runutils.cpp

Ocr source code. provides an Ocr engine that works in C++ language.

          else if(split_line[0] == "PD_NUM_MATCHES") {

               int value = atoi(split_line[1].c_str());
               if(value < 1) {

                    string error_msg = "PD_NUM_MATCHES: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }
               
               parse_dialog_num_matches = value;

          }
          else if(split_line[0] == "IGNORE_EDGES_TOP") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "IGNORE_EDGES_TOP: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               scan_ignore_edges_top = value;

          }
          else if(split_line[0] == "IGNORE_EDGES_BOTTOM") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "IGNORE_EDGES_BOTTOM: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               scan_ignore_edges_bottom = value;

          }
          else if(split_line[0] == "IGNORE_EDGES_LEFT") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "IGNORE_EDGES_LEFT: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               scan_ignore_edges_left = value;

          }
          else if(split_line[0] == "IGNORE_EDGES_RIGHT") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "IGNORE_EDGES_RIGHT: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               scan_ignore_edges_right = value;

          }
          else if(split_line[0] == "SPACE_INT_PADDING") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "SPACE_INT_PADDING: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               space_int_padding = value;

          }
          else if(split_line[0] == "LINEH_INT_PADDING") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "LINEH_INT_PADDING: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               line_height_int_padding = value;

          }
          // training config:
          else if(split_line[0] == "TRCFG_GRID_SIZE_X") {

               int value = atoi(split_line[1].c_str());
               if(value < 1) {

                    string error_msg = "TRCFG_GRID_SIZE_X: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }
               
               trcfg_grid_size_x = value;

          }
          else if(split_line[0] == "TRCFG_GRID_SIZE_Y") {

               int value = atoi(split_line[1].c_str());
               if(value < 1) {

                    string error_msg = "TRCFG_GRID_SIZE_Y: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_grid_size_y = value;

          }
          else if(split_line[0] == "TRCFG_GRIDCELL_SIZE_X") {

               int value = atoi(split_line[1].c_str());
               if(value < 1) {

                    string error_msg = "TRCFG_GRIDCELL_SIZE_X: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_gridcell_size_x = value;

          }
          else if(split_line[0] == "TRCFG_GRIDCELL_SIZE_Y") {

               int value = atoi(split_line[1].c_str());
               if(value < 1) {

                    string error_msg = "TRCFG_GRIDCELL_SIZE_Y: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_gridcell_size_y = value;

          }
          else if(split_line[0] == "TRCFG_INTERSECT_PX") {

               int value = atoi(split_line[1].c_str());
               if(value < 1) {

                    string error_msg = "TRCFG_INTERSECT_PX: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_intersect_px = value;

          }
          else if(split_line[0] == "TRCFG_SPACE_INT_LOWER") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "TRCFG_SPACE_INT_LOWER: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_space_int.lower = value;

          }
          else if(split_line[0] == "TRCFG_SPACE_INT_UPPER") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "TRCFG_SPACE_INT_UPPER: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_space_int.upper = value;

          }
          else if(split_line[0] == "TRCFG_LINE_HEIGHT_INT_LOWER") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "TRCFG_LINE_HEIGHT_INT_LOWER: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_line_height_int.lower = value;

          }
          else if(split_line[0] == "TRCFG_LINE_HEIGHT_INT_UPPER") {

               int value = atoi(split_line[1].c_str());
               if(value < 0) {

                    string error_msg = "TRCFG_LINE_HEIGHT_INT_UPPER: value out of range";
                    log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
                    continue;

               }

               trcfg_line_height_int.upper = value;

          }
          else {

               string error_msg = string("in parse_from_config at line #") +
                                  int_to_string(line_num);
               log_error(ERROR_INVALID_CONFIG_OPTION, error_msg.c_str(), NULL);
               fclose(fp);
               return false;

          }

     } // while

     fclose(fp);

     return true;

}

bool write_config_file(string filename) {

     FILE *fp = fopen(filename.c_str(), "w");
     if(fp == NULL) {

          log_error(ERROR_OPENING_FILE,"in write_config_file",strerror(errno));
          return false;

     }

     // config:
     string output_line = string("CFG_PROFILE = ") + cfg_profile + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("CFG_PROFILE_DIR = ") + cfg_profile_dir + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("CFG_SAVE_DIR = ") + cfg_save_dir + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("CFG_SCAN_DIR = ") + cfg_scan_dir + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("CFG_LOG_DIR = ") + cfg_log_dir + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("CFG_LOG_FILE = ") + cfg_log_file + "\n\n";
     fputs(output_line.c_str(), fp);

     // misc config:
     output_line = string("GRID_RGB_INT_RL = ") + 
                   int_to_string(grid_rgb_int.Rl) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("GRID_RGB_INT_RU = ") + 
                   int_to_string(grid_rgb_int.Ru) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("GRID_RGB_INT_GL = ") + 
                   int_to_string(grid_rgb_int.Gl) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("GRID_RGB_INT_GU = ") + 
                   int_to_string(grid_rgb_int.Gu) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("GRID_RGB_INT_BL = ") + 
                   int_to_string(grid_rgb_int.Bl) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("GRID_RGB_INT_BU = ") + 
                   int_to_string(grid_rgb_int.Bu) + "\n\n";
     fputs(output_line.c_str(), fp);

     // parse config:
     output_line = string("ZI_PR0_PG = ") + 
                   double_to_string(ZERO_INST_PR0_PGIVEN) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("ZI_PR1_PG = ") + 
                   double_to_string(ZERO_INST_PR1_PGIVEN) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("ZI_PR0_PGC = ") + 
                   double_to_string(ZERO_INST_PR0_PGC) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("ZI_PR1_PGC = ") + 
                   double_to_string(ZERO_INST_PR1_PGC) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("PRSF = ") + double_to_string(PR_SCALING_FACTOR)+'\n';
     fputs(output_line.c_str(), fp);
     output_line = string("PD_NUM_MATCHES = ") +
                   int_to_string(parse_dialog_num_matches) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("IGNORE_EDGES_TOP = ") +
                   int_to_string(scan_ignore_edges_top) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("IGNORE_EDGES_BOTTOM = ") +
                   int_to_string(scan_ignore_edges_bottom) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("IGNORE_EDGES_LEFT = ") +
                   int_to_string(scan_ignore_edges_left) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("IGNORE_EDGES_RIGHT = ") +
                   int_to_string(scan_ignore_edges_right) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("SPACE_INT_PADDING = ") + 
                   int_to_string(space_int_padding) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("LINEH_INT_PADDING = ") +
                   int_to_string(line_height_int_padding) + '\n';
     fputs(output_line.c_str(), fp);

     // training config:
     output_line = string("TRCFG_GRID_SIZE_X = ") + 
                   int_to_string(trcfg_grid_size_x) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_GRID_SIZE_Y = ") + 
                   int_to_string(trcfg_grid_size_y) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_GRIDCELL_SIZE_X = ") + 
                   int_to_string(trcfg_gridcell_size_x) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_GRIDCELL_SIZE_Y = ") + 
                   int_to_string(trcfg_gridcell_size_y) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_INTERSECT_PX = ") + 
                   int_to_string(trcfg_intersect_px) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_SPACE_INT_LOWER = ") + 
                   int_to_string(trcfg_space_int.lower) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_SPACE_INT_UPPER = ") + 
                   int_to_string(trcfg_space_int.upper) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_LINE_HEIGHT_INT_LOWER = ") + 
                   int_to_string(trcfg_line_height_int.lower) + '\n';
     fputs(output_line.c_str(), fp);
     output_line = string("TRCFG_LINE_HEIGHT_INT_UPPER = ") + 
                   int_to_string(trcfg_line_height_int.upper) + '\n';
     fputs(output_line.c_str(), fp);

     fclose(fp);

     return true;

}

/////// string utils:
string trim_whitespace(string input) {

     vector<string> whitespace_chars;
     whitespace_chars.push_back("\n");
     whitespace_chars.push_back(" ");
     whitespace_chars.push_back("\t");

     for(int i = 0; i < whitespace_chars.size(); i++) {
	     string cur_char = whitespace_chars[i];
          int cur_pos = 0;
          while(cur_pos != -1) {
               cur_pos = input.find_first_of(cur_char, 0);
               if(cur_pos != -1)
                    input = input.replace(cur_pos, 1, "");
          }
     }

     return input;
}

string strip_newlines(string input) {

     vector<string> whitespace_chars;
     whitespace_chars.push_back("\n");
     //whitespace_chars.push_back("\t");

     for(int i = 0; i < whitespace_chars.size(); i++) {
	     string cur_char = whitespace_chars[i];
          int cur_pos = 0;
          while(cur_pos != -1) {
               cur_pos = input.find_first_of(cur_char, 0);
               if(cur_pos != -1)
                    input = input.replace(cur_pos, 1, "");
          }
     }

     return input;
}

string strip_unused_chars(string input) {

     int cur_char;
     for(int i = 0; i < input.length(); ) {

          cur_char = (int) input[i];
          if(((cur_char >= 97) && (cur_char <= 122)) ||
             ((cur_char >= 65) && (cur_char <= 90)) || ((char) cur_char == ' '))
               i++;
          else 
               input = input.replace(i, 1, "");

     }

     return input;

}

vector<string> split_string(string line, char delimeter) {

     vector<string> r_val;
     int cur_pos = 0;//, next_pos = 0;
     string temp_val = "";//, first_pos = "";
     
     while(line.length() != 0) {
          cur_pos = line.find_first_of(delimeter, 0);
          if(cur_pos == -1) {
               r_val.push_back(line);
               return r_val;
          }
          else {
               if(cur_pos != 0)
                    r_val.push_back(line.substr(0, cur_pos));
               else
                    r_val.push_back(line.substr(0, 1));
                              
               if(line.length() == cur_pos + 1)
                    return r_val;
               
               line = line.substr(cur_pos + 1, line.length() - cur_pos - 1);
          }

     }

     return r_val;

}

string int_to_string(int val) {

     char int_val[10];
     sprintf(int_val, "%i", val);
     return string(int_val);

}

string double_to_string(double val) {

     char double_str[16];
     sprintf(double_str, "%05f", val);
     return string(double_str);

}

// error checking:
bool directory_exists(string dir_name) {

     DIR *dir = opendir(dir_name.c_str());