tesseract::BoxChar Class Reference

#include <boxchar.h>

Public Member Functions
	BoxChar (const char *utf8_str, int len)
	~BoxChar ()
const std::string &	ch () const
const Box *	box () const
const int &	page () const
void	set_rtl_index (int index)
const int &	rtl_index () const
void	AddBox (int x, int y, int width, int height)
void	set_page (int page)
std::string *	mutable_ch ()
Box *	mutable_box ()
bool	operator< (const BoxChar &other) const
void	GetDirection (int *num_rtl, int *num_ltr) const
void	ReverseUnicodesInBox ()

Static Public Member Functions
static void	TranslateBoxes (int xshift, int yshift, std::vector< BoxChar *> *boxes)
static void	PrepareToWrite (std::vector< BoxChar *> *boxes)
static void	InsertNewlines (bool rtl_rules, bool vertical_rules, std::vector< BoxChar *> *boxes)
static void	InsertSpaces (bool rtl_rules, bool vertical_rules, std::vector< BoxChar *> *boxes)
static void	ReorderRTLText (std::vector< BoxChar *> *boxes)
static bool	ContainsMostlyRTL (const std::vector< BoxChar *> &boxes)
static bool	MostlyVertical (const std::vector< BoxChar *> &boxes)
static int	TotalByteLength (const std::vector< BoxChar *> &boxes)
static void	RotateBoxes (float rotation, int xcenter, int ycenter, int start_box, int end_box, std::vector< BoxChar *> *boxes)
static void	WriteTesseractBoxFile (const std::string &name, int height, const std::vector< BoxChar *> &boxes)
static std::string	GetTesseractBoxStr (int height, const std::vector< BoxChar *> &boxes)

Detailed Description

Definition at line 36 of file boxchar.h.

Constructor & Destructor Documentation

BoxChar()

tesseract::BoxChar::BoxChar	(	const char *	utf8_str,
		int	len
	)

Definition at line 41 of file boxchar.cpp.

    : ch_(utf8_str, len), box_(nullptr), page_(0), rtl_index_(-1) {}

~BoxChar()

tesseract::BoxChar::~BoxChar

(

)

Definition at line 44 of file boxchar.cpp.

{ boxDestroy(&box_); }

Member Function Documentation

AddBox()

void tesseract::BoxChar::AddBox	(	int	x,
		int	y,
		int	width,
		int	height
	)

Definition at line 46 of file boxchar.cpp.

                                                        {
  box_ = boxCreate(x, y, width, height);
}

box()

const Box* tesseract::BoxChar::box ( ) const

inline

Definition at line 44 of file boxchar.h.

{ return box_; }

ch()

const std::string& tesseract::BoxChar::ch ( ) const

inline

Definition at line 43 of file boxchar.h.

{ return ch_; }

ContainsMostlyRTL()

bool tesseract::BoxChar::ContainsMostlyRTL ( const std::vector< BoxChar *> &  boxes )

static

Definition at line 260 of file boxchar.cpp.

                                                                {
  int num_rtl = 0, num_ltr = 0;
  for (int i = 0; i < boxes.size(); ++i) {
    boxes[i]->GetDirection(&num_rtl, &num_ltr);
  }
  return num_rtl > num_ltr;
}

GetDirection()

void tesseract::BoxChar::GetDirection	(	int *	num_rtl,
		int *	num_ltr
	)		const

Definition at line 52 of file boxchar.cpp.

                                                           {
  // Convert the unichar to UTF32 representation
  std::vector<char32> uni_vector = UNICHAR::UTF8ToUTF32(ch_.c_str());
  if (uni_vector.empty()) {
    tprintf("Illegal utf8 in boxchar string:%s = ", ch_.c_str());
    for (int c = 0; c < ch_.size(); ++c) {
      tprintf(" 0x%x", ch_[c]);
    }
    tprintf("\n");
    return;
  }
  for (char32 ch : uni_vector) {
    UCharDirection dir = u_charDirection(ch);
    if (dir == U_RIGHT_TO_LEFT || dir == U_RIGHT_TO_LEFT_ARABIC ||
        dir == U_ARABIC_NUMBER || dir == U_RIGHT_TO_LEFT_ISOLATE) {
      ++*num_rtl;
    } else if (dir != U_DIR_NON_SPACING_MARK && dir != U_BOUNDARY_NEUTRAL) {
      ++*num_ltr;
    }
  }
}

GetTesseractBoxStr()

std::string tesseract::BoxChar::GetTesseractBoxStr ( int  height, const std::vector< BoxChar *> &  boxes  )

static

Definition at line 327 of file boxchar.cpp.

                                                                          {
  std::string output;
  char buffer[kMaxLineLength];
  for (size_t i = 0; i < boxes.size(); ++i) {
    const Box* box = boxes[i]->box_;
    if (box == nullptr) {
      tprintf("Error: Call PrepareToWrite before WriteTesseractBoxFile!!\n");
      return "";
    }
    int nbytes =
        snprintf(buffer, kMaxLineLength, "%s %d %d %d %d %d\n",
                 boxes[i]->ch_.c_str(), box->x, height - box->y - box->h,
                 box->x + box->w, height - box->y, boxes[i]->page_);
    output.append(buffer, nbytes);
  }
  return output;
}

InsertNewlines()

void tesseract::BoxChar::InsertNewlines ( bool  rtl_rules, bool  vertical_rules, std::vector< BoxChar *> *  boxes  )

static

Definition at line 112 of file boxchar.cpp.

                                                         {
  int prev_i = -1;
  int max_shift = 0;
  for (size_t i = 0; i < boxes->size(); ++i) {
    Box* box = (*boxes)[i]->box_;
    if (box == nullptr) {
      if (prev_i < 0 || prev_i + 1 < i || i + 1 == boxes->size()) {
        // Erase null boxes at the start of a line and after another null box.
        do {
          delete (*boxes)[i];
          boxes->erase(boxes->begin() + i);
          if (i == 0) break;
        } while (i-- == boxes->size() && (*boxes)[i]->box_ == nullptr);
      }
      continue;
    }
    if (prev_i >= 0) {
      Box* prev_box = (*boxes)[prev_i]->box_;
      int shift = box->x - prev_box->x;
      if (vertical_rules) {
        shift = box->y - prev_box->y;
      } else if (rtl_rules) {
        shift = -shift;
      }
      if (-shift > max_shift) {
        // This is a newline. Since nothing cares about the size of the box,
        // except the out-of-bounds checker, minimize the chance of creating
        // a box outside the image by making the width and height 1.
        int width = 1;
        int height = 1;
        int x = prev_box->x + prev_box->w;
        int y = prev_box->y;
        if (vertical_rules) {
          x = prev_box->x;
          y = prev_box->y + prev_box->h;
        } else if (rtl_rules) {
          x = prev_box->x - width;
          if (x < 0) {
            tprintf("prev x = %d, width=%d\n", prev_box->x, width);
            x = 0;
          }
        }
        if (prev_i + 1 == i) {
          // New character needed.
          BoxChar* new_box = new BoxChar("\t", 1);
          new_box->AddBox(x, y, width, height);
          new_box->page_ = (*boxes)[i]->page_;
          boxes->insert(boxes->begin() + i, new_box);
          ++i;
        } else {
          (*boxes)[i - 1]->AddBox(x, y, width, height);
          (*boxes)[i - 1]->ch_ = "\t";
        }
        max_shift = 0;
      } else if (shift > max_shift) {
        max_shift = shift;
      }
    }
    prev_i = i;
  }
}

InsertSpaces()

void tesseract::BoxChar::InsertSpaces ( bool  rtl_rules, bool  vertical_rules, std::vector< BoxChar *> *  boxes  )

static

Definition at line 177 of file boxchar.cpp.

                                                       {
  // After InsertNewlines, any remaining null boxes are not newlines, and are
  // singletons, so add a box to each remaining null box.
  for (size_t i = 1; i + 1 < boxes->size(); ++i) {
    Box* box = (*boxes)[i]->box_;
    if (box == nullptr) {
      Box* prev = (*boxes)[i - 1]->box_;
      Box* next = (*boxes)[i + 1]->box_;
      ASSERT_HOST(prev != nullptr && next != nullptr);
      int top = std::min(prev->y, next->y);
      int bottom = std::max(prev->y + prev->h, next->y + next->h);
      int left = prev->x + prev->w;
      int right = next->x;
      if (vertical_rules) {
        top = prev->y + prev->h;
        bottom = next->y;
        left = std::min(prev->x, next->x);
        right = std::max(prev->x + prev->w, next->x + next->w);
      } else if (rtl_rules) {
        // With RTL we have to account for BiDi.
        // Right becomes the min left of all prior boxes back to the first
        // space or newline.
        right = prev->x;
        left = next->x + next->w;
        for (int j = i - 2;
             j >= 0 && (*boxes)[j]->ch_ != " " && (*boxes)[j]->ch_ != "\t";
             --j) {
          prev = (*boxes)[j]->box_;
          ASSERT_HOST(prev != nullptr);
          if (prev->x < right) {
            right = prev->x;
          }
        }
        // Left becomes the max right of all next boxes forward to the first
        // space or newline.
        for (size_t j = i + 2;
             j < boxes->size() && (*boxes)[j]->box_ != nullptr &&
             (*boxes)[j]->ch_ != "\t";
             ++j) {
          next = (*boxes)[j]->box_;
          if (next->x + next->w > left) {
            left = next->x + next->w;
          }
        }
      }
      // Italic and stylized characters can produce negative spaces, which
      // Leptonica doesn't like, so clip to a positive size.
      if (right <= left) right = left + 1;
      if (bottom <= top) bottom = top + 1;
      (*boxes)[i]->AddBox(left, top, right - left, bottom - top);
      (*boxes)[i]->ch_ = " ";
    }
  }
}

MostlyVertical()

bool tesseract::BoxChar::MostlyVertical ( const std::vector< BoxChar *> &  boxes )

static

Definition at line 270 of file boxchar.cpp.

                                                             {
  int64_t total_dx = 0, total_dy = 0;
  for (size_t i = 1; i < boxes.size(); ++i) {
    if (boxes[i - 1]->box_ != nullptr && boxes[i]->box_ != nullptr &&
        boxes[i - 1]->page_ == boxes[i]->page_) {
      int dx = boxes[i]->box_->x - boxes[i - 1]->box_->x;
      int dy = boxes[i]->box_->y - boxes[i - 1]->box_->y;
      if (abs(dx) > abs(dy) * kMinNewlineRatio ||
          abs(dy) > abs(dx) * kMinNewlineRatio) {
        total_dx += dx * dx;
        total_dy += dy * dy;
      }
    }
  }
  return total_dy > total_dx;
}

mutable_box()

Box* tesseract::BoxChar::mutable_box ( )

inline

Definition at line 55 of file boxchar.h.

{ return box_; }

mutable_ch()

std::string* tesseract::BoxChar::mutable_ch ( )

inline

Definition at line 54 of file boxchar.h.

{ return &ch_; }

operator<()

bool tesseract::BoxChar::operator< ( const BoxChar &  other ) const

inline

Definition at line 59 of file boxchar.h.

                                             {
    if (box_ == nullptr) return true;
    if (other.box_ == nullptr) return false;
    return box_->x < other.box_->x;
  }

page()

const int& tesseract::BoxChar::page ( ) const

inline

Definition at line 45 of file boxchar.h.

{ return page_; }

PrepareToWrite()

void tesseract::BoxChar::PrepareToWrite ( std::vector< BoxChar *> *  boxes )

static

Definition at line 97 of file boxchar.cpp.

                                                       {
  bool rtl_rules = ContainsMostlyRTL(*boxes);
  bool vertical_rules = MostlyVertical(*boxes);
  InsertNewlines(rtl_rules, vertical_rules, boxes);
  InsertSpaces(rtl_rules, vertical_rules, boxes);
  for (unsigned int i = 0; i < boxes->size(); ++i) {
    if ((*boxes)[i]->box_ == nullptr) tprintf("Null box at index %u\n", i);
  }
  if (rtl_rules) {
    ReorderRTLText(boxes);
  }
}

ReorderRTLText()

void tesseract::BoxChar::ReorderRTLText ( std::vector< BoxChar *> *  boxes )

static

Definition at line 235 of file boxchar.cpp.

                                                       {
  // Ideally we need the inverse of the algorithm used by ResultIterator.
  // For now, let's try a sort that reverses original positions for RTL
  // characters, otherwise by x-position. This should be much closer to
  // correct than just sorting by x-position.
  int num_boxes = boxes->size();
  for (int i = 0; i < num_boxes; ++i) {
    int num_rtl = 0, num_ltr = 0;
    (*boxes)[i]->GetDirection(&num_rtl, &num_ltr);
    if (num_rtl > num_ltr) {
      (*boxes)[i]->set_rtl_index(i);
      (*boxes)[i]->ReverseUnicodesInBox();
    }
  }
  BoxCharPtrSort sorter;
  size_t end = 0;
  for (size_t start = 0; start < boxes->size(); start = end + 1) {
    end = start + 1;
    while (end < boxes->size() && (*boxes)[end]->ch_ != "\t") ++end;
    std::sort(boxes->begin() + start, boxes->begin() + end, sorter);
  }
}

ReverseUnicodesInBox()

void tesseract::BoxChar::ReverseUnicodesInBox

(

)

Definition at line 76 of file boxchar.cpp.

                                   {
  std::vector<char32> unicodes = UNICHAR::UTF8ToUTF32(ch_.c_str());
  std::reverse(unicodes.begin(), unicodes.end());
  ch_ = UNICHAR::UTF32ToUTF8(unicodes);
}

RotateBoxes()

void tesseract::BoxChar::RotateBoxes ( float  rotation, int  xcenter, int  ycenter, int  start_box, int  end_box, std::vector< BoxChar *> *  boxes  )

static

Definition at line 299 of file boxchar.cpp.

                                                      {
  Boxa* orig = boxaCreate(0);
  for (int i = start_box; i < end_box; ++i) {
    BOX* box = (*boxes)[i]->box_;
    if (box) boxaAddBox(orig, box, L_CLONE);
  }
  Boxa* rotated = boxaRotate(orig, xcenter, ycenter, rotation);
  boxaDestroy(&orig);
  for (int i = start_box, box_ind = 0; i < end_box; ++i) {
    if ((*boxes)[i]->box_) {
      boxDestroy(&((*boxes)[i]->box_));
      (*boxes)[i]->box_ = boxaGetBox(rotated, box_ind++, L_CLONE);
    }
  }
  boxaDestroy(&rotated);
}

rtl_index()

const int& tesseract::BoxChar::rtl_index ( ) const

inline

Definition at line 47 of file boxchar.h.

{ return rtl_index_; }

set_page()

void tesseract::BoxChar::set_page ( int  page )

inline

Definition at line 52 of file boxchar.h.

{ page_ = page; }

set_rtl_index()

void tesseract::BoxChar::set_rtl_index ( int  index )

inline

Definition at line 46 of file boxchar.h.

{ rtl_index_ = index; }

TotalByteLength()

int tesseract::BoxChar::TotalByteLength ( const std::vector< BoxChar *> &  boxes )

static

Definition at line 289 of file boxchar.cpp.

                                                             {
  int total_length = 0;
  for (size_t i = 0; i < boxes.size(); ++i)
    total_length += boxes[i]->ch_.size();
  return total_length;
}

TranslateBoxes()

void tesseract::BoxChar::TranslateBoxes ( int  xshift, int  yshift, std::vector< BoxChar *> *  boxes  )

static

Definition at line 83 of file boxchar.cpp.

                                                         {
  for (size_t i = 0; i < boxes->size(); ++i) {
    BOX* box = (*boxes)[i]->box_;
    if (box != nullptr) {
      box->x += xshift;
      box->y += yshift;
    }
  }
}

WriteTesseractBoxFile()

void tesseract::BoxChar::WriteTesseractBoxFile ( const std::string &  name, int  height, const std::vector< BoxChar *> &  boxes  )

static

Definition at line 320 of file boxchar.cpp.

                                                                      {
  std::string output = GetTesseractBoxStr(height, boxes);
  File::WriteStringToFileOrDie(output, filename);
}

---

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

• /usr/src/tesseract-ocr.master/src/training/boxchar.h
• /usr/src/tesseract-ocr.master/src/training/boxchar.cpp