/*
 * csvtotab -- convert comma-separated values to tab-separated values
 *
 * RFC 4180 requires CRLF as line endings; this program also accepts
 * CR or LF. RFC 4180 requires the same number of fields in each line,
 * this program does not check for that.
 *
 * Line endings on output are LF, independent of the input.
 *
 * Copyright © 2014 World Wide Web Consortium
 * See http://www.w3.org/Consortium/Legal/2002/copyright-software-20021231
 *
 * Author: Bert Bos <bert@w3.org>
 * Created: 3 April 2014
 */

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <err.h>
#include <getopt.h>

#define VERSION "0.1"

static struct option longopts[] = {
  {"help",	no_argument,		NULL, 'h'},
  {"separator",	required_argument,	NULL, 's'},
  {"version",	no_argument,		NULL, 'v'}
};

static char sep = ',';


/* version -- print program version and exit */
static void version(const char *prog)
{
  printf("%s %s\n", prog, VERSION);
  exit(0);
}


/* usage -- print usage message on stderr and exit */
static void usage(const char *prog)
{
  fprintf(stderr, "Usage: %s [-h] [-v] [-s char] [file [file...]]\n", prog);
  exit(1);
}


/* help -- print usage information and exit */
static void help(const char *prog)
{
  printf("\
Usage: %s [options] [--] [csv-file [csv-file...]]\n\
    Version: %s\n\
    Options:\n\
    -s, --separator=<char>\n\
        The character that separates fields in the input. Default comma (,)\n\
    -h, --help\n\
        Show this help text and exit.\n\
    -v, --version\n\
        Show the program version nummer and exit.\n\
    The arguments are CSV files. If not files are given or if the name\n\
    is -, the program reads from standard input.\n",
	 prog, VERSION);
  exit(0);
}


/* process -- read file f as a CSV and output tab-separated to stdout */
static void process(FILE *f, const char *name)
{
  enum {START, COMMA, IN_FIELD, QUOTE, QUOTED, CR} state = START;
  int c;

  /* Remove the BOM, if any */
  if ((c = fgetc(f)) != 0xEF) {
    ungetc(c, f);
  } else if ((c = fgetc(f)) != 0xBB) {
    putchar(0xEF); state = IN_FIELD; ungetc(c, f);
  } else if ((c = fgetc(f)) != 0xBF) {
    putchar(0xEF); putchar(0xBB); state = IN_FIELD; ungetc(c, f);
  }

  while ((c = fgetc(f)) != EOF) {
    if (c == sep) {
      switch (state) {
      case QUOTED: putchar(c); break;
      case CR: fputs("\n\t", stdout); state = COMMA; break;
      default: putchar('\t'); state = COMMA; break;
      }
    } else if (c == '"') {
      switch(state) {
      case IN_FIELD: errx(1, "Found a quote (\") inside a field");
      case QUOTE: putchar('"'); state = QUOTED; break;
      case QUOTED: state = QUOTE; break;
      case CR: putchar('\n'); state = QUOTED; break;
      default: state = QUOTED; break;
      }
    } else if (c == '\t') {
      switch (state) {
      case QUOTE: errx(1, "Found a lone quote (\") in a quoted field");
      case QUOTED: fputs("\\t", stdout); break;
      case CR: fputs("\n\\t", stdout); state = IN_FIELD; break;
      default: fputs("\\t", stdout); state = IN_FIELD; break;
      }
    } else if (c == '\r') {
      switch (state) {
      case QUOTED: fputs("\\r", stdout); break;
      case CR: putchar('\n'); break;
      default: state = CR; break;
      }
    } else if (c == '\n') {
      switch (state) {
      case QUOTED: fputs("\\n", stdout); break;
      case CR: fputs("\n", stdout); state = START; break;
      default: putchar('\n'); state = START; break;
      }
    } else if (c == '\\') {
      switch (state) {
      case QUOTED: fputs("\\\\", stdout); break;
      case QUOTE: errx(1, "Found a lone quote (\") in a quoted field");
      case CR: fputs("\n\\\\", stdout); state = IN_FIELD; break;
      default: fputs("\\\\", stdout); state = IN_FIELD;
      }
    } else {
      switch (state) {
      case QUOTE: errx(1, "Found a lone quote (\") in a quoted field");
      case QUOTED: putchar(c); break;
      case CR: putchar('\n'); putchar(c); state = IN_FIELD; break;
      default: putchar(c); state = IN_FIELD; break;
      }
    }
  }
  switch (state) {
  case START: break;
  case QUOTED: errx(1, "End of file inside a quoted field");
  case CR:
  case QUOTE:
  case IN_FIELD:
  case COMMA: putchar('\n'); break;
  }
}


int main(int argc, char *argv[])
{
  FILE *f;
  int i, c;

  while ((c = getopt_long(argc, argv, "hs:v", longopts, NULL)) != -1)
    switch (c) {
    case 'h': help(argv[0]); break;
    case 's': if (!*optarg||optarg[1]) usage(argv[0]); sep = *optarg; break;
    case 'v': version(argv[0]); break;
    default: usage(argv[0]);
    }

  if (sep == '"' || sep == '\r' || sep == '\n')
    errx(1, "Cannot use this separator.");

  if (optind == argc)
    process(stdin, "<STDIN>");
  else
    for (i = optind; i < argc; i++)
      if (argv[i][0] == '-' && !argv[i][1]) process(stdin, "<STDIN>");
      else if (!(f = fopen(argv[i], "r"))) err(1, NULL);
      else process(f, argv[i]);

  return 0;
}