/*
 * tabtocsv -- convert tab-separated values to comma-separated values
 *
 * 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 <err.h>
#include <unistd.h>
#include <getopt.h>
#include <stdbool.h>
#include <stdlib.h>

#define VERSION "0.1"

static struct option longopts[] = {
  {"crlf",	no_argument,		NULL, 'c'},
  {"help",	no_argument,		NULL, 'h'},
  {"rfc4180",	no_argument,		NULL, 'r'},
  {"separator",	required_argument,	NULL, 's'},
  {"version",	no_argument,		NULL, 'v'}
};

static bool strict = false;
static bool crlf = false;
static char sep = ',';

#define out(c) putchar((c))
#define outs(s) fputs((s), stdout)


/* 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] [-r] [-s char] [-c] [file [file...]]\n",
	  prog);
  exit(1);
}


/* help -- print explanation of command line options and exit */
static void help(const char *prog)
{
  printf("\
Usage: %s [options] [--] [file [file...]]\n\
    Convert tab-separated files to comma-separated.\n\
    Version: %s\n\
    Options:\n\
    -r, --rfc4180\n\
        Strict conformance to RFC 4180: refuse control characters\n\
        (ASCII 0-31), use a comma (,) as separator and CR-LF as line\n\
        ending. Implies --crlf. Incompatible with --separator.\n\
    -c, --crlf\n\
        Force line endings to be CR-LF. Default is to keep the\n\
        line endings of the input.\n\
    -s, --separator=<character>\n\
        The character to insert between fields. Default is a comma (,).\n\
    -h, --help\n\
        Print this help text and exit.\n\
    -v, --version\n\
        Print the progam version number and exit.\n\
    The arguments are files with lines of tab-separated fields.\n\
    Without arguments, or if the argument is -, the lines are read\n\
    from standard input.\n",
	 prog, VERSION);
  exit(0);
}


/* process -- read tab-separated file f and write comma-separated to stdout */
static void process(FILE *f, const char *name)
{
  enum {START, TAB, CR, IN_FIELD, BACKSLASH} 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 (strict && c < 0x20 && c != '\r' && c != '\n' && c != '\t')
      errx(1, "Control characters in input are not allowed in --rfc4180 mode");
    switch (state) {
    case START:
      switch (c) {
      case '\\': out('"'); state = BACKSLASH; break;
      case '\t': out(sep); state = TAB; break;
      case '"': outs("\"\"\""); state = IN_FIELD; break;
      case '\r': state = CR; break;
      case '\n': if (crlf) out('\r'); out('\n'); break;
      default: out('"'); out(c); state = IN_FIELD; break;
      }
      break;
    case TAB:
      switch (c) {
      case '\\': out('"'); state = BACKSLASH; break;
      case '\t': out(sep); break;
      case '"': outs("\"\"\""); state = IN_FIELD; break;
      case '\r': state = CR; break;
      case '\n': if (crlf) out('\r'); out('\n'); state = START; break;
      default: out('"'); out(c); state = IN_FIELD; break;
      }
      break;
    case CR:
      switch (c) {
      case '\\': out('\r'); if (crlf) out('\n'); out('"'); state = BACKSLASH; break;
      case '\t': out('\r'); if (crlf) out('\n'); out(sep); state = TAB; break;
      case '"': out('\r'); if (crlf) out('\n'); outs("\"\"\""); state = IN_FIELD; break;
      case '\r': out('\r'); if (crlf) out('\n'); state = CR; break;
      case '\n': outs("\r\n"); state = START; break;
      default: out('\r'); if (crlf) out('\n'); out('"'); out(c); state = IN_FIELD; break;
      }
      break;
    case IN_FIELD:
      switch (c) {
      case '\\': state = BACKSLASH; break;
      case '\t': out('"'); out(sep); state = TAB; break;
      case '"': outs("\"\""); break;
      case '\r': out('"'); state = CR; break;
      case '\n': out('"'); if (crlf) out('\r'); out('\n'); state = START; break;
      default: out(c); break;
      }
      break;
    case BACKSLASH:
      switch (c) {
      case 't':
	if (strict) errx(1, "Tab (\\t) is not allowed in --rfc4180 mode");
	out('\t'); state = IN_FIELD; break;
      case 'r': out('\r'); state = IN_FIELD; break;
      case 'n': out('\n'); state = IN_FIELD; break;
      case '\\': out('\\'); state = IN_FIELD; break;
      case '"': outs("\\\"\""); state = IN_FIELD; break;
      case '\n': outs("\\\""); if (crlf) out('\r'); out('\n'); state = START; break;
      case '\r': outs("\\\""); state = CR; break;
      case '\t': outs("\\\""); out(sep); state = TAB; break;
      default: out('\\'); out(c); state = IN_FIELD; break;
      }
      break;
    }
  }
  switch (state) {
  case START: break;
  case TAB: if (crlf) outs("\r\n"); break;
  case CR: out('\r'); if (crlf) out('\n'); break;
  case IN_FIELD: out('"'); if (crlf) outs("\r\n"); break;
  case BACKSLASH: out('\\'); if (crlf) outs("\r\n"); break;
  }
}


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

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

  if (strict && sep != ',')
    errx(1, "--rfc4180 and --separator cannot be used together.");
  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;
}