boxread.cpp File Reference

#include "boxread.h"
#include <cstring>
#include <locale>
#include <sstream>
#include <string>
#include "errcode.h"
#include "fileerr.h"
#include <tesseract/genericvector.h>
#include <tesseract/helpers.h>
#include "rect.h"
#include <tesseract/strngs.h>
#include "tprintf.h"
#include <tesseract/unichar.h>

Go to the source code of this file.

Functions
FILE *	OpenBoxFile (const STRING &fname)
bool	ReadAllBoxes (int target_page, bool skip_blanks, const STRING &filename, GenericVector< TBOX > *boxes, GenericVector< STRING > *texts, GenericVector< STRING > *box_texts, GenericVector< int > *pages)
bool	ReadMemBoxes (int target_page, bool skip_blanks, const char *box_data, bool continue_on_failure, GenericVector< TBOX > *boxes, GenericVector< STRING > *texts, GenericVector< STRING > *box_texts, GenericVector< int > *pages)
bool	ReadNextBox (int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)
bool	ReadNextBox (int target_page, int *line_number, FILE *box_file, STRING *utf8_str, TBOX *bounding_box)
bool	ParseBoxFileStr (const char *boxfile_str, int *page_number, STRING *utf8_str, TBOX *bounding_box)
void	MakeBoxFileStr (const char *unichar_str, const TBOX &box, int page_num, STRING *box_str)

Function Documentation

MakeBoxFileStr()

void MakeBoxFileStr	(	const char *	unichar_str,
		const TBOX &	box,
		int	page_num,
		STRING *	box_str
	)

Definition at line 249 of file boxread.cpp.

                                     {
  *box_str = unichar_str;
  box_str->add_str_int(" ", box.left());
  box_str->add_str_int(" ", box.bottom());
  box_str->add_str_int(" ", box.right());
  box_str->add_str_int(" ", box.top());
  box_str->add_str_int(" ", page_num);
}

OpenBoxFile()

FILE* OpenBoxFile

(

const STRING &

fname

)

Definition at line 54 of file boxread.cpp.

                                       {
  std::string filename = BoxFileName(fname.c_str());
  FILE* box_file = nullptr;
  if (!(box_file = fopen(filename.c_str(), "rb"))) {
    CANTOPENFILE.error("read_next_box", TESSEXIT, "Can't open box file %s",
                       filename.c_str());
  }
  return box_file;
}

ParseBoxFileStr()

bool ParseBoxFileStr	(	const char *	boxfile_str,
		int *	page_number,
		STRING *	utf8_str,
		TBOX *	bounding_box
	)

Definition at line 181 of file boxread.cpp.

                                                           {
  *bounding_box = TBOX();       // Initialize it to empty.
  *utf8_str = "";
  char uch[kBoxReadBufSize];
  const char *buffptr = boxfile_str;
  // Read the unichar without messing up on Tibetan.
  // According to issue 253 the utf-8 surrogates 85 and A0 are treated
  // as whitespace by sscanf, so it is more reliable to just find
  // ascii space and tab.
  int uch_len = 0;
  // Skip unicode file designation, if present.
  const auto *ubuf = reinterpret_cast<const unsigned char*>(buffptr);
  if (ubuf[0] == 0xef && ubuf[1] == 0xbb && ubuf[2] == 0xbf)
      buffptr += 3;
  // Allow a single blank as the UTF-8 string. Check for empty string and
  // then blindly eat the first character.
  if (*buffptr == '\0') return false;
  do {
    uch[uch_len++] = *buffptr++;
  } while (*buffptr != '\0' && *buffptr != ' ' && *buffptr != '\t' &&
           uch_len < kBoxReadBufSize - 1);
  uch[uch_len] = '\0';
  if (*buffptr != '\0') ++buffptr;
  int x_min = INT_MAX;
  int y_min = INT_MAX;
  int x_max = INT_MIN;
  int y_max = INT_MIN;
  *page_number = 0;
  std::stringstream stream(buffptr);
  stream.imbue(std::locale::classic());
  stream >> x_min;
  stream >> y_min;
  stream >> x_max;
  stream >> y_max;
  stream >> *page_number;
  if (x_max < x_min || y_max < y_min) {
    tprintf("Bad box coordinates in boxfile string! %s\n", ubuf);
    return false;
  }
  // Test for long space-delimited string label.
  if (strcmp(uch, kMultiBlobLabelCode) == 0 &&
      (buffptr = strchr(buffptr, '#')) != nullptr) {
    strncpy(uch, buffptr + 1, kBoxReadBufSize - 1);
    uch[kBoxReadBufSize - 1] = '\0';  // Prevent buffer overrun.
    chomp_string(uch);
    uch_len = strlen(uch);
  }
  // Validate UTF8 by making unichars with it.
  int used = 0;
  while (used < uch_len) {
    tesseract::UNICHAR ch(uch + used, uch_len - used);
    int new_used = ch.utf8_len();
    if (new_used == 0) {
      tprintf("Bad UTF-8 str %s starts with 0x%02x at col %d\n",
              uch + used, uch[used], used + 1);
      return false;
    }
    used += new_used;
  }
  *utf8_str = uch;
  if (x_min > x_max) Swap(&x_min, &x_max);
  if (y_min > y_max) Swap(&y_min, &y_max);
  bounding_box->set_to_given_coords(x_min, y_min, x_max, y_max);
  return true;  // Successfully read a box.
}

ReadAllBoxes()

bool ReadAllBoxes	(	int	target_page,
		bool	skip_blanks,
		const STRING &	filename,
		GenericVector< TBOX > *	boxes,
		GenericVector< STRING > *	texts,
		GenericVector< STRING > *	box_texts,
		GenericVector< int > *	pages
	)

Definition at line 71 of file boxread.cpp.

                                             {
  GenericVector<char> box_data;
  if (!tesseract::LoadDataFromFile(BoxFileName(filename.c_str()).c_str(), &box_data))
    return false;
  // Convert the array of bytes to a string, so it can be used by the parser.
  box_data.push_back('\0');
  return ReadMemBoxes(target_page, skip_blanks, &box_data[0],
                      /*continue_on_failure*/ true, boxes, texts, box_texts,
                      pages);
}

ReadMemBoxes()

bool ReadMemBoxes	(	int	target_page,
		bool	skip_blanks,
		const char *	box_data,
		bool	continue_on_failure,
		GenericVector< TBOX > *	boxes,
		GenericVector< STRING > *	texts,
		GenericVector< STRING > *	box_texts,
		GenericVector< int > *	pages
	)

Definition at line 87 of file boxread.cpp.

                                             {
  STRING box_str(box_data);
  GenericVector<STRING> lines;
  box_str.split('\n', &lines);
  if (lines.empty()) return false;
  int num_boxes = 0;
  for (int i = 0; i < lines.size(); ++i) {
    int page = 0;
    STRING utf8_str;
    TBOX box;
    if (!ParseBoxFileStr(lines[i].c_str(), &page, &utf8_str, &box)) {
      if (continue_on_failure)
        continue;
      else
        return false;
    }
    if (skip_blanks && (utf8_str == " " || utf8_str == "\t")) continue;
    if (target_page >= 0 && page != target_page) continue;
    if (boxes != nullptr) boxes->push_back(box);
    if (texts != nullptr) texts->push_back(utf8_str);
    if (box_texts != nullptr) {
      STRING full_text;
      MakeBoxFileStr(utf8_str.c_str(), box, target_page, &full_text);
      box_texts->push_back(full_text);
    }
    if (pages != nullptr) pages->push_back(page);
    ++num_boxes;
  }
  return num_boxes > 0;
}

ReadNextBox() [1/2]

bool ReadNextBox	(	int *	line_number,
		FILE *	box_file,
		STRING *	utf8_str,
		TBOX *	bounding_box
	)

Definition at line 134 of file boxread.cpp.

                                                       {
  return ReadNextBox(-1, line_number, box_file, utf8_str, bounding_box);
}

ReadNextBox() [2/2]

bool ReadNextBox	(	int	target_page,
		int *	line_number,
		FILE *	box_file,
		STRING *	utf8_str,
		TBOX *	bounding_box
	)

Definition at line 142 of file boxread.cpp.

                                                       {
  int page = 0;
  char buff[kBoxReadBufSize];   // boxfile read buffer
  char *buffptr = buff;
 
  while (fgets(buff, sizeof(buff) - 1, box_file)) {
    (*line_number)++;
 
    buffptr = buff;
    const auto *ubuf = reinterpret_cast<const unsigned char*>(buffptr);
    if (ubuf[0] == 0xef && ubuf[1] == 0xbb && ubuf[2] == 0xbf)
      buffptr += 3;  // Skip unicode file designation.
    // Check for blank lines in box file
    if (*buffptr == '\n' || *buffptr == '\0') continue;
    // Skip blank boxes.
    if (*buffptr == ' ' || *buffptr == '\t') continue;
    if (*buffptr != '\0') {
      if (!ParseBoxFileStr(buffptr, &page, utf8_str, bounding_box)) {
        tprintf("Box file format error on line %i; ignored\n", *line_number);
        continue;
      }
      if (target_page >= 0 && target_page != page)
        continue;  // Not on the appropriate page.
      return true;  // Successfully read a box.
    }
  }
  fclose(box_file);
  return false;  // EOF
}