main.c

agrep

		  CONSTANT = ON;
		  argv++;
		  strcat(Pattern, argv[0]);
		  if(argc > 1 && argv[1][0] == '-') {
			fprintf(stderr, "%s: -k should be the last option in the command\n", Progname);
			exit(2);
		  }
		  break;
       case 'l' : FILENAMEONLY = ON;
                  break;
       case 'n' : LINENUM = ON;  /* output prefixed by line no*/
                  break;
       case 'v' : INVERSE = ON;  /* output no-matched lines */
                  break;
       case 't' : OUTTAIL = ON;  /* output from tail of delimiter */
                  break;
       case 'B' : BESTMATCH = ON;
                  break;
       case 'w' : WORDBOUND = ON;/* match to words */
		  if(WHOLELINE) {
			fprintf(stderr, "%s: illegal option combination\n", Progname);
			exit(2);
		  }
                  break;
       case 'y' : NOPROMPT = ON;
		  break;
       case 'I' : I = atoi(argv[0]+2);  /* Insertion Cost */
	          JUMP = ON;
                  break;
       case 'S' : S = atoi(argv[0]+2);  /* Substitution Cost */
                  JUMP = ON;
                  break;
       case 'D' : DD = atoi(argv[0]+2); /* Deletion Cost */
                  JUMP = ON;
                  break;
       case 'G' : FILEOUT = ON; 
		  COUNT = ON;
		  break;
       default  : if (isdigit(c)) {
		    APPROX = ON;
                    D = atoi(argv[0]+1);
                    if (D > MaxError) {
                      fprintf(stderr,"%s: the maximum number of errors is %d \n", Progname, MaxError);
                      exit(2);
                    }
                  } else {
                    fprintf(stderr, "%s: illegal option  -%c\n",Progname, c);
		    usage();
                  }
       } /* switch(c) */
  } /* while (--argc > 0 && (*++argv)[0] == '-') */

  if (FILENAMEONLY && NOFILENAME) {
    fprintf(stderr, "%s: -h and -l options are mutually exclusive\n",Progname);
  }
  if (COUNT && (FILENAMEONLY || NOFILENAME)) {
      FILENAMEONLY = OFF; 
      if(!FILEOUT) NOFILENAME = OFF;
  }
  if (!(PAT_FILE) && Pattern[0] == '\0') { /* Pattern not set with -e option */
    if (argc == 0) usage();
    strcpy(Pattern, *argv); 
    argc--;
    argv++;
  }
  Numfiles = 0;  
  fd = 3; /* make sure it's not 0 */
  if (argc == 0)  /* check Pattern against stdin */
    fd = 0;
  else {
    if (!(Textfiles = (CHAR **)malloc(argc * sizeof(CHAR *) ))) {
      fprintf(stderr, "%s: malloc failure (you probably don't have enough memory)\n", Progname);
      exit(2);
    }
    while (argc--)  { /* one or more filenames on command line -- put
                          the valid filenames in a array of strings */    
/*
      if ((filetype = check_file(*argv)) != ISASCIIFILE) {
	if(filetype == NOSUCHFILE) fprintf(stderr,"%s: %s: no such file or directory\n",Progname,*argv);
        argv++;
*/
      
      if ((filetype = check_file(*argv)) == NOSUCHFILE) {
	if(filetype == NOSUCHFILE) fprintf(stderr,"%s: %s: no such file or directory\n",Progname,*argv);
        argv++;
      } else { /* file is ascii*/
        if (!(Textfiles[Numfiles] = (CHAR *)malloc((strlen(*argv)+1)))) {
          fprintf(stderr, "%s: malloc failure (you probably don't have enough memory)\n", Progname);
          exit(2);
        }
        strcpy(Textfiles[Numfiles++], *argv++);
	   } /* else */
     } /* while (argc--) */
  } /* else */
  checksg(Pattern, D);       /* check if the pattern is simple */
  strcpy(OldPattern, Pattern);
  if (SGREP == 0) {
      preprocess(D_pattern, Pattern);
      strcpy(old_D_pat, D_pattern);
      M = maskgen(Pattern, D);
  }
  else M = strlen(OldPattern);
  if (PAT_FILE)  prepf(fp);
  if (Numfiles > 1) FNAME = ON;
  if (NOFILENAME) FNAME = 0;
  num_of_matched = 0;
  compat(); /* check compatibility between options */
  if (fd == 0) {
    if(FILENAMEONLY) {
       fprintf(stderr, "%s: -l option is not compatible with standard input\n", Progname);
       exit(2);  
    }
    if(PAT_FILE) mgrep(fd);
    else {
    	if(SGREP) sgrep(OldPattern, strlen(OldPattern), fd, D);
   	else      bitap(old_D_pat, Pattern, fd, M, D);
    }
    if (COUNT) {
	if(INVERSE && PAT_FILE) printf("%d\n", total_line-num_of_matched);
	else printf("%d\n", num_of_matched);
    }
  } 
  else {
    for (i = 0; i < Numfiles; i++, close(fd), num_of_matched = 0) {
      	strcpy(CurrentFileName, Textfiles[i]);
      	if ((fd = open(Textfiles[i], 0)) <= 0) {
            fprintf(stderr, "%s: can't open file %s\n",Progname, Textfiles[i]);
      	} 
	else { 
	     	if(PAT_FILE) mgrep(fd);
	     	else {
             		if(SGREP) sgrep(OldPattern, strlen(OldPattern), fd, D);
             		else      bitap(old_D_pat, Pattern, fd, M, D);
	        }
        	if (num_of_matched) NOMATCH = OFF;
        	if (COUNT && !FILEOUT) {
			if(INVERSE && PAT_FILE) {
 		    		if(FNAME) printf("%s: %d\n", CurrentFileName, total_line - num_of_matched);
				else printf("%d\n", total_line - num_of_matched);
			}
			else {
 		    		if(FNAME) printf("%s: %d\n", CurrentFileName, num_of_matched);
				else printf("%d\n", num_of_matched);
			}
        	}  /* if COUNT */
		if(FILEOUT && num_of_matched) {
			file_out(CurrentFileName);
		}
      	} /* else */
    }  /* for i < Numfiles */
    if(NOMATCH && BESTMATCH) {
	if(WORDBOUND || WHOLELINE || LINENUM || INVERSE) { 
		SGREP = 0;	
      		preprocess(D_pattern, Pattern);
      		strcpy(old_D_pat, D_pattern);
      		M = maskgen(Pattern, D);
	}
	COUNT=ON; D=1;
	while(D<M && D<=MaxError && num_of_matched == 0) {
    		for (i = 0; i < Numfiles; i++, close(fd)) {
      			strcpy(CurrentFileName, Textfiles[i]);
      			if ((fd = open(Textfiles[i], 0)) > 0) {
	     			if(PAT_FILE) mgrep(fd);
	     			else {
             		    		if(SGREP) sgrep(OldPattern,strlen(OldPattern),fd,D);
             		    		else bitap(old_D_pat,Pattern,fd,M,D);
      				}
			} 
   		}  /* for i < Numfiles */
		D++;
	} /* while */
	if(num_of_matched > 0) {
		D--; COUNT = 0;
		if(NOPROMPT) goto GO_AHEAD;
		if(D==1) fprintf(stderr, "best match has 1 error, ");
		else fprintf(stderr, "best match has %d errors, ", D);
		fflush(stderr);
		if(num_of_matched == 1) fprintf(stderr,"there is 1 match, output it? (y/n)");
		else fprintf(stderr,"there are %d matches, output them? (y/n)", num_of_matched);
		scanf("%c",&c);
		if(c != 'y') goto CONT;
GO_AHEAD:
    		for (i = 0; i < Numfiles; i++, close(fd)) {
      			strcpy(CurrentFileName, Textfiles[i]);
      			if ((fd = open(Textfiles[i], 0)) > 0) {
	     			if(PAT_FILE) mgrep(fd);
	     			else {
             		    		if(SGREP) sgrep(OldPattern,strlen(OldPattern),fd,D);
             		    		else bitap(old_D_pat,Pattern,fd,M,D);
				}
      			} 
   		}  /* for i < Numfiles */
		NOMATCH = 0;
	}
   }
 }
CONT:
 if(EATFIRST) {
      printf("\n");
      EATFIRST = OFF;
 }
 if(num_of_matched) NOMATCH = OFF;
 if(NOMATCH) exit(1);
 exit(0);
} /* end of main() */

	   
file_out(fname)
char *fname;
{
int num_read;
int fd;
int i, len;
CHAR buf[4097];
	if(FNAME) {
		len = strlen(fname);
		putchar('\n');
		for(i=0; i< len; i++) putchar(':');
		putchar('\n');
		printf("%s\n", CurrentFileName);
		len = strlen(fname);
		for(i=0; i< len; i++) putchar(':');
		putchar('\n');
		fflush(stdout);
	}
	fd = open(fname, 0);
	while((num_read = read(fd, buf, 4096)) > 0) 
		write(1, buf, num_read);
}


usage()
{
    	fprintf(stderr, "usage: %s [-#cdehiklnpstvwxBDGIS] [-f patternfile] pattern [files]\n", Progname); 
 	printf("\n");	
	fprintf(stderr, "summary of frequently used options:\n");
	fprintf(stderr, "-#: find matches with at most # errors\n");
	fprintf(stderr, "-c: output the number of matched records\n");
	fprintf(stderr, "-d: define record delimiter\n");
	fprintf(stderr, "-h: do not output file names\n");
	fprintf(stderr, "-i: case-insensitive search, e.g., 'a' = 'A'\n");
	fprintf(stderr, "-l: output the names of files that contain a match\n");
	fprintf(stderr, "-n: output record prefixed by record number\n");
	fprintf(stderr, "-v: output those records containing no matches\n");
	fprintf(stderr, "-w: pattern has to match as a word, e.g., 'win' will not match 'wind'\n");
	fprintf(stderr, "-B: best match mode. find the closest matches to the pattern\n"); 
	fprintf(stderr, "-G: output the files that contain a match\n");
 	printf("\n");	

    	exit(2);
}

checksg(Pattern, D) 
CHAR *Pattern; int D;
{                          
  char c;
  int i, m;
  m = strlen(Pattern);
  if(!(PAT_FILE) && m <= D) {
      fprintf(stderr, "%s: size of pattern must be greater than number of errors\n", Progname);
      exit(2);
  }
  SIMPLEPATTERN = ON;
  for (i=0; i < m; i++) 
  {
      switch(Pattern[i])
      {
         case ';' : SIMPLEPATTERN = OFF; break;
         case ',' : SIMPLEPATTERN = OFF; break;
         case '.' : SIMPLEPATTERN = OFF; break;
         case '*' : SIMPLEPATTERN = OFF; break;
         case '-' : SIMPLEPATTERN = OFF; break;
         case '[' : SIMPLEPATTERN = OFF; break;
         case ']' : SIMPLEPATTERN = OFF; break;
         case '(' : SIMPLEPATTERN = OFF; break;
         case ')' : SIMPLEPATTERN = OFF; break;
         case '<' : SIMPLEPATTERN = OFF; break;
         case '>' : SIMPLEPATTERN = OFF; break;
         case '^' : if(D > 0) SIMPLEPATTERN = OFF; 
		    break;
         case '$' : if(D > 0) SIMPLEPATTERN = OFF; 
		    break;
         case '|' : SIMPLEPATTERN = OFF; break;
         case '#' : SIMPLEPATTERN = OFF; break;
         case '\\' : SIMPLEPATTERN = OFF; break;
         default  : break;
      }
  }
  if (CONSTANT) SIMPLEPATTERN = ON;
  if (SIMPLEPATTERN == OFF) return;
  if (NOUPPER && D) return;     
  if (JUMP == ON) return;
  if (I == 0) return;
  if (LINENUM) return;
  if (DELIMITER) return;   
  if (INVERSE) return;
  if (WORDBOUND && D > 0) return;  
  if (WHOLELINE && D > 0) return;  
  if (SILENT) return;     /* REMINDER: to be removed */
  SGREP = ON;
  if(m >= 16) DNA = ON;
  for(i=0; i<m; i++) {
	c = Pattern[i];
	if(c == 'a' || c == 'c' || c == 't' || c == 'g' ) ;
	else DNA = OFF;
  }
  return;
}

output (buffer, i1, i2, j)  
register CHAR *buffer; int i1, i2, j;
{
register CHAR *bp, *outend;
	if(i1 > i2) return;
        num_of_matched++;
        if(COUNT)  return;
        if(SILENT) return;
	if(OUTTAIL) {
              i1 = i1 + D_length;
              i2 = i2 + D_length;
        }
        if(DELIMITER) j = j+1;
        if(FIRSTOUTPUT) {
           if (buffer[i1] == '\n')  {
               i1++;
               EATFIRST = ON;
           }
           FIRSTOUTPUT = 0;
        }
        if(TRUNCATE) {
           fprintf(stderr, "WARNING!!!  some lines have been truncated in output record #%d\n", num_of_matched-1);
        }
        while(buffer[i1] == '\n' && i1 <= i2) {
	   printf("\n");
           i1++;
        }
        if(FNAME == ON) printf("%s: ", CurrentFileName);
        if(LINENUM) printf("%d: ", j-1); 
	bp = buffer + i1;
	outend = buffer + i2;
	while(bp <= outend) putchar(*bp++);
}

/* end of main.c */