blob: 7a44d10886442770b2542619fec581cf4f8c433d [file] [log] [blame]
"Robert P. J. Day"63fc1a92006-07-02 19:47:05 +00001/* vi: set sw=4 ts=4: */
Erik Andersen7ab9c7e2000-05-12 19:41:47 +00002/*
Mark Whitley807f0fd2000-08-02 18:30:11 +00003 * cut.c - minimalist version of cut
Erik Andersen7ab9c7e2000-05-12 19:41:47 +00004 *
Eric Andersen8ec10a92001-01-27 09:33:39 +00005 * Copyright (C) 1999,2000,2001 by Lineo, inc.
Eric Andersenc7bda1c2004-03-15 08:29:22 +00006 * Written by Mark Whitley <markw@codepoet.org>
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +00007 * debloated by Bernhard Fischer
Erik Andersen7ab9c7e2000-05-12 19:41:47 +00008 *
"Robert P. J. Day"801ab142006-07-12 07:56:04 +00009 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
Erik Andersen7ab9c7e2000-05-12 19:41:47 +000010 */
11
Denis Vlasenkob6adbf12007-05-26 19:00:18 +000012#include "libbb.h"
Erik Andersen7ab9c7e2000-05-12 19:41:47 +000013
Denis Vlasenko99912ca2007-04-10 15:43:37 +000014/* This is a NOEXEC applet. Be very careful! */
15
16
Mark Whitleyb6967632001-05-18 23:04:51 +000017/* option vars */
Denis Vlasenko6ca409e2007-08-12 20:58:27 +000018static const char optstring[] ALIGN1 = "b:c:f:d:sn";
Denis Vlasenko3831c912007-11-23 07:26:15 +000019#define CUT_OPT_BYTE_FLGS (1 << 0)
20#define CUT_OPT_CHAR_FLGS (1 << 1)
21#define CUT_OPT_FIELDS_FLGS (1 << 2)
22#define CUT_OPT_DELIM_FLGS (1 << 3)
23#define CUT_OPT_SUPPRESS_FLGS (1 << 4)
Mark Whitleyb6967632001-05-18 23:04:51 +000024
25struct cut_list {
26 int startpos;
27 int endpos;
28};
29
Rob Landleybc68cd12006-03-10 19:22:06 +000030enum {
31 BOL = 0,
32 EOL = INT_MAX,
33 NON_RANGE = -1
34};
Mark Whitleyb6967632001-05-18 23:04:51 +000035
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +000036/* growable array holding a series of lists */
37static struct cut_list *cut_lists;
38static unsigned int nlists; /* number of elements in above list */
Mark Whitleyb6967632001-05-18 23:04:51 +000039
40
41static int cmpfunc(const void *a, const void *b)
42{
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +000043 return (((struct cut_list *) a)->startpos -
44 ((struct cut_list *) b)->startpos);
Mark Whitleyb6967632001-05-18 23:04:51 +000045
46}
47
Denis Vlasenko3831c912007-11-23 07:26:15 +000048static void cut_file(FILE *file, char delim)
Mark Whitley807f0fd2000-08-02 18:30:11 +000049{
Mark Whitleyb6967632001-05-18 23:04:51 +000050 char *line = NULL;
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +000051 unsigned int linenum = 0; /* keep these zero-based to be consistent */
Mark Whitley807f0fd2000-08-02 18:30:11 +000052
53 /* go through every line in the file */
Denis Vlasenko8ee649a2008-03-26 20:04:27 +000054 while ((line = xmalloc_fgetline(file)) != NULL) {
Mark Whitleyb6967632001-05-18 23:04:51 +000055
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +000056 /* set up a list so we can keep track of what's been printed */
57 char * printed = xzalloc(strlen(line) * sizeof(char));
58 char * orig_line = line;
59 unsigned int cl_pos = 0;
60 int spos;
61
Mark Whitleyb6967632001-05-18 23:04:51 +000062 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
Denis Vlasenko372686b2006-10-12 22:42:33 +000063 if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +000064 /* print the chars specified in each cut list */
65 for (; cl_pos < nlists; cl_pos++) {
66 spos = cut_lists[cl_pos].startpos;
67 while (spos < strlen(line)) {
68 if (!printed[spos]) {
69 printed[spos] = 'X';
70 putchar(line[spos]);
71 }
72 spos++;
73 if (spos > cut_lists[cl_pos].endpos
74 || cut_lists[cl_pos].endpos == NON_RANGE)
75 break;
76 }
77 }
78 } else if (delim == '\n') { /* cut by lines */
79 spos = cut_lists[cl_pos].startpos;
Mark Whitleyb6967632001-05-18 23:04:51 +000080
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +000081 /* get out if we have no more lists to process or if the lines
82 * are lower than what we're interested in */
83 if (linenum < spos || cl_pos >= nlists)
84 goto next_line;
85
86 /* if the line we're looking for is lower than the one we were
87 * passed, it means we displayed it already, so move on */
88 while (spos < linenum) {
89 spos++;
90 /* go to the next list if we're at the end of this one */
91 if (spos > cut_lists[cl_pos].endpos
92 || cut_lists[cl_pos].endpos == NON_RANGE) {
93 cl_pos++;
94 /* get out if there's no more lists to process */
95 if (cl_pos >= nlists)
96 goto next_line;
97 spos = cut_lists[cl_pos].startpos;
98 /* get out if the current line is lower than the one
99 * we just became interested in */
100 if (linenum < spos)
101 goto next_line;
102 }
103 }
104
105 /* If we made it here, it means we've found the line we're
106 * looking for, so print it */
107 puts(line);
108 goto next_line;
109 } else { /* cut by fields */
110 int ndelim = -1; /* zero-based / one-based problem */
111 int nfields_printed = 0;
112 char *field = NULL;
113 const char delimiter[2] = { delim, 0 };
114
115 /* does this line contain any delimiters? */
116 if (strchr(line, delim) == NULL) {
Denis Vlasenko372686b2006-10-12 22:42:33 +0000117 if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000118 puts(line);
119 goto next_line;
120 }
121
122 /* process each list on this line, for as long as we've got
123 * a line to process */
124 for (; cl_pos < nlists && line; cl_pos++) {
125 spos = cut_lists[cl_pos].startpos;
126 do {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000127 /* find the field we're looking for */
128 while (line && ndelim < spos) {
129 field = strsep(&line, delimiter);
130 ndelim++;
131 }
132
133 /* we found it, and it hasn't been printed yet */
134 if (field && ndelim == spos && !printed[ndelim]) {
135 /* if this isn't our first time through, we need to
136 * print the delimiter after the last field that was
137 * printed */
138 if (nfields_printed > 0)
139 putchar(delim);
140 fputs(field, stdout);
141 printed[ndelim] = 'X';
142 nfields_printed++; /* shouldn't overflow.. */
143 }
144
145 spos++;
146
147 /* keep going as long as we have a line to work with,
148 * this is a list, and we're not at the end of that
149 * list */
150 } while (spos <= cut_lists[cl_pos].endpos && line
151 && cut_lists[cl_pos].endpos != NON_RANGE);
152 }
Mark Whitley807f0fd2000-08-02 18:30:11 +0000153 }
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000154 /* if we printed anything at all, we need to finish it with a
155 * newline cuz we were handed a chomped line */
156 putchar('\n');
Denis Vlasenko372686b2006-10-12 22:42:33 +0000157 next_line:
Mark Whitleyb6967632001-05-18 23:04:51 +0000158 linenum++;
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000159 free(printed);
160 free(orig_line);
Mark Whitley807f0fd2000-08-02 18:30:11 +0000161 }
162}
163
Denis Vlasenko9b49a5e2007-10-11 10:05:36 +0000164int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Denis Vlasenko68404f12008-03-17 09:00:54 +0000165int cut_main(int argc ATTRIBUTE_UNUSED, char **argv)
Mark Whitley807f0fd2000-08-02 18:30:11 +0000166{
Denis Vlasenko3831c912007-11-23 07:26:15 +0000167 char delim = '\t'; /* delimiter, default is tab */
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000168 char *sopt, *ltok;
Mark Whitley807f0fd2000-08-02 18:30:11 +0000169
Denis Vlasenko67b23e62006-10-03 21:00:06 +0000170 opt_complementary = "b--bcf:c--bcf:f--bcf";
Denis Vlasenkofe7cd642007-08-18 15:32:12 +0000171 getopt32(argv, optstring, &sopt, &sopt, &sopt, &ltok);
Denis Vlasenko9f739442006-12-16 23:49:13 +0000172// argc -= optind;
173 argv += optind;
Denis Vlasenko372686b2006-10-12 22:42:33 +0000174 if (!(option_mask32 & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
175 bb_error_msg_and_die("expected a list of bytes, characters, or fields");
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000176
Denis Vlasenko372686b2006-10-12 22:42:33 +0000177 if (option_mask32 & CUT_OPT_DELIM_FLGS) {
Denis Vlasenko3831c912007-11-23 07:26:15 +0000178 if (ltok[0] && ltok[1]) { /* more than 1 char? */
Eric Andersenc9e70242003-06-20 09:16:00 +0000179 bb_error_msg_and_die("the delimiter must be a single character");
180 }
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000181 delim = ltok[0];
Mark Whitley807f0fd2000-08-02 18:30:11 +0000182 }
Mark Whitley807f0fd2000-08-02 18:30:11 +0000183
Mark Whitleyb6967632001-05-18 23:04:51 +0000184 /* non-field (char or byte) cutting has some special handling */
Denis Vlasenko372686b2006-10-12 22:42:33 +0000185 if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
Denis Vlasenko3831c912007-11-23 07:26:15 +0000186 static const char _op_on_field[] ALIGN1 = " only when operating on fields";
187
Denis Vlasenko372686b2006-10-12 22:42:33 +0000188 if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000189 bb_error_msg_and_die
190 ("suppressing non-delimited lines makes sense%s",
191 _op_on_field);
Mark Whitleyb6967632001-05-18 23:04:51 +0000192 }
Eric Andersen8876fb22003-06-20 09:01:58 +0000193 if (delim != '\t') {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000194 bb_error_msg_and_die
195 ("a delimiter may be specified%s", _op_on_field);
Mark Whitleyb6967632001-05-18 23:04:51 +0000196 }
Mark Whitley807f0fd2000-08-02 18:30:11 +0000197 }
198
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000199 /*
200 * parse list and put values into startpos and endpos.
201 * valid list formats: N, N-, N-M, -M
202 * more than one list can be separated by commas
203 */
204 {
205 char *ntok;
206 int s = 0, e = 0;
207
Denis Vlasenko62a90cd2008-03-17 09:07:36 +0000208 /* take apart the lists, one by one (they are separated with commas) */
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000209 while ((ltok = strsep(&sopt, ",")) != NULL) {
210
211 /* it's actually legal to pass an empty list */
Denis Vlasenko3831c912007-11-23 07:26:15 +0000212 if (!ltok[0])
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000213 continue;
214
215 /* get the start pos */
216 ntok = strsep(&ltok, "-");
Denis Vlasenko3831c912007-11-23 07:26:15 +0000217 if (!ntok[0]) {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000218 s = BOL;
219 } else {
Denis Vlasenko13858992006-10-08 12:49:22 +0000220 s = xatoi_u(ntok);
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000221 /* account for the fact that arrays are zero based, while
222 * the user expects the first char on the line to be char #1 */
223 if (s != 0)
224 s--;
225 }
226
227 /* get the end pos */
Denis Vlasenko3831c912007-11-23 07:26:15 +0000228 if (ltok == NULL) {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000229 e = NON_RANGE;
Denis Vlasenko3831c912007-11-23 07:26:15 +0000230 } else if (!ltok[0]) {
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000231 e = EOL;
232 } else {
Denis Vlasenko3831c912007-11-23 07:26:15 +0000233 e = xatoi_u(ltok);
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000234 /* if the user specified and end position of 0, that means "til the
235 * end of the line */
236 if (e == 0)
237 e = EOL;
238 e--; /* again, arrays are zero based, lines are 1 based */
239 if (e == s)
240 e = NON_RANGE;
241 }
242
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000243 /* add the new list */
Denis Vlasenko372686b2006-10-12 22:42:33 +0000244 cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
245 cut_lists[nlists-1].startpos = s;
246 cut_lists[nlists-1].endpos = e;
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000247 }
248
249 /* make sure we got some cut positions out of all that */
250 if (nlists == 0)
251 bb_error_msg_and_die("missing list of positions");
252
253 /* now that the lists are parsed, we need to sort them to make life
254 * easier on us when it comes time to print the chars / fields / lines
255 */
256 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
257 }
258
Denis Vlasenko3831c912007-11-23 07:26:15 +0000259 {
260 int retval = EXIT_SUCCESS;
Denis Vlasenko3831c912007-11-23 07:26:15 +0000261
Denis Vlasenko62a90cd2008-03-17 09:07:36 +0000262 if (!*argv)
263 *--argv = (char *)"-";
Bernhard Reutner-Fischer73561cc2006-08-28 23:31:54 +0000264
Denis Vlasenko9f739442006-12-16 23:49:13 +0000265 do {
Denis Vlasenko62a90cd2008-03-17 09:07:36 +0000266 FILE *file = fopen_or_warn_stdin(*argv);
Denis Vlasenko3831c912007-11-23 07:26:15 +0000267 if (!file) {
268 retval = EXIT_FAILURE;
269 continue;
Mark Whitley807f0fd2000-08-02 18:30:11 +0000270 }
Denis Vlasenko3831c912007-11-23 07:26:15 +0000271 cut_file(file, delim);
Denis Vlasenko62a90cd2008-03-17 09:07:36 +0000272 fclose_if_not_stdin(file);
Denis Vlasenko9f739442006-12-16 23:49:13 +0000273 } while (*++argv);
Denis Vlasenko3831c912007-11-23 07:26:15 +0000274
275 if (ENABLE_FEATURE_CLEAN_UP)
276 free(cut_lists);
277 fflush_stdout_and_exit(retval);
Mark Whitley807f0fd2000-08-02 18:30:11 +0000278 }
Mark Whitley807f0fd2000-08-02 18:30:11 +0000279}