1 /* cut - remove parts of 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) 'cut' command
14 as faithfully as possible; however, I haven't run a full set of regression
15 tests. Thus, the user of this program accepts full responsibility for any
16 effects or loss; in particular, the author is not responsible for any losses,
17 explicit or incidental, that may be incurred through use of this program.
19 I ask that any bugs (and, if possible, fixes) be reported to me when
20 possible. -David Ihnat (312) 784-4544 ignatz@homebru.chi.il.us
22 POSIX changes, bug fixes, long-named options, and cleanup
23 by David MacKenzie <djm@gnu.ai.mit.edu>.
25 Rewrite cut_fields and cut_bytes -- Jim Meyering (meyering@comco.com).
29 -b byte-list Print only the bytes in positions listed
31 Tabs and backspaces are treated like any
32 other character; they take up 1 byte.
34 --characters=character-list
35 -c character-list Print only characters in positions listed
37 The same as -b for now, but
38 internationalization will change that.
39 Tabs and backspaces are treated like any
40 other character; they take up 1 character.
43 -f field-list Print only the fields listed in FIELD-LIST.
44 Fields are separated by a TAB by default.
47 -d delim For -f, fields are separated by the first
48 character in DELIM instead of TAB.
50 -n Do not split multibyte chars (no-op for now).
53 -s For -f, do not print lines that do not contain
54 the field separator character.
56 The BYTE-LIST, CHARACTER-LIST, and FIELD-LIST are one or more numbers
57 or ranges separated by commas. The first byte, character, and field
60 A FILE of `-' means standard input. */
64 /* Get isblank from GNU libc. */
73 #include <sys/types.h>
77 #define FATAL_ERROR(s) \
85 /* Append LOW, HIGH to the list RP of range pairs, allocating additional
86 space if necessary. Update local variable N_RP. When allocating,
87 update global variable N_RP_ALLOCATED. */
89 #define ADD_RANGE_PAIR(rp, low, high) \
92 if (n_rp >= n_rp_allocated) \
94 n_rp_allocated *= 2; \
95 (rp) = (struct range_pair *) xrealloc ((rp), \
96 n_rp_allocated * sizeof (*(rp))); \
98 rp[n_rp].lo = (low); \
99 rp[n_rp].hi = (high); \
113 /* This buffer is used to support the semantics of the -s option
114 (or lack of same) when the specified field list includes (does
115 not include) the first field. In both of those cases, the entire
116 first field must be read into this buffer to determine whether it
117 is followed by a delimiter or a newline before any of it may be
118 output. Otherwise, cut_fields can do the job without using this
120 static char *field_1_buffer;
122 /* The number of bytes allocated for FIELD_1_BUFFER. */
123 static int field_1_bufsize;
125 /* The largest field or byte index used as an endpoint of a closed
126 or degenerate range specification; this doesn't include the starting
127 index of right-open-ended ranges. For example, with either range spec
128 `2-5,9-', `2-3,5,9-' this variable would be set to 5. */
129 static unsigned int max_range_endpoint;
131 /* If nonzero, this is the index of the first field in a range that goes
133 static unsigned int eol_range_start;
135 /* In byte mode, which bytes to output.
136 In field mode, which DELIM-separated fields to output.
137 Both bytes and fields are numbered starting with 1,
138 so the zeroth element of this array is unused.
139 A field or byte K has been selected if
140 (K <= MAX_RANGE_ENDPOINT and PRINTABLE_FIELD[K])
141 || (EOL_RANGE_START > 0 && K >= EOL_RANGE_START). */
142 static int *printable_field;
148 /* Output characters that are in the given bytes. */
151 /* Output the given delimeter-separated fields. */
155 /* The name this program was run with. */
158 static enum operating_mode operating_mode;
160 /* If nonzero do not output lines containing no delimeter characters.
161 Otherwise, all such lines are printed. This option is valid only
163 static int suppress_non_delimited;
165 /* The delimeter character for field mode. */
168 /* Nonzero if we have ever read standard input. */
169 static int have_read_stdin;
171 /* If nonzero, display usage information and exit. */
172 static int show_help;
174 /* If nonzero, print the version on standard output then exit. */
175 static int show_version;
177 static struct option const longopts[] =
179 {"bytes", required_argument, 0, 'b'},
180 {"characters", required_argument, 0, 'c'},
181 {"fields", required_argument, 0, 'f'},
182 {"delimiter", required_argument, 0, 'd'},
183 {"only-delimited", no_argument, 0, 's'},
184 {"help", no_argument, &show_help, 1},
185 {"version", no_argument, &show_version, 1},
193 fprintf (stderr, _("Try `%s --help' for more information.\n"),
198 Usage: %s [OPTION]... [FILE]...\n\
202 Print selected parts of lines from each FILE to standard output.\n\
204 -b, --bytes=LIST output only these bytes\n\
205 -c, --characters=LIST output only these characters\n\
206 -d, --delimiter=DELIM use DELIM instead of TAB for field delimiter\n\
207 -f, --fields=LIST output only these fields\n\
209 -s, --only-delimited do not print lines not containing delimiters\n\
210 --help display this help and exit\n\
211 --version output version information and exit\n\
213 Use one, and only one of -b, -c or -f. Each LIST is made up of one\n\
214 range, or many ranges separated by commas. Each range is one of:\n\
216 N N'th byte, character or field, counted from 1\n\
217 N- from N'th byte, character or field, to end of line\n\
218 N-M from N'th to M'th (included) byte, character or field\n\
219 -M from first to M'th (included) byte, character or field\n\
221 With no FILE, or when FILE is -, read standard input.\n\
223 puts (_("\nReport bugs to textutils-bugs@gnu.ai.mit.edu"));
225 exit (status == 0 ? EXIT_SUCCESS : EXIT_FAILURE);
228 /* The following function was copied from getline.c, but with these changes:
229 - Read up to and including a newline or TERMINATOR, whichever comes first.
230 The original does not treat newline specially.
231 - Remove unused argument, OFFSET.
232 - Use xmalloc and xrealloc instead of malloc and realloc.
233 - Declare this function static. */
235 /* Always add at least this many bytes when extending the buffer. */
238 /* Read up to (and including) a newline or TERMINATOR from STREAM into
239 *LINEPTR (and null-terminate it). *LINEPTR is a pointer returned from
240 xmalloc (or NULL), pointing to *N characters of space. It is
241 xrealloc'd as necessary. Return the number of characters read (not
242 including the null terminator), or -1 on error or EOF. */
245 getstr (char **lineptr, int *n, FILE *stream, char terminator)
247 int nchars_avail; /* Allocated but unused chars in *LINEPTR. */
248 char *read_pos; /* Where we're reading into *LINEPTR. */
250 if (!lineptr || !n || !stream)
256 *lineptr = xmalloc (*n);
266 register int c = getc (stream);
268 /* We always want at least one char left in the buffer, since we
269 always (unless we get an error while reading the first char)
270 NUL-terminate the line buffer. */
272 assert (*n - nchars_avail == read_pos - *lineptr);
273 if (nchars_avail < 1)
280 nchars_avail = *n + *lineptr - read_pos;
281 *lineptr = xrealloc (*lineptr, *n);
284 read_pos = *n - nchars_avail + *lineptr;
285 assert (*n - nchars_avail == read_pos - *lineptr);
288 if (feof (stream) || ferror (stream))
290 /* Return partial line, if any. */
291 if (read_pos == *lineptr)
300 if (c == terminator || c == '\n')
301 /* Return the line. */
305 /* Done - NUL terminate and return the number of chars read. */
308 return read_pos - *lineptr;
312 print_kth (unsigned int k)
314 return ((0 < eol_range_start && eol_range_start <= k)
315 || (k <= max_range_endpoint && printable_field[k]));
318 /* Given the list of field or byte range specifications FIELDSTR, set
319 MAX_RANGE_ENDPOINT and allocate and initialize the PRINTABLE_FIELD
320 array. If there is a right-open-ended range, set EOL_RANGE_START
321 to its starting index. FIELDSTR should be composed of one or more
322 numbers or ranges of numbers, separated by blanks or commas.
323 Incomplete ranges may be given: `-m' means `1-m'; `n-' means `n'
324 through end of line. Return nonzero if FIELDSTR contains at least
325 one field specification, zero otherwise. */
327 /* FIXME-someday: What if the user wants to cut out the 1,000,000-th field
328 of some huge input file? This function shouldn't have to alloate a table
329 of a million ints just so we can test every field < 10^6 with an array
330 dereference. Instead, consider using a dynamic hash table. It would be
331 simpler and nearly as good a solution to use a 32K x 4-byte table with
332 one bit per field index instead of a whole `int' per index. */
335 set_fields (const char *fieldstr)
337 unsigned int initial = 1; /* Value of first number in a range. */
338 unsigned int value = 0; /* If nonzero, a number being accumulated. */
339 int dash_found = 0; /* Nonzero if a '-' is found in this field. */
340 int field_found = 0; /* Non-zero if at least one field spec
341 has been processed. */
343 struct range_pair *rp;
345 unsigned int n_rp_allocated;
350 rp = (struct range_pair *) xmalloc (n_rp_allocated * sizeof (*rp));
352 /* Collect and store in RP the range end points.
353 It also sets EOL_RANGE_START if appropriate. */
357 if (*fieldstr == '-')
359 /* Starting a range. */
361 FATAL_ERROR (_("invalid byte or field list"));
373 else if (*fieldstr == ',' || ISBLANK (*fieldstr) || *fieldstr == '\0')
375 /* Ending the string, or this field/byte sublist. */
380 /* A range. Possibilites: -n, m-n, n-.
381 In any case, `initial' contains the start of the range. */
384 /* `n-'. From `initial' to end of line. */
385 eol_range_start = initial;
390 /* `m-n' or `-n' (1-n). */
392 FATAL_ERROR (_("invalid byte or field list"));
394 /* Is there already a range going to end of line? */
395 if (eol_range_start != 0)
397 /* Yes. Is the new sequence already contained
398 in the old one? If so, no processing is
400 if (initial < eol_range_start)
402 /* No, the new sequence starts before the
403 old. Does the old range going to end of line
404 extend into the new range? */
405 if (value + 1 >= eol_range_start)
407 /* Yes. Simply move the end of line marker. */
408 eol_range_start = initial;
412 /* No. A simple range, before and disjoint from
413 the range going to end of line. Fill it. */
414 ADD_RANGE_PAIR (rp, initial, value);
417 /* In any case, some fields were selected. */
423 /* There is no range going to end of line. */
424 ADD_RANGE_PAIR (rp, initial, value);
432 /* A simple field number, not a range. */
433 ADD_RANGE_PAIR (rp, value, value);
438 if (*fieldstr == '\0')
445 else if (ISDIGIT (*fieldstr))
447 /* FIXME: detect overflow? */
448 value = 10 * value + *fieldstr - '0';
452 FATAL_ERROR (_("invalid byte or field list"));
455 max_range_endpoint = 0;
456 for (i = 0; i < n_rp; i++)
458 if (rp[i].hi > max_range_endpoint)
459 max_range_endpoint = rp[i].hi;
462 /* Allocate an array large enough so that it may be indexed by
463 the field numbers corresponding to all finite ranges
464 (i.e. `2-6' or `-4', but not `5-') in FIELDSTR. */
466 printable_field = (int *) xmalloc ((max_range_endpoint + 1) * sizeof (int));
467 memset (printable_field, 0, (max_range_endpoint + 1) * sizeof (int));
469 /* Set the array entries corresponding to integers in the ranges of RP. */
470 for (i = 0; i < n_rp; i++)
473 for (j = rp[i].lo; j <= rp[i].hi; j++)
475 printable_field[j] = 1;
484 /* Read from stream STREAM, printing to standard output any selected bytes. */
487 cut_bytes (FILE *stream)
489 unsigned int byte_idx; /* Number of chars in the line so far. */
494 register int c; /* Each character from the file. */
512 if (print_kth (byte_idx))
520 /* Read from stream STREAM, printing to standard output any selected fields. */
523 cut_fields (FILE *stream)
526 unsigned int field_idx;
527 int found_any_selected_field;
528 int buffer_first_field;
530 found_any_selected_field = 0;
533 /* To support the semantics of the -s flag, we may have to buffer
534 all of the first field to determine whether it is `delimited.'
535 But that is unnecessary if all non-delimited lines must be printed
536 and the first field has been selected, or if non-delimited lines
537 must be suppressed and the first field has *not* been selected.
538 That is because a non-delimited line has exactly one field. */
539 buffer_first_field = (suppress_non_delimited ^ !print_kth (1));
543 if (field_idx == 1 && buffer_first_field)
547 len = getstr (&field_1_buffer, &field_1_bufsize, stream, delim);
553 /* If the first field extends to the end of line (it is not
554 delimited) and we are printing all non-delimited lines,
556 if (field_1_buffer[len - 1] != delim)
558 if (suppress_non_delimited)
564 fwrite (field_1_buffer, sizeof (char), len, stdout);
565 /* Make sure the output line is newline terminated. */
566 if (field_1_buffer[len - 1] != '\n')
573 /* Print the field, but not the trailing delimiter. */
574 fwrite (field_1_buffer, sizeof (char), len - 1, stdout);
575 found_any_selected_field = 1;
580 if (print_kth (field_idx))
582 if (found_any_selected_field)
584 found_any_selected_field = 1;
586 while ((c = getc (stream)) != delim && c != '\n' && c != EOF)
593 while ((c = getc (stream)) != delim && c != '\n' && c != EOF)
611 else if (c == '\n' || c == EOF)
613 if (found_any_selected_field
614 || !(suppress_non_delimited && field_idx == 1))
619 found_any_selected_field = 0;
625 cut_stream (FILE *stream)
627 if (operating_mode == byte_mode)
633 /* Process file FILE to standard output.
634 Return 0 if successful, 1 if not. */
637 cut_file (char *file)
641 if (!strcmp (file, "-"))
648 stream = fopen (file, "r");
651 error (0, errno, "%s", file);
660 error (0, errno, "%s", file);
663 if (!strcmp (file, "-"))
664 clearerr (stream); /* Also clear EOF. */
665 else if (fclose (stream) == EOF)
667 error (0, errno, "%s", file);
674 main (int argc, char **argv)
676 int optc, exit_status = 0;
678 program_name = argv[0];
679 setlocale (LC_ALL, "");
680 bindtextdomain (PACKAGE, LOCALEDIR);
681 textdomain (PACKAGE);
683 operating_mode = undefined_mode;
685 /* By default, all non-delimited lines are printed. */
686 suppress_non_delimited = 0;
691 while ((optc = getopt_long (argc, argv, "b:c:d:f:ns", longopts, NULL)) != -1)
700 /* Build the byte list. */
701 if (operating_mode != undefined_mode)
702 FATAL_ERROR (_("only one type of list may be specified"));
703 operating_mode = byte_mode;
704 if (set_fields (optarg) == 0)
705 FATAL_ERROR (_("missing list of positions"));
709 /* Build the field list. */
710 if (operating_mode != undefined_mode)
711 FATAL_ERROR (_("only one type of list may be specified"));
712 operating_mode = field_mode;
713 if (set_fields (optarg) == 0)
714 FATAL_ERROR (_("missing list of fields"));
719 /* Interpret -d '' to mean `use the NUL byte as the delimiter.' */
720 if (optarg[0] != '\0' && optarg[1] != '\0')
721 FATAL_ERROR (_("the delimiter must be a single character"));
729 suppress_non_delimited = 1;
739 printf ("cut (%s) %s\n", GNU_PACKAGE, VERSION);
746 if (operating_mode == undefined_mode)
747 FATAL_ERROR (_("you must specify a list of bytes, characters, or fields"));
749 if (delim != '\0' && operating_mode != field_mode)
750 FATAL_ERROR (_("a delimiter may be specified only when operating on fields"));
752 if (suppress_non_delimited && operating_mode != field_mode)
753 FATAL_ERROR (_("suppressing non-delimited lines makes sense\n\
754 \tonly when operating on fields"));
760 exit_status |= cut_file ("-");
762 for (; optind < argc; optind++)
763 exit_status |= cut_file (argv[optind]);
765 if (have_read_stdin && fclose (stdin) == EOF)
767 error (0, errno, "-");
770 if (ferror (stdout) || fclose (stdout) == EOF)
771 error (EXIT_FAILURE, errno, _("write error"));
773 exit (exit_status == 0 ? EXIT_SUCCESS : EXIT_FAILURE);