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 Fischer
 *
 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
 */

#include "busybox.h"

/* option vars */
static const char optstring[] = "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)

static char delim = '\t';	/* delimiter, default is tab */

struct cut_list {
	int startpos;
	int endpos;
};

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

/* growable array holding a series of lists */
static struct cut_list *cut_lists;
static unsigned int nlists;	/* number of elements in above list */


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 *line = NULL;
	unsigned int linenum = 0;	/* keep these zero-based to be consistent */

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

		/* set up a list so we can keep track of what's been printed */
		char * printed = xzalloc(strlen(line) * sizeof(char));
		char * orig_line = line;
		unsigned int 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 < strlen(line)) {
					if (!printed[spos]) {
						printed[spos] = 'X';
						putchar(line[spos]);
					}
					spos++;
					if (spos > cut_lists[cl_pos].endpos
						|| 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 (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 < 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 (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;
			const char delimiter[2] = { delim, 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);
	}
}

static const char _op_on_field[] = " only when operating on fields";

int cut_main(int argc, char **argv)
{
	char *sopt, *ltok;

	opt_complementary = "b--bcf:c--bcf:f--bcf";
	getopt32(argc, argv, optstring, &sopt, &sopt, &sopt, &ltok);
//	argc -= optind;
	argv += optind;
	if (!(option_mask32 & (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 (option_mask32 & BB_GETOPT_ERROR)
		bb_error_msg_and_die("only one type of list may be specified");

	if (option_mask32 & CUT_OPT_DELIM_FLGS) {
		if (strlen(ltok) > 1) {
			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 (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
		if (option_mask32 & 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 (strlen(ltok) == 0)
				continue;

			/* get the start pos */
			ntok = strsep(&ltok, "-");
			if (ntok == NULL) {
				bb_error_msg
					("internal error: ntok is null for start pos!?\n");
			} else if (strlen(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 */
			ntok = strsep(&ltok, "-");
			if (ntok == NULL) {
				e = NON_RANGE;
			} else if (strlen(ntok) == 0) {
				e = EOL;
			} else {
				e = xatoi_u(ntok);
				/* 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;
			}

			/* if there's something left to tokenize, the user passed
			 * an invalid list */
			if (ltok)
				bb_error_msg_and_die("invalid byte or field list");

			/* add the new list */
			cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
			cut_lists[nlists-1].startpos = s;
			cut_lists[nlists-1].endpos = e;
		}

		/* 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(struct cut_list), cmpfunc);
	}

	/* argv[0..argc-1] should be names of file to process. If no
	 * files were specified or '-' was specified, take input from stdin.
	 * Otherwise, we process all the files specified. */
	if (argv[0] == NULL || LONE_DASH(argv[0])) {
		cut_file(stdin);
	} else {
		FILE *file;

		do {
			file = fopen_or_warn(argv[0], "r");
			if (file) {
				cut_file(file);
				fclose(file);
			}
		} while (*++argv);
	}
	if (ENABLE_FEATURE_CLEAN_UP)
		free(cut_lists);
	return EXIT_SUCCESS;
}
1	/* 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	* debloated by Bernhard Fischer
8	*
9	* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
10	*/
11
12	#include "busybox.h"
13
14	/* option vars */
15	static const char optstring[] = "b:c:f:d:sn";
16	#define CUT_OPT_BYTE_FLGS (1<<0)
17	#define CUT_OPT_CHAR_FLGS (1<<1)
18	#define CUT_OPT_FIELDS_FLGS (1<<2)
19	#define CUT_OPT_DELIM_FLGS (1<<3)
20	#define CUT_OPT_SUPPRESS_FLGS (1<<4)
21
22	static char delim = '\t'; /* delimiter, default is tab */
23
24	struct cut_list {
25	int startpos;
26	int endpos;
27	};
28
29	enum {
30	BOL = 0,
31	EOL = INT_MAX,
32	NON_RANGE = -1
33	};
34
35	/* growable array holding a series of lists */
36	static struct cut_list *cut_lists;
37	static unsigned int nlists; /* number of elements in above list */
38
39
40	static int cmpfunc(const void a, const void b)
41	{
42	return (((struct cut_list *) a)->startpos -
43	((struct cut_list *) b)->startpos);
44
45	}
46
47	static void cut_file(FILE * file)
48	{
49	char *line = NULL;
50	unsigned int linenum = 0; /* keep these zero-based to be consistent */
51
52	/* go through every line in the file */
53	while ((line = xmalloc_getline(file)) != NULL) {
54
55	/* set up a list so we can keep track of what's been printed */
56	char * printed = xzalloc(strlen(line) * sizeof(char));
57	char * orig_line = line;
58	unsigned int cl_pos = 0;
59	int spos;
60
61	/* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
62	if (option_mask32 & (CUT_OPT_CHAR_FLGS \| CUT_OPT_BYTE_FLGS)) {
63	/* print the chars specified in each cut list */
64	for (; cl_pos < nlists; cl_pos++) {
65	spos = cut_lists[cl_pos].startpos;
66	while (spos < strlen(line)) {
67	if (!printed[spos]) {
68	printed[spos] = 'X';
69	putchar(line[spos]);
70	}
71	spos++;
72	if (spos > cut_lists[cl_pos].endpos
73	\|\| cut_lists[cl_pos].endpos == NON_RANGE)
74	break;
75	}
76	}
77	} else if (delim == '\n') { /* cut by lines */
78	spos = cut_lists[cl_pos].startpos;
79
80	/* get out if we have no more lists to process or if the lines
81	* are lower than what we're interested in */
82	if (linenum < spos \|\| cl_pos >= nlists)
83	goto next_line;
84
85	/* if the line we're looking for is lower than the one we were
86	* passed, it means we displayed it already, so move on */
87	while (spos < linenum) {
88	spos++;
89	/* go to the next list if we're at the end of this one */
90	if (spos > cut_lists[cl_pos].endpos
91	\|\| cut_lists[cl_pos].endpos == NON_RANGE) {
92	cl_pos++;
93	/* get out if there's no more lists to process */
94	if (cl_pos >= nlists)
95	goto next_line;
96	spos = cut_lists[cl_pos].startpos;
97	/* get out if the current line is lower than the one
98	* we just became interested in */
99	if (linenum < spos)
100	goto next_line;
101	}
102	}
103
104	/* If we made it here, it means we've found the line we're
105	* looking for, so print it */
106	puts(line);
107	goto next_line;
108	} else { /* cut by fields */
109	int ndelim = -1; /* zero-based / one-based problem */
110	int nfields_printed = 0;
111	char *field = NULL;
112	const char delimiter[2] = { delim, 0 };
113
114	/* does this line contain any delimiters? */
115	if (strchr(line, delim) == NULL) {
116	if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
117	puts(line);
118	goto next_line;
119	}
120
121	/* process each list on this line, for as long as we've got
122	* a line to process */
123	for (; cl_pos < nlists && line; cl_pos++) {
124	spos = cut_lists[cl_pos].startpos;
125	do {
126	/* find the field we're looking for */
127	while (line && ndelim < spos) {
128	field = strsep(&line, delimiter);
129	ndelim++;
130	}
131
132	/* we found it, and it hasn't been printed yet */
133	if (field && ndelim == spos && !printed[ndelim]) {
134	/* if this isn't our first time through, we need to
135	* print the delimiter after the last field that was
136	* printed */
137	if (nfields_printed > 0)
138	putchar(delim);
139	fputs(field, stdout);
140	printed[ndelim] = 'X';
141	nfields_printed++; /* shouldn't overflow.. */
142	}
143
144	spos++;
145
146	/* keep going as long as we have a line to work with,
147	* this is a list, and we're not at the end of that
148	* list */
149	} while (spos <= cut_lists[cl_pos].endpos && line
150	&& cut_lists[cl_pos].endpos != NON_RANGE);
151	}
152	}
153	/* if we printed anything at all, we need to finish it with a
154	* newline cuz we were handed a chomped line */
155	putchar('\n');
156	next_line:
157	linenum++;
158	free(printed);
159	free(orig_line);
160	}
161	}
162
163	static const char _op_on_field[] = " only when operating on fields";
164
165	int cut_main(int argc, char **argv)
166	{
167	char sopt, ltok;
168
169	opt_complementary = "b--bcf:c--bcf:f--bcf";
170	getopt32(argc, argv, optstring, &sopt, &sopt, &sopt, &ltok);
171	// argc -= optind;
172	argv += optind;
173	if (!(option_mask32 & (CUT_OPT_BYTE_FLGS \| CUT_OPT_CHAR_FLGS \| CUT_OPT_FIELDS_FLGS)))
174	bb_error_msg_and_die("expected a list of bytes, characters, or fields");
175	if (option_mask32 & BB_GETOPT_ERROR)
176	bb_error_msg_and_die("only one type of list may be specified");
177
178	if (option_mask32 & CUT_OPT_DELIM_FLGS) {
179	if (strlen(ltok) > 1) {
180	bb_error_msg_and_die("the delimiter must be a single character");
181	}
182	delim = ltok[0];
183	}
184
185	/* non-field (char or byte) cutting has some special handling */
186	if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
187	if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) {
188	bb_error_msg_and_die
189	("suppressing non-delimited lines makes sense%s",
190	_op_on_field);
191	}
192	if (delim != '\t') {
193	bb_error_msg_and_die
194	("a delimiter may be specified%s", _op_on_field);
195	}
196	}
197
198	/*
199	* parse list and put values into startpos and endpos.
200	* valid list formats: N, N-, N-M, -M
201	* more than one list can be separated by commas
202	*/
203	{
204	char *ntok;
205	int s = 0, e = 0;
206
207	/* take apart the lists, one by one (they are separated with commas */
208	while ((ltok = strsep(&sopt, ",")) != NULL) {
209
210	/* it's actually legal to pass an empty list */
211	if (strlen(ltok) == 0)
212	continue;
213
214	/* get the start pos */
215	ntok = strsep(&ltok, "-");
216	if (ntok == NULL) {
217	bb_error_msg
218	("internal error: ntok is null for start pos!?\n");
219	} else if (strlen(ntok) == 0) {
220	s = BOL;
221	} else {
222	s = xatoi_u(ntok);
223	/* account for the fact that arrays are zero based, while
224	* the user expects the first char on the line to be char #1 */
225	if (s != 0)
226	s--;
227	}
228
229	/* get the end pos */
230	ntok = strsep(&ltok, "-");
231	if (ntok == NULL) {
232	e = NON_RANGE;
233	} else if (strlen(ntok) == 0) {
234	e = EOL;
235	} else {
236	e = xatoi_u(ntok);
237	/* if the user specified and end position of 0, that means "til the
238	* end of the line */
239	if (e == 0)
240	e = EOL;
241	e--; /* again, arrays are zero based, lines are 1 based */
242	if (e == s)
243	e = NON_RANGE;
244	}
245
246	/* if there's something left to tokenize, the user passed
247	* an invalid list */
248	if (ltok)
249	bb_error_msg_and_die("invalid byte or field list");
250
251	/* add the new list */
252	cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
253	cut_lists[nlists-1].startpos = s;
254	cut_lists[nlists-1].endpos = e;
255	}
256
257	/* make sure we got some cut positions out of all that */
258	if (nlists == 0)
259	bb_error_msg_and_die("missing list of positions");
260
261	/* now that the lists are parsed, we need to sort them to make life
262	* easier on us when it comes time to print the chars / fields / lines
263	*/
264	qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
265	}
266
267	/* argv[0..argc-1] should be names of file to process. If no
268	* files were specified or '-' was specified, take input from stdin.
269	* Otherwise, we process all the files specified. */
270	if (argv[0] == NULL \|\| LONE_DASH(argv[0])) {
271	cut_file(stdin);
272	} else {
273	FILE *file;
274
275	do {
276	file = fopen_or_warn(argv[0], "r");
277	if (file) {
278	cut_file(file);
279	fclose(file);
280	}
281	} while (*++argv);
282	}
283	if (ENABLE_FEATURE_CLEAN_UP)
284	free(cut_lists);
285	return EXIT_SUCCESS;
286	}