src/paste.c

   1 /* paste - merge lines of files
   2    Copyright (C) 1997-2015 Free Software Foundation, Inc.
   3    Copyright (C) 1984 David M. Ihnat
   4
   5    This program is free software: you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation, either version 3 of the License, or
   8    (at your option) any later version.
   9
  10    This program is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13    GNU General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
  17
  18 /* Written by David Ihnat.  */
  19
  20 /* The list of valid escape sequences has been expanded over the Unix
  21    version, to include \b, \f, \r, and \v.
  22
  23    POSIX changes, bug fixes, long-named options, and cleanup
  24    by David MacKenzie <djm@gnu.ai.mit.edu>.
  25
  26    Options:
  27    --serial
  28    -s                           Paste one file at a time rather than
  29                                 one line from each file.
  30    --delimiters=delim-list
  31    -d delim-list                Consecutively use the characters in
  32                                 DELIM-LIST instead of tab to separate
  33                                 merged lines.  When DELIM-LIST is exhausted,
  34                                 start again at its beginning.
  35    A FILE of '-' means standard input.
  36    If no FILEs are given, standard input is used. */
  37
  38 #include <config.h>
  39
  40 #include <stdio.h>
  41 #include <getopt.h>
  42 #include <sys/types.h>
  43 #include "system.h"
  44 #include "error.h"
  45 #include "fadvise.h"
  46
  47 /* The official name of this program (e.g., no 'g' prefix).  */
  48 #define PROGRAM_NAME "paste"
  49
  50 #define AUTHORS \
  51   proper_name ("David M. Ihnat"), \
  52   proper_name ("David MacKenzie")
  53
  54 /* Indicates that no delimiter should be added in the current position. */
  55 #define EMPTY_DELIM '\0'
  56
  57 /* If nonzero, we have read standard input at some point. */
  58 static bool have_read_stdin;
  59
  60 /* If nonzero, merge subsequent lines of each file rather than
  61    corresponding lines from each file in parallel. */
  62 static bool serial_merge;
  63
  64 /* The delimiters between lines of input files (used cyclically). */
  65 static char *delims;
  66
  67 /* A pointer to the character after the end of 'delims'. */
  68 static char const *delim_end;
  69
  70 static struct option const longopts[] =
  71 {
  72   {"serial", no_argument, NULL, 's'},
  73   {"delimiters", required_argument, NULL, 'd'},
  74   {GETOPT_HELP_OPTION_DECL},
  75   {GETOPT_VERSION_OPTION_DECL},
  76   {NULL, 0, NULL, 0}
  77 };
  78
  79 /* Set globals delims and delim_end.  Copy STRPTR to DELIMS, converting
  80    backslash representations of special characters in STRPTR to their actual
  81    values. The set of possible backslash characters has been expanded beyond
  82    that recognized by the Unix version.
  83    Return 0 upon success.
  84    If the string ends in an odd number of backslashes, ignore the
  85    final backslash and return nonzero.  */
  86
  87 static int
  88 collapse_escapes (char const *strptr)
  89 {
  90   char *strout = xstrdup (strptr);
  91   bool backslash_at_end = false;
  92
  93   delims = strout;
  94
  95   while (*strptr)
  96     {
  97       if (*strptr != '\\')      /* Is it an escape character? */
  98         *strout++ = *strptr++;  /* No, just transfer it. */
  99       else
 100         {
 101           switch (*++strptr)
 102             {
 103             case '0':
 104               *strout++ = EMPTY_DELIM;
 105               break;
 106
 107             case 'b':
 108               *strout++ = '\b';
 109               break;
 110
 111             case 'f':
 112               *strout++ = '\f';
 113               break;
 114
 115             case 'n':
 116               *strout++ = '\n';
 117               break;
 118
 119             case 'r':
 120               *strout++ = '\r';
 121               break;
 122
 123             case 't':
 124               *strout++ = '\t';
 125               break;
 126
 127             case 'v':
 128               *strout++ = '\v';
 129               break;
 130
 131             case '\\':
 132               *strout++ = '\\';
 133               break;
 134
 135             case '\0':
 136               backslash_at_end = true;
 137               goto done;
 138
 139             default:
 140               *strout++ = *strptr;
 141               break;
 142             }
 143           strptr++;
 144         }
 145     }
 146
 147  done:
 148
 149   delim_end = strout;
 150   return backslash_at_end ? 1 : 0;
 151 }
 152
 153 /* Report a write error and exit.  */
 154
 155 static void write_error (void) ATTRIBUTE_NORETURN;
 156 static void
 157 write_error (void)
 158 {
 159   error (EXIT_FAILURE, errno, _("write error"));
 160   abort ();
 161 }
 162
 163 /* Output a single byte, reporting any write errors.  */
 164
 165 static inline void
 166 xputchar (char c)
 167 {
 168   if (putchar (c) < 0)
 169     write_error ();
 170 }
 171
 172 /* Perform column paste on the NFILES files named in FNAMPTR.
 173    Return true if successful, false if one or more files could not be
 174    opened or read. */
 175
 176 static bool
 177 paste_parallel (size_t nfiles, char **fnamptr)
 178 {
 179   bool ok = true;
 180   /* If all files are just ready to be closed, or will be on this
 181      round, the string of delimiters must be preserved.
 182      delbuf[0] through delbuf[nfiles]
 183      store the delimiters for closed files. */
 184   char *delbuf = xmalloc (nfiles + 2);
 185
 186   /* Streams open to the files to process; NULL if the corresponding
 187      stream is closed.  */
 188   FILE **fileptr = xnmalloc (nfiles + 1, sizeof *fileptr);
 189
 190   /* Number of files still open to process.  */
 191   size_t files_open;
 192
 193   /* True if any fopen got fd == STDIN_FILENO.  */
 194   bool opened_stdin = false;
 195
 196   /* Attempt to open all files.  This could be expanded to an infinite
 197      number of files, but at the (considerable) expense of remembering
 198      each file and its current offset, then opening/reading/closing.  */
 199
 200   for (files_open = 0; files_open < nfiles; ++files_open)
 201     {
 202       if (STREQ (fnamptr[files_open], "-"))
 203         {
 204           have_read_stdin = true;
 205           fileptr[files_open] = stdin;
 206         }
 207       else
 208         {
 209           fileptr[files_open] = fopen (fnamptr[files_open], "r");
 210           if (fileptr[files_open] == NULL)
 211             error (EXIT_FAILURE, errno, "%s", quotef (fnamptr[files_open]));
 212           else if (fileno (fileptr[files_open]) == STDIN_FILENO)
 213             opened_stdin = true;
 214           fadvise (fileptr[files_open], FADVISE_SEQUENTIAL);
 215         }
 216     }
 217
 218   if (opened_stdin && have_read_stdin)
 219     error (EXIT_FAILURE, 0, _("standard input is closed"));
 220
 221   /* Read a line from each file and output it to stdout separated by a
 222      delimiter, until we go through the loop without successfully
 223      reading from any of the files. */
 224
 225   while (files_open)
 226     {
 227       /* Set up for the next line. */
 228       bool somedone = false;
 229       char const *delimptr = delims;
 230       size_t delims_saved = 0;  /* Number of delims saved in 'delbuf'. */
 231       size_t i;
 232
 233       for (i = 0; i < nfiles && files_open; i++)
 234         {
 235           int chr IF_LINT ( = 0);       /* Input character. */
 236           int err IF_LINT ( = 0);       /* Input errno value.  */
 237           bool sometodo = false;        /* Input chars to process.  */
 238
 239           if (fileptr[i])
 240             {
 241               chr = getc (fileptr[i]);
 242               err = errno;
 243               if (chr != EOF && delims_saved)
 244                 {
 245                   if (fwrite (delbuf, 1, delims_saved, stdout) != delims_saved)
 246                     write_error ();
 247                   delims_saved = 0;
 248                 }
 249
 250               while (chr != EOF)
 251                 {
 252                   sometodo = true;
 253                   if (chr == '\n')
 254                     break;
 255                   xputchar (chr);
 256                   chr = getc (fileptr[i]);
 257                   err = errno;
 258                 }
 259             }
 260
 261           if (! sometodo)
 262             {
 263               /* EOF, read error, or closed file.
 264                  If an EOF or error, close the file.  */
 265               if (fileptr[i])
 266                 {
 267                   if (ferror (fileptr[i]))
 268                     {
 269                       error (0, err, "%s", quotef (fnamptr[i]));
 270                       ok = false;
 271                     }
 272                   if (fileptr[i] == stdin)
 273                     clearerr (fileptr[i]); /* Also clear EOF. */
 274                   else if (fclose (fileptr[i]) == EOF)
 275                     {
 276                       error (0, errno, "%s", quotef (fnamptr[i]));
 277                       ok = false;
 278                     }
 279
 280                   fileptr[i] = NULL;
 281                   files_open--;
 282                 }
 283
 284               if (i + 1 == nfiles)
 285                 {
 286                   /* End of this output line.
 287                      Is this the end of the whole thing? */
 288                   if (somedone)
 289                     {
 290                       /* No.  Some files were not closed for this line. */
 291                       if (delims_saved)
 292                         {
 293                           if (fwrite (delbuf, 1, delims_saved, stdout)
 294                               != delims_saved)
 295                             write_error ();
 296                           delims_saved = 0;
 297                         }
 298                       xputchar ('\n');
 299                     }
 300                   continue;     /* Next read of files, or exit. */
 301                 }
 302               else
 303                 {
 304                   /* Closed file; add delimiter to 'delbuf'. */
 305                   if (*delimptr != EMPTY_DELIM)
 306                     delbuf[delims_saved++] = *delimptr;
 307                   if (++delimptr == delim_end)
 308                     delimptr = delims;
 309                 }
 310             }
 311           else
 312             {
 313               /* Some data read. */
 314               somedone = true;
 315
 316               /* Except for last file, replace last newline with delim. */
 317               if (i + 1 != nfiles)
 318                 {
 319                   if (chr != '\n' && chr != EOF)
 320                     xputchar (chr);
 321                   if (*delimptr != EMPTY_DELIM)
 322                     xputchar (*delimptr);
 323                   if (++delimptr == delim_end)
 324                     delimptr = delims;
 325                 }
 326               else
 327                 {
 328                   /* If the last line of the last file lacks a newline,
 329                      print one anyhow.  POSIX requires this.  */
 330                   char c = (chr == EOF ? '\n' : chr);
 331                   xputchar (c);
 332                 }
 333             }
 334         }
 335     }
 336   free (fileptr);
 337   free (delbuf);
 338   return ok;
 339 }
 340
 341 /* Perform serial paste on the NFILES files named in FNAMPTR.
 342    Return true if no errors, false if one or more files could not be
 343    opened or read. */
 344
 345 static bool
 346 paste_serial (size_t nfiles, char **fnamptr)
 347 {
 348   bool ok = true;       /* false if open or read errors occur. */
 349   int charnew, charold; /* Current and previous char read. */
 350   char const *delimptr; /* Current delimiter char. */
 351   FILE *fileptr;        /* Open for reading current file. */
 352
 353   for (; nfiles; nfiles--, fnamptr++)
 354     {
 355       int saved_errno;
 356       bool is_stdin = STREQ (*fnamptr, "-");
 357       if (is_stdin)
 358         {
 359           have_read_stdin = true;
 360           fileptr = stdin;
 361         }
 362       else
 363         {
 364           fileptr = fopen (*fnamptr, "r");
 365           if (fileptr == NULL)
 366             {
 367               error (0, errno, "%s", quotef (*fnamptr));
 368               ok = false;
 369               continue;
 370             }
 371           fadvise (fileptr, FADVISE_SEQUENTIAL);
 372         }
 373
 374       delimptr = delims;        /* Set up for delimiter string. */
 375
 376       charold = getc (fileptr);
 377       saved_errno = errno;
 378       if (charold != EOF)
 379         {
 380           /* 'charold' is set up.  Hit it!
 381              Keep reading characters, stashing them in 'charnew';
 382              output 'charold', converting to the appropriate delimiter
 383              character if needed.  After the EOF, output 'charold'
 384              if it's a newline; otherwise, output it and then a newline. */
 385
 386           while ((charnew = getc (fileptr)) != EOF)
 387             {
 388               /* Process the old character. */
 389               if (charold == '\n')
 390                 {
 391                   if (*delimptr != EMPTY_DELIM)
 392                     xputchar (*delimptr);
 393
 394                   if (++delimptr == delim_end)
 395                     delimptr = delims;
 396                 }
 397               else
 398                 xputchar (charold);
 399
 400               charold = charnew;
 401             }
 402           saved_errno = errno;
 403
 404           /* Hit EOF.  Process that last character. */
 405           xputchar (charold);
 406         }
 407
 408       if (charold != '\n')
 409         xputchar ('\n');
 410
 411       if (ferror (fileptr))
 412         {
 413           error (0, saved_errno, "%s", quotef (*fnamptr));
 414           ok = false;
 415         }
 416       if (is_stdin)
 417         clearerr (fileptr);     /* Also clear EOF. */
 418       else if (fclose (fileptr) == EOF)
 419         {
 420           error (0, errno, "%s", quotef (*fnamptr));
 421           ok = false;
 422         }
 423     }
 424   return ok;
 425 }
 426
 427 void
 428 usage (int status)
 429 {
 430   if (status != EXIT_SUCCESS)
 431     emit_try_help ();
 432   else
 433     {
 434       printf (_("\
 435 Usage: %s [OPTION]... [FILE]...\n\
 436 "),
 437               program_name);
 438       fputs (_("\
 439 Write lines consisting of the sequentially corresponding lines from\n\
 440 each FILE, separated by TABs, to standard output.\n\
 441 "), stdout);
 442
 443       emit_stdin_note ();
 444       emit_mandatory_arg_note ();
 445
 446       fputs (_("\
 447   -d, --delimiters=LIST   reuse characters from LIST instead of TABs\n\
 448   -s, --serial            paste one file at a time instead of in parallel\n\
 449 "), stdout);
 450       fputs (HELP_OPTION_DESCRIPTION, stdout);
 451       fputs (VERSION_OPTION_DESCRIPTION, stdout);
 452       /* FIXME: add a couple of examples.  */
 453       emit_ancillary_info (PROGRAM_NAME);
 454     }
 455   exit (status);
 456 }
 457
 458 int
 459 main (int argc, char **argv)
 460 {
 461   int optc;
 462   bool ok;
 463   char const *delim_arg = "\t";
 464
 465   initialize_main (&argc, &argv);
 466   set_program_name (argv[0]);
 467   setlocale (LC_ALL, "");
 468   bindtextdomain (PACKAGE, LOCALEDIR);
 469   textdomain (PACKAGE);
 470
 471   atexit (close_stdout);
 472
 473   have_read_stdin = false;
 474   serial_merge = false;
 475
 476   while ((optc = getopt_long (argc, argv, "d:s", longopts, NULL)) != -1)
 477     {
 478       switch (optc)
 479         {
 480         case 'd':
 481           /* Delimiter character(s). */
 482           delim_arg = (optarg[0] == '\0' ? "\\0" : optarg);
 483           break;
 484
 485         case 's':
 486           serial_merge = true;
 487           break;
 488
 489         case_GETOPT_HELP_CHAR;
 490
 491         case_GETOPT_VERSION_CHAR (PROGRAM_NAME, AUTHORS);
 492
 493         default:
 494           usage (EXIT_FAILURE);
 495         }
 496     }
 497
 498   if (optind == argc)
 499     argv[argc++] = bad_cast ("-");
 500
 501   if (collapse_escapes (delim_arg))
 502     {
 503       /* Don't use the quote() quoting style, because that would double the
 504          number of displayed backslashes, making the diagnostic look bogus.  */
 505       error (EXIT_FAILURE, 0,
 506              _("delimiter list ends with an unescaped backslash: %s"),
 507              quotearg_n_style_colon (0, c_maybe_quoting_style, delim_arg));
 508     }
 509
 510   if (!serial_merge)
 511     ok = paste_parallel (argc - optind, &argv[optind]);
 512   else
 513     ok = paste_serial (argc - optind, &argv[optind]);
 514
 515   free (delims);
 516
 517   if (have_read_stdin && fclose (stdin) == EOF)
 518     error (EXIT_FAILURE, errno, "-");
 519   return ok ? EXIT_SUCCESS : EXIT_FAILURE;
 520 }