1 /* paste - merge lines of files
2 Copyright (C) 1984 by David M. Ihnat
4 This program is a total rewrite of the Bell Laboratories Unix(Tm)
5 command of the same name, as of System V. It contains no proprietary
6 code, and therefore may be used without violation of any proprietary
7 agreements whatsoever. However, you will notice that the program is
8 copyrighted by me. This is to assure the program does *not* fall
9 into the public domain. Thus, I may specify just what I am now:
10 This program may be freely copied and distributed, provided this notice
11 remains; it may not be sold for profit without express written consent of
13 Please note that I recreated the behavior of the Unix(Tm) 'paste' command
14 as faithfully as possible, with minor exceptions; however,
15 I haven't run a full set of regression tests. Thus, the user of
16 this program accepts full responsibility for any effects or loss;
17 in particular, the author is not responsible for any losses,
18 explicit or incidental, that may be incurred through use of this program.
20 I ask that any bugs (and, if possible, fixes) be reported to me when
21 possible. -David Ihnat (312) 784-4544 ignatz@homebru.chi.il.us
23 The list of valid escape sequences has been expanded over the Unix
24 version, to include \b, \f, \r, and \v.
26 POSIX changes, bug fixes, long-named options, and cleanup
27 by David MacKenzie <djm@gnu.ai.mit.edu>.
31 -s Paste one file at a time rather than
32 one line from each file.
33 --delimiters=delim-list
34 -d delim-list Consecutively use the characters in
35 DELIM-LIST instead of tab to separate
36 merged lines. When DELIM-LIST is exhausted,
37 start again at its beginning.
38 A FILE of `-' means standard input.
39 If no FILEs are given, standard input is used. */
45 #include <sys/types.h>
53 static char *collapse_escapes ();
54 static int paste_parallel ();
55 static int paste_serial ();
58 /* Indicates that no delimiter should be added in the current position. */
59 #define EMPTY_DELIM '\0'
61 /* Element marking a file that has reached EOF and been closed. */
62 #define CLOSED ((FILE *) -1)
64 /* Element marking end of list of open files. */
65 #define ENDLIST ((FILE *) -2)
67 /* Name this program was run with. */
70 /* If nonzero, we have read standard input at some point. */
71 static int have_read_stdin;
73 /* If nonzero, merge subsequent lines of each file rather than
74 corresponding lines from each file in parallel. */
75 static int serial_merge;
77 /* The delimeters between lines of input files (used cyclically). */
80 /* A pointer to the character after the end of `delims'. */
81 static char *delim_end;
83 /* If non-zero, display usage information and exit. */
86 /* If non-zero, print the version on standard output then exit. */
87 static int show_version;
89 static struct option const longopts[] =
91 {"serial", no_argument, 0, 's'},
92 {"delimiters", required_argument, 0, 'd'},
93 {"help", no_argument, &show_help, 1},
94 {"version", no_argument, &show_version, 1},
103 int optc, exit_status;
104 char default_delims[2], zero_delims[3];
106 program_name = argv[0];
109 delims = default_delims;
110 strcpy (delims, "\t");
111 strcpy (zero_delims, "\\0");
113 while ((optc = getopt_long (argc, argv, "d:s", longopts, (int *) 0))
122 /* Delimiter character(s). */
123 if (optarg[0] == '\0')
124 optarg = zero_delims;
139 printf ("paste - %s\n", version_string);
149 delim_end = collapse_escapes (delims);
152 exit_status = paste_parallel (argc - optind, &argv[optind]);
154 exit_status = paste_serial (argc - optind, &argv[optind]);
155 if (have_read_stdin && fclose (stdin) == EOF)
156 error (1, errno, "-");
157 if (ferror (stdout) || fclose (stdout) == EOF)
158 error (1, errno, "write error");
162 /* Replace backslash representations of special characters in
163 STRPTR with their actual values.
164 The set of possible backslash characters has been expanded beyond
165 that recognized by the Unix version.
167 Return a pointer to the character after the new end of STRPTR. */
170 collapse_escapes (strptr)
173 register char *strout;
175 strout = strptr; /* Start at the same place, anyway. */
179 if (*strptr != '\\') /* Is it an escape character? */
180 *strout++ = *strptr++; /* No, just transfer it. */
186 *strout++ = EMPTY_DELIM;
223 /* Perform column paste on the NFILES files named in FNAMPTR.
224 Return 0 if no errors, 1 if one or more files could not be
228 paste_parallel (nfiles, fnamptr)
232 int errors = 0; /* 1 if open or read errors occur. */
233 /* Number of files for which space is allocated in `delbuf' and `fileptr'.
234 Enlarged as necessary. */
235 int file_list_size = 12;
236 int chr; /* Input character. */
237 int line_length; /* Number of chars in line. */
238 int somedone; /* 0 if all files empty for this line. */
239 /* If all files are just ready to be closed, or will be on this
240 round, the string of delimiters must be preserved.
241 delbuf[0] through delbuf[file_list_size]
242 store the delimiters for closed files. */
244 int delims_saved; /* Number of delims saved in `delbuf'. */
245 register char *delimptr; /* Cycling pointer into `delims'. */
246 FILE **fileptr; /* Streams open to the files to process. */
247 int files_open; /* Number of files still open to process. */
248 int i; /* Loop index. */
249 int opened_stdin = 0; /* Nonzero if any fopen got fd 0. */
251 #ifdef lint /* Suppress `used before initialized' warning. */
255 delbuf = (char *) xmalloc (file_list_size + 2);
256 fileptr = (FILE **) xmalloc ((file_list_size + 1) * sizeof (FILE *));
258 /* Attempt to open all files. This could be expanded to an infinite
259 number of files, but at the (considerable) expense of remembering
260 each file and its current offset, then opening/reading/closing. */
262 for (files_open = 0; files_open < nfiles; ++files_open)
264 if (files_open == file_list_size - 2)
266 file_list_size += 12;
267 delbuf = (char *) xrealloc (delbuf, file_list_size + 2);
268 fileptr = (FILE **) xrealloc (fileptr, (file_list_size + 1)
271 if (!strcmp (fnamptr[files_open], "-"))
274 fileptr[files_open] = stdin;
278 fileptr[files_open] = fopen (fnamptr[files_open], "r");
279 if (fileptr[files_open] == NULL)
280 error (1, errno, "%s", fnamptr[files_open]);
281 else if (fileno (fileptr[files_open]) == 0)
286 fileptr[files_open] = ENDLIST;
288 if (opened_stdin && have_read_stdin)
289 error (1, 0, "standard input is closed");
291 /* Read a line from each file and output it to stdout separated by a
292 delimiter, until we go through the loop without successfully
293 reading from any of the files. */
297 /* Set up for the next line. */
302 for (i = 0; fileptr[i] != ENDLIST && files_open; i++)
304 line_length = 0; /* Clear so we can easily detect EOF. */
305 if (fileptr[i] != CLOSED)
307 chr = getc (fileptr[i]);
308 if (chr != EOF && delims_saved)
310 fwrite (delbuf, sizeof (char), delims_saved, stdout);
320 chr = getc (fileptr[i]);
324 if (line_length == 0)
326 /* EOF, read error, or closed file.
327 If an EOF or error, close the file and mark it in the list. */
328 if (fileptr[i] != CLOSED)
330 if (ferror (fileptr[i]))
332 error (0, errno, "%s", fnamptr[i]);
335 if (fileptr[i] == stdin)
336 clearerr (fileptr[i]); /* Also clear EOF. */
337 else if (fclose (fileptr[i]) == EOF)
339 error (0, errno, "%s", fnamptr[i]);
347 if (fileptr[i + 1] == ENDLIST)
349 /* End of this output line.
350 Is this the end of the whole thing? */
353 /* No. Some files were not closed for this line. */
356 fwrite (delbuf, sizeof (char), delims_saved, stdout);
361 continue; /* Next read of files, or exit. */
365 /* Closed file; add delimiter to `delbuf'. */
366 if (*delimptr != EMPTY_DELIM)
367 delbuf[delims_saved++] = *delimptr;
368 if (++delimptr == delim_end)
374 /* Some data read. */
377 /* Except for last file, replace last newline with delim. */
378 if (fileptr[i + 1] != ENDLIST)
382 if (*delimptr != EMPTY_DELIM)
383 putc (*delimptr, stdout);
384 if (++delimptr == delim_end)
395 /* Perform serial paste on the NFILES files named in FNAMPTR.
396 Return 0 if no errors, 1 if one or more files could not be
400 paste_serial (nfiles, fnamptr)
404 int errors = 0; /* 1 if open or read errors occur. */
405 register int charnew, charold; /* Current and previous char read. */
406 register char *delimptr; /* Current delimiter char. */
407 register FILE *fileptr; /* Open for reading current file. */
409 for (; nfiles; nfiles--, fnamptr++)
411 if (!strcmp (*fnamptr, "-"))
418 fileptr = fopen (*fnamptr, "r");
421 error (0, errno, "%s", *fnamptr);
427 delimptr = delims; /* Set up for delimiter string. */
429 charold = getc (fileptr);
432 /* `charold' is set up. Hit it!
433 Keep reading characters, stashing them in `charnew';
434 output `charold', converting to the appropriate delimiter
435 character if needed. After the EOF, output `charold'
436 if it's a newline; otherwise, output it and then a newline. */
438 while ((charnew = getc (fileptr)) != EOF)
440 /* Process the old character. */
443 if (*delimptr != EMPTY_DELIM)
444 putc (*delimptr, stdout);
446 if (++delimptr == delim_end)
450 putc (charold, stdout);
455 /* Hit EOF. Process that last character. */
456 putc (charold, stdout);
462 if (ferror (fileptr))
464 error (0, errno, "%s", *fnamptr);
467 if (fileptr == stdin)
468 clearerr (fileptr); /* Also clear EOF. */
469 else if (fclose (fileptr) == EOF)
471 error (0, errno, "%s", *fnamptr);
483 fprintf (stderr, "Try `%s --help' for more information.\n",
488 Usage: %s [OPTION]... [FILE]...\n\
493 -d, --delimiters=LIST reuse characters from LIST instead of TABs\n\
494 -s, --serial paste one file at a time instead of in parallel\n\
495 --help display this help and exit\n\
496 --version output version information and exit\n\
498 With no INPUT, or when INPUT is -, read standard input.\n\