BGI/BASE/src/2bwt/FASTAToText.c

   1 /* FASTAtoText.c        FASTA to Text
   2
   3    Copyright 2004, Wong Chi Kwong, all rights reserved.
   4
   5    This module convert FASTA DNA format to plain text format.
   6
   7    This software may be used freely for any purpose. However, when distributed,
   8    the original source must be clearly stated, and, when the source code is
   9    distributed, the copyright notice must be retained and any alterations in
  10    the code must be clearly marked. No warranty is given regarding the quality
  11    of this software.
  12
  13 */
  14
  15 #include <stdio.h>
  16 #include "TypeNLimit.h"
  17 #include <stdlib.h>
  18 #include <time.h>
  19
  20 int main(int argc, char** argv) {
  21
  22     int c;
  23     FILE* inputFile;
  24     FILE* outputFile;
  25     char uppercase[256];
  26     char lowercase[256];
  27     char dna[4] = {'A', 'C', 'G', 'T'};
  28
  29     if (argc != 3 && argc != 4) {
  30         fprintf(stderr, "%s <input> <output>\n", argv[0]);
  31         fprintf(stderr, "%s <input> <output> a\n", argv[0]);
  32         fprintf(stderr, "%s <input> <output> n\n", argv[0]);
  33         fprintf(stderr, "%s <input> <output> r\n", argv[0]);
  34         return 1;
  35     }
  36     inputFile = (FILE*)fopen64(argv[1], "r");
  37     if (inputFile == NULL) {
  38         fprintf(stderr, "Input file not exists!\n");
  39         return 1;
  40     }
  41     outputFile = (FILE*)fopen64(argv[2], "r");
  42     if (outputFile != NULL) {
  43         fprintf(stderr, "Output file exists!\n");
  44         return 1;
  45     }
  46     outputFile = (FILE*)fopen64(argv[2], "wb");
  47
  48     srand(clock());
  49
  50     for (c=0; c<256; c++) {
  51         uppercase[c] = (char)c;
  52     }
  53     for (c='a'; c<='z'; c++) {
  54         uppercase[c] = (char)c + 'A' - 'a';
  55     }
  56     for (c=0; c<256; c++) {
  57         lowercase[c] = (char)c;
  58     }
  59     for (c='A'; c<='Z'; c++) {
  60         lowercase[c] = (char)c + 'a' - 'A';
  61     }
  62
  63     c = fgetc(inputFile);
  64     while (c == '\n' || c == '\r') {
  65         c = fgetc(inputFile);
  66     }
  67     while (c == '>') {
  68         while (c != '\n' && c != '\r' && c != EOF) {
  69             c = fgetc(inputFile);
  70         }
  71         while (c == '\n' || c == '\r') {
  72             c = fgetc(inputFile);
  73         }
  74     }
  75
  76     while (c != EOF) {
  77         if (argc == 4 || (c != 'n' && c != 'N')) {
  78             if ((c == 'n' || c == 'N') && (argc == 4 && argv[3][0] == 'a')) {
  79                 if (c == 'n') {
  80                     fputc('a', outputFile);
  81                 } else {
  82                     fputc('A', outputFile);
  83                 }
  84             } else {
  85                 if ((c == 'n' || c == 'N') && (argc == 4 && argv[3][0] == 'r')) {
  86                     if (c == 'n') {
  87                         fputc(lowercase[dna[rand() % 4]], outputFile);
  88                     } else {
  89                         fputc(uppercase[dna[rand() % 4]], outputFile);
  90                     }
  91                 } else {
  92                     fputc(c, outputFile);
  93                 }
  94             }
  95         }
  96         c = fgetc(inputFile);
  97         while (c == '\n' || c == '\r') {
  98             c = fgetc(inputFile);
  99         }
 100         while (c == '>') {
 101             while (c != '\n' && c != '\r' && c != EOF) {
 102                 c = fgetc(inputFile);
 103             }
 104             while (c == '\n' || c == '\r') {
 105                 c = fgetc(inputFile);
 106             }
 107         }
 108     }
 109
 110     fclose(inputFile);
 111     fclose(outputFile);
 112
 113 }