tesseract::BLOCK Class Reference

#include <ocrblock.h>

Inheritance diagram for tesseract::BLOCK:

Public Member Functions
	BLOCK ()
	BLOCK (const char *name, bool prop, int16_t kern, int16_t space, TDimension xmin, TDimension ymin, TDimension xmax, TDimension ymax)
	~BLOCK ()=default
void	set_stats (bool prop, int16_t kern, int16_t space, int16_t ch_pitch)
void	set_xheight (int32_t height)
	set char size More...
void	set_font_class (int16_t font)
	set font class More...
bool	prop () const
	return proportional More...
bool	right_to_left () const
void	set_right_to_left (bool value)
int32_t	fixed_pitch () const
	return pitch More...
int16_t	kern () const
	return kerning More...
int16_t	font () const
	return font class More...
int16_t	space () const
	return spacing More...
const char *	name () const
	return filename More...
int32_t	x_height () const
	return xheight More...
float	cell_over_xheight () const
void	set_cell_over_xheight (float ratio)
ROW_LIST *	row_list ()
	get rows More...
void	compute_row_margins ()
PARA_LIST *	para_list ()
C_BLOB_LIST *	blob_list ()
	get blobs More...
C_BLOB_LIST *	reject_blobs ()
FCOORD	re_rotation () const
void	set_re_rotation (const FCOORD &rotation)
FCOORD	classify_rotation () const
void	set_classify_rotation (const FCOORD &rotation)
FCOORD	skew () const
void	set_skew (const FCOORD &skew)
const ICOORD &	median_size () const
void	set_median_size (int x, int y)
Image	render_mask (TBOX *mask_box)
TBOX	restricted_bounding_box (bool upper_dots, bool lower_dots) const
void	reflect_polygon_in_y_axis ()
void	rotate (const FCOORD &rotation)
void	sort_rows ()
	decreasing y order More...
void	compress ()
	shrink white space More...
void	check_pitch ()
	check proportional More...
void	compress (const ICOORD vec)
	shrink white space and move by vector More...
void	print (FILE *fp, bool dump)
	dump whole table More...
BLOCK &	operator= (const BLOCK &source)
Public Member Functions inherited from tesseract::ELIST_LINK
	ELIST_LINK ()
	ELIST_LINK (const ELIST_LINK &)
void	operator= (const ELIST_LINK &)

Public Attributes
PDBLK	pdblk
	Page Description Block. More...

Friends
class	BLOCK_RECT_IT

Detailed Description

Definition at line 32 of file ocrblock.h.

Constructor & Destructor Documentation

BLOCK() [1/2]

tesseract::BLOCK::BLOCK ( )

inline

Definition at line 37 of file ocrblock.h.

: re_rotation_(1.0f, 0.0f), classify_rotation_(1.0f, 0.0f), skew_(1.0f, 0.0f) {}

BLOCK() [2/2]

tesseract::BLOCK::BLOCK	(	const char *	name,
		bool	prop,
		int16_t	kern,
		int16_t	space,
		TDimension	xmin,
		TDimension	ymin,
		TDimension	xmax,
		TDimension	ymax
	)

BLOCK::BLOCK

Constructor for a simple rectangular block.

Parameters

name	filename
prop	proportional
kern	kerning
space	spacing
xmin	bottom left
xmax	top right

Definition at line 34 of file ocrblock.cpp.

    : pdblk(xmin, ymin, xmax, ymax)
    , filename(name)
    , re_rotation_(1.0f, 0.0f)
    , classify_rotation_(1.0f, 0.0f)
    , skew_(1.0f, 0.0f) {
  ICOORDELT_IT left_it = &pdblk.leftside;
  ICOORDELT_IT right_it = &pdblk.rightside;
 
  proportional = prop;
  kerning = kern;
  spacing = space;
  font_class = -1; // not assigned
  cell_over_xheight_ = 2.0f;
  pdblk.hand_poly = nullptr;
  left_it.set_to_list(&pdblk.leftside);
  right_it.set_to_list(&pdblk.rightside);
  // make default box
  left_it.add_to_end(new ICOORDELT(xmin, ymin));
  left_it.add_to_end(new ICOORDELT(xmin, ymax));
  right_it.add_to_end(new ICOORDELT(xmax, ymin));
  right_it.add_to_end(new ICOORDELT(xmax, ymax));
}

~BLOCK()

tesseract::BLOCK::~BLOCK ( )

default

Member Function Documentation

blob_list()

C_BLOB_LIST * tesseract::BLOCK::blob_list ( )

inline

get blobs

Definition at line 123 of file ocrblock.h.

                           {
    return &c_blobs;
  }

cell_over_xheight()

float tesseract::BLOCK::cell_over_xheight ( ) const

inline

Definition at line 104 of file ocrblock.h.

                                  {
    return cell_over_xheight_;
  }

check_pitch()

void tesseract::BLOCK::check_pitch

(

)

check proportional

BLOCK::check_pitch

Check whether the block is fixed or prop, set the flag, and set the pitch if it is fixed.

Definition at line 164 of file ocrblock.cpp.

                        { // check prop
  //      tprintf("Missing FFT fixed pitch stuff!\n");
  pitch = -1;
}

classify_rotation()

FCOORD tesseract::BLOCK::classify_rotation ( ) const

inline

Definition at line 135 of file ocrblock.h.

                                   {
    return classify_rotation_; // Apply this before classifying.
  }

compress() [1/2]

void tesseract::BLOCK::compress

(

)

shrink white space

BLOCK::compress

Delete space between the rows. (And maybe one day, compress the rows) Fill space of block from top down, left aligning rows.

Definition at line 128 of file ocrblock.cpp.

                     { // squash it up
#define ROW_SPACING 5
 
  ROW_IT row_it(&rows);
  ROW *row;
  ICOORD row_spacing(0, ROW_SPACING);
 
  ICOORDELT_IT icoordelt_it;
 
  sort_rows();
 
  pdblk.box = TBOX(pdblk.box.topleft(), pdblk.box.topleft());
  pdblk.box.move_bottom_edge(ROW_SPACING);
  for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
    row = row_it.data();
    row->move(pdblk.box.botleft() - row_spacing - row->bounding_box().topleft());
    pdblk.box += row->bounding_box();
  }
 
  pdblk.leftside.clear();
  icoordelt_it.set_to_list(&pdblk.leftside);
  icoordelt_it.add_to_end(new ICOORDELT(pdblk.box.left(), pdblk.box.bottom()));
  icoordelt_it.add_to_end(new ICOORDELT(pdblk.box.left(), pdblk.box.top()));
  pdblk.rightside.clear();
  icoordelt_it.set_to_list(&pdblk.rightside);
  icoordelt_it.add_to_end(new ICOORDELT(pdblk.box.right(), pdblk.box.bottom()));
  icoordelt_it.add_to_end(new ICOORDELT(pdblk.box.right(), pdblk.box.top()));
}

compress() [2/2]

void tesseract::BLOCK::compress

(

const ICOORD

vec

)

shrink white space and move by vector

BLOCK::compress

Compress and move in a single operation.

Definition at line 175 of file ocrblock.cpp.

  {
  pdblk.box.move(vec);
  compress();
}

compute_row_margins()

void tesseract::BLOCK::compute_row_margins

(

)

Definition at line 330 of file ocrblock.cpp.

                                {
  if (row_list()->empty() || row_list()->singleton()) {
    return;
  }
 
  // If Layout analysis was not called, default to this.
  POLY_BLOCK rect_block(pdblk.bounding_box(), PT_FLOWING_TEXT);
  POLY_BLOCK *pblock = &rect_block;
  if (pdblk.poly_block() != nullptr) {
    pblock = pdblk.poly_block();
  }
 
  // Step One: Determine if there is a drop-cap.
  //           TODO(eger): Fix up drop cap code for RTL languages.
  ROW_IT r_it(row_list());
  ROW *first_row = r_it.data();
  ROW *second_row = r_it.data_relative(1);
 
  // initialize the bottom of a fictitious drop cap far above the first line.
  int drop_cap_bottom = first_row->bounding_box().top() + first_row->bounding_box().height();
  int drop_cap_right = first_row->bounding_box().left();
  int mid_second_line = second_row->bounding_box().top() - second_row->bounding_box().height() / 2;
  WERD_IT werd_it(r_it.data()->word_list()); // words of line one
  if (!werd_it.empty()) {
    C_BLOB_IT cblob_it(werd_it.data()->cblob_list());
    for (cblob_it.mark_cycle_pt(); !cblob_it.cycled_list(); cblob_it.forward()) {
      TBOX bbox = cblob_it.data()->bounding_box();
      if (bbox.bottom() <= mid_second_line) {
        // we found a real drop cap
        first_row->set_has_drop_cap(true);
        if (drop_cap_bottom > bbox.bottom()) {
          drop_cap_bottom = bbox.bottom();
        }
        if (drop_cap_right < bbox.right()) {
          drop_cap_right = bbox.right();
        }
      }
    }
  }
 
  // Step Two: Calculate the margin from the text of each row to the block
  //           (or drop-cap) boundaries.
  PB_LINE_IT lines(pblock);
  r_it.set_to_list(row_list());
  for (r_it.mark_cycle_pt(); !r_it.cycled_list(); r_it.forward()) {
    ROW *row = r_it.data();
    TBOX row_box = row->bounding_box();
    int left_y = row->base_line(row_box.left()) + row->x_height();
    int left_margin;
    const std::unique_ptr</*non-const*/ ICOORDELT_LIST> segments_left(lines.get_line(left_y));
    LeftMargin(segments_left.get(), row_box.left(), &left_margin);
 
    if (row_box.top() >= drop_cap_bottom) {
      int drop_cap_distance = row_box.left() - row->space() - drop_cap_right;
      if (drop_cap_distance < 0) {
        drop_cap_distance = 0;
      }
      if (drop_cap_distance < left_margin) {
        left_margin = drop_cap_distance;
      }
    }
 
    int right_y = row->base_line(row_box.right()) + row->x_height();
    int right_margin;
    const std::unique_ptr</*non-const*/ ICOORDELT_LIST> segments_right(lines.get_line(right_y));
    RightMargin(segments_right.get(), row_box.right(), &right_margin);
    row->set_lmargin(left_margin);
    row->set_rmargin(right_margin);
  }
}

fixed_pitch()

int32_t tesseract::BLOCK::fixed_pitch ( ) const

inline

return pitch

Definition at line 81 of file ocrblock.h.

                              {
    return pitch;
  }

font()

int16_t tesseract::BLOCK::font ( ) const

inline

return font class

Definition at line 89 of file ocrblock.h.

                       {
    return font_class;
  }

kern()

int16_t tesseract::BLOCK::kern ( ) const

inline

return kerning

Definition at line 85 of file ocrblock.h.

                       {
    return kerning;
  }

median_size()

const ICOORD & tesseract::BLOCK::median_size ( ) const

inline

Definition at line 147 of file ocrblock.h.

                                    {
    return median_size_;
  }

name()

const char * tesseract::BLOCK::name ( ) const

inline

return filename

Definition at line 97 of file ocrblock.h.

                           {
    return filename.c_str();
  }

operator=()

BLOCK & tesseract::BLOCK::operator=

(

const BLOCK &

source

)

BLOCK::operator=

Assignment - duplicate the block structure, but with an EMPTY row list.

Definition at line 222 of file ocrblock.cpp.

  {
  this->ELIST_LINK::operator=(source);
  pdblk = source.pdblk;
  proportional = source.proportional;
  kerning = source.kerning;
  spacing = source.spacing;
  filename = source.filename; // STRINGs assign ok
  if (!rows.empty()) {
    rows.clear();
  }
  re_rotation_ = source.re_rotation_;
  classify_rotation_ = source.classify_rotation_;
  skew_ = source.skew_;
  return *this;
}

para_list()

PARA_LIST * tesseract::BLOCK::para_list ( )

inline

Definition at line 119 of file ocrblock.h.

                         {
    return &paras_;
  }

print()

void tesseract::BLOCK::print	(	FILE *	fp,
		bool	dump
	)

dump whole table

BLOCK::print

Print the info on a block

Parameters

fp	file to print on
dump	print full detail

Definition at line 188 of file ocrblock.cpp.

  {
  ICOORDELT_IT it = &pdblk.leftside; // iterator
 
  pdblk.box.print();
  tprintf("Proportional= %s\n", proportional ? "TRUE" : "FALSE");
  tprintf("Kerning= %d\n", kerning);
  tprintf("Spacing= %d\n", spacing);
  tprintf("Fixed_pitch=%d\n", pitch);
  tprintf("Filename= %s\n", filename.c_str());
 
  if (dump) {
    tprintf("Left side coords are:\n");
    for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
      tprintf("(%d,%d) ", it.data()->x(), it.data()->y());
    }
    tprintf("\n");
    tprintf("Right side coords are:\n");
    it.set_to_list(&pdblk.rightside);
    for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
      tprintf("(%d,%d) ", it.data()->x(), it.data()->y());
    }
    tprintf("\n");
  }
}

prop()

bool tesseract::BLOCK::prop ( ) const

inline

return proportional

Definition at line 71 of file ocrblock.h.

                    {
    return proportional;
  }

re_rotation()

FCOORD tesseract::BLOCK::re_rotation ( ) const

inline

Definition at line 129 of file ocrblock.h.

                             {
    return re_rotation_; // How to transform coords back to image.
  }

reflect_polygon_in_y_axis()

void tesseract::BLOCK::reflect_polygon_in_y_axis

(

)

BLOCK::reflect_polygon_in_y_axis

Reflects the polygon in the y-axis and recompute the bounding_box. Does nothing to any contained rows/words/blobs etc.

Definition at line 104 of file ocrblock.cpp.

                                      {
  pdblk.poly_block()->reflect_in_y_axis();
  pdblk.box = *pdblk.poly_block()->bounding_box();
}

reject_blobs()

C_BLOB_LIST * tesseract::BLOCK::reject_blobs ( )

inline

Definition at line 126 of file ocrblock.h.

                              {
    return &rej_blobs;
  }

render_mask()

Image tesseract::BLOCK::render_mask ( TBOX *  mask_box )

inline

Definition at line 155 of file ocrblock.h.

                                    {
    return pdblk.render_mask(re_rotation_, mask_box);
  }

restricted_bounding_box()

TBOX tesseract::BLOCK::restricted_bounding_box	(	bool	upper_dots,
		bool	lower_dots
	)		const

Definition at line 88 of file ocrblock.cpp.

                                                                          {
  TBOX box;
  // This is a read-only iteration of the rows in the block.
  ROW_IT it(const_cast<ROW_LIST *>(&rows));
  for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
    box += it.data()->restricted_bounding_box(upper_dots, lower_dots);
  }
  return box;
}

right_to_left()

bool tesseract::BLOCK::right_to_left ( ) const

inline

Definition at line 74 of file ocrblock.h.

                             {
    return right_to_left_;
  }

rotate()

void tesseract::BLOCK::rotate

(

const FCOORD &

rotation

)

BLOCK::rotate

Rotate the polygon by the given rotation and recompute the bounding_box.

Definition at line 81 of file ocrblock.cpp.

                                         {
  pdblk.poly_block()->rotate(rotation);
  pdblk.box = *pdblk.poly_block()->bounding_box();
}

row_list()

ROW_LIST * tesseract::BLOCK::row_list ( )

inline

get rows

Definition at line 111 of file ocrblock.h.

                       {
    return &rows;
  }

set_cell_over_xheight()

void tesseract::BLOCK::set_cell_over_xheight ( float  ratio )

inline

Definition at line 107 of file ocrblock.h.

                                          {
    cell_over_xheight_ = ratio;
  }

set_classify_rotation()

void tesseract::BLOCK::set_classify_rotation ( const FCOORD &  rotation )

inline

Definition at line 138 of file ocrblock.h.

                                                     {
    classify_rotation_ = rotation;
  }

set_font_class()

void tesseract::BLOCK::set_font_class ( int16_t  font )

inline

set font class

Definition at line 67 of file ocrblock.h.

                                    {
    font_class = font;
  }

set_median_size()

void tesseract::BLOCK::set_median_size ( int  x, int  y  )

inline

Definition at line 150 of file ocrblock.h.

                                     {
    median_size_.set_x(x);
    median_size_.set_y(y);
  }

set_re_rotation()

void tesseract::BLOCK::set_re_rotation ( const FCOORD &  rotation )

inline

Definition at line 132 of file ocrblock.h.

                                               {
    re_rotation_ = rotation;
  }

set_right_to_left()

void tesseract::BLOCK::set_right_to_left ( bool  value )

inline

Definition at line 77 of file ocrblock.h.

                                     {
    right_to_left_ = value;
  }

set_skew()

void tesseract::BLOCK::set_skew ( const FCOORD &  skew )

inline

Definition at line 144 of file ocrblock.h.

                                    {
    skew_ = skew;
  }

set_stats()

void tesseract::BLOCK::set_stats ( bool  prop, int16_t  kern, int16_t  space, int16_t  ch_pitch  )

inline

set space size etc.

Parameters

prop	proportional
kern	inter char size
space	inter word size
ch_pitch	pitch if fixed

Definition at line 56 of file ocrblock.h.

                                                                           {
    proportional = prop;
    kerning = static_cast<int8_t>(kern);
    spacing = space;
    pitch = ch_pitch;
  }

set_xheight()

void tesseract::BLOCK::set_xheight ( int32_t  height )

inline

set char size

Definition at line 63 of file ocrblock.h.

                                   {
    xheight = height;
  }

skew()

FCOORD tesseract::BLOCK::skew ( ) const

inline

Definition at line 141 of file ocrblock.h.

                      {
    return skew_; // Direction of true horizontal.
  }

sort_rows()

void tesseract::BLOCK::sort_rows

(

)

decreasing y order

BLOCK::sort_rows

Order rows so that they are in order of decreasing Y coordinate

Definition at line 115 of file ocrblock.cpp.

                      { // order on "top"
  ROW_IT row_it(&rows);
 
  row_it.sort(decreasing_top_order);
}

space()

int16_t tesseract::BLOCK::space ( ) const

inline

return spacing

Definition at line 93 of file ocrblock.h.

                        {
    return spacing;
  }

x_height()

int32_t tesseract::BLOCK::x_height ( ) const

inline

return xheight

Definition at line 101 of file ocrblock.h.

                           {
    return xheight;
  }

Friends And Related Function Documentation

BLOCK_RECT_IT

friend class BLOCK_RECT_IT

friend

Definition at line 35 of file ocrblock.h.

Member Data Documentation

pdblk

PDBLK tesseract::BLOCK::pdblk

Page Description Block.

Definition at line 185 of file ocrblock.h.

---

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

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