            bottom, top,
            min_bottom, max_bottom, min_top, max_top,
            sub_thresh_bot, sub_thresh_top,
            sup_thresh_bot);
  }
  return retval;
}

set_adjust_factor()

void WERD_CHOICE::set_adjust_factor ( float  factor )

inline

Definition at line 305 of file ratngs.h.

                                       {
    adjust_factor_ = factor;
  }

set_blob_choice()

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

Definition at line 293 of file ratngs.cpp.

                                                                  {
  unichar_ids_[index] = blob_choice->unichar_id();
  script_pos_[index] = tesseract::SP_NORMAL;
  state_[index] = blob_count;
  certainties_[index] = blob_choice->certainty();
}

set_certainty()

void WERD_CHOICE::set_certainty ( float  new_val )

inline

Definition at line 368 of file ratngs.h.

                                           {
    certainty_ = new_val;
  }

set_dangerous_ambig_found_()

void WERD_CHOICE::set_dangerous_ambig_found_ ( bool  value )

inline

Definition at line 362 of file ratngs.h.

                                              {
    dangerous_ambig_found_ = value;
  }

set_length()

void WERD_CHOICE::set_length ( int  len )

inline

Definition at line 377 of file ratngs.h.

                                  {
    ASSERT_HOST(reserved_ >= len);
    length_ = len;
  }

set_permuter()

void WERD_CHOICE::set_permuter ( uinT8  perm )

inline

Definition at line 371 of file ratngs.h.

                                       {
    permuter_ = perm;
  }

set_rating()

void WERD_CHOICE::set_rating ( float  new_val )

inline

Definition at line 365 of file ratngs.h.

                                        {
    rating_ = new_val;
  }

set_unichar_id() [1/2]

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

inline

Definition at line 355 of file ratngs.h.

                                                               {
    assert(index < length_);
    unichar_ids_[index] = unichar_id;
  }

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

Definition at line 460 of file ratngs.h.

                                                                       {
    assert(index < length_);
    unichar_ids_[index] = unichar_id;
    state_[index] = blob_count;
    certainties_[index] = certainty;
    script_pos_[index] = tesseract::SP_NORMAL;
    rating_ += rating;
    if (certainty < certainty_) {
      certainty_ = certainty;
    }
  }

set_unichars_in_script_order()

bool WERD_CHOICE::set_unichars_in_script_order ( bool  in_script_order )

inline

Definition at line 527 of file ratngs.h.

                                                          {
    return unichars_in_script_order_ = in_script_order;
  }

set_x_heights()

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

inline

Definition at line 338 of file ratngs.h.

                                                                {
    min_x_height_ = min_height;
    max_x_height_ = max_height;
  }

SetAllScriptPositions()

void WERD_CHOICE::SetAllScriptPositions

(

tesseract::ScriptPos

position

)

Definition at line 612 of file ratngs.cpp.

                                                                   {
  for (int i = 0; i < length_; ++i)
    script_pos_[i] = position;
}

SetScriptPositions() [1/2]

void WERD_CHOICE::SetScriptPositions	(	bool	small_caps,
		TWERD *	word
	)

Definition at line 531 of file ratngs.cpp.

                                                                 {
  // Since WERD_CHOICE isn't supposed to depend on a Tesseract,
  // we don't have easy access to the flags Tesseract stores.  Therefore, debug
  // for this module is hard compiled in.
  int debug = 0;

  // Initialize to normal.
  for (int i = 0; i < length_; ++i)
    script_pos_[i] = tesseract::SP_NORMAL;
  if (word->blobs.empty() || word->NumBlobs() != TotalOfStates()) {
    return;
  }

  int position_counts[4];
  for (int i = 0; i < 4; i++) {
    position_counts[i] = 0;
  }

  int chunk_index = 0;
  for (int blob_index = 0; blob_index < length_; ++blob_index, ++chunk_index) {
    TBLOB* tblob = word->blobs[chunk_index];
    int uni_id = unichar_id(blob_index);
    TBOX blob_box = tblob->bounding_box();
    if (state_ != NULL) {
      for (int i = 1; i <  state_[blob_index]; ++i) {
        ++chunk_index;
        tblob = word->blobs[chunk_index];
        blob_box += tblob->bounding_box();
      }
    }
    script_pos_[blob_index] = ScriptPositionOf(false, *unicharset_, blob_box,
                                               uni_id);
    if (small_caps && script_pos_[blob_index] != tesseract::SP_DROPCAP) {
      script_pos_[blob_index] = tesseract::SP_NORMAL;
    }
    position_counts[script_pos_[blob_index]]++;
  }
  // If almost everything looks like a superscript or subscript,
  // we most likely just got the baseline wrong.
  if (position_counts[tesseract::SP_SUBSCRIPT] > 0.75 * length_ ||
      position_counts[tesseract::SP_SUPERSCRIPT] > 0.75 * length_) {
    if (debug >= 2) {
      tprintf("Most characters of %s are subscript or superscript.\n"
              "That seems wrong, so I'll assume we got the baseline wrong\n",
              unichar_string().string());
    }
    for (int i = 0; i < length_; i++) {
      ScriptPos sp = script_pos_[i];
      if (sp == tesseract::SP_SUBSCRIPT || sp == tesseract::SP_SUPERSCRIPT) {
        position_counts[sp]--;
        position_counts[tesseract::SP_NORMAL]++;
        script_pos_[i] = tesseract::SP_NORMAL;
      }
    }
  }

  if ((debug >= 1 && position_counts[tesseract::SP_NORMAL] < length_) ||
      debug >= 2) {
    tprintf("SetScriptPosition on %s\n", unichar_string().string());
    int chunk_index = 0;
    for (int blob_index = 0; blob_index < length_; ++blob_index) {
      if (debug >= 2 || script_pos_[blob_index] != tesseract::SP_NORMAL) {
        TBLOB* tblob = word->blobs[chunk_index];
        ScriptPositionOf(true, *unicharset_, tblob->bounding_box(),
                         unichar_id(blob_index));
      }
      chunk_index += state_ != NULL ? state_[blob_index] : 1;
    }
  }
}

SetScriptPositions() [2/2]

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

Definition at line 602 of file ratngs.cpp.

                                                 {
  ASSERT_HOST(length == length_);
  if (positions != script_pos_) {
    delete [] script_pos_;
    script_pos_ = new ScriptPos[length];
    memcpy(script_pos_, positions, sizeof(positions[0]) * length);
  }
}

shallow_copy()

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

Definition at line 395 of file ratngs.cpp.

                                                              {
  ASSERT_HOST(start >= 0 && start <= length_);
  ASSERT_HOST(end >= 0 && end <= length_);
  if (end < start) { end = start; }
  WERD_CHOICE retval(unicharset_, end - start);
  for (int i = start; i < end; i++) {
    retval.append_unichar_id_space_allocated(
        unichar_ids_[i], state_[i], 0.0f, certainties_[i]);
  }
  return retval;
}

state()

int WERD_CHOICE::state ( int  index ) const

inline

Definition at line 315 of file ratngs.h.

                                    {
    return state_[index];
  }

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.

Definition at line 430 of file ratngs.cpp.

                                                                     {
  *word_str = "";
  if (word_lengths_str != NULL) *word_lengths_str = "";
  for (int i = 0; i < length_; ++i) {
    const char *ch = unicharset_->id_to_unichar_ext(unichar_ids_[i]);
    *word_str += ch;
    if (word_lengths_str != NULL) {
      *word_lengths_str += strlen(ch);
    }
  }
}

TotalOfStates()

int WERD_CHOICE::TotalOfStates

(

)

const

Definition at line 700 of file ratngs.cpp.

                                     {
  int total_chunks = 0;
  for (int i = 0; i < length_; ++i) {
    total_chunks += state_[i];
  }
  return total_chunks;
}

unichar_id()

UNICHAR_ID WERD_CHOICE::unichar_id ( int  index ) const

inline

Definition at line 311 of file ratngs.h.

                                                {
    assert(index < length_);
    return unichar_ids_[index];
  }

unichar_ids()

const UNICHAR_ID* WERD_CHOICE::unichar_ids ( ) const

inline

Definition at line 308 of file ratngs.h.

                                               {
    return unichar_ids_;
  }

unichar_lengths()

const STRING& WERD_CHOICE::unichar_lengths ( ) const

inline

Definition at line 544 of file ratngs.h.

                                        {
    this->string_and_lengths(&unichar_string_, &unichar_lengths_);
    return unichar_lengths_;
  }

unichar_string()

const STRING& WERD_CHOICE::unichar_string ( ) const

inline

Definition at line 537 of file ratngs.h.

                                       {
    this->string_and_lengths(&unichar_string_, &unichar_lengths_);
    return unichar_string_;
  }

unichars_in_script_order()

bool WERD_CHOICE::unichars_in_script_order ( ) const

inline

Definition at line 531 of file ratngs.h.

                                        {
    return unichars_in_script_order_;
  }

unicharset()

const UNICHARSET* WERD_CHOICE::unicharset ( ) const

inline

Definition at line 296 of file ratngs.h.

                                       {
    return unicharset_;
  }

UpdateStateForSplit()

void WERD_CHOICE::UpdateStateForSplit

(

int

blob_position

)

Definition at line 688 of file ratngs.cpp.

                                                       {
  int total_chunks = 0;
  for (int i = 0; i < length_; ++i) {
    total_chunks += state_[i];
    if (total_chunks > blob_position) {
      ++state_[i];
      return;
    }
  }
}

Member Data Documentation

kBadRating

const float WERD_CHOICE::kBadRating = 100000.0

static

Definition at line 271 of file ratngs.h.

---

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

• /home/stweil/src/github/tesseract-ocr/tesseract/ccstruct/ratngs.h
• /home/stweil/src/github/tesseract-ocr/tesseract/ccstruct/ratngs.cpp