#include <ratngs.h>

Inheritance diagram for tesseract::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_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

bool	empty () const

unsigned	length () const

float	adjust_factor () const

void	set_adjust_factor (float factor)

const std::vector< UNICHAR_ID > &	unichar_ids () const

UNICHAR_ID	unichar_id (unsigned index) const

unsigned	state (unsigned index) const

ScriptPos	BlobPosition (unsigned index) const

float	rating () const

float	certainty () const

float	certainty (unsigned 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 (unsigned index, MATRIX *ratings) const

MATRIX_COORD	MatrixCoord (unsigned index) const

void	set_unichar_id (UNICHAR_ID unichar_id, unsigned 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 (unsigned len)

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

void	init (unsigned 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, unsigned index)

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

bool	contains_unichar_id (UNICHAR_ID unichar_id) const

void	remove_unichar_ids (unsigned index, int num)

void	remove_last_unichar_id ()

void	remove_unichar_id (unsigned index)

bool	has_rtl_unichar_id () const

void	reverse_and_mirror_unichar_ids ()

void	punct_stripped (unsigned start_core, unsigned end_core) const

void	GetNonSuperscriptSpan (int start, int end) const

WERD_CHOICE	shallow_copy (unsigned start, unsigned end) const

void	string_and_lengths (std::string word_str, std::string word_lengths_str) const

std::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

std::string &	unichar_string ()

const std::string &	unichar_string () const

const std::string &	unichar_lengths () const

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

void	SetAllScriptPositions (ScriptPos position)

int	GetTopScriptID () const

void	UpdateStateForSplit (int blob_position)

unsigned	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 tesseract::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 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 258 of file ratngs.h.

Constructor & Destructor Documentation

◆ WERD_CHOICE() [1/5]

tesseract::WERD_CHOICE::WERD_CHOICE ( const UNICHARSET * unicharset )

inline

Definition at line 263 of file ratngs.h.

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

◆ WERD_CHOICE() [2/5]

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

inline

Definition at line 266 of file ratngs.h.

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

◆ WERD_CHOICE() [3/5]

tesseract::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

Definition at line 269 of file ratngs.h.

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

◆ WERD_CHOICE() [4/5]

tesseract::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.

Definition at line 213 of file ratngs.cpp.

    : unicharset_(&unicharset) {
  std::vector<UNICHAR_ID> encoding;
  std::vector<char> lengths;
  std::string cleaned = unicharset.CleanupString(src_string);
  if (unicharset.encode_string(cleaned.c_str(), true, &encoding, &lengths, nullptr)) {
    lengths.push_back('\0');
    std::string src_lengths = &lengths[0];
    this->init(cleaned.c_str(), src_lengths.c_str(), 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]

tesseract::WERD_CHOICE::WERD_CHOICE ( const WERD_CHOICE & word )

inline

Definition at line 275 of file ratngs.h.

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

◆ ~WERD_CHOICE()

tesseract::WERD_CHOICE::~WERD_CHOICE ( )

default

WERD_CHOICE::~WERD_CHOICE

Member Function Documentation

◆ adjust_factor()

float tesseract::WERD_CHOICE::adjust_factor ( ) const

inline

Definition at line 290 of file ratngs.h.

                              {
    return adjust_factor_;
  }

◆ append_unichar_id()

void tesseract::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 447 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 tesseract::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 428 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 * tesseract::WERD_CHOICE::blob_choices	(	unsigned	index,
		MATRIX *	ratings
	)		const

Definition at line 274 of file ratngs.cpp.

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

◆ BlobPosition()

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

inline

Definition at line 306 of file ratngs.h.

                                               {
    if (index >= length_) {
      return SP_NORMAL;
    }
    return script_pos_[index];
  }

◆ certainty() [1/2]

float tesseract::WERD_CHOICE::certainty ( ) const

inline

Definition at line 315 of file ratngs.h.

                                 {
    return certainty_;
  }

◆ certainty() [2/2]

float tesseract::WERD_CHOICE::certainty ( unsigned index ) const

inline

Definition at line 318 of file ratngs.h.

                                               {
    return certainties_[index];
  }

◆ contains_unichar_id()

bool tesseract::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 309 of file ratngs.cpp.

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

◆ ContainsAnyNonSpaceDelimited()

bool tesseract::WERD_CHOICE::ContainsAnyNonSpaceDelimited ( ) const

inline

Definition at line 488 of file ratngs.h.

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

◆ dangerous_ambig_found()

bool tesseract::WERD_CHOICE::dangerous_ambig_found ( ) const

inline

Definition at line 348 of file ratngs.h.

                                     {
    return dangerous_ambig_found_;
  }

◆ debug_string()

std::string tesseract::WERD_CHOICE::debug_string ( ) const

inline

Definition at line 479 of file ratngs.h.

                                 {
    std::string word_str;
    for (unsigned i = 0; i < length_; ++i) {
      word_str += unicharset_->debug_str(unichar_ids_[i]);
      word_str += " ";
    }
    return word_str;
  }

◆ DisplaySegmentation()

void tesseract::WERD_CHOICE::DisplaySegmentation ( TWERD * word )

Definition at line 728 of file ratngs.cpp.

                                                 {
  // Number of different colors to draw with.
  const int kNumColors = 6;
  static ScrollView *segm_window = nullptr;
  // Check the state against the static prev_drawn_state.
  static std::vector<int> prev_drawn_state;
  bool already_done = prev_drawn_state.size() == length_;
  if (!already_done) {
    prev_drawn_state.clear();
    prev_drawn_state.resize(length_);
  }
  for (unsigned 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 == nullptr) {
    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 (unsigned c = 0; c < length_; ++c) {
    auto 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();
  segm_window->Wait();
}

◆ double_the_size()

void tesseract::WERD_CHOICE::double_the_size ( )

inline

Make more space in unichar_id_ and fragment_lengths_ arrays.

Definition at line 372 of file ratngs.h.

                                {
    if (reserved_ > 0) {
      reserved_ *= 2;
    } else {
      reserved_ = 1;
    }
    unichar_ids_.resize(reserved_);
    script_pos_.resize(reserved_);
    state_.resize(reserved_);
    certainties_.resize(reserved_);
  }

◆ empty()

bool tesseract::WERD_CHOICE::empty ( ) const

inline

Definition at line 284 of file ratngs.h.

                     {
    return length_ == 0;
  }

◆ GetNonSuperscriptSpan()

void tesseract::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 tesseract::WERD_CHOICE::GetTopScriptID ( ) const

Definition at line 631 of file ratngs.cpp.

                                      {
  unsigned max_script = unicharset_->get_script_table_size();
  std::vector<unsigned> sid(max_script);
  for (unsigned 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).
  unsigned max_sid = 0;
  for (unsigned 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();
  }
  return max_sid;
}

◆ has_rtl_unichar_id()

bool tesseract::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 411 of file ratngs.cpp.

                                           {
  for (unsigned 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 tesseract::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.

Definition at line 238 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 (unsigned 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;
}

◆ init() [2/2]

void tesseract::WERD_CHOICE::init ( unsigned 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 386 of file ratngs.h.

                                      {
    reserved_ = reserved;
    if (reserved > 0) {
      unichar_ids_.resize(reserved);
      script_pos_.resize(reserved);
      state_.resize(reserved);
      certainties_.resize(reserved);
    } else {
      unichar_ids_.clear();
      script_pos_.clear();
      state_.clear();
      certainties_.clear();
    }
    length_ = 0;
    adjust_factor_ = 1.0f;
    rating_ = 0.0;
    certainty_ = FLT_MAX;
    min_x_height_ = 0.0f;
    max_x_height_ = FLT_MAX;
    permuter_ = NO_PERM;
    unichars_in_script_order_ = false; // Tesseract is strict left-to-right.
    dangerous_ambig_found_ = false;
  }

◆ IsAllSpaces()

bool tesseract::WERD_CHOICE::IsAllSpaces ( ) const

inline

Definition at line 497 of file ratngs.h.

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

◆ length()

unsigned tesseract::WERD_CHOICE::length ( ) const

inline

Definition at line 287 of file ratngs.h.

                                 {
    return length_;
  }

◆ make_bad()

void tesseract::WERD_CHOICE::make_bad ( )

inline

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

Definition at line 419 of file ratngs.h.

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

◆ MatrixCoord()

MATRIX_COORD tesseract::WERD_CHOICE::MatrixCoord ( unsigned index ) const

Definition at line 286 of file ratngs.cpp.

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

◆ max_x_height()

float tesseract::WERD_CHOICE::max_x_height ( ) const

inline

Definition at line 324 of file ratngs.h.

                                    {
    return max_x_height_;
  }

◆ min_x_height()

float tesseract::WERD_CHOICE::min_x_height ( ) const

inline

Definition at line 321 of file ratngs.h.

                                    {
    return min_x_height_;
  }

◆ operator+=()

WERD_CHOICE & tesseract::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 462 of file ratngs.cpp.

                                                              {
  ASSERT_HOST(unicharset_ == second.unicharset_);
  while (reserved_ < length_ + second.length()) {
    this->double_the_size();
  }
  const std::vector<UNICHAR_ID> &other_unichar_ids = second.unichar_ids();
  for (unsigned 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 & tesseract::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 499 of file ratngs.cpp.

                                                             {
  while (reserved_ < source.length()) {
    this->double_the_size();
  }
 
  unicharset_ = source.unicharset_;
  const std::vector<UNICHAR_ID> &other_unichar_ids = source.unichar_ids();
  for (unsigned 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_t tesseract::WERD_CHOICE::permuter ( ) const

inline

Definition at line 331 of file ratngs.h.

                                  {
    return permuter_;
  }

◆ permuter_name() [1/2]

const char * tesseract::WERD_CHOICE::permuter_name ( ) const

Definition at line 267 of file ratngs.cpp.

                                             {
  return kPermuterTypeNames[permuter_];
}

◆ permuter_name() [2/2]

const char * tesseract::WERD_CHOICE::permuter_name ( uint8_t permuter )

static

Definition at line 189 of file ratngs.cpp.

                                                       {
  return kPermuterTypeNames[permuter];
}

◆ print() [1/2]

void tesseract::WERD_CHOICE::print ( ) const

inline

Definition at line 561 of file ratngs.h.

                     {
    this->print("");
  }

◆ print() [2/2]

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

WERD_CHOICE::print

Print WERD_CHOICE to stdout.

Definition at line 689 of file ratngs.cpp.

                                             {
  tprintf("%s : ", msg);
  for (unsigned 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 (unsigned i = 0; i < length_; ++i) {
    tprintf("\t%s", ScriptPosToString(script_pos_[i]));
  }
  tprintf("\nstr");
  for (unsigned i = 0; i < length_; ++i) {
    tprintf("\t%s", unicharset_->id_to_unichar(unichar_ids_[i]));
  }
  tprintf("\nstate:");
  for (unsigned i = 0; i < length_; ++i) {
    tprintf("\t%d ", state_[i]);
  }
  tprintf("\nC");
  for (unsigned i = 0; i < length_; ++i) {
    tprintf("\t%.3f", certainties_[i]);
  }
  tprintf("\n");
}

◆ print_state()

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

Definition at line 716 of file ratngs.cpp.

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

◆ punct_stripped()

void tesseract::WERD_CHOICE::punct_stripped	(	unsigned *	start,
		unsigned *	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 367 of file ratngs.cpp.

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

◆ rating()

float tesseract::WERD_CHOICE::rating ( ) const

inline

Definition at line 312 of file ratngs.h.

                              {
    return rating_;
  }

◆ remove_last_unichar_id()

void tesseract::WERD_CHOICE::remove_last_unichar_id ( )

inline

Definition at line 455 of file ratngs.h.

                                       {
    --length_;
  }

◆ remove_unichar_id()

void tesseract::WERD_CHOICE::remove_unichar_id ( unsigned index )

inline

Definition at line 458 of file ratngs.h.

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

◆ remove_unichar_ids()

void tesseract::WERD_CHOICE::remove_unichar_ids	(	unsigned	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 325 of file ratngs.cpp.

                                                            {
  ASSERT_HOST(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 (unsigned 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 tesseract::WERD_CHOICE::reverse_and_mirror_unichar_ids ( )

reverse_and_mirror_unichar_ids

Reverses and mirrors unichars in unichar_ids.

Definition at line 349 of file ratngs.cpp.

                                                 {
  for (unsigned 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 tesseract::WERD_CHOICE::ScriptPositionOf	(	bool	print_debug,
		const UNICHARSET &	unicharset,
		const TBOX &	blob_box,
		UNICHAR_ID	unichar_id
	)

static

Definition at line 599 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 tesseract::WERD_CHOICE::set_adjust_factor ( float factor )

inline

Definition at line 293 of file ratngs.h.

                                       {
    adjust_factor_ = factor;
  }

◆ set_blob_choice()

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

Definition at line 297 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 tesseract::WERD_CHOICE::set_certainty ( float new_val )

inline

Definition at line 357 of file ratngs.h.

                                           {
    certainty_ = new_val;
  }

◆ set_dangerous_ambig_found_()

void tesseract::WERD_CHOICE::set_dangerous_ambig_found_ ( bool value )

inline

Definition at line 351 of file ratngs.h.

                                              {
    dangerous_ambig_found_ = value;
  }

◆ set_length()

void tesseract::WERD_CHOICE::set_length ( unsigned len )

inline

Definition at line 366 of file ratngs.h.

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

◆ set_permuter()

void tesseract::WERD_CHOICE::set_permuter ( uint8_t perm )

inline

Definition at line 360 of file ratngs.h.

                                         {
    permuter_ = perm;
  }

◆ set_rating()

void tesseract::WERD_CHOICE::set_rating ( float new_val )

inline

Definition at line 354 of file ratngs.h.

                                        {
    rating_ = new_val;
  }

◆ set_unichar_id() [1/2]

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

inline

Definition at line 437 of file ratngs.h.

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

◆ set_unichar_id() [2/2]

void tesseract::WERD_CHOICE::set_unichar_id	(	UNICHAR_ID	unichar_id,
		unsigned	index
	)

inline

Definition at line 344 of file ratngs.h.

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

◆ set_unichars_in_script_order()

bool tesseract::WERD_CHOICE::set_unichars_in_script_order ( bool in_script_order )

inline

Definition at line 509 of file ratngs.h.

                                                          {
    return unichars_in_script_order_ = in_script_order;
  }

◆ set_x_heights()

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

inline

Definition at line 327 of file ratngs.h.

                                                                {
    min_x_height_ = min_height;
    max_x_height_ = max_height;
  }

◆ SetAllScriptPositions()

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

Definition at line 592 of file ratngs.cpp.

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

◆ SetScriptPositions()

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

Definition at line 528 of file ratngs.cpp.

                                                                            {
  // Initialize to normal.
  for (unsigned i = 0; i < length_; ++i) {
    script_pos_[i] = tesseract::SP_NORMAL;
  }
  if (word->blobs.empty() || word->NumBlobs() != TotalOfStates()) {
    return;
  }
 
  unsigned position_counts[4] = {0, 0, 0, 0};
 
  int chunk_index = 0;
  for (unsigned 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_.empty()) {
      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 (4 * position_counts[tesseract::SP_SUBSCRIPT] > 3 * length_ ||
      4 * position_counts[tesseract::SP_SUPERSCRIPT] > 3 * 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().c_str());
    }
    for (unsigned i = 0; i < length_; i++) {
      ScriptPos sp = script_pos_[i];
      if (sp == tesseract::SP_SUBSCRIPT || sp == tesseract::SP_SUPERSCRIPT) {
        ASSERT_HOST(position_counts[sp] > 0);
        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().c_str());
    int chunk_index = 0;
    for (unsigned 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_.empty() ? 1 : state_[blob_index];
    }
  }
}

◆ shallow_copy()

WERD_CHOICE tesseract::WERD_CHOICE::shallow_copy	(	unsigned	start,
		unsigned	end
	)		const

Definition at line 393 of file ratngs.cpp.

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

◆ state()

unsigned tesseract::WERD_CHOICE::state ( unsigned index ) const

inline

Definition at line 303 of file ratngs.h.

                                              {
    return state_[index];
  }

◆ string_and_lengths()

void tesseract::WERD_CHOICE::string_and_lengths	(	std::string *	word_str,
		std::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 427 of file ratngs.cpp.

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

◆ TotalOfStates()

unsigned tesseract::WERD_CHOICE::TotalOfStates ( ) const

Definition at line 676 of file ratngs.cpp.

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

◆ unichar_id()

UNICHAR_ID tesseract::WERD_CHOICE::unichar_id ( unsigned index ) const

inline

Definition at line 299 of file ratngs.h.

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

◆ unichar_ids()

const std::vector< UNICHAR_ID > & tesseract::WERD_CHOICE::unichar_ids ( ) const

inline

Definition at line 296 of file ratngs.h.

                                                          {
    return unichar_ids_;
  }

◆ unichar_lengths()

const std::string & tesseract::WERD_CHOICE::unichar_lengths ( ) const

inline

Definition at line 533 of file ratngs.h.

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

◆ unichar_string() [1/2]

std::string & tesseract::WERD_CHOICE::unichar_string ( )

inline

Definition at line 519 of file ratngs.h.

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

◆ unichar_string() [2/2]

const std::string & tesseract::WERD_CHOICE::unichar_string ( ) const

inline

Definition at line 526 of file ratngs.h.

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

◆ unichars_in_script_order()

bool tesseract::WERD_CHOICE::unichars_in_script_order ( ) const

inline

Definition at line 513 of file ratngs.h.

                                        {
    return unichars_in_script_order_;
  }

◆ unicharset()

const UNICHARSET * tesseract::WERD_CHOICE::unicharset ( ) const

inline

Definition at line 281 of file ratngs.h.

                                       {
    return unicharset_;
  }

◆ UpdateStateForSplit()

void tesseract::WERD_CHOICE::UpdateStateForSplit ( int blob_position )

Definition at line 664 of file ratngs.cpp.

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

Member Data Documentation

◆ kBadRating

const float tesseract::WERD_CHOICE::kBadRating = 100000.0

static

Definition at line 260 of file ratngs.h.

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

/media/home/debian/src/github/tesseract-ocr/tesseract/src/ccstruct/ratngs.h
/media/home/debian/src/github/tesseract-ocr/tesseract/src/ccstruct/ratngs.cpp

Public Member Functions

Static Public Member Functions

Static Public Attributes

Detailed Description

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()

◆ empty()

◆ 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()

◆ shallow_copy()

◆ state()

◆ string_and_lengths()

◆ TotalOfStates()

◆ unichar_id()

◆ unichar_ids()

◆ unichar_lengths()

◆ unichar_string() [1/2]

◆ unichar_string() [2/2]

◆ unichars_in_script_order()

◆ unicharset()

◆ UpdateStateForSplit()

Member Data Documentation

◆ kBadRating