tesseract::LigatureTable Class Reference

#include <ligature_table.h>

Public Member Functions
std::string	AddLigatures (const std::string &str, const PangoFontInfo *font) const
std::string	RemoveLigatures (const std::string &str) const
std::string	RemoveCustomLigatures (const std::string &str) const
const LigHash &	norm_to_lig_table () const
const LigHash &	lig_to_norm_table () const

Static Public Member Functions
static LigatureTable *	Get ()

Protected Member Functions
	LigatureTable ()
void	Init ()

Protected Attributes
LigHash	norm_to_lig_table_
LigHash	lig_to_norm_table_
int	min_lig_length_
int	max_lig_length_
int	min_norm_length_
int	max_norm_length_

Static Protected Attributes
static std::unique_ptr< LigatureTable >	instance_

Detailed Description

Definition at line 55 of file ligature_table.h.

Constructor & Destructor Documentation

LigatureTable()

tesseract::LigatureTable::LigatureTable ( )

protected

Definition at line 78 of file ligature_table.cpp.

                                                                {

Member Function Documentation

AddLigatures()

std::string tesseract::LigatureTable::AddLigatures	(	const std::string &	str,
		const PangoFontInfo *	font
	)		const

Definition at line 174 of file ligature_table.cpp.

                   {
      result += str[i];
      step = 1;
    }
  }
  result += str.substr(i, len - i);
  return result;
}
 
}  // namespace tesseract

Get()

LigatureTable * tesseract::LigatureTable::Get ( )

static

Definition at line 70 of file ligature_table.cpp.

                                                                {

Init()

void tesseract::LigatureTable::Init ( )

protected

Definition at line 81 of file ligature_table.cpp.

                                                                {
        norm_to_lig_table_[normed8] = lig8;
        lig_to_norm_table_[lig8] = normed8;
        if (min_lig_length_ == 0 || lig_length < min_lig_length_)
          min_lig_length_ = lig_length;
        if (lig_length > max_lig_length_)
          max_lig_length_ = lig_length;
        if (min_norm_length_ == 0 || norm_length < min_norm_length_)
          min_norm_length_ = norm_length;
        if (norm_length > max_norm_length_)
          max_norm_length_ = norm_length;
      }
    }
    // Add custom extra ligatures.
    for (int i = 0; UNICHARSET::kCustomLigatures[i][0] != nullptr; ++i) {
      norm_to_lig_table_[UNICHARSET::kCustomLigatures[i][0]] =
          UNICHARSET::kCustomLigatures[i][1];
      int norm_length = strlen(UNICHARSET::kCustomLigatures[i][0]);
      if (min_norm_length_ == 0 || norm_length < min_norm_length_)
        min_norm_length_ = norm_length;
      if (norm_length > max_norm_length_)
        max_norm_length_ = norm_length;
 
      lig_to_norm_table_[UNICHARSET::kCustomLigatures[i][1]] =
          UNICHARSET::kCustomLigatures[i][0];
    }
  }
}
 
std::string LigatureTable::RemoveLigatures(const std::string& str) const {
  std::string result;
  UNICHAR::const_iterator it_begin = UNICHAR::begin(str.c_str(), str.length());
  UNICHAR::const_iterator it_end = UNICHAR::end(str.c_str(), str.length());
  char tmp[5];
  int len;
  for (UNICHAR::const_iterator it = it_begin; it != it_end; ++it) {
    len = it.get_utf8(tmp);
    tmp[len] = '\0';
    LigHash::const_iterator lig_it = lig_to_norm_table_.find(tmp);
    if (lig_it != lig_to_norm_table_.end()) {
      result += lig_it->second;
    } else {
      result += tmp;
    }
  }
  return result;

lig_to_norm_table()

const LigHash& tesseract::LigatureTable::lig_to_norm_table ( ) const

inline

Definition at line 72 of file ligature_table.h.

        :
  LigatureTable(const LigatureTable&);

norm_to_lig_table()

const LigHash& tesseract::LigatureTable::norm_to_lig_table ( ) const

inline

Definition at line 69 of file ligature_table.h.

        :
  LigatureTable(const LigatureTable&);

RemoveCustomLigatures()

std::string tesseract::LigatureTable::RemoveCustomLigatures

(

const std::string &

str

)

const

Definition at line 148 of file ligature_table.cpp.

           {
      result += tmp;
    }
  }
  return result;
}
 
std::string LigatureTable::AddLigatures(const std::string& str,
                                        const PangoFontInfo* font) const {
  std::string result;
  int len = str.size();
  int step = 0;
  int i = 0;
  for (i = 0; i < len - min_norm_length_ + 1; i += step) {
    step = 0;
    for (int liglen = max_norm_length_; liglen >= min_norm_length_; --liglen) {
      if (i + liglen <= len) {
        std::string lig_cand = str.substr(i, liglen);
        LigHash::const_iterator it = norm_to_lig_table_.find(lig_cand);
        if (it != norm_to_lig_table_.end()) {
          tlog(3, "Considering %s -> %s\n", lig_cand.c_str(),
               it->second.c_str());
          if (font) {
            // Test for renderability.

RemoveLigatures()

std::string tesseract::LigatureTable::RemoveLigatures

(

const std::string &

str

)

const

Definition at line 129 of file ligature_table.cpp.

                                                                         {
  std::string result;
  UNICHAR::const_iterator it_begin = UNICHAR::begin(str.c_str(), str.length());
  UNICHAR::const_iterator it_end = UNICHAR::end(str.c_str(), str.length());
  char tmp[5];
  int len;
  int norm_ind;
  for (UNICHAR::const_iterator it = it_begin; it != it_end; ++it) {
    len = it.get_utf8(tmp);
    tmp[len] = '\0';
    norm_ind = -1;
    for (int i = 0;
         UNICHARSET::kCustomLigatures[i][0] != nullptr && norm_ind < 0; ++i) {
      if (!strcmp(tmp, UNICHARSET::kCustomLigatures[i][1])) {
        norm_ind = i;
      }
    }

Member Data Documentation

instance_

std::unique_ptr< LigatureTable > tesseract::LigatureTable::instance_

staticprotected

Definition at line 82 of file ligature_table.h.

lig_to_norm_table_

LigHash tesseract::LigatureTable::lig_to_norm_table_

protected

Definition at line 84 of file ligature_table.h.

max_lig_length_

int tesseract::LigatureTable::max_lig_length_

protected

Definition at line 86 of file ligature_table.h.

max_norm_length_

int tesseract::LigatureTable::max_norm_length_

protected

Definition at line 88 of file ligature_table.h.

min_lig_length_

int tesseract::LigatureTable::min_lig_length_

protected

Definition at line 85 of file ligature_table.h.

min_norm_length_

int tesseract::LigatureTable::min_norm_length_

protected

Definition at line 87 of file ligature_table.h.

norm_to_lig_table_

LigHash tesseract::LigatureTable::norm_to_lig_table_

protected

Definition at line 83 of file ligature_table.h.

---

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

• /tesseract/src/training/ligature_table.h
• /tesseract/src/training/ligature_table.cpp