busybox/coreutils/cut.c

/* vi: set sw=4 ts=4: */
/*
 * cut.c - minimalist version of cut
 *
 * Copyright (C) 1999,2000,2001 by Lineo, inc.
 * Written by Mark Whitley <markw@codepoet.org>
 * debloated by Bernhard Reutner-Fischer
 *
 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
 */

#include "libbb.h"

/* This is a NOEXEC applet. Be very careful! */


/* option vars */
static const char optstring[] ALIGN1 = "b:c:f:d:sn";
#define CUT_OPT_BYTE_FLGS     (1 << 0)
#define CUT_OPT_CHAR_FLGS     (1 << 1)
#define CUT_OPT_FIELDS_FLGS   (1 << 2)
#define CUT_OPT_DELIM_FLGS    (1 << 3)
#define CUT_OPT_SUPPRESS_FLGS (1 << 4)

struct cut_list {
	int startpos;
	int endpos;
};

enum {
	BOL = 0,
	EOL = INT_MAX,
	NON_RANGE = -1
};

static int cmpfunc(const void *a, const void *b)
{
	return (((struct cut_list *) a)->startpos -
			((struct cut_list *) b)->startpos);

}

static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists)
{
	char *line;
	unsigned linenum = 0;	/* keep these zero-based to be consistent */

	/* go through every line in the file */
	while ((line = xmalloc_fgetline(file)) != NULL) {

		/* set up a list so we can keep track of what's been printed */
		int linelen = strlen(line);
		char *printed = xzalloc(linelen + 1);
		char *orig_line = line;
		unsigned cl_pos = 0;
		int spos;

		/* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
		if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
			/* print the chars specified in each cut list */
			for (; cl_pos < nlists; cl_pos++) {
				spos = cut_lists[cl_pos].startpos;
				while (spos < linelen) {
					if (!printed[spos]) {
						printed[spos] = 'X';
						putchar(line[spos]);
					}
					spos++;
					if (spos > cut_lists[cl_pos].endpos
					/* NON_RANGE is -1, so if below is true,
					 * the above was true too (spos is >= 0) */
					/* || cut_lists[cl_pos].endpos == NON_RANGE */
					) {
						break;
					}
				}
			}
		} else if (delim == '\n') {	/* cut by lines */
			spos = cut_lists[cl_pos].startpos;

			/* get out if we have no more lists to process or if the lines
			 * are lower than what we're interested in */
			if (((int)linenum < spos) || (cl_pos >= nlists))
				goto next_line;

			/* if the line we're looking for is lower than the one we were
			 * passed, it means we displayed it already, so move on */
			while (spos < (int)linenum) {
				spos++;
				/* go to the next list if we're at the end of this one */
				if (spos > cut_lists[cl_pos].endpos
				 || cut_lists[cl_pos].endpos == NON_RANGE
				) {
					cl_pos++;
					/* get out if there's no more lists to process */
					if (cl_pos >= nlists)
						goto next_line;
					spos = cut_lists[cl_pos].startpos;
					/* get out if the current line is lower than the one
					 * we just became interested in */
					if ((int)linenum < spos)
						goto next_line;
				}
			}

			/* If we made it here, it means we've found the line we're
			 * looking for, so print it */
			puts(line);
			goto next_line;
		} else {		/* cut by fields */
			int ndelim = -1;	/* zero-based / one-based problem */
			int nfields_printed = 0;
			char *field = NULL;
			char delimiter[2];

			delimiter[0] = delim;
			delimiter[1] = 0;

			/* does this line contain any delimiters? */
			if (strchr(line, delim) == NULL) {
				if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
					puts(line);
				goto next_line;
			}

			/* process each list on this line, for as long as we've got
			 * a line to process */
			for (; cl_pos < nlists && line; cl_pos++) {
				spos = cut_lists[cl_pos].startpos;
				do {
					/* find the field we're looking for */
					while (line && ndelim < spos) {
						field = strsep(&line, delimiter);
						ndelim++;
					}

					/* we found it, and it hasn't been printed yet */
					if (field && ndelim == spos && !printed[ndelim]) {
						/* if this isn't our first time through, we need to
						 * print the delimiter after the last field that was
						 * printed */
						if (nfields_printed > 0)
							putchar(delim);
						fputs(field, stdout);
						printed[ndelim] = 'X';
						nfields_printed++;	/* shouldn't overflow.. */
					}

					spos++;

					/* keep going as long as we have a line to work with,
					 * this is a list, and we're not at the end of that
					 * list */
				} while (spos <= cut_lists[cl_pos].endpos && line
						&& cut_lists[cl_pos].endpos != NON_RANGE);
			}
		}
		/* if we printed anything at all, we need to finish it with a
		 * newline cuz we were handed a chomped line */
		putchar('\n');
 next_line:
		linenum++;
		free(printed);
		free(orig_line);
	}
}

int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int cut_main(int argc UNUSED_PARAM, char **argv)
{
	/* growable array holding a series of lists */
	struct cut_list *cut_lists = NULL;
	unsigned nlists = 0;	/* number of elements in above list */
	char delim = '\t';	/* delimiter, default is tab */
	char *sopt, *ltok;
	unsigned opt;

	opt_complementary = "b--bcf:c--bcf:f--bcf";
	opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, &ltok);
//	argc -= optind;
	argv += optind;
	if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
		bb_error_msg_and_die("expected a list of bytes, characters, or fields");

	if (opt & CUT_OPT_DELIM_FLGS) {
		if (ltok[0] && ltok[1]) { /* more than 1 char? */
			bb_error_msg_and_die("the delimiter must be a single character");
		}
		delim = ltok[0];
	}

	/*  non-field (char or byte) cutting has some special handling */
	if (!(opt & CUT_OPT_FIELDS_FLGS)) {
		static const char _op_on_field[] ALIGN1 = " only when operating on fields";

		if (opt & CUT_OPT_SUPPRESS_FLGS) {
			bb_error_msg_and_die
				("suppressing non-delimited lines makes sense%s",
				 _op_on_field);
		}
		if (delim != '\t') {
			bb_error_msg_and_die
				("a delimiter may be specified%s", _op_on_field);
		}
	}

	/*
	 * parse list and put values into startpos and endpos.
	 * valid list formats: N, N-, N-M, -M
	 * more than one list can be separated by commas
	 */
	{
		char *ntok;
		int s = 0, e = 0;

		/* take apart the lists, one by one (they are separated with commas) */
		while ((ltok = strsep(&sopt, ",")) != NULL) {

			/* it's actually legal to pass an empty list */
			if (!ltok[0])
				continue;

			/* get the start pos */
			ntok = strsep(&ltok, "-");
			if (!ntok[0]) {
				s = BOL;
			} else {
				s = xatoi_u(ntok);
				/* account for the fact that arrays are zero based, while
				 * the user expects the first char on the line to be char #1 */
				if (s != 0)
					s--;
			}

			/* get the end pos */
			if (ltok == NULL) {
				e = NON_RANGE;
			} else if (!ltok[0]) {
				e = EOL;
			} else {
				e = xatoi_u(ltok);
				/* if the user specified and end position of 0,
				 * that means "til the end of the line" */
				if (e == 0)
					e = EOL;
				e--;	/* again, arrays are zero based, lines are 1 based */
				if (e == s)
					e = NON_RANGE;
			}

			/* add the new list */
			cut_lists = xrealloc_vector(cut_lists, 4, nlists);
			/* NB: startpos is always >= 0,
			 * while endpos may be = NON_RANGE (-1) */
			cut_lists[nlists].startpos = s;
			cut_lists[nlists].endpos = e;
			nlists++;
		}

		/* make sure we got some cut positions out of all that */
		if (nlists == 0)
			bb_error_msg_and_die("missing list of positions");

		/* now that the lists are parsed, we need to sort them to make life
		 * easier on us when it comes time to print the chars / fields / lines
		 */
		qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
	}

	{
		int retval = EXIT_SUCCESS;

		if (!*argv)
			*--argv = (char *)"-";

		do {
			FILE *file = fopen_or_warn_stdin(*argv);
			if (!file) {
				retval = EXIT_FAILURE;
				continue;
			}
			cut_file(file, delim, cut_lists, nlists);
			fclose_if_not_stdin(file);
		} while (*++argv);

		if (ENABLE_FEATURE_CLEAN_UP)
			free(cut_lists);
		fflush_stdout_and_exit(retval);
	}
}
1	niro	532	/* vi: set sw=4 ts=4: */
2			/*
3			* cut.c - minimalist version of cut
4			*
5			* Copyright (C) 1999,2000,2001 by Lineo, inc.
6			* Written by Mark Whitley <markw@codepoet.org>
7	niro	816	* debloated by Bernhard Reutner-Fischer
8	niro	532	*
9			* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
10			*/
11
12	niro	816	#include "libbb.h"
13	niro	532
14	niro	816	/* This is a NOEXEC applet. Be very careful! */
15
16
17	niro	532	/* option vars */
18	niro	816	static const char optstring[] ALIGN1 = "b:c:f:d:sn";
19			#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)
24	niro	532
25			struct cut_list {
26			int startpos;
27			int endpos;
28			};
29
30			enum {
31			BOL = 0,
32			EOL = INT_MAX,
33			NON_RANGE = -1
34			};
35
36			static int cmpfunc(const void a, const void b)
37			{
38			return (((struct cut_list *) a)->startpos -
39			((struct cut_list *) b)->startpos);
40
41			}
42
43	niro	816	static void cut_file(FILE file, char delim, const struct cut_list cut_lists, unsigned nlists)
44	niro	532	{
45	niro	816	char *line;
46			unsigned linenum = 0; /* keep these zero-based to be consistent */
47	niro	532
48			/* go through every line in the file */
49	niro	816	while ((line = xmalloc_fgetline(file)) != NULL) {
50	niro	532
51			/* set up a list so we can keep track of what's been printed */
52	niro	816	int linelen = strlen(line);
53			char *printed = xzalloc(linelen + 1);
54			char *orig_line = line;
55			unsigned cl_pos = 0;
56	niro	532	int spos;
57
58			/* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
59			if (option_mask32 & (CUT_OPT_CHAR_FLGS \| CUT_OPT_BYTE_FLGS)) {
60			/* print the chars specified in each cut list */
61			for (; cl_pos < nlists; cl_pos++) {
62			spos = cut_lists[cl_pos].startpos;
63	niro	816	while (spos < linelen) {
64	niro	532	if (!printed[spos]) {
65			printed[spos] = 'X';
66			putchar(line[spos]);
67			}
68			spos++;
69			if (spos > cut_lists[cl_pos].endpos
70	niro	816	/* NON_RANGE is -1, so if below is true,
71			* the above was true too (spos is >= 0) */
72			/* \|\| cut_lists[cl_pos].endpos == NON_RANGE */
73			) {
74	niro	532	break;
75	niro	816	}
76	niro	532	}
77			}
78			} else if (delim == '\n') { /* cut by lines */
79			spos = cut_lists[cl_pos].startpos;
80
81			/* get out if we have no more lists to process or if the lines
82			* are lower than what we're interested in */
83	niro	816	if (((int)linenum < spos) \|\| (cl_pos >= nlists))
84	niro	532	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	niro	816	while (spos < (int)linenum) {
89	niro	532	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	niro	816	\|\| cut_lists[cl_pos].endpos == NON_RANGE
93			) {
94	niro	532	cl_pos++;
95			/* get out if there's no more lists to process */
96			if (cl_pos >= nlists)
97			goto next_line;
98			spos = cut_lists[cl_pos].startpos;
99			/* get out if the current line is lower than the one
100			* we just became interested in */
101	niro	816	if ((int)linenum < spos)
102	niro	532	goto next_line;
103			}
104			}
105
106			/* If we made it here, it means we've found the line we're
107			* looking for, so print it */
108			puts(line);
109			goto next_line;
110			} else { /* cut by fields */
111			int ndelim = -1; /* zero-based / one-based problem */
112			int nfields_printed = 0;
113			char *field = NULL;
114	niro	984	char delimiter[2];
115	niro	532
116	niro	984	delimiter[0] = delim;
117			delimiter[1] = 0;
118
119	niro	532	/* does this line contain any delimiters? */
120			if (strchr(line, delim) == NULL) {
121			if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
122			puts(line);
123			goto next_line;
124			}
125
126			/* process each list on this line, for as long as we've got
127			* a line to process */
128			for (; cl_pos < nlists && line; cl_pos++) {
129			spos = cut_lists[cl_pos].startpos;
130			do {
131			/* find the field we're looking for */
132			while (line && ndelim < spos) {
133			field = strsep(&line, delimiter);
134			ndelim++;
135			}
136
137			/* we found it, and it hasn't been printed yet */
138			if (field && ndelim == spos && !printed[ndelim]) {
139			/* if this isn't our first time through, we need to
140			* print the delimiter after the last field that was
141			* printed */
142			if (nfields_printed > 0)
143			putchar(delim);
144			fputs(field, stdout);
145			printed[ndelim] = 'X';
146			nfields_printed++; /* shouldn't overflow.. */
147			}
148
149			spos++;
150
151			/* keep going as long as we have a line to work with,
152			* this is a list, and we're not at the end of that
153			* list */
154			} while (spos <= cut_lists[cl_pos].endpos && line
155	niro	816	&& cut_lists[cl_pos].endpos != NON_RANGE);
156	niro	532	}
157			}
158			/* if we printed anything at all, we need to finish it with a
159			* newline cuz we were handed a chomped line */
160			putchar('\n');
161			next_line:
162			linenum++;
163			free(printed);
164			free(orig_line);
165			}
166			}
167
168	niro	816	int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
169			int cut_main(int argc UNUSED_PARAM, char **argv)
170	niro	532	{
171	niro	816	/* growable array holding a series of lists */
172			struct cut_list *cut_lists = NULL;
173			unsigned nlists = 0; /* number of elements in above list */
174			char delim = '\t'; /* delimiter, default is tab */
175	niro	532	char sopt, ltok;
176	niro	816	unsigned opt;
177	niro	532
178			opt_complementary = "b--bcf:c--bcf:f--bcf";
179	niro	816	opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, &ltok);
180	niro	532	// argc -= optind;
181			argv += optind;
182	niro	816	if (!(opt & (CUT_OPT_BYTE_FLGS \| CUT_OPT_CHAR_FLGS \| CUT_OPT_FIELDS_FLGS)))
183	niro	532	bb_error_msg_and_die("expected a list of bytes, characters, or fields");
184
185	niro	816	if (opt & CUT_OPT_DELIM_FLGS) {
186			if (ltok[0] && ltok[1]) { /* more than 1 char? */
187	niro	532	bb_error_msg_and_die("the delimiter must be a single character");
188			}
189			delim = ltok[0];
190			}
191
192			/* non-field (char or byte) cutting has some special handling */
193	niro	816	if (!(opt & CUT_OPT_FIELDS_FLGS)) {
194			static const char _op_on_field[] ALIGN1 = " only when operating on fields";
195
196			if (opt & CUT_OPT_SUPPRESS_FLGS) {
197	niro	532	bb_error_msg_and_die
198			("suppressing non-delimited lines makes sense%s",
199			_op_on_field);
200			}
201			if (delim != '\t') {
202			bb_error_msg_and_die
203			("a delimiter may be specified%s", _op_on_field);
204			}
205			}
206
207			/*
208			* parse list and put values into startpos and endpos.
209			* valid list formats: N, N-, N-M, -M
210			* more than one list can be separated by commas
211			*/
212			{
213			char *ntok;
214			int s = 0, e = 0;
215
216	niro	816	/* take apart the lists, one by one (they are separated with commas) */
217	niro	532	while ((ltok = strsep(&sopt, ",")) != NULL) {
218
219			/* it's actually legal to pass an empty list */
220	niro	816	if (!ltok[0])
221	niro	532	continue;
222
223			/* get the start pos */
224			ntok = strsep(&ltok, "-");
225	niro	816	if (!ntok[0]) {
226	niro	532	s = BOL;
227			} else {
228			s = xatoi_u(ntok);
229			/* account for the fact that arrays are zero based, while
230			* the user expects the first char on the line to be char #1 */
231			if (s != 0)
232			s--;
233			}
234
235			/* get the end pos */
236	niro	816	if (ltok == NULL) {
237	niro	532	e = NON_RANGE;
238	niro	816	} else if (!ltok[0]) {
239	niro	532	e = EOL;
240			} else {
241	niro	816	e = xatoi_u(ltok);
242			/* if the user specified and end position of 0,
243			* that means "til the end of the line" */
244	niro	532	if (e == 0)
245			e = EOL;
246			e--; /* again, arrays are zero based, lines are 1 based */
247			if (e == s)
248			e = NON_RANGE;
249			}
250
251			/* add the new list */
252	niro	816	cut_lists = xrealloc_vector(cut_lists, 4, nlists);
253			/* NB: startpos is always >= 0,
254			* while endpos may be = NON_RANGE (-1) */
255			cut_lists[nlists].startpos = s;
256			cut_lists[nlists].endpos = e;
257			nlists++;
258	niro	532	}
259
260			/* make sure we got some cut positions out of all that */
261			if (nlists == 0)
262			bb_error_msg_and_die("missing list of positions");
263
264			/* now that the lists are parsed, we need to sort them to make life
265			* easier on us when it comes time to print the chars / fields / lines
266			*/
267	niro	984	qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
268	niro	532	}
269
270	niro	816	{
271			int retval = EXIT_SUCCESS;
272	niro	532
273	niro	816	if (!*argv)
274			--argv = (char )"-";
275
276	niro	532	do {
277	niro	816	FILE file = fopen_or_warn_stdin(argv);
278			if (!file) {
279			retval = EXIT_FAILURE;
280			continue;
281	niro	532	}
282	niro	816	cut_file(file, delim, cut_lists, nlists);
283			fclose_if_not_stdin(file);
284	niro	532	} while (*++argv);
285	niro	816
286			if (ENABLE_FEATURE_CLEAN_UP)
287			free(cut_lists);
288			fflush_stdout_and_exit(retval);
289	niro	532	}
290			}