commands/paste/paste.c

   1 /* paste - laminate files               Author: David Ihnat */
   2
   3 /* Paste - a recreation of the Unix(Tm) paste(1) command.
   4  *
   5  * syntax: paste file1 file2 ... paste -dLIST file1 file2 ... paste -s [-dLIST]
   6  * file1 file2 ...
   7  *
   8  * Copyright (C) 1984 by David M. Ihnat
   9  *
  10  * This program is a total rewrite of the Bell Laboratories Unix(Tm) command of
  11  * the same name, as of System V.  It contains no proprietary code, and
  12  * therefore may be used without violation of any proprietary agreements
  13  * whatsoever.  However, you will notice that the program is copyrighted by
  14  * me.  This is to assure the program does *not* fall into the public domain.
  15  * Thus, I may specify just what I am now: This program may be freely copied
  16  * and distributed, provided this notice remains; it may not be sold for
  17  * profit without express written consent of the author. Please note that I
  18  * recreated the behavior of the Unix(Tm) 'paste' command as faithfully as
  19  * possible, with minor exceptions (noted below); however, I haven't run a
  20  * full set of regression * tests.  Thus, the user of this program accepts
  21  * full responsibility for any effects or loss; in particular, the author is
  22  * not responsible for any losses, explicit or incidental, that may be
  23  * incurred through use of this program.
  24  *
  25  * The changes to the program, with one exception, are transparent to a user
  26  * familiar with the Unix command of the same name.  These changes are:
  27  *
  28  * 1) The '-s' option had a bug in the Unix version when used with multiple
  29  *    files.  (It would repeat each file in a list, i.e., for
  30  *
  31  *              paste -s file1 file2 file3
  32  *
  33  * it would list
  34  *
  35  *       <file1\n><file1\n><file2\n><file1\n><file2\n><file3\n>
  36  *
  37  * I fixed this, and reported the bug to the providers of the command in Unix.
  38  *
  39  * 2) The list of valid escape sequences has been expanded to include \b,\f,
  40  *    and \r.  (Just because *I* can't imagine why you'd want to use them
  41  *    doesn't mean I should keep them from you.)
  42  *
  43  * 3) There is no longer any restriction on line length.
  44  *
  45  * I ask that any bugs (and, if possible, fixes) be reported to me when
  46  * possible.  -David Ihnat (312) 784-4544 ihuxx!ignatz
  47  */
  48
  49 /* Modified to run under MINIX 1.1 by David O. Tinker  (416) 978-3636
  50  * (utgpu!dtinker) Sept. 19, 1987
  51  */
  52
  53 /* Modified to conform to POSIX 1003.2/Draft10 standard 23rd Sept. 1990
  54  * Changes:
  55  *      - the arguments can be in any order
  56  *      - removed the ToUpper function
  57  * by Thomas Brupbacher (tobr@mw.lpc.ethz.ch)
  58  */
  59
  60 #include <errno.h>
  61 #include <ctype.h>
  62 #include <stdlib.h>
  63 #include <string.h>
  64 #include <stdio.h>
  65
  66 /* I'd love to use enums, but not everyone has them.  Portability, y'know. */
  67 #define NODELIM         1
  68 #define USAGE           2
  69 #define BADFILE         3
  70 #define TOOMANY         4
  71
  72 #define TAB             '\t'
  73 #define NL              '\n'
  74 #define BS              '\b'
  75 #define FF              '\f'
  76 #define CR              '\r'
  77 #define DEL             '\177'
  78 #define SPACE           ' '
  79 #define BACKSLASH       '\\'
  80
  81 #define _MAXSZ          512
  82 #define _MAXFILES       12
  83 #define CLOSED          ((FILE *)-1)
  84 #define ENDLIST         ((FILE *)-2)
  85
  86 char *cmdnam;
  87
  88 short int sflag;
  89 static char default_delims[] = {TAB};   /* default delimiter string      */
  90 char *delims;                   /* the pointer to the delimiters        */
  91 int number_of_delims = 1;       /* number of delimiters to use           */
  92
  93 int main(int argc, char **argv);
  94 void docol(int nfiles, char **fnamptr);
  95 void doserial(int nfiles, char **fnamptr);
  96 void delimbuild(char *strptr);
  97 void prerr(int etype, char *estring);
  98
  99 int main(argc, argv)
 100 int argc;
 101 char **argv;
 102 {
 103   char **arg_ptr;               /* used to save argv, needed for docol() etc */
 104   int num_files = 0;            /* Number of filenames specified on cmd line */
 105   sflag = 0;
 106   delims = default_delims;      /* use default delimiters */
 107
 108   cmdnam = *argv;
 109
 110   if (argc >= 2) {
 111
 112         /* Skip invocation name */
 113         argv++;
 114         argc--;
 115
 116         /* Save argv */
 117         arg_ptr = argv;
 118         /* First, parse input options */
 119
 120         while (argc-- > 0) {
 121                 if (argv[0][0] == '-' && argv[0][1] != '\0') {
 122                         switch (argv[0][1]) {
 123                             case 'd':
 124                                 /* Delimiter character(s) */
 125                                 if (*(++argv) == '\0')
 126                                         prerr(NODELIM, "");
 127                                 else
 128                                         delimbuild(*(argv));
 129                                 argc--;
 130                                 break;
 131
 132                             case 's':   sflag++;        break;
 133
 134                             default:    prerr(USAGE, "");
 135                         }
 136                         argv++;
 137                 } else {
 138                         num_files++;
 139                         argv++;
 140                 }
 141         }
 142
 143         /* If there are more than MAX_FILES files on the command
 144          * line, exit with error message.                                                */
 145         if (num_files > _MAXFILES) prerr(TOOMANY, "");
 146
 147         /* If no files specified, simply exit.  Otherwise, if not the
 148          * old '-s' option, process all files. If '-s', then process
 149          * files one-at-a-time. */
 150
 151         if (!sflag)
 152                 docol(num_files, arg_ptr);      /* Column paste */
 153         else
 154                 doserial(num_files, arg_ptr);   /* Serial paste */
 155
 156         exit(0);
 157   } else
 158         prerr(USAGE, "");
 159   return(0);
 160 }
 161
 162 void docol(nfiles, fnamptr)
 163 int nfiles;
 164 char **fnamptr;
 165 {
 166   char iobuff[_MAXSZ];          /* i/o buffer for the fgets */
 167   short int somedone;           /* flag for blank field handling */
 168
 169   /* There is a strange case where all files are just ready to be
 170    * closed, or will on this round.  In that case, the string of
 171    * delimiters must be preserved.  delbuf[1] ->delbuf[MAXFILES+1]
 172    * provides intermediate storage for closed files, if needed;
 173    * delbuf[0] is the current index.
 174    */
 175   char delbuf[_MAXFILES + 2];
 176
 177   FILE *fileptr[_MAXFILES + 1];
 178
 179   int filecnt;                  /* Set to number of files to process */
 180   register char *delimptr;      /* Cycling delimiter pointer */
 181   int index;                    /* Working variable */
 182   int strend;                   /* End of string in buffer */
 183
 184   /* Perform column paste.  First, attempt to open all files. (This
 185    * could be expanded to an infinite number of files, but at the
 186    * (considerable) expense of remembering the file and its current
 187    * offset, then opening/reading/closing.  The commands' utility
 188    * doesn't warrant the effort; at least, to me...)
 189    */
 190
 191   for (filecnt = 0; (nfiles > 0); fnamptr++) {
 192         if ((fnamptr[0][0] == '-') && (fnamptr[0][1] != '\0')) {
 193                 if (fnamptr[0][1] == 'd') fnamptr++;
 194         } else {
 195                 nfiles--;
 196                 if (fnamptr[0][0] == '-') {
 197                         fileptr[filecnt++] = stdin;
 198                 } else {
 199                         fileptr[filecnt] = fopen(fnamptr[0], "r");
 200                         if (fileptr[filecnt++] == NULL)
 201                                 prerr(BADFILE, *fnamptr);
 202                 }
 203         }
 204   }
 205
 206   fileptr[filecnt] = ENDLIST;   /* End of list. */
 207
 208   /* Have all files.  Now, read a line from each file, and output to
 209    * stdout.  Notice that the old 511 character limitation on the line
 210    * length no longer applies, since this program doesn't do the
 211    * buffering.  Do this until you go through the loop and don't
 212    * successfully read from any of the files.
 213    */
 214   for (; filecnt;) {
 215         somedone = 0;           /* Blank field handling flag */
 216         delimptr = delims;      /* Start at beginning of delim list */
 217         delbuf[0] = 0;          /* No squirreled delims */
 218
 219         for (index = 0; (fileptr[index] != ENDLIST) && filecnt; index++) {
 220                 /* Read a line and immediately output. If it's too
 221                  * big for the buffer, then dump what was read and go
 222                  * back for more.
 223                  *
 224                  * Otherwise, if it is from the last file, then leave
 225                  * the carriage return in place; if not, replace with
 226                  * a delimiter (if any)
 227                  */
 228
 229                 strend = 0;     /* Set so can easily detect EOF */
 230
 231                 if (fileptr[index] != CLOSED)
 232                         while (fgets(iobuff, (_MAXSZ - 1),
 233                                         fileptr[index]) != NULL) {
 234                                 strend = strlen(iobuff);/* Did the buf fill? */
 235
 236                                 if (strend == (_MAXSZ - 1)) {
 237                                         /* Gosh, what a long line. */
 238                                         fputs(iobuff, stdout);
 239                                         strend = 0;
 240                                         continue;
 241                                 }
 242
 243                                 /* Ok got whole line in buffer. */
 244                                 break;  /* Out of loop for this file */
 245                         }
 246
 247                 /* Ended either on an EOF (well, actually NULL
 248                  * return-- it *could* be some sort of file error,
 249                  * but but if the file was opened successfully, this
 250                  * is unlikely. Besides, error checking on streams
 251                  * doesn't allow us to decide exactly what went
 252                  * wrong, so I'm going to be very Unix-like and
 253                  * ignore it!), or a closed file, or a received line.
 254                  * If an EOF, close the file and mark it in the list.
 255                  * In any case, output the delimiter of choice.
 256                  */
 257
 258                 if (!strend) {
 259                         if (fileptr[index] != CLOSED) {
 260                                 fclose(fileptr[index]);
 261                                 fileptr[index] = CLOSED;
 262                                 filecnt--;
 263                         }
 264
 265                         /* Is this the end of the whole thing? */
 266                         if ((fileptr[index + 1] == ENDLIST) && !somedone)
 267                                 continue;       /* EXITS */
 268
 269                         /* Ok, some files not closed this line. Last file? */
 270                         if (fileptr[index + 1] == ENDLIST) {
 271                                 if (delbuf[0]) {
 272                                         fputs(&delbuf[1], stdout);
 273                                         delbuf[0] = 0;
 274                                 }
 275                                 putc((int) NL, stdout);
 276                                 continue;       /* Next read of files */
 277                         } else {
 278                                 /* Closed file; setup delim */
 279                                 if (*delimptr != DEL) {
 280                                         delbuf[0]++;
 281                                         delbuf[delbuf[0]] = *delimptr++;
 282                                         delbuf[delbuf[0] + 1] = '\0';
 283                                 } else
 284                                         delimptr++;
 285                         }
 286
 287                         /* Reset end of delimiter string if necessary */
 288                         if (*delimptr == '\0') delimptr = delims;
 289                 } else {
 290                         /* Some data read. */
 291                         somedone++;
 292
 293                         /* Any saved delims? */
 294                         if (delbuf[0]) {
 295                                 fputs(&delbuf[1], stdout);
 296                                 delbuf[0] = 0;
 297                         }
 298
 299                         /* If last file, last char will be NL. */
 300                         if (fileptr[index + 1] != ENDLIST) {
 301                                 if (*delimptr == DEL) {
 302                                         delimptr++;
 303                                         iobuff[strend - 1] = '\0';/* No delim*/
 304                                 } else
 305                                         iobuff[strend - 1] = *delimptr++;
 306                         }
 307                         if (*delimptr == '\0') delimptr = delims;
 308
 309                         /* Now dump the buffer */
 310                         fputs(iobuff, stdout);
 311                         fflush(stdout);
 312                 }
 313         }
 314   }
 315 }
 316
 317 void doserial(nfiles, fnamptr)
 318 int nfiles;
 319 char **fnamptr;
 320 {
 321   /* Do serial paste.  Simply scarf characters, performing
 322    * one-character buffering to facilitate delim processing.
 323    */
 324
 325   register int charnew, charold;
 326   register char *delimptr;
 327
 328   register FILE *fileptr;
 329
 330   for (; nfiles != 0; fnamptr++) {
 331         if ((fnamptr[0][0] == '-') && (fnamptr[0][1] != '\0')) {
 332                 if (fnamptr[0][1] == 'd') fnamptr++;
 333         } else {
 334                 if (fnamptr[0][0] == '-') {
 335                         fileptr = stdin;
 336                 } else {
 337                         fileptr = fopen(*fnamptr, "r");
 338
 339                         if (fileptr == NULL) prerr(BADFILE, *fnamptr);
 340                 }
 341
 342                 /* The file is open; just keep taking characters,
 343                  * stashing them in charnew; output charold,
 344                  * converting to the appropriate delimiter character
 345                  * if needful.  After the EOF, simply output
 346                  * 'charold' if it's a newline; otherwise, output it
 347                  * and then a newline.
 348                  */
 349
 350                 delimptr = delims;      /* Set up for delimiter string */
 351
 352                 if ((charold = getc(fileptr)) == EOF) {
 353                         /* Empty file! */
 354                         putc(NL, stdout);
 355                         fflush(stdout);
 356                         continue;       /* Go on to the next file */
 357                 }
 358
 359                 /* Ok, 'charold' is set up.  Hit it! */
 360
 361                 while ((charnew = getc(fileptr)) != EOF) {
 362                         /* Ok, process the old character */
 363                         if (charold == NL) {
 364                                 if (*delimptr != DEL)
 365                                         putc((int) *delimptr++, stdout);
 366
 367                                 /* Reset pointer at end of delimiter string */
 368                                 if (*delimptr == '\0') delimptr = delims;
 369                         } else
 370                                 putc(charold, stdout);
 371
 372                         charold = charnew;
 373                 }
 374
 375                 /* Ok, hit EOF.  Process that last character */
 376
 377                 putc(charold, stdout);
 378                 if ((char) charold != NL) putc(NL, stdout);
 379                 fflush(stdout);
 380                 nfiles--;
 381         }
 382   }
 383 }
 384
 385 void delimbuild(strptr)
 386 char *strptr;
 387 {
 388   /* Process the delimiter string into something that can be used by
 389    * the routines.  This involves, primarily, collapsing the backslash
 390    * representations of special characters into their actual values,
 391    * and terminating the string in a manner that the routines can
 392    * recognize.  The set of possible backslash characters has been
 393    * expanded beyond that recognized by the vanilla Unix(Tm) version.
 394    */
 395
 396   register char *strout;
 397
 398   delims = strptr;              /* delims now points to argv[...]        */
 399   strout = strptr;              /* Start at the same place, anyway */
 400
 401   while (*strptr) {
 402         if (*strptr != '\\')    /* Is it an escape character? */
 403                 *strout++ = *strptr++;  /* No, just transfer it */
 404         else {
 405                 strptr++;       /* Get past escape character */
 406
 407                 switch (*strptr) {
 408                     case '0':   *strout++ = DEL;        break;
 409
 410                     case 't':   *strout++ = TAB;        break;
 411
 412                     case 'n':   *strout++ = NL; break;
 413
 414                     case 'b':   *strout++ = BS; break;
 415
 416                     case 'f':   *strout++ = FF; break;
 417
 418                     case 'r':   *strout++ = CR; break;
 419
 420                     case '\\':
 421                         *strout++ = BACKSLASH;
 422                         break;
 423
 424                     default:    *strout++ = *strptr;
 425                 }
 426
 427                 strptr++;
 428         }
 429
 430   }
 431   *strout = '\0';               /* Heaven forfend that we forget this! */
 432 }
 433
 434 void prerr(etype, estring)
 435 int etype;
 436 char *estring;
 437 {
 438   switch (etype) {
 439       case USAGE:
 440         fprintf(stderr, "%s : Usage: %s [-s] [-d <delimiters>] file1 file2 ...\n", cmdnam, cmdnam);
 441         break;
 442
 443       case NODELIM:
 444         fprintf(stderr, "%s : no delimiters\n", cmdnam);
 445         break;
 446
 447       case BADFILE:
 448         fprintf(stderr, "%s : %s : cannot open\n", cmdnam, estring);
 449         break;
 450
 451       case TOOMANY:
 452         fprintf(stderr, "%s : too many files\n", cmdnam);
 453         break;
 454   }
 455   exit(1);
 456 }