// HsTraceDump.c
// - extract readable text in HsTraceDatabase.PDB
// by HAL <http://hal9.blog3.fc2.com> 2007/1/13

#include <stdio.h>
#include <string.h>

#define BUFSZ 8

typedef enum {NL_DOS, NL_UNIX, NL_MAC } NewLine_t;

void newline(NewLine_t newlineMode, FILE* ofp){
  switch(newlineMode){
  case NL_UNIX: fputc('\n', ofp); return;
  case NL_MAC : fputc('\r', ofp); return;
  default:
  case NL_DOS : fprintf(ofp, "\r\n"); return;
  }
}

void usage(char* argv0){
  fprintf(stderr, "USAGE:\n");
  fprintf(stderr, "\t%s [-a][-A][-u][-m][-o output.txt] HsTraceDatabase.PDB\n", argv0);
  fprintf(stderr, "option:\n");
  fprintf(stderr, "\t-a : print all readable chars with newline\n");
  fprintf(stderr, "\t-A : print all readable chars without newline\n");
  fprintf(stderr, "\t-u : (Unix) LF as newline (default: (Dos) CR-LF as newline)\n");
  fprintf(stderr, "\t-m : (Mac) CR as newline\n");
  fprintf(stderr, "\t-o : specify output file name (default: STDOUT)\n");
}

int main(int argc, char** argv){
  int argi = 1;
  FILE *ifp;
  FILE *ofp = stdout;
  char buf[BUFSZ];
  enum {WORK_SEARCH, WORK_PRINT} workMode = WORK_SEARCH;
  enum {PRINT_NORMAL, PRINT_ALL_WITH_NEWLINE, PRINT_ALL} printMode = PRINT_NORMAL;
  NewLine_t nlMode = NL_DOS;
  
  // check arg
  if(argc < 2){
    usage(argv[0]);
    exit(1);
  }
  
  // check option
  for(argi=1; argi<argc-1; argi++){
    if(strlen(argv[argi]) == 2 && argv[argi][0] == '-'){
      switch(argv[argi][1]){
        case 'a':
          printMode = PRINT_ALL_WITH_NEWLINE;
          break;

        case 'A':
          printMode = PRINT_ALL;
          break;

      case 'u':
	nlMode = NL_UNIX;
	break;

      case 'm':
	nlMode = NL_MAC;
	break;

        case 'o':
          if(argi<argc-2){
            argi++;
            ofp = fopen(argv[argi], "w");
            if(ofp == NULL) ofp = stdout;
            break;
          }
          //NOT BREAK

        default:
          usage(argv[0]);
          exit(1);
      }
    }
    else{
      usage(argv[0]);
      exit(1);
    }
  }

  ifp = fopen(argv[argi], "rb");
  if(ifp == NULL){
    fprintf(stderr, "ERR: read error %s\n", argv[argi]);
    usage(argv[0]);
    exit(1);
  }

  switch(printMode){
    case PRINT_ALL:
    case PRINT_ALL_WITH_NEWLINE:
      while(!feof(ifp)){
        buf[0] = fgetc(ifp);
        if(!isprint(buf[0])) 
          continue;

        if(buf[0] == '<' && printMode == PRINT_ALL_WITH_NEWLINE) newline(nlMode, ofp);
        fputc(buf[0], ofp);
        if(buf[0] == '>' && printMode == PRINT_ALL_WITH_NEWLINE) newline(nlMode, ofp);
      }
      break;

    default:
    case PRINT_NORMAL:
      while(!feof(ifp)){
        buf[0] = fgetc(ifp);
        if(!isprint(buf[0])) 
          continue;
        
        if(workMode == WORK_SEARCH){ // search pattern: "<Text>"
          if(buf[0] == '<'){
            if(fgets(buf, 6, ifp) != NULL && strncmp(buf, "Text>", 5) == 0)
              workMode = WORK_PRINT;
            else
              fseek(ifp, -5, SEEK_CUR);
          }
          continue;
        }
        else if(workMode == WORK_PRINT){  // search pattern: "</Text>"
          if(buf[0] == '<'){
            if(fgets(buf, 7, ifp) != NULL && strncmp(buf, "/Text>", 6) == 0){
              workMode = WORK_SEARCH;
              newline(nlMode, ofp);
              continue;
            }
            else{
              fseek(ifp, -6, SEEK_CUR);
              buf[0] = '<';
            }
          }
          fputc(buf[0], ofp);
        }
      }
      break;
  }

  fclose(ifp);
  if(ofp != stdout) 
    fclose(ofp);

  exit(0);
}